tbc
Authors: Borgeaud, MauriceOrganisations: ESA, Italy
tbc
Authors: Borgeaud, Mauricetbc
Authors: Engdahl, MarcusTBC
Authors: Potin, Pierretbc
Authors: Rosich, Betlemtbc
Authors: Hajduch, GuillaumeBIOMASS was approved as the 7th Earth Explorer mission in May 2013 at the Earth Observation Programme Board in Svalbard, Norway. The overall objective of the mission is to reduce the uncertainty in the worldwide spatial distribution and dynamics of forest biomass in order to improve current assessments and future projections of the global carbon cycle. Its payload consists of a fully-polarimetric left-looking P-band SAR in order to reach the main objective to provide consistent global estimates of forest biomass, forest disturbance and re-growth parameters. The BIOMASS mission lasts 5 years, and consists of two phases, i.e. a tomographic and an interferometric phase. The Preliminary Design Review was completed in 2017 and the Critical Design Review will take place during Q2-2021. The satellite is due for launch in Q3-2023. The BIOMASS payload is a P-band Synthetic Aperture Radar (SAR), which will operate in Quad-Pol mode, in which the V-polarisation and H-polarisation pulses are transmitted alternatively and both the V- and H- polarisation backscattered signals are received simultaneously. The main mode is similar to a standard stripmap acquisition whereby the subswaths are accessed through a spacecraft roll maneuver. The Biomass payload consists of two elements, i.e. (1) the main electrical part of the instrument providing all of the functionality up to the Feed Array which transmits/receives the SAR signals as part of the offset fed antenna system; and; (2) the 12m Large Deployable Reflector (LDR). BIOMASS will be the first SAR sensor to operate in P-band from space thus facing strong (given the exploited low-frequency) distortions (e.g. polarimetric scattering matrix and geometric distortions) due to the propagation of the radar waveforms through the ionosphere. This calls for an important role of the cal/val activities, especially during the commissioning phase, to be able to properly characterise, calibrate and validate the system despite the presence of ionosphere related distortions in the acquired data. The presentation will cover the mission and system in more detail and will include the most recent status thereof.
Authors: Rommen, Björn; Willemsen, Philip; Leanza, Antonio; Hélière, Florence; Carbone, Adriano; Scipal, KlausTBC
Authors: Davidson, Malcom; Kubanek, Julia; Iannini, Lorenzo; Furnell, Rob; Gebert, Nico; Osborne, Steve; Geudtner, Dirktbc
Authors: Davidson, Malcom; Kubanek, Julia; Torres, Ramon; Geudtner, Dirk; Bibby, Davidtbc
Authors: Lopez-Dekker, Paco (1); Prats, Pau (2); Hooper, Andy (3); Biggs, Juliet (4); Mouginot, Jeremie (5); Kääb, Andy (6); Rott, Helmut (7)Wide swath radar imaging requires that the time interval to collect each radar pulse echo is large, and often can exceed the interpulse period. As it is difficult to both transmit and receive from the same antenna simultaneously, there will be “blind ranges” when the receive and transmit times overlap. This leads to gaps in the radar echo and thus degradation of system performance. Today most wide swath systems address this by segmenting the swath in range, such as in ScanSAR or TOPS mode operation, so that each subswath is within the range ambiguity limit. Some groups have started experimenting with variations of sweepSAR technology, in which the receive antenna tracks the radar echo across the swath, realizing a complete wide swath range scan but leading to the echo gaps. Here we look at minimizing the effect of blind ranges by varying the radar PRF. If the pulse times are selected randomly, the main effect is to raise the noise floor of the echoes. We develop a simple conceptual model of the noise added by random dithering, and find that the noise level depends not only on the fraction of pulses lost, but also on the amount of oversampling in the azimuth direction. Oversampling helps system performance as it can recover a portion of the missing azimuth spectrum- that portion that is lower in Doppler frequency. From these two parameters we can compute the noise, and add it to the other system noises in the error budget. These predict that even for losses around 10%the added noise is small. We also compute sample radar products by artificially inducing dithering as the images are generated. The simuated products, which are derived from initial actual data, follow the model. We find that geocoded magnitude images, interferograms, and correlation images show strong robustness with respect to dithering, demonstrating that choosing essentially random PRFs allows for accurate generation of SAR and InSAR data products while retaining wide swath, fine resolution, coverage. Our model reopresents a useful tool that allows system designers to obtain complete ground converage with wide swath imaging at high quality, by providing a way to balance operational parameters optimally.
Authors: Zebker, Howard AllanThrough extensive analysis of Sentinel-1 data, we observe a systematic, yet temporally fading, signal in the multilooked interferograms which especially compromises the reliability of InSAR-derived deformation velocity maps. The observed error is attributed to the systematic variation in the scattering properties of the sub resolution scatterers [1]. The phenomenon can be explained, for instance, by moisture variation, vegetation growth etc. Further research is needed to interpret the physical source and model the interferometric phase of such systematic signals. Complementary to the latter line of research, this work sheds light on other aspects of the problem. Here we initially demonstrate: the presence of the systematic effects in interferograms; the propagated error to the deformation velocity maps; the effect of processing algorithms on mitigating the error. More importantly we introduce an efficient processing scheme and propose an intermediate InSAR product which significantly reduces the observed effects. Furthermore, a phase quality measure is introduced which allows predicting the error budgets of deformation velocity maps in presence of the fading signal. Initially the bias was demonstrated for a four-year archive of Sentinel-1 over Sicily-Italy, as reported in [2]. In Sicily, the presence of the systematic errors is shown and the relation between the average phase bias of the interferograms to their temporal baseline is explored. From this study, we concluded that the multilooked interferograms contain a signal which attenuates with the temporal baseline, hence the name fading signal. In the current contribution, we report our observations on a larger and more coherent 5-year time series over Qinghai province-China in Tibetan plateau. The error in multilooked interferograms as well as the deformation velocity map is explored. The new test site in Tibetan plateau confirms the observation of fading signal in short temporal baseline interferograms. Note that the InSAR community mostly exploits these interferograms for Big Data processing. We further experiment the propagation of bias to the deformation velocity estimation by including different subset of interferograms in the analysis. For the first test site in Sicily by inclusion of up to 30-day interferograms we observe a deformation bias of 6.5 mm/yr. The bias decreases to 3 mm/yr when including up to 60-day interferograms. It further reduces to 0.25 mm/yr if all possible interferogram pairs within the data stack are exploited. For the more coherent test site in Tibetan plateau, we still observe a bias of -0.98 mm/yr by inclusion of up to 56-day interferograms. Similar to Sicily, the inclusion of more interferograms monotonically decreases the bias. Our recommendation for mitigation of the errors is to exploit all the possible interferograms and reconstruct the consistent phase component in a data adaptive manner. The reconstruction is performed by imposing the simple model of phase triangulation [3]. It shall be stressed that the major role in the suppression of the fading signal in phase triangulation is played by the inclusion of the long temporal baseline interferograms in the estimation. To address the demand for Big Data processing, we further introduce our efficient phase triangulation scheme [4]. [1] De Zan et al. 2015: https://ieeexplore.ieee.org/document/7147797 [2] Ansari et. al. 2020: Study of Systematic Bias in Measuring Surface Deformation with SAR Interferometry (to appear in IEEE Transactions on Geoscience and Remote Sensing subject to minor revisions) [3] Monti-Guarnieri and Tebaldini 2008: https://ieeexplore.ieee.org/document/4685949 [4] Ansari et al. 2017: https://ieeexplore.ieee.org/document/8024151
Authors: Ansari, Homa; De Zan, Francesco; Parizzi, AlessandroMulti-looking or spatial averaging is a common practice in InSAR time-series algorithms to reduce the stochastic noise over a neighborhood of distributed scatterers in InSAR measurements. However, multi-looking may break consistency among a triplet of interferometric phases formed from three acquisitions leading to a residual phase called closure phase errors. Understanding the cause of the closure phase errors in multi-looked InSAR measurements and the impact of closure phase errors on the performance of the InSAR time-series algorithms is crucial to quantify the uncertainty of the ground displacement time-series derived from InSAR measurements. It is reported that a fading signal exists in multi-looked InSAR measurements and result in systematic bias in InSAR time-series if only short temporal span subsets are used in the analysis [Ansari et.al. 2020]. The observed bias has been attributed to the lack of phase consistency in multi-looked interferograms. However, the observational reports on the bias lack a theoretical basis to relate the closure phase to the observed bias. Moreover, the cause of the closure phase errors remains controversial and the previous studies contradict on relating the closure phase to a physical process such as soil moisture change [DeZan et al, 2015] or to attribute the closure phase to the pure statistical properties of the SAR measurements [Molan et al, 2020]. In this work, we develop a model that can consistently explain both closure phase errors and fading signals in multi-looked interferometric measurements. We show that closure phase errors are indicators of local processes inside a multi-look window. We adopt a semi-quantitative approach where we represent inhomogeneity with two groups of measurements with distinct Probability Density Functions (PDFs). Though it is possible to describe inhomogeneity with a continuous distribution such as the work presented by Zwieback and Meyer, 2020, we find a simple dyad model sufficient to provide useful insights. We show that fading signals are manifestations of local processes that satisfy the following two requirements: (1) the process alters both phases and correlations of radar measurements, and (2) the correlation of the process diminishes with time. Further, we propose a method to estimate the systematic bias in the InSAR time-series with generalized closure phase measurements. We validate our model with a case study in Barstow-Bristol trough, California. We find systematic differences on the order of cm/year between InSAR time-series results using subsets of varying maximum temporal baseline. We show that these biases can be identified and accounted for with closure phase measurements.
Authors: Zheng, Yujie (1); Fattahi, Heresh (2); Agram, Piyush (1); Simons, Mark (1,2)ABSTRACT Non-stationary coherence and rapid soil movement in alternating directions makes correctly tracking the deformation phase very difficult when using standard distributed scatterer (DS) time-series InSAR techniques [Morishita & Hanssen 2015, Samiei-Esfahani et al. 2016]. We find that unwrapping in the time domain presents a particularly great obstacle for these conventional techniques. For instance, peaty soils can exhibit rapid uplift during rainy conditions which often causes phase ambiguity resolution routines to estimate deformation in the wrong direction [Heuff & Hanssen 2020]. However, it is possible to reduce the set of possible solutions for the unwrapped phase by applying some simple physics-based constraints to the system. Such treatment is made possible when analysing the scene from a data-driven perspective. In this analysis, we consider multilooked regions as the basic unit of measure in our DS InSAR processing, which for simplicity we will refer to here as points. By clustering like-behaved DS “points” together based on their unwrapped phase time-series, we find groups of similarly deforming regions within the scene. The clustering of points is accomplished quickly and efficiently using the t-distributed stochastic neighbour embedding (t-SNE) algorithm [van der Maaten & Hinton 2008, Van de Kerkhof et al. 2019]. This is an “off-the-shelf” dimensionality-reduction algorithm which allows the user to reduce high-dimensional data by embedding it onto a two-dimensional plane. Positions on this plane are determined by the relative similarity of the signals being compared. What this means for our application is that InSAR time-series which display similar phase-temporal behaviour will be grouped together by the algorithm. By cross-referencing these t-SNE groups with maps of soil type, groundwater level and with precipitation signals, we find alternative deformation hypotheses with which we can improve the phase unwrapping solution. This can be as simple as ruling out an unwrapping direction based on the soil type and the presence of rain between epochs. Our current work includes an assessment of the InSAR coherence matrix as a measure of confidence in the phase unwrapping solution so as to weight the standard vs contextual unwrapping solutions. By ruling out low confidence unwrapping solutions deemed unphysical, we are able to significantly mitigate the errors present in the final deformation time series. REFERENCES F. Heuff and R. F. Hanssen, InSAR Phase Reduction Using the Remove-Compute-Restore Method, IGARSS 2020 Y. Morishita and R. F. Hanssen, Temporal Decorrelation in L-, C-, and X-band Satellite Radar Interferometry for Pasture on Drained Peat Soils, IEEE Transactions on Geoscience and Remote Sensing, 2015, 53, 1096-1104 S. Samiei-Esfahany, J. E. Martins, F. van Leijen and R. F. Hanssen, Phase Estimation for Distributed Scatterers in InSAR Stacks Using Integer Least Squares Estimation, IEEE Transactions on Geoscience and Remote Sensing, 2016, 54, 5671-5687 B. van de Kerkhof, V. Pankratius, L. Chang, R. Van Swol, R. F. Hanssen, Individual scatterer model learning for satellite interferometry. IEEE transactions on geoscience and remote sensing, 2019, 58(2):1273-80. L. van der Maaten and G. Hinton, Visualizing data using t-SNE, Journal of Machine Learning Research, 2008
Authors: Conroy, Philip; Bruna, Marc; Hanssen, Ramon FInSAR phase observations are wrapped and known only modulo 2pi. Recovering the unambiguous phase from the wrapped phase by adding integer numbers of 2pi, known as phase unwrapping, is a key step to obtain the continuous phase field for DEM generation and surface deformation mapping. Wrong integer numbers of 2pi added to the wrapped phase during phase unwrapping, to which we refer as unwrapping errors, could bias the height or displacement measurements. A better understanding of this error and its contribution to the uncertainty budget is crucial to quantify the uncertainty of InSAR displacement products. Moreover, a realistic InSAR uncertainty model that accounts for unwrapping errors is needed for the performance analysis of future SAR missions during design, development and operation. We quantify the uncertainty of InSAR observations due to unwrapping errors using a statistical approach with data simulations. For this purpose, we simulate realistic interferometric phase contributions from various sources, including decorrelation using the observed coherence from C- (Sentinel-1) and L-band (ALOS-1) missions, dry and wet tropospheric delay using ERA5 and spatial spectra database from InSAR/MODIS/MERIS, and long spatial wavelength contribution from ionosphere using GNSS-based Total Electron Content. We realize the simulation over multiple datasets with different noise characteristics, unwrap each realization with different algorithms including the Minimum-Cost-Flow and evaluate unwrapping errors in terms of likelihoods at pixels, regions and SAR scene levels. Results of the simulation realizations in different regions demonstrate that connected components are reliable indicators of potential unwrapping errors with
Authors: Yunjun, Zhang (1); Fattahi, Heresh (2); Agram, Piyush (1); Rosen, Paul (2); Simons, Mark (1)Synthetic Aperture Radar (SAR) can measure the relative range change between the satellite antenna and the ground surface in two ways: a) using interferometry via exploiting the coherent nature of the complex electromagnetic wave or b) using speckle tracking via estimating the relative shift between image pairs referred to as offsets. The latter is usually based on the cross-correlation of complex (coherent speckle tracking) or intensity (incoherent speckle tracking) images. Although estimating displacements with relatively lower spatial resolution and precision, SAR offsets have some advantages compared to SAR interferometry. First, SAR offsets do not require phase unwrapping, which is computationally expensive and error prone. Second, SAR offsets are spatially absolute measure, without the need of a reference point if biases are carefully corrected for, making it easy to stitch/mosaic and desirable for large scale displacement mapping. Third, the simpler workflow of the offset mapping makes it big data friendly and more easily to scale the estimation on continental or global extents. SAR offsets have traditionally been used to map fast surface deformation processes, such as ice shelves and streams, earthquakes, and landslides. With the advance of GPU-based cross-correlation technique and the availability of long time-series of SAR images, it is of interest to evaluate the potential of using SAR offset time-series to measure slow surface deformation such as tectonic and volcanic processes. The uncertainty of ground displacements estimated from SAR offsets is governed by the uncertainty of the offset estimation and other noise sources including the orbital error, atmospheric propagation delays and ground surface motions such as the tidal and loading effects. We demonstrate the impact of noise reductions in the SAR offset time-series. More specifically, we correct for the ionospheric delay using the public IGS Total Electron Content (TEC) maps or JPL’s higher temporal and spatial resolution TEC maps, for the tropospheric delay using the ERA-5 weather re-analysis dataset and for the solid Earth tides following the International Earth Rotation and Reference Systems Service conventions. We evaluate the quality of the offset velocity using the offset triangulation residual and quantify the uncertainty of SAR offset time-series and velocity through the error propagation in a linear observation system. Results from the range offset time-series of a stack of Sentinel-1 data over southern California shows the standard deviation of the linear velocity reduces from 0.5 cm/year to 0.2 cm/year after the noise reduction. We observe a systematic bias of ~10 cm in the range offset estimated between the Sentinel-1 A and B sensors. The velocity field across the southern San Andreas Fault system shows ~1.5 cm/year ground displacement in LOS direction which is compatible with the total expected plate motion across the plate boundary and consistent with the independent measurements from GNSS analysis.
Authors: Yunjun, Zhang (1); Fattahi, Heresh (2); Brancato, Virginia (2); Rosen, Paul (2); Simons, Mark (1)Interferometric Synthetic Aperture Radar (InSAR) technique has been proven to be capable of monitoring displacements associated with earthquakes, volcanos, glacier movements, landslides and so on. However, we can retrieve only one-dimensional (1-D) displacement along the line-of-sight (LOS) direction with the InSAR technique, yielding possible misinterpretation of geohazards which generally occur in three-dimensional (3-D) framework. At present, the estimations of three-dimensional (3-D) displacements with InSAR are basically based on the Weight Least Squares (WLS) method, which integrates heterogeneous InSAR measurements provided by different satellites/tracks/algorithms on a pixel-by-pixel basis. The reliability of this method depends on the exact accuracies/weights of the heterogeneous InSAR measurements, which is however quite difficult to determine due to the complicated inherent errors in the InSAR measurements. Here, we present a novel method for mapping the accurate 3-D displacements from heterogeneous InSAR measurements based on Strain Model (SM) and Variance Component Estimation (VCE) (SM-VCE) algorithm. This method can be implemented by generally three steps. Firstly, the observation functions are established based on the SM, where the SM represents the spatial correlation of the homogeneous points' displacements when a portion of earth’s surface is deformed by a geodynamic process (e.g., intrusion of magma or activity of fault). Then, the exact weights of the heterogeneous InSAR measurements can be determined by the VCE algorithm with the SM-based observation functions. In this way, the accurate 3-D displacements can be finally estimated with the WLS method. Particularly, we also propose a SM-based Adaptive Neighborhood (SMAD) determination strategy to only involve the homogeneous displacement measurements in the spatial domain when establishing the observation functions. Simultaneously, the precisions of the involved heterogeneous InSAR observations and the inferred 3-D displacements can be quantitatively provided by this method from a posteriori way. The proposed method is assessed with both simulated and real case studies. In the case study of the 2007 eruption of Kilauea Volcano (Hawai’i), improvements of 51.2%, 22.4%, and 18.5% have been achieved for the derived east, north, and up displacements, respectively, compared with those derived from the classical weighted least squares method. With respect to the 2016 Tottori earthquake, we estimated the 3-D displacements based on the SM-VCE method from ascending/descending left-/right-looking InSAR measurements, and conducted the first precision assessment for InSAR measurements in a posteriori way. To further estimate the 3-D displacements associated with the 2016 Mw7.8 Kaiköura earthquake which is known as an extraordinary complex, more than 20 fault ruptured event, an SMAD determination strategy was proposed to assist establishing the observation functions. The comparison with the GPS observation clearly demonstrates the capability of the presented method in retrieving 3-D displacements from heterogeneous InSAR measurements.
Authors: Liu, Jihong (1); Hu, Jun (1); Li, Zhiwei (1); Zhu, Jianjun (1); Wu, Lixin (1); Sun, Qian (2)Whilst most medium-large magnitude earthquakes are detected by global networks of seismometers, many other important transient deformation phenomena, for example slow-slip events, are difficult or impossible to detect using seismological data. It is important to characterize when and where such events have occurred in order to accurately estimate seismic hazard and improve our understanding of fault mechanics. InSAR offers a potential alternative for the detection of transient small-magnitude or aseismic phenomena, and the Sentinel-1 constellation provides the frequent, regular and global SAR coverage required for this task, with data acquired over the majority of global tectonic belts on a weekly-to-daily basis. However, both the size of this dataset and the large magnitude of atmospheric and other nuisance signals, relative to tectonic signals of interest, makes this task difficult and precludes systematic manual analysis. In order to address this issue, here we have developed a new deep-learning based approach for the automatic identification of transient deformation events in noisy time-series of InSAR images, without requiring supervision or labelling of known example events. To achieve this, we have adopted an anomaly detection framework where anomalies correspond to any transient phenomena that deviates from the ‘normal’ spatio-temporal pattern of phase-change arising from a combination of atmospheric signals, time-constant deformation-rates, orbital errors and other unwanted signals. Our novel workflow learns such patterns in the InSAR dataset, leveraging the unique three-dimensional structure of the interferogram stack and its relationship to the unknown 2D fields of nuisance non-tectonic signals that correspond to individual SAR acquisition dates (epochs). Our unsupervised and event-agnostic approach is preferable to previously-published supervised, task-oriented approaches because it focuses on learning from sequential spatial changes in phase over time for all cases. In detail, our framework consists of fully convolutional autoencoders that embed the feature encoding of a sequence of interferograms and then decode them to an estimation of their corresponding epochs. The autoencoders consist of a combination of time-distributed convolutions, convolutional LSTM (Long Short-Term Memory) cells and fully connected layers that are trained on an InSAR dataset of a fixed size (for each iteration there are 26 interferograms made up from 9 epochs covering a 14 km by 12 km area). For every sequence we train our autoencoder with feature integration from its previous sequence, which ensures temporal consistency within the whole sequence. During testing, when an anomalous sequence is passed to the network, it reconstructs it with large residuals that serve as an indicator for detection of an anomaly. To train and test our method, we use InSAR data from several Sentinel-1 tracks in Turkey, obtained from the LiCSAR processing system developed by the UK’s Centre for the Observation and Modelling of Earthquakes, Volcanoes and Tectonics. Our initial findings show that our unsupervised and event-agnostic pipeline accurately detects both real and synthesized anomalous signals and recovers both the spatio-temporal structure of flagged deformation events and the time-series of non-deformation ‘nuisance’ signals. Our method therefore offers significant potential for future automated detection of transient deformation from InSAR big-data, and will likely be of wide utility for detection of small and slow deformation signals caused by a variety of processes, ranging from tectonic, to volcanic, to anthropogenic.
Authors: Shakeel, Anza (1,3); Walters, Richard (1,3); Al Moubayed, Noura (2); Allen, Mark (3)Satellite radar interferometry (InSAR) is a powerful technique for monitoring deformation phenomena. While deformation phenomena occur in a three-dimensional (3D) world, one of the limitations of the InSAR phase observations is that they are only sensitive to the projection of the 3D displacement vector onto the radar line-of-sight (LoS) direction. To uniquely estimate the three displacement components, we would require at least three spatiotemporally coinciding independent (STCI) LoS observations, i.e., scatterers on an object that is not subject to internal deformations, observed at the same time. More importantly, the system of equations needs to have a full rank coefficient matrix. Unfortunately, in most practical situations at most two STCI LoS observations are available, resulting in an underdetermined system with an infinite amount of possible solutions. Within the InSAR literature we encounter different approaches to address the underdeterminancy problem, unfortunately often with either mathematical or semantic flaws. Their impact reaches from quantitative errors in the reported studies, mismatches in comparative studies with other geodetic techniques, a lack of trust in the technology by end users, to plain confusion. Here we perform an in-depth literature review of InSAR studies, classifying main categories of InSAR fallacies and analysing their impact. Moreover, we propose both a uniform nomenclature and an alternative approach to the standard 3D decomposition problem using the concept of a strap-down reference system. Fallacies on projection, decomposition, attribution, and assumption Attribution error: Attributing a line of sight estimate to a vertical displacement. In some studies, no projection statements are given, and LoS observations are directly interpreted as vertical displacements. This is a false statement that results in a systematic underestimation (bias) of vertical displacements of up to 30%. Projection error: Projecting the LoS displacement estimations onto the vertical, and presenting this as ‘vertical displacements’, or the equivalent situation in a particular horizontal direction. While ‘projection onto the vertical’ is a correct statement, ‘vertical displacement’ is not, since it relies on the assumption that any non-vertical displacement component of the 3D vector is non-existent. Moreover, while the term ‘projection onto the vertical’ is in all cases correct, both geometrically as well as semantic, the term ‘vertical displacement’ can only be correctly interpreted if the assumption of a non-existent horizontal component is correct and mentioned. Since this assumption is in many cases incorrect (landslides, subsidence bowls), it leads to a biased estimation instead of a more noisy estimation, which often has a bigger impact an less chance of being detected. Decomposition error: Ignoring the null-space in the 3D solution space using only one or two viewing geometries, i.e., ascending and descending, and subsequently equating a not-measurable displacement vector component to a not-existing displacement vector components. In such cases, it is assumed that the lack of sensitivity in the north-south component for near-polar orbits is equivalent to the absence of a north-south component, by simply removing the component from the decomposition equation. However, this assumption fails since the heading of the ascending and descending track are not completely parallel and neglecting the north-south component will results in biased estimates for the east-west and vertical component. Therefore, these approaches are very dependent on the actual magnitude of the north-south displacements. Flawed assumptions The problem of estimating 3D displacement vectors observed by only one or two viewing geometries can only be solved by adding additional information, in the form of assumptions. These need to be explicitly stated in the documentation, but also in the final products. However, in many cases, these assumptions are either lacking, misstated, incorrect, or implausible. The consequence of flawed assumptions typically result in biased results, rather than noisy results. Correct approaches and semantics LoS unaltered as the final product . This product is obviously correct, as it does not attempt to do any projection, attribution, or decomposition. The drawback of the LoS product is that it is typically hard to interpret, especially for untrained end users. As potential vertical and horizontal displacement components are projected onto the LoS and superposed, it remains ‘invisible’ what happens in the real world. “Projection-onto” products. Another correct option is to present the projection of the LoS onto the, e.g., vertical direction. In this case, it must be made clear that the results are displacement projections and not the vertical displacements itself to avoid misinterpretation. 2D Decomposition with valid, plausible, and explicit assumptions While assumptions that east-west displacements are more likely than north-south displacements are very implausible in most of the cases, there are several assumptions for which plausibility would be hardly disputed. For example, for gravity-induced downslope displacements, it is near-impossible that there is a displacement component parallel to the elevation contours, it is near-impossible that for a subsidence-bowl there is a displacement component parallel to the LoS displacement contours, and it is very unlikely to have significant secular displacements in the longitudinal direction of horizontal line infrastructure (Chang et al. 2018). Unification and nomenclature It is clear that there is need for further unification in the way in which InSAR results are computed and communicated, and for insight in the consequences of particular choices in terms of accuracy and precision. We propose a systematic approach that gives physically more relevant estimates for the deformation signal for cases where only two LoS observations are available. We also propose a method to give an estimate for the bias that is a result of the method. Strap-down reference system We propose to use a local, strap-down, right-handed Cartesian coordinate system that is fixed to the deformation phenomenon with transversal, longitudinal, and normal (TLN) components, extending the work of Chang et al. (2018). For many practical cases, such as line-infrastructure, landslides, or subsidence bowls, analysis of the main driving forces supports the assumption that significant deformations in the longitudinal direction are unlikely. As a result, it is possible to consider the real world displacement vector in two dimensions, as long as we choose the orientation of our local TLN coordinate system correct. A misalignment error of the frame leads to an assumption-error: the assumption of zero longitudinal displacements is no longer valid, resulting in biased estimates for the transversal and normal displacement component. We found that it is possible the express the uncertainty for the transversal and normal displacement components in the uncertainty of the alignment of our frame. Chang, L., Dollevoet, R.P. and Hanssen, R.F., 2018. Monitoring line-infrastructure with multisensor SAR interferometry: Products and performance assessment metrics. IEEE journal of selected topics in applied earth observations and remote sensing, 11(5), pp.1593-1605.
Authors: Brouwer, Wietske; Hanssen, RamonDescribing the precision of InSAR data in the form of second statistical moments or a noise covariance matrix, and using the inverse of the covariance matrix as a proper weight matrix in estimation/modeling processes is a common practice in InSAR-based Geo-modeling. In this view, the correct description of the spatial or temporal noise components is indispensable for stochastic modeling and the construction of the full covariance matrix of InSAR observations. While there are various studies in the literature regarding noise characteristics in InSAR measurements, the effect of timeseries processing on the spatio-temporal variability of the noise components have been overlooked in the available InSAR stochastic models. Note that InSAR processing steps, mainly the spatio-temporal atmospheric filtering, affects InSAR deformation results, and consequently it alters the spatio-temporal noise structure of InSAR deformation estimates. Therefore, stochastic properties of the final InSAR deformation timeseries are significantly affected by the filtering step. As the filtering step and its setting varies from case to case, it is difficult to derive a generic stochastic model for InSAR deformation timeseries. Here, we propose to derive InSAR stochastic model parameters directly from final InSAR timeseries, in a data-driven approach. The key concept of the proposed method is to estimate the noise components from InSAR timeseries over a deformation signal-free assumed area. In other words, we use InSAR data from an assumedly stable area, and then we reduce/subtract all the potential residual signals in order to isolate the effect of the measurement noise and residual atmospheric signal delay components. We provide an analytical formulation and a step-wise algorithm to calculate the variances of all spatio-temporal deformation measurements and covariances among them. To illustrate the approach, the measurement noise properties of InSAR data over the Groningen area, The Netherlands, for both RadarSAT2 and Sentinel-1 datasets were investigated through a spatio-temporal variogram analysis over a stable (signal-free) area. The results showed the presence of three different noise components in the data: i) white noise (nugget effect), ii) spatially correlated noise, and iii) temporally correlated noise. The modeling of the empirical variograms showed that the temporal and the spatial components can be modeled in a dis-joint manner. For both the RadarSAT2 and Sentinel-1 datasets, the parameters of the noise components were estimated. Furthermore, we propose a simple analytical approximation to propagate the covariance matrix of InSAR data to the covariance matrix of InSAR reduced datasets. Note that, due to large volumes of InSAR datasets (and consequently their large covariance matrices), an effective data reduction method is usually applied in practice on the data. Hence, for describing the quality of a reduced dataset, the covariance matrix of InSAR data should be propagated to the covariance matrix of the reduced dataset. The main objective here is to derive a simple analytical approximate equation for the elements of the covariance matrix of the reduced dataset. We only considered averaging-based reduction methods, i.e. the methods that are based on unweighted averaging either in time, in space, or in both. The proposed approach has been tested and validated via a simulated study and was demonstrated on real data. The proposed approximation can construct covariance matrices with the correct structure. However, it usually underestimates the (co)variance values, especially for small averaging intervals. The underestimation is mainly linear and can be mitigated by calibrating a scaling factor.
Authors: Samiei Esfahany, Sami (1,2); van Leijen, Freek (2); Hanssen, Ramon (2)The result of a multi-temporal InSAR analysis is usually a sparse grid of points for which displacement data (corresponding to differential range variations) are estimated over the time span of the radar data-stack. Over the last two decades, different techniques have been developed. Some algorithms rely on the identification of pointwise, persistent (or “permanent”), scatterers (PS) [1], while others exploit distributed scatterers (DS), exhibiting good phase coherence in a large number of SAR interferograms [2]. More recently, following the approach outlines in [3], a growing number of algorithms aim to exploit both families of radar targets (PS and DS), trying to maximize the spatial density of measurement points over the area of interest and properly combining the information associated to all possible interferometric pairs. Interestingly enough, amplitude data (and sometimes even the most basic of their statistical parameters: i.e. the mean and the standard deviation) are rarely presented as a standard output of an InSAR analysis and the displacement time series of the measurement points are rarely shown together with their associated amplitude values. Yet, a synoptic analysis can prove extremely interesting and informative. In this paper, we argue that amplitude SAR data should be considered as a key information source supporting precision assessment and quality checks (e.g. outlier removal) during multi-temporal InSAR data processing, where the attention of radar specialists is usually focused on phase data only. RCS values should be considered independently of the family of scatterers (PS or DS) targeted as measurement points. In fact, the statistical analysis of amplitude data can significantly increase the quality of InSAR results [4]. Of course, rather than just tools for automatic quality checks, time series analysis of the amplitude values can pave the way to new algorithms capable of managing semi-PS or temporary coherent scatterers, a topic which is already attracting a considerable degree of attention within the InSAR community. In fact, multi-year InSAR monitoring services should not be designed based on the hypothesis that all measurement points identified at the beginning of the project will stay coherent during the whole monitoring phase. Coherent scatterers can disappear, while new good radar targets can become available during the project. The statistical analysis of amplitude time series can then become a very important tool for properly managing temporary coherent scatterers. Rather than providing a set of recipes ready for implementation, this paper aims at triggering further discussions among InSAR practitioners, showing results obtained from the analysis of large data-sets acquired by different sensors, at different frequencies and spatial resolutions. We claim that the statistical analysis of amplitude data, should not be overlooked and should become a standard tool for quality check before data validation and calibration with GNSS data. BIBLIOGRAPHIC REFERENCES [1] Ferretti, A., Prati, C. and Rocca, F. “Permanent scatterers in SAR interferometry.” IEEE Transactions on geoscience and remote sensing, 39(1), pp.8-20, 2001. [2] Berardino, P., Fornaro, G., Lanari, R. and Sansosti, E. “A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms.” IEEE Transactions on geoscience and remote sensing, 40(11), pp.2375-2383, 2002. [3] Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F. and Rucci, A. “A new algorithm for processing interferometric data-stacks: SqueeSAR.” IEEE Transactions on Geoscience and Remote Sensing, 49(9), pp. 3460-3470, 2011. [4] Ferretti, A., Colesanti, C., Perissin, D., Prati, C. and Rocca, F. December. “Evaluating the effect of the observation time on the distribution of SAR permanent scatterers.” In Proc. Fringe, pp. 1-5, 2003.
Authors: Ferretti, Alessandro; Fumagalli, Alfio; Novali, Fabrizio; Rucci, AlessioIn radar interferometry, the method of storage for coherent scatterers and their attributes directly influences the ability for interpretation. Typically, modern InSAR time-series methods produce millions of measurement points, which can be point scatterers as well as distributed scatterers (Feretti et al. 2001, Berardino et al. 2002). Particularly in the built environment, these measurement points can exhibit a multitude of displacement characteristics, which depend on the particular driving mechanisms involved. This hampers an unambiguous interpretation of the data. Considering the complexity of InSAR data, and their typical high data-volume requirements, the need for a consistent and queryable spatial data-platform becomes relevant. Our proposed method uses the concept of a space-time matrix to store the coherent scatterers, implemented by means of a relational spatial database. With the dimensionality of the system accounted for, the stored InSAR data are partitioned in modules of the displacement time series, inherent scatterer and processing-related attributes, and their corresponding contextual attributes. The modularity of the system relies on the use of (primary and foreign) key constraints for the set up inter-table relations. The database implementation supports both spatial and non-spatial data in various formats, i.e. InSAR data is usually delivered in delineated format (e.g., csv, spreadsheets), whereas most of the used (spatial) contextual data consist of vector and/or raster-based files and services. Furthermore, data and context-driven queries and join procedures are facilitated with the use of spatial indices, such as a Generalized Search Tree (GiST). Our method is currently used to study the deformation behavior of the Delfland and Schieland areas, the Netherlands, using InSAR data, processed from Sentinel-1 and TerraSAR-X SAR acquisitions together with contextual attributes from the Dutch base registries and other relevant open data sources. As such, the contextualization of coherent scatterers in a modular framework helps us to better understand and “label” scattering behavior of physical objects that are present in both the natural and built environment. P. Berardino, G. Fornaro, R. Lanari, and E. Sansosti. A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing, 40(11):2375—2383, 2002. A. Ferretti, C. Prati, and F. Rocca. Permanent scatterers in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 39(1):8–20, January 2001.
Authors: Bruna, Marc FD; van Leijen, Freek J; Hanssen, Ramon FMulti-temporal InSAR (MTI) displacement time series are usually characterized by the merit figure known as temporal phase coherence, which is the vector sum on the complex residues left after removing a model phase trend, which is imposed a priori. Most MTI detection algorithms work in fact by detecting stable scatterers through maximization of the temporal coherence, so the choice of the temporal model is doubly important. Many processing chains make use of a linear model (constant velocity), while others use a wider repertoire, such as polynomial or periodic functions [1, 2]. Some MTI algorithms do not explicitly impose an a priori phase model, but nevertheless rely on some smoothness criteria in the temporal dimension to identify the stable pixels [3]. Clearly, then, the temporal coherence figure is not an exhaustive parameter to ascertain a posteriori the presence of useful information in a given MTI pixel time series, and it should be used with caution as a stable point detection criterion, as time series with low noise, following models different from the ones postulated in the detection step, will exhibit low coherence values, together with uninteresting, noisy time series. Moreover, one may be interested in discovering temporal trends which do not follow any of the predefined models used in the processing. This is especially true in cases when PS time series are produced independently and subsequently passed to application specialists, with the task of extracting useful information about the territory. It should also be considered that MTI datasets are increasingly made available over wide areas, such as entire national [4, 5] and even continental scales, with computational loads which discourage extended use of a priori models more complex than the simplest ones. Thus, typically, additional tests have to be performed to recognize other “interesting” but non-modeled trends, and some automated approaches to this task have been proposed to date [6]. We propose here the fuzzy entropy (FuzEn), a quantity originally developed for medical time series analysis, as a viable parameter to characterize multi-temporal InSAR time series, in order to isolate smooth, or temporarily smooth trends without a predefined model. FuzEn [7] basically measures the degree of regularity of a given time series by comparing short sub-sequences of samples according to a given distance measure, typically the Chebyshev distance in the original formulation, which can be easily substituted by e.g. a distance defined on the circle to deal with possible unwrapping errors. Being a measure of disorder in a time series, FuzEn exhibits homogeneously low values for a large class of displacement models, such as seasonal, parabolic or piecewise linear signals, or series with a few discontinuities, while increasing for more chaotic trends, dominated by noise. It appears therefore suited as a model-free discriminative parameter to isolate a few meaningful MTI pixels time series within large datasets. It also exhibits good potential as an a priori criterion for stable scatterer detection. The calculation of FuzEn has low computational cost and can thus be easily performed in batch, as a pre-screening filter. In the presentation, some results over simulated data and some examples on a real dataset are shown, with interesting performances which hint to possible large-scale implementations. References [1] M. Crosetto, O. Monserrat, M. Cuevas-González, N. Devanthéry, G. Luzi, and B. Crippa, “Measuring thermal expansion using X-band persistent scatterer interferometry,” ISPRS J. Photogramm. Remote. Sens., vol. 100, pp. 84–91, may 2015. [2] Y. Morishita and R. F. Hanssen, “Deformation Parameter Estimation in Low Coherence Areas Using a Multisatellite InSAR Approach,” IEEE Transactions on Geosci. Remote. Sens., vol. 53, no. 8, pp. 4275–4283, 2015. [3] P. Berardino, G. Fornaro, R. Lanari, and E. Sansosti, “A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms,” IEEE Transactions on Geosci. Remote. Sens., vol. 40, no. 11, pp. 2375–2383, 2002. [4] M. Caro Cuenca, R. F. Hanssen, A. J. Hooper, and M. Arikan, “Surface deformation of the whole Netherlands after PSI analysis,” in FRINGE Workshop Proceedings, pp. 1–27, 2013. [5] M. Costantini, et al., “Analysis of surface deformations over the whole Italian territory by interferometric processing of ERS, Envisat and COSMO-SkyMed radar data,” Remote. Sens. Environ., vol. 202, pp. 250–275, dec 2017. [6] M. Berti, A. Corsini, S. Franceschini, and J. P. Iannacone, “Automated classification of Persistent Scatterers Interferometry time series,” Natural Hazards and Earth System Science, vol. 13, no. 8, pp. 1945–1958, Aug. 2013. [7] Weiting Chen, Zhizhong Wang, Hongbo Xie, and Wangxin Yu, “Characterization of Surface EMG Signal Based on Fuzzy Entropy,” IEEE Transactions on Neural Syst. Rehabil. Eng., vol. 15, no. 2, pp. 266–272, 2007.
Authors: Refice, Alberto; Bovenga, Fabio; Pasquariello, GuidoSAR interferometry (InSAR) is nowadays an established tool for measurement of millimetric ground deformations from satellite. Several applications and operational projects have shown its effectiveness and confirmed the achievable accuracy, using data from the various available sensors. Recently, the European satellite constellation Sentinel-1 has made possible the systematic observation of very large areas (the swath of a single image is about 250 km x 200 km, and the revisit time 6 days). The possible size of areas analysed by InSAR is therefore moving from regional to national scale or even continental scale: projects like the European Ground Motion Service for the analysis of ground motion over the whole Europe are on the track. On the other hand, X-band SAR constellations such as Cosmo-SkyMed first and second generation, TerraSAR-X and PAZ, make possible to obtain much more detailed measurements on relatively smaller areas (swaths smaller than 40 km x 40 km). Last, L-band SAR satellites such as ALOS and SAOCOM are capable of measuring, although with lower precision, larger deformations, also in correspondence of vegetated areas. In this scenario, where the areas to monitor are getting bigger and the time of observation longer (but high spatial and temporal resolution and high accuracy for the deformation measurements must be kept), a robust and accurate data fusion method is necessary for homogenizing the InSAR measurements obtained from different sensors with different acquisition characteristics. This is the subject of this work. In order to make the discussion more clear, in the following we will call cluster the set of displacement measurements obtained by a single SAR processing, referring to the same spatial and temporal reference. A cluster is a sequence of displacement maps (one for each acquisition date of the input stack), which we will call layer. It is important to note that the displacement measurements obtained by InSAR represent the projection along the sensor line of sight (LOS) of the three-dimensional ground deformations. In addition, InSAR provides relative and not absolute measurements: they are referred spatially to a point properly chosen within the analysed frame, and temporally to an acquisition date of the stack. Among the other things, it is also worth remembering that, in order to mitigate atmospheric and orbital artefacts, interferometry techniques require specific spatio-temporal filtering steps. Unfortunately, atmospheric and orbital noise are mixed with the deformation signal of interest, and cannot be completely separated. In practice, the filtering procedure has the side effect of a deformation signal attenuation: the smaller is the cluster spatial extension, the larger is the attenuation effect. Therefore, clusters with significant differences in the spatial dimension could have relevant differences in terms of the capability to detect large-scale ground deformations. For the previously described reasons, it is clear that measurements belonging to different clusters, even if covering the same area, cannot be merged or mosaicked without applying a proper data fusion procedure. In this work, we propose a technique for optimally fusing the InSAR measurements obtained from stacks of SAR images collected by different sensors (with different characteristics in term of acquisition geometry, spatial and temporal resolution, spatial coverage, achievable accuracy, etc…). The algorithm reconstructs (when the acquisition geometries of the different clusters make it possible) the horizontal and vertical components of the ground deformation. The proposed approach also makes it possible to optimally integrates, when available, independent in situ measurements (e.g., from GNSS, optical levelling, ground based SAR, etc.). The proposed integration and harmonization algorithm exploits the fact that the displacement measurements obtained by the different sensors must be consistent, in order to: - Determine the temporal constants to be applied to each point of a cluster in order to refer the measurements to a single date, and the spatial constant to be applied to each layer in order to have a common spatial reference - Recover large-scale deformation patterns (typically removed by spatio-temporal filters during InSAR processing) - Integrate external in situ measurements (e.g., from GNSS, optical levelling, ground based SAR, etc.), which can be useful to absolutely calibrate InSAR measurements and further constraint the solution. Given N sets of terrain displacement measurements (clusters) obtained by independent SAR interferometry processing, and, if available, a set of in situ measurements (T), the proposed method: - Calibrate in the space domain the N input clusters by using appropriate correction functions (parametric surfaces S) to be applied to each of the input layers in order to simultaneously minimize the differences between layers in the overlapping areas and (if available) the differences between layers and external ground deformation measurements - Align the N sets in the time domain by using specific calibration constants to refer the N clusters to a common reference time. In order to find a unique solution, a further condition needs to be enforced. We have made experiments considering either one or the other of the following two conditions: - Minimizing the corrections to be applied to each layer - Minimizing the estimated displacement for most part of the observed area. As a result of the fusion procedure, measurements with reduced noise are obtained in the areas of overlap between the different sets of measurements. In the remaining areas, an optimal mosaicking of the different datasets is obtained. The application to simulated data have demonstrated the effectiveness of the proposed method to reconstruct the actual displacement in a controlled environment (see Annex). The method has been also applied to the fusion of a set of real InSAR measurements obtained from different sensors at different wavelength (COSMO-SkyMed, Sentinel-1, Radarsat-2), covering very long periods (more than 10 years) and very large areas, and considering also the integration of GNSS data (the results on real data will be presented at the workshop).
Authors: Vecchioli, Francesco; Malvarosa, Fabio; Minati, Federico; Trillo, Francesco; Costantini, MarioGiven a set of three interferometrically compatible synthetic aperture radar (SAR) images, three interferograms can be estimated by cross-correlating these images in a circular way. The phase of the product of these three interferograms (i.e. the wrapped sum of the three interferometric phases) is called the closure phase. A common and fundamental implicit assumption in SAR interferometry is that this closure phase is zero, at least in the expected value sense. This phase consistency assumption is linked to a geometric interpretation of the interferometric phase. At a single pixel level, the phase consistent assumption always holds, while after spatial averaging, the multilooked phases are never fully consistent. Current studies generally attribute such phase inconsistencies to random phase noise associated to the finite interferometric coherence. However, as shown in recent publications, in many cases, the interferometric triplets show a significant phase closure that is spatially and temporally correlated. Compared with the land cover map, the statistics of the phase closure and land cover types are closely related and indicate a geophysical origin of the closures. In this paper, we analyze the statistics of the phase closure for Sentinel-1 time-series over a large region in southern France and relate it to the land cover type. In order to separate phase inconsistencies of geophysical origin from those caused by noise, we compare the phase closure to the standard deviation of the phase closure obtained from a numerical model describing the null-hypothesis, in which phase closures are caused by phase estimation uncertainties. Therefore, the statistical significance of the phase closure is quantitatively assessed and a threshold is identified to isolate phase inconsistencies. Based on the land cover map, phase inconsistencies are classified for each land cover type. For a specific land cover type, statistic properties analysis of both phase inconsistencies and coherence are carried out. The temporal behavior of phase inconsistencies is illustrated, the percentage of classified phase inconsistencies to the overall phase closures is explored, and the magnitude of phase inconsistencies is presented. This research improves our understanding of non-geometric contributions to the interferometric SAR phases and paves the way for further studies aiming at relating the phase closure with geophysical variables.
Authors: Yuan, Yan; Hadler, Noah; Steele-Dunne, Susan; Lopez Dekker, PacoUsage of the open source SNAP toolbox continues to grow and see widespread adoption within academia, research and commercial organizations. A common architecture for all Sentinel Toolboxes is being jointly developed by Brockmann Consult, SkyWatch and C-S called the Sentinel Application Platform (SNAP). The SNAP architecture is ideal for Earth Observation processing by providing Extensibility, Portability, Modular Rich Client Platform, Generic EO Data Abstraction, Tiled Memory Management, and a Graph Processing Framework. Maintenance and development of the Sentinel-1 Toolbox (S1TBX) in SNAP is led by the teams at SkyWatch and Sensar. Over the past year, new development has been added to support new missions including SAOCOM, Iceye, Capella and Cosmo Second Generation. With the flurry of new commercial SAR providers launching constellations in the short term, expect SNAP to continually add support for new missions, abstracting their data and making all the existing tools for SAR processing available. S1TBX provides a wide variety of SAR tools including coregistration, calibration, filtering, classification, de-noising, slice product assembling, TOPSAR deburst and sub-swath merging as well as a suite of tools for interferometry and polarimetry. New functionality has been developed for: Spotlight Interferometry InSAR Network Definition Tool Interface with PyRate Automatic retrieval of the new Copernicus DEM Automatic search of download of Sentinel orbit files from ESA’s new GNSS hub Faraday Rotation Correction for L-band Multitemporal Composites of Radiometric Terrain Corrected Images Transform between Beta0, Sigma0 and Gamma0 New Polarimetric Decompositions Soil Moisture Toolkit New and existing users will appreciate the ongoing expansion of the library of tutorials explaining how to make the most of S1TBX capabilities. Recently, the toolbox has been increasingly used for backend processing in operational cloud-based environments. This trend is expected to grow as data providers do more of the analysis ready data processing in the cloud and less is done on the desktop. The team intends to support these activities with new cloud-native tools and ARD workflows.
Authors: Veci, Luis (1); House, Lisa (1); Lu, Jun (1); McVittie, Alex (1); Aguilera, Esteban (2); Hernandez, Carlos (2); Engdahl, Marcus (3); Fitrzyk, Magdalena (3)The SAR2CUBE project, launched in February 2020, is defined to satisfy simultaneously two objectives. The first one is to facilitate the use of SAR products in the scientific EO community and to promote them as relevant EO assets and the second is to improve the feasibility of hosting SAR products for a data provider. The Sentinel missions within the Copernicus program have defined a new playground with an extraordinary and unique amount of EO information. In particular, the radar pair defined by the twins Sentinel-1A and Sentinel-1B is offering a constant stream of SAR data since they were launched, late 2014 and early 2016 respectively. However, the interferometric capabilities provided by this source are underused. The particular nature of the complex interferometric data often presents a barrier to incorporate these data within the processing chains. The obvious nature of other kinds of sensors, such as optical or multi-spectral data, facilitates the incorporation of these products into different analysis frameworks. To reduce the entry-level barrier of the InSAR-derived products the SAR2CUBE project is designed to provide both SAR and InSAR analysis-ready data (ARD) specifically defined to achieve efficiency and flexibility in processing and analysing this valuable source of information. The first step on the scientific part of this project has been the definition of all the required information that has to be stored in the data cubes, including both the original SLC data from Sentinel-1 as well as auxiliary data that is used during the workflow to be able to compute an analysis ready data product, including for example a digital terrain model (DEM) or precise orbit information. All pre-processing steps can be employed without altering the nature of the original data with the developed SAR SLC datacube. Further layers for efficient geo-coding have been added as well. The second part of the project deals with the implementation of on-the-fly processors for ARD data products based on the developed data model. For different types of analysis different levels of filter might need to be applied, depending on robustness of dealing with noisy data. In some cases, a very strong speckle filter might be desired to provide a smooth image; in other cases, a certain level of noise might be tolerated in order to minimize the reduction of spatial detail. This requires the adaption of existing methods for a data cube ready implementation as opposed to working in the traditional file system, but also provides the opportunity for possible novel methods, fully utilizing the access to complete time series in the cube domain. See figure 1 for a general workflow of the idea. During Fringe we will provide an introduction to the project, the basic ideas behind it and a first proof of concept implementation based on the open data cube and the openEO API for accessing and processing of the data.
Authors: Jacob, Alexander (1); Claus, Michele (1); Centolanza, Giuseppe (2); Moral, Francisco (3); Vicente-Guijalba, Fernando (2); Mougnaud, Philippe (4)Analysis Ready Data (ARD) are satellite Earth Observation data sets that have had several necessary pre-processing steps already applied prior to being disseminated. Adopting these ‘higher level’ products enable users without prerequisite knowledge or dedicated processing resources to skip these time consuming steps and begin time series analysis. As such, providing ARD products effectively lowers the barrier to entry in to Earth Observation data analysis. The Committee on Earth Observation Satellites’ Land Surface Imaging Virtual Constellation (CEOS LSI-VC) has been leading the development of a definition and framework for ARD. The CEOS ARD for Land (CARD4L) framework provides the specifications for Earth Observation data to be analysis ready. Specifications for products across the optical, thermal, and radar domains and a process for product assessment against the specifications have been developed, to help data producers to generate CARD4L compliant products and for users to be able to adopt these products for a range of applications. CEOS LSI-VC is working with domain experts, EO data users and providers to implement the CARD4L framework. CARD4L products are intended to be flexible, accessible and suitable for a wide range of users and applications, including particularly time series analysis and multi-sensor application development. They are also intended to support rapid ingestion and exploitation through high-performance computing, cloud computing and other future data architectures. However they may not be suitable for all purposes, and are not intended as a ‘replacement’ for other types of satellite products. Since its inception in the 1990’s and 2000’s, differential InSAR has become a widely used remote sensing technique in the geosciences, reaching a level of maturity where most expert users implement the same procedures to turn a stack of SAR single look complex (SLC) products into a network of geocoded unwrapped interferograms. The InSAR processing methodology is therefore an ideal candidate for standardisation under CARD4L, to enable routine generation of InSAR ARD products for dissemination to the community, including non-experts. In fact, several groups around the world are already disseminating routinely generated Sentinel-1-derived InSAR products (one example is the COMET-LiCS portal). The usual procedures implemented in differential InSAR processing include co-registering the stack of SLCs, forming a differential interferogram, removing the orbital and topographic contributions of the phase signal, filtering, unwrapping, and finally, geocoding. The output product of this differential InSAR processing (a geocoded unwrapped interferogram) is then typically the input to further time series processing to derive the displacement history and average displacement rate (velocity) for certain pixels in the InSAR stack. Under the CARD4L initiative, Geoscience Australia is leading an effort to develop the specifications for a suite of InSAR ARD products: wrapped and unwrapped interferograms, and interferometric coherence products. In this contribution we will describe the CARD4L initiative and present to the assembled community of InSAR domain experts the current ARD specification. We intend that dissemination in this forum will assist in gathering feedback that will improve the ARD specification and lead to more widespread engagement with the InSAR community.
Authors: Garthwaite, Matthew C.; Wang, Lan-Wei; Thankappan, MedhavyCopernicus Programme’s Sentinel-1 SAR constellation images most of the land masses, with a revisit time of 6-24 days, in the Interferometric Wide (IW) swath Terrain Observation by Progressive Scanning (TOPS) mode. The S1 constellation has generated more than 10PB of Level-1 products since September 2014, and the size of this archive is expected to grow 3-4 fold over the next decade as more instruments are added to the constellation. Despite excellent global coverage and temporal sampling, continental and global scale processing campaigns with Sentinel-1 data have been hampered due to scalability of underlying SLC/GRD data access mechanisms (Appendix F of [1]). In this work, we present an efficient data access mechanism developed by Descartes Labs Inc. for Sentinel-1 SLC products and examples from the near real-time global backscatter and InSAR products that this capability has enabled. I. Efficient SLC data access We exploited the excellent burst synchronization properties of S1 TOPS acquisitions to build a global burst footprint map (See Figure 1) for all IW mode V-transmit data (~295K bursts), by analyzing all S1 annotation files from start of the mission till July 2020. We use this global burst map with every S1 SLC zip file released by ESA to label all burst footprints contained in it (new previously un-imaged footprints are added to the global map when first encountered), extract the SAR metadata for each burst and store it in json format for immediate access and identify the byte ranges of each individual burst in the zip files, using techniques developed in the neuroimaging and genomics communities [2], to enable random access to any burst from cloud storage buckets or public archives with minimal data transfer/ egress. Note that our approach requires us to be able to access the complete S1 SLC zip file at least once in order to derive the exact byte ranges for the bursts. Once indexed, we are able to pull in metadata and imagery for any individual TOPS burst in ~3s from cloud storage or ~7s (without authentication time) from Alaska Satellite Facility. We have also developed tools to export these individual tools to formats compatible with open source tools like SNAP, GMTSAR and ISCE for use in various processing pipelines. We have been keeping up with ESA’s live stream of SLCs since mid September 2020 and have indexed all the historical SLCs over numerous global AOIs with their metadata available for immediate use within large scale processing pipelines. II. Geocoded bursts Once we addressed the SLC data access bottleneck, the next challenge to tackle was the use of custom Range-Doppler projections specific to every individual SLC image in SAR/InSAR processing workflows. InSAR workflows often pick one acquisition per imaging geometry as a reference and form a coregistered stack on this reference image’s grid. This imaging geometry is typically incompatible with the set of standard map projections like lat/lon or UTM or polar stereographic that are used in the geospatial community and within GIS frameworks. To simplify our processing pipelines, we geocoded the SLC data with slant range phase corrections over an aligned UTM grid - 10 meter Northing and 2.5 meter Easting, similar to [3]. This then allows us to perform all interferometric and normalization operations in geographic space using simple raster operations. Note that our approach can easily incorporate a priori azimuth / slant range offset models during geocoding, if needed. Combined with our efficient data access mechanism, geocoding enables us to generate a coregistered stack for any burst footprint in the world in a few minutes. We have leveraged this newly developed capability to build two global data pipelines - one for SAR backscatter and one for InSAR coherence/ wrapped phase (see Figure 2). Both these pipelines have been keeping up with the live SLC data stream from ESA since late Jan 2021. We currently delete the geocoded bursts once we derive the necessary global products but can regenerate them very efficiently as needed by our higher resolution InSAR pipelines for infrastructure monitoring. III. Global SAR backscatter product Our Global SAR backscatter product is available at a posting of 10 meters on same UTM grid as Sentinel-2 and represents thermal noise corrected 𝜎0,E observations in dB. This product was derived using geocoded bursts, resulting in InSAR-grade coregistration of imagery acquired on different passes. We also provide three additional static global layers which we generated using ESA’s SNAP tool Correction factor to transform 𝜎0,E in dB to 𝛾0,T (within 0.1 dB over most areas) Layover Shadow mask Local Incidence angle We conducted numerous experiments with metadata from stacks of bursts to conclude that due to the small orbital tube characteristics for the Sentinel-1 mission, a static correction layer is more than sufficient for our applications where we desire Radiometrically Terrain Corrected backscatter data. Also, note that our global backscatter product is derived from the same source as our global InSAR product - geocoded SLC bursts, which results in better coregistration between these products. This may not be the case if these were generated using different approaches/ software -e.g., backscatter product with GRDs and the InSAR product with SLCs. We will show examples of products from our live pipeline within our Viewer environment and present examples of our static layer based Terrain Correction, during the talk. IV. Global InSAR product Our Global InSAR product is available at a posting of 20m on the same UTM grid as Sentinel-2 and is generated using a Gaussian Filter of wavelength 80m (resolution ~50m) for multilooking and coherence estimation. Our coregistration approach is purely geometric, like GMTSAR, for the global product and the processing setup ensures phase continuity at burst boundaries (see Figure 3). We currently generate all interferometric pairs with a temporal baseline of less than 25 days for live stream and have processed all pairs less than 13 days for historical data (before Nov 2020). In addition to coherence, we preserved the wrapped phase at 20m which users can combine with the backscatter product to further average down spatially using simple band math (see Figure 3). Our burst based approach also lets users combine a consistent set of bursts to form their own standardized frames for mosaicking before phase unwrapping. We will show examples of how co-seismic interferograms can be easily accessed and visualized from our global product in our Jupyter Lab-based workbench environment, during the talk. V. Conclusions Descartes Labs Inc has developed a very efficient random access mechanism for Sentinel-1 TOPS bursts, which enables it to generate global radar backscatter and InSAR products efficiently using geocoded bursts as an intermediate product. On a normal day of operations, these global layers are nominally updated well within 3-4 hours of SLCs being copied over from Copernicus programme’s scihub portal into our system. The developed data access mechanism that Descartes Labs Inc has developed eliminates SLC data access as the bottleneck in continental/ global scale processing / reprocessing campaigns. Pipeline throughput is now determined by the amount of compute resources and optimization of the analysis algorithms assigned to different pipelines rather than data access. For targeted infrastructure monitoring studies, we are able to generate coregistered, geocoded stacks of SLCs for any AOI in the world in a few minutes. References EGMS Specification and Implementation plan, Copernicus Land Monitoring Service, 2020. https://land.copernicus.eu/user-corner/technical-library/egms-specification-and-implementation-plan Z. Rajna, A. Keskinarkaus, V. Kiviniemi and T. Seppanen "Speeding up the file access of large compressed NIfTI neuroimaging data", Engineering in Medicine and Biology Society (EMBC), 2015 37th Annual International Conference of the IEEE, Milan, 2015, pp. 654-657. Zebker, H.A., 2017. User-friendly InSAR data products: Fast and simple time series processing. IEEE Geoscience and Remote Sensing Letters, 14(11), pp.2122-2126.
Authors: Calef, Matthew; Warren, Michael S.; Agram, Piyush; Arko, ScottInterferometric Synthetic Aperture Radar (InSAR) has proven to be a powerful technique for studying how the Earth's surface is deforming. ESA’s Sentinel-1 and the upcoming NISAR mission offer observations at high spatial-temporal sampling (~20m every ~6-12 days starting October 2014 and ~2021, respectively) over large ground swaths (~350x250 km)—an unprecedented volume of data. However, the complexity in specialized InSAR processing methods and its expensive prerequisite computing power/space, as well as the already immense, growing volume of SAR data makes this technique often cumbersome and not widely accessible. JPL’s Advanced Rapid Imaging and Analysis (ARIA) project has been automatically generating standard InSAR displacement products at a 90 m posting from the Sentinel-1 mission by leveraging the InSAR Scientific Computing Environment (ISCE) software and processing in a cloud environment. These ARIA standardized InSAR products are universally compatible with most GIS software and processing environments, allowing users to circumvent the use of specialized radar processing software altogether and making InSAR products more accessible and manageable for science applications. Since the 2018 AGU Fall Meeting, ARIA has shared these open-access products to the public through its products page (http://aria-products.jpl.nasa.gov/) and the NASA Alaska Satellite Facility Distributed Active Archive Center data search page (https://search.asf.alaska.edu/#/) as part of the Getting Ready for NISAR project. We will provide an overview of the current archive, which includes complete coverage for California, amongst other geographical regions. We will also provide updates on our ongoing development of the open-source ARIA-tools package (https://github.com/aria-tools/ARIA-tools) which allows community users to download, manage, and manipulate these products, and prepare extracted layers for time-series analysis. Our complementary suite of comprehensive, interactive Jupyter notebooks collectively outline the structure of the standard products and range of functionality provided through ARIA-tools (https://github.com/aria-tools/ARIA-tools-docs). We will demonstrate the use of these ARIA standardized InSAR products to generate science-grade surface displacement time-series across the entire state of California. We do this by leveraging ARIA-tools in combination with the Miami INsar Time-series software in PYthon (https://github.com/insarlab/MintPy) package. In total, we processed 9 Sentinel-1 tracks over California using the Small Baseline method to obtain complete coverage from both ascending and descending geometries. The frequent acquisitions and duration of the Sentinel-1 record allows us to maintain good coverage with accuracies on the order of a few mm/yr at 20-100 km distance from the reference point. We are able to capture various fault strands of the San Andreas fault system such as the main San Andreas, San Gregoria, Hayward, and Calaveras faults moving at rates from 7 to 2 mm/yr, as well as the large-scale anthropogenic surface displacements related to agricultural activity and pumping in the San Joaquin and Sacramento Delta basins. We validate our results by comparing these observations with the dense Global Navigation Satellite Systems (GNSS) network of stations available in California.
Authors: Bekaert, David (1); Sangha, Simran (1,2); Funning, Gareth (3); Team, Aria (1)The NASA-ISRO Synthetic Aperture Radar (NISAR) is a repeat-pass radar mission that will acquire fully-polarimetric SAR data in an innovative acquisition mode known as quasi-quad pol (QQP) [1]. It consists of simultaneously operating a HH/HV and a VH/VV dual-pol modes in the lowest and upper parts of the transmitted frequency spectrum (i.e., 20 or 40 MHz on the upper part and 5 or 20 MHz on the lower part). To mitigate the impact of ionospheric phase delay, operational repeat-pass NISAR algorithms will likely exploit a modified version of the range split-spectrum technique, wherein the range frequency spectrum of the radar signal is divided into two smaller sub-bands to form two SAR images at a lower resolution and with different center frequencies and at different polarizations (e.g., HH and VV or HV and VH) [2]. The use of sub-band interferograms at different polarizations makes the ionospheric phase estimation vulnerable to changes in the polarimetric scattering properties of the observed scene introducing spurious polarimetric-dependent terms in the interferometric phase [2]. To understand the impact of NISAR QQP acquisition mode on ionospheric phase estimation, we use airborne SAR data collected with the NASA-Jet Propulsion Laboratory (JPL)'s UAVSAR radar system and document observations of L-band interferometric phase discrepancies between HH and VV interferograms for various land cover types. We analyze and quantify the impact of these spurious polarimetric-dependent terms on QQP ionospheric phase estimates and provide insights on their implications for geodetic products derived from QQP data. We observed discrepancies between HH and VV SAR interferograms over bare soil and agricultural land cover types. For bare soil areas, we detect HH-VV interferometric phase discrepancies up to 30 degrees over 12 days, corresponding to a 10 cm bias at L-band. Over comparable time intervals, changes in vegetation vitality introduce HH-VV interferometric phase inconsistencies beyond 90 degrees for vertically-oriented agricultural fields (e.g., wheat and barley crops). Using the values of HH-VV interferometric phase discrepancies observed in the UAVSAR data, we simulate the ionospheric phase screen for QQP repeat-pass measurements over the San Andreas Valley, California, USA. This area offers an excellent test case to evaluate the accuracy of QQP ionospheric phase estimates since this region is characterized by a mixture of bare soil patches (e.g., desert) embedded in an agricultural landscape with a variety of cultivations. We observed that the bias introduced by HH-VV interferometric phase discrepancies is one order of magnitude bigger than the typical deformations observed over the San Andreas Valley and sufficient to significantly impair the scientific requirement set for NISAR interferometric geodetic products. Based on the results from the UAVSAR data analysis, we recommend using the same polarization on the main and side-bands of the NISAR operational science modes (e.g., single-pol or dual-pol) to avoid potential biases in the ionospheric phase estimates. References [1] Rosen, Paul A.,et al.. "The NASA-ISRO SAR mission-An international space partnership for science and societal benefit." In 2015 IEEE Radar Conference (RadarCon), pp. 1610-1613. IEEE, 2015. [2] Zwieback, Simon, and Irena Hajnsek. "Influence of vegetation growth on the polarimetric zero-baseline DInSAR phase diversity—Implications for deformation studies." IEEE Transactions on Geoscience and Remote Sensing 54, no. 5 (2016): 3070-3082.
Authors: Brancato, Virginia; Fattahi, HereshThe split-spectrum method (SSM) can largely isolate and correct for the ionospheric contribution in the L-band interferometric synthetic aperture radar (InSAR). The standard SSM is performed on the assumption of only the first-order ionospheric dispersive effect, which is proportional to the total electron content (TEC). It is also known that during extreme atmospheric events, either originated from the ionosphere or in the troposphere, other dispersive effects do exist and potentially provide new insights into the dynamics of the atmosphere. We apply L-band InSAR into heavy rain cases and examine the applicability and limitation of the standard SSM. Since no events such as earthquakes responsible for surface deformation took place, the non-dispersive component is apparently attributable to the large water vapor associated with heavy rain, whereas there are spotty anomalies in the dispersive component that are closely correlated with the heavy rain area. The ionosonde and Global Navigation Satellite System rate of total electron content index (ROTI) map both show little anomalies during the heavy rain, which suggests few ionospheric disturbances during the SAR imaging period. Therefore, we interpret that the spotty anomalies in the dispersive component of the standard SSM during heavy rain are originated in the troposphere. While we can consider two physical mechanisms, one is runaway electron avalanche and the other is the dispersive effect due to rain, from comparisons with the observations from the ground-based lightning detection network and rain gauge data, we conclude that the rain dispersive effect is spatiotemporally favorable. We further propose a formulation to examine if another dispersive phase than the first-order TEC effect is present and apply it to the heavy rain cases as well as two extreme ionospheric sporadic-E events. Our formulation successfully isolates the presence of another dispersive phase during heavy rain that is in positive correlation with the local rain rate. In comparison with other dispersive phases during Sporadic-E episodes, the dispersive heavy rain phases seem to have the same order of magnitude with the ionospheric higher-order effects.
Authors: Furuya, Masato; Setiawan, NaufalSAR images benefit from excellent geometric accuracy due to accurate time measurements in range and precise orbit determination in azimuth. Moreover, the interferometric phase of each single pixel can be exploited to achieve differential range measurements for the reconstruction of topography and the observation of earth surface deformation and surface motions. But these measurements are influenced by the spatial and temporal variability of the atmospheric conditions, by Earth dynamics, and by SAR processor approximations, which may lead to overall displacements shifts of up to several meters. These effects become visible in various SAR applications including the retrieval of ice velocity applying offset tracking and various InSAR applications, which might require several post-processing steps and external information for correction. In this paper we present the Enhanced Timing Annotation Dataset (ETAD) for Sentinel-1 recently developed in a joint effort by ESA and DLR based on research results and processor prototypes available at DLR [1-3]. ETAD is a novel and flexible product for correcting the SAR fast and slow time annotations in standard Sentinel-1 IWS TOPS products. It accounts for the most relevant effects including tropospheric delays based on 3D ECMWF operational analysis data, ionospheric delays based on TEC maps inferred from GNSS, solid Earth tides calculated following geodetic conventions, and corrections of IPF SAR processor approximations. The data are converted to range and azimuth time corrections and are provided in a 200 m resolution grid matching the burst outline of Sentinel-1 TOPS mode data. The processor to compute the ETAD product is now ready to be integrated into S-1 operational environment and to validate the products. The impact of the ETAD product is evaluated by two applications exploiting widely used techniques, namely SAR interferometry and feature tracking (FT) applied for monitoring velocity on ice sheets and ice caps [4]. In the first part of the evaluation experiment we tested the impact on the retrieval of ice velocity products using FT in Greenland ice sheet and Iceland glaciers and non-glaciated areas. The time correction was applied on burst-level to the Sentinel-1 image pairs before offset tracking processing, and alternatively as a post processing step to correct the offset tracking results. Both methods provided similar level of improvements and showing almost unbiased displacement measurements which was validated for stable terrain. Large scale biases in the results were removed in all cases leading to reduced velocity errors, especially for short 6-day baselines. This potentially reduces the uncertainty for ice velocity retrieval for Greenland and Antarctic ice sheet where stable control points are not always available. In the second part of the experiment we investigate the impact of ETAD products on interferometric applications such as landslides and InSAR velocity retrieval. In this experiment, a series of ETAD products was used to simulate atmospheric phase screens that are then subtracted from S-1 interferograms. As test area the steep fjords and valleys in Norway that are severely prone to landslides, were selected. In this area NORCE, PPO.labs and the Geological Survey of Norway (NGU) operate a national ground motion service which will be supported by the upcoming European Ground Motion Service (EGMS) in the near future. In these regions atmospheric effects are in general visible in interferograms and hamper the deformation measurements when the atmospheric conditions change between SAR acquisitions. These tropospheric phase errors may introduce several fringes indicating a false motion and many data sets are required to reduce this error by InSAR time-series methods. Initial analysis has demonstrated that about 95% of the 66 produced atmospheric interferograms show clearly the atmospheric effects observed in the SAR interferograms. After subtraction of ETAD-based phase screens from the SAR interferograms, the stratification signal is significantly reduced and some of the large-scale trends are removed. In the presentation we will explain the basis of the ETAD correction method, how to apply it for offset tracking and InSAR applications, and we will show results of both evaluation experiments. [1] Balss, U., Gisinger, C., Eineder, M.: Measurements on the Absolute 2-D and 3-D Localization Accuracy of TerraSAR-X. Remote Sensing, vol. 10, no. 4, pp. 1-21, 2018. Doi: 10.3390/rs10040656 [2] Gisinger, C., Suchandt, S., Breit, H., Balss, U., Lachaise, M., Fritz, T., Eineder, M., Miranda, N.: Towards Operational SAR Imaging Geodesy: An Extended Time Annotation Dataset for Sentinel-1 Image Products. Very High-resolution Radar & Optical Data Assessment workshop, 18.-22. Nov. 2019, Frascati, Italy. [3] Gisinger, C., Schubert, A., Breit, H., Garthwaite, M., Balss, U., Willberg, M., Small, D., Eineder, M., Miranda, N.: In-Depth Verification of Sentinel-1 and TerraSAR-X Geolocation Accuracy using the Australian Corner Reflector Array. IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 2, pp. 1154-1181, 2021. DOI: 10.1109/TGRS.2019.2961248 [4] Nagler, T., Rott, H., Hetzenecker, M., Wuite, J., Potin, P.: The Sentinel-1 Mission: New Opportunities for Ice Sheet Observations. Remote Sens., vol. 7, pp. 9371-9389, 2015. DOI: 10.3390/rs70709371
Authors: Gisinger, Christoph (1); Nagler, Thomas (2); Libert, Ludivine (2); Marinkovic, Petar (3); Larsen, Yngvar (4); Miranda, Nuno (5); Valentino, Antonio (5); Breit, Helko (1); Suchandt, Steffen (1); Balss, Ulrich (1); Krieger, Lukas (1); Fritz, Thomas (1); Eineder, Michael (1)The Synthetic Aperture Radar (SAR) and Interferometry SAR (InSAR) data are commonly provided in a regular grid, known as range-Doppler or radar grid, whose two axes are along and across the SAR platform's track and is naturally closest to the imaging geometry of the SAR acquisition. With the precise information of the SAR platform and a Digital Elevation Model (DEM), each slant-range, azimuth time coordinates can be mapped to the ground (forward mapping) or the ground coordinates can be mapped to the radar coordinates (inverse mapping). Inaccurate mapping between the radar and map coordinates leads to mis-alignment of SAR data and cause partial or complete interferometric decorrelation. Also, inaccurate mapping causes incorrect geocoding which affects interpretation of the multi-temporal geocoded SAR data (phase or backscatter amplitude) or causes decorrelation of the interferograms formed from geocoded SAR SLCs. The precise orbits of the modern SAR satellites (e.g., cm-level accuracy of Sentinel-1 orbits), the precise pointing and the current global DEMs (e.g., SRTM DEMs) ideally are sufficient to achieve sub-pixel accuracy for coregistration or for geocoding at a pixel spacing comparable to the SAR native resolution. However, the precise mapping between the radar and geographic map grids requires precise slant range information from the radar phase center to the targets on the ground. The accuracy of the slant range can be affected by the propagation delay of the signal through a planet's atmosphere. The propagation of the microwave signal through Earth’s atmosphere with components from ionosphere and troposphere may cause a group range delay, which can bias the geometrical mapping. We demonstrate the impact of the ionospheric range delay on the geolocation of SAR data. For this purpose, we estimate the time-series of differential range delay from InSAR observations and compare with expected range delay from different global and regional GNSS-based TEC maps including the available IGS Total Electron Content (TEC) maps, and the JPL’s higher resolution TEC maps. We demonstrate the impact of the above SAR orbit’s TEC (top-side TEC) on the range delay computation using GNSS data onboard the Copernicus Sentinel missions. Results from stacks of ALOS-1 and ALOS-2 data demonstrates that the observed range delay (up to 12 m) in InSAR data is most consistent with the ionospheric range delay derived from high-resolution GNSS-based TEC maps of JPL which reduces the residual delay to ~2 m. The results from Stacks of Sentinel-1 over Northern Chile shows that ionosphere dominates range delay and that GNSS-based TEC data can successfully predict the observed SAR range delay on the ascending tracks (~6PM local time). Using the GNSS receivers onboard the Sentinel-1 platfoirms, we demonstrate the impact of the topside TEC on the absolute geolocation of low-frequency SAR data assisted with GNSS TEC measurements. Using GNSS receivers onboard the Sentinel-1 we demonstrate the magnitude of the top-side TEC and its temporal and spatial variations. We show that after ionospheric range delay correction, the residual range delay can be mainly explained by the solid earth tide and tropospheric propagation delay. We discuss the implications of this correction on the operational production of the level-2 SAR and InSAR products from NASA-ISRO SAR (NISAR) mission.
Authors: Fattahi, Heresh (1); Yunjun, Zhang (2); Pi, Xiaoqing (1); Rosen, Paul (1); Agram, Piyush (2); Aoki, Yosuke (3)Persistent Scatterer Interferometry (PS-InSAR) enables – apart from height and motion determination - the estimation of differential slant delays due to varying water vapor in the atmosphere. Moreover, single observations of zenith delays (ZD) can be estimated from a stack of interferograms under certain assumptions and with respect to an arbitrary constant, which appears as a rank deficiency in the estimation procedure. In particular for modern SAR satellite imagery, zenith delay maps are characterized by a high spatial resolution that can be used as input to water vapor tomography. A common approach to overcome the rank deficiency of this estimation process is the so called ‘zero-mean-assumption’, which assumes for each PS-point a zero average delay over all acquisition times. However, this approach neglects local variations, for example height dependent variations due to stratospheric layering or specific regional characteristics. Even a conventional DEM-dependent stratification approach can only accomplish a rough approximation of the true situation. Additionally, the resulting delays are still relative but for the usage in weather models, absolute values need to be determined. In this work, we reconsider different constraints for the conversion from differential slant delays obtained from Sentinel-1 PS-InSAR stacks to single observation zenith wet delays (ZWD). We show that the introduction of additional information, e.g. from weather models such as ERA5, can help to mitigate those limitations. Those models provide atmospheric parameters on a coarse grid. With interpolation to the specific PS-points at the SAR acquisition times, true-mean values of the wet delays can be determined for each PS. Such mean values include regional variabilities and systematic height dependent differences. Long-wavelength phase gradients in PSI stacks can be caused by several influences, such as orbital ramps, processing factors, earth tides and tidal effects or atmospheric signals. The distinguishment of those contributions is only possible with further knowledge. As the ERA5 model describes the atmospheric delays at long-wavelengths reliably, it is a suitable dataset to prune the long-wavelength signal leaving only the atmospheric component. The applied corrections influence the absolute values in the regional context and in the long-wavelength range, but do not lower the advantage of the spatially high-resolution SAR-data and thus, allows the analysis of local variations with absolute ZWD. The proposed upgrading of the zero-mean assumption does not require any commercial data, as both the Sentinel-1 data and weather model data from ERA5 are freely available online. Our area of investigation is the Western Region of the Upper Rhine Graben, where the topography is varying from flat areas in the Rhine valley to flanking low mountain ranges with high potential for significant tropostatic and turbulent components of phase delays. We show the results of the upgrading to the ZWD estimation and discuss the plausibility and suitability of different estimation approaches. The validation of the resulting absolute ZWD is performed using GNSS data, processed with GAMIT software. That data includes local characteristics at the GNSS sites, which might be contained in the InSAR dataset, but not in the ERA5 model. The work is part of the AtmoWater project, targeting the improvement of water vapour estimations and modeling using PSI, GNSS, tomographic approaches and weather modeling strategies.
Authors: Kamm, Bettina; Schenk, Andreas; Yuan, Peng; Hinz, StefanEstimation of ground deformation via time series analysis of SAR interferometric phase is challenging in rural, vegetated areas. A change in the geometric and/or electromagnetic profile of a scatterer, between successive SAR images, will cause a shift in the corresponding measurement of interferometric phase. These Scattering Phase Changes (SPC) can hence cause misinterpretation or obfuscation of ground deformation measurements if unaccounted for. This SPC signal is typically treated as a zero-mean noise [1] to be minimised by e.g. spatial averaging of the interferograms [2, 3]. However, authors have highlighted common situations where a non-zero mean SPC may be expected. These may reflect changes in soil moisture content [4–6] or in vegetative canopy height or biomass [7, 8]. To avoid inaccurate ground deformation measurements, these phenomena require identification in the interferometric phase data and any non-zero SPCs require estimation to avoid being misinterpreted as ground movement [7, 8]. This work explores an approach aimed at identifying and estimating SPC directly from the SAR data set, using time-series behaviour of measurable phenomena, such as polarimetric amplitudes, coherence, and closure phases, to infer areas of consistent patterns in scattering behaviour. The work focuses on Sentinel-1 data from its Interferometric Wide (IW) swath mode, because of its high temporal resolution and widespread application in ground deformation studies. Data from areas with consistent scattering patterns exhibit consistent phase offsets relative to estimates of the atmospheric phase screen (APS). APS estimates were derived by Kriging interpolation from more stable, persistent scatterers [9, 10]. The SPCs are correlated over far larger distances than can be explained by errors in APS estimation. They are highly unlikely to represent ground deformation, as the sign and rates of apparent movement do not agree with any known geological phenomena. To validate that these consistent phase offsets are a product of scattering change phenomena, we used independent a priori land cover maps, provided by the UK Centre for Ecology and Hydrology [11], to extract crop types from InSAR data for the Vale of Glamorgan, South Wales, United Kingdom. This area was chosen because it is a dominantly rural, agricultural area, but with numerous towns and villages allowing for good APS estimation, the topography is moderately undulating, and the geology is relatively simple. Around 25,000 fields were extracted and classified according to common crop patterns between 2016-2019, years for which the crop maps were available. Within each class of fields, i.e. each set sharing a particular crop pattern, there is significant correlation in the measured phase offsets. Comparatively, there are significant differences in the mean phase offsets between different classes. This suggests that SPC varies based on crop patterns, and that to some extent it is predictable. We thus conclude that differences in scatterer type and changes in scatterer behaviour are the most likely causes of the observed phase offsets. This supports the case for targeted estimation of non-zero mean SPCs in InSAR deformation studies. Here we propose an SPC correction, and apply this to a Sentinel-1 2015-2020 InSAR dataset over South Wales. Scatterers are first classified according to their amplitude and coherence behaviour, and the average SPC behaviour is then estimated for each class. The SPC estimates are then used to correct the measured coherence matrices, prior to application of a standard approach [2] towards phase triangulation and model inversion to derive the ground measurements. The correction yields a significant improvement in the coherence of the resultant ground deformation measurements, yielding far greater point density as compared to the uncorrected data. Large regional uplift effects have been detected over areas of former coal mining, consistent with a previous InSAR study [12]. SPC-corrected results further demonstrate a significant improvement in the range of deforming surface features which are identifiable in the data, by highlighting ongoing deformation at several known debris slides, spoil heaps, settlement of made ground, and other geotechnical phenomena of interest. References: [1] F. Rocca, “Modeling Interferogram Stacks,” IEEE Trans. Geosci. Remote Sensing, vol. 45, no. 10, pp. 3289–3299, Oct. 2007, doi: 10.1109/TGRS.2007.902286. [2] A. Ferretti, A. Fumagalli, F. Novali, C. Prati, F. Rocca, and A. Rucci, “A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR,” IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, p. 12, 2014. [3] F. Cigna and A. Sowter, “The relationship between intermittent coherence and precision of ISBAS InSAR ground motion velocities: ERS-1/2 case studies in the UK,” Remote Sensing of Environment, vol. 202, pp. 177–198, Dec. 2017, doi: 10.1016/j.rse.2017.05.016. [4] F. D. Zan, A. Parizzi, P. Prats-Iraola, and P. Lopez-Dekker, “A SAR interferometric model for soil moisture,” p. 8. [5] S. Zwieback, S. Hensley, and I. Hajnsek, “Assessment of soil moisture effects on L-band radar interferometry,” Remote Sensing of Environment, vol. 164, pp. 77–89, Jul. 2015, doi: 10.1016/j.rse.2015.04.012. [6] B. Barrett, P. Whelan, and E. Dwyer, “Detecting changes in surface soil moisture content using differential SAR interferometry,” p. 23. [7] S. Zwieback and I. Hajnsek, “Influence of Vegetation Growth on the Polarimetric Zero-Baseline DInSAR Phase Diversity—Implications for Deformation Studies,” IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 5, pp. 3070–3082, May 2016, doi: 10.1109/TGRS.2015.2511118. [8] F. De Zan and G. Gomba, “Vegetation and soil moisture inversion from SAR closure phases: First experiments and results,” Remote Sensing of Environment, vol. 217, pp. 562–572, Nov. 2018, doi: 10.1016/j.rse.2018.08.034. [9] A. Ferretti, C. Prati, and F. Rocca, “Permanent Scatterers in SAR Interferometry,” IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, vol. 39, no. 1, p. 13, 2001. [10] F. Meyer, B. Kampes, R. Bamler, and J. Fischer, “Methods for Atmospheric Correction in INSAR Data,” p. 7. [11] Center for Ecology and Hydrology, “Land Cover plus: Crops (2019),” EDINA Environment Digimap Service, Nov. 22, 2019. [12] L. Bateson, F. Cigna, D. Boon, and A. Sowter, “The application of the intermittent SBAS (ISBAS) InSAR method to the South Wales Coalfield, UK,” International Journal of Applied Earth Observation and Geoinformation, vol. 34, pp. 249–257, Feb. 2015, doi: 10.1016/j.jag.2014.08.018.
Authors: Agar, Stewart Andrew (1); Lawrence, James (1); Mason, Philippa (1); Ghail, Richard (2)Interferometric Synthetic Aperture Radar (InSAR) is a powerful tool for monitoring ground deformation associated with earthquakes, volcanoes, landslides, and different anthropogenic activities. Until recently, the accuracy of the estimated deformation had been assumed to depends only on a number of well-known parameters including tropospheric and ionospheric delays, phase‐unwrapping errors and phase decorrelation. However, recently an additional source of phase noise has been identified, which is strongest in short interval multillooked interferograms and unlike other sources of the noise, leads to biased, non zero loop clousre phases. This is problematic for time-series analysis because short-interval interferograms may be the only ones that maintain coherence for some areas. It was observed in previous studies that changes in soil moisture and in the water content of vegetation as well as the vegetation growth may lead to such phase inconsistencies. The study aims at studying of the phase bias phenomenon in InSAR data. We first investigated the charactrisitics of the phase bias by constructing daisy chain sums of the interferograms covering an identical 1-year time period, but using different time period interferograms. It was observed that the short baseline interferograms are more susceptible to the phase bias. However, moving toward longer baseline daisy chains, the bias begins to disappear. We also investigated the effect of filtering and multi-looking in the resulting bias. We find that the filtering of the interferograms, in addition to multi-looking, can significantly increase the bias in the estimated time-series displacement. We, then, propose a novel empirical method for quantifying and correcting phase bias. The proposed method is based on the assumption that the bias in the interferograms are linearly related to the sum of the bias terms in the shorter interferograms spanning the same time. We tested the algorithm over a study area in west of Turkey, which is regularly monitored by Sentinel-1 A and B. The land cover is heterogeneous which allows to investigate the bias effect in different land covers ranging from more coherent urban areas to the agricultural and forest areas. We compared the results to one of the phase linking approaches, which employs the full coherency matrix. The experimental results show that the proposed method can significantly decrease the bias in the estimated velocities.
Authors: Maghsoudi, Yasser (1); Lazecky, Milan (1); Morishita, Yu (2); Hooper, Andy (1); Wright, Tim (1)Geometric and temporal decorrelation impact the quality of interferometric SAR (synthetic aperture radar) pairs and make interferometric information ambiguous. A number of approaches (e.g., single master baseline network, minimal number connection baseline network) to geometric and temporal baseline network construction have been proposed and demonstrated in order to obtain more meaningful and reliable InSAR time series products such as deformation time series per InSAR measurement point. Yet, a cross assessment and comparison among these popular approaches are not fully explored. Therefore, this study proposes a barycenter-like assessment method, aiming at estimating, assessing, and comparing InSAR products which are generated under various spatio-temporal baseline networks. This study tests six well-known baseline networks, those are single master (SM) baseline network, short temporal (ST) baseline network, minimum spanning tree (MST) baseline network, minimal number connection (MNC) baseline network, redundant baseline network, and traveling salesman problem (TSP) baseline network. These six networks are included in distinct processing strategies, thereby the detected InSAR measurement points can be temporal/constantly coherent point-wise scatterers, and/or distributed scatterers. To ensure the InSAR products with the six networks are comparable, we downsample every InSAR measurement point on a common resolution-cell-size level. We also apply data alignment by using a common spatio-temporal reference point. As such, all temporal dynamics of InSAR measurement points are projected on a uniform reference system. The standard quality control metrics, i.e., precision, accuracy, and computational load, are used to assess and compare the InSAR products. This research chose Beijing as a test site in the experiment. The used data were 21 TerraSAR-X images acquired from 22-04-2011 to 09-05-2013 in stripmap mode, covering the deformation funnel in the Beijing area of 231 km2. A stable coherent point near Tiananmen Square and existed in all the baseline network results was chosen as the reference point. The in-situ measurements of two leveling benchmarks were used in double-difference comparison to assess the accuracy and the posterior variance is adapted to evaluate the precision of InSAR products. Putting all these performances into the barycenter-like assessment and assigning a designed weight (the weight could be adjusted by end-user) to each criterion, the result shows the most suitable baseline network in Beijing case is the TSP baseline network. As the TSP baseline network is usually seldom employed, we plan to keep evaluating the performance of the barycenter-like assessment by applying it to several other test sites.
Authors: Zhang, Xu; Chang, LingSynthetic Aperture Radar Interferometry (InSAR) is a remote-sensing tool that is capable for precisely mapping the topography and the surface deformation with high resolution. Decorrelation is one of the main limitations for InSAR, which affects the data processing, resulted in unreliable products. Therefore, masking decorrelation areas is crucial for the procedure to retrieve information from interferograms. Two main methods are often used for generating decorrelation masks: manually drawing one or setting a coherence threshold. However, manually drawing masks is unfeasible when the decorrelation areas are fuzzy with tiny boundaries. On the other hand, setting a single coherence threshold is hard, if not impossible, to mask out all the decorrelations without causing loss of valid phase. This is because the estimation bias and wide coherence range of interferograms. Here, we propose an end-to-end deep learning semantic segmentation model, Selective Kernel Res-Attention UNet, for generating decorrelation masks with applications to the TanDEM-X interferograms. The encoder of the network is on the basic of backbone of ResNet18. We replaced 3 by 3 convolution kernel in the basic block of ResNet18 with Selective Kernel Block (SK Block), and added Atrous Spatial Pyramid Pooling (ASPP) Block to the last layer, which enables network to extract feature with an adaptive receptive field, thus improving the capability of network for learning tiny and complex decorrelation boundaries. In decoder, we employed UNet decoder structure combining SK block and three times down-sampling scheme to replenish the features of former layers. The structure of the network is shown in Fig. 1. We trained the network with a few TanDEM-X interferogram samples with carefully drawing masks. We carried out several experiments to determine the optimal training strategy, including the optimal sample size, batch size, loss function and down-sampling scheme, to maximize model segmentation performance. Afterwards, we compared performance of the proposed model with other classical segmentation models (FCN-8s, ResUNet50, and DeepLabV3+) on the same dataset. Our evaluation shows that the proposed model has the best segmentation performance in semantic segmentation evaluation metrics with MIoU and MPA, where MIoU and MPA outperform other segmentation models by 3.26 %, 3.6% respectively, and has the fastest inferring speed, 0.5271s on sample size of 1024 by 1024 pixels, which is at least 50% faster than the other classical segmentation models. We applied the proposed model to mask out TanDEM-X interferograms of Wuhan District, Muztagata Glacier in China, and Kīlauea crater in Hawaii. The results show that proposed model can generate a fine and accurate mask for decorrelation regions, and even for those fuzzy, tiny and complex boundaries, as shown in Fig. 2 and Fig. 3 for the Wuhan case. Comparing with coherence threshold method, the proposed model can clearly mask out all the decorrelation regions, rarely causing loss of valid phase or destroys the continuity of valid phase. Comparing with other classical semantic segmentation models, the proposed model exhibits better performance especially for those tiny and complex decorrelation boundaries with less computational time. Our proposed model shows promising results in generating masks for TanDEM-X interferograms, and can achieve masking quickly and effectively. We can then train the network with more complicated interferograms with temporal decorrelation areas, such as that from Sentinel-1 interferograms. We believe such network can greatly saves time and labor in the big-data era of InSAR.
Authors: Zhang, Qi (1); Wang, Teng (1); Jiang, Houjun (2); Shi, Xuguo (3)Despite the large volume of available satellite data in displacement measurement, data incompleteness is still a commonly encountered issue. This phenomenon is mainly due to surface changes of the observed targets and/or technical limitations of the displacement estimation methods (e.g. InSAR, offset tracking). By generating spatial and temporal discontinuity, data incompleteness can hinder the full quantification of the displacement and then the thorough understanding of the underlying physical process. Therefore, advanced gap filling methods seem of particular interest for handling spatio-temporal gaps in displacement measurement time series. We propose two approaches, namely Expectation-Maximization Empirical Orthogonal Functions (EM-EOF) and extended Expectation-Maximization Empirical Orthogonal Functions (extended EM-EOF) respectively, to impute missing data in displacement time series. Both approaches rely on an Expectation-Maximization type iterative resolution scheme and lean on the low rank structure of the covariance matrix of the displacement time series. Starting from an appropriate initialisation of missing data, in an iterative way, the EM-EOF method decomposes the temporal covariance of the displacement time series into EOF modes and reconstructs the displacement time series by keeping the leading EOF modes that represent most of the variability of the displacement behavior. The extended EM-EOF follows the same principle as the EM-EOF method, but works on the spatio-temporal covariance of the displacement time series. The number of leading EOF modes is dertermined by three criteria, including the root mean squre error of the reconstruction with respect to the cross validation data, the error redaction rate and a confidence index estimated from the eigenvalue spectrum of the covariance. The efficiency of both methods has been proven in both synthetic simulations and real data applications. The latter corresponds to Sentinel-1 interferograms time series over the Gorner and Miage glaciers in the French Alps, covering the period between November 2016 and March 2017 and Sentinel-2 offset tracking time series over the Fox glacier in the Southern Alps of New Zealand, covering the period of 2018. Both methods are capable of dealing with cases with complex displacement behavior (e.g. oscillations with different frequencies, exponential behavior, etc.), spatio-temporally correlated noise, low signal to noise ratio and large data gaps. The extended EM-EOF method outperforms the EM-EOF method in case of small time series size and large correlated data gaps.
Authors: Hippert-Ferrer, Alexandre (1); Yan, Yajing (2); Bolon, Philippe (2)While orbital SAR systems have demonstrated excellent capabilities for mapping of surface deformation and applications based on polarimetry, there are some important limitations to consider. With respect to the observation geometry, the polar orbits used show deficits in measuring deformation perpendicular to the direction of the line of sight. Furthermore, the region of interest may fall in an area of layover or shadow when situated in steep terrain. Consider also that the repeat observation interval determined by the orbit may be insufficient to measure fast deformation. Temporal decorrelation of vegetated surfaces at C-Band and experience with L-Band sensors such as ALOS-2 suggest that L-Band is preferred in this aspect. Based on these considerations, and the desire to have an instrument that is light and compact to permit deployment on the widest range of platforms; Gamma designed and built the Gamma L-Band Synthetic Aperture Radar (Gamma SAR L). This instrument is adaptable to different mobile platforms including aircraft, UAVs, automobiles, ships, and a linear rail scanner designed by Gamma. UAVs and ground-based platforms using the Gamma SAR L can operate in geometries where they can look up onto slopes, thereby avoiding layover and shadow that preclude observations from high altitude aircraft or satellites. While the satellite repeat orbit cycle constrains the minimum time interval between observations, observations by UAVs or ground-based platforms can be repeated as required. The system architecture was developed to be readily adaptable to other frequencies without significant increase of power or mass. The Gamma SAR L is a compact (41x33x17 cm) and light weigh (7.65 kg) L-Band FMCW radar with 4 independent receiver channels and 2 transmitter output ports. The Gamma SAR L supports fully polarimetric and multi-baseline interferometric radar acquisition configurations. When using a dual receivers, up to 6 hours of raw data can be recorded by the on-board 1 TB SSD memory. The transmitter output power of up to 8 Watts permitting high resolution imaging beyond 20 kilometers in slant range. The radar noise-equivalent sigma0 (-40 dB at 6 km slant range, 200 MHz bandwidth) permits accurate fully-polarimetric measurements. Direct Digital Synthesis of the transmitter chirp supports use of any chirp in the frequency range between 1200 to 1400 MHz. A custom designed quad-channel 14-bit ADC sampling at 100 MHz has programmable decimation factors between 1 and 20, with a maximum aggregate data rate of 80 Mbytes/s to the SSD. The instrument has low power consumption of less than 80 W when actively transmitting at full power. Essential to the operation of the Gamma SAR instrument on mobile platforms is availability of track information with high accuracy as provided by a GNSS/INU system mounted near the antennas and a GNSS base-station that records the local GNSS signal. The blended INU and differential GNSS radar track solution is used as input to Gamma’s GPU-based time-domain back-projection (TDBP) radar processor that functions with arbitrary radar tracks to produce terrain-geocoded SLC images. The practical application of time-domain back projection using accurate track solutions provided by the GNSS/INU make repeat-track differential interferometry very applicable as demonstrated by this system. The radar has been deployed using a 12-meter linear rail, on automobiles, and on a helicopter UAV. We show differential interferograms of deformation acquired at the Brienzauls landslide in Switzerland. Data were acquired on 4 days over a 2-week interval. Interferograms at 0.5-meter sample spacing demonstrate high correlation even with snow cover and clearly map the landslide extent and provide quantitative measurements of the velocity along the line of sight.
Authors: Werner, Charles (1); Frey, Othmar (1); Wegmüller, Urs (1); Kurz, Andreas (2)For the small gradient deformation and the large gradient deformation, SAR interferometry and SAR intensity-based offset tracking method play an important role in revealing historical displacement of the Earth’s surface. The small baseline subsets (SBAS) InSAR technology can retrieval deformation time series by mitigating the effects of the decorrelation of the short-temporal-space baseline interferograms. Meanwhile, pixels offset small baseline subsets (PO-SBAS) method can also provide deformation time series in both slant-range and azimuth directions based on the short-temporal-space baseline SAR images. With the advancement of the SAR satellites, the onboard Sentinel-1A/B, ALSO/PALSAR-2, TerraSAR-X and TanDEM-X, COSMO-SkyMed, Radarsat-2, and newly planned NASA-ISRO SAR (NISAR), and Germany Tandem-L SAR satellites will continue to provide unprecedented SAR data. However, it is still changeable to estimate the deformation time series in near real-time, which makes SAR/InSAR method fail to meet the demand of fast deformed objects monitoring including landslides and civil infrastructures. In recent years, the parallel computation strategy and the cloud-computing-based approach have been introduced to improve the computational efficiency, for the SBAS InSAR processing, namely P-SBAS. On the other hand, the sequential filtering method was proposed to reconstruct new time series interferograms by block compression, in which it can update deformation time series dynamically after phase unwrapping. Besides, the Kalman filter method was proposed to updating the deformation parameter by integrating the predictive models. Unlike the above-mentioned methods, we proposed a series of dynamic deformation parameters estimation strategy with sequential least squares to SBAS-SAR/InSAR observations, which can not only increase the computational efficiency significantly, but keep the same precision of the deformation time series as traditional SBAS method. Specifically, the new method is operated in two steps, firstly, the archived SAR data are processed with SBAS method to estimate the DEM error, deformation rate, time series, and their cofactor matrixes of each pixel. Then, sequential least squares method is conducted deformation parameter to update the deformation parameters (including deformation rate, and deformation time series), once new SAR image is acquired. To this end, the former estimated parameters and their cofactor matrixes rather than original unwrapped interferograms are involved in this step, which can update the deformation parameters as quickly as possible. Both phase based interferograms and SAR intensity offset measurements are considered to update the small gradient deformation and large gradient deformation parameters, respectively. Finally, both the simulated and the real SAR data have verified the performance of the proposed method, which indicates it can be regarded as an effective SAR/InSAR data processing tool in the coming era of SAR big data.
Authors: Wang, Baohang (1); Zhao, Chaoying (1); Zhang, Qin (1); Lu, Zhong (2)“Sentinel-1 Companion Satellite” (Sentinel-1CS) foresees to provide information for forest monitoring services through a pair of a passive C-Band receiver in constellation with Sentinel-1 to acquire bi-static single-pass across track image pairs. In order to demonstrate the feasibility of the proposed concept, bi-static tomographic C-Band observations and retrievals for agriculture applications need to be developed using ground-based measurements under controlled conditions. In this work we present the development status of a tomographic ground-based radar setup which simulates the measurement geometry of the Sentinel-1 and its companion. Though the ground-based synthetic aperture radar system is built around the S1 sensor characteristics, it provides the possibility to enhance SAR acquisition parameters and to adjust accurately the interferometric baseline until the right solution of vegetation mapping is enabled; the flexibility required to explore possible solutions justifies a ground-based experiment since any change in space-based configurations is either very expensive or impossible. The radar system will be deployed in the field, on the top of a metallic tower, achieving an incidence angle of approximately 40° and providing radar data on agricultural crops. Pairs of SAR images will be formed by using a transmitter with 800MHz-bandwidth pulse and a dual channel receiver. While mono-static configuration is built from the transmitter and first channel of the receiver, the bi-static data are provided by the second channel of the receiver, which achieves a variable baseline. The scanning frontend is also provided with a rotational stage which permits controlled angular displacements for speckle reduction. The radar transceiver is split in two, physically separated units: backend and frontend, respectively. The backend unit is part of the main radar cabinet, which is situated on an easily accessible (low) level of the tower. The cabinet at this level comprises most of the radar parts which are to be serviced and surveilled. Frontend unit is situated on the upper level of the tower and comprises the displacement rails of the scanning device, their associated controllers, antennas and minimal signal conditioning for both transmitter and receiver channels. Signal transmission between backend and frontend is provided by optical fiber transceivers, in order to mitigate the effect of the temperature variations and signal attenuation over long coaxial cables. As the system is intended to correlate the radar data with the bio-physical properties of the crops by building tomographic images over phenological cycles, it is provided with in-situ surveillance cameras, meteorological and soil sensors.
Authors: Moldovan, Adrian-Septimiu (1); Poncos, Valentin (1); Toma, Stefan-Adrian (1); Teleagă, Delia (1); Șerban, Florin (1); Moise, Cristian (3); Davidson, Malcolm (2)In timeseries InSAR algorithms, constructing all possible interferometric combinations is both time-consuming and computationally expensive. On the other hand, using a only a subset with minimum required number of interferograms (i.e., without redundancy) usually does not provide sufficient precision for final estimated phase timeseries and neither it is capable of detecting/identifying potential unwrapping errors. In practice, it is a common practice to use redundant subset of interferograms (e.g., in SBAS methodology) to improve the precision and to create the capability of detection of unwrapping errors. However, in the existing algorithms, there is no generic and standard approach for selecting interferometric combinations. Considering the similarities between the problem of interferogram selection and the design of geodetic networks, the main objective here is to present an optimal approach for interferogram subset selection in InSAR stacks using the concept of geodetic network design. To reach this objective, we first define the network mathematical model and the equivalent of standard geodetic criteria of precision, reliability, and cost for the problem of interferogram selection. Then, an objective function is defined (comprising the three criteria) and the problem is solved as an optimization/minimization over the objective function. The problem of equivalent single-master (ESM) phase estimation (known as phase inversion in some literature) for each pixel is defined as the mathematical framework of the design, following with the introduction of the three design criteria as: Precision: the closeness of the covariance matrix of the estimated phases with an criterion matrix is introduced as a precision criterion. Here, the lower Cramer-Rao Bound (CRB) of estimated phases (derived from coherence matrix of each pixel) is considered as the criterion matrix. Reliability: the concept of internal reliability and minimum detectable error for each observation in the timeseries is introduced as the reliability criterion. Considering the minimum possibility of unwrapping errors (i.e., one phase cycle), the MDB is fixed to the value of one phase cycle. Cost: the number of interferograms is defined as the cost criterion. That is, the lower the number of selected interferograms, the more optimal (cheaper) the network. These three criteria are then blended in a weighted multi-objective function. To solve the problem (i.e. minimization of the proposed objective function), we use the Genetic algorithm. The optimization is solved in two steps: first the network is designed to reach the defined precision criterion, and then, in the second step, the network is densified further to reach the reliability criterion. The reason to solve the problem in a step-wise manner is to guarantee that we reach absolutely to the both criteria and also to speed up the processing time. The effect of different parameters (e.g., coherence of scatterers, multilooking, baseline lengths, non-centrality parameter, satellite revisit time, calculation approach of covariance matrix of observations, and iterative algorithm of design) is tested on simulated data in seven different scenarios. The proposed method is then applied over real data of Envisat satellite of Tehran region. To assess the performance of the method regarding the precision criterion, the residual phases over a presumably stable area are used to compute an empirical covariance matrix of the derived timeseries. The comparison of the empirical covariance matrix with the criterion matrix of the design confirms that the algorithm reach the expected precision. Also, the reliability of the designed network and the ability of detection of unwrapping error is tested and assessed in the study area.
Authors: Moghbeli-Damaneh, Mosayeb; Samiei Esfahany, SamiSentinel-1A & -1B offer a six-day revisit cycle and unprecedented coverage of Europe, with freely available data. This capability of the Sentinel-1 mission addresses limitations of cost and data availability, and provides research and commercial opportunities e.g. a European ground motion map. A European Ground Motion Service (EU-GMS) is currently under development, sponsored by the European Environment Agency, to provide consistent and reliable information on ground motion over Europe and across national borders, with millimetre accuracy. The ground motion results will be derived from time series analyses of Sentinel-1 data, most likely using different Persistent Scatterer (PS) and Distributed Scatterer (DS) InSAR approaches. To make the outputs useful for operational applications, quality assessment of ground motion maps is a fundamental priority, and an important aspect of quality assessment is data consistency. The nationwide/international ground motion map will be likely processed by multiple suppliers and their products can differ in terms of different metrics, such as density and coverage of measurement points, estimated deformation rate, and time series. Therefore, there is a need to assess and ensure consistency of InSAR results. Our main goal in this research is to develop and test a fair and robust methodology capable of assessing the similarities and differences between results from different InSAR processing chains, and to recommend a validation strategy for any nationwide/international (e.g. UK/EU) ground motion map. We base our approach on the Terrafirma Validation Project (EU/ESA Global Monitoring for Environment and Security (GMES) programme) (Crosetto et al. 2008), but tackle its limitations as follows: 1) As end-users require geocoded InSAR data, we compare all the datasets in geographic rather than radar coordinates. 2) We avoid assuming that any reference InSAR processing is the “truth”. 3) We define several polygons with different land cover types and stability. 4) We do not limit the time series processing to PSI algorithms and are open to any other methodologies e.g. both PS and DS InSAR processing. 5) We work with Sentinel-1 imagery. Our approach can be split into pre-processing and inter-comparison stages. The pre-processing stage includes checking global consistency between the InSAR datasets, re-referencing in the time and space domains, making an identical grid and defining different polygons. Then, the deformation velocities and time series, density and coverage of measurement pixels are compared and analysed by extracting some meaningful statistics. We use InSAR results from the Clyde Gateway of the Glasgow City Region in UK to test our methods. This is an area of particular interest to the Natural Environment Research Council (NERC) as it is the British Geological Survey (BGS) geothermal energy research field test site of the UKGEOS project (Bateson and Novellino 2019). We have access to multiple Sentinel-1 InSAR data products for this area, including data from SatSense, processed using a modified RapidSAR algorithm (Spaans and Hooper 2016), from TRE-ALTAMIRA, processed using the SqueeSAR algorithm (Ferretti et al. 2011), and from GAMMA-IPTA, processed using PSI at BGS. We used these datasets as well as our own analysis of Sentinel-1 using the Stanford Method for Persistent Scatterers (StaMPS) algorithm (Hooper et al. 2007). The results show that all the InSAR datasets detect similar deformation signals in the deforming polygon with all velocities consistent at the 1.1 mm/yr level (1 sigma). However, the InSAR products are not completely identical. One of the most striking differences between different InSAR methods is density and coverage of selected pixels. In general, the results of our comparison show that those methods that take advantage of both PS and DS, and benefit from making all possible interferograms, are more successful at extracting the maximum information (density and/or coverage) from the SAR stack. However, due to the short baseline of the Sentinel-1 interferograms, some DS pixels can remain coherent in a single-master interferogram network and would be identified as PS pixels in some PS InSAR processing methods. In addition to considering both PS and DS, other factors such as the temporal sampling of signal, the configuration of the interferometric network, whether oversampling of the original images is applied, and the specific thresholds imposed on signal-to-noise ratio (SNR) for pixel selection, can all have a major impact on the density of measurements. There are also some systematic effects in difference maps between different InSAR products, which are mainly due to different approaches to dealing with long wavelength trends and atmospheric phase screens (APS). Different precise geocoded coordinates for the common selected pixels is another discrepancy between the InSAR datasets. Some qualitative indicators including spatial resolution, frequency of update and latency period are the source of inconsistencies between the InSAR providers. We discuss the reasons for these differences and make some recommendations for any future nationwide/international InSAR product based on our comparison results. Any future national or international ground motion service using Sentinel-1 InSAR will need to instigate a validation process to ensure data meet minimum standards and are consistent across borders. We propose some requirements for the validation activities. References: Bateson, L., & Novellino, A. (2019). Open Report: Glasgow Geothermal Energy Research Field Site - Ground motion survey report British Geological Survey, Available:http://nora.nerc.ac.uk/id/eprint/524555/1/OR18054.pdf Crosetto, M., Monserrat, O., & Agudo, M. (2008b). Validation of existing processing chains in Terrafirma stage 2: Process analysis Report-Part 2: IG inter-comparison. ESA GMES Service Element, Institut de Geomatica. Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F., & Rucci, A. (2011). A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR. IEEE Transactions on Geoscience and Remote Sensing, 49, 9,3460-3470, doi:10.1109/TGRS.2011.2124465. Hooper, A., Segall, P., & Zebker, H. (2007). Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcán Alcedo, Galápagos. Journal of Geophysical Research: Solid Earth, 112, B7,doi:10.1029/2006JB004763. Spaans, K., & Hooper, A. (2016). InSAR processing for volcano monitoring and other near-real time applications. Journal of Geophysical Research: Solid Earth, 121, 4,2947-2960,doi:https://doi.org/10.1002/2015JB012752.
Authors: Sadeghi, Zahra (1); Wright, Tim.J. (1); Hooper, Andrew.J. (1); Jordan, Colm (2); Novellino, Alessandro (2); Bateson, Luke (2); Biggs, Juliet (3)Bank gully erosion of the lakeshore causes a great problem of silting of the dams in a semi-arid climate. To model and monitor the sedimentary dynamics of gullies, it is common to use numerical terrain models, generated using passive remote sensing data. The cost of acquiring these data often remains high even with the democratization of UAVs.This study was carried out on the gullies of the banks of the Rambla de Algeciras lake to estimate their sedimentary balance. In 2018, the sedimentary balance of a gully was calculated using three UAV photography missions. Sentinel-1 images were used during the same period of aerial photography to establish a comparison between the UAV results and SAR interferometry (PS-InSAR). It should be noted that the slopes of the gullies affected by the layover phenomenon are very particular. In these areas the SAR radar signal phase is the sum of the phase of the three components of the gully (the head, the slope, and the gully bed). Before applying the InSAR technique on the Sentinel-1 images we first studied the stability of the coherence using the CCD technique before and after the start of erosion. The results showed that the coherence on the gully slopes remained high enough to exploit the phase and calculate the ground displacement on the layover areas. After this step, we modified the PS algorithm implemented in the StaMPS software. To consider the pixels of the layover areas as PS pixels we used Amplitude Dispersion Index (ADI) and Mean Amplitude (MA) to select the PS candidates. To estimate the accuracy of the PSI results we performed a comparison with the results obtained by the UAV.The results show that the SAR radar data can measure with high accuracy the ground displacement on the gullies. But these results are related to the configuration of the SAR images and the geomorphology of the gullies. For bank gullies connected with the lake or characterized by wide slopes, the PSI technique indicates negative values. On the other hand, the PSI technique estimates positive values for gullies that are characterized by accumulation areas wider than the gully slope. This shows that SAR radar interferometry can measure the sediment balance in gullies on lakeshore with a high degree of accuracy.
Authors: Radouane, Hout; Véronique, MalevalMulti-temporal interferometric synthetic aperture radar (MT-InSAR) techniques [1]-[4] are nowadays well recognized as useful tools for the detection and monitoring of the Earth’s surface temporal changes. In this work, the fundamentals of error noise propagation and perturbation theories are applied to address the ground displacement products’ theoretical error bounds of the small baseline (SB) differential interferometric synthetic aperture radar algorithms. The different contributions of the unwrapped phases of the identified set of SB InSAR data pairs are thoroughly investigated in order to elucidate the different sources of decorrelation in a set of SB differential SAR interferograms. Specifically, the role of thetime-correlated and time-uncorrelated disturbance phase terms in the set of SB interferograms is discussed, to study how these terms could affect the quality of the reconstructed ground deformation time-series via the multi-temporal SB InSAR techniques. As a matter of fact, the time inconsistencies in the set of unwrapped SB interferograms might lead to erroneous InSAR products. For instance, in the case of the Small Baseline Subset (SBAS) [5] algorithm, the temporal coherence is typically used as a quality index of the retrieved ground deformation time-series [6]. In particular, the temporal coherence values could be used to eventually identify a group of reliable and well-processed SAR pixels after the SBAS inversion. Invoking the principles of error noise propagation [7] and perturbation theory [8], the ground deformation InSAR products’ relative error is eventually retrieved. It has also mathematically and statistically demonstrated that this error depends on the properties of the SB unwrapped phases and the design matrices of the identified networks of SB interferograms. Experiments were carried out considering sets of ERS, ENVISAT, Cosmo-SkyMed SAR images on different areas of the world. The main outcome of this investigation is a mathematical framework that can be used to quantify the errors of Mt-InSAR SB measurements in different operational context. Further analyses are still required to characterize the covariance matrix of the phase unwrapping and phase triplet mistakes in a set of SB interferograms. [1] Ferretti, A.; Prati, C.; Rocca, F. Nonlinear Subsidence Rate Estimation Using Permanent Scatterers in Differential SAR Interferometry. IEEE Trans. Geosci. Remote Sens. 2000, 38, 2202–2212. [2] Ferretti, A.; Fumagalli, A.; Novali, F.; Prati, C.; Rocca, F.; Rucci, A. A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR. IEEE Trans. Geosci. Remote Sens. 2011, 49, 3460–3470. [3] Fornaro, G.; Verde, S.; Reale, D.; Pauciullo, A. CAESAR: An Approach Based on Covariance Matrix Decomposition to Improve Multibaseline–Multitemporal Interferometric SAR Processing. IEEE Trans. Geosci. Remote Sens. 2015, 53, 2050–2065. [4] Hetland, E.A.; Musé, P.; Simons, M.; Lin, Y.N.; Agram, P.S.; DiCaprio, C.J. Multiscale InSAR Time Series (MInTS) Analysis of Surface Deformation. J. Geophys. Res. Solid Earth 2012, 117. [5] Berardino, P.; Fornaro, G.; Lanari, R.; Sansosti, E. A New Algorithm for Surface Deformation Monitoring Based on Small Baseline Differential SAR Interferograms. IEEE Trans. Geosci. Remote Sens. 2002, 40, 2375–2383. [6] Pepe, A.; Lanari, R. On the Extension of the Minimum Cost Flow Algorithm for Phase Unwrapping of Multitemporal Differential SAR Interferograms. IEEE Trans. Geosci. Remote Sens. 2006, 44, 2374–2383. [7] Papoulis, A.; Pillai, S.U. Probability, Random Variables and Stochastic Processes with Errata Sheet, 4th ed.; McGraw-Hill Education: Boston, MA, USA, 2015; ISBN 978-0-07-122661-5. [8] Chandrasekaran, S.; Ipsen, I.C.F. On the Sensitivity of Solution Components in Linear Systems of Equations. Siam J. Matrix Anal. Appl. 1995, 16, 93–112
Authors: Pepe, AntonioSatellite‐radar imagery is used increasingly to map Earth‐surface displacements, providing unprecedented insights into geohazards and crustal changes. The expansion of globally consistent satellite‐radar imagery presents new opportunities to measure Earth‐surface displacements on intercontinental scales. However, although the coverage of radar imagery is now global, applications to global‐scale processes are not underway. A fundamental obstacle to this is the need to transform satellite‐based displacement maps from measuring changes relative to an arbitrary point to being constrained within a globally consistent reference frame. In a new approach, the international network of Very Long Baseline Interferometry radio telescopes is shown to be a unique, unexplored, yet readily available link to unify satellite‐radar measurements on a global scale. Using the global telescope network requires no installations of additional geodetic infrastructure or ongoing ground surveys. Here we present proof‐of‐concept experiments using telescopes on two continents to demonstrate that these instruments simultaneously provide high‐intensity reflections in satellite‐radar imagery (similar to trihedral corner reflectors), while simultaneously acting as direct ties to a global reference frame. Automated tracking of radar satellites requires only minor additions to existing telescope operations alongside ongoing schedules for geodesy and astrometry. The approach is therefore immediately ready to implement globally, impacting upon the rapidly growing numbers and diversity of scientists using satellite radar to address geohazards on ever‐increasing scales. Such advances in approaches to the integration of relative satellite‐radar measurements with absolute geodetic methods comprise a first step toward using InSAR on a truly global scale. Globally unified InSAR measurements have unique potential to deliver a complete and consistent assessment of the land component of relative sea‐level change. This phenomenon is globally ubiquitous impacting upon growing coastal populations and records of sea‐level change, but remains little studied. Similarly, the spatial resolution, coverage, and accuracy of InSAR measurements are now sufficient to assess the fundamental mechanics of the way in which continents deform. Utilizing a unified set of global measurements expands the field of view from discrete studies of faults or plate boundaries, to the scale of entire plates, constraining the distribution and rate of strain accumulation in Earth's crust at a level of detail simply unachievable by ground‐based methods. Strain rates (predominantly estimated from cGNSS) are a key input to global forecasts of earthquake hazard, and advances to these estimates from globally unified InSAR will impact most considerably in regions where cGNSS coverage is limited.
Authors: Parker, Amy Laura (1); McCallum, Lucia (2); Featherstone, Will E (1); McCallum, Jamie N (2); Hass, Rüdiger (3)Quantum Annealer (QA) is well-suited for a certain class of optimization problems which can be expressed as a QuadraticUnconstrained Binary Optimization (QUBO) problem. A QUBO problem belongs to the family of Integer Programmingproblems which are called the NP-hard optimization problems. Feasible solutions of such problems can be found by usingclassical optimization techniques. However, studies claim that QA can find a feasible global solution that is faster than aclassical annealer for QUBO problems. Hence, it appears promising to program and use the QA-to-QUBO approachfor Earth Observation. In search of the QUBO problem in the domain of Earth Observation, we examined severalInteferometric Synthetic Aperture Radar (InSAR) applications and identified a residue connection problem in the phaseunwrapping procedure. In particular, we consider the residue connection problem with multiples of 2pi as a QUBOproblem. For this practical problem, we studied how to formulate this QUBO problem, and we examined the challengesto program the D-Wave quantum annealer, in particular, embedding the QUBO problem into our QA architecture with aso-called Pegasus topology, and the annealing parameter settings in the D-Wave quantum annealer. We then analysed theparameter effects on finding the global minimum of the residue connection problem. From these results, we derived andenhanced our insight for programming future quantum annealers; for instance, choosing real-world problems in EarthObservation, conceiving the embedding procedure, and the tuning of the annealing parameters.
Authors: Otgonbaatar, Soronzonbold; Datcu, MihaiThe advanced Differential Synthetic Aperture Radar (SAR) Interferometry (DInSAR) [Rosen et al. 2000] is a multi-temporal technique, that allows to estimate deformation time-series with centimeter to millimeter accuracy from a stack of SAR interferograms relevant to an area of interest on the Earth surface [Xue et al., 2020]. Although the multi-temporal DInSAR approaches have different technical and implementation details, however they all have to deal with the phase unwrapping (PhU) problem to retrieve accurate deformation time series. Hence PhU remains one of the main problems in multi-temporal DInSAR and is referred to as an ill-posed problem. In particular, PhU errors can threaten the correctness of retrieved deformation time series relevant to significant non-linear displacement signals related to seismic events, volcanic unrest and anthropogenic hazards. Considering the impact of PhU errors on the retrieval of deformation time series, several methods have been developed for the mitigation of PhU errors in multi-temporal DInSAR techniques. We proposed a new method, that exploits the compressive sensing theory [Donoho, 2006] to estimate and mitigate the PhU errors in the interferograms within multi-temporal DInSAR approaches. Compressive Sensing (CS) theory, relatively new in signal processing, offers an effective way to address inverse problems characterized by sparse solutions, as in the case of PhU error estimation. The technique, under appropriate conditions, allows robust and stable reconstruction of signal from an under-sampled set of noisy measurements [Donoho, 2006; Rani et al., 2018]. Very generally speaking, CS is based on the sparsity concept to solve L0-normminimization problems by using L1-norm minimization method. Indeed, although L0 minimization problems occur in several scientific fields, their solution is hard to find and is defined as NP-hard problem. CS theory demonstrates that the cases in which the solution of the L0-norm problem is a sparse vector it can be converted in a L1-norm minimization problem, that can be effectively addressed by using solvers from linear programming. The CS principles have been successfully applied in several contexts related to the analysis and processing of SAR data [Tello Alonso et al., 2010; Zhu and Bamler, 2010]. Moreover, some works on advanced DInSAR methods also demonstrate some advantages in using L1-norm minimization to reconstruct deformation time-series [Lauknes and Zebker, 2011; Xu and Sandwell, 2020]. However, although there are some studies that exploit L1-norm minimization for multi-temporal DInSAR applications, CS theory in multi-temporal DInSAR applications is not broadly explored yet. To address the problem of retrieving PhU errors in multi-temporal DInSAR approaches, we benefit from a set of interferometric pairs selected through the Delaunay triangulation method applied in the spatial/temporal baseline plane. Subsequently, by imposing the phase closure condition on the generated interferometric network, we retrieve an underdetermined linear system that can be effectively solved by applying the CS theory. Indeed, the number of PhU errors is usually very small if compared to the total number of interferometric pairs, hence the PhU error vector can be considered as sparse vector, i.e., a vector in which number of non-zero elements is significantly smaller than number of its elements. To retrieve the PhU errors, we adopt an L1 estimator based on the Iterative Reweighted Least Squares (IRLS) method [Burrus, 2012], properly modified to increase its performances on real DInSAR cases by taking into account the baseline characteristics of the interferometric pairs. The developed technique can be effectively applied to correct interferograms unwrapped with any PhU method; the only requirement is the selection of interferometric pairs that must be based on a triangulated network. Moreover, as the proposed procedure is based on pixel-by-pixel analysis, i.e., all pixels are analyzed independently, all the pixels can be computed simultaneously. Therefore, the algorithm is inherently suitable to be implemented through parallel programming techniques to reduce the computational time. To demonstrate the effectiveness of the proposed algorithm, among several multi-temporal DInSAR approaches, we tailored our method to the Small BAseline Subset (SBAS) processing chain, a well-known multi-temporal DInSAR algorithm able to reconstruct deformation time series with centimeter to millimeter accuracy [Berardino et al., 2002; Casu et al., 2014]. The SBAS algorithm has initially been proposed to retrieve low to medium spatial resolution DInSAR products (i.e., deformation time-series and velocity maps) relevant to extended areas by processing multi-look interferograms. Subsequently, the SBAS approach has been extended to work on Full Resolution (FR) SAR interferograms to retrieve high resolution deformation time-series [Lanari et al., 2004]. The FR SBAS algorithm has also been enhanced to jointly process SAR data acquired by different systems characterized by slightly different carrier frequencies, as in the case of ERS-1/2 and ASAR-ENVISAT systems [Bonano et al., 2012]. The full resolution SBAS algorithm allows to retrieve very localized deformation phenomena that characterizes natural and anthropogenic hazards, such as landslides, faults, urban areas and single man-made infrastructures. It has been applied in several application scenarios and its accuracy has been assessed by benefitting from independent ground based measurements [Arangio et al., 2013; Manunta et al., 2008]. Benefitting from the proposed CS-based PhU error estimator, we enhanced the FR SBAS processing chain to estimate and mitigate the PhU errors, mainly associated with strong non-linear deformation patterns. Particularly, we applied the proposed CS-based approach at regional and local scale analyses of the FR SBAS processing chain. In regional scale, we estimate and then mitigate the multiple errors affecting the unwrapped multi-temporal interferograms, whereas in the local scale analysis we unwrap the non-linear phase component [Lanari et al., 2004]. To assess the performances of the proposed approach, we carry out a deep experimental analysis based on both simulated and real data. In particular, we use TOPS Sentinel-1 dataset, consisting 171 images, acquired over Central Italy between March 2015 and December 2018. With this dataset, we perform an experimental analysis to assess the performance of CS based regional scale SBAS approach. To demonstrate the capability of proposed technique, we perform regional scale analysis of the area affected by three main seismic events occurred between 2016 and 2017. Fig. 1 shows the geocoded mean deformation velocity map (retrieved by both CS based (left) and conventional (right) SBAS approach) superimposed on optical image of the area. By visual inspection of Figure 1, it is quite evident the significant increase of the number of coherent points detected the the CS-based method. The second analysis is based on a stripmap COSMO-SkyMed (CSK) dataset acquired from descending orbits over the urban area of Rome (Italy) from 2011 to 2019. The dataset is composed of 129 scenes and is analyzed with the local scale SBAS approach to retrieve deformations affecting the city of Rome at full spatial resolution scale. To assess the performance of proposed approach in local scale analysis, we process the historical city center of Rome; a particular is shown in Figure 2 where we show the Ostiense Station, where deformation phenomena related to soil compaction are still on going. It is quite evident that the CS based approach (left) is able to detect the spatial details about the ongoing phenomenon, which are not clear in the conventional FR SBAS local scale analysis (right). The achieved results clearly demonstrate the effectiveness and potentialities of the proposed approach in enhancing the multi-temporal DInSAR approaches such as the FR SBAS processing chain. References P. A. Rosen et al., “Synthetic aperture radar interferometry,” Proc. IEEE, vol. 88, no. 3, pp. 333–382, 2000. F. Xue, X. Lv, F. Dou and Y. Yun, "A Review of Time-Series Interferometric SAR Techniques: A Tutorial for Surface Deformation Analysis," in IEEE Geoscience and Remote Sensing Magazine, vol. 8, no. 1, pp. 22-42, March 2020, doi: 10.1109/MGRS.2019.2956165. D. L. Donoho, “Compressed Sensing,” IEEE Trans. Inf. Theory, vol. 52, no. 4, pp. 1289–1306, 2006. M. Rani, S. B. Dhok, and R. B. Deshmukh, “A Systematic Review of Compressive Sensing : Concepts , Implementations and Applications,” IEEE Access, vol. 6, pp. 4875–4894, 2018. M. Tello Alonso, P. López-Dekker, J. J. Mallorquí, “A novel strategy for radar imaging based on compressive sensing,” IEEE Trans. Geosci. Remote Sens., vol. 48, no. 12, pp. 4285-4295, 2010. X. X. Zhu and R. Bamler, “Tomographic SAR Inversion by L 1 -Norm Sensing Approach,” vol. 48, no. 10, pp. 3839–3846, 2010. T. Lauknes and H. Zebker, “InSAR Deformation Time Series Using an L1 Norm Small-baseline Approach,” Geosci. Remote, vol. 49, no. c, pp. 1–11, 2011. X. Xu and D. T. Sandwell, “InSAR : Correcting for Earth Tides and Phase,” IEEE Trans. Geosci. Remote Sens., vol. 58, no. 1, pp. 726–733, 2020. C. S. Burrus, “Iterative Reweighted Least Squares,” OpenStax CNX, 17 dic 2012. [Available online: http://cnx.org/contents/92b90377-2b34-49e4-b26f-7fe572db78a1@12] P. Berardino, G. Fornaro, R. Lanari, and E. Sansosti, “A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms,” IEEE Trans. Geosci. Remote Sens., vol. 40, no. 11, pp. 2375–2383, 2002. F. Casu, S. Elefante, P. Imperatore, I. Zinno, M. Manunta, C. De Luca, and R. Lanari, “SBAS-DInSAR parallel processing for deformation time-series computation,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 7, no. 8, 2014. R. Lanari, O. Mora, M. Manunta, J. J. Mallorquí, P. Berardino, and E. Sansosti, “A Small-Baseline Approach for Investigating Deformations on Full-Resolution Differential SAR Interferograms,” IEEE Trans. Geosci. Remote Sens., vol. 42, no. 7, pp. 1377–1386, 2004. M. Bonano, M. Manunta, M. Marsella, and R. Lanari, “Long-term ERS/ENVISAT deformation time-series generation at full spatial resolution via the extended SBAS technique,” Int. J. Remote Sens., vol. 33, no. 15, pp. 4756–4783, 2012. S. Arangio, F. Calò, M. Di Mauro, M. Bonano, M. Marsella, and M. Manunta, “An application of the SBAS-DInSAR technique for the assessment of structural damage in the city of Rome,” Struct. Infrastruct. Eng. Maintenance, Manag. Life-Cycle Des. Perform., no. October, pp. 1–15, Oct. 2013. M. Manunta, M. Marsella, G. Zeni, M. Sciotti, S. Atzori, and R. Lanari, “Two‐scale surface deformation analysis using the SBAS‐DInSAR technique: a case study of the city of Rome, Italy,” Int. J. Remote Sens., vol. 29, no. 6, pp. 1665–1684, Mar. 2008.
Authors: Muhammad, Yasir (1,2); Manunta, Michele (1)In this work, we take the opportunity of a large dataset of Sentinel-1 acquisitions processed within the Centre for Observation and Modelling of Earthquakes, Volcanoes and Tectonics (COMET) LiCSAR system to investigate (spatio-)temporal variations of spectral diversity (SD) values that were calculated per each LiCSAR frame epoch over the Alpine-Himalayan Belt. A frame SD value results as an average of SD values calculated in overlapping areas within TOPS bursts in the frame. A standard LiCSAR frame contains 39 bursts in total and covers an area of approx. 220x250 km. The SD values are provided as a (sub) pixel shift daz in the azimuth direction. The shift is due to several factors that can be partially modelled. We take into consideration: - daz due to physical ground motion in azimuth direction (approx. N-S motion) daz_defo that is considered to consist of plate tectonic motion daz_pdefo and the elastic deformation due to solid Earth tides, daz_tdefo, - azimuthal phase ramp induced by tropospheric delay daz_tropo, - azimuthal phase ramp induced by ionospheric phase advance (daz_iono) of the SAR carrier frequency daz_riono and induced by the dispersive effect of ionosphere on different frequencies of used spectrum at overlap edges of Sentinel-1 TOPS bursts, daz_diono, - azimuthal phase ramp daz_error induced by inaccuracies in the input data (e.g. orbital inaccuracies) and processing (e.g. focusing): daz = daz_defo + daz_tropo - daz_iono + daz_error In our analysis, we convert daz sub-pixel values into metres on the ground and use basic models to estimate separate factors: - daz_pdefo - estimated by ITRF2014 model of plate motion, - daz_tdefo - estimated by model of solid Earth tides, as implemented by ‘solid’/GMT earthtide, - daz_iono - estimated by IRI2016 model of TEC (currently used only for daz_riono part), - daz_tropo - can be estimated by zenith total delay from GACOS (not applied to the whole dataset). Though we expect inaccuracies within applied models (that can be estimated by variance expected per each daz factor), this analysis clarifies the overview on sources of the systematic signals observed in the presented time series of daz values. As daz values are related to the global reference frame used by Sentinel-1 orbit ephemerides, the physical large-scale N-S motion captured by the analysis can be considered ‘absolute’. Finally, the results of the analysis draws attention to using daz values back to improve some of the applied models.
Authors: Lazecky, Milan; Hooper, AndrewToday AI4EO is mainly focused on data analysis. On this line, we aim towards enlarging the AI paradigm with Computational Sensing methods for bistatic configurations, integrating the physical sensor and mission models in the information extraction processing chain. In this presentation we highlight the latest advances of interferometric processing for a bistatic synthetic aperture radar (SAR) system for space-ground sensing. The methodology is developed around a ground-based stationary bistatic SAR receiver that uses a SAR satellite as transmitter of opportunity (e.g., Sentinel-1A/B, TerraSAR-X). The ground receiver has at least two channels: one channels that receives directly the transmitted pulses through an antenna pointing towards the satellite (reference channel), whereas the others receive the reflected signals through an antenna pointing towards the imaged scene. In such a bistatic configuration, the same area on the ground can be illuminated from several orbits, and if the satellite operates in a multi-swath mode, for each orbit from several sub-swaths (the ground receiver can capture pulses even if the satellite’s main lobe is not pointed towards it). The available multi-angle/multi-orbit data can be used in various ways to identify, characterize, classify and track targets by exploiting the enhanced observation diversity in comparison to solely using spaceborne monostatic data. The main results that will be highlighted at the workshop were developed in the Research Centre for Spatial Information (CEOSpaceTech) within two ESA projects (COBIS and TomoSAR-1B) and span various hardware/software aspects: ground receiver architecture design, synchronization between the transmitter of opportunity and the stationary receiver, bistatic SAR focusing with classical and compressed sensing-based algorithms, multi-aperture focusing used to increase azimuth resolution, electronic target (transponder) development for testing and/or calibration, and specific InSAR algorithms: repeat-pass/single-pass bistatic SAR interferometry, bistatic permanent scatterers (PS) analysis and single-pass bistatic SAR tomography. The currently available ground receiver consists in a modular platform based on a PXI Express chassis, and ongoing work is focused on the development of a miniaturized multi-channel portable remotely controlled receive platform based on an ADRV9009 transceiver chip. With such a ground platform, the developed InSAR algorithms will be more easily evaluated on several scenes with variable content imaged from different acquisition geometries and new directions could be addressed (e.g., extensive validation of bistatic PS processing and bistatic PolInSAR). Additionally, a portable bistatic SAR receiver could be used as test bed in experimental campaigns for the study of Sentinel-1 companion missions (e.g., HARMONY) and for the generation of test data for future mission-agnostic SAR processors.
Authors: Anghel, Andrei (1); Cacoveanu, Remus (1); Ciuca, Madalina (1,2); Rosu, Filip (1); Focsa, Adrian (1,3); Datcu, Mihai (1,4)The basic principle of exploiting interferometric synthetic aperture radar (InSAR) to reconstruct digital elevation model (DEM) is to obtain image pairs in the imaging area with certain incidence angle difference which is caused by two pairs of antennas (or repeated observation of one antenna) of the synthetic aperture radar (SAR) system, and extract DEM according to its interferometric phase. InSAR has the capability of topographic mapping with high accuracy under all-day, all-weather conditions, therefore it has become one of the important means of DEM reconstruction. With the development of InSAR and the increasing demand of topographic mapping in various application, the requirements of DEM precision become higher. According to InSAR model, the precision of DEM reconstruction is proportional to interferometric baseline in the case of other parameters fixed. Therefore using the InSAR to achieve high precision DEM extraction need long baseline. However, while implementing high precision DEM reconstruction, the long baseline also introduces other problems. In rugged terrains the interference fringe obtained by long baseline is very dense which is easy to cause phase ambiguity (or even loss), and once the interferometric phase is lost, the correct phase cannot be obtained after phase unwrapping. Using low accuracy external DEM during InSAR process can improve the reliability and accuracy of DEM reconstruction. We study on how to use external DEM to achieve high precision InSAR processing in this manuscript. There are four main steps to extract DEM by InSAR, namely InSAR data acquisition, image processing, coherent processing and elevation reconstruction. Combined with the existing research results, this manuscript proposes an overall framework of interferometric processing based on external DEM which includes two approaches distinguished by the position of exploiting external DEM, that is, external DEM is introduced into SAR image processing and coherent processing respectively. And interferometric technology is used to refine the low accuracy external DEM to obtain the higher accuracy DEM. There are differences between the interferometric models of two approaches, so this manuscript introduces the implementation process of the two approaches and analyzes their respective characteristics. In the first method, backprojection (BP) algorithm is adopted to introduce the external DEM into image processing, considering that BP algorithm has the advantages of good focusing performance, high motion compensation accuracy and easy introduction of external terrain. The BP algorithm is used to obtain the images for subsequent interferometric processing. Then the interferometric model based on BP algorithm is built to obtain high accuracy DEM. This approach has many advantages. The image pairs obtained by BP algorithm are in the same imaging grid, so image registration has been realized during the BP algorithm image processing. As external DEM information is used in the image processing, the elevation information of external DEM is implied in the SAR images, which can reduce the density of interferometric fringes and even avoid phase unwrapping. In conclusion, this method has the potential to simplify the interferometric processing and improve the accuracy of DEM reconstruction. In the second method, the more efficient frequency domain algorithm, such as RD algorithm and Chirp Scaling algorithm, is adopted to obtain the range Doppler images. Then during the coherent processing procedure of image pairs, external DEM information is exploited to improve the accuracy of elevation reconstruction. External DEM can be used in registration, phase unwrapping and baseline parameters estimation to improve their accuracy, so as to optimize interferometric processing performance and obtain high accuracy DEM. Finally, preliminary experiments, including simulation experiment and the airborne data processing, have been carried out to verify the effectiveness of InSAR processing based on external DEM. Airborne repeat pass data which were acquired at C-band (5.4 GHz) over a test site in Zhuhai, China are used in the airborne data processing, and the Shuttle Radar Topography Mission (SRTM) DEM of illuminated area is used as external DEM.
Authors: Hu, Xiaoning (1,2); Xiang, Maosheng (1,2); Wang, Bingnan (1,2); Chong, Jinsong (1,2)Flexible mobile mapping of surface displacements with repeat-pass interferometry at L-band from custom moving platforms such as cars and UAVs has been a rather unexplored field.In this contribution we address this topic comprehensively by means of surface displacements and coherence measurements obtained with our novel car-borne and UAV-borne L-band SAR system setupat three different test sites in Switzerland. The reduced temporal decorrelation at L-band is an important advantage and a complementary property as compared to high-frequency (quasi-)stationary systems. While the sensitivity to line-of-sight displacements is lower, the longer wavelength permits to acquire longer interferometric time intervals also in natural terrain and in adverse conditions, in which the decorrelation time at X- or Ku-band (the frequency of many stationary systems) can be in the order of minutes or less. Terrestrial synthetic aperture radar acquisitions from a car driving on a road or acquisitions from a UAV allow to obtain synthetic aperture lengths of 100m and more which yields high-resolution SAR imagery also at lower frequency such as L-band. At the same time the view geometry can be chosen to offer line-of-sight views to landslides that complement the view geometries available from spaceborne SAR systems. Then, using a time-domain back-projection image focusing approach, it is ensured that even for curvilinear SAR data acquisition paths (e.g. a car driving along a curved road)well focused SAR images and therefore high-quality interferometric data with good spatial resolution can be obtained. Based on these properties we show that such a mobile InSAR system fills a current gap in terms of available InSAR system for displacement monitoring.In our presentation, we briefly introduce our recently developed L-band InSAR acquisition and processing systemwith attention to various INS/GNSS setups used in the experiments and to the time-domain back-projection based processing approach used, which allows to directly focus the SAR data to single-look complex data in map coordinates (GEOSLC) and subsequent generation of topographic-phase-free differential interferograms in map coordinates. As a main focus we show the potential and discuss the challenges and the limitations of this mobile mapping system for terrestrial InSAR displacement measurements at L-band with high spatial resolution.We do so with the help of three repeat-pass interferometry showcases:1) car-borne mapping of surface displacements of fast-moving land slide and surrounding area from a distance of a few hundred meters as well as a few kilometers distance2) car-borne mapping of surface displacements of a glacier. 3) UAV-borne mapping of surface displacements of a steep slope with various land covers using different spatial baselines. See also a few examples of the experimental setups and of the interferometric results obtained in the additional pdf uploaded with the abstract. Special attention is given to a few particularities that come along with such mobile mapping data acquisition scenarios and which can naturally be handled by the time-domain SAR imaging approach, such as:- variable synthetic aperture length- range-varying azimuth resolution- space-variant mean “squint angle”- differential (topography-free) interferogram formation directly in map coordinates (through TDBP image formation).We also discuss the impact of (and the requirements on) spatial baselines in the context of the implemented acquisition system setup for successful displacement mapping for different natural terrain/land cover.All three test scenarios show that both UAV-borne and car-borne displacement measurements are feasible with high quality over various natural terrain and with temporal baselines of up to 2 weeks and longer.
Authors: Frey, Othmar (1,2); Werner, Charles (1); Manconi, Andrea (1); Coscione, Roberto (2)The project TecVolSA (Tectonics and Volcanoes in South America) aims at developing an intelligent Earth Observation (EO) data processing system for monitoring the earthquake cycle and volcanic events in South America. The Remote Sensing Technology Institute of DLR participates to this project together with GFZ (German Research Centre for Geosciences). The project is partially financed by Helmholtz. So far we have processed about 40 Sentinel-1 slices covering the Atacama Desert with mixed Permanent Scatterer and Distributed Scatterer (PS/DS) techniques. The area is very dry and the spatio-temporal coverage is excellent. Tropospheric correction have been applied using ECMWF ERA5 data, hence improving the performance in observing both topography related and large scale deformation signals. The current results reveal, as expected, plenty of interesting signals to be interpreted (see attached figure for an overview of the velocity field). Preliminary GPS cross-validation, thanks to data freely available from the Geodetic Nevada Laboratory, confirm that the InSAR relative error in the estimated velocities is in the order of 1 mm/yr at large scale (>100 km) and confirms the large scale signal related to the subduction of the Nazca plate (see attached figure). More GNSS validation will be possible with additional GPS stations. The challenge of the project is the separation of different contributions to the InSAR measurements: apart from the tectonic effects, there are contributions coming from volcanic unrest, atmospheric delays, moisture effects, snow, flank instability (likely downhill creep or solifluction related to permafrost, see attached figure), salt lake growth, mining, and likely more. We are dealing with this complexity with a diversity of tools: physical modeling and statistical analysis, deep neural networks, and expert knowledge. GFZ contributes process knowledge, historic seismic data, in-situ motion measurements and observations and 4D geophysical modelling codes for producing a diverse database for the training of neural networks in order to autonomously discover significant events in noisy data. We tackle the problem as a semi-supervised multi-class classification approach where the labeling of the known deformation phenomena is provided by GFZ. Signals for which the source of deformation is unknown are identified and clustered automatically using advanced unsupervised machine-learning techniques. Therefore, we leverage from the advantages of both supervised and unsupervised learning and improve the accuracy for detection and classification of different deformation sources. The networks and AI-based methods are developed at DLR. This new approach (InSAR + Artificial Intelligence) should be able to process the massive data stream of the Copernicus Sentinel-1 SAR mission. South America was selected because manifold geophysical signals can be expected there in short time scales and plenty of in-situ data are available. This project will complement the current model-based geophysical research by a data-driven AI-based approach. Training and applying this intelligent system shall improve our understanding of geophysical processes related to natural and anthropogenic hazards. At a later stage the system shall be scalable to global processing capacity. Future developments on the InSAR processing will include ionospheric corrections based on split-spectrum and mosaicking of the velocity and displacement series. Some issues with the L1 processor are hindering the deployment of the split-spectrum technique. Stacks from the ascending geometry are already being processed and will help the geophysical interpretation.
Authors: De Zan, Francesco (1); Ansari, Homa (1); Parizzi, Alessandro (1); Shau, Robert (1); Eineder, Michael (1); Montazeri, Sina (1); Navarro Sanchez, Victor (1); Walter, Thomas (2)GNSS and DInSAR are nowadays the most common space-borne technologies for the study of the solid Earth deformations. In particular, the GNSS is an active L-band microwave technology that was introduced for civil and scientific scopes in the early 1990s. Using high-quality data acquired all around the globe, it helps the scientific community in better understanding our planet, measuring the slow and relentless temporal evolution of Earth’s crustal plates. In the last years, the number of permanent GNSS stations has increased, although not uniformly, on the local, regional and global network scale. The GNSS continuously operating networks provide measurements in near real-time for different purposes, both geodetic and geodynamic, at the national or local levels [1-2]. On the other hand, the DInSAR is a microwave remote sensing technique playing nowadays a crucial role in the investigation of Earth surface deformation phenomena with high level of accuracy [3]. In the last 20 years Advanced DInSAR approaches have been developed. These approaches allow the retrieval of the temporal evolution of the surface deformations by exploiting huge SAR datasets collected since the ‘90s. Among these, the Small BAseline Subset (SBAS) algorithm [4] and in particular its parallel computing version (P-SBAS), is an advanced DInSAR approach, aimed at computing displacement time-series with sub-centimetre accuracy [5]. On April 3rd, 2014, the Sentinel-1A (S-1A) sensor, which is the first of a family of satellites designed to provide C-Band (~5.5 cm wavelength) SAR data, developed within the European Copernicus Programme and primarily devoted to interferometric applications [6]. In addition, the twin system Sentinel-1B (S-1B) was launched on April 25th, 2016, halving the sensors interferometric revisit time from 12 to 6 days, thus achieving an unprecedented temporal sampling for the DInSAR analysis. In addition, the Sentinel-1 constellation is characterized by a new acquisition modality named Terrain Observation by Progressive Scans (TOPS) that allows to catch radar images with a very big swath (about 250 km square). This peculiarity has a strong impact in the DInSAR analysis since the low frequency displacement signals, mainly due to the movements of crustal plates, are more evident, especially if compared with to the old SAR sensor generation that acquired radar images with about one hundred kilometres of swath length. In this work we present a methodology to integrate the GNSS measurements within the P-SBAS processing chain to retrieve information about both the high and low spatial frequency deformation patterns. In addition, by exploiting the GNSS stations we have developed a procedure to clean out possible residual deformation phase artefacts affecting the national scale DInSAR displacements time-series produced at national scale to increase the quality of the generated results. This approach has been applied on 175 descending S1 datasets, processed with the P-SBAS algorithm, covering almost the whole European territory from March 2015 up to December 2018. Rationale of the proposed approach The GNSS stations measurements usually provide very precise three-dimensional absolute measures with millimetre accuracies of the deformation field (East, North and Vertical deformation components), however, the spatial distance among these stations is often too large to completely capture localized displacement patterns, while they are particularly appropriate for the study and comprehension of the low spatial frequency deformation fields due to the tectonic migration. On the other hand, the DInSAR technique generates one-dimensional measurements in the radar Line of Sight (LOS), but with a wide spatial density. This peculiarity makes the DInSAR an incomparable technology for reconstructing high spatial frequency deformation patterns. Accordingly, the complementary use of both technologies could be suitable for the entire reconstruction of the deformation signals affecting a specific area of interest (low and high spatial frequency deformation contributions).The developed techniques work on the estimation of the low spatial frequency deformation components and residues to increase the quality of the DInSAR analysis. Since our approach make use of the GNSS information for the estimation of the low spatial frequency deformation signals, we have to identify, among hundreds of available GNSS stations, the ones affected by regional tectonic deformation trends, only. To perform this screening operation, we developed an automatic procedure that is essentially based on two sub-procedures: GNSS stations time-series filtering. Evaluation of the spatial similarity index of the GNSS stations. Moreover, for this analysis we benefit from the continuous GNSS position time-series provided by the Nevada Geodetic Laboratory at the University of Nevada, Reno, USA (UNR-NGL). This free and open service allows the scientific community to exploit thousands of GNSS stations around the world, processed in a standard way and daily updated. The GNSS stations time series can be sometimes perturbed by the presence of several jumps, which can be generated by different causes, both anthropic and natural. Software modifications or physical changes of the antenna as well as the occur of a seismic event, are the main reasons for the presence of jumps within position time series. Sometimes the GNSS time series providers make available to the scientific community auxiliary files where the jumps dates are highlighted, otherwise it is fundamental to implement advanced procedures able to identify the presence of strong temporal gradients. The tectonic plate velocity, especially for the planar components (East-West and North-South), can be considered constant in time, and the developed technique is targeted to the estimation of these values as accurate as possible. To this aim the presented algorithm is able to automatically filter the GNSS stations time series identifying for each station the most reliable temporal range for the estimation of the tectonic trend. After the GNSS time series filtering, the second step of the procedure for the integration of these ground-based displacement measurements within the P-SBAS processing chain is the automatic identification of the GNSS stations that are affected by displacements due to the local tectonic only, and thus to reject from the DInSAR processing the stations that sense also some local displacement effects. To this aim, the proposed algorithm takes advantage of a strong spatial correlation of the tectonic trends. Whose deformations are spatially correlated for tens kilometres square. For this reason, to select the GNSS stations to be efficiently used in the P-SBAS processing, we can evaluate a velocities similarity index of the GNSS stations that fall within a specific correlation window. The spatial similarity index is designed to be equal to 1 when the stations deformation velocity is exactly the same, while the similarity decreases for values that far off 1. Moreover, it is important noting that the GNSS technology provides a high accuracy level for the planar components but not for the vertical one. Accordingly, to above, the vertical deformation component is excluded from the similarity analysis. The GNSS stations for which the similarity index for both east-west and north-south velocity is included between 0.95 and 1.05, are considered for the low spatial frequency component estimation within the P-SBAS processing chain. Finally, the correlation window used for the GNSS stations selection is fixed to 30 km2. Experimental Results In Figure 1 we represent the planar components of the GNSS stations that cover the whole European territory and that are compliant with the similarity conditions above described. It is clear that the stations selected for the P-SBAS DInSAR processing are strongly spatially correlated and no evident singularities are present. The developed algorithm has been applied over almost the whole European territory, in Figure 2 we show the area covered by the P-SBAS displacement analysis, computed by exploiting the entire S1 data archive acquired from descending orbits during the period March 2015 – December 2018. The total number of performed P-SBAS processing is 175, and the extent of the analyzed area is of about 5,000,000 km2. As additional remark, we underscore that with this approach it is possible to separately exploit the high spatial frequency deformation component (provided by DInSAR technique) or the overall signal obtained adding the low spatial frequency component to the previous one, thus increasing the range of the value-added results offered by these advanced DInSAR analyses. References [1] Dow JM, Neilan RE, Rizos C (2009) The International GNSS Service in a changing landscape of Global Navigation Satellite Systems. J Geod 83:191–198 [2] Mueller II, Beutler G (1992) The International GPS Service for Geodynamics—development and current structure. Proceedings of the 6th Symposium on Satellite Positioning. Ohio State University, Columbus, OH, pp 823–835 [3] R. Burgmann, P.A. Rosen, and E.J. Fielding, “Synthetic aperture radar interferometry to measure Earth’s surface topography and its deformation”. Annu Rev Earth Planet Sci , 28, 169–209, 2000 [4] P. Berardino, G. Fornaro, R. Lanari, and E. Sansosti, “A new Algorithm for Surface Deformation Monitoring based on Small Baseline Differential SAR Interferograms”, IEEE Trans. Geosci. Remote Sens. 40, 11, 2375-2383, 2002. [5] F. Casu, S. Elefante, P. Imperatore, I. Zinno, M. Manunta, C. De Luca, and R. Lanari, “SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 7, no. 8, pp. 3285–3296, 2014 [6] N. Yague-Martinez, P. Prats-Iraola, F. Rodriguez Gonzalez, R. Brcic, R. Shau, D. Geudtner, M. Eineder, and R. Bamler, “Interferometric Processing of Sentinel-1 TOPS Data”, IEEE Trans. Geosci. Remote Sens., 54, 4, 2220-2234, 2016
Authors: De Luca, Claudio; Bonano, Manuela; Casu, Francesco; De Novellis, Vincenzo; Manunta, Michele; Manzo, Mariarosaria; Onorato, Giovanni; Zinno, Ivana; Lanari, RiccardoSAR tomography has been considered in the last decade as one of the leading technique in SAR imaging that delivers a 3D representation of the earth. By performing the same concept of tomography implemented in medicine, SAR Tomography uses several acquisitions with different angle of view to synthesize a small aperture around the target to assure a reconstruction in elevation. Compare to InSAR, the SAR Tomography retrieves not just the heights of the targets like the InSAR but also the position of each scatters along the slant rang. The resolution attained in elevation follows the same level of analysis as the one provided by SAR in azimuth, which is directly related to the total baselines aperture. Due to the small elevation aperture of the new high resolution space-borne SAR systems, the resolution in elevation represents one of the major restraint of SAR Tomography. The non-uniform distribution of baseline and the limits of acquisitions can also be considered as the primary problems for the SAR tomography reconstruction. As a result, several reconstruction methods have been published on this matter to improve the SAR Tomography reconstruction. The elevation resolution achieved by these classical SAR Tomography reconstruction techniques is significantly larger than the high resolution in range/azimuth’s acquire by the new space-borne SAR systems such as TerraSAR-X, TanDEM-X and COSMO-Skymed. In the recent years, a new sampling theory known as Compressive sensing was implemented in this domain which go beyond the achieved elevation resolution ‘Rayleigh resolution’ with a small number of acquisitions by taking advantage of the random distribution of the baselines and the prior knowledge regarding the sparsity of the reflectivity profile in elevation. Most of the Compressive sensing reconstruction methods are based on the L1 norm minimization due to its potential to approach the sparse solution. however, the reconstructed profile by the L1 norm still has a lot of outliers due to the violation of the fundamentals conditions of compressive sensing such as RIP in measurements matrix. To overcome this problem a new reconstruction method named SL1MMER based on the L1 & L2 norm minimization with model selection was proposed (Xiao Xiang Zhu & Richard Bamler 2011). We have introduced in this study an adaptive threshold in order to select the predominant backscattering mechanism (s) within a resolution cell and therefore a feasible reconstruction by L1 & L2. The proposed process was tested on simulated targets and on images acquired by Terra SAR X. We compared and evaluated our results with the mentioned reconstruction method (SL1MMER) Scale-down by L1 norm Minimization, Model selection, and Estimation Reconstruction, (SVD) Singular value decomposition (BF) Conventional beamforming and (NLS) Nonlinear least squares.
Authors: Daoud, Ishak; Karima, Hadj Rabah; Aichouche, Belhadj aissaEarth orbiting satellites, such as Sentinel 1A‐B, build up an ever‐growing set of synthetic aperture radar images of the ground. This conceptually allows for real‐time monitoring of ground displacements using Interferometric Synthetic Aperture Radar (InSAR) with a dense coverage and a fine resolution, notably in tectonically active regions such as fault zones or over volcanoes. However, existing small-baseline time series techniques will inevitably become overwhelmed by the rapid accumulation of images, as they require increasing computing power and memory as the amount of observations grows. We propose a Kalman filter for InSAR time series analysis (KFTS), an efficient method to rapidly update preexisting time series of displacement with data as they are made available, with limited computational cost. KFTS solves together for the evolution of interferometric phase change at each time step together with the parameters of a time dependent model of ground deformation and associated uncertainties. KFTS iteratively projects the a priori knowledge of deformation at a future time, a forecast that is later adjusted when new interferograms are available. Synthetic tests of KFTS reveal exact agreement with the equivalent weighted least squares solution, as expected, and a convergence of descriptive model parameter after the assimilation of about 1 year of data. We include the impact of sudden deformation events such as earthquakes or slow slip events on the time series of displacement. First tests of the KFTS on ENVISAT data over Mt. Etna (Sicily) and Sentinel 1 data around the Chaman fault (Afghanistan, Pakistan) show precise (± 0.05 mm) retrieval of interferometric phase change when data are sufficient. Otherwise, the optimised time dependent model is used to forecast phase change. This simple time dependent model of deformation only fills the gaps when data is missing. Good agreement is found with classic time series analysis solution and GPS‐derived time series. Based on the study of the Chaman fault, we also implemented an automatic parametrisation to model co-seismic displacement due to moderate magnitude earthquakes (Mw~5) affecting a fraction of the processed area. Accurate estimates are conditioned to the proper parametrisation of errors so that models and observations can be combined with their respective uncertainties. However, perfect agreement with the data is guaranteed if the error associated with interferograms (i.e. interferometric misclosure) is several orders of magnitude smaller than mismodelling error, which is included in the process noise associated with the last added SAR acquisition. The quality of the estimate can be assessed throughout the iterative process and improved by looking at the innovation term, which informs about the model predictive value, and the root mean square error, which evaluates phase accuracy with respect to interferograms. KFTS is designed as a flexible tool and is freely available to process ongoing InSAR time series.
Authors: Dalaison, Manon (1); Jolivet, Romain (1,2)Compact and affordable radar transponders are an attractive alternative to corner reflectors (CR) for SAR interferometry (InSAR) deformation monitoring, datum connection and geodetic data integration. The ERC-C radar transponders, manufactured by Metasensing B.V., are the first commercially available transponders for the Sentinel-1 and Radarsat-2 C-band SAR satellites. As these instruments came quite recently to the market, their precise radiometric, geometric, and InSAR phase stability characteristics, essential for InSAR geodesy, are not yet known. Moreover, the impact of the secular or seasonal effects, such as variations in temperature and humidity, on the suitability for long-term use at permanent geodetic stations has yet to be proven. This work aims to address these issues using short baseline experiments, with five ECR-C units and six corner reflectors deployed at test sites in The Netherlands and Slovakia. The Sentinel-1 time series of ECR-C that we analyze are over 1.5 years long. For the tested ECR-C units, we observe a high average Radar Cross Section (RCS) of over 42 dBm2 across 15 degrees of elevation misalignment. Such RCS is comparable to a triangular trihedral corner reflector with a leg length of 2.0 m. However, the RCS from the ECR-C showed much higher temporal variations (0.3 - 1.2 dBm2 standard deviation) than would be observed from a corner reflector of equivalent RCS. RCS variations show a partial correlation with surface temperature changes. From precise SAR positioning of ECR-C units, we observe aspect-dependent and the unit-dependent internal electronic delay between 1.4 - 2.0 m in the slant-range, showing the need for individual calibrations for geodetic applications requiring high absolute positioning accuracy. The precision of the InSAR phase double-differences over short baselines between ECR-C and stable reference corner reflectors is 0.5 - 1.0 mm (one sigma). However, InSAR double-differences show a distinct correlation with surface temperature, causing seasonal variations up to +/- 3 mm. These should be modelled and corrected if the ECR-C is to be used for high precision InSAR applications. This is one of the main drawbacks of compact radar transponders compared to bulky corner reflectors. Our main conclusion is that compact radar transponders are a viable alternative to corner reflectors for locations where installation of corner reflector is not possible or otherwise not practical.
Authors: Czikhardt, Richard (1); van der Marel, Hans (2); Papco, Juraj (1)Landslides, tsunamis, and volcanic eruptions occur over numerous temporal and spatial scales. Although these phenomena are often studied individually, there is frequent interconnectivity between disaster types. The GeoSciFramework project, funded by the NSF Office of Advanced Cyberinfrastructure and NSF EarthCube programs, aims to improve intermediate-to-short term forecasts of catastrophic natural hazard events, allowing researchers to instantly detect when an event has occurred and reveal the more suppressed, long-term motions of Earth's surface at unprecedented scales. These goals will be primarily accomplished by training machine learning algorithms to recognize patterns across various data signals during geophysical events, including Interferometric Synthetic Aperture Radar (InSAR), seismic, GNSS, tide gauge, and gas-emission data, which are all effective tools used to characterize earthquakes, tsunamis, and volcanic eruptions. Differential InSAR (DInSAR) time-series analysis quantifies line-of-sight (LOS) ground deformation at mm-cm spatial resolution. Repeatedly acquired scenes from different satellites using various wavelengths are integrated into a single time series of ground deformation using the Multidimensional Small Baseline Subset (MSBAS) method [Samsonov and D’Oreye, 2012]. Furthermore, InSAR processing can be combined with Global Navigation Satellite System (GNSS) data to obtain 3-D ground surface motions [Samsonov and Tiampo, 2006]. Relevant sources of error, such as topographic and atmospheric effects must be assessed appropriately. To account for topographic error, each scene is fitted to a 5- or 10-m Digital Elevation Model (DEM), acquired either from the United State Geological Survey (USGS) National Map or created using optical satellite images. Atmospheric corrections can be accounted for by applying a high-resolution numerical weather model such as WRF simulations [Bredemeyer et al., 2018; Jung et al., 2014] or by exploiting GNSS measurements [Li et al., 2005; Rosen et al., 1996; Gonzalez et al., 2015]. Here, we compare time series products generated under three different processing techniques. The first, an automated version of InSAR processing using the small baseline subset (SBAS) method performed in parallel on systems such as Generic Mapping Tool SAR (GMT5SAR) and the Generic InSAR Analysis Toolbox (GIAnT). The second method will resemble the first but will implement different processing systems for performance comparison using the InSAR Scientific Computing Environment (ISCE) and the Miami InSAR Time Series Software in Python (MintPy). The final strategy, developed by Drs. Zheng and Zebker from Stanford University, concentrates on the topographic phase component of the SAR signal so that simple cross multiplication returns an observation sequence of interferograms in geographic coordinates [Zebker, 2017]. Our results provide high-resolution views of ground motions and measure LOS deformation over both short and long periods of time.
Authors: Corsa, Brianna (1); Tiampo, Kristy (1); Kelevitz, Krisztina (2); Baker, Scott (3); Meertens, Charles (3); Mencin, David (3)Time series InSAR is routinely used to monitor ground deformation for multiple processes, but issues remain with selecting pixels from which useful deformation signals can be extracted. Persistent scatterer (PS) and small baseline approaches are commonly applied, but are not optimal for pixels that sometimes lose coherence completely. These pixels often belong to rural areas absent of strong scatterers. “Similar time-series interferometric phase” (STIP, Narayan et al., 2018), has been proposed as an interesting new approach for candidate PS pixel selection, and here we investigate whether a similar approach can be extended to pixels with variable coherence characteristics. The STIP algorithm works on the basis of summing the phase correlation between a central pixel and a neighbouring pixel within a defined window (over the interferogram stack) for different time lags. When the maximum summed correlation between the central and neighbouring pixel occurs at zero time lag, then this pair of pixels are considered siblings, and the central pixel gains a STIP sibling. Repeating this for all pixels in the neighbourhood determines how many siblings each pixel has. By using a threshold, pixels with enough siblings are identified as candidate PS pixels. These pixels can then go on to be used in time-series analysis of ground deformation. Our investigations so far have focussed on optimal determination of the thresholding for STIP siblings and the number of interferograms required to robustly identify STIP pixels. By analysing the results of applying the STIP algorithm to synthetic noise data, we can separate the results when applied to real interferograms into two different populations: noise pixels and coherent pixels. This provides a probabilistic way to set the threshold for STIP siblings without relying on an arbitrary value. The algorithm does not require correlation between the central and neighbouring pixels throughout the entire stack of interferograms to be picked as siblings. This presents the opportunity for a way to constrain the number of interferograms the pair of pixels should be coherent in, and an additional way to threshold the candidate pixels. Our current efforts are focussing on using STIP siblings to identify coherence of pixels in individual interferograms, in a similar approach to RapidSAR.
Authors: Connolly, Jacob (1); Hooper, Andy (1); Wright, Tim (1); King, Stuart (2); Bekaert, David (3); Ingleby, Tom (4)Global rapid full resolution InSAR derived from Sentinel-1 geocoded bursts, leveraging scalable computingKelly Olsen, Matthew Calef, Michael S. Warren, Piyush Agram, Scott ArkoDescartes Labs Inc., Santa Fe, NM (USA) The Copernicus Program’s Sentinel-1 SAR constellation has been imaging most of the Earth’s land masses with a revisit time of 6 to 24 days since September 2014. SLC bursts are organized in InSAR compatible stacks, most of which have at least a year of data, thereby enabling historical time series analysis. Pixel selection in full resolution InSAR (cf. [1]) requires this historical data, in particular a stack of co-registered SLC. Analysis about the present is contingent on historical collections. Historical and ongoing Sentinel-1 collections address this contingency, enabling rapid full resolution InSAR anywhere there are sufficiently many “good” pixels. In this work we present a method allowing rapid assessment of an AOI for full resolution InSAR and rapid generation of analysis. Data pipelines A common method for preparing data for InSAR analysis methods, such as full resolution InSAR, is to choose one SLC as the reference SLC, and to co-register the other SLCs to the Range-Doppler coordinate system implicit in the reference SLC. One drawback to this approach is that this coordinate system is arbitrary and not supported by geospatial tools. Instead we co-register to a UTM grid with a 10m Northing resolution and a 2.5 meter Easting resolution. This geocoding approach follows [2]. Our choice of resolution aims to reflect the overall point-density and anisotropy of a typical Sentinel-1 IW mode Range-Doppler sample grid projected onto UTM. SLC phase for a pixel reflects propagation distance modulo wavelength and the scattering properties of the area associated with that pixel. This phase appears nearly random and current image compression methods are not effective. For this reason it is prohibitively expensive to store a permanent global library of geocoded bursts at the geocoded resolution. Instead we have developed a mechanism [3] to geocode rapidly historical stacks of imagery. We have not implemented enhanced spectral diversity, but retain the ability to incorporate a priori offset models during geocoding, Many of the areas we would like to monitor are small enough to fit within a single burst. Data preparation consists of identifying the stacks of InSAR compatible bursts. This can be done with visualization tools (Figure 1) or with programmatic access. The stack ID provides a list of bursts to be geocoded. The list of geocoding tasks are sent to a Kubernetes cluster, where geocoding takes about 2 CPU-minutes per burst for a single polarization. This provides a means whereby we can rapidly generate a stack of co-registered historical SLCs necessary for point-selection that can be accessed with geospatial APIs. Full resolution InSAR methods Our full resolution InSAR methods leverage published work. Pixel selection is based on amplitude dispersion ([1]), or phase estimation ([4] and [5]) followed by temporal coherence requirements ([6, Section II]). We’ve implemented several unwrapping methods. In one method we follow [7] to estimate deformation trends and DEM error, remove the corresponding phase, use a minimum cost-flow (MCF) approach to unwrap the residual, and then add the deformation trend back. In another method we implement something akin to, but simpler than, [8]. Here we form a Delaunay triangulation of bursts plotted in a scaled temporal/perpendicular-baseline space, and unwrap edges from the spatial Delaunay triangulation. These edges are then used to recover deformations at selected pixels through a second unwrapping step. Results are already in UTM and can be viewed with any number of tools. For an AOI with a few thousand “good” pixels, analysis can be performed and visualized within a few minutes; Figure 2 shows an example. Using these methods with Sentinel-1 imagery has two drawbacks. The first is that minimum cost flows methods ensure that the difference between wrapped and unwrapped phases is an integer multiple of 2 pi. However, visual inspection of the results often leads to the subjective conclusion that the unwrapper chose the wrong integer. The second is that there are AOIs that support too few “good” pixels. Broadly assessing error bars can be challenging absent ground truth. Uncertainty analysis One component of error is sensitivity to noise in the input phase. Our approach is to take the input stack of co-registered data, extract noise models that can be used to form synthetic stacks that have the same noise properties. We use scalable computing to run our full resolution InSAR analysis on 100 stacks and can look at the spread in deformation histories. For Distributed Scatterer (DS) methods we use the sample correlation matrix as a covariance matrix to produce a synthetic phase, which we apply to the observed amplitude, this can be considered a simplification of [9]. For Persistent Scatterer (PS) methods we use a Rice distribution informed by the amplitude dispersion to generate phase noise. We can compare noise statistics for real and synthetic data as is shown in Figure 3 to vet the accuracy of our noise models. Our experience is that observed deformation history is often within one-sigma of the ensemble mean deformation, as is shown in Figure 3. This use of synthetic data allows us to assess our methods, but also further down select to pixels where the output of our methods are insensitive to the noise generated by our models -- this down selection is shown on the right panels of Figure 3. There are other reasons why InSAR results may differ from actual deformation, but this approach addresses errors due to phase noise. Conclusions Sentinel-1’s historical stacks of InSAR compatible bursts enable full resolution InSAR analysis in much of the world. Rapid geocoding and scalable compute allow full resolution InSAR to be done quickly, and for sensitivity to noise to be assessed rapidly. References Ferretti, Alessandro, Claudio Prati, and Fabio Rocca. "Permanent scatterers in SAR interferometry." IEEE Transactions on geoscience and remote sensing 39.1 (2001): 8-20. Zebker, Howard A. "User-friendly InSAR data products: Fast and simple time series processing." IEEE Geoscience and Remote Sensing Letters 14.11 (2017): 2122-2126. Calef, Matthew, Warren, Mike, Agram, Piyush, Arko, Scott, “Global near real-time backscatter and InSAR products derived from Sentinel-1 geocoded bursts” Fringe 2021 Guarnieri, Andrea Monti, and Stefano Tebaldini. "On the exploitation of target statistics for SAR interferometry applications." IEEE Transactions on Geoscience and Remote Sensing 46.11 (2008): 3436-3443. Ansari, Homa, Francesco De Zan, and Richard Bamler. "Sequential estimator: Toward efficient InSAR time series analysis." IEEE Transactions on Geoscience and Remote Sensing 55.10 (2017): 5637-5652. Ferretti, Alessandro, et al. "A new algorithm for processing interferometric data-stacks: SqueeSAR." IEEE transactions on geoscience and remote sensing 49.9 (2011): 3460-3470. Pepe, Antonio, and Riccardo Lanari. "DEM correction and mean surface displacement rate retrieval from a stack of wrapped multi-temporal DInSAR interferograms." 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2017. Costantini, Mario, Fabio Malvarosa, and Federico Minati. "A general formulation for redundant integration of finite differences and phase unwrapping on a sparse multidimensional domain." IEEE Transactions on Geoscience and Remote Sensing 50.3 (2011): 758-768. Agram, P. S., and M. Simons. "A noise model for InSAR time series." Journal of Geophysical Research: Solid Earth 120.4 (2015): 2752-2771.
Authors: Olsen, Kelly (1); Calef, Matthew Thomas (2); Warren, Michael S. (2); Agram, Piyush (2); Arko, Scott (2)The launch of Sentinel-1 has brough new opportunities for SAR-based earth observation with respect to both data coverage and volume. Especially the field of InSAR has immensely benefited from the mission in many application domains. However, while most studies focus on the detection of deformation and surface changes, only a small number deal with the derivartion of height information in general. The reasons are of both technical nature (narrow orbital tube of Sentnel-1) and lack of understanding of critical parameters related to the selection of suitable data pairs. This contribution gives a short summary on the technical workflow within the ESA SNAP toolbox and the impact of different variables on the expected quality of generated digital elevation models. Futhermore, it reviews existing studies dealing with the derivation of surface heights by means of Sentinel-1 interferometry and compares their application, techniques, as well as the quality of the results. The results indicate that a large number of current studies neglect basic InSAR principles and scientific standards. The review helps to identify both best practice examples and also to highlight current shortcomings. Additonally, experieces from the SNAP user forum are collected to underline challenges related to the objective. The findings are gathered to define standardized recommendations as a base for future research and scientific discourse. The attached figure shows the impact of the perpendicular baseline on the interferogram (top) and the resulting DEM (bottom, hillshade) quality given as root mean square error.
Authors: Braun, Andreas (1,2)Multi-temporal SAR interferometry (MTInSAR) techniques are able to derive displacement maps and displacement time series over coherent objects on the Earth, used for monitoring either geophysical ground deformation or structural instabilities. Nowadays, several datasets are available at different wavelengths, spatial resolutions, and revisit time, collectively covering long time periods (even more than 20 years). In particular, short revisit times (e.g. from Sentinel-1 and COSMO-SkyMed constellations), by improving the temporal sampling, make theoretically possible to catch high-rate and non-linear kinematics, which typically characterize warning signals related to, for instance, landslides or pre-failure of artificial infrastructures. However, MTInSAR algorithms generally fit the displacement signal by using a linear model, which is computationally convenient, and also, more robust than higher-order models for what concerns the errors affecting the InSAR phase. Moreover, the analysis of the MTInSAR products is often performed by only considering the mean displacement rate, computed over the monitoring period, which is the information typically displayed on the displacement maps. Therefore, in order to fully exploit the content of MTInSAR products, methods are needed for automatically identifying relevant changes along displacement time series, and, consequently, classifying the targets on the ground according to their kinematic regime. This also allows performing a more reliable ground deformation spatial analysis, by distinguishing among spatial patterns of different kinematics (linear, bi-linear, quadratic, discontinuous and periodic). Recently, approaches have been proposed to tackle this problem, which use different strategies, based on Principal Component Analysis (Chaussard et al., 2014), statistical tests (e.g., Berti et al., 2013), or more sophisticated probabilistic multiple hypotheses testing (Chang and Hanssen, 2016). Our work proposes a new set of rules based on statistical characterization of displacement time series, which allows, under certain constraints, recognising automatically different kinematic classes, and estimating the relevant parameters useful for target characterization in time. We introduce a new statistical test based on the Fisher distribution aimed at evaluating the reliability of a displacement model with a certain statistical confidence. We also studied the reliability of other tests already introduced in literature and used for comparing two different models, namely the Akaike Information Criterion, the Bayesian Information Criterion, and the Fisher test. A performance analysis has been carried out by simulating time series with different characteristics in terms of kinematic (stepwise linear with different breakpoints and velocities), level of noise, signal length and temporal sampling. The displacement estimations performed by using the different tests have been compared. Finally, a procedure for selecting the optimum displacement model has been defined according to the output of the performance analysis. The procedure has been also tested by using real datasets coming from Sentinel-1 and COSMO-SkyMED missions, and covering areas affected by non-linear displacements and slope failures. Acknowledgments This work was supported in part by the Italian Ministry of Education, University and Research, D.D. 2261 del 6.9.2018, Programma Operativo Nazionale Ricerca e Innovazione (PON R&I) 2014–2020 under Project OT4CLIMA; and in part by Regione Puglia, POR Puglia FESR-FSE 204-2020 - Asse I - Azione 1.6 under Project DECiSION (p.n. BQS5153). References Berti, M., Corsini, A., Franceschini, S., Iannacone, J. P., 2013. Automated classification of Persistent Scatterers Interferometry time series. Nat. Hazards Earth Syst. Sci. 13(8), 1945–1958. Chang, L., Hanssen, R.F., 2016. A Probabilistic Approach for InSAR Time-Series Postprocessing. IEEE Trans. Geosci. and Remote Sens. 54(1), 421–430. https://doi.org/10.1109/TGRS.2015.245903 Chaussard, E., Bürgmann, R., Shirzaei, M., Fielding, E. J., Baker, B., 2014. Predictability of hydraulic head changes and characterization of aquifer-system and fault properties from InSAR-derived ground deformation. J. Geophys. Res. Solid Earth 119, 6572–6590. https://doi.org/10.1002/2014JB011266.
Authors: Bovenga, Fabio (1); Pasquariello, Guido (1); Refice, Alberto (1); Nutricato, Raffale (2); Nitti, Davide Oscar (2); Chiaradia, Maria Teresa (2)Phase unwrapping is considered as a critical step in the InSAR and DInSAR processing. Indeed, the interferometric phase is ambiguous since its value is known modulo 2π. Hence, a robust phase unwrapping algorithm is crucial to reconstruct the elevation or the displacement information. Many approaches and algorithms were studied and developed for efficient phase reconstruction with minimum error propagation. One of the first approaches is the branch-cut algorithm which consists of unwrapping the phase following a discontinuous path that avoids going through branch-cuts. Phase jumps induce discontinuities which, once quantified, form a sparse grid of residues with positive and negative polarities. The branch-cuts map is constructed by connecting these residues so that the sum of the cuts residues is equal to zero. The branch-cuts map is used as a mask that prevents the unwrapping ‘s path from crossing the branch-cuts. This widely used method has proven its reliability. However, one of its limitations is that in areas with an important phase jumps, the branch-cuts formed can isolate some parts of the interferogram causing an inconsistency in the result since those areas remain wrapped. Further research proposed hybrid methods that aim to unwrap the branch-cuts while limiting the error propagation. Within this scope, we propose a two-steps approach that improves the result of the branch-cuts method through a first phase unwrapping guided by a quality map, but always using the mask of the cuts, and a second one that unwraps the branch-cuts progressively. In the second step, we use a technique inspired by the way flat areas, depressions and uneven terrains are processed in Digital Model Elevation. The flow from a flat area takes the path of the lowest gradient. On the other hand, in an area of depression to escape the flow we create locally flat areas from the average gradient. By analogy with this principle, our algorithm unwraps the pixels under cuts by analyzing and interpolating their unwrapped and already processed neighbors as following; i) 75% of a Pixel-Cut neighborhood is already unwrapped, ii) only the central pixel is unwrapped at each iteration, iii)The integer ambiguity assigned to the central pixel depends on the frequency of occurrence of its neighbors integers ambiguities. We tested our method on Sentinel 1 data and compared the obtained results with other commonly used phase unwrapping algorithms in most InSAR and DInSAR processing tools and show the efficiency of our algorithm with evaluation metrics. We retained that our algorithm's unwrapping is quite similar to that obtained by MCF(Snaphu) and that it presents superior results in areas with a high density of residues.
Authors: Belhadj-aissa, Saoussen; Hocine, Faiza; Daoud, Ishak; Belhadj-aissa, MostefaPolarimetric persistent scatterer interferometry (PolPSI) enhance interferograms' phase quality by adequately combining the available different polarization channels (e.g., HH, VV, HV and VH) into an improved one, for which polarimetric optimizations are exploited. The amplitude dispersion (DA) is a commonly used phase quality metric in PSI and in PolPSI optimization oriented to deterministic scatterers. The resolution of PolSAR images is assumed to influence the performance of the PolPSI optimization based on DA, but this issue has not been studied in detail yet. In this paper, the impact of SAR images' resolution in DA-based PolPSI is investigated. To this end, the Exhaustive Search Polarimetric Optimization (ESPO) algorithm has been used with PolSAR datasets with different resolutions and different polarimetric combinations. Specifically, a set of 31 original fully polarimetric RADARSAT-2 images over Barcelona with resolutions around 5 m, in both range and azimuth, is used to generate additional datasets with different spatial resolutions and polarimetric channels. A total of 25 data sets with five resolutions (i.e., 5 m, 7.5 m, 10 m, 15 m, and 20 m) and five polarimetric combinations (i.e., HH, full-pol, HH+VV, HH+VH and VV+VH) have been generated from the original dataset. DA values associated with the generated datasets have been estimated to analyze the influence of SAR images’ resolution on the performance of the polarimetric optimization, excluding the single HH channel case. Moreover, the temporal mean of the phase coherence has also been employed for completing this analysis. At the same time, to assess the impact of resolution on final PolPSI products, ground deformation results based on the 25 datasets have also been obtained. The results over Barcelona show that for all polarimetric combinations, full- or dual-pol, the improvement of polarimetric optimization on pixels' phase qualities generally decreases with the reduction of SAR images' resolution. On the other hand, for PSI applications, the polarimetric optimization can improve PS pixels' density with respect to that of HH channel for all the employed data sets. Moreover, in relative terms, this improvement is more significant as the resolution worsens. In other words, the lower the resolution the higher the improvement in pixels density. Thus, low resolution PolSAR data sets can be largely benefited of the polarimetric optimization techniques.
Authors: Zhao, Feng (1); Mallorqui, Jordi J. (2); Lopez-Sanchez, Juan M. (3)Tropospheric delays are still the main error source of satellite-based Interferometric Synthetic Aperture Radar (InSAR) mapping of the Earth’s surface movements. Recent studies have demonstrated the potential of using global atmospheric models (GAMs) to reduce InSAR tropospheric delays. However, the importance of appropriate weighting strategies in GAM-corrections has been largely ignored. Here we develop a new GAM-based tropospheric correction method by incorporating spatial stochastic models of the troposphere into the weighting strategy of the correction. The method determines the correlation between a pixel of interest and the GAM grid locations (3D) flexibly according to the spatial variabilities of the tropospheric random field, instead of subjectively using an inverse distance method or using a local spline function. Also, considering the horizontal heterogeneities of the tropospheric field, our new method estimates the integral of the tropospheric delays along the satellite line-of-sight (LOS) direction directly, instead of calculating the projected zenith-delays. The new method can be generalized to all GAMs, and we implement it with the latest ECMWF (European Center for Medium-Range Weather Forecasts) ERA5 reanalysis data in this study. We have tested the performance of the method with multiple Sentinel-1 interferograms from different geographical environments, including both mid- and high-latitude areas, mountainous and flat terrains, as well as in tropical and temperate climates. In addition, our test examples also encompass both areas with earthquake movements and volcanic unrest. Based on these examples and the results, we highlight (1) the importance of considering the tropospheric spatial variability in GAM-corrections; (2) the importance of considering the horizontal heterogeneities when estimating the LOS delays; and (3) the importance of correcting the tropospheric delays when mapping long-wavelength or small-magnitude deformation using InSAR.
Authors: Cao, Yunmeng; Jónsson, SigurjónOver the years, various satellites like ERS-1, ERS-2 and Envisat have been in use for the interferometric capability for a wide range of geophysical and environmental applications. Utilizing these Synthetic Aperture Radar (SAR) acquisitions, repeated approximately from the same point in space at different times, Interferometric SAR (InSAR) gives us the differences in path length in the scale of the carrier wavelength, due to changes in wavelength. With the launches of Sentinel-1A and 1B satellites in 2014 and 2016 respectively, the availability of SAR data from every part of the world has been increased many folds. The interferometric phase is affected by differences in propagation delays through troposphere or ionosphere in the time of SAR image acquisitions. One of the main problems while measuring surface deformation using InSAR is to correct the interferogram from the tropospheric phase delay. To make this correction, the interferometric phase is compared to the atmospheric delays derived from the ERA-Interim global meteorological model and the Global Navigation Satellite System (GNSS). However, in remote areas or countries with very less GNSS stations, correction by the ERA-Interim model becomes the only choice. The problem with using ERA-Interim model or GACOS for interferogram correction is that after the tropospheric correction, there are still some residuals left in the interferograms, which can be mainly attributed to turbulent troposphere. This is due to the spatial resolution of the atmospheric models by ERA-Interim, which are coarser than the resolution of the S‑1 interferograms. Therefore, they are not able to completely remove the turbulent tropospheric effect from the interferograms. One method to extract the features from the interferograms corrupted by noise, and to remove the turbulence noise from these interferograms, is to implement a convolutional neural network which can learn and generate the corrected version of the interferogram free from external noise. In this project, we have applied a recurrent neural network (RNN) algorithm for denoising raw InSAR time series. We have used the unwrapped differential phase derived from Sentinel-1 interferograms. We have implemented a noise aware RNN on the interferometric phase to remove the noise and detect the transient signals from the InSAR time series.
Authors: Ghosh, Binayak (1); Motagh, Mahdi (1,2); Haghighi, Mahmud Haghshenas (2); Maghsudi, Setareh (3)The Persistent Scatterer SAR Interferometry (PS-InSAR) technique, Global Navigation Satellite Systems (GNSS) networks and precise levelling provide a unique database for the detection of recent displacements of the Earth’s surface. However, each technique has its own characteristics in terms of sampling rates, spatial resolution, reference frames and processing models. We focus on the development of a fusion technique that takes advantage of the geodetic measurement methods mentioned in order to resolve and determine the small tectonic surface movements in the area of the Upper Rhine Basin and its surroundings. A main challenge of this fusion is the unification of the measured surface deformation vectors of these techniques, since InSAR measures the projection of the deformation in the line of sight. Besides the calculation of the horizontal and vertical components derived from ascending and descending data, the focus of this presentation is on the correction of atmospheric noise. In scope of this work, Sentinel-1A data from the ascending orbits 15 and descending orbit 66 in northern Switzerland is evaluated in the period April 2015 and February 2020. We provide the complete processing chain using SNAP (Sentinel Application Platform, ESA) and StaMPS (Stanford Method for Persistent Scatterers, A. Hooper et al.). The reduction of the atmospheric noise is calculated with TRAIN (Toolbox for Reducing Atmospheric InSAR Noise, D. Bekaert). We apply phase-based techniques like linear and powerlaw corrections as well as weather model-based techniques using the models GACOS (Generic Atmospheric Correction Online Service for InSAR) and ERA-Interim. As northern Switzerland has flat as well as steep terrain with elevations between 200m and 2000, this area is well suited to investigate these different atmospheric correction approaches. The calculation of the horizontal and vertical components is implemented using a kriging interpolation of the results from the Persistent Scatterer analysis of the ascending and descending orbits. The derived components are validated with the results of a GNSS network processing. The data of the evaluated GNSS permanent stations is provided by GURN (GNSS Upper Rhine Graben Network). The validation of the PS results with the GNSS results is an intermediate step of the data fusion methodology developed at the Geodetic Institute at KIT. In a next step, the data sets must be brought onto a common grid and combined by means of weightings. Thus, validation at this stage of the work helps to find the correct method for interpolation and suitable weighting parameters. The presentation will compare the different atmospheric reduction techniques and will discuss the reliability of the horizontal and vertical velocity fields considering tectonic and anthropogenic processes in the investigation area. Furthermore, the validation of the results with regard to data fusion is discussed and the methodology for data fusion is presented.
Authors: Heck, Alexandra; Westerhaus, Malte; Sumaya, Hael; Weisgerber, Jakob; Even, Markus; Kutterer, Hansjörg; Heck, BernhardIn InSAR analysis, the effect of microwave propagation delay in the Earth's atmosphere such as the neutral atmospheric delay and the ionospheric delay is recognized as the primary noise for surface deformation researches like Earthquake source modeling, tectonic fault motion, and volcanic activity monitoring. Although, for the ionospheric delay, we can now apply the range split spectrum method (SSM) to effectively mitigate it, the mitigation of the neutral atmospheric delay noise remains difficult and is the research problem to be solved. Recently, Arief and Heki (2020) developed a new method to retrieve two-dimensional Zenith Wet Delay (ZWD) distribution at sea level based on the GNSS ZWD and delay gradient derived from the Japanese GNSS network named GEONET. Here we proposed a new InSAR delay correction method based on modifying the Arief and Heki's method and applied it to the ALOS-2 ScanSAR interferograms to evaluate its effectiveness. In our study, we used 5-minute interval GNSS PPP data provided by the Nevada Geodetic Laboratory in Nevada University, Reno. Since InSAR atmospheric delay contains both hydrostatic and wet components, we estimated two-dimensional Zenith Total Delay (ZTD) distribution at sea level instead of ZWD, and we simultaneously estimated height dependence of ZTD as a linear function. The model consists of the regularly distributed grids with 5 km interval and the height dependence. The retrieval of ZTD distribution is performed by the least square inversion with the smoothing constraint. The retrieved ZTD is then projected onto the InSAR line-of-sight direction and calculated a difference of two epochs to generate an InSAR delay model. Interferograms were generated by RINC ver.0.41r using 16 ALOS-2 ScanSAR level 1.1 full-aperture data covering Kanto Plain in Japan. We applied the SSM to all of interferograms to correct the ionospheric delay noise before applying the proposed tropospheric delay correction. The result of applying proposed correction method showed that the correction effectively reduced the phase variance, especially in the long-wavelength phase variation. The phase standard deviation (STD) in the whole scene decreased from 35.95 mm to 25.84 mm by applying the proposed GNSS-based correction method. For comparing effectiveness of the proposed method with existing methods, we also calculated the phase STD derived by applying the GACOS model and the numerical weather model-based correction using the Japan Meteorological Agency's Meso-scale model data. The result of comparison showed that the proposed GNSS-based method most reduced the whole-scene phase STD. The GACOS model decreased the STD to 30.96 mm, and the JMA MSM decrease to 27.71 mm, respectively. We then calculate the distance-dependence of the phase STD based on the variogram model. The variogram derived from all the interferograms showed the superiority of the GNSS-based correction, although the STD in distance shorter than 20 km seemed no differences between correction methods.
Authors: Kinoshita, YoheiA major source of error for spaceborne Interferometric Synthetic Aperture Radar (InSAR) is the phase delay in radio signal propagation through the troposphere. To overcome this, we released the first generation of Generic Atmospheric Correction Online Service (GACOS 1.0) in the FRINGE workshop in Helsinki, Finland on June 2017. GACOS utilises operational high resolution ECMWF data (0.125 degree grid, 137 vertical levels, 6-hour interval) with an iterative tropospheric decomposition model and its applications to globally-distributed interferograms, encompassing both flat and mountainous topographies, mid-latitude and near polar regions, monsoon and oceanic climate systems, suggests a phase precision and displacement accuracy of approximately 1 cm for the corrected interferograms. GACOS has the following notable features: (i) global coverage, (ii) all-weather, all-time useability, (iii) correction maps available in near real-time, and (iv) indicators to assess the correction performance and feasibility. To further improve the performance of GACOS to better serve the InSAR community, a new generation (GACOS 2.0) has been being developed by: (i) improving the temporal resolution by integrating the newly published 1-hour ERA-5 weather model and the 5-minute GPS tropospheric delay estimates; (ii) developing an API system to facilitate automatic data processing; and (iii) enhancing GACOS based on regional/local datasets (such as national weather model and regional GPS network). Based on the globally distributed GPS network and the MODIS PWV product, the performance of GACOS 2.0 at different regions of the world is evaluated with its elevation and latitude dependency being concluded which could be served as a new performance indicator. All these features will contribute to a simplified time series analysis method (i.e. relying less on spatial temporal filters) to reduce the computational burden, provided that the majority of the atmospheric error has been mitigated by GACOS 2.0.
Authors: Li, Zhenhong (1,2); Yu, Chen (2)InSAR has the ability to provide a high spatial resolution measurement (few meters’ spacing) of how the Earth's surface is deforming; however, its retrieval accuracy is strongly limited by propagation delays through the ionosphere and troposphere. Variable degrees of success have been observed when correcting for propagation delays using weather and reanalysis models. Ongoing developments on assimilating (e.g., 3D/4D-VAR) and new higher spatial-temporal remote sensing observations (e.g. InSAR, radiometers, ground based radar, and spatially-dense GNSS networks) are enabling the atmospheric community to make significant strides towards improved high spatial-temporal resolution weather model forecasting and reanalysis products. For example, the High-Resolution Rapid Refresh (HRRR) weather model for North America, which assimilates dense ground-based RADAR every 15 min, has a spatial model resolution of 3 km, versus ECMWF-operational model at 9km and ERA-5 at 31 km. Assimilation of high-resolution observations into a high resolution model allows for capturing more complex processes. One should also consider inaccuracies in the weather model simulations; however, it is to be expected that these will keep on improving over time. Space-borne sensors with large incidence angles (such as Sentinel-1 and NISAR) will span 3-5 HRRR resolution cells versus 1 cell for ERA-5. Current state-of-the-art InSAR tropospheric correction packages are not well suited to take advantage of models under 10 km spatial resolution as radar slant delays are computed by projecting zenith delays into the radar line-of-sight. We introduce RAiDER (“Raytracing Atmospheric Delay Estimation for RADAR”), a state-of-the-art open-source troposphere correction tool for Interferometric Synthetic Aperture Radar (InSAR). RAiDER uses ray-tracing in the line-of-sight direction to directly obtain the slant delay, leveraging ongoing improvements in spatial-temporal sampling and accuracy of weather models. It is available as an open-source Python package on GitHub (https://github.com/dbekaert/RAiDER), and includes command-line tools for generating tropospheric delay estimates, correcting interferograms, and statistically analyzing the results. It adopts standardized data formatting and design optimized for storage, following NISAR’s data convention to define delays on a 3D cube, with additional options for delay results output on a 2D raster or at specified station locations. We present preliminary results using RAiDER to compare ERA-5 and HRRR weather models, including comparison against GNSS-derived tropospheric delays and a single interferogram containing primarily atmospheric noise.
Authors: Maurer, Jeremy (1,2); Bekaert, David (2); Sangha, Sim (2); Fattahi, Heresh (2)This work presents the results of an experiment aiming to the study of problem of atmospheric phase delay effects on the estimation of soil moisture by means of SAR interferometry. The rationale of the experiment is that if we properly model the spatial variability of the water vapor and its contribution to the interferometric phase, the remaining interferometric component can be related mostly to soil moisture variations, when no displacements occur in the scene. To simplify the experimental conditions, we selected a flat area to avoid the artifacts due to topography and a bare soil parcel to avoid volumetric surface scattering. The experiment was carried out in a farm located approximately 20 km east of Lisbon, Portugal, close to the Tagus River estuary. A set of five soil moisture sensors was deployed and set to record soil moisture in an hourly basis providing a 3-month long times series of in-situ measurements. A set of SAR images were used, consisting of 17 C-band Sentinel-1 A/B images, acquired between 10 January 2019 and 16 April 2019. The time series of SAR images were interferometrically processed using the set of short temporal baseline interferograms with a time difference between master and slave images of 6 days. With a short temporal baseline the surface deformation contribution is minimized, remaining the temporal changes in the spatial distribution of troposphere water vapor and the soil moisture changes. In this study the atmospheric phase delay (APD) was mitigated using the approach proposed by Mateus et al, 2020 [1] in which the atmospheric phase delay is estimated from a time series of phase measurements calibrated with a minimum of four GNSS stations. The interferometric phase was calibrated with a nearby persistent scatter and the atmospheric path delay contribution removed resulting in a residual phase. The power spectral density of the interferometric phase before (InSAR phase) and after the atmospheric effects’ mitigation (Res phase), as well as those obtained by the ERA5 model (ERA5 phase) and InSAR derived path delay (APD) are presented in Figure 1. Both INSAR phase and APD phase have the same energy up to spatial frequencies lower than 6e-4 (m-1) (wavelength greater than 1.6 km). The residual phase has less energy in the longer wavelengths and the same energy in the shorter wavelengths. This can be explained by the fact that the long wavelength contribution from the atmospheric path delay was removed and at least mitigated from the interferometric phase. Results based on C-band Sentinel-1 data pointed out a clear relationship between temporal changes of interferometric phase, after mitigation of atmospheric propagation effects, and soil moisture. The R2 coefficient from the regression between soil moisture and the phase has improved from 0.14 to 0.66, when the atmospheric path delay contribution is removed from the interferometric phase. Both residual phase and coherence information were inverted to estimate the soil moisture using the model proposed by De Zan (2014) [2]. The reference soil moisture for the first date is the soil moisture measured by the in-situ sensor. The estimated soil moisture for the entire area is shown in Figure 2. The map shows with high spatial resolution detailed soil moisture changes across the agricultural parcels. Keywords: Soil Moisture, Sentinel-1, Water Vapor in Atmosphere, Satellite Interferometry, Power Spectrum Analysis References: [1] P. Mateus, J. Catalão, G. Nico, e P. Benevides, «Mapping Precipitable Water Vapor Time Series From Sentinel-1 Interferometric SAR», IEEE Transactions on Geoscience and Remote Sensing, vol. 58, n. 2, pp. 1373–1379, Fev. 2020, doi: 10.1109/TGRS.2019.2946077. [2] F. De Zan, A. Parizzi, P. Prats-Iraola, e P. López-Dekker, «A SAR Interferometric Model for Soil Moisture», IEEE Transactions on Geoscience and Remote Sensing, vol. 52, n. 1, pp. 418–425, Jan. 2014, doi: 10.1109/TGRS.2013.2241069.
Authors: Mira, Nuno (1,2); Catalão, João (2); Nico, Giovanni (3); Mateus, Pedro (2)Microwave propagation delay caused by temporal and spatial differences in dry and wet atmospheric conditions, and redistribution of atmospheric liquid contents and ionospheric electron density lead to distort the land deformation signal measured by Interferometry. Nowadays, despite of an unprecedented increase in possibility to use InSAR for land deformation measuring, still applying atmospheric and ionospheric corrections, which need auxiliary datasets, appear as an obstacle for the fast and on-time use of SAR Interferometry. A set of knowledge on atmospheric and ionospheric delay patterns would help us to foresight how these delays would distort our estimations from InSAR. At this point, Interferometry applicants can specify the most influencing factors on InSAR processing within their studies. By ignoring unimportant factors or make approximation for them, one can apply a timely and opportune process of InSAR for land deformation measuring. Depending on the acquisition time and location of used InSAR data, the atmospheric and ionospheric factors affect the Interferometry process in different ways with different magnitudes. The atmospheric signal, particularly in the neighboring of large water bodies, may have significant impacts on InSAR measurements and therefore need to be investigated carefully. In this study, we have investigated the impacts of atmospheric and ionospheric factors over the detection of multiscale land deformations. Using five Sentinel-1A single look complex images, acquired at span of 4.5 months, in 2019 and 2020, we studied two distinct regions of study in the eastern coast of the Caspian Sea, Centered at 53.15E and 40.25N (Fig.1). For preventing the land cover and land use related noises, we choose a low populated area with rare human constructions and free of forest and vegetation covers. Despite of small temporal baselines, it is expected that in the both study regions, the InSAR pairs include a large scale land vertical deformation inducing by the Caspian Sea water loading. However, in one of the region, tectonic land deformations are also expected to be caused by an earthquake of 4.9 Richter degree magnitudes occurred on 6 Jan 2020. Utilizing the InSAR images, seven interferograms are generated and vertical land deformations are extracted for the region of studies. Accordingly, for the local time of image pairs, atmospheric and ionospheric delays are calculated and the equivalent vertical displacements are subtracted from InSAR derived land deformation. We calculate dry and wet atmospheric delays from vertical temperature and humidity profiles. For this purpose, we utilized hourly ERA5 reanalysis. Using reanalysis of specific cloud, rain and snow water contents, we also calculated delays caused by liquid particles within could, rain and snow. In this context, the specific rain and snow water contents are compared to half-hourly precipitation records from the Global Precipitation Model (GPM) which provides a better time resolution. We corrected for ionospheric delay with analyzing hourly IONEX dataset from the GNSS data analysis center. Our results show that, the deformations caused by the earthquake are detectable before dry any atmospheric and ionospheric corrections are applied to the inteferometry results. However, the large scale loading effect enhanced only after ionospheric delay is applied. Filtering small scale deformation and outliers helped to detect west-east sloped deformation induced by water loading. In order to investigate long term spatial and temporal patterns we calculated and analyzed four years time series (Jan 2017 to Dec 2020) of dry and wet atmospheric, and ionospheric delays in different latitudes and longitudes. Results show that, dry atmospheric delay has large spatial (order of tens km) and temporal (order of weeks) wavelengths. Its spatial and temporal patterns are mainly determined by surface topography and air temperature respectively. However, wet atmospheric delay and liquid delay may rapidly vary both spatially and temporally. The integrated liquid delays are small and very sparse. Although, the dry atmospheric delay is about 8 times greater than wet delay, however, due to large variability it is the wet atmospheric delay that principally controls the atmospheric delay. Based on our long term analysis wet delay conforms in average about 80 percent of time variability of atmospheric delay. Dry and wet atmospheric delays show inverse seasonal trends with the maximum in winter and summer for dry and wet delays respectively (Fig.2). Ionospheric delay shows a very large spatial wavelength in the order of hundreds kilometers. As it depends on the sun and moon angles, the stationary part of ionospheric delay strongly but regularly varies in time and by the geographical latitudes. Thus, ionospheric delay shows one diurnal and, depending on the latitude, one or two seasonal cycles. The results show that the annual cycle of ionospheric delay include two seasonal cycles in Equator, switching to a single cycle in higher latitudes. According to season, the sign of trends are reverse in the northern and southern hemispheres (Fig.3). Four years time series show that the amplitude of stationary component in both diurnal and seasonal cycle in the region of study is about 6 TEC (Total Electron Content) units (Fig.4). In order to evaluate the results we used GPS records of total zenith path delay (ZPD) obtained by GNSS data center. This GPS network provides very high resolution records (5 minutes interval), but it is limited to very sparsely distributed ground stations which are located relatively far from our study regions. Although, GPS delay data provide low resolution for application in InSAR derived measures; however, in the absence of other insitu measurements we benefited using GNSS ZPD records. The data from three GPS stations of Atyrau, Tehran and Aruch are compared to the total delay derived from REA-5 profiles, for the same time and locations. Comparison between our estimation of total delays with GPS zenith delays showed a mean RMS value of 17 mm. Conclusions: Depending on the spatial scale of the land deformation which InSAR applicants expect to detect, each of dry/wet atmospheric and ionospheric factors could have a particular importance to be mitigated or ignored. In small land deformation scales, such as landsides and small earthquakes, surface deformations are locally limited. In such cases, the global land deformation measured in a Sentinel scene is due to large scale delays. Subtracting the global component of land deformation lets to map the local deformations. Therefore, for small land deformation scales, only wet atmospheric and liquid delays would be enough to ensure the InSAR measurement accuracy. In the other side, for large land deformation scales, such as surface loading or isostatic adjustment, the impact of large scale factors such as dry atmospheric and ionospheric delays should be prioritized. In such cases, liquid delay due to sparse cloud parcels or local precipitation can be filtered. In the case where small and large scale deformations are needed to be detected, more complicated contributions from atmospheric and ionospheric effects could lead to mismeasurements from InSAR. In such cases, all delay components should be considered.
Authors: Moradi, AyoubSynthetic Aperture Radar (SAR) is an imaging technique in which the earth’s surface backscattering can be characterized spatially. The backscatter information in terms of the scattering matrix is decomposed to get the account of the scattering mechanism shown by the target(s). However, de-oriented targets with radar line of sight may distort the polarization information and affect the extracted parameter of interest. The de-oriented scattering matrix has changed intensity values. Therefore, the process of de-orientation or POA compensation can also induce a change in the higher-order statistical products of PolSAR data. Intrinsically, POA is defined as the angle between the major axis and X-axis of the polarization ellipse of the polarized wave. There is another factor which changes the polarization basis of the transmitted wave, called Faraday Rotation. It is one of the polarimetric distortion factors and depends on the characteristic of the ionosphere. The scattering matrix is rotated by Faraday Rotation Angle (FRA) to compensate for the effect of Faraday Rotation. However, the scattering response including the distortions, from the same target at a different time is also different. For the same target if the acquisition is differed by a spatial or temporal baseline, then both polarimetric and interferometric information of the target can be represented by the PolInSAR cross-correlation matrix. PolInSAR data is derived from vector interferometry technique using polarization state as a projection vector. This complex data is quantified into meaningful information in terms of PolInSAR coherence. This coherence is also the correlation between scattering response from the same target with a different spatial or temporal baseline of acquisition. Since PolInSAR coherence is polarization-dependent, therefore, the temporal change in polarimetric distortion can be analyzed using PolInSAR data, which is the aim of the present study. Apparently, Faraday rotation and POA shift can affect the PolInSAR data and cause a change in the received backscatter intensities. To quantify the temporal change of polarimetric distortion, a comparative analysis has been made to analyze the effect of Faraday rotation and POA compensation on PolInSAR coherence estimation. Further analysis has been done to quantify the effect of Faraday rotation on POA estimation. For the present study, RADARSAT-2 PolInSAR data Dehradun region, Uttrakhand, India has been used. The repeat-pass data has a temporal baseline of 24 days. The results show that Faraday rotation has no significant effect on the estimation of PolInSAR coherence. Since POA shift is the distortion induced due to scatterer, more variation in the coherence estimates has been observed which signifies the sensitivity of coherence with the physical characteristics of the target. POA compensation has significantly changed the coherence estimates. The derived coherences have increased values irrespective of the projection vector used. However, the change in HH and VV channel coherence is more as compared to HV channel coherence. Furthermore, the results also show that Faraday rotation has a little or negligible effect on POA estimation.
Authors: Shafai, Shahid Shuja (1); Kumar, Shashi (2); Aghababaei, Hossein (1); Kulshrestha, Anurag (1)An adaptive fusion of multiple InSAR tropospheric delay correction methods Li Zhanga , Jie Dongb , Lu Zhanga , Mingsheng Liaoa a State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079 China b School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079 China Atmospheric propagation phase delay is still the main challenge in monitoring surface deformation using space-borne Synthetic Aperture Radar Interferometry (InSAR). The tropospheric atmospheric delay often contaminates the interesting deformation signal, making the deformation interpretation harder. The empirical phase-elevation models (Wicks et al. 2002, Bekaert, Hooper and Wright 2015), meteorological reanalysis models (Doin et al. 2009, Jolivet et al. 2011, Jolivet et al. 2014), numerical weather forecast models (Gong et al. 2010), and space-based multispectral data (Li, Muller J.P and P 2005, Li et al. 2006, Li et al. 2012) generally are used to correct tropospheric delay under various circumstances. However, these tropospheric delay correction methods present uneven effects due to the different data parameters, processing algorithms, and external conditions. In this study, we proposed an adaptive fusion of tropospheric delay estimates derived from ERA5, GACOS, WRF, MERRA-2, NARR, MODIS, Linear, and Powerlaw to achieve a more accurate and reliable tropospheric delay correction. The spatially varying scaling algorithm (Lin, Andrew and John 2019) was employed to deal with the spatially lateral variation of tropospheric delay. This fusion method was integrated into the standard Small Baseline Subset (SBAS) time series analysis procedure. To evaluate the performance of the fusion method, the Los Angeles of Southern California was selected as the study area. We first used one ENVISAT ASAR data stack from 2007 to 2009 to assess the reliability of the proposed fusion method through comparing with the homochromous MERIS measurements, and then extended it to ascending and descending Sentinel-1 data stacks from 2015 to 2018. The result of Sentinel-1 datasets showed that over 80% of the interferograms in both the ascending and descending tracks have the smallest residual phase Root Mean Square (RMS) value after correction by the fusion method, compared to any single tropospheric correction method. The deformation time series InSAR of Sentinel-1 data after fusion correction was validated against the GPS measurements with 125 GPS SCIGN stations. We found that the Root Mean Square Error (RMSE) of the original time series is reduced by more than 60% with the average exceeding 20% after fusion correction. Overall, the performance of fusion method outperformed other tropospheric delay correction methods. Moreover, contributed by the spatial changing scaling algorithm, the fusion method alleviates not only the stratified delay, but also the turbulent mixing delay. Reference: Bekaert, D. P. S., A. Hooper & T. J. Wright (2015) A spatially variable power law tropospheric correction technique for InSAR data. Journal of Geophysical Research: Solid Earth, 120, 1345–1356. Doin, M. P., C. Lasserre, G. Peltzer, O. Cavalié & C. Doubre (2009) Corrections of stratified tropospheric delays in SAR interferometry: Validation with global atmospheric models. Journal of Applied Geophysics, 69, 35-50. Gong, W., F. Meyer, P. W. Webley, D. Morton & S. Liu. 2010. Performance analysis of atmospheric correction in InSAR data based on the Weather Research and Forecasting Model (WRF). In 2010 IEEE International Geoscience and Remote Sensing Symposium, 2900-2903. Jolivet, R., P. S. Agram, N. Y. RE Lin, M. Simons, M.-P. Doin, G. Peltzer & Z. Li (2014) Improving InSAR geodesy using Global Atmospheric Models. Journal of Geophysical Research: Solid Earth, 119, 2324-2341. Jolivet, R., R. Grandin, C. Lasserre, M. P. Doin & G. Peltzer (2011) Systematic InSAR tropospheric phase delay corrections from global meteorological reanalysis data. Geophysical Research Letters, 38. Li, Z., E. J. Fielding, P. Cross & J.-P. Muller (2006) Interferometric synthetic aperture radar atmospheric correction: Medium Resolution Imaging Spectrometer and Advanced Synthetic Aperture Radar integration. Geophysical Research Letters, 33, 272–288. Li, Z., Muller J.P & C. P (2005) Interferometric synthetic aperture radar (InSAR) atmospheric correction: GPS, Moderate Resolution Imaging Spectroradiometer (MODIS), and InSAR integration. Journal of Geophysical Research: Solid Earth, 110. Li, Z., P. Pasquali, A. Cantone, A. Singleton, G. Funning & D. Forrest (2012) MERIS Atmospheric Water Vapor Correction Model for Wide Swath Interferometric Synthetic Aperture Radar. IEEE Geoscience and Remote Sensing Letters, 9, 257-261. Lin, S., H. Andrew & E. John (2019) A Spatially Varying Scaling Method for InSAR Tropospheric Corrections Using a High‐Resolution Weather Model. Journal of Geophysical Research: Solid Earth, 124, 4051-4068. Wicks, C. W., D. Dzurisin, S. E. Ingebritsen, W. Thatcher, Z. Lu & J. Iverson (2002) Magmatic activity beneath the quiescent Three Sisters volcanic center, central Oregon Cascade Range, USA. Geophysical Research Letters, 29.
Authors: Zhang, Li (1); Dong, Jie (2); Zhang, Lu (1); Liao, Mingsheng (1)The advanced Differential Synthetic Aperture Radar (SAR) Interferometry (DInSAR) [1] technique referred to as Small BAseline Subset (SBAS) algorithm [2], [3] has deeply proven its effectiveness to perform surface displacement analyses in both natural and anthropic hazard scenarios [4]-[6], thanks to the well-known capability to generate mean deformation velocity maps and displacement time series over wide areas with sub-centimeter accuracy. One key point of the SBAS approach, which contributed to its widespread use throughout the DInSAR community, is the capability to perform different spatial scales interferometric analyses, referred to regional and local ones [7], with various spatial resolutions [8]. This is accomplished on the one hand by dealing with multi-look (low resolution) interferograms (regional scale analysis with a typical resolution of about 30-90 m), capable to perform extended spatial scale DInSAR analyses relevant to a wide range of natural deformation phenomena (mainly volcanic and seismic events); on the other hand, it is possible to work with full resolution interferograms generated from SLC data (local scale analysis with a spatial resolution of about 3-10 m), in order to zoom in areas where localized deformations occur by benefiting from the very high resolution characteristics of the SAR acquisitions. This latter is particularly suitable to retrieve the space-time characteristics of deformation phenomena in anthropogenic contexts (extended urban areas, historical and archaeological sites, oil-gas extraction, transportation infrastructures), with reasonable computational requirements. These algorithmic advancements have paved the way to the concurrent development of new SAR sensors and satellite missions, characterized by different frequency bands, revisit times and ground coverage, which have been collecting hundreds of thousands of full resolution SAR data over the past two decades. Such a large availability of huge, long-term SAR data stacks, each of them composed of hundreds of full resolution SAR images, requires to face some issues related to their storage, handling, processing and management. For instance, the increase of the available SAR datasets (in terms of number and size of the images) implies an exponential growth of the processing time needed to achieve accurate displacement time series, with a drastic increase of data processing load and complexity that must be properly taken into account. A successful solution has already been adopted for the regional scale interferometric analysis through the Parallel SBAS (P-SBAS) processing chain [9]-[11], which consists in an automatic and parallel implementation of the low resolution SBAS workflow, based on the exploitation of multi-core and multi-thread parallel computing strategies. Unfortunately, the size of the full resolution SBAS datasets, which generally comprise hundreds of DInSAR interferometric pairs and hundreds of millions pixels to process at the sensor full spatial resolution, is about two orders of magnitude greater than that of the corresponding low resolution ones. Concerning the overall computation time and data load, this represents a serious bottleneck if addressed with a conventional (i.e. sequential) implementation approach. Accordingly, to speed up the overall full resolution SBAS processing chain at reasonable time frames (e.g. less than 24 hours) and achieve high efficiency in terms of scalability and computing performances, the implementation of more advanced solutions of parallel computing, based on the exploitation of advanced ICT distributed computing environments and on the development of much more advanced DInSAR methodologies for maximizing the information related to these huge amount of DInSAR data, are recommended and foreseen. In this framework, a great boost may come from the exploitation of the Graphical Processing Unit (GPU) technology, which is specifically designed to accelerate deep learning, analytics, and engineering applications [12]. Indeed, a GPU has a massive parallel architecture consisting of thousands of smaller but more efficient cores, designed for handling multiple tasks simultaneously, whereas a CPU comprises a few cores optimized to speed up the sequential processing. In this work, we present an advanced parallel implementation of the full resolution SBAS approach (full resolution P-SBAS) based on the exploitation of both multi-core/CPU and GPU parallel processing strategies, aimed at generating displacement time series and corresponding deformation velocity maps related to extended urban areas but at the scale of single buildings and infrastructures. Concerning the multi-core strategy, similarly to [9], the implemented full resolution P-SBAS processing chain exploits parallel CPUs for the steps with image granularity (SAR acquisitions or interferograms) or when spatial analyses are involved. Indeed, in the multi-core approach, each step computed on a single spatial layer can be treated independently, leading to a set of independent computations concurrently running on different data, which can be naturally distributed to multiple cores. This kind of processing involves a limited number of parallel jobs (some hundreds), needs large availability of memory (from a few tens to hundred GBytes) and implements complex operations on large data (e.g., spatial filtering of full resolution interferograms, oversampling and under sampling operations). On the other hand, the GPU parallel processing is particularly suitable for the processing steps characterized by pixel granularity (e.g., phase unwrapping operation and SVD method). In particular, it deals with the independent processing of single pixels, starting from a common data structure, within the different interferograms (temporal analysis), leading to a set of independent computations concurrently running on the several cores of the GPU. This kind of processing requires a very large number of parallel jobs (up to hundreds of millions), needs small memory for each job (a few KBytes) and involves simple operations on small parts of the full resolution data (mainly matrix and vector multiplications). One major bottleneck to face when dealing with CPU-GPU processing is represented by the lower speed of the host memory (CPU) compared to that of the GPU. Accordingly, to obtain good computing performances, we re-engineered some steps of the full resolution SBAS workflow, in order to minimize the data transfer between the host and GPU memories of each module, leaving on GPU memory intermediate results. The implemented processing chain, designed to effectively handle large datasets of Stripmap SAR data acquired by the X-band COSMO-SkyMed (CSK) constellation, can be easily adapted also for Sentinel-1 data acquired in the TOPS mode. It has been applied to a large number of full resolution ascending and descending CSK data stacks (spatial resolution around 3x3 meters), which have been acquired over extended Italian urban areas in the stripmap mode within the framework of the MAP Italy project and made available by the Italian Space Agency. In particular, to assess the computing performances of the proposed approach, we carry out an experimental analysis by processing a sequence of 129 ascending CSK images acquired over the extended Roma (Italy) urban area during the March 2009 – March 2019 time interval through the Amazon Web Service (AWS) environment. In particular, on AWS instances we benefit from Virtual Machines (VMs) equipped with 1 to 8 GPUs NVIDIA Tesla V100. This availability allows us to carry out an extensive performance analysis on the scalability and computing time of the developed Full Resolution P-SBAS processing chain. The preliminary achieved results show that the proposed parallel full resolution implementation can process a full CSK frame (40x40 km2) dataset in less than 12 hours. References [1] Massonnet, D. et al., The displacement field of the Landers earthquake mapped by radar interferometry, Nature, vol. 364, no.6433, pp. 138– 142, Jul. 1993. [2] Berardino, P., Fornaro, G., Lanari, R., and Sansosti, E., A new Algorithm for Surface Deformation Monitoring based on Small Baseline Differential SAR Interferograms. IEEE Trans. Geosci. Remote Sens, 40, pp.2375-2383, 2002 [3] Lanari, R., Mora, O., Manunta, M., Mallorquí, J.J., Berardino, P., and Sansosti, E., A small baseline approach for investigating deformations on full resolution differential SAR interferograms. IEEE Trans. Geosci. Remote Sens., 42, 1377-1386, 2004 [4] Lanari, R., Casu, F., Manzo, M., Zeni, G., Berardino, P., Manunta, M., Pepe, A., An Overview of the Small BAseline Subset Algorithm: a DInSAR Technique for Surface Deformation Analysis, Pure and Applied Geophysics, 164, 4, pp. 637-661, 2007 [5] Sansosti, E., Berardino, P., Bonano, M., Calo`, F., Castaldo, R., Casu, F., Manunta, M., Manzo, M., Pepe, A., Pepe, S., Solaro, G., Tizzani, P., Zeni, G., and Lanari, R., How second generation SAR systems are impacting the analysis of ground deformation. Int J Appl Earth Obs Geoinf 28:1–11. doi:10.1016/j.jag.2013.10.007, 2014 [6] Bonano, M., Manunta, M., Pepe, A., Paglia, L., and Lanari, R., From previous C-Band to New X-Band SAR systems: assessment of the DInSAR mapping improvement for deformation time-series retrieval in urban areas. IEEE Trans Geosci Remote Sens 51 (4):1973–1984, 2013 [7] Bonano, M., Manunta, M., Marsella, M., and Lanari, R., Long-term ERS/ENVISAT deformation time-series generation at full spatial resolution via the extended SBAS technique. Int J Remote Sens 33:4756–4783, 2012 [8] Solari, L., Ciampalini, A., Raspini, F., Bianchini, S., Zinno, I., Bonano, M., Manunta, M., Moretti, S. and Casagli, N., Combined use of C-and X-Band SAR data for subsidence monitoring in an urban area. Geosciences 7:21. doi: 10.3390/geosciences7020021, 2017 [9] Casu et al., SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., 7,8, 3285–3296, 2014 [10] Zinno, et al., National scale surface deformation time series generation through advanced DInSAR processing of sentinel-1 data within a cloud computing environment, IEEE Trans. Big Data, doi: 10.1109/TBDATA.2018.2863558, 2018 [11] Manunta M., et al., The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment, IEEE Trans. Geosci. Remote Sens., 57, 9, 2019, doi: 10.1109/TGRS.2019.2904912., 2019 Corporation, NVIDIA. nvidia. URL: www.nvidia.com
Authors: Bonano, Manuela (1); Buonanno, Sabatino (1); Lanari, Riccardo (1); Yasir, Muhammad (1,2); Zinno, Ivana (1); Manunta, Michele (1)LiCSAR is an operational system built for large-scale interferometric processing of Sentinel-1 data, developed within the Centre for Observation and Modelling of Earthquakes, Volcanoes and Tectonics (COMET). Utilising Centre for Environmental Data Analysis (CEDA) JASMIN computing facility, LiCSAR automatically produces geocoded wrapped and unwrapped interferograms and coherence estimates, for large regions globally, and processing can be prioritised following an earthquake or during volcanic crises. LiCSAR products are open and freely accessible. As of March 2021, more than 407,000 interferometric (InSAR) pairs have been generated by processing 127,000 Sentinel-1 acquisitions in the LiCSAR system, prioritising areas of the Alpine-Himalayan tectonic belt, the East African Rift, and global volcanoes. The InSAR products of 0.001° resolution (WGS-84) are available for download at https://comet.nerc.ac.uk/comet-lics-portal. The dataset is increasing by ~10-15,000 interferograms per month. The products are being ingested also into the European Plate Observing System (EPOS, https://ics-c.epos-ip.org) and the UK’s CEDA Archive (http://archive.ceda.ac.uk). The processing solutions of LiCSAR are constructed with a (MySQL) metadata database at its core. The data processing chain uses functionality of GAMMA software, with various extensions from relevant open-source tools, e.g. GMT for visualisation and solid Earth tide correction. We have also developed LiCSBAS (https://github.com/yumorishita/LiCSBAS), an open-source tool for InSAR time series and velocity analysis, which works directly with the LiCSAR products. We provide pre-processed data from COMET’s GACOS (http://ceg-research.ncl.ac.uk/v2/gacos) and include tools for their basic processing to partially correct InSAR signal for atmospheric phase screens. The LiCSAR Event responder generates interferograms following earthquakes in a rapid mode and ensures prioritised monitoring during volcanic crises. The mass production of interferometric products in easy-to-use formats, their ingestion into standard archives, and the availability of post-processing tools are aimed at accelerating the uptake of InSAR data by non-expert users. In this contribution, we will present an overview of the products, algorithms, processing, storage and event trigger solutions implemented in LiCSAR and present the LiCSBAS processing tool. We have developed a specific solution for unique burst identification, enabling us to form custom units (“frames”). Dimensions of a standard frame is approx. 220x250 km. We routinely generate interferometric products from 3-8 short-time connections per acquisition date (“epoch”). Additionally, we are generating frame-based InSAR time series (primarily using NSBAS inversion implemented within LiCSBAS), maps of tectonic strain, and developing volcanic alert systems, and other monitoring tools that work directly with LiCSAR products. We will show several case studies that use LiCSAR products to measure tectonic and volcanic deformation, including response to latest seismic and volcanic events. We have developed an expanding catalogue of earthquake deformation using the LiCSAR processing system. Following magnitude 5.5+ shallow continental and larger subduction zone earthquakes, we create an individual event webpage that indicates which pending data will be processed. Interferometric products are then served through a web portal to end users and form a long-term catalogue of the coseismic deformation for these events. The LiCSAR Earthquake Data Provider is triggered by USGS alerts and acts in near-real time, with latencies from when data has been ingested by ESA of a few hours. Depending upon the size of the event, the processing of postseismic data continues for weeks to months following the mainshock. There are 140 triggered events covering the past year (2020), with new events continually being added.
Authors: Lazecky, Milan (1); Maghsoudi, Yasser (1); Morishita, Yu (1,2); Albino, Fabien (3); Juncu, Daniel (1); Elliott, John (1); Spaans, Karsten (4); Gonzalez, Pablo (5,6); Greenall, Nicholas (7); Hatton, Emma (1); McDougall, Alistair (7); Walters, Richard (8); Watson, Scott (1); Weiss, Jonathan (1,9); Hooper, Andrew (1); Wright, Tim (1)The Aosta Valley Autonomous Region (Regione Autonoma Valle d'Aosta), since 2010 has acquired the interferometric radar data of the RADARSAT platform for the period 2003-2010 (subsequently processed with the SqueeSARTM technique) and the PS processed by ERS satellite of the period 1992-2001. The processing of these InSAR data has been combined and integrated by aerial photointerpretation, allowing to update the regional section of the Inventory of Italian Landslide Phenomena (IFFI). Thanks to the new Sentinel-1 missions and in co-operation with the University of Florence Earth Science Department, since 2018, a continuous PS monitoring service has been implemented throughout the entire regional territory, with a scheduled time resolution of 14 days. The PS monitoring service has enabled the identification of new active deformation areas in some of which the Aosta Valley Geological Survey has activated instrumental field follow-up, adopting integrated solutions for the calibration and validation of the satellite data (remote monitoring) with ground data (contact monitoring ). Furthermore, the availability of satellite data has made it possible to implement a new procedure which, starting from the screening of the territory, also known as PS mapping and PS monitoring, (Casagli et al. 2016), allows to quantitatively identify active deformation areas. Based on the internal procedure of the Geological Service, which is under validation during 2020, the identification of sectors in motion and acceleration can be followed by an upgrade to an higher level of attention towards the active area, which requires field investigations and more sophisticated monitoring (e.g.: drone survey, GNSS measurements, simple instrumentation installation, geological and geotechnical survey, etc.). Among these activities, an analysis was carried out at the Allesaz test site (Challand-Saint-Anselme) using the TRE-Altamira FASTVEL service in the Geohazards TEP platform, combined with the discontinuous on-site monitoring data provided by a network of GNSS points on the ground. The hamlet of Allesaz in the Eastern Aosta Valley is located on a terrace of fluvio-glacial origin. The site structural setting was altered by the Late Pleistocenic-Early Olocenic gravitational processes that followed the post-glacial decompression. The actual morphology is actually being shaped by the erosional processed caused by the River Evançon and by the detrital-torrential dynamics by the side-tributary Allesaz creek. The Allesaz site is described in the Regional Disaster Cadastre as "complex landslide". Several comparative tests were performed varying the coherence value (0.2, 0.3 and 0.4), in order to identify which value is most suitable to be used for any future needs in an emergency situation (e.g. Fontainemore slope). The first results of the GNSS measures are consistent with the information coming from the PS processing. A correct calibration of the system integrated in the first phase of experimentation will allow to temporarily delay onsite measures and for better programming of human and economic resources. Moreover the comparison between the PSinSAR monitoring datasets provided by the Regional PSMonitoring Service and the data from the FASTVEL service provided by TEP show a good consistence in the identification of the most active deformation areas on the investigated site. These outcomes highlight that the new InSAR services embedded in the TEP platform are now ready to be used for an operative application in the ground motion assessment in support to the civil protection authorities.
Authors: Thuegaz, Patrick; Bertolo, Davide; Stra, MichelThe Geohazard Exploitation Platform (GEP) is a web-based platform developed by the European Space Agency (ESA) that enables the exploitation of satellite Earth Observation (EO) data for geohazards analysis. GEP provides several tools and processing services, including radar interferometry (InSAR) that is an effective technique to monitor geological processes such as volcanism, land subsidence or landslides. The Sentinel-1 CNR-IREA SBAS service is one of the GEP InSAR thematic apps that consist on a processing chain for the generation of Earth time series of displacement and mean velocity maps of displacement. In this work, we present two study cases, regarding landslide detection and monitoring, for which GEP has been a great support for our research. The present work has been developed in the framework of the RISKCOAST project (Ref: SOE3/P4/E0868), founded by the Interreg SUDOE program. The first study case is referred to the Rules Reservoir, located in Southern Spain, where unstable slopes (i.e. landslides) represented a particular challenge for the reservoir construction and at present time, for its management (Reyes-Carmona et al. 2020). We applied DInSAR (Differential Interferometric Synthetic Aperture Radar) by exploiting Sentinel-1 (A and B) images to derive velocity maps and Time Series of displacement (TS) of the reservoir area. For such analysis, we used a Persistent Scatterer Interferometry chain of the Geomatics Division (PSIG) of the Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) (see Devanthéry et al. 2014 for further details). Thus, we processed ascending orbit images and we detected three landslides within the reservoir slopes that represent different potential hazards for the reservoir (Reyes-Carmona et al. 2020). Two of these landslides are of rotational type (i.e. Lorenzo-1 and Rules Viaduct landslides) that reaches displacement rates in Line of Sight (LOS) of -35 mm/year while the other landslide shows a translational behaviour (i.e. El Arrecife Landslide) with displacement rates ranging from -10 to -55 mm/year. Afterwards, we used the Sentinel-1 CNR-IREA SBAS service to analyse the Rules Reservoir area using ascending and descending orbit images. Results from both orbits images show the same three landslides within the reservoir slopes. For descending orbits, we obtained mean LOS velocities up to 3 mm/year in the El Arrecife Landslide and LOS velocities reaching -2 mm/year in both Lorenzo-1 and Rules Viaduct landslides. For ascending orbits, we obtained almost similar velocity rates to those we obtained using the PSIG processing (Reyes-Carmona et al. 2020). The second study case is Sierra Nevada, a mountain range of 90 km length and 15-30 km width located in Southern Spain. It reaches 3478 m in elevation, being the highest peak of the Iberian Peninsula. The local relief is 3000 m at just 35 km from Sierra Nevada to the coastline, what leads to high topographic gradients and thus, to a deep fluvial incision that triggers abundant slope instability processes in the area (Fernández et al. 1997). A landslide inventory (Chacón et al. 2007) compiles several landslides in Sierra Nevada, but up to date, no information about their activity has been reported. Thus, we made use of the Sentinel-1 CNR-IREA SBAS service and we obtained the first InSAR results of Sierra Nevada at regional scale. Combining the ascending and descending orbit processing, we were able to detect several active landslides in the area. These two examples illustrate the usefulness of the Sentinel-1 CNR-IREA SBAS GEP service as a tool to support the detection and monitoring of active landslides. If we already have InSAR data of an area, like the described case of the Rules Reservoir, obtaining InSAR results by the GEP allows to evaluate the consistence of both results. Moreover, obtaining the descending orbit data allowed us to have a more comprehensive vision of the displacement patterns of the already reported landslides in the Rules Reservoir. The Sentinel-1 CNR-IREA SBAS service is also useful to obtain preliminary InSAR data in an unexplored area, such as the described case of Sierra Nevada and thus, we are able to evaluate whether it is worth doing further research using InSAR techniques. This service provides results in just 24 hours, what makes possible to perform quick analysis and processing in both ascending and descending orbits. The most remarkable disadvantage of using the Sentinel-1 CNR-IREA SBAS GEP service is that the user cannot control any processing parameter. This fact could lead to reduce the reliability of the InSAR results in some cases. Despite of this, we conclude that the Sentinel-1 CNR-IREA SBAS service has satisfactory proven its effectiveness not only in complementing previously obtained InSAR data but also in being the starting point for performing preliminary InSAR analysis in new areas. References Chacón, J.; Irigaray, T.; Fernández, T. Los movimientos de ladera de la provincia de Granada. In Atlas 658 Riesgos Naturales en la Provincia de Granada, 1st ed.; Ferrer, M., Ed.; Diputación de Granada-Geological 659 Survey of Spain (IGME), Madrid, Spain. 2007; pp. 45–82. Fernández, T.; Brabb, E.; Delgado, F.; Martin-Algarra, A.; Irigaray, C.; Estévez, A., Chacón-Montero, J. Rasgos geológicos y movimientos de ladera en el sector Ízbor-Vélez Benaudalla de la cuenca del río 655 Guadalfeo (Granada). In IV Simposio Nacional sobre taludes y laderas inestables, Granada, Spain, 11-14. November 1997; Alonso, E., Ed.; 795-808. 657 Devanthéry, N.; Crosetto, M.; Monserrat, O.; Cuevas-González, M.; Crippa, B. An approach to persistent 676 scatterer interferometry. Remote Sensing 2014, 6, 6662–6679. Reyes-Carmona, C.; Barra, A.; Galve, J.P.; Monserrat, O.; Pérez-Peña, J.V.; Mateos, R.M.; Notti, D.; Ruano, P.; Millares, A.; López-Vinielles, J.; Azañón, J.M. Sentinel-1 DInSAR for Monitoring Active Landslides in Critical Infraestructures: The Case of the Rules Reservoir (Southern Spain). Remote Sensing 2020, 12, 809.
Authors: Reyes-Carmona, Cristina (1); Barra, Anna (2); Galve, Jorge Pedro (3); Monserrat, Oriol (2); López-Vinielles, Juan (1); Béjar-Pizarro, Marta (1); Ezquerro, Pablo (1); Mateos, Rosa María (1); Herrera, Gerardo (1); Azañón, José Miguel (3)The European Space Agency (ESA), with the Sentinel-1 mission, has introduced an important change in the management of SAR data, freely opening the data archives to the scientific community. This opportunity has provided a huge amount of data to heterogeneous users, from SAR experts to volcanologists and geologists, and together with the advent of large new initiatives and projects of global interest has led to the growth in size and importance of the satellite EO and geohazards community. The geohazard platform or GEP offers this expanding community a unique set of tools to forge new applications in direct collaboration with large numbers of actors. In particular, the community will benefit from a cloud-based workspace, allowing advanced EO data exploitation activities and offering access to a broad range of shared processing tools. Each partner brings their tools and processing chains, but also has access in the same workspace to large data sets and shared processing tools. In this framework ARESYS, in cooperation with Terradue, has put in place a Sentinel-1 SAR data processing service called Sentinel-1 SAR PIPE. This service allows to produce interferograms and coherence maps starting from Sentinel-1 Level-0 (RAW) data, supporting both STRIPMAP and TOPSAR acquisition modes. The main advantages of such service are: the data stack consistency, achieved starting the processing directly from Level-0 (RAW) data and then ensuring a homogeneous reprocessing of all the analysed images; the processing efficiency, ensured limiting the analysis only to the area of interest identified by the user, instead of processing entire frames of data (250x250km or 400x400km respectively for TOPSAR IW and EW acquisition modes). In addition, all the main advantages of the GEP framework are ensured, i.e.: easy and fast data access to the entire Sentinel-1 data archive; simple and powerful web-based service interface; data visualization and results download. Thanks to the way it has been designed and to its modularity, this service can be expanded in the future adding new features and further processing levels and options. Between the others, one of the more promising features will be the introduction of a new Level-0 (RAW) data processing kernel based on a time domain approach (Back Projection), which will allow first of all to obtain highly accurate focused data, removing all the approximation intrinsic in the usage of standard frequency domain approaches, but also to skip the classical steps of data coregistration and geocoding, focusing the data directly on the user-selected Region Of Interest in geographical coordinates and over a predefine Digital Elevation Model. In order to overcome one of the main drawbacks of such kernel, related to low time performances, another future evolution of Sentinel-1 SAR PIPE will be the introduction of GPU processing. This paper is aimed at showing some sample results obtained using Sentinel-1 SAR PIPE service on GEP. In particular, areas affected by earthquakes or other geological events have been selected and analysed.
Authors: Piantanida, Riccardo (1); Prandi, Giorgio (1); Penati, Mattia (1); Aletti, Matteo (1); Pacini, Fabrizio (2); Rossi, Cesare (2); Donnini, Livio (2)The Copernicus POD (Precise Orbit Determination) Service is part of the Copernicus Processing Data Ground Segment (PDGS) of the Copernicus Sentinel-1, -2 and -3 missions. A GMV-led consortium is operating the Copernicus POD (CPOD) Service being in charge of generating precise orbital products and auxiliary data files for their use as part of the processing chains of the respective Sentinel PDGS. The two Copernicus satellites Sentinel-1A and Sentinel-1B are SAR (Synthetic Aperture Radar) satellites, launched in April 2014 and 2016, respectively. POD of the satellites is done based on the dual frequency high precision GPS data from the on-board receivers. Three different orbit products are provided for both satellites. The new PREORB product has a latency of maximum 30 minutes with an accuracy requirement better than 1 m in 2D. It provides an extrapolation of four orbital revolutions to the future. The near real-time (NRT) orbit product has a latency of maximum three hours and an accuracy requirement of 10 cm in 2D. The non-time critical (NTC) orbit product has a latency requirement of less than 20 days and a very high accuracy requirement of 5 cm in 3D. The orbit accuracy validation is mainly done by cross-comparing the CPOD orbits with independent orbit solutions provided by the Copernicus POD Quality Working Group. This is essential to monitor and to even improve the orbit accuracy, because for Sentinel-1 this is the only possibility to externally assess the quality of the orbits. Lacking such an external validation method it took a long time to detect the usage of erroneous GPS antenna reference point (ARP) coordinates for both satellites. Several major updates have been performed during the last year to provide state-of-the-art precise orbit products: 6 May 2020: Major background model update and the switch to carrier phase integer ambiguity fixing 29/30 July 2020: ARP configuration has been corrected Feb 2021: Orbit parametrization update Not only these major changes, but also several updates during the previous mission years made a reprocessing of the S1 NTC orbit product necessary for both satellites. This paper presents the Copernicus POD Service in terms of operations and orbital accuracy achieved for all orbital products for Sentinel-1A and -1B. Focus is led to the validation of the reprocessing results, which now provide a continuous and consistent orbit time series. This includes the comparison against kinematic and combined orbits, and the analysis of the frequencies of the orbital differences, a method that can help to improve the orbit modelling.
Authors: Peter, Heike (1); Fernández, Marc (2); Berzosa, Javier (2); Fernández, Jaime (2); Féménias, Pierre (3)In the present study a geospatial intelligence application is proposed as a useful operational tool in the immediate post-earthquake phase of infrastructure inspection that may have been affected by the co-seismic ground deformation due to very strong earthquake event. The basic principle of geospatial intelligence (GEOINT) is to organize and combine all available data around a geographical location on Earth and then exploit it in order to prepare products that can be easily used by planners, emergency responders and decision makers. As a case study, the Durres (Central Albania) earthquake of 26/11/2019 (3h and 54 min local time), M=6.4 with the epicenter (41.38N, 19.47E) 32 km NWN from the country's capital Tirana and few kilometers N from the city o Durres has been investigated.Aiming to achieve this goal, free SLC SAR images of the Sentinel-1 Copernicus satellite from Copernicus open Hub (https://scihub.copernicus.eu) in both geometry of acquisition (Table 1) were used to create the differential interferograms. Then vector data concerning the local infrastructure was collected from open sources (Table 2). The main steps of the methodology used is composed by the following steps: Generation of co-seismic differential interferograms using ascending orbit data (interferometric pair) over the investigated area as well as the descending orbit data. Phase unwrapping and converting both unwrapped phases to LOS displacement Derivation of East-West and Up-Down components based on the above products (DInSAR LOS displacement) The new deformation maps were introduced and combined with the exposure data in GIS environment to extract the deformation for each exposure type (roads, railways, dams, hospitals e.t.c). Table 1. Sentinel 1 SLC data used Number of Pair Satellite Sub-swath Number Sensing Day Orbit 1st (master) Sentinel-1A IW1 20/11/2019 Ascending 1st (slave) Sentinel-1A IW1 02/12/2019 Ascending 2nd (master) Sentinel-1A IW2 19/11/2019 Descending 2nd (slave) Sentinel-1A IW2 01/12/2019 Descending Table 2 Exposure data and sources Polygonal, linear and point import of exposure Source Roads OpenStreetMap Railway OpenStreetMap Port SEARATES Dams Identification through Google Earth Airport Tirana International Airport (Google Earth) Archeological Sites UNESCO World Heritage Center Hospital Hospital Websites Bridges Identification through Google Earth Hydroelectric Power Plant International Hydropower Association Βoundaries of the cities of Tirana and Durres CORINE Land Cover (CLC) version 2018 The combination of the co-seismic deformation maps knowledge with the exposure geospatial information GEOINT products that are obtained, could lead to a post disaster (response phase) operational tool for field infrastructure inspections.In conclusion, this operational tool could actas a guide for engineers, giving priority to infrastructure detected in regions that exhibit the maximum co-seismic ground deformation.
Authors: Tompolidi, Athanasia-Maria (1); Angelou, Dimitrios (1); Karavias, Andreas (1); Bafi, Despoina (1); Markogiannaki, Olga (2); Parcharidis, Issaak (1)Mining-induced mass movements and surface deformations are very frequent phenomena in mining areas worldwide. In open-pit mining, ground movements potentially lead to slope failures that can produce considerable damage and economic losses. Identifying the triggering factors and detecting unstable slopes and precursory displacements, which can be done by exploiting remote sensing data, is essential to reduce their impact. Within this context, satellite Synthetic Aperture Radar Interferometry (InSAR) (Ferretti et al. 2001), represents a powerful technique for the remote monitoring of open-pit slopes. The present work illustrates the new possibilities offered by the application of FASTVEL ( Ferretti et al. 2001, Berardino et al. 2002), a web-based unsupervised InSAR processing service available on the Geohazards Exploitation Platform (GEP), to characterize ground movements in an open-pit mining area. We illustrate this approach by evaluating a massive landslide occurred in Las Cruces, an active open-pit mine in southwest Spain, in January 2019, that affected the north slope of the pit. For this study, both ascending and descending data were processed spanning January 2018 - January 2019 (i.e. one year before the landslide). The results were obtained by processing 61 Sentinel-1 ascending images (spatial resolution of 3 x 14 m), and 61 descending images. The results in ascending geometry contain 26,721 measurement points (MPs), mostly showing negative line-of-sight (LOS) ground velocities (movement away from the satellite). The north dump exhibits the maximum negative velocity, with a value of -15.7 cm/year, whereas the north slope shows a maximum velocity of -5.8 cm/year. The map in descending geometry contains 27,492 MPs and shows several areas moving away from the satellite as well, corresponding to waste dumps distributed around the open pit. Once more, the north dump exhibits the maximum negative velocity, with a value of -16.6 cm/year, but in contrast to the ascending results, the map in descending geometry shows positive LOS velocities (movement towards the satellite) over the north slope, with a maximum velocity of 6.0 cm/year. The opposite direction of the movements measured along both satellite LOSs over the north slope indicates that the actual displacements have a strong horizontal component. These results were cross-validated using the commercial SqueeSAR® (Ferretti et al. 2011) processing chain. Both ascending and descending data were processed covering the period from 4 January 2018 to 17 January 2019. A set of 63 and 62 Sentinel-1 ascending and descending images were processed. Ascending results contain 18,814 measurement points (MPs), showing only negative LOS velocities over the study area. The south dump exhibits the maximum negative velocity, with a value of -20.4 cm/year, whereas the north slope shows a maximum velocity of -13.2 cm/year. The map in descending geometry contains 19,475 MPs and shows several areas moving away from the satellite as well. Again, the south dump exhibits the maximum negative velocity, with a value of -23.0 cm/year, and in consonance with FASTVEL results, the map in descending geometry shows positive LOS velocities over the north slope, with a maximum velocity of 7.7 cm/year. Visual comparison of ascending results shows larger negative displacements in SqueeSAR®. In addition, FASTVEL results are more stable outside the active areas. This is observed in the resulting absolute difference, with large areas exceeding 2.5 cm/year displacement velocities on the north and south dumps. Global RMSE (Root Mean Square Error) difference reaches 2.97 cm/year for the ascending results, slightly over the stability thresholds. Due to the acquisition geometry (LOS direction) and the slope orientation, the ascending geometry is less sensitive to displacements (63° orientation difference) and its results, less accurate. Descending results fit better and present a lower RMSE (2.28 cm/year), below the stability threshold. This is explained by the fact that descending geometry is more sensitive over the north slope (47° orientation difference), although there are small areas in both the north dump and the north slope showing considerably high differences. The analysis presented herein supports the use of FASTVEL to perform preliminary analysis for the characterization of ground movements in mining areas, although this is constrained by certain limitations in comparison to SqueeSAR®. Note that mining areas represent a challenging environment from the InSAR processing point of view due to a series of limitations mainly related to the temporary decorrelation produced by the strong topographic variations commonly associated to mining operations. Here we show a case where the unsupervised FASTVEL service was successfully applied to characterize ground movements in an open-pit mine prior to a slope failure. However, some caution should be exercised when using InSAR data from unsupervised processing services, considering that very few parameters can be selected during the processing and the results provide average velocities over multilooked pixels. For monitoring purposes, the preliminary results obtained with these tools, should be followed by InSAR processing using supervised tools (e.g. SqueeSAR®) where processing parameters can be adapted to each particular case and InSAR time series at full resolution are obtained. Berardino, Paolo, Gianfranco Fornaro, Riccardo Lanari, and Eugenio Sansosti. 2002. “A New Algorithm for Surface Deformation Monitoring Based on Small Baseline Differential SAR Interferograms.” IEEE Transactions on Geoscience and Remote Sensing 40(11): 2375–83. https://ieeexplore.ieee.org/abstract/document/1166596/ (February 17, 2020). Ferretti, Alessandro et al. 2011. “A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR.” IEEE Transactions on Geoscience and Remote Sensing 49(9): 3460–70. Ferretti, Alessandro, Claudio Prati, and Fabio Rocca. 2001. “Permanent Scatterers in SAR Interferometry.” IEEE Transactions on Geoscience and Remote Sensing 39(1): 8–20.
Authors: López-Vinielles, Juan (1,2); Ezquerro, Pablo (1); Fernández-Merodo, José A. (1); Béjar-Pizarro, Marta (1); Monserrat, Oriol (3); Barra, Anna (3); Blanco, Pablo (4); García-Robles, Javier (4); Filatov, Antón (4); Reyes-Carmona, Cristina (1); Sarro, Roberto (1); Mateos, Rosa M. (1); Azañón, José M. (5); Galve, Jorge P. (5); Herrera, Gerardo (1)In the framework of the french research infrastructure « Data Terra » and its Data and Services cluster for Solid Earth « ForM@Ter », CNES has developped a new service dedicated to InSAR processing of Sentinel-1 data. This service is named FLATSIM : ForM@Ter LArge-scale multi-Temporal Sentinel-1 InterferoMetry. It allows the systematic production of interferograms from Sentinel-1 data, as well as displacement time series, over large geographic areas (more than 250000 km2). It is based on the InSAR “New Small temporal and spatial BASelines” calculation chain (NSBAS, Doin et al., 2011, Grandin, 2015). The scientific objectives are to allow the measurement of deformations of the Earth's surface at the continental scale, and to ensure the spatio-temporal monitoring of critical regions prone to natural hazards (for instance large active fault zones and magmatic systems, landslides at the scale of large massifs). We present here (1) the architecture of the FLATSIM processing chain at CNES, (2) the distributed products, including stacks of co-registered interferograms, displacement time series and mean velocity maps in radar and ground geometry, as well as a set of auxiliary products or files such as interferogram network, atmospheric delay maps used to correct the interferograms, spatial and temporal coherence maps, etc, (3) first results on several test study areas all over the world, including the eastern part of the tibetan plateau, where Sentinel-1 data from 1200 km-long tracks acquired along two descending and two ascending orbits were processed (validation is made by comparing average velocities in overlapping areas of adjacent tracks), as well as (4) products distribution and service access policy and perspectives. References : - Doin, M.-P., Guillaso, S., Jolivet, R., Lasserre, C., Lodge, F., Ducret, G., Grandin, R., Presentation of the small-baseline NSBAS processing chain on a case example : the Etna deformation monitoring from 2003 to 2010 using ENVISAT data, Proceedings of the European Space Agency Symposium « Fringe », Frascati, Italy, 2011. - Grandin R., Interferometric processing of SLC Sentinel-1 TOPS data, Proceedings of the European Space Agency Symposium « Fringe », Frascati, Italy, 2015.
Authors: Donadieu, Joëlle (1); Durand, Philippe (1); Proy, Catherine (1); Clesse, Dominique (2); Deschamps-Ostanciaux, Emilie (3); Diament, Michel (3); Grandin, Raphaël (3); Pointal, Elisabeth (3); Doin, Marie-Pierre (4); Laurent, Christophe (4); Pathier, Erwan (4); Thollard, Franck (4); Lasserre, Cécile (5); Lemrabet, Laëtitia (5)Synthetic aperture radar (SAR) interferometry (InSAR) is a powerful technology that makes it possible to measure, with millimeter-scale precision, ground motions (typically induced by landslides, subsidence and earthquake or volcanic phenomena) using series of SAR data acquired from satellites. This enables, for example, monitoring of the stability of slopes, mining areas, buildings and infrastructures. Satellite InSAR technology is the basis of the European Ground Motion Service (EGMS), which is funded by the European Commission and forms an essential element of the Copernicus Land Monitoring Service (CLMS) managed by the European Environment Agency. The EGMS constitutes the first application of the InSAR technology to high-resolution monitoring of ground deformations over an entire continent, based on full-resolution processing of the whole archive of past and future Sentinel-1 (S1) satellite acquisitions over most parts of Europe. Upscaling from existing national precursor services to pan-European scale is a very challenging but important task, although low-resolution datasets have been recently produced at this scale. To this aim, the EGMS counts on leading expertise on InSAR technology and ground motion service provision, and utilises the most advanced persistent scatterer (PS) and distributed scatterer (DS) InSAR processing algorithms, as well as adequate techniques exploiting the overlaps between adjacent processing areas in order to ensure seamless harmonization between the adjacent S-1 tracks. Moreover, the realization of a high-quality GNSS model and its use, to tie the ground motion products to an established geodetic reference frame (ETRF2014), is also foreseen within EGMS. To foster maximum use of the service by the growing Copernicus user community and the public at large, the EGMS will also provide tools for visualization, exploration, analysis and download of the ground deformation measurements, as well as elements to promote best practices and user uptake. The EGMS will add a new and unique European-wide geospatial layer to the Copernicus Land Monitoring Service (CLMS) portfolio.
Authors: Costantini, Mario (1); Minati, Federico (1); Trillo, Francesco (1); Ferretti, Alessandro (2); Novali, Fabrizio (2); Passera, Emanuele (2); Dehls, John (3); Larsen, Yngvar (4); Marinkovic, Petar (5); Eineder, Michael (6); Brcic, Ramon (6); Siegmund, Robert (7); Kotzerke, Paul (7); Kenyeres, Ambrus (8); Proietti, Sergio (1); Solari, Lorenzo (9,10); Andersen, Henrik S. (10)The Geohazard Eexploitation Pplatform (GEP) is a cloud processing service platform created developed by Terradue and funded by European Space Agency (ESA), offering the possibility to access a certain number of InSAR processing software’sservices; like GMTSAR, SNAP, GAMMA, DIAPASON, etc… This, in complement with the possibility to process free open data offered by ESA both SAR (ERS, Envisat, Copernicus Sentinel-1) and optical (Copernicus Sentinel-2). All these utilities come with default parameters and requires a minimum level of intervention of the user, this facilitates the task for non-experimented experienced researchers in the InSAR field. InSAR Experience is still required for proper interpretation and analysis of InSAR results. The estimation of displacement times series from SAR data, requires dealing with advanced algorithms like PSInSAR and SBAS. GEP offers two processing services for Multi-Temporal InSAR analysis (MTI); Grid Processing On Demand (G-POD) for ERS and Envisat data, and Parallel SBAS (P-SBAS) for Sentinel-1 SLC data. Dealing with a large amount of SAR data (>30 images) especially in the case of Sentinel-1, is time and resources consuming, without neglecting the time considered for downloading the data. The user friendly tools P-SBAS provided by GEP helps to process automatically a set on 100 Sentinel-1 SLC images in less than 72 hours. The P-SBAS algorithm was applied on two different Sentinel-1 data sets covering the region of Oran, the second important city in Algeria. The Tracks 103 ascending (56 images) and 110 descending (104 images) were used to generate displacement times series and mean annual velocities in both geometries. Oran city is characterized by an active subsidence in its coastal area, where some landslides were mapped these late years. A former study by our team, about using Envisat data (2003-2010) to monitor the active subsidence in the area surrounding the city, showed a continuous subsidence in the suburban’s, where water pumping from underground in frequent. The results obtained by P-SBAS on GEP confirm the stability of the city, where the mapped subsidence attained 3 to 5 mm/yr. But the surrounding area is still suffering from an important subsidence that can reach 1 cm/yr. In certain regions, especially at the coast where it is important to continually monitor this area to prevent landslides that can be triggered by this subsidence, as it was the case in August 2017, when a bridge failed due to a landslide.
Authors: Hasni, Kamel; Gourine, Bachir; Allal, Saddam-HousseynThe Earth’s subaerial volcanoes pose a variety of threats, yet the vast majority remain unmonitored. However, with the advent of the latest synthetic aperture radar (SAR) satellites, interferometric SAR has evolved into a tool that can be used to monitor the majority of these volcanoes. Whilst challenges such as the automatic creation of interferograms have been addressed and progress has been made in the automatic identification of signals in individual interferograms, there remain challenges in the flagging of low rate, topographically correlated or unusual deformation We present a new algorithm named LiCSAlert, developed to monitor all volcanoes globally using Sentinel-1 data. Automatic creation of interferograms is performed by LiCSAR, and time series are formed using LiCSBAS, before the LiCSAlert algorithm endeavours to determine the latent signals present in a baseline time series. Changes in either the spatial or temporal nature of these signals are then used to determine if a volcano is undergoing unusual deformation. Of particular importance for volcano monitoring using InSAR is the ability to differentiate between signals caused by changes in the atmosphere, and those caused by deformation. Our algorithm is designed to mitigate this through characterising the atmospheric signals observed in the baseline data, and we present results of a comparison between the application of our algorithm to raw interferograms, and to interferograms corrected using the results of weather models. The LiCSAlert algorithm also contains a deep learning module which is able to differentiate between spatial signals within the baseline data that are caused by deformation, and those caused by changes in the atmosphere. We present the results of using this module at volcanoes which feature steady state deformation, and show that it can be used to detect subtle changes that are important for volcano monitoring.
Authors: Gaddes, Matthew; Hooper, Andrew; Ebmeier, SusannaWe are communicating recent integration of the SNAPPING surface motion mapping service for Sentinel-1 mission on the GEP platform in support to the scientific community as well as EO practitioners. The service is built on ESA SNAP and StaMPS packages that have already demonstrated numerous successful investigations of geohazard phenomena. SNAPPING is well-tailored in terms of EO data manipulation and parallelization on cloud resources, enabling users to respond to the ever increasing volume of satellite data and high computational requirements. The service generates average motion rate maps and full displacement time series at reduced spatial resolution, making it suitable not only for various research application domains, but also when rapid and low cost inspection of an area at medium resolution is of interest.
Authors: Foumelis, Michael (1); Delgado Blasco, Jose Manuel (2); Brito, Fabrice (3); Pacini, Fabrizio (3); Pishehvar, Panteha (3)After the earthquakes of 1978 in the vicinity of Thessaloniki, 1981 in the eastern Gulf of Corinth and 1986 near Kalamata, a European consortium leaded by French and Greek research and educational centers decided to initiate a long term observational project aiming to better understand the physics of the earthquakes and seek for possible precursory transients of any type. The consortium decided to establish a dense array of sensors around the western gulf of Corinth (Greece), which is one of the areas with the highest seismic hazard in Europe. The cities of Patras and Aigion and other towns were destroyed several times since the antiquity by earthquakes and, in some cases, by tsunamis. The historical earthquake catalogue of the area reports on average five to ten events of magnitude larger than 6 per century. The Corinth Rift Laboratory (CRL, http://crlab.eu) gathers several tens of scientists from many European countries and is now one of the Near Fault Observatories of the European Plate Observing System (EPOS, http://www.epos-ip.org). The observational array in the field includes more than 80 instruments than can be queried in real-time, most of them being seismic and GNSS stations. With the development of the InSAR and high-resolution optical space missions, remote sensing has currently an increasing role in the observatory. Remote sensing does not have the real-time capability. However, the time lag can be relatively short for the space products to be useful for the monitoring of a crisis. In addition, the data and products of space missions, especially InSAR, contain products that cannot be obtained in the field and are sometimes crucial to understand the physical processes. In the domain of InSAR, the strategy of the observatory is based on a complementary approach, involving on the one hand the routine processing of interferograms (using the ESA SNAP in the computers of the observatory, as well as specific post-processing chains) and, on the other hand, the use of the Geohazards Thematic Exploitation Platform (G-TEP). These two approaches are not redundant, but organized to be complementary. For the observatory needs, the G-TEP can provide and gather in a well-organized manner products routinely produced by different software, with a double benefit for the observatory: (1) no need to maintain computer and software facilities and (2) possibility to compare the solutions obtained by different software. However, there is one major disadvantage, from the metrology and science point of view, which is the lack of visibility on the algorithms used by the G-TEP software. In the forthcoming years we intend to strengthen our links with the G-TEP team and the space agencies to turn this remote sensing component stronger in the CRL NFO. The G-TEP has another important interest which is the potential conviviality that permits to disseminate much easier information to the public and to schools. Furthermore, it permits to put easily the hands of students on the available tools and products. Since 2016 a yearly summer school is organized in the framework of the NFO CRL for the master students of the partner Universities. This is a summer school called CRL-School, tailored to teach in the lab and in the field the major components and theoretical background of the observations performed in the NFO. Space observations occupy an important role in the school with the presence of experts from space agencies and the G-TEP consortium. Particular efforts are made to analyze with the master students the geological features and their dynamics, in an approach combining the remote sensing and the field investigation.
Authors: Elias, Panagiotis (1); Briole, Pierre (2); Mathot, Emmanuel (3); Foumelis, Michael (4); Ganas, Athanassios (5); Beck, Christian (6); Mouratidis, Antonios (7); Valkaniotis, Sotiris (8); Panagis, Akis (9); Parcharidis, Issaak (10); Kaviris, George (11); Klein, Emilie (2); Karamitros, Ioannis (5); De Luca, Claudio (12); Bally, Philippe (13)With the launch and successful commissioning of the RADARSAT Constellation Mission (RCM), Canada has entered a new era in remote sensing. Capable of collecting vast quantities of high-resolution, wide-area Synthetic Aperture Radar (SAR) data with a rapid four-day revisit cycle, RCM can provide users with high volumes of SAR data. This SAR data, as well as derived ground deformation products, can provide critical information for a wide variety of end users. Efficient processing of this data, for expert and non-expert users alike, remains an important goal for the continuing RCM mission. In support of this goal, the Canada Centre for Mapping and Earth Observation (CCMEO) has developed a system generating standard and advanced deformation and change detection products from Synthetic Aperture Radar (SAR) data acquired by RADARSAT-2 and RCM satellites using Differential Interferometric Synthetic Aperture Radar (DInSAR) processing methodology. The development of this system is supported through the Canadian Space Agency (CSA) Data Utilization Application Plan Program (DUAP). The system is designed to be compatible with remote High Performance Computing (HPC) cluster environment and is both scalable and customizable. In this presentation, we will describe this automated system, from the data ingestion pipeline through to the output derived deformation and time series products. We will explore the motivations of the RCM Data Utilization Application Plan program, including its goals and intended applications. The graphical user interface used to select data and processing options will also be presented. The capabilities of the system will be demonstrated through case studies using both RCM and RADARSAT-2 data. These studies will help to demonstrate the intended goals of the system. A set of representative monitoring sites will be discussed and the results of the analysis by the automated system across these sites will be explored. Derived interferometric products will be presented for both Stripmap and ScanSAR beam modes. The effects of temporal decorrelation for a 4-day repeat pass system are investigated and compared with simple decay models. The use of RCM Compact-Pol data for DInSAR processing is also assessed. Lastly, we will summarize the status of the system and discuss current limitations and future goals.
Authors: Dudley, Jonathan; Samsonov, SergeyCorner reflectors and radar transponders are commonly used for radiometric and geometric Synthetic Aperture Radar (SAR) calibration, SAR interferometry (InSAR) applications over areas with few natural coherent scatterers, and InSAR datum connection and geodetic integration. Despite the current abundance of regular SAR time series, no free and open-source software (FOSS) dedicated to analyzing SAR time series of corner reflectors exists. Here, we present the GECORIS: A FOSS Python Geodetic Corner Reflector InSAR Toolbox. The GECORIS allows efficient and automatic: (i) estimation of the clutter level of a site before a corner reflector installation, (ii) estimation of the Radar Cross Section (RCS) to track a corner reflector’s performance and detect outliers, (iii) estimation of the Signal-to-Clutter Ratio (SCR) to predict the positioning precision and the InSAR phase variance, (iv) estimation of the InSAR displacement time series of a corner reflector network. Using the GECORIS, we analyze one-year of the Sentinel-1 time series of five areas in Slovakia affected by landslides. Twenty-four double back-flipped trihedral corner reflectors (CR) were carefully deployed at these sites to form a reference network, guaranteeing reliable displacement information over the critical landslide zones. To confirm the measurement quality, we show that the temporal average Signal-to-Clutter Ratio (SCR) of the CR is better than 20 dB. Using ascending and descending reflectors within a single instrument enables the assumption-less decomposition of the observed cross-track line-of-sight (LOS) displacements into the vertical and horizontal displacement components. The observed CR motions in vertical and east-west directions vary from several millimetres up to 3 centimetres, with an average standard deviation better than 0.5 mm. Since the GECORIS is efficiently working on the SAR image patches containing reflectors, regular updates can be performed as soon as a recent Sentinel-1 acquisition is available. Therefore, it allows tracking the corner reflector’s measurement quality and operationally monitor their displacement and identify possible problems, for example, reflector damage or debris accumulation. Using the GECORIS, even less-experienced users from wider geodetic, SAR and InSAR communities can readily analyze SAR time series of their corner reflectors in a standard and concise way. We consider this tool especially useful in the context of emerging national and international initiatives, installing corner reflectors collocated with continuously operating GNSS reference stations to provide an absolute geodetic reference for large-scale InSAR analysis. Acknowledgements: Corner reflectors employed in this study were built thanks to the Slovakian geological task ‘Monitoring of landslide deformations’, which is solved within the Environmental Quality Operational Program, priority axis 3: Support of risk management, emergency management and resilience to emergencies affected by climate change, which is directly related to the task CMS–GF.
Authors: Czikhardt, Richard (1); van der Marel, Hans (2); Papco, Juraj (1); Ondrejka, Peter (3)The COMET LiCSAR system automatically processes Sentinel-1 data to derive InSAR products for tectonic regions and volcanoes [1]. LiCSAR data are used to assess tectonic velocities and strain, earthquake rupture zones, volcanic deformation, and have potential applications to mass movement and cryospheric research. Products are disseminated through an online portal (https://comet.nerc.ac.uk/comet-lics-portal/), which enables location-based search and download of the processed data, and visualisation of quick-look images. Here, we present recent developments to the web portal, including the analysis of displacement time series, visualisation of coseismic interferograms, and investigation of volcanic deformation. We discuss future developments to the web tools and invite user feedback on the LiCSAR system. The LiCSAR portal displays the current InSAR product coverage using an interactive Leaflet map, with options to overlay the location of active volcanoes and the Global Active Faults database. Data download is provided by links to storage at the Centre for Environmental Data Analysis (CEDA) and updates are tweeted by @COMET_database #LiCSAR. Recent developments include displacement time series data for an area of Turkey processed using LiCSBAS [2], which allows the user to interactively plot five years of displacement data. Time series availability will increase when the LiCSBAS module is initiated for more frames, initially focussing on volcanoes and the Alpine-Himalayan belt. The COMET Volcano Deformation Database [3], which is in development, will allow users to plot, analyse, and export displacement time series for over 1000 volcanoes. Machine learning will aid in the identification of deformation signals [4], which will help observatories monitor and respond to volcanic unrest. We will present the current status of the LiCSAR portal, web tools, and the results of user feedback on these systems, including testing of the Volcano Deformation Database by a group of volcano experts. We conclude by discussing the implementation of the database, and linking the LiCSAR earthquake responder to the web portal, which will improve the access to InSAR products following major earthquakes. We will also evaluate future capabilities to analyse LiCSAR data through cloud-based platforms such as Google Earth Engine. References: 1. Lazecky, M., Spaans, K., Gonzalez, P. J., Maghsoudi, Y., Morishita, Y., Albino, F., Elliott, J., Greenall, N., Hatton, E., Hooper, A., Juncu, D., McDougall, A., Walters, R., Watson, C. S., Weiss, J. R. and Wright, T. J. . LiCSAR: An Automatic InSARTool for Measuring and Monitoring Tectonic and Volcanic Activity, Remote Sensing (in prep). 2. Morishita, Y.; Lazecky, M.; Wright, T.J.; Weiss, J.R.; Elliott, J.R.; Hooper, A. LiCSBAS: An Open-Source InSAR Time Series Analysis Package Integrated with the LiCSAR Automated Sentinel-1 InSAR Processor. Remote Sensing 2020, 12, 424. 3. Ebmeier, S.K.; Andrews, B.J.; Araya, M.C.; Arnold, D.W.D.; Biggs, J.; Cooper, C.; Cottrell, E.; Furtney, M.; Hickey, J.; Jay, J., et al. Synthesis of global satellite observations of magmatic and volcanic deformation: implications for volcano monitoring & the lateral extent of magmatic domains. Journal of Applied Volcanology 2018, 7, 2, doi:10.1186/s13617-018-0071-3. 4. Anantrasirichai, N.; Biggs, J.; Albino, F.; Hill, P.; Bull, D. Application of Machine Learning to Classification of Volcanic Deformation in Routinely Generated InSAR Data. Journal of Geophysical Research: Solid Earth 2018, 123, 6592-6606, doi:10.1029/2018jb015911.
Authors: Watson, C. Scott (1); Lazecky, Milan (1); Ebmeier, Susanna (1); Rigby, Richard (2); Burns, Helen (2); Morishita, Yu (1,3); Maghsoudi, Yasser (1); Elliott, John (1); Wright, Tim (1)The grounding line of marine-terminating glaciers is the boundary where ice starts to become afloat in the ocean. Knowledge of the position of this boundary is important for our understanding of ice sheet mass balance, glacier dynamics, and their contributions to sea level rise. The state of the art method to delineate the grounding line is through differential interferometric synthetic-aperture radar interferograms. Data availability changed dramatically in recent years. Before the launch of Sentinel-1 and other SAR missions of the same generation, SAR acquisitions in coastal Antarctica were campaign based and few and far between. As a result, few suitable acquisitions were made to form double difference interferograms. Grounding lines were mostly generated through manual digitizations of the tide dependent phase information. Where multiple acquisitions were available over a year, we find fluctuation of the grounding line position that is not related to advance or retreat, but is driven by tides. We define the width of the area of grounding line position change as the grounding zone. This information is relevant to ice sheet models. A consequence of wider grounding zones is that seawater must intrude over considerably longer distances beneath grounded ice at high tide than expected. Water intrusions in turn imply that ice probably melts at tidal frequencies in these broad regions. The Sentinel-1 mission plan includes ongoing 6- or 12-day acquisitions in most coastal regions of Antarctica. As a result, many options for double difference interferograms exist for a given year. Manual interpretation of the data is therefore no longer feasible. We present a fully-convolutional neural network that automatically and efficiently delineates grounding lines at a large scale. In addition to a grounding line position estimate, our method also provides uncertainty estimates. Our procedure detects grounding lines within 232 m in 100-m posting interferograms, which is comparable to the performance achieved by human experts. Having many grounding measurements per year for the entire observation area enables us to measure the grounding zone width everywhere. We demonstrate that grounding zones are one order magnitude wider than expected from hydrostatic equilibrium, which justifies the need to map grounding lines repeatedly and comprehensively to inform numerical models. The complete mapping of grounding lines and the corresponding grounding zone around Antarctica performed by the ML algorithm provides a new basis for the analysis of grounding line dynamics in Antarctica that was not available previously. We expect that the product will be of considerable interest for the scientific community, in particular for numerical ice sheet modelers and ocean modelers, and will help improve projections of ice sheet evolution and contributions to sea level rise from models.
Authors: Mohajerani, Yara (1,2,3); Scheuchl, Bernd (1); Jeong, Seongsu (1); Chen, Hanning (1); Milillo, Pietro (1); Rignot, Eric (1,4); Velicogna, Isabella (1)Sentinel-1 SAR mission. Over land surfaces and inland ice the Sentinel-1 SAR operates in Interferometric Wide Swath (IWS) mode applying Terrain Observation by Progressive Scans (TOPS) technology providing a spatial resolution of about 3 m and 22 m in slant range and azimuth, respectively, over a swath of 250 km width. The Sentinel-1 A/B constellation, with its dedicated polar acquisition scheme, is the basis for monitoring ice flow velocity of the Greenland and Antarctic ice sheets at unprecedented spatial and temporal sampling, currently by applying the offset tracking technique. A continuous observational record of the ice sheet margins since October 2014, augmented by dedicated ice sheet wide mapping campaigns, enables the operational monitoring of key parameters such as ice velocity and calving fluxes. In 2019 additional tracks were added to the regular acquisition scheme, covering the slow-moving interior of the Greenland Ice Sheet by crossing ascending and descending acquisitions. This offers the opportunity for regular application of the InSAR technique for improving ice velocity products in particular in slow moving sections of ice sheets. We developed and implemented an InSAR processing line for generation of ice velocity maps from crossing orbits of Sentinel-1 IW TOPS data. Processing of 2 years of Sentinel-1 data over Greenland shows that coherence is largely preserved for 6-day time spans during periods without surface melt. Topographic phase flattening is applied using the GIMP DEM resampled to 50 m pixel spacing. The major challenge in TOPS interferometry are phase jumps at burst boundaries affecting displacement in along-track direction. We developed and implemented a method to reduce the phase jumps before debursting. After debursting and phase unwrapping, ice velocity maps with 50 m pixel spacing are generated combining Sentinel-1 acquisitions from crossing orbits, applying data stacking over 2 months periods and using external reference points for velocity calibration. The interferometric ice velocity product is not affected by ionospheric strikes which are evident especially in slow moving areas in the corresponding offset tracking ice velocity products. For areas covered by only one orbit direction (ascending or descending) the interferometric ice velocity retrieval uses as additional information the flow direction from multi-annual Sentinel-1 ice velocity maps generated with offset tracking. On the tongues of major outlet glaciers the high velocity causes decorrelation of the phase signal. These areas are filled with ice velocity data generated by offset tracking. We will present the novel Greenland ice sheet wide ice velocity map (50 m pixel spacing) generated by means of Sentinel-1 SAR interferometry, complemented by offset tracking in fast moving sections of outlet glaciers for the winter period 2019/20 and 2020/21. Additionally, we show dense time series with 6 days intervals of InSAR ice velocity maps for selected regions in Northern Greenland and Antarctica. The performance of the product is evaluated using GPS measurements. Improvements of the InSAR based products versus the current ice velocity product, based on offset tracking, will be highlighted.
Authors: Nagler, Thomas; Libert, Ludivine; Hetzenecker, Markus; Keuris, Lars; Wuite, Jan; Rott, HelmutMapping ice velocity is essential for many glaciological applications such as the study of ice dynamics, sliding at the base or inversion of ice thickness distribution. During the last decade, the number of available satellite observations has increased significantly, allowing for far more frequent measurements of the glacier speed from synthetic aperture radar and optical observations using tracking algorithms. Despite this increase in observations, each sensor comes with its own limitations in mapping glacier displacement. Here, we show that fusioning radar observations obtained by Sentinel-1/ESA with optical images such as Sentinel-2/ESA or Landsat-8/NASA allows for a comprehensive mapping of the >250’000 glaciers around the world, or 98% of glacierized areas outside the ice sheets. This new comprehensive mapping highlights the importance of combining multi-sensor satellite data, i.e SAR and optical, to produce the most exhaustive ice velocity datasets, hence revealing the ice flow of glaciers that have never been mapped before. The new picture of glacier dynamics is finally combined with with single pass interferometry DEM from DLR's TanDEM-x and optical stereo DEM from ASTER to revise ice thickness distribution and volume of glacier around the world with a simple inversion approach that uses the Shallow Ice Approximation. We discuss the quality of our global glacier surface flow velocity product and of our new ice volume reconstruction with respect to existing state of the art estimates and quantify the impact of our results in terms of sea level rise and water resources.
Authors: Millan, Romain (1); Mouginot, Jeremie (1,2); Rabatel, Antoine (1); Morlighem, Mathieu (2)Surface ice velocity is a fundamental characteristic of glaciers and ice sheets that quantifies the transport of ice. Changes in ice dynamics have a major impact on ice sheet mass balance and its contribution to sea level rise. Prior comprehensive mappings employed speckle and feature tracking techniques, optimized for fast‐flow areas, with precision of 2‐5 m/year, hence limiting our ability to describe ice flow in the slow interior. This level of precision is suitable for calculating ice volume fluxes of fast moving glaciers along the periphery of the ice sheets, where ice speed ranges from hundreds to thousands of meters per year, but it is not sufficient to constrain physical processes in the interior regions, such as flow changes, and changes in strain rate or forcebalance. Indeed, about 60% of Antarctica is flowing at a speed slower than 10 m/yr. A SAR interferogram is formed by differencing the complex amplitude (magnitude and phase expressed as a complex number) of radar data acquired a few days apart along the same relative orbit of the spacecraft. The resulting interferogram contains signals associated with ice motion, surface topography, and image noise, which includes ionospheric noise, residual error in interferometric baseline, and perturbations associated with water vapor. After correcting for the different effects in order to obtain only the displacement in light of sight, the phases were unwrapped, referenced and calibrated against existing datasets to remove residual baseline errors. The unwrapped phases are then stored in a geographic database constructed with netCDF cubes as multiple layers, with their own heading, incidence angle, date of acquisition, and precision levels. Dividing Antarctica in cubes makes it possible to process data in parallel, hence improving computational efficiency of the mapping process. The unwrapped phase uses C-Band sensors (l ~ 5.5 cm) with 1461 tracks from RADARSAT-1 & -2, 76 from ERS-1/2, 319 from Envisat ASAR, and L-Band sensors (l ~ 23.6 cm): 556 from ALOS/PALSAR and 3 from ALOS2/PALSAR2. To form a three-dimensional ice flow map from the cubes, we used the surface-parallel flow assumption. For each cube, ascending and descending passes with a difference in track direction greater than 25º are combined to form the horizontal components (vx, vy) of the ice-flow. Surface slopes needed for estimating the vertical component (vz) of the flow are generated using the high-resolution TanDEM-X DEM smoothed with an averaging filter of 5 to 10 ice thickness. We obtain a vector map of ice velocity using the interferometric phase from multiple satellite synthetic aperture radars resulting in 10 times higher precision in speed (20 cm/year) and direction (5°) over 80% of Antarctica. Precision mapping over areas of slow motion (
Authors: Mouginot, Jeremie (1,2); Rignot, Eric (2,3,4); Scheuchl, Bernd (2); Milillo, Pietro (2); Millan, Romain (1); Kääb, Andreas M. (5); Mulder, Gert (6); Lopez-Dekker, Paco (6)Grounding zones mark the boundary between grounded and floating parts of a marine terminating ice sheet. This is the most appropriate boundary for estimating the balance between outgoing ice fluxes and snow accumulation and is the point past which the ocean can influence and interact with the ice sheet. Recent findings highlight how grounding zones can migrate by several kilometers within a single tidal cycle depending on the shape, geometry of the bed, surface slope, and glacier thickness. Here, we present evidence of how grounding line dynamics affect glacier flow and how it affects uncertainties in sea level projections. The proposed synergistic approach enables the characterization of grounding line migrations, together with their impact on thinning and ice bottom melt, and sheds new lights on the physical processes controlling grounding line migration. We employ data from the second generation of SAR systems e.g. the European ESA Sentinel-1A/B constellation, the Italian COSMO- SkyMed (CSK) and COSMO-Second Generation (CSG) constellations, the German TanDEM-X (TDX) formation to monitor grounding line retreat using short repeat-time interferometry and accurate InSAR digital elevation models (DEM) on the Amundsen Sea Embayment (ASE), West Antarctica. The ASE is a marine-based ice sheet with a retrograde bed, containing enough ice to rise global sea level by 120 cm. Several studies have inferred the mechanical properties of portions of the ASE using observationally constrained numerical models, but these studies offer only temporal snapshots of basal mechanics owing to a dearth of observational time series. Prior attempts of grounding lines mapping have been limited because few space-borne SAR missions offer the short-term repeat-pass capability required to map the differential vertical displacement of floating ice at tidal frequencies with sufficient detail to resolve grounding line boundaries in areas of fast ice deformation. Using data characterized by short temporal baselines and TDX DEMs, we collected frequent, high-resolution grounding line measurements starting from 2015. We compare the results with ERS data spanning 1996-2011. We find that in these areas grounding line retreat has been fueled by the enhanced intrusion of warm, salty, subsurface ocean water of circumpolar deep water origin onto the continental shelf, beneath the floating ice shelf, to reach the glacier grounding zone and melt it from below at rates varying from 50 to 150 m/yr. The retreat rate varies depending on the magnitude of ice melt by the ocean, the rate of ice thinning, and the shape of the glacier surface and bed topography.
Authors: Milillo, Pietro (1,2); Rignot, Eric (1,3); Rizzoli, Paola (2); Scheuchl, Bernd (1); Mouginot, Jeremie (4); Bueso Bello, Jose Luis (2); Prats Iraola, Pau (2); Zink, Manfred (2); Dini, Luigi (5)Synthetic Aperture Radar (SAR) Remote Sensing already proved as an ideal solution to determine surface displacements, thanks to its day-and-night and cloud-free characteristics, either with differential SAR interferometry or speckle tracking. Additionally, the increasing number of SAR data made available to users (Radarsat Constellation Mission, Cosmo Skymed, SAOCOM, Sentinel-1, and so forth) allows performing time-series displacement monitoring using (M)SBAS or PSI techniques. As it is well established, using SAR images at different times from slightly different points of view, we can observe surface movements by differential phase shift measurements between acquisitions. This differential interferometry approach is used by techniques like DInSAR, SBAS, MSBAS, PSI, MAI, BOI, and so forth. DInSAR allows determining displacements according to the line of sight of the sensor with a centimetric or even millimetric accuracy, with a sensor at several hundreds of kilometers distance. This partly explained the success of DInSAR. To reconstruct the bi- or tri-dimensional displacements, we need other measures from different viewing geometries. Combining a great number of images from different orbits, sensors and/or viewing angles, we can reconstruct the full vectorial components of the surface deformation. Unfortunately, a time series approach is not achievable everywhere on Earth. In particular, Antarctica has many geographical areas where only a limited number of acquisition geometries is available. Besides, techniques based on SAR interferometry are limited by other factors. Among them, the magnitude of the displacements can introduce a decorrelation such that the wavefronts combination emitted from two different times does not give a coherent signal. This temporal decorrelation is particularly remarkable in coastal regions of Antarctica, where the revisit time of Sentinel-1 (6 or 12 days, depending on the region) allows the scatterers to move from one picture element to another. In these cases, it is possible to use another family of techniques, based on the tracking of feature elements at the surface, i.e. speckle tracking technique. In speckle tracking, the technique uses two SAR images at different acquisition times. According to a defined spatial sampling, we search in the second image a translation in picture elements that maximizes the local correlation. From this translation is deduced a bi-dimensional displacement and, in fine, a velocity. This technique is less precise than InSAR-based methods but is less impacted by displacement-induced decorrelation, while it also directly brings a 2D velocity field. The limits of applicability between InSAR and Speckle tracking are not fixed and, when the two options are possible, we would always opt for phase-based measurements thanks to their incredible accuracy. It is in this context that coherence tracking was born by taking the best of the two approaches. Coherence tracking determines bidimensional displacements by maximizing the quality of the interferogram at a local scale, through the coherence estimation. The use of the phase in a tracking approach allows recovering the location of ground scatterers in the second image. Then, it is possible to determine a tracked interferogram that contains the displacement along the line-of-sight, with the expected interferometric accuracy. Coherence tracking is one way to circumvent the issue of temporal decorrelation induced by fast-moving areas. Using coherence tracking with Sentinel-1 TOPSAR Acquisition mode is not straightforward. TOPSAR introduces a phase bias to be taken into account in the interferometric processing. By steering its sensor during the acquisition, Sentinel-1 contains in its signal a strong azimuthal phase ramp. While this phase ramp can be canceled out in classical interferometry, this is not the case in a tracked interferogram. In this research, we present the coherence tracking technique and the added-value brought by the phase information in offset tracking methods. We then explain how to adapt the approach in TOPSAR data, in particular with Sentinel-1. Derived 2D velocity fields of Ice Shelves in East Antarctica are presented. More precisely, we are focusing on the Roi Baudouin Ice Shelf, in Dronning Maud Land. Results are finally compared to traditional approaches.
Authors: Glaude, Quentin (1,2); Derauw, Dominique (2); Barbier, Christian (2); Pattyn, Frank (1)Permafrost is an essential component of the terrestrial cryosphere that is often undervalued in climate change scenarios. Permafrost degradation contributes to global warming by releasing greenhouse gases previously trapped in the frozen ground and has direct impacts on infrastructure and ecosystems. Permafrost state is typically documented by in-situ instruments but the scarce network makes remote and large polar regions difficult to comprehensively document. This leads to large unknowns in the state and projection of future changes, which justify the development of complementary remote sensing products in these areas. In permafrost landscapes, the seasonal ground freeze and thaw induce heave and subsidence within the active layer. The variability of the ground thermal regime, water content and ground properties leads to uneven distribution, amplitude and timing of these displacements. Based on Sentinel-1, we applied a Small Baseline Subset (SBAS) method to retrieve seasonal InSAR time series in Adventdalen (Spitsbergen, Svalbard). We showed that subsidence/heave can be measured with a 6-day sampling and the temporal patterns of the displacement follow the variations of ground temperature measured in boreholes. The series over flat areas can be well described by the Stefan equation with a composite model that combines thawing and freezing degree-day indices and the heave onset matches the start of the active layer freeze-back. The conclusions of the initial study highlighted potentials for the development of alternative InSAR products in areas characterized by cyclic patterns. In the follow-up research, we are mapping the Day of Year (DOY) of the subsidence maxima in three different regions of Western Svalbard: Adventdalen, Kapp Linné and Ny-Ålesund. We identify geographical zonation of the ground dynamics, suggesting that SBAS series and DOY products can indirectly document the changes of the ground thermal regime and complement traditional monitoring networks.
Authors: Rouyet, Line (1,2,3); Liu, Lin (4); Lauknes, Tom Rune (1); Christiansen, Hanne Hvidtfeldt (3); Strand, Sarah Marie (3); Larsen, Yngvar (1)The Copernicus Sentinel-1 Synthetic Aperture Radar (SAR) system, featuring a 6-day repeat pass period and the Interferometric Wide Swath (IW) acquisition mode, is currently the most commonly applied system for acquiring ice sheet and outlet glacier motion measurements. Such measurements are routinely generated using offset tracking techniques. These methods have the advantage of being applicable even on fast-flowing outlet glaciers, although in the ice-sheet interior significant temporal averaging is required to reduce the measurement errors and observe the underlying ice velocities. Furthermore, the achievable spatial resolution is, at best, several hundred meters. Conversely, differential SAR interferometry (DInSAR) generally achieves high accuracy and high spatial resolution (on the order of tens of meters), while not being applicable on very fast-flowing outlet glaciers. An obvious synergy thus lies in applying both offset tracking and interferometric techniques to the Sentinel-1 data archive, yielding velocity measurements of high accuracy and resolution in the interior parts of the Greenland ice sheet as well as an extensive coverage of outlet glaciers provided by offset tracking. While such a combination of SAR techniques has been exploited in the generation of Greenland/Antarctic ice velocity maps [1,2], Sentinel-1 interferometry is not currently applied for the generation of such products. This is mainly due to a feature of Sentinel-1’s main acquisition mode, namely Terrain Observation by Progressive Scans (TOPS), in which the azimuth antenna steering introduces a coupling between interferometric phase and azimuth registration [3]. This in turn complicates interferometric processing and requires a highly accurate azimuth coregistration procedure. To this end, Extended Spectral Diversity has been successfully utilized in scenes that are nearly stationary [4], however this is not straightforward for ice sheets, which are non-stationary. If not accounted for, the azimuth motion component of the ice sheet will cause interferometric phase gradients across each burst, resulting in phase discontinuities at burst boundaries. We present a Sentinel-1 interferometric processing chain, which reduces the azimuth coupling to the line-of-sight phase signal through a spatially adaptive coregistration refinement incorporating azimuth velocity measurements [5]. The latter are based on available multi-year ice velocity mosaics, optionally supplemented by Burst-Overlap Multi-Aperture Interferometry [5,6]. The DInSAR processing chain is demonstrated for a large drainage basin in Northeast Greenland, which includes the Northeast Greenland Ice Stream (NEGIS), and integrated with state-of-the-art offset tracking measurements. In the ice sheet interior the combined DInSAR and offset tracking ice velocity product provides a spatial resolution of 50 m x 50 m and 1-sigma accuracies of 0.18 m/y and 0.44 m/y in the x and y components respectively, compared to EastGRIP GPS measurements. References [1] I. Joughin, Ice-sheet velocity mapping: A combined interferometric and speckle-tracking approach, Annals of Glaciology 34 (2002) [2] J. Mouginot, E. Rignot, B. Scheuchl, and R. Millan, Comprehensive annual ice sheet velocity mapping using Landsat-8, Sentinel-1, and RADARSAT-2 data, Remote Sensing 9 (2017) [3] N. Yague-Martinez, P. Prats-Iraola, F. Rodriguez, R. Brcic, R. Shau, D. Geudtner, M. Eineder, and R. Bamler, Interferometric Processing of Sentinel-1 TOPS Data, IEEE Transactions on Geoscience and Remote Sensing 54 (2016) [4] R. Scheiber, M. Jager, P. Prats-Iraola, F. De Zan, and D. Geudtner, Speckle tracking and interferometric processing of TerraSAR-X TOPS data for mapping nonstationary scenarios, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 8 (2014) [5] Andersen, J. K., Kusk, A., Merryman Boncori, J. P., Hvidberg, C. S., Grinsted, A., Improved Ice Velocity Measure-ments with Sentinel-1 TOPS Interferometry, Remote Sensing 12(12), (2020) [6] Kusk, A., Andersen, J. K., Merryman Boncori, J. P., Burst Overlap Coregistration for Sentinel-1 TOPS DInSAR IceVelocity Measurements, in press, IEEE Geoscience and Remote Sensing Letters, (2021)
Authors: Andersen, Jonas Kvist (1); Kusk, Anders (1); Merryman Boncori, John Peter (1); Solgaard, Anne Munck (2); Hvidberg, Christine Schøtt (3); Grinsted, Aslak (3)Fifteen years ago we carried out dGPS field surveys and mapped surface changes in the Hornbreen-Hambergbreen (H-H) icy isthmus, also known as Mendeleev Plain in South Spitsbergen, Svalbard, using satellite interferometry (ERS-1/2) and altimetry (ICESat) data. Local vertical movements with amplitudes of a few cm per day, measured on the flat and smooth glacier surface, were associated with accelerated basal melting and tidal effects and showed that a relatively large part of the H-H system was afloat. We therefore assumed that there was a subglacial strait and predicted the collapse of the H-H ice bridge by 2020. Now is 2021 and the H-H ice bridge that connects Sörkapp Land to the main island of Spitsbergen is still intact, although it is getting thinner and narrower over time. To explain the reason for our failed long-term forecast, we decided to repeat remote sensing surveys of the ice bridge. This time it was done using the repeat-pass Sentinel-1-A/B SAR interferograms controlled with concurrent altimetry data from the ICESat-2 satellite. Preference was given to the ICESat-2 ATL-06 land ice elevation data recorded on the days with low snow cover on the glacier surface and SAR interferograms with horizontal polarization, short spatial (several tens of meters) and temporal (6 days) baseline. All elevation tracks were corrected for the geoid height (31.8 m at 77.0°N and 16.8°E) and co-registered to SAR interferograms using a straightforward transformation, precise orbits and the Sentinel-SAR sensor model implemented in the ENVI SARscape software. In the resultant composite products every height spot within each altimetric transect is given corresponding interferometric phase and coherence values (See Annex). Such a combination made it possible to reliably estimate the length of interferometric baselines, to determine interferometric phase offsets and to remove them over homogeneous glacier surfaces. In addition, the process of geocoding and visual interpretation of the interferometric motion phase was supported with precise glacier elevation values, which is a luxury for ice bridge surveys. The current height of the ice walls on the flanks of the bridge was measured as 40 m a.s.l. in the west and 50 m a.s.l. in the east. Bathymetric data show the maximum depths of 96 m in Hambergbukta and 93 m in Brepollen, which means that the ice thickness of the front parts of Hambergbreen and Hornbreen does not exceed 130 m and 150 m, respectively. The maximum height of the ice bridge along its lowest axis was measured as 111 m compared to 130 m in 2006. We determined the mean value of the surface lowering in the area of the ice bridge to be 20 m or 1.5 m / a, which is three times the average lowering rate for present-day glaciation in southern Spitsbergen specified by J. Bamber et al. in 2005. The width of the ice bridge along its arcuate lower axis decreased from 8.8 km (2004) to 5.4 km (2016) and 4.8 km (2019). The present minimum width is 4.5 km. Hornbreen's current frontal velocity of 28.8 cm/day, measured in Sentinel-1 interferograms of 2020 with a transferential approach, is slightly lower than that determined in the field at 31.5 cm/day in 2006. The overall length of calving fronts along the bridge’s flanks decreased from 8.2 to 7.8 kilometres, which is still 1.2 times longer than the total width of all inlets feeding the icy isthmus. Under the current climate warming and the significant loss of ice due to calving, surface ablation and basal melting, the H-H ice bridge is melting much faster than the inland glacier parts. It is very likely that there is a narrow subglacial channel that connects the Greenland and Barents seas. In composed interferometric products, points with low interferometric coherence and greatest changes in elevation demarcate the approximate position and configuration of this subglacial strait. The width of the potential strait in the narrowest part is less than one kilometre. The interferometric analysis of SAR data shows that, currently, the entire area of the ice bridge is set in motion as a result of glacier flow and tidal effects. Several small islands and banks support the H-H ice bridge in its central part, which explains the durability of this object and leaves open the question of the possible appearance in the region of a new island (Sörkappland) with a total area of around 1270 km². We discuss the effects of long-term glacioclimatic and oceanographic trends, tidal effects, hydrometeorological conditions and inherent limitations of satellite radar interferometry and lidar altimetry on the validity and value of our results. Some differences in the results of our remote measurements from previous terrestrial surveys in the Mendeleev Plain with ground penetrating radar from A. Pälli (2003), M. Grabiec (2017) et al. are explained by methodological differences and natural changes in the region.
Authors: Sharov, Aleksey (1); Nikolskiy, Dmitry (2)Damage such as deeply crevassed areas and open fractures weakens the structure of an ice shelf. It can promote feedback processes such as ice shelf thinning, leading to speed up and increased shearing, thereby further enhancing damage and it might be an indication for ice shelf disintegration. The complex response of ice shelf to rifting, the difficulty to predict this response and the resulting uncertainties in mass balance models and sea level rise predictions highlight the need for monitoring damage evolution. Nowadays, the majority of Antarctic ice shelves are routinely monitored with optical and radar satellites, providing dense image time series that allow the systematic surveying of fracture opening, propagation and widening. In particular, the acquisition strategy of Sentinel-1 provides a continuous coverage of the margins of Antarctica with a 6-day repeat, which enables the systematic surveillance of fracturing on ice shelves with radar images.Here we show how differential SAR interferometry can also be used for detecting new cracks and tracking crack propagation: active rifts cause different stress fields, flow velocities and tidal responses between two regions on an ice shelf that appear as phase discontinuities in an interferogram. The interferometric phase pattern provides information about the dynamical response of an ice shelf to active cracks which is used to detect stress and early indications for fracturing that are not yet visible in amplitude images. We present an automatic method for delineating ice shelf fractures using Sentinel-1 interferometry. The method relies on the fact that regions of an ice shelf separated by an active crack show fringe patterns with different orientations and fringe rates, and active cracks correspond to spatial discontinuities of the phase gradient that can be mapped using an edge detection procedure. We demonstrate our method to monitor the evolution and propagation of different cracks and fracturing systems of the Brunt Ice Shelf using a dataset of 6-day repeat Sentinel-1 interferograms acquired between September 2020 and March 2021. During this period, a new crack, the North Rift, developed and rapidly propagated eventually resulting in the calving of iceberg A74 (1270 km²) in February 2021. We show the evolution of the fracture that led to the calving event and intercompare the automatically delineated cracks with Sentinel-1 amplitude images and Sentinel-2 data. The proposed method has the potential for detecting precursor signs of iceberg calving, as the quasi full extent of the crack could be detected in interferograms some weeks before the iceberg broke off. This work is carried out in the framework of the ESA Polar+ Ice Shelve Project.
Authors: Libert, Ludivine; Wuite, Jan; Nagler, ThomasConserving the Amazonian rainforest is among the most important goals to preventing severe climate change. In turn, understanding and quantifying forest loss is crucial to predicting future environmental changes. While monitoring the Amazonian ecosystem with optical satellite remote sensing frequently is almost impossible due to cloud cover, frequent acquisitions of SAR data have opened new avenues to detect and characterize forest changes. As a part of the Sentinel-1 for Science Amazonas project (http://project.gisat.cz/s14scienceAmazonas/), we aim to develop, test and validate an operational-level Multi-temporal forest Change Detection (MCD) algorithm using Interferometric Wide (IW) swath Single Look Complex (SLC) time series. We use C-band SAR backscatter and InSAR coherence estimates at a 12-day frequency in dual polarization mode (VV and VH) for detecting forest loss between 2015 and 2020. Methodologically, two approaches were tested to develop a multi-temporal, pixel level, coherence change detection algorithm for Sentinel-1 IW SLC dense time series on the area of Manaus, Brazil - (a) a thresholding and decision tree approach, and (b) a temporal convolutional neural network (tCNN) architecture. The Thresholding and Decision-Tree Approach aims to pin-point discontinuities in the time series of S1 statistical descriptors. Herein, the most important descriptors and change threshold values for them are found to be able to produce binary change maps for each time-point in the S1 time series. A set of statistical descriptors or ‘statcubes’ with same resolution and dimensions for each 12-day period (e.g. R-squared value of linear trend fit on VV-VH backscatter ratio data) are first created. Thresholds are then applied to these statcubes to identify discontinuities in the time series. Preliminary results for the approach show that most changes and roughly half of the change area were detected by the MCD. Only very small events (< 1 ha) had a low detection rate. The false positive rate was 0.3%. The (monthly) temporal accuracy of the MCD was very high with only a few outliers. The Deep Learning Approach, in which the data can be input as raw time series (no human interaction for feature engineering), is aimed at classifying the area of interest into a minimum of continuous forest and yearly deforestation using the full S1 time series information. A train/test/validation dataset that was created based on PRODES yearly deforestation dataset as well as visual interpretation of NICFI Planet high resolution optical data (https://www.planet.com/nicfi/), providing a monthly estimate of the occurrence of deforestation. The overall accuracy of the tCNN approach for the test dataset reaches up to 96% for yearly deforestation detection. Visual interpretation of the test polygons showed that mostly edge pixels of events and single pixels within larger unchanged forest areas were misclassified. Both results also highly depend on the quality of the used forest / no forest masks. Both approaches provided promising first results, with both facing their own but also shared challenges. One big challenge is provided by the decision of what is the true state of a pixel in the Amazonian rainforest. Reference data is often incomplete and hard to improve even with the help of high resolution optical data mosaics. Both approaches are currently being further refined to be applied to the whole Amazonas basin, which will then be the basis for generating biomass loss maps.
Authors: Wittke, Samantha (1,2); Joshi, Neha (3); Karila, Kirsi (1); Puttonen, Eetu (1); Karjalainen, Mika (1)One-year long time series of interferometric coherence measured with Sentinel-1 satellites are exploited as input features for crop classification in a study carried out in the framework of the ESA-funded project SInCohMap. An agricultural area in Sevilla (Spain), in which 17 crop species are present, is employed in this study. The analysis covers different options regarding temporal baselines, polarisation, and combination with radiometric data (backscattering coefficient). The contribution of coherence is interpreted by inspecting the time series: in presence of fully developed crops it is very low due to temporal decorrelation, whereas it increases in the early vegetative stages. In addition, it reaches high values when the ground is bare soil. The difference between VH and VV is due to the different sources of backscattered signal (i.e., direct scattering from ground or from the vegetation volume), as well as to different values of signal-to-noise ratio (SNR), which clearly affects coherence at some stages, especially those characterised by low radar echoes. The time pattern of coherence is associated with the crop calendar (i.e., sowing and harvest dates), which is typical of each crop type, hence helping crop classification. Results show that both radiometric and interferometric features provide notable classification accuracy when used individually (overall accuracy lies between 70% and 80%). It is found that the shortest temporal baseline coherences (6 days) and the use of all available intensity images perform best, hence proving the advantage of the 6-day revisit time provided by the Sentinel-1 constellation with respect to longer revisit times (e.g., 12 days). It is also shown that dual-pol data always provide better classification results than single-pol ones. More importantly, when both coherence and backscattering coefficient are jointly used, a significant increase of accuracy is obtained (greater than 7% in overall accuracy). Individual accuracies of all crop types are increased, and an overall accuracy above 86% is reached. This proves that both features provide complementary information, and that the combination of interferometric and radiometric radar data constitute a solid information source for this application.
Authors: Mestre-Quereda, Alejandro (1); Lopez-Sanchez, Juan M (2); Jacob, Alexander W (3); Engdahl, Marcus E. (4); Yam, Luis (5)The increasing availability of high-resolution Synthetic Aperture Radar (SAR) data has motivated and enabled artificial intelligence approaches, especially for land-cover and land-use from remote sensing data. In this work, we extended the Convolutional Neural Network (CNN) presented in [1] to map forests at a large-scale and monitor the Amazon rainforest using bistatic TanDEM-X data. From TanDEM-X acquisitions, it is possible to derive amplitude as well as coherence images. By exploiting the bistatic interferometric coherence, concretely the volume correlation factor, it is possible to distinguish forested areas from non-vegetated ones, as demonstrated for the generation of the global TanDEM-X Forest/Non-Forest Map, that was based on a supervised clustering algorithm [2]. In the actual study for forest mapping over the Amazon rainforest, the U-Net presented in [1] has been improved by including relevant information on the acquisition geometry as input features, such as height of ambiguity (related to the perpendicular baseline) and the local incidence angle. Moreover, given the special environment presented by the Amazon region, with the presence of many river beds, the U-Net has also been extended for multi-layer semantic segmentation, providing three classes: forest, non-forest, and water. The U-Net has been trained from scratch to avoid any type of transfer learning from previous works, by implementing an ad-hoc strategy with allows the model to generalize well on all different acquisition geometries. Mainly images acquired in 2011 have been used for the training, to minimize the temporal distance to the used independent reference, a forest map based on Landsat data from 2010. TanDEM-X images acquired in 2012 have been considered too, to account for the high variability in the interferometric acquisition geometry. In total, 455 images have been used for training and 320 for testing, covering three ranges of incidence angles as in [2], as well as heights of ambiguity between 20 and 120 m. Moreover, according to the reference data, the considered images present a forest content between 30% and 70% to account for a balanced class training. Extra images have been considered to train the U-Net on the detection of water. By applying the proposed method on single TanDEM-X images, we achieved a significant performance improvement in the test images with respect to the clustering approach developed in [2], with an F-score increase of 0.13. The improvement of classification accuracy makes it possible to skip the weighted mosaicking of overlapping images used in the clustering approach for achieving a good final accuracy at large scale. Moreover, no external references are necessary to filter out waterbodies, as done for the forest/non-forest map in [2]. In this way, we were able to generate three time-tagged mosaics over the Amazon rainforest utilizing the nominal TanDEM-X acquisitions between 2011 and 2017, just by averaging the single image maps classified by the ad-hoc trained CNN. These mosaics can be exploited to monitor forest coverage changes over the years and follow deforestation patterns. By increasing the number of TanDEM-X acquisitions over the Amazonas and applying the trained CNN, it would be possible to perform a near-real-time forest monitoring over selected hot-spot areas. [1] Antonio Mazza, Francescopaolo Sica, Paola Rizzoli, and Giuseppe Scarpa, “TanDEM-X forest mapping using convolutional neural networks,” Remote Sensing MDPI, vol. 11, 12 2019. [2] M. Martone, P. Rizzoli, C. Wecklich, C. Gonzalez, J.-L. Bueso-Bello, P. Valdo, D. Schulze, M. Zink, G. Krieger, and A. Moreira, “The Global Forest/Non-Forest Map from TanDEM-X Interferometric SAR Data,” Remote Sensing of Environment, vol. 205, pp. 352–373, Feb. 2018.
Authors: Bueso-Bello, Jose-Luis; Pulella, Andrea; Sica, Francescopaolo; Rizzoli, PaolaThe availability of short temporal baseline from Sentinel-1 SAR allows to monitor critical crop growth periods using interferometric coherence. The short temporal baseline was not possible in earlier SAR satellite systems. The potential of interferometric coherence has been evaluated earlier for crop growth dynamics. However, many critical phenological stages remain under-evaluated, possibly due to the large temporal baseline. In this study, we have analyzed Sentinel-1 SAR interferometric coherence information at the parcel level for structurally different crops. The openly available crop parcel-level data over The Netherlands is used for the identification of different crops and parcel boundaries. The spatially averaged parcel level interferometric coherence values are calculated from the 6 day pairs of Sentinel-1 in VV and VH polarizations. A total of 32 pairs of Sentinel-1 single look complex (SLC) products of interferometric wide swath data are analyzed during the period of April to December 2019. Structurally different crops such as potato, sugar beet, maize, and winter wheat from the province of Flevoland are used in this study. Initial findings suggest that the coherence values are high (> 0.4) at the early transplanting stage and harvesting period when most of the backscatter response is from the exposed soil surface in VV polarization. Coherence values are relatively lower (
Authors: Kumar, Vineet (1); Huber, Manuel (2); Rommen, Bjorn (2); C. Steele-Dunne, Susan (3)Tropical landscapes are relevant in their contribution to global climate regulation as potential carbon sinks or sources. The sequestered and emitted carbon is commonly approximated with the aboveground biomass (AGB). However, the estimation of AGB is to date mostly focussed on the estimation of one single point in time, where the accuracy is limited, particularly on large spatial scales. The interferometric information (i.e. coherence and height) of high-frequency synthetic aperture radar (SAR) systems are considered particularly useful to estimate vegetation height and AGB. It is frequently assumed that the interferometric height of X-band systems like TanDEM-X represent the canopy surface height. Consequently, these interferometric SAR (InSAR) heights can be combined with terrain information to estimate vegetation canopy height and subsequently AGB. No spaceborne system exists to date to estimate the terrain height consistently on a global scale, and thus the combination of TanDEM-X InSAR height and terrain information is normally limited to small spatial coverages. The potential to estimate AGB and in particular its change with such an approach is limited. In contrast, TanDEM-X InSAR heights can be directly compared over time. The differences can be assumed as differences in the canopy height, assuming that the TanDEM-X InSAR heights represent the canopy surface height at a single point in time. Thus, the canopy height differences estimated by calculating the difference between bi- or multi-temporal TanDEM-X InSAR heights can be related to AGB differences. However, it was frequently found that the X-band signal penetrates into the canopy. This results in the fact that considering the X-band InSAR height as an approximation of the canopy surface height is not correct. More importantly, the penetration depth can differ between different acquisitions depending on the acquisition properties and properties on the ground (e.g. moisture), which would result in pseudo-changes in the difference calculation of InSAR height models at different points in time. In our study, we used two TanDEM-X acquisitions (from 2012 and 2019) covering a dynamic tropical area in Sumatra, Indonesia. We derived the InSAR heights for both acquisition dates individually and calculated their difference. In addition, we assessed the penetration depth of the individual InSAR heights and modelled the penetration to compensate potential pseudo-changes. The absolute accuracy of the individual TanDEM-X heights was assessed with a LiDAR height model used as reference. The TanDEM-X height differences were further related to ground-based AGB estimations from 2012 and 2019. This resulted in a significant linear relationship between the height models and AGB differences, where the penetration compensated height models had a higher coefficient of determination and accuracy compared to the original InSAR heights. However, the accuracy was generally high in both cases with relative root mean square errors below 15%. This suggests that X-band heights from TanDEM-X can be used to estimate canopy height differences and subsequently AGB changes on large spatial scales. However, the differences in penetration depth should not be neglected in order to avoid pseudo-changes and to estimate also small changes linked to land degradation or forest growth.
Authors: Schlund, Michael (1); Kotowska, Martyna M. (2); Brambach, Fabian (3); Wessel, Birgit (4); Camarretta, Nicolò (5); Surati Jaya, I Nengah (6); Erasmi, Stefan (7)Long dormant volcanoes are challenging, both for their hazard and for understanding their physical processes. Nevertheless, their monitoring has experienced a revolution over the past 25 years thanks to the advent of satellite-based remote-sensing methods. Here, we present geodetic observations (interferometric synthetic aperture radar - InSAR time series from SAR sensors onboard the ALOS, ALOS-2, Radarsat-2, and Sentinel-1 satellites) and thermal observations (radiance time series retrieved from the moderate resolution imaging spectroradiometers -MODIS sensors- onboard the Terra and Aqua satellites) of a newly discovered center of unrest at Domuyo volcano, Argentina. Located in the southern Andes, Domuyo is a little-studied 4,700 m elevation volcano constructed of dacitic-rhyolitic domes and flows. Its most recent magmatic eruptions are Pleistocene to possibly Holocene, although it features an active hydrothermal system that has been the epicenter of at least three gas-driven explosions over the last two decades. Our geodetic analysis reveals that Domuyo was the location of gentle to null subsidence from 2008 to 2014, which abruptly began inflating in 2014, continuing into 2019. During the current inflation interval complementary seismicity of low to moderate magnitude has been recorded near the volcanic edifice. Inflation has been roughly linear, ~13-15 cm/year in the satellite line-of-sight (LOS), with an elliptical spatial pattern roughly 30 km in extent. Modeling using the Caltech/JPL-developed AlTar code resolves a sub-horizontal flattened ellipsoidal source at ~6.7 km depth with full axes lengths of 10.6 x 6.7 x 2 km. In contrast, our thermal analysis, based on a new algorithm that captures the diffuse heating of the ground, reveals that the thermal output of Domuyo volcano was relatively steady from 2008 to 2012. Then, it declined at a roughly steady rate (~-0.25 °C/yr) through late 2016, with an abrupt increase of thermal emissions over the past two years (~ 0.40 °C/yr). Our results beg the question: what are the physical mechanisms that can account for simultaneous inflation and reduced thermal output while also accounting for a time lag between both observables? New analysis of the InSAR time series in early 2020 finds that inflation is slowing. We present possible mechanisms based on physical modeling that couples magma injection and gas diffusion through the overlying crust, with either top-down or bottom-up models possible depending on the relative lags between the thermal and geodetic time series. In addition we will extend the thermal and InSAR time series into 2021. We will also present preliminary dynamical modeling using COMSOL Multi-physics of the surface deformation and thermal time series.
Authors: Lundgren, Paul (1); Girona, Tarsilo (2); Bato, Mary Grace (1); Realmuto, Vincent (1); Samsonov, Sergey (3); Cardona, Carlos (4); Franco, Luis (4); Gurrola, Eric (1); Aivazis, Michael (5); Pulvirenti, Fabio (1); Feigl, Kurt (6)Cerro Azul is one of the most active volcanoes in the western Galápagos Islands, but its unrest episodes are poorly studied. Unrest, which started in 2007, culminated in two eruptive phases from 29 May to 11 June 2008. We investigate this unrest and the associated eruptions using interferometric synthetic aperture radar (InSAR) data and geodetic modelling. To overcome the unwrapping errors affecting some of our InSAR data, we propose a new method, based on the wrapped phase differences among nearby pixels, to invert the wrapped phase data directly. Our results highlight how the eruption was preceded by long‐term pre‐eruptive inflation (October 2007–April 2008). During the first eruptive phase, most of the magma responsible for the inflation fed the lateral propagation of a radial dike, which caused a first deflation of the magmatic reservoir. During the second eruptive phase, the further lateral propagation of the dike fed a radial eruptive fissure at the base of the edifice, causing further deflation of the magmatic reservoir. From the first to the second eruptive phase, the radial dike changed its strike propagating toward a topographic low between Cerro Azul and Sierra Negra.
Authors: Galetto, Federico (1,2); Hooper, Andrew (3); Bagnardi, Marco (3,4); Acocella, Valerio (1)We present a retrospective radar backscatter and interferometric phase analysis of Kilauea’s lava lake and new lava flows between 2010 and 2013. Volcanic eruptions change the backscattering properties of the ground surface that can observed and measured by Synthetic Aperture Radar (SAR) backscatter. As radar backscatter does not depend on cloud-free days or solar radiation it has significant advantages for near real-time monitoring and understanding ongoing volcanic eruptions and can aid in the monitoring of remote volcanoes. Radar backscatter has been shown to be useful for detecting dome growth, mapping lava flows and the emplacement of explosive deposits. However, radar backscatter is currently less widely used for volcano monitoring than radar phase measurements. In part this is because of the ambiguity in the data interpretation: there is not a simple link between the magnitude or sign of the change in radar backscatter and the physical properties of the fresh volcanic deposits. We present an analysis of a temporally dense dataset of high-resolution SAR images to demonstrate the applicability of SAR backscatter for observing and measuring the effusive activity at Kīlauea, Hawai’i between 2010 – 2013. During this eruptive period there were three characteristically different lava flows; Kamoamoa Fissure Eruption (Episode 59, March 2011), a breakout in August 2011 (Episode 60) and the Peace Day flow (Episode 61, September 2011) accompanied by inflation and collapse events at Puʻu ʻŌʻō crater. At Puʻu ʻŌʻō crater, we test the geometrical and morphological limitations of using radar shadows to extract accurate height measurements and investigate the relationship between lava lake expression in radar backscatter and the characteristics of lava lake activity. We demonstrate the use of principal component analysis and speckle filters to reduce noise in the backscatter and aid in detecting volcanic activity changes in timeseries. Through the construction of a simple model for backscatter change, we track the emplacement of the lava flows, and the development and cooling down of the flow field. We integrate our backscatter analysis with maps of interferometric phase coherence, and show that backscatter provides unique information about flow development, especially during the period of post-emplacement phase incoherence.
Authors: Dualeh, Edna W. (1); Ebmeier, Susanna K. (1); Wright, Tim J. (1); Poland, Michael P. (2)On 12 January 2020, the Philippine Institute of Volcanology and Seismology (PHIVOLCS) raised the alert status of Taal volcano from level 2 at 14h30 PST (Philippine Standard Time) to level 4 by 19h30 PST due to the rapid escalation of the volcanic activity, the latter meaning a hazardous eruption is imminent within hours to days. At the time of writing this abstract, the alert level has been lowered to 3 as the eruptive activity has begun to wane. Here we will present: 1) pre-eruptive interferometric synthetic aperture radar (InSAR) time-series from Sentinel-1, ALOS and ALOS-2 data between 2007 and 2019, and 2) co-eruptive displacements derived from both InSAR and/or pixel-offset analyses starting 9 January 2020. On one hand, we are able to constrain the geometry and the location of the deformation source using the pre-eruptive time-series and the Caltech-JPL-developed AlTar v2.0 Bayesian inversion software. Preliminary results show an inflating prolate (cigar-shaped, long-axis sub-horizontal) body located at ~4-5 km depth, beneath the NE portion of Volcano island. On the other hand, from the co-eruptive dataset, we inferred a deflating source beneath NE of the Volcano island and a roughly 21 km x 8.6 km vertically dipping dyke with ~3.2 m opening that extends southwestward from the centre of Taal caldera. An important aspect of this work is the near-real-time deliverance of the processed InSAR data, model and analyses to PHIVOLCS while the eruption was happening. Deformation data are correlated to ongoing uplift and subsidence particularly in the observed ground manifestations near the coastal areas. We will highlight the significant role of open-access and timely acquired remote sensing data as well as discuss the potential “adverse effect” during an on-going volcanic crisis.
Authors: Bato, Mary Grace (1); Lundgren, Paul (1); Pinel, Virginie (2); Solidum, Renato (3); Daag, Arturo (3); Cahulogan, Mabelline (3)We present early results from the Committee on Earth Observing Satellites (CEOS) Volcano Demonstrator, including InSAR measurements of volcanic deformation and quantitative information extracted from SAR backscatter. Hazards from volcanic eruptions pose risks to the lives and livelihood of local populations, with potential global impacts to businesses, agriculture, and air travel. The 2015 Global Assessment of Risk report notes that ∼800 million people are estimated to live within 100 km of ∼1400 subaerial volcanoes identified as having eruption potential. However, only 55% of these volcanoes have any type of ground-based monitoring. The only methods currently available to monitor these unmonitored volcanoes are space-based systems that provide a global view. CEOS developed a 4-year pilot project (2013-2017) to demonstrate how satellite observations of ground deformation can be used to monitor large numbers of volcanoes cost-effectively, particularly in areas with scarce instrumentation and/or difficult access. One component of the CEOS volcano pilot was to systematically observe the ∼320 Holocene active volcanoes within Latin America with the goal to link the volcano observatories that are governmentally responsible for volcano monitoring with data providers at the international space agencies (ESA, CSA, ASI, DLR, JAXA, NASA, CNES, CONAE) and remote sensing experts who analyze the data. Starting in October 2019, the 3-year Volcano Demonstrator project encompasses target volcanoes with high risk and the greatest need for remote sensing in Latin America, Southeast Asia, and Africa — encompassing about half of the world’s potentially active subaerial volcanoes. The CEOS demonstrator links diverse volcano observatories around the world, international space agencies and remote sensing experts. The demonstrator is particularly focussed on providing high resolution SAR datasets (e.g., CSK and TSX) to volcano observatories and collaborating researchers. Using examples from recent eruptions and unrest in Indonesia (Merapi and Sinabung), the Caribbean (St Vincent) and South America (Sangay), we demonstrate the value of using the whole constellation of civilian SAR systems, and especially high resolution SAR imagery. This is particularly important in the context of increasingly available automatically processed Sentinel-1 imagery. One goal of including more regions is to understand the heterogeneous needs of volcano observatories around the world given their diverse capabilities as well as volcanoes with different manifestations of activity and environmental conditions. Here we discuss cases where the demonstrator scientists have made deformation measurements and observations that have been used by observatories to monitor volcanoes, complement/validate small monitoring networks and respond to crises.
Authors: Ebmeier, Susanna (1); Poland, Mike (2); Pritchard, Matt (3); Biggs, Juliet (4); Hamling, Ian (5); Aoki, Yosuke (6); Delgado, Francisco (7); Albino, Fabien (2); Dualeh, Edna (1); Bemelmans, Mark (4); Espin Bedon, Pedro (1); Grandin, Raphael (9); Lundgren, Paul (8); Wauthier, Christelle (10); Amelung, Falk (11); Shreve, Tara (12); Danzeglocke, Jens (13); Zoffoli, Simona (14)Mangrove forests are extremely important for the coastal stabilization by reducing erosion that occurs through currents, storm surges, tides and waves. The complex root system of mangroves also makes its forests serve as an attractive nursery to many fish species and organisms seeking nutriment and sanctuary against predators. However, until now there has been little effort to map, monitor and observe the mangrove forests along the Red Sea coasts and compare the rehabilitation results and/or experiences and to analyse what makes a mangrove (re-)development sustainable. In this study, we have been working on mapping the mangrove forests using remote sensing and artificial intelligence techniques to better understanding what the ecosystem conditions of mangroves, in order to detect which mangroves are intact versus, which mangroves might be degraded, also the environmental drivers that could be affect the different types of mangrove forest and what this means for its ecosystem. This study conducted analyses to locate the mangrove forests sites along the Sinai peninsula coasts to support the implementation of restoration projects. It was conducted to map and monitor the mangrove forest areas via satellite imagery “Sentinel1 and Sentinel2”, geographic information system “GIS” and artificial intelligence “AI” techniques, and try to assess the potential benefit of these methods using remote sensing for the management and restoration of mangrove forests along the coasts. The end of this study will understand how to create a mangrove extent map using a random classification and create a time series for mangrove extent change. We will use google earth engine “GEE” as an cloud-based geospatial processing platform followed by a time series analysis, then a demo of performing a time series analysis for Nabq section coasts as an example study site, then go over to set up and map it using satellite imagery and filter a sentinel composite, then will be constructing a random forest model, also doing a time series comparison and new random forest classification for that comparison, then go through calculated mangrove area from our resulting maps. Compared with in-Situ surveys, remote sensing provides a synoptic view and, using the artificial intelligence techniques, makes it possible to look at accurate data and present an interactive map of mangrove forests.
Authors: Adel, Mohamed (1,2); Asser, Aya (1); Beheary, Mokhtar (2)The forests of Central Yakutia (Sakha Republic, eastern Siberia, Russia) are vulnerable natural ecosystems that undergo both technogenic and natural transformations. On the one hand, forests are subjected to a high anthropogenic impact, manifested through a large number of deforested areas on the territory of the Yakutia (Sakha) Republic. Deforested areas lead to changes in the upper layer of permafrost. On the other hand, it is the occurrence of many wildfires that significantly change forest ecosystems in a short time. For example, in July 2019, the area of the wildfires in Yakutia was 1.13 million hectares, while in other regions of Russia in total was 1.56 million hectares. After wildfires in the forests, their structure and composition are completely transformed, and then stabilize slowly towards the original ecosystem. Each stage of the post-fire succession is characterized by the presence of certain plant communities in the forest cover, which are gradually replaced by a more typical plant species with certain life cycles and ecology. This is the basis for distinguishing natural and disturbed plant communities in satellite images. Traditional data sources for the studies of the state of forests are optical satellite images and, in recent years, images from the unmanned aerial vehicles. Radar images are less common, but they can provide additional information about the current state and structure of vegetation. The availability of Sentinel-1 SAR imagery allows us to analyze the dynamics of the forest ecosystems and assess the suitability of using C-band radar images to study the dynamics and structure of the forest. Our research objectives were to assess the current state of the Central Yakutian forests, especially the wildfires areas, through a comprehensive analysis of remote sensing data of different spectral ranges and spatial resolutions, validated against ground-based measurements, to accurately assess the state and structure of the forest communities. This study is based on Sentinel-1 SAR C-band data (SLC and GRDH products) and is complemented with results of our field studies in July-August 2019. The field studies included both visual and instrumental measurements of forest ecosystems, as well as UAV surveys. As a result of field measurements, detailed descriptions of the structure and tree species of forests were obtained for the selected key sites (11 in total). As additional sources of information, we used images of the optical range (Sentinel 2 MSI) and full-polarimetric ALOS-2 PALSAR-2 L-band data (acquired on 27/07/2019). The main methods of data analysis were multitemporal Sentinel-1 data (GRDH product) analysis, including analysis of the seasonal сhange of backscattering coefficient values (σ0) and coherence analysis. The seasonal cycle of change of the σ0 demonstrated that σ0 values for deforested and burnt areas differ significantly from healthy forest ecosystems. The difference can reach 10 dB in the co-polarization (VV). The σ0 values of larch forests are slightly higher than those of pine forests. Birch forests do not differ significantly from larch forests in its values. Based on the results of coherence analysis of Sentinel-1 SAR data and decomposition of ALOS-2 PALSAR-2 images, the main types of forest communities in Central Yakutia were identified. The principle of using coherence analysis is based on the fact that forest ecosystems with different species composition have different seasonal variability of coherence values. The analysis of coherence values was performed based on consecutive pairs of Sentinel-1 SAR images (SLC product) for the summer and winter period. Weather conditions were analyzed on the day before and on the day of acquisition during SAR image selection. It was found that for winter images of coherence values, the boundaries of forest ecosystems are almost not defined. Values of coherence for summer pairs allow to separate birch and deciduous forests from pine and wildfire spots. Unfortunately, a more detailed classification of forest types by species composition and age was not possible at the current stage of the study. The results of decomposition of full-polarimetric ALOS-2 PALSAR -2 data were used to analyze the state of forests. Such a comprehensive analysis provide significantly more information about the types of forests, its condition and age (in the case of wildfires and deforested areas) in the territory of Central Yakutia. Thus, the first results of assessing the suitability of radar data for analyzing the forest state in Central Yakutia were obtained. It was found that the analysis of the multitemporal C-band data (Sentinel-1 SAR) allows us to identify different types of forest ecosystems, as well as wildfires and deforested areas of different ages. The coherence analysis methods have shown that the boundaries of the different forest ecosystems are poorly defined based on coherence values only. Using L-band radar data allows a more detailed analysis of the structure and forest state. For a comprehensive analysis of the state of forests, it is necessary to use radar data of different frequency ranges. *Field data acquisitions were supported by the he Russian-British project “Multiplatform remote sensing of the impact of climate change on northern forests of Russia”, funded by the British Council (Grant no. 352397111) and the Ministry of Science and Higher Education of the Russian Federation (project RFMEFI61618X0099). Radar imagery analysis was funded by RFBR according to the research project № 18-05-60121-Arctic Ram Avtar and provision of the ALOS 2 PALSAR-2 data are supported by the Human Resource Development Platform for Japan-Russia Economic Cooperation and Personnel Exchange (HaRP Program)
Authors: Mikhaylyukova, Polina (1); Tutubalina, Olga (1); Sozontova, Anna (1); Avtar, Ram (2)Every year the Ministry of agriculture of Russia implements monitoring of field works (crops sowing, harvesting, and etc.) over the whole country for regulation and management of agricultural markets. The main way of this information acquisition is its collection from agricultural producers and then sequential generalization at the municipal, regional and, finally, federal levels. Different subjective factors may affect the reliability of this information, so, a search for tools of its verification becomes an important task. Satellite remote sensing methods can be appropriate for this purpose. In this study a time series of interferometric coherence was examined as a source of information about the agricultural activity within the fields because its well-known sensitivity to the changes of the surface between two observations. Amur Oblast (Russian Far East) was selected as an area of interest. The main crops here are soy (~80% of the sown area in the region) and cereals (spring wheat, spring barley, corn and others). A multi-temporal data set of complex (SLC) VV-polarization Sentinel-1 scenes obtained from two orbits (32 and 105) in descending geometry was selected. The data was acquired during the snow-free period (April-October) of 2019. The shortest interval between one orbit acquisitions was primarily 12 days, and 5-7 days between acquisitions from both orbits. Interferometric coherence was estimated between consecutive acquisition dates for both orbits. Comparison of coherence time series and a set of optical images (Sentinel-2, Landsat-8/-7, Terra) showed that the behavior of the former over the agricultural fields can be in general described as follows: 1) High coherence values are observed over the fields with bare soil which was not disturbed between two acquisitions; 2) Temporal decorrelation and, as a result, low coherence values (0,3 and less) are observed over the fields covered by vegetation in both dates or over the fields where significant changes have occurred between acquisition, such as crops harvesting, pre-sowing and after-harvesting soil ploughing, and etc. A map of crops harvesting paces was created for a test site based on the coherence dynamics which showed a good agreement with the base map created using the frequent (2-8 days) Sentinel-2 observations. Generally, Sentinel-1 interferometric coherence seems to be a promising instrument which allows agricultural activity revealing with high accuracy (less than 6,5 days for most of the agricultural lands of Russia). The material is prepared on the topic of the state task No. AAAA-A19-119022190168-8 and supported partly by RFBR (project No. 18-07-00816).
Authors: Troshko, Ksenia A. (1,2); Denisov, Pavel V. (2)ESA’s Geohazards Exploitation Platform (GEP) [1] provides the scientific research community with a body of robust tools based on established techniques of Synthetic Aperture Radar (SAR) image processing, that enable users to select Sentinel-1 Interferometric Wide swath images from the available collection, and run processing jobs without the need to download the data on local computers and use in-house computing resources. This paper focuses on the opportunities and perspectives that SAR-based change detection processing tools – such as the SNAC - SNAP S-1 GRD Amplitude Change [2] and COIN – Coherence and Intensity change for Sentinel-1 [3] –, currently open not only to support specialist studies of natural hazards, but also for exploitation of Sentinel-1 data and derived products by a wider user community. SAR-based change detection tools are generally more intuitive than differential and multi-interferogram SAR processing techniques, and could be more easily accessible and exploited by less expert users. As such, they could represent an appealing computing facility for a wider GEP user community. For example, environmental scientists interested in mapping land surface changes, and niche communities like archaeological remote sensing scientists that are more used to proximity and/or airborne surveying techniques to investigate cultural landscapes and assess the impact of natural and anthropogenic hazards on local heritage. However, the use of SAR change detection techniques is still limited mostly to SAR experts, given the lack of familiarity of other communities with SAR data in general [4, 5]. This also matches with the user feedback that we collected during session ‘D2.05: Cultural & Natural Heritage’ at the last ESA Living Planet Symposium 2019 in Milan. Delegates coming from more humanities-related backgrounds [e.g., 6] called for easier accessibility to existing satellite data, archived data and products and processing routines to increase the capability of users to work on satellite images (particularly SAR data) and generate value‐added products. In this context, as part of the GEP Early Adopters Programme and the Geohazards Lab initiative, we have run tests for detection and mapping of natural hazard events (e.g. flash-floods) that occurred in the last couple of years across different locations in the Middle East. In this geographic region, natural hazards are currently gathering less attention than anthropogenic hazards and threats related to the current conflict situation. Nevertheless, we have found evidence in satellite imagery suggesting that archive Sentinel-1 images have captured events, and therefore can provide a valuable stock of geospatial information for documentation and assessment of impact on local settlements and infrastructure. We will show how the short temporal revisit of Sentinel-1 Interferometric Wide swath mode data, alongside the spectral analysis of coeval cloud-free Sentinel-2 optical images, can allow for the investigation of two main types of situations [7]: (i) events that were poorly documented in either scientific or grey literature, as well as in social media and international information networks; and (ii) events for which no documentation or on-the-ground evidence is available. We will discuss these use-cases that have been designed, also accounting for the constraints that practitioners coming from different backgrounds have already reported in literature. For example, prior to the launch of most of the Sentinel satellites, the time and effort required to download, process and analyse satellite imagery were perceived as obstacles for near real-time satellite monitoring [8]. In more recent times, accessibility to near real-time free satellite imagery has become a key factor to significantly reduce the overall timeframe of data analysis [9] and has therefore led to increased data exploitation to monitor hazard events. Through the discussion of the results, we aim to demonstrate the real opportunities for future GEP users. REFERENCES [1] Foumelis M., Papadopoulou T., Bally P., Pacini F., Provost F., Patruno J. 2019. Monitoring Geohazards Using On-Demand and Systematic Services on ESA’s Geohazards Exploitation Platform. Proceedings of the IEEE International Geoscience and Remote Sensing Symposium - IGARSS 2019, 5457-5460. doi: 10.1109/IGARSS.2019.8898304. [2] Terradue 2020. SNAC - SNAP S-1 GRD Amplitude Change — Geohazards Thematic Exploitation Platform 2.1 documentation. Available at: https://terradue.github.io/doc-tep-geohazards/tutorials/rss_snap_s1_snac.html [3] Terradue 2020. COIN – Coherence and Intensity change for Sentinel-1 — Geohazards Thematic Exploitation Platform 2.1 documentation. Available at: https://terradue.github.io/doc-tep-geohazards/tutorials/rss_snap_s1_coin.html. [4] Tapete D., Cigna F. 2017. Trends and perspectives of space-borne SAR remote sensing for archaeological landscape and cultural heritage applications. Journal of Archaeological Science: Reports 14, 716–726. doi: 10.1016/j.jasrep.2016.07.017. [5] Cuca B., Hadjimitsis D.G. 2017. Space technology meets policy: An overview of Earth Observation sensors for monitoring of cultural landscapes within policy framework for Cultural Heritage. Journal of Archaeological Science: Reports, 14, 727–733. doi: 10.1016/j.jasrep.2017.05.001. [6] Rączkowski W., Ruciński D. 2019. Cooling down enthusiasm: potential vs. practice in application of EO techniques in archaeological research and heritage management – have lessons been learned? ESA Living Planet Symposium, Milan, 17 May 2019. [7] Tapete D., Cigna F. 2020. Poorly known 2018 floods in Bosra UNESCO site and Sergiopolis in Syria unveiled from space using Sentinel-1/2 and COSMO-SkyMed. Scientific Reports, 10, 12307. doi: 10.1038/s41598-020-69181-x [8] Witmer F.D.W. 2015. Remote sensing of violent conflict: eyes from above. International Journal of Remote Sensing, 36(9), 2326–52. doi: 10.1080/01431161.2015.1035412. [9] Zwijnenburg W., Hochhauser D., Dewachi O., Sullivan R., Nguyen V.-K. 2019. Solving the jigsaw of conflict-related environmental damage: Utilizing open-source analysis to improve research into environmental health risks. Journal of Public Health, fdz107. doi: 10.1093/pubmed/fdz107.
Authors: Tapete, Deodato; Cigna, FrancescaThe collaborative research project ALGORITHMS (2019-2022) between the Italian Space Agency (ASI) and the Institute of Applied Physics of the National Research Council of Italy (IFAC–CNR) aims to develop innovative algorithms to estimate the main hydrological parameters (e.g. soil moisture content, vegetation properties, snow water equivalent) [1]. In particular, ALGORITHMS focuses on the estimation of geophysical parameters that can feed into the analysis of the entire hydrological cycle and relate to: (a) soil and agricultural-forest vegetation, to monitor cultivations; (b) snow, with specific regard to extent and thickness of snowpack and its wet/dry conditions. The novelty of the project lies in the experimental testing of algorithms on instrumented sites located in northern and central Italy that provide an assorted distribution of agricultural, grassland and pasture land cover types, as well as long-lasting snow cover, and are covered by very long Synthetic Aperture Radar (SAR) time series with short revisit time (up to 1 day in the case of COSMO-SkyMed) and different polarizations. It is in this regard that the present paper aims to showcase the value of using VV-pol. Sentinel-1 Interferometric Wide (IW) swath, HH-pol. COSMO-SkyMed Stripmap Himage, VV/VH and HH/VV dual-pol. COSMO-SkyMed Stripmap PingPong SAR imagery, to improve the existing capabilities for the retrieval of: (i) Snow Water Equivalent (SWE) in Alpine areas; (ii) crop phenology and (iii) land cover classification in agricultural zones; and (iv) Soil Moisture Content (SMC) mapping at regional scale. We therefore aim to present the results of the first year and half of the project, with regard to the following research streams: (i) sensitivity analysis of X-band σ° to SWE, and tests of SWE retrieval with Support Vector Regression (SVR) [2] and Artificial Neural Network (ANN) algorithms, exploiting the whole archive of 2013-2015 HH-pol. COSMO-SkyMed Stripmap Himage winter images over the South Tyrol vs. the nearly simultaneous in-situ SWE measurements kindly provided by the Hydrographic Office of the Province of Bolzano; (ii) land cover/crop type classification by means of an ensemble of Convolutional Neural Network (CNN) classifiers operating in the time-sensor domain (each independently returning its prediction and the most voted class being returned as the prediction of the ensemble), that have been developed also accounting for the lesson learnt on machine learning methods tested on forests [3]. CNN are tested on VV-pol. Sentinel-1 IW, Sentinel-2 (Bands 3 to 8), and HH-pol. COSMO-SkyMed Stripmap Himage datasets vs. available ground-truth maps; (iii) analysis of the temporal trends of σ° from Sentinel-1 VV-pol., COSMO-SkyMed Stripmap Himage and HH/VV dual-pol. COSMO-SkyMed Stripmap PingPong across an assorted selection of surveyed crops (wheat, sunflower, corn, beans, sorghum) vs. precipitation data, soil and vegetation properties collected in-situ at the time of satellite imagery acquisitions, and bespoke airborne radiometric surveys; (iv) sensitivity analysis of σ° to SMC and Plant Water Content (PWC) for the full range of VV-pol. Sentinel-1 IW, HH-pol. COSMO-SkyMed Stripmap Himage and HH/VV dual-pol. COSMO-SkyMed Stripmap PingPong datasets collected during ALGORITHMS, alongside tests and validation of SMC maps of improved spatial resolution and enhanced accuracy, generated by merging SMAP and Sentinel-1 data through ANN methods. This work is carried out by ASI, IFAC-CNR and EURAC, in the framework of the 2019-2022 project ‘Development of algorithms for estimation and monitoring of hydrological parameters from satellite and drone’, funded by ASI under grant agreement n.2018-37-HH.0. REFERENCES [1] Tapete D., Cigna F., Paloscia S., Santi E., Pettinato S., Fontanelli G., Chiarito E., Notarnicola C., Cuozzo G., Jacob A., De Gregorio L., Rossi M. 2020. Development of algorithms for the estimation of hydrological parameters combining COSMO-SkyMed and Sentinel time series with in situ measurements. 2020 IEEE Mediterranean and Middle-East Geoscience and Remote Sensing Symposium 2020 - M2GARSS 2020, 9-11 Mar 2020, Tunis, Tunisia, pp. 53-56. doi: 10.1109/M2GARSS47143.2020.9105313 [2] De Gregorio L., Cigna F., Cuozzo G., Jacob A., Paloscia S., Pettinato S., Santi E., Tapete D., Bruzzone L., Notarnicola C. 2019. SWE retrieval by exploiting COSMO-SkyMed X-band SAR imagery and ground data through a machine learning approach. Proc. SPIE 11154, Active and Passive Microwave Remote Sensing for Environmental Monitoring III, 111540M, 8 Oct. 2019, doi:10.1117/12.2550824 [3] Lapini A., Pettinato S., Santi, E., Paloscia S., Fontanelli G., Garzelli A. 2020. Comparison of Machine Learning Methods Applied to SAR Images for Forest Classification in Mediterranean Areas. Remote Sensing, 12(3), 369. doi: 10.3390/rs12030369
Authors: Tapete, Deodato (1); Cigna, Francesca (1); Paloscia, Simonetta (2); Santi, Emanuele (2); Pettinato, Simone (2); Fontanelli, Giacomo (2); Lapini, Alessandro (2); Chiarito, Eugenia (2,3); Notarnicola, Claudia (3); Cuozzo, Giovanni (3); Jacob, Alexander (3); De Gregorio, Ludovica (3); Rossi, Mattia (3)As one of the effective tools for monitoring ground surface deformation, Interferometric Synthetic Aperture Radar (SAR, InSAR) is characterized by high-resolution, large-scale, high-precision, all-weather and all-day. However, InSAR is very sensitive to vegetation coverage on the ground since serious decorrelated noise would be caused by the vegetation changes during the time interval of SAR images. Therefore, the accuracy of the InSAR deformation monitoring results in the vegetation area cannot be guaranteed. In recent decades, many SAR satellites had been launched, which provide a wide space for the data selection in monitoring surface deformation. In addition, the parameters, such as polarization modes, incident angles and spatial resolutions, can be customized by the users. Integrating multi-temporal, multi-orbit, multi-frequency and multi-polarization SAR data will be a trend for monitoring surface deformation. Therefore, it is quite useful to assess the performance of SAR data prior to its acquisition for achieving the optimal configuration of SAR data. As a co-product of InSAR, coherence is generally used as the index for assessing the decorrelation of InSAR measurement. In vegetation area, the decorrelation is mainly contributed by the temporal decorrelation caused by vegetation change. Therefore, it is natural to establish the relationship between vegetation cover and coherence. Recent studies had demonstrated that coherence was affected by height and density of vegetation, and a non-linear relationship could be found between coherence and some vegetation indices. However, the quantitative relationship between coherence and vegetation coverage is still not clear. Normalized Difference Vegetation Index (NDVI), which can be retrieved from optical satellite images, is a commonly used measurement of vegetation coverage. It had been found that the InSAR coherence is highly related to the NDVI. This provides an opportunity for mapping InSAR decorrelation with NDVI. In this study, the quantitative relationship between InSAR coherence and NDVI is determined by using the ALOS-1/PALSAR-1 and Landsat5 images acquired in Meitanba area, Hunan Province, China. Since it is difficult to find the relationship based on the raw data due to the gross errors, a sampling procedure is proposed based on a correlation estimating model for the extraction of observation samples from InSAR coherence and NDVI. A novel three-stage model is then established for mapping InSAR coherence with NDVI, which reveals the interaction between InSAR decorrelation and vegetation coverage. The performance of the proposed model is finally evaluated by using the similar data in Longhui area, Hunan Province, China. The results demonstrate that the novel three-stage model can be used to quantitatively estimate the InSAR coherence of L-band ALOS-1/PALSAR-1 data by exploiting the NDVI derived from Landsat5 data.
Authors: Sun, Qian (1); Chen, Yaogang (2); Hu, Jun (2)Climate change (on a global scale) has already had observable effects on the environment. The glaciers have shrunk, the ice in the lakes is breaking earlier, the animal hibernation period has been modified and the trees are blooming sooner than expected. The effects that scientists had predicted in the past and that are being observed are: loss of sea ice, accelerated sea level rise and longer and more intense heat waves, among others (NASA, 2019). Costa Rica, due to its geographical location (in the intertropical zone) is located in one of the hot spots for the effects of climate change. In recent years, events such as droughts in the Guanacaste area or floods in the southern and Caribbean areas of the country are becoming more frequent. This has direct repercussions on the quality of life of the population, since there is a loss of crops, which influences the price of food or water rationing due to the scarcity of it. It has also been observed that the country is affected by natural events that did not previously impact, such as the direct passage of Hurricane Otto in 2018. After the bibliographic review made by the proponents, it was determined that the application of radar image information is scarce in Costa Rica, except for some specific examples, such as mapping the forest coverage in Parque Nacional Corcovado and in the Mangrove of Bahia Drake , using the NASA EcoSAR sensor in 2014. In order to fully harness the products of the Sentinel 1 satellite, a research project was initiated in 2020 at Universidad Nacional in Costa Rica with the goal of identifying the change in land use at sites of interest using radar images, which allows the analysis of variations in urban areas, natural coverages and water bodies. Also within the project there is an objective associated with the determination of the speed of vertical changes in the land surface in the South Caribbean of Costa Rica, through the application of the PSI technique, with the purpose of contributing to understanding the local geodynamic processes, which is consider to be influencing other processes such as coastal erosion in this region. To obtain the results of the mentioned phenomena, SLC images will be used, which will be processed using the SNAP and StaMPS program. In order to remove the atmospheric phase, the application of TRAIN will be used. Likewise, the GRD products will be used for the analysis of the change of the bodies of water, specifically applied to the analysis of the variations in the surface of Lago Arenal, due to its importance as a source of electricity generation and contribution to the system of irrigation for the Guanacaste area. NASA. (2019). What is Climate Change? (S. Callery, Editor, & NASAs Jet Propulsion Laboratory) Recuperado el 21 de Marzo de 2019, de https://climate.nasa.gov/
Authors: Paniagua Jiménez, Diana Ninette; Valverde Calderón, Jose Francisco; Barrantes Castillo, GustavoMt. Etna is located in the eastern part of Sicily and it can be considered as the largest and most active volcano in Europe. On 16 February 2021, the activity of Etna volcano clearly intensified, with lava fountaining and strong ash emissions reaching a height of about 10 km. Since then, the volcanic activity has been increasing esponentially, with up to 12 paroxysm events in less than one month. Many of the surrounding municipalities have been facing damages for the high quantity of ash, which spread over buildings, fields and streets. For the past decades, satellite SAR Interferometry (InSAR) has proven to be a relevant tool to monitor wide areas interested by surface dynamics. From slow millimetric-scale phenomena triggered by human activities to sudden sub-metric and centimetric deformation phenomena related to natural hazards, InSAR can provide both temporal series information and relative single measurement related to such deformation events. The InSAR ability to provide reliable deformation measures depends entirely on the stability of the scattering measurements. In particular, Persistent Scattering Interferometry (PSI) approaches evaluate the behavior of the stable points to be analyzed using the temporal dimension, usually exploiting the statistical nature of either the coherence or the amplitude. In the case of classic InSAR approaches only static information is obtained as a measurement of the quality of the scattering mechanisms. Furthermore, in addition to surface displacement estimation, some derived features from InSAR have proven to contain relevant information in regards of the land cover content and variation, giving insights regarding those areas that can be difficult to be reached. Specifically, coherent change detectors proved to be an important tool for dealing with disaster monitoring, such as flooding, burning areas. Volcanic eruptions are among those sudden events where rapid change of the surface occurs. Thus, in order to have a more reliable information related to a particular event been analyzed with InSAR, in this work it is proposed to evaluate the impact of exploiting a coherence change detector in combination of classic InSAR. In particular, the dataset chosen for the analyses will be Sentinel-1A/B Interferometric Wide (IW) acquisitions in both orbits ascending and descending, tracks 44 and 124, respectively. Thus, single interferograms with very short temporal baseline will be performed, with the aim of exploiting areas with very rapid surface changes. This information can be then used in future analyses to perform PSI method, focusing on stable points and thus easing pure deformation estimation.
Authors: Orlandi, DianaForests play a vital role in maintaining a balance in the ecosystem, majorly through carbon sequestration in order to mitigate climate change. Therefore the analysis and estimation of biophysical parameters like aboveground biomass of forest areas are of prime importance. Aboveground biomass constitutes the leaves, branches and trunk of trees. Microwave signals are being used to retrieve backscatter information and coherence contribution of natural targets. Synthetic Aperture Radar (SAR) sensors transmit microwave signals and receive the backscattered radiation from the targets on the Earth’s surface. These backscatters can be decomposed in order to extract the individual scattering mechanisms from the targets within each SAR resolution cells. In this paper, we used polarimetry and interferometry coherently for a better analysis of the target that is, the forest region. We have identified the coherence channel, which is best suited for the estimation of biophysical parameters of forests by analysing the dependency of coherences of different polarization basis with respect to ground-to-volume scattering ratio for forest region. This analysis identifies the coherences present in the dominance of ground or volume backscatter contributions. A pair of fully polarimetric C-band Radarsat-2 data acquired in January and February 2019 with a temporal baseline of 24 days was used. The study area selected for this work was Malhan Range of Dehradun Forest Division, Uttarakhand, India. The ground-to-volume scattering ratio was estimated from the double-bounce, and volume scattering powers retrieved as a result of the PolSAR based Yamaguchi four-component decomposition modelling. And, the coherences for different polarization basis were extracted from the PolInSAR based complex coherence images. For the coherences in the linear basis, HV coherence was observed to have a lower range in the y-axis compared to the HH and VV coherences. This behaviour indicates the decrease in coherence due to higher volume decorrelation for randomly oriented dipole scatterers. In the Pauli basis, the distribution of HV+VH coherence was similar to that of HV coherence. Whereas, the HH-VV coherence was found to be centred at the 0 dB value and independent of the ground-to-volume scattering ratio. And, the optimal coherences were observed to have an equiprobable distribution with no dependence on the ground-to-volume scattering ratio. Only the coherence range was observed to decrease in the case for optimal coherences from optimal-1 to optimal-2 and finally, it was the least for optimal-3. The cross-pol coherences HV and HV+VH were identified to represent the biophysical parameters best as they have less sensitivity to ground scattering for a forest region. Therefore, these coherences were considered to be suited for further analysis of the biophysical parameters for forest areas. It was also observed that the co-pol coherences, HH, VV, HH+VV and HH-VV had higher range of values representing stable scatterers. The coherences were also observed to change in different polarization channels, even for the same target media.
Authors: Mukhopadhyay, Ritwika (1,2); Kumar, Shashi (1); Aghababaei, Hossein (2); Kulshrestha, Anurag (2)Urban settlements of big cities called metropolises are growing uncontrolled all over the world. It is known that there are many reasons for this. Many reasons such as in-border and cross-border migrations, relocate for universities, shopping centres, entertainment venues, concert halls and fairgrounds, which attract people, are in big city centres, causing the already crowded big cities to become more crowded. The rapid growth of urban areas of more crowded cities is also increasing disproportionately. As of December 31, 2020, Turkey's population amounted to 83 million 614 thousand 362 people. The proportion of those living in provincial and district centres in Turkey was 93% in 2020. Antalya is one of the important metropolises where crowded life progresses rapidly. The city of Antalya has grown more than 25 percent in population in the last ten years. This rate has made Antalya Turkey's most high-immigration cities. Antalya city in Turkey has been seen as the most appropriate to address the urban boundary's change. Determining urban areas' boundaries is necessary (from the point of view) information to public authorities. In light of this information, cities' infrastructure and superstructure needs can be seen precisely and clearly. It provides significant contributions to municipalities' town and regional planning works, following the zoning plan applications and determining illegal urbanization. Among the different methods of deciding urban area boundaries, the most robust SAR data is considered Interferometric Coherence. In this study, border extraction has been carried out by using multitemporal SAR data to determine the uncontrolled growth in Antalya city centre and district centres, monitor it, and take precautions if possible. In the study where Sentinel-1A (Band C) data were used, Interferometric Coherence for each year was produced and exported as raster data. It was possible to have information about Antalya city centre by determining the visual changes from the city borders data of 3 different years. As a result of the study, how much and in which directions urban boundaries occur been put forward. By using higher resolution SAR data in future studies and matching the urbanization plans to be obtained from the municipalities with the generated urbanization visual, the potential more significant benefits of examining urban changes will be demonstrated to local governments.
Authors: Makineci, Hasan BilgehanArctic region is attracting increasing attention of researchers due to rapid changes. An important prerequisite for tracking multi-year changes is to consider the seasonal rhythm. Regular surveys of Arctic region by Sentinel-1 provide an opportunity to detailed study of the region seasonal changes over the one season, as well as differences over several years. SAR data is of primary importance for this region due to SAR independence of natural illumination and clouds. Coherence is one of the parameters calculated during interferometric processing of synthetic aperture radar (SAR) remote sensing data. On the one hand, coherence is a means for evaluating the quality of the generated interferograms, which determine the accuracy of the final results of such processing, i.e. digital elevation models and surface displacements maps. As the coherence values decrease, the reliability of final elevation values descends, the areas with low coherence values are unsuitable for processing the interferometry data. So, one of the crucial questions of repeat-pass interferometry technique is the minimization of factors causing the decrease of coherence, above all the selection of optimal period of radar observations. On the other hand, as the coherence is calculated for each element of the radar data, thus forming a map of coherence, it can be used as an additional source of thematic information about the state and properties of a surface. High coherence values are characteristic of stable areas that have not changed between data acquisitions, and vice versa, the coherence is low in changed areas. As an area of interest two test sites were selected. They are located near the meteorological stations «Ostrov Kotelny» and «Proliv Sannikova» (Kotelny Island, The New Siberian Islands archipelago, Russia) to more accurately relate coherence to real changes in the surface state observed at the stations. The Kotelny island is one of the largest Arctic islands and its landscapes diversity is rather well-defined for these high latitudes. The test sites are located within the ice complex lowland covered by tundra vegetation. Seasonal changes of coherence calculated using 2-year (May 2017 – June 2019, revisit time – 12 days) VV-polarization Sentinel‑1B/C-band SAR data set were investigated. Coherence values varied within 0,13…0,8 during the period on study. During the cold (with stable negative air temperature, and steady snow cover) and warm (mainly with positive air temperatures, absence of stable snow cover) periods coherence oscillations were in the range of 0,4 (rarely – 0,3)…0,8. During the transition period (when air temperature crosses 0ºC, snow cover forms or melts down) a sharp decrease in coherence to 0.2 or less was observed. Detailed comparison of coherence values over the sites with available additional information (vegetation indices obtained using optical data, meteorological data – air temperature, snow height and other) is presented in the report. Normalized radar cross-section (NRCS) values across the test sites were also analyzed in the study. It is shown that NRCS values are subjected to less variability compared with coherence, i.e. during the cold period it varies within ‑13…‑14 dB and during the warm period – within ‑7…‑9 dB. Significant NRCS values changes (up to 5…7 dB) are observed just during the interim periods. Generally this fact makes the coherence an important additional feature characterizing the condition of the surface of Arctic region. The material is prepared on the topic of the state task No. AAAA-A19-119022190168-8 and supported partly by RFBR (projects No. 18-07-00816 and No. 18-05-60221).
Authors: Baldina, Elena A. (1); Troshko, Ksenia A. (2,3)More than 6000 landslides occurred around Atsuma town in Hokkaido immediately after the Iburi-Tobu earthquake and the heavy rainfalls of the Jebi typhoon in September of 2018. Large-scale surface changes and the remarkable number of landslides made this case suitable for testing change detection techniques based on various space-borne SAR sensors. Recent papers were focusing on using the high-resolution, L-band Palsar-2 sensor (Aimati et al. 2019) for landslide detection purposes, evaluating the interferometric coherence and intensity differences (Fujiwara et al. 2019, Aimati et al. 2019). Despite the ongoing research, the capability of Sentinel-1 coherence and its effectiveness on change detection applications depending on the target conditions here addresses for landslide detection. The aim of our work is to apply a methodology to map landslides of the above-mentioned area and test the feasibility and reliability of the applied workflow using the C-band Sentinel-1 sensor. Whole stacks of Sentinel-1 images in both ascending and descending geometry from 2018 were used for interferometric analysis to map coherence changes of the area. The coherence over the study area has been calculated for subsequent acquisitions previously filtered through a Boxcar filtering approach. Flat areas as well as layover and shadow ones were masked using a digital elevation model to avoid misleading results. Moreover, images affected by snow cover were also rejected from the image stacks. Pre-, co- and post-event coherence differences have been calculated. Furthermore, temporal descriptors of the coherence series were computed to better describe the temporal evolution of the coherence on the surfaces affected by landslides. These temporal descriptors of the pre-, co-, and post-event periods were compared. Thereafter, coherence differences have been classified and combined with ground truth data. According to the classification results, S1 has limitations in detecting collapsed slopes, since single coherence differences provided poor agreement with control data. On the other hand, the results were significantly improved using multi-temporal descriptors of coherence. Half of the false positives of the final result were identified in the 30-45 m buffer zone of the successfully mapped landslides, highlighting how the spatial resolution of the S1 is not so accurate landslide detection. To improve the obtained results and increase the spatial capability of Sentinel-1 SAR on effectively mapping surface detection, the coherence will be calculated through SAR acquisitions previously filtered through adaptive non-local filtering methods. The expected improvement in the success rate will be calculated as proposed in the previously applied workflow.
Authors: Kovacs, Istvan Peter (1); Tessari, Giulia (2); Ogushi, Fumitaka (3); Pasquali, Paolo (2)The C-band Sentinel-1 A/B satellites in TOPS mode provide unprecedented opportunities for continuous radar mapping of the Earth with enhanced revisit frequency. The reliability for routine operational services relies on very high azimuth co-registration accuracy. However, while the enhanced spectral diversity (ESD) technique achieves a co-registration accuracy better than 0.001 of a pixel, the accuracy might still be degraded over low coherence area due to fast decorrelation and abrupt loss of coherence, especially for time series acquisitions. To overcome the limitation of decorrelation, we develop a two-step algorithm to improve the ESD estimation capability, and simultaneously mitigate abrupt loss of coherence in time series. In the first step, the double sample in the overlap region between consecutive bursts is used for coherence estimation. Under the assumption that the same decorrelation mechanism holds in both burst pairs, larger sample size will impose more accurate coherence estimation with smaller bias and variance. The accurate coherence observations stretch the contrast for coherent and incoherent targets and therefore avoid the ESD phase artifacts caused by excessive sample average over pure noise area. Next, we use all pairs shortest path (APSP) algorithm under the framework of graph theory to maximize the quality of temporal network. Different with conventional methods that heuristically choose a S-BAS subset of interferograms with better coherence, APSP check all interferograms one by one in a prior network and automatically replace pairs with bad coherence by the combination of neighboring pairs with better coherence. Compared with S-BAS subset, this method can avoid abrupt loss of coherence and meanwhile preserving the network consistency (no isolated subset). Compared with MST (minimum spanning tree) presented in previous studies, APSP can provide redundant observations which are very important during least squares adjustment. We further use APSP algorithm to evolve spatio-temporal network of co-registered Sentinel-1 SAR stack for line of sight (LOS) velocity estimation over low coherence area. We perform InSAR time series analysis and evaluate the effectiveness of proposed method using data acquired on descending track over representative low coherence region in southwestern China. For those images suffering from significant coherence limitation, We found that this method can obtain better co-registration accuracy than those obtained from previous network ESD method. By cross validation from descending Cosmo-SkyMed data, 39% uncertainty reduction on LOS velocity estimation can be achieved when applying co-registered Sentinel-1 stack with high accuracy and spatio-temporal network evolution.
Authors: Jiang, Mi (1); Tian, Xin (2)Emissions of native hydrogen (H2), until recently considered point-specific and localized at the plate boundaries, are finally quite common, especially in the intra-cratonic environments in the heart of continents. These emission zones are objects of primary interest, since, if it proved to be exploitable, this natural hydrogen could constitute a source of energy whose combustion would produce only water. The study of these emanations showed that they could sometimes lead to the formation of elliptic structures, associated with a local depression of the topography and a modification of the vegetation. These phenomena, visible by satellites, have been highlighted in several regions of the globe, as in Russia (Larin et al., 2015) or in the United States (Zgonnik et al., 2015). The objective of Sen4H2 project, jointly lead by Terradue and IFPEN, is to use data from several satellite missions, mainly from European Space Agency (ESA), to characterize the evolution of these topographic structures and identify the main descriptors of the presence of hydrogen. For this, in addition to the two regions mentioned above, two other sites were also studied, in the region of Minas Gervais in Brazil, equipped with permanent sensors for a continuous measurement of H2 emissions (Prinzhofer et al., 2019) and in the region of Bourakebougou in Mali. Among the descriptors investigated, we measure for each structure the deformation with radar data in interferometry, follow the development of ellipses by visible images and study the impact on the vegetation by the contribution of the near infrared. This approach could then be deployed to perform a "remote" search of emission areas in other parts of the globe. Presently, the first achievement of the sen4H2 project was to build a data catalog associated with different structures where hydrogen emissions were characterized. The produced dataset covers five identified structures of different size and contains over a thousand of processed satellite images, for acquisitions between 2002 and 2019 by several Earth observation satellites at different wavelengths, from visible and near infrared to radar. This collection is a promising asset in the perspective of improving the understanding on hydrogen continental emissions. In particular, the Sentinel-1 InSAR processing chains are making use of the SNAP toolbox. Coherence and Intensity (COIN) change detection products are processed as geocoded composites of coherence and amplitude images from Sentinel-1 TOPSAR IW SLC pairs. Amplitude Change (SNAC) products are processed as RGB composite of backscattering from a pair Sentinel-1 GRD IW and EW products (e.g. pre- and post-event). This was achieved through the use of dedicated web services for serving the collection of EO-based products and Jupyter Notebooks allowing to access, post-process and visualize the entire data collection. The use of Jupyter Notebooks makes particular sense in the context of the sen4H2 project, as they allow users without any prior programming skills to visualize, understand and adjust to a certain extent a set of post-processing algorithms to their specific needs. The existing notebooks thus represent a stepping stone to the more systematic data analysis required by the processes of interest, that will be capitalized on during the next steps of the sen4H2 project. References: Larin N., Zgonnik V., Rodina S., Deville E., Prinzhofer A. (2015) Natural molecular hydrogen seepage associated with surficial, rounded depressions on the European craton in Russia. Natural Resources Research, 24(3):369–383. Zgonnik V., Beaumont V., Deville E., Larin N., Pillot D., Farrell K. (2015) Evidence for natural molecular hydrogen seepage associated with Carolina bays (surficial, ovoid depressions on the Atlantic Coastal Plain, Province of the USA). Progress in Earth and Planetary Science, 2(31). Prinzhofer A., Moretti I., Françolin J., Pacheco C., D’Agostino A., Werly J., Rupin F. (2019) Natural hydrogen continuous emission from sedimentary basins: The example of a Brazilian H2-emitting structure. International Journal of Hydrogen Energy, 44:5676-5685.
Authors: Ducret, Gabriel (1); Caumont, Hervé (2); Deville, Eric (1); Brito, Fabrice (2); Sissmann, Olivier (1); Pacini, Fabrizio (2); Bachaud, Pierre (1)Research at the nexus of machine/deep learning and geoscience is on the rise, with countless applications. Computational geoscience is often conducted via deep learning models like convolutional neural networks (CNNs), which are notoriously uninterpretable. Satellite imagery InSAR is a commonly used data source for training these artificial intelligence models on, as they provide useful insights into earth scientific aspects such as the cryosphere, natural hazards, oceanic sciences, and geomorphology. Whether these machine learning models are being utilized for infrastructure damage assessment after natural disasters, plastic detection in oceans, gaining insights into urban sprawl, or various other useful applications, they are largely not transparent and therefore their predictions are hard to trace by humans. Keeping end users of such technologies in mind, it is important to be able to decode the pipeline that leads the models to make their predictions; black box models can lead to unforeseen biases, which can become dangerous upon deployment. To address the uncertainty of what is occurring behind the scenes, we propose performing ablation studies, changing the modalities of input that are utilized over many trials. For instance, in the case of predicting damage levels (categories) of buildings after natural disasters using satellite imagery, we experiment by gradually adding more types of training data each time: first, we only train on a post-disaster image; secondly, we add a pre-disaster image in tandem; thirdly, we further add the type of disaster (hurricane, tornado, fire, etc.); fourthly, we incorporate information about the damage status of neighboring buildings. Each time, as we add more information for the model to take into account in its training process, we also learn how the accuracy metrics on the testing set fluctuate. Ablation studies like this allow us to determine what aspects of the labeled image data are useful and enhance performance. Other steps we propose to address this overall uncertainty is to produce saliency maps, providing end users with visual tools that display where objects are detected in a geoscientific imagery assessment problem. Our preliminary findings in the space of interpretable natural hazards models, based on Sentinel-1 InSAR data, include that ordinal cross-entropy loss in the most useful criterion for optimizing these models and that modalities such as the pre-disaster and post-disaster image (for change detection) and the disaster type each increase the efficacy of the model when included as input. The gradient class-activation maps we directly produce also pave the way for deployable technology, which is a novel contribution upon the current literature. Our work and the ensuing discussions further the important mission of making neural network decision frameworks based on remote sensing and Sentinel-1 imagery more widely available to a diverse audience, gradually clearing up uncertainty that is perpetuated by black box models.
Authors: Chen, Thomas Y.Interferometric SAR observations of surface deformation are a valuable tool for investigating the dynamics of earthquakes, volcanic activity, landslides, glaciers, etc. To evaluate the accuracy of deformation measurements obtained from different existing or potential spaceborne InSAR configurations (different wavelengths, spatial resolutions, look geometries, repeat intervals, etc.), NASA is developing the Science Performance Model (SPM). The SPM allows for simulating different InSAR configurations and considers the major error sources affecting the accuracy of deformation measurements, such as ionospheric and tropospheric propagation delays or the effects of spatial and temporal decorrelation. In this NASA-funded study, we are generating global temporal coherence maps for four seasons with a spatial resolution of 3 arcsec using Sentinel-1 6- and 12-day repeat-pass imagery to complement the SPM with spatially detailed information on the effect of temporal decorrelation at C-band. Global processing of one year of Sentinel-1 Interferometric Wide Swath (IW) repeat-pass observations acquired between December 2019 and November 2020 to calculate all possible 6-, 12-, 18-, 24-, 36-, and 48-day repeat-pass coherence images (6- and 18-day repeat-pass where available) requires fast data access and sufficient compute resources to complete such scale of processing (~250 000 SLCs) in a few weeks. We implement the global S1 coherence processor using established solutions for processing Sentinel-1 SLC data. Input data are streamed from the Sentinel-1 SLC archive of the Alaska Satellite Facility and processed with the InSAR processing software developed by GAMMA Remote Sensing (www.gamma-rs.ch) coupled with cloud-scaling processing software employing Amazon Web Services developed by Earth Big Data LLC (earthbigdata.com). The processing is done on a per relative orbit basis and includes co-registration of SLCs to a common reference SLC, calculation of differential interferograms including slope-adaptive range common band filtering, and coherence estimation with adaptive estimation windows, which ensure a low coherence estimation bias of < 0.05. To account for the steep azimuth spectrum ramp in each burst, most of the processing steps are performed in the original burst geometry of the S1 SLCs so that information in the overlap areas of adjacent bursts are processed separately. Terrain-corrected geocoding to the 3x3 arcsec target resolution and simulation of topographic phase relies on S1 precision orbit information and the GLO-90-F Copernicus DEM. Alongside the coherence imagery, backscatter images are processed to radiometrically-terrain-corrected, RTC, level. Seasonal composites of 6- to 48-day coherence imagery as well as RTC backscatter are generated. Based on the coherence values, coherence decay rates are determined per season with a simple exponential model. The processing of the individual coherence images, RTC backscatter images, seasonal coherence and backscatter composites as well as the pixel-level coherence decay modeling results are to be completed by mid-2021. The processing methodology and intermediate results will be presented at the FRINGE Workshop.
Authors: Cartus, Oliver (1); Kellndorfer, Josef (2); Oveisgharan, Shadi (3); Osmanoglu, Batu (4); Rosen, Paul (3); Wegmüller, Urs (1)Extreme rain events have been increasing in frequency worldwide, even in arid and hyper-arid regions. Heavy precipitation in normally arid regions can have devastating consequences for local communities. One way in which the scientific community is working to better characterize these events is through the study of soil moisture variability and storage. SAR data, with its high spatial resolution and relative insensitivity to cloud cover, is poised to become a key tool in the next generation of soil moisture research. Here, we intercompare the effect of soil moisture on both amplitude and phase following two large cyclones that impacted the southern Arabian Peninsula in May (Cyclone Mekunu) and October (Cyclone Luban) of 2018. The eastern edge of both rain events are roughly co-located, and the data show distinct boundaries between regions that experienced rain and those that did not. These two conditions allow us to compare differing degrees of ground surface wetting across short spatial scales, and investigate the repeatability of the soil moisture response after rain events in the same region. We examine the impact soil moisture has on the phase and magnitude of differently-polarized SAR returns, and explore the implications of spatially averaged complex data. This study highlights the utility of analyzing soil moisture changes in hyper-arid regions, such that it allows for clear separation of the impact of soil moisture variability on InSAR phase from other effects, such as surface roughness and vegetation. Preliminary results show that the magnitude and longevity of soil moisture-related signals following Cyclone Mekunu and Cyclone Luban are related to the land surface temperature, as well as the total precipitation at a given location. Countries that are adjacent to the Arabian Sea and northwestern Indian Ocean continue to feel the impacts of a changing climate, with another unusually large storm (Cyclone Gati) hitting the Horn of Africa in late 2020. With the rapid expansion of an already extensive SAR catalog, it has become apparent that we need to work towards a more integrated understanding of the impact of soil moisture on all aspects of SAR data, including phase, magnitude, polarity, downlooking, filtering, and coherence.
Authors: Burgi, Paula; Lohman, RowenaThis study analyzes the information content of various derivatives of VHR radar image products over the city of Hue in Central Vietnam. 24 PAZ images, acquired in Spotlight mode in both ascending and descending orbits were used to compute calibrated radar cross section (Sigma0, n=24), interferometric coherence of different temporal and perpendicular baselines (n=30) and complex intensity which is computed from the combination of real and imaginary parts of the reference and support image of an interferometric pair and contains less speckle and also information from two acquisitions (n=30). These grouped derivatives were used as features to test their contribution for the detection of dwelling density, the allocation of built-up areas and the classification of land use and landcover. Reference data were provided by Vietnamese project partners and values of 5000 reference points were extracted based on a stratified random sampling to include representative training data for all land cover types. Building density (ranging between 0 and 2.6 buildings within a radius of 10 meters): A feature ranking based on the ReliefF measure confirmed that coherence layers have a distinctively higher predictive value for the building density than complex intensity and Sigma0. However, no correlation between perpendicular or temporal baseline and the prediction accuracy has been found. None of the feature groups was able to reliably predict building density based on the 24 images based on linear regression (R² between 0.0 and 0.16). Built-up area (binary: built-up and open): The feature ranking again confirmed the superiority of coherence and complex intensity over Sigma0 data, but indicates no clear trend between baseline, orbit direction and ReliefF. A decision tree based on all three feature groups resulted in a classification accuracy 0.84 and a F1 score of 0.82. The most important feature to split the training data was a coherence product with a temporal baseline of 11 days and a perpendicular baseline of 22 m. While complex intensities were also included in the tree, calibrated Sigma0 was only rarely used to split the samples. Land cover (9 classes: crops, forest, open, paved, public, residential, vegetation, water, other): Coherence-based features resulted in highest ReliefF values compared to complex intensity (slightly lower) and Sigma0 (clearly lower). A random forest classification based resulted in an overall classification accuracy of 32.5 % with strong variation between the classes. The classes with the highest accuracies were water (UA: 83 %) crops (PA: 81.7 %), and residential (PA: 66.8 %), while classes with less distinct physical characteristics (vegetation, public or paved) performed comparably poor. Concludingly, orbit direction and perpendicular baseline did not have significant impacts on the classification results, but coherence (especially combinations of several temporal baselines) were found most suitable to characterize urban features in the case city.
Authors: Braun, Andreas (1); Felix, Bachofer (2)Airborne Synthetic Aperture Radar (SAR), with its unique capability of great flexibility as well as imaging independent of solar illumination and weather conditions, has been widely used in ground activity monitoring. And Frequency Modulation Continuous Wave (FMCW) have wide bandwidth, small volume, light weight and low cost. FMCW SAR combines the characteristic of FMCW and SAR, it can be applied to small aircraft with low platform cost and small UAV without casualties. With the development of society, security defense, ground search and surveillance, threat activity detection on ground, ground activity management and other human activity all become an important topic of social security. In addition, in military applications, the need of managing military sites and detecting artificial targets in various climatic environments is also growing. To solve this problem, SAR Interferometry (InSAR) is a good way. In this paper, we focus on FMCW SAR for subtle human activity detection. There are two kinds of detection method. One is incoherent change detection method, the other is coherent change detection (CCD) method. For the difference between these two methods, the former can only detect larger change, and the latter can detect smaller change. Aiming at small-scale human activity detection, so this paper employs the coherent change detection method to achieve the subtle change detection between two data acquisitions. Coherent change detection method utilizes the phase difference between the two complex images to detect the change area in the scene. For the traditional coherent change detection method based on a single threshold, it can cause inaccurate detection, such as local false detection. Because the coherence present different level in different area, so we adopt a coherent change detection method based on adaptive threshold to detect human activity area in this paper. In detail, we divide the co-registered image into blocks, and regarding the average of the coherence as the adaptive threshold of each block. And the number of blocks is also discussed, generally blocks is enough. The method adopted in this paper is further validated by the actual airborne Ka band repeat-pass Interferometric Synthetic Aperture Radar data that acquired by the Institute of Electronics, Chinese Academy of Sciences . In order to further illustrate the effectiveness of the method, we compare the result with it acquired by the coherent change detection method based on a single threshold. The results show that the method adopted by this paper have better detection performance than the traditional coherent change detection method based on a single threshold.
Authors: Wang, Zhongbin (1,2); Xiang, Maosheng (1,2); Wang, Bingnan (1,2); Chong, Jinsong (1,2); Wang, Shuai (1,2)Ecuador has many volcanoes in areas where it is challenging to install in-situ instruments (such as tiltmeters or GPS) because they are located on islands or areas where vegetation is dense. Volcanoes Sierra Negra and Wolf (in the Galápagos Islands) and Antisana, on the mainland, are monitored with seismic sensors and in the last years with InSAR techniques with the use of SENTINEL-1 images. The Galápagos Islands (Ecuador) have many volcanoes with significant rates of deformation. Sierra Negra volcano, located in the southern portion of Isabela Island, had a massive eruption in July 2018, and it showed an intense trend of inflation before and after the eruption. DModels software has been applied to locate the source of deformation, magma volume, and depth. From July 2019 to July 2020, an incoming rate of 25 Mm3/yr was detected using the SBAS method with SENTINEL-1 data. Sphere and sill geometries were used to study the source of deformation. A variogram is generated by the software for the sill geometry and indicates a misfit between the model and data near zero, indicating that the model is sound and adequately explains the registered LOS's magnitude (line of sight). Modeling results indicate a depth of 2.8 km below the caldera's surface, 25x10^6 m^3, a sill radius of 2.5 km, and a location for the source near the caldera's center. Wolf volcano also displayed intense inflationary trends; in 2015, it had one of the most significant eruptions in the Galápagos Islands since the onset of routine satellite-based volcano monitoring. An InSAR (Interferometric SAR) approach is applied to study the deformation at Wolf volcano with Sentinel-1 images acquired from November 2016 to December 2020 in both orbits. The observed actual deformation rates and variations of source volumes pose issues for the ongoing unrest and possible forthcoming eruptions. In continental Ecuador, there are many volcanoes near cities. One example is at Antisana volcano, which has a sizeable glacier (~15 km2), affecting towns and Quito's water supply if it melted. An InSAR (Interferometric SAR) approach is applied to study the possible deformation at Antisana volcano using Sentinel-1 images acquired from July 2019 to August 2020 when a minor inflationary signal of 15 mm was picked up on the western flank. This deformation is minimal compared to other continental volcanoes, and there could be some doubts about the changes. However, COMET Project processing also registered inflationary trends around Antisana volcano (https://comet.nerc.ac.uk/comet-volcano-portal/). In summary, InSAR processing using ESA's SENTINEL-1 images has let us monitor volcanoes on islands and in continental areas in different conditions of vegetation and locations of difficult access. The use of modeling software allows the estimation of parameters of the source of deformation.
Authors: Aguaiza, SantiagoSurface deformation at dome building volcanoes often precedes vicious explosive eruptions that may be accompanied by dome collapse and break-off of instable blocks leading to devastating pyroclastic flows. Surface displacement may vary on the scale of mm up to several meters related to mid-crustal storage changes or near-surface intrusions, respectively. After four years of dormancy of Bezymianny volcano in Kamchatka, its 2016-2017 eruption series began with effusive activity followed by three successively stronger explosive eruptions. For the onset of the eruptive succession, we employ pixel offset tracking between 39 adjacent and co-registered descending TerraSAR-X (TSX) spotlight imagery by retaining the highest resolution. We could quantify locally defined summit deformation associated with long lasting (>7 months) precursory plug extrusion exceeding 30 m of deformation in LOS, predominantly without observed seismicity. Our results also uncover recurrent morphometric changes related to periodic expansion of the northern dome flank. The TSX amplitude data suggest that the observed different dome growth mechanisms were controlled by the magma discharge rate and the cooling rate of the upper conduit, which deflected intrusions into structurally weaker parts of the upper dome. Since Bezymianny is frequently covered with snow and newly deposited ash, Interferometric Synthetic Aperture Radar (InSAR) was neither effective to detect precursory nor eruption-simultaneous ground deformation during the first three eruptions. Yet for the 20 December 2017 eruption, we employed InSAR on adjacent multilooked two-pass TSX scenes acquired in ascending and descending orbital geometry. After atmospheric delay correction by using the Generic Atmospheric Correction Online Service (GACOS), our interferometric results unveil persistent mm-scale deformation of the outer flanks of the edifice whereby the polarity of the signal inverted from before to after the eruption. We interpret the resulting displacement patterns prior and during the eruption as inflation-type and deflation-type deformation, respectively. Bayesian inversion of the corrected InSAR observations preceding the last eruption indicate a source model depth that is in agreement with seismicity derived from a temporary seismic network installed close to the eruptive centre, as well as with petrologic evidences for a very shallow reservoir. In this presentation we discuss the broader implications of this deformation signal along with its causes, providing a valuable contribution to our understanding of the shallow magma plumbing system beneath Bezymianny.
Authors: Mania, René; Walter, Thomas; Cesca, SimoneChiles and Cerro Negro are two nearly co-joined strato-volcanoes located at the boundary between Ecuador and Colombia and situated beside several communities and small cities. Since 2013, anomalous seismic activity was recorded in locations around the volcanoes, announcing the beginning of a period of unrest that exists to this day. Initially, the ground deformation was retrieved using geodetic GNSS data and later in the last five years, with Interferometric Synthetic Aperture Radar (InSAR) observations. Its results reveal areas around the volcanoes with ascending displacements related to probable inflationary trends and are often coherent with upticks in seismicity. The most discernible deformation is located in the Potrerillos basin, an old volcanic caldera that displays a continuous uplift of around 2.5 – 3 cm/yr over an 8 x 6 km2 area. A second area on the southwest flank of Chiles volcano presents an inflationary pattern averaging ~2 – 2.5 cm/yr. Finally, Cerro Negro volcano itself has a fluctuating trend between stability, inflation, and subsidence over time, averaging 1 – 1.5 cm/yr. Although swarm episodes characterize seismic activity, the surface deformation rates for Potrerillos and Chiles volcano stand without significant changes nor fluctuations. InSAR ground deformation and times series were obtained from processing images of the ESA’s Sentinel-1 and DLR’s TerraSAR-X missions with ascending and descending paths, showing that temporal products are consistent in space and deformation rates. The retrieved deformation patterns and the changing seismic activity reveals that deformation at the Chiles/Cerro Negro volcanoes and in the Potrerillos basin possibly represent significant hazards to the local communities should a significant change occur.
Authors: Yépez, Marco (1); Mothes, Patricia (1); Trasatti, Elisa (2); Tolomei, Cristiano (2); Atsori, Simone (2)Volcanic unrest in calderas can be exhibited through a variety of different mechanisms, such changes in seismicity, ground deformation, thermal radiance, and/or gas emissions. However, not all caldera unrest results in explosive caldera-forming volcanic activity. Alternative activity may include a period of quiescence, persistent degassing, or effusive activity in the form of lava flows or appearance of lava lakes. In December 2015, the Nicaraguan basaltic caldera Masaya exhibited signs of unrest with the formation of a lava lake at the summit. Geodetic studies spanning the 2015-2016 period, such as Interferometric Synthetic Aperture Radar (InSAR) [Stephens & Wauthier, 2018, GRL] and precision leveling data [Murray et al., 2017, IAVCEI 2017; Rymer et al., 2017, IAVCEI 2017], identified a previously undetected magma reservoir (Masaya Central Reservoir, MCR) located 3 km north of the active Santiago pit crater. Furthermore, gas geochemistry in-conjunction with geodetic data indicated changes in magma supply from depth occurring within the MCR in the weeks leading up to the appearance of the lava lake [Aiuppa et al., 2018, GCubed.; Stephens et al., 2020, RS]. This study takes advantage of the increase in available SAR datasets to examine the long-term evolution of ground deformation within the caldera and the sub-surface conditions of the MCR during different periods of volcanic activity. We focus on ground deformation at Masaya between 2011-2019. The areas surrounding Masaya and Managua City experienced several regional tectonic earthquakes during this period, and Masaya itself exhibited small throat-clearing eruptions, periods of seismic tremor, passive degassing, incandescence and lava lake activity. With access to over 700 scenes with different wavelengths, Line-Of-Sight geometries, and orbit directions from RADARSAT-2, COSMO-SkyMed and Sentinel-1 datasets, we apply the Multidimensional Small BAseline Subset (MSBAS) time-series approach [Samsonov & d’Oreye, 2012, GJI; Samsonov & d’Oreye, 2017, CJRS; Samsonov, 2019, GC] to improve the temporal resolution of ground deformation at Masaya. This approach allows us to separate the relative contributions of the vertical and E-W horizontal components to the deformation field. Plotting up the component results centered over the MCR, we divide the MSBAS time-series displacements into several time periods according to observed general trends. Cumulative displacement maps in the vertical and E-W horizontal components were created for each time period. As the source location for MCR appears to be stable, particularly during the period of lava lake activity, we define the input source geometry of MCR according to the joint inversion performed by Stephens & Wauthier [2018]. Simple linear least-squares inversion is performed with the deformation source embedded in a uniform, isotropic, homogeneous, and elastic half-space. The source location and depth are kept constant in order to obtain the volume change of MCR during the different periods of unrest and volcanic activity. The results from this work emphasize the importance of long-term deformation monitoring of caldera regions to improve our understanding of the relationship between the magma plumbing system and observed volcanic activity. Volume change in particular is a useful and more accessible metric to help define changes in volcanic activity compared to using changes in ground displacement alone. In-conjunction with other remote sensing datasets and observations, this approach will provide a more detailed framework on volcanic behaviour for forecasting in other caldera settings.
Authors: Stephens, Kirsten (1); Wauthier, Christelle (1,2)One of the most significant volcanic eruptions to occur on the Canary Islands, was the 1730–1736 eruption on Lanzarote. During the 2055-day eruption between 3 –5 km2 (Carracedo, 2014) of material was erupted over an area greater than 200 km2 (Solana et al., 2004). Since then there has only been one other volcanic eruption in 1824 and present-day activity is limited to anomalous heat flow over part of the 1730—1736 lava flows (Montanas del Fuego). Very few studies have looked at long term (centuries) volcanic deformation, but previous ERS InSAR (Gonzalez and Fernandez, 2011) and GNSS (Riccardi et al, 2018) measurements of the past few decades on Lanzarote showed an area of subsidence spatially correlated with the high heat flow. Here we used over 400 Sentinel-1 and Envisat ascending and descending images to construct a time series of line of sight displacements and calculate linear deformation rates over the whole Island. The ascending and descending deformation rates were decomposed to give vertical and E-W rates. Our preliminary results show a constant subsidence rate of 6--7 mm/yr associated with the south-east portion of the Timanfaya lava flows over the 28 year period (1992 – 2020) covered by the Sentinel, Envisat and ERS data. The phase bias contribution of the shorter interferograms was also explored for the Sentinel 1 data using a range of short and longer temporal baseline networks. The phase bias is generally present across the whole island, but there were two significant areas – positive bias at Uga equivalent to ~5 mm/yr and negative bias at Guiatiza equivalent to ~-7 mm/yr. The deformation signal observed at Timanfaya remains approximately constant with a maximum rate of -7 mm/yr, but the spatial extend and variability decreases as the temporal baselines of the interferograms are increased. To explain the subsidence at Timanfaya, we explored thermal contraction and mechanical compaction of the lava flow, as well as a historic magmatic intrusion. Our analysis shows that the constraints of an intrusion are more realistic than those required for lava flow deformation after almost 300 years. Regardless of the mechanism, the signal at Timanfaya shows that volcanic deformation can occur on longer timescales than previously thought and is still measurable centuries after volcanic events.
Authors: Purcell, Victoria Jane (1); Elliott, John (1); Ebmeier, Susanna (1); Gonzalez, Pablo (2); Reddin, Eoin (1); Watson, Andrew (1); Morishita, Yu (3)Volcano monitoring at volcano observatories still relies mainly on ground-based observations despite the potential added value of remote sensing data. We describe here our efforts to integrate in near real time the information derived from Sentinel-1 satellites into the monitoring devices at BBPTKG in charge of Merapi volcano survey. Merapi (7°32.5’ S and 110°26.5’ E) located in the densely populated Province of Yogyakarta in Central Java is one of the most active volcanoes in Indonesia. The recent eruptive history of Merapi is characterized by two eruptive styles: 1) recurrent effusive growth of viscous lava domes, with gravitational collapses producing pyroclastic flows known as « Merapi-type nuées ardentes » (VEI 2); 2) more exceptional explosive eruptions of relatively large size (VEI 3-4), associated with column collapse pyroclastic flows reaching distances larger than 15 km from the summit. The eruptive periodicity is 4 to 5 years for the effusive events and 50 to 100 years for the explosive ones. The last explosive event (VEI 3-4) occurred in November 2010 and was followed by a period of limited activity. On August 11, 2018, a new dome was observed inside the summit crater, thus marking the start of a new phase of effusive activity. A new period of unrest then started in mid-October 2020 leading to an eruption on 4 January 2021. Deformation is currently being recorded by tiltmeters together with a network of 10 permanent GNSS stations. GNSS data are automatically processed and inverted for a pressure source at depth. Since 2019, a deep source of pressurisation (around 8 km depth) appears intermittently with a limited amplitude (below 1 cm of displacement). Such a small and transient signal is challenging to detect by InSAR. Sentinel-1 data are acquired over the volcano every 12 days on descending track 76 and every 6 days on ascending track 127. Sentinel-1 data are automatically uploaded to a local server at BPPTKG. Interferograms and coherence images are then produced using the NSBAS processing chain (Doin et al, 2012) and automatically integrated to WebObs (Beauducel et al., Frontiers, 2020), an integrated web-based system for monitoring, to enable detection of potential rapid and significant changes in signal. Mean velocity maps are also produced as well as time series of surface displacement at given location, allowing direct comparison with GNSS measurements. No significant displacement was evidenced by InSAR until November 2020, when the descending time series showed a strong displacement of the north-eastern part of the summit area away from the satellite. The objective is to integrate the temporal information on surface displacement provided by Sentinel-1 data into WebObs and combine it with GNSS information in a joint inversion. References: *Beauducel et al., WebObs: The Volcano Observatories Missing Link Between Research and Real-Time Monitoring, Front. Earth Sci., 2020, https://doi.org/10.3389/feart.2020.00048 *M. -P. Doin, F. Lodge, S. Guillaso, R. Jolivet, C. Lasserre, G. Ducret, R. Grandin, E. Pathier, V. Pinel , Presentation of the small baseline NSBAS processing chain on a case example: The Etna deformation monitoring from 2003 to 2010 using Envisat data, Fringe Workshop Proceedings, ESA Special Publication SP-697, 2012.
Authors: Pinel, Virginie (1); Beauducel, François (1,2,3); Putra, Raditya (3); Sulistiyani, Sulis (3); Nandaka, Gusti Made Agung (3); Nurnaning, Aisyah (3); Santoso, Agus Budi (3); Humaida, Hanik (3); Doin, Marie-Pierre (1); Bascou, Pascale (1); Thollard, Franck (1); Laurent, Christophe (1)The 22nd of December 2018 volcanogenic tsunami triggered by the Anak Krakatau (Indonesia) lateral flank collapse has been one of the deadliest volcanic phenomena in almost 30 years, with over 400 fatalities and 16,000 people displaced. Understanding volcanogenic tsunamis triggering processes, and mitigating their effects, still remains a major challenge. The volcano morphological changes associated with the Krakatau collapse event were captured in unprecedented detail by both optical (Sentinel-2) and radar satellite imagery (Sentinel-1, TerraSAR-X and COSMO-SkyMed) and eyewitness accounts. We have combined this material with historic observations on the growth history and eruptive activity of the volcano along with sedimentological, petrographic, bathymetric and ecological datasets to reconstruct the main volcanic processes during the pre- and post-collapse period (May 2018 – November 2019) affecting Anak Krakatau and its archipelago. Analysis of the shoreline changes from Sentinel-2 imagery show that the net change in size of the Anak Krakatau Island was an increase in area, from 2.84 km2 to 3.19 km2,between 15 May 2018 – 1 November 2019. This is despite the lateral collapse almost halved the island area to ~1.7±0.3 km2 as evidenced in the COSMO-SkyMed image of 23 December 2018, by a NW-SE scarp that bisected the whole island. Regrowth of the island occurred rapidly, through vigorous volcanic activity, which led to the subsequently increased island area. Analysis of Digital Terrain Model (DTM) in combination with bathymetric and seismic reflection data reveal a total landslide deposit (0.214±0.036 km3) dominated by large intact blocks, emplaced over 1.5 km into the adjacent basin, that triggered a secondary debris flow (0.022±0.005 km3). Our findings are consistent with en-masse lateral collapse with a failure volume ≥0.16 km3, resolving ambiguities in previous subaerial reconstructions of the event. These observations allows for an accurate reconstruction of the volcanic unrest period by providing new insights and constraints for reconstructing the volcano external architecture and for understanding the tsunami source mechanisms. In turn, these insights could help improve our understanding of tsunami and volcanic source mechanisms and precursory signals especially along highly densely populated south east Asia coastlines.
Authors: Novellino, Alessandro (1); Engwell, Samantha (1); Grebby, Stephen (2); Hunt, James (3); Watt, Sebastian (4); Ebmeir, Susanna (5); Cassidy, Michael (6); Day, Simon (7); Madden-Nadeau, Amber (6); Grilli, Stephan (8)Magmatic intrusions generally produce ground deformation that can be studied by geodetic techniques. Many dikes and sills emplacements (sometimes associated with eruptions) in different tectonic settings have been analyzed through InSAR in the past two decades. However, post-intrusive behavior was studied in only a few cases. Here we analyze the post-diking deformation at Harrat Lunayyir, which is a monogenetic volcanic field located in western Saudi Arabia on the eastern margin of the Red Sea Rift. Between April and July 2009, an intensive seismic swarm with thousands of earthquakes was registered in the area with many events with a magnitude above 4 and the largest earthquake of Mw 5.7. InSAR data showed that the earthquake swarm was accompanied by a dike intrusion that stopped ~1km below the surface, estimated to be ~7 km in length and with an opening of up to 4 m. Above the intrusion, a ~10 km long and ~5 km wide graben formed during the activity with up to 1 m of fault slip on the border faults. Within the graben, continuous seismicity has been registered to the present day generally with Ml3. The seismicity has been decreasing through time, both in terms of earthquake number and energy released. In February 2017, a new seismic swarm occurred ~60 km north of Harrat Lunayyir and another swarm started in October 2018, about 30 km southwest of the volcanic field. The last one is still ongoing with a few events per week and Ml
Authors: Nobile, Adriano (1); Cao, Yunmeng (1); Youssof, Mohammad (1); Trippanera, Daniele (1); Passarelli, Luigi (2); Jónsson, Sigurjón (1)In the face of a volcanic eruption, it is important to know the affected regions to manage an emergency. Volcanic ashfall affects buildings, vegetation and harvest, water bodies, livestock and population. Aiming to identify and map the regions affected by volcanic ashfall, we adapted and applied a Temporal Decorrelation Model (TDM) [1] using Synthetic Aperture Radar (SAR) C band data acquired by the Sentinel-1 constellation. Coherence maps derived from SAR interferometry (InSAR) are useful for change detection but these maps alone can be limited for distinguishing the changes resulting from the presence of volcanic ash on the ground from changes generated by other causes such as rain, snow, wind or the seasons. To this end, the implementation of TDM aims to improve our capabilities to discriminate among both kinds of changes. The eruption of Taal volcano (Philippines) on January 12, 2020 was selected as a case study. We prepared a dataset of 93 SAR images acquired between January 02, 2017 and February 16, 2020 in interferometric wide (IW) beam mode and Level 1 Single Look Complex (SLC), polarization VV and ascending orbit. We arranged these scenes into two subsets, one that contained 89 images acquired before the eruption and another with 4 images acquired after the eruption. Subsequently, we generated 3916 pre-eruption interferometric pairs (reference pairs) and 362 post-eruption pairs. First, the total coherence is estimated for each pixel with the TDM by using as input the pre-eruption coherence maps. Temporal decorrelation changes are related to changes in the dielectric and physical structure of the scatterers and in the location where these changes occur (i.e. volume and ground). The estimation takes into account three aspects: (1) the random motion of the vegetation structure, (2) the temporally correlated dielectric changes (seasonal and normal decreasing of the coherence over time) and (3) the temporally uncorrelated dielectric changes (random). Considering these aspects, the total temporal coherence depends on three Pair-Invariant variables: (1) the Ground to Volume Ratio, (2) the Characteristic Time for Ground and (3) the Characteristic Time for Volume. The total temporal coherence for each pixel is also dependent on two Pair-Variant random variables: (1) the Random Dielectric Changes of Ground, (2) the Random Dielectric Changes of Volume. Thus, the number of model parameters becomes twice the total number of interferometric pairs (N) plus three Pair-Invariant variables (2N + 3); in this sense, coherence depending on the land cover, the properties of the vegetation and the dominant scattering. [1] Second, the extracted model parameters are used to estimate the affected area. A Cumulative Distribution Function (CDF) is constructed by using the random dielectric changes of ground and volume variables for each pixel. This CDF is the core for the estimation because it contains statistical information on how natural occurrence or randomness affect the coherence of each pixel [1]. Third, the whole procedure explained above is implemented again but with the post-eruption subset in order to generate a new post-eruption CDF. Finally, both CDFs are compared in order to assess their similarity and subsequently estimate the probability that each pixel has been affected by ashfall. Thus, a probability map is generated, where pixels with probabilities closer to one indicate the presence of volcanic ash on the ground. Our implementation of the TDM resulted in a probability map that shows higher values that are in agreement with the distribution of volcanic ash observed on the ground by the volcano observatory of the Philippine Institute of Volcanology and Seismology (https://www.phivolcs.dost.gov.ph/) and with the dispersion of volcanic ash in the atmosphere observed from Himawari-8 satellite data. However, areas where coupled effects occur, like crustal deformation and ashfall, are more complex to interpret. Subsequent work will aim at correlating the probabilities with the presence and thickness of ashfall deposits and at implementing the TDM with both C and L Band SAR data. References: [1] J. Jung, D. Kim, M. Lavalle, and S.-H. Yun, “Coherent Change Detection Using InSAR Temporal Decorrelation Model: A Case Study for Volcanic Ash Detection,” IEEE Trans. Geosci. Remote Sens., vol. 54, no. 10, pp. 5765–5775, Oct. 2016.
Authors: Naranjo, Camilo (1,2); Euillades, Pablo (1,3); Toyos, Guillermo (1,4); Euillades, Leonardo (1,3); Villarosa, Gustavo (1,2)Kamchatka is the most active area of volcanism and seismicity in Russian Federation and probably of the Pacific fire ring as well. It hosts up to 30 volcanoes main part of which are active, and bordered at the East by more than 700 km-long part of the Kuril-Kamchatka subduction zone. Main part of this territory is sparsely populated, seismic, geodetic, and other networks are not dense, hence satellite monitoring is very important tool here. On the other hand, this mountain area covered by snow at least 6 months, what is a challenge for SAR applications. We present some results obtained by our group applying DInSAR and PSInSAR techniques to ENVISAT, Sentinel-1A and ALOS-2 images. Plosky Tolbachik eruption (2012-2013). A huge fissure eruption occurred at the Plosky Tolbachik volcano in 2012-2013, which results in formation of broad lava fields. Three series of Sentinel-1A images were used as initial data: 12 from May 14 to October 17, 2017, 12 from May 21 to September 30, 2018, 11 from May 16 to September 13, 2019. Applying SBAS technology (SARScape software) we found numerous persistent scatterers, mostly located in the southern part of the Tolbachik valley on the Leningrad and Tolud lava fields. For these periods the average rate of the LOS displacement was about 200 mm/year. In addition, one can clearly see a jump in the time series in June 2017, which coincided with eruption of the Bezymyanny volcano at June 16, 2017. We constructed a thermal model of lava subsidence taking into account real data on lava composition and eruption scenario. The model explains well subsidence at the main part of the lava field except regions around fissure, where lava moved within tubes below solid lid (Mikhailov et al., 2020). Activation of Bolshaya Udina volcanoe (2017-2018). Since the end of 2017, strong seismic activity has been detected near Bolshaya and Malaya Udina volcanoes. This high seismicity was related to activation of the volcanoes caused by uplift of lava from deep reservoir to the surface (Koulakov et al., 2019). Detailed analysis of seismology data leads us to an opposite scenario, that seismic activation was a result of lava movement from a shallower reservoir down to a deeper one. To analyze surface displacements during summer of 2017 and 2018, we used 22 Sentinel-1A images from track 60 of descending orbit (10 images from June 07 to September 23, 2017 and 12 images from May 21 to September 30, 2018) as well as three images of the ALOS-2 PALSAR-2 satellite (October 4, 2016, June 13, 2016, and October 2, 2018). Time-series of the LOS displacements were calculated using the SBAS method in SARScape software, and SRTM3v4 DEM. Persistent scatterers were found mainly in the lower part of the volcano Bolshaya Udina. TS are similar for both time-periods even in 2018 the LOS velocities are slightly higher at almost all PSs. Hence, SAR data do not support the hypothesis of magma uplift from depth to shallower reservoir (Senyukov et al. 2020). Activation of Kizimen volcanoe (2009-2011). We also investigated Kizimen active volcano, located within the Kronotsky national park, 265 kilometers from the city of Petropavlovsk-Kamchatsky. Kizimen has been silent during 80 years; the last eruption occurred in 1928-1929. In April 2009, the first activation of seismicity was registered in the Kizimen area, and a noticeable increase occurred in July 2009. The seismicity of the volcano in 2009 far exceeded the background one during the period 2003-2008. The eruptive activity started at the end of October 2010, growing up to its maximum in December. At the end of February 2011, a lava flow about 200–300 m-long was formed at the eastern slope of the volcano. The explosive activity slowly ceased, and stopped in December 2011. We used 34 ENVISAT images (17 images from track 59 of descending orbit from July 30, 2004 to October 17, 2010 and 17 images from track 288 of descending orbit from September 09, 2004 to August 29, 2010) and StaMPS / MTI software to investigate surface displacements at the Kizimen volcano. Time series clearly show an area of “uplift” to the east of Kizimen. Here LOS displacements started in summer of 2005, and since September 2008 positive LOS displacement became much faster. Hence, SAR data demonstrated that the deformations of the Kizimen slopes began long before the seismic activation. Large size of the area of positive LOS displacements (“uplifts”) suggests that during 2005 - 2008 magma moved up from relatively big depth and rate of its movement has been growing up since 2008. Olyutorsk earthquake (April 20, 2006). Using SAR interferometry, we also updated the model of the Olyutorsk Mw=7.6 April 20, 2006 earthquake and its strongest aftershock of Mw 6.6 that occurred on April 29 suggested in (Mikhailov et al., 2017). The new model of this earthquake is based on joint inversion of SAR and GPS data. As a result, we conclude that SAR interferometry is an efficient tool for monitoring volcanic activity and studying earthquakes even in very complicated environment of Kamchatka (mountains, snow, dense vegetation etc.) when combining modern efficient methods of SAR data processing, effective dumping of noise and artifacts, with comprehensive analysis of all available satellite and terrestrial data. Acknowledgments. We thank A. Hooper for sharing his StaMPS / MTI software. We are grateful to the Japan Space Agency JAXA for the ALOS satellite images provided as part of the ER2A2N075 project. Financial support. Our work was supported by the Megagrant from the Ministry of Science and Higher Education of the Russian Federation under project no. 14.W03.31.0033 “Geophysical studies, monitoring, and forecasting the development of catastrophic geodynamical processes in the Far East of the Russian Federation” and by the state contract of the Kamchatka Branch of Geophysical Survey RAS within the research project no. АААА-А19-119031590060-3 “Complex geophysical studies of the volcanoes of Kamchatka and the northern Kuril Islands to detect signs of an impending eruption, as well as to predict its dynamics with an assessment of the ash hazard for aviation.” Study of Vera Timofeeva was partly supported by Russian Foundation for Basic Research (grant no. 19-35-90092). We dedicate this paper to the memory of our colleague Elena Kiseleva, a wonderful scientist and outstanding person, who died on November 8, 2019. REFERENCES Koulakov I., Komzeleva V., Abkadyrov I., Kugaenko Y., El Khrepy S., Al Arifi N. Unrest of the Udina volcano in Kamchatka inferred from the analysis of seismicity and seismic tomography // Journal of Volcanology and Geothermal Research, 2019, V. 379, p. 45–59. https://doi.org/10.1016/j.jvolgeores.2019.05.006 Mikhailov V.O., Kiseleva E.A., Arora K., Timoshkina E.P., Smirnov V.B., Chadda R., Ponomarev A.V., Shrinagesh D. New Data on the Olyutorskii Earthquake Acquired via SAR Interferometry // Journal of Volcanology and Seismology, 2017, V. 12 (3), p. 213-220. DOI 10.1134/S0742046318030053 Mikhailov V.O., Volkova M.S., Timoshkina E.P., Shapiro N.M., Babayantz I.P., Dmitriev P.N., Khairetdinov S.A. Analysis of displacements of the lava flow surface of the 2012–2013 Tolbachik fissure eruption by SAR interferometry // Geophysical Research, 2020, V.21 (4), p. 21-34. https://doi.org/10.21455/gr2020.4-2 Senyukov S.L., Mikhailov V.O., Nuzhdina I.N., Kiseleva E.A., Droznina S.Ya., Timofeeva V.A., Volkova M.S., Shapiro N.M., Kozhevnikova T.Yu., Nazarova Z.A., Sobolevskaya O.V. 2020. A Joint Study of Seismicity and SAR Interferometry Observations for Assessing the Possibility of an Eruption of the Dormant Bolshaya Udina Volcano // Journal of Volcanology and Seismology, 2020, V. 14 (5), p. 305–317. DOI: 10.1134/S074204632005005X
Authors: Mikhailov, Valentin (1); Volkova, Maria (1); Timofeeva, Vera (1); Shapiro, Nikolay (1,4); Timoshkina, Elena (1); Senyukov, Sergey (1,2); Dmitriev, Pavel (1); Babayants, Igor (3)La Soufrière is an active stratovolcano on the island country of Saint Vincent and the Grenadines, with a crater lake inside the summit caldera. The last major historical eruption was recorded in 1902 and caused the loss of about 1600 lives. La Soufrière recently became active and a new lava dome in the main crater has been slowly growing. On 29th of December, 2020 the alert level for the volcano has been elevated to Orange, due to increased activity at the site such as gas and steam eruption. The monitoring of surface deformation with multi-temporal synthetic aperture radar (SAR) data can provide vital clues to assess the potential risk of an upcoming eruption at La Soufrière. In this study, ALOS-2 L-band SAR images from October 2014 to March 2021 have been analyzed in dual-pol fine mode (SM3) over the island of St. Vincent. Interferograms generated in this area commonly have low-coherence values caused by heavy vegetation, except for the summit area. In order to compensate for the temporal decorrelation of interferograms, the small baseline subset (SBAS) method was adopted to estimate the time series surface displacements of the volcano. Although there is a data gap of about one year prior to the most recent acquisitions, approximately 6-8 cm of ground movements in an upward direction have been observed according to the time series analysis over the past year. In addition, the uplift and subsidence of the dome area is detected using the time series analysis. The flanks of the volcano do not show significant deformation in the most recent acquisitions, even though the growth of the new lava dome from the amplitude time series images is clearly visible. Moreover, we reconstructed the three-dimensional surface displacement field using ascending and descending interferometric pairs acquired in January 2020 and January 2021, to better assess the recent volcanic activity. Surface displacements in line-of-sight and along-track directions by means of conventional and multiple-aperture SAR interferometry (MAI) techniques were integrated to retrieve the 3D components of surface deformation. The resulting 3D maps reveal the behavior of the volcano more clearly. The upward motions of areas at the summit have been identified despite lack of deformation in more recent data. It is anticipated that a more accurate magma model can be estimated using these 3D deformation maps.
Authors: Jo, Minjeong (1,2); Yun, Sang-Ho (3); Osmanoglu, Batuhan (1); Macorps, Elodie (1,4)About 730 volcanoes are located on the East African Rift (EAR), with 38 showing evidence of eruptive or unrest activity during the Holocene. It is not feasible to deploy and maintain permanent ground monitoring stations (GNSS, seismometers) over such a large number of volcanoes and therefore satellite radar interferometry (InSAR) is a great solution for detecting volcanic unrest at regional scale. Previous InSAR surveys have already detected volcanic unrest in the EAR providing constraints on the magma recharge in the rift system, which helps to better assess volcanic hazards (Biggs 2009, 2011). However, at that time, InSAR real-time monitoring was not possible due to several limitations such as the temporal resolution, the cost of the data or the processing time. With ESA’s Sentinel-1 mission launched in 2014, InSAR real-time monitoring is now feasible as data are freely available with a revisit of 12 days in Africa. In the past years, our research group COMET has developed LïCSAR, a system for the automated processing of Sentinel-1 SAR data, which provides updated InSAR products (e.g. wrapped/unwrapped interferograms, coherence and amplitude maps) to the community (http://comet.nerc.ac.uk/COMET-LiCS-portal/). At present time, LïCSAR has processed more than 55000 interferograms in the EAR, and we use this dataset to systematically produce InSAR time series in the EAR for the period 2014-2019. Our preliminary analysis showed that 9 volcanic systems are currently deforming with a large diversity in the spatio-temporal patterns detected: (i) subsidence related to contraction of magma bodies at Gada Ale and Dallol; (ii) large inflation of large magmatic systems at Corbetti and Tullu Moje; (iii) small inflation of shallow reservoirs at Nabro and Suswa; (iv) rapid inflation due to magma intrusions at Erta Ale and Fentale and (v) subsidence due to geothermal exploitation at Olkaria. Further investigations will include the modelling of the signals detected as well as the development of simple algorithms for the automatic detection of future anomalies or trend changes in the time series. Our final objective is to integrate our results to the COMET Volcano Database, a webtool initiative providing updated interferograms, time series and machine learning-based classifications.
Authors: Albino, Fabien; Biggs, JulietMiyakejima volcano is one of the active basaltic-andesitic volcanos in the Izu-Bonin arc, Japan. A caldera was formed on the summit of Miyakejima volcano during an eruption in 2000, which was the most recent explosive eruption. The caldera formation was driven by underpressure in a magma chamber and the collapse of a cylindrical structure caused by magma withdrawal associated with a dike intrusion prior to the 2000 phreatomagmatic eruption at the summit. Previous geological surveys have categorized the Miyakejima caldera as funnel-type and suggested that the caldera evolution is mainly driven by gravity. Recent satellite SAR observations have indicated a rapid deceleration of the caldera subsidence in mid-2009 based on SAR time-series analysis of PALSAR data acquired during 2006-2011 (Ozawa and Ueda, 2011). However, few spatiotemporal variations in seismicity on Miyakejima volcano or few significant variations in the GNSS baseline change rate were identified in the period. While the previous study proposed a flat elongated source in the northeast-southwest direction to reproduce the crustal deformations near the summit excluding the caldera, the physical interpretation of the rapid termination of the caldera subsidence is still poorly understood. We expect that examining the time-series of crustal deformation on the Miyakejima caldera would contribute to understanding the spatiotemporal caldera evolution during inter-eruption periods. In this study, we applied conventional multi-temporal InSAR (MTI) analysis to PALSAR/PALSAR-2 data to investigate the time-series of crustal deformation on Miyakejima volcano during the periods of 2006-2011 and 2014-2019. The PALSAR/PALSAR-2 MTI data suggested that the location of the maximum subsidence moved from the central part to the southernmost part of the caldera in mid-2009, that is, the spatial characteristics of caldera subsidence varied from the concentric patten (funnel-like) to the north-south asymmetric subsidence with a hinge at the northernmost caldera (trapdoor-like). The central caldera was the location of the initiation of the caldera collapse associated with the 2000 episode, and active fumaroles with volcanic gas and steam emission are located at the southernmost caldera. We also confirmed that the line-of-sight change rate at the central caldera decelerate in mid-2009 as the previous study showed. Therefore, the deceleration of caldera subsidence proposed by Ozawa and Ueda (2011) can be interpreted as a portion of the spatial variation of the caldera subsidence. Although the north-south asymmetric subsidence can be related to volcanic fluids, such as heated groundwater at the ambient aquifer, we have no way to explain the spatial variation in the caldera subsidence. We also found SAR image distortions (foreshortening or layover) of the caldera wall caused by the observation geometry of SAR image acquisition, however, the spatial variation of caldera subsidence is plausible because the spatial feature of caldera subsidence varied for the north-south direction, and the PALSAR/PALSAR-2 images were distorted for the east-west direction.
Authors: Himematsu, Yuji (1); Aoki, Yosuke (2); Ozawa, Taku (1)Radar Interferometry (InSAR) has been widely used to detect ground deformation associated to volcanic activity in the last ten years. The launch of Sentinel-1 in 2014 allowed the beginning of deformation monitoring systems based on this technique thanks to the free and open access strategy followed by the Copernicus Program. On the other hand, web-based platforms which allow download, visualization and processing of images have revolutionized the managing of InSAR workflows and analysis. In 2007 National Geographic Institute of Spain, following its responsibility ordered by law, started to develop a volcano monitoring system in The Canary Islands which include several geodetic techniques to provide the most complete picture of deformation possible. In this communication we focus on the monitoring of long-term processes where slow deformation velocities are expected (several mm per year). We have used the open platforms GEP (Geohazard Explotation Platform) and GPOD (Grid Processing on Demand) to process Sentinel-1 and Envisat data over the Canary Islands with a dual purpose. On one hand, resources needed to process large stacks of images can exceed possibilities of a medium size centre. These services provide results making use of the platform computational resources and applying largely tested methodologies. Results have provided the most relevant slow deformations during Envisat and Sentinel-1 life time, so they have allowed the detection of moving areas which can be studied in detail afterwards. On the other hand, as there are plenty of advanced InSAR methodologies available for the scientific community, these platforms provided us an excellent opportunity to validate their performance with external data. We present velocity and time series results obtained for Envisat and Sentinel-1 over Tenerife and Lanzarote islands applying P-SBAS (Parallel Small Baseline Subset) and FASTVEL approaches. These results have been validated with GPS data and compared with deformation maps obtained in the project “Sentinel-1 for geohazard prevention and forecasting” (Safety) which make use of different advanced InSAR strategies. This experience measn the beginning of a service based on the monitoring of long-term processes with InSAR techniques beyond the scope of volcanic monitoring in the Canary Islands to move to the study of geophysical phenomena in the Iberia Peninsula and influence area.
Authors: Gonzalez-Alonso, Elena (1); Fernández-García, Anselmo (1); López-Gea, José (2); Botija, Diego (2)Volcanic flank collapse has been responsible for over 20,000 casualties in the past 400 years, and is one of the most dangerous hazards affecting communities and infrastructure near volcanoes. In Central America, more than 40 volcanic debris avalanches resulting from flank collapse have been identified, 75% of which travelled over 10 km away from their source. In Guatemala, all but one volcano with elevation greater than 2000 m have experienced collapse, including Pacaya Volcano, an active basaltic stratovolcano where flank collapse took place between 0.6-1.6 ka, producing a 0.65 km3 debris avalanche that travelled 25 km away from the cone. Since then, the modern cone has built up within the collapse scarp and remained active since 1961. Slow long-term slip at volcanoes can lie below the detection threshold of conventional differential InSAR. At Pacaya, transient motion of the southwest flank associated to large eruptions in 2010 and 2014 has been identified through InSAR, however, due to a lack of continuous ground-based instrumentation and the decreased InSAR signal quality with time due to temporal decorrelation, the study of potential longer-term low amplitude slip signals with conventional InSAR has been challenging. With InSAR time-series analysis, on the other hand, lower displacement rates can be captured over longer periods, by combining information from multiple, partly overlapping, shorter time interval interferograms that jointly span the study period. Using this method, we reveal creep of Pacaya’s southwest flank between 2011 and 2013, a time when the volcano was relatively quiescent, followed by an increase in slip rate coincident with the timing of a large eruption in 2014. Results from separate geodetic data inversions for the quiescent and active periods support that the observed flank motion could result from slip on a southwest-dipping detachment fault, with the addition of a dike intrusion feeding the 2014 eruption. Static stress change analysis confirms that dike intrusion could punctuate slip on the modeled detachment fault, causing the observed increase in slip rate. Slip on the preferred detachment fault is also found to favor tensile fracturing above its top and below the volcano summit. Our results highlight the value of InSAR time-series analysis for detecting long-term flank motion and the requirement for studies to cover long enough intervals of volcanic quiescence as well as of magmatic activity in order to identify any potential gravity-dominated long-term instability as well as assess the effect of magmatic forces.
Authors: Gonzalez Santana, Judit (1); Wauthier, Christelle (1,2)Understanding volcanic unrest is crucial to forecasting eruptions. At active mafic calderas unrest culminates in eruption more frequently than at felsic calderas. However, the mafic caldera of Alcedo Volcano (Ecuador) has experienced repeated episodes of unrest without erupting, since at least 1992, when geodetic monitoring began. Here we investigate the unrest that occurred between 2007 and 2011 usinginterferometric synthetic aperture radar (InSAR) data and geodetic modeling. We observe an initial asymmetric uplift of the southern caldera floor (~30 cm of vertical motion) from 2007 to 2009, followed by subsidence of the uplifted area and contemporary uplift of the northwestern caldera rim between Januaryand June 2010. Finally, from June 2010 through March 2011, caldera uplift resumed. The first uplift episode is best explained by inflation of a sill and the activation of an inner ring fault. Successive caldera subsidence and rim uplift are compatible with the withdrawal of magma from the previously inflated sill and its northwestern migration. The resumption of uplift is consistent with the repressurization of the sill. This evolution suggests episodic magma emplacement in a shallow reservoir beneath the caldera, with aborted lateral magma migration, probably due to the discontinuous supply from depth. This short‐term deformation pattern matches well geological observations showing a longer‐term (hundreds of years at least) asymmetric uplift of the caldera floor, culminating in a weak resurgence of ~30 m. We propose that the monitored episodes of uplift represent short‐term stages of the rarely observed incremental growth of a resurgent basaltic caldera.
Authors: Galetto, Federico (1,2); Bagnardi, Marco (3,4); Acocella, Valerio (1); Hooper, Andrew (3)The Central Geophysical Observatory belonging to the National Geographic Institute of Spain has among its responsibilities: planning and management of surveillance and communication systems for volcanic activity institutions in the national territory and determination of associated hazards, as well as systems management of observation in the field of geodynamics, geophysics, volcanology, gravimetry and geomagnetism and the realization of related works and studies. In this framework of responsibilities, observation systems are multidisciplinary, including deformation, seismology, gravimetry, geochemistry and geomagnetism techniques. Among deformation measurement techniques, Spaceborne SAR interferometry (InSAR) is combined with others such as GNSS, inclinometers or robotic total stations. GNSS and tiltmeters provide daily and subdaily solutions in certain locations, while InSAR can monitor deformation process over broader areas, despite its lower temporal frequency, which makes InSAR a good complement to GNSS and tilt measurements. In recent years, a fully automatic processing methodology has been developed to obtain interferograms with each new image acquired by the Sentinel 1 Satellites (either A or B) over the Canary Islands. This automatic processing is included in the volcanic monitoring system in the Canary Islands, allowing to have a new interferogram for each island every 6 or 12 days. The methodology is based on the processing with SNAP software, and the SNAP's Graph Builder tool has been used to generate graphs that link different tools in unique processes. Coherence, displacement and interferometric phase maps are obtained as final products whithout the intervention of an operator. This poster presents the improvement of this methodology to be used in other areas and with other sensors, with the intention of being able to answer to any geophysical phenomena (volcanic or not) that could occur and be included among the responsibilities of the IGN. Moreover, other improvements in the methodology are explained, such as the projection of GNSS data to the Line of Sight (LOS) direction and the efforts focused to eliminate atmospheric artifacts. The projection of GNSS data to LOS direction allows to compare and validated InSAR data in these location where GNSS stations are placed. On the other hand, atmospheric artifacts are specially frequent in the Canarian archipielago, so its elimination is essential to make the use of interferograms in volcano monitoring as useful as possible.
Authors: Fernández-García, Anselmo (1); González-Alonso, Elena (1); García-Cañada, Laura (1); Lamolda, Héctor (1,2); López-Gea, José (3)We present Sentinel-1 measurements of uplift at Sangay Volcano, Ecuador, spanning a period of intense eruption in September 2020. Sangay (5230 m) is the most active volcano in southern Ecuador and has had eruptive activity intermittently since 1628. The most recent eruptive episode began in May 2019 and continues through February 2021,and is characterized by continuous explosions/emissions that formed 1-10 km high ash plumes, lava flows of several km length, and pyroclastic flows emitting from the summit. The volcano is located in a remote location, surrounded by rainforest, limiting access to install ground-based monitoring stations. In Ecuador, Synthetic Aperture Radar Interferometry (InSAR) is an especially useful technique for monitoring large-scale surface deformation at remote volcanoes, and is an essential complement to ground-basned , providing constraints on magma locations and volumes. We present Sentinel-1 time series using LICSBAS and LiCSAR interferograms between August 2019 and September 2020 from 60 and 40 images (descending and ascending tracks, respectively). We apply GACOS atmospheric corrections, and focus our analysis on the two last notable eruptive events on 08 June and on 19 September 2020. Line Of Sight (LOS) uplift of ~70 mm spanning the August 2019 to September 2020 period is measured in both asending and descending imagery, which we decompose to estimate to east-west and up-down components of motion. We will present preliminary analytical source modelling results to used assess depth and volume change of the best-fit geodetic source, and incorporating a topographic correction. We also present a comparison of the Sentinel-1 observations with those from CosmoSkyMed imagery spanning the September 2020 event, and discuss the suitability of both datasets for volcano monitoring at Sangay.
Authors: Espín Bedón, Pedro Alejandro (1,2); Ebmeier, Susanna (1); Wright, Tim (1); Elliott, John (1); Mothes, Patricia (2); Andrade, Daniel (2)Santiaguito volcano is a complex of four volcanic domes which are located at the flank of Santa Maria’s volcanic edifice in Guatemala. Activity at Santiaguito volcano began in 1922, 20 years after an eruption of Santa Maria with VEI 6. Today, only the oldest of the four volcanic domes, Caliente, is still active. At Santiaguito volcano, eruptions occur daily with varying frequency (e.g. 10-15 times per day) and also rockfalls, pyroclastic flows and lava flows are common features of Santiaguito’s activity. The lava is blocky, and the longest lava flow is almost 4km long. Here we present new digital elevation models (DEMs) of Santiaguito with a resolution of approximately 6.5m in north-south and 4-6m in east-west direction from data acquired between September 2011 and April 2019. The DEMs were generated from TanDEM-X data using the DORIS software (Delft Object-oriented Radar Interferometric Software). As the TanDEM-X mission consists of two satellites which acquire data simultaneously, the data are very well suited for volcanic areas, as only a single pass over the volcano is necessary to generate one DEM. This way atmospheric disturbances, temporal decorrelation and decorrelation due to deformation are negligible. We analyse a series of DEMs to determine elevation changes and associated volume estimates of lava flows focussing on the southern flank of the volcano as well as its crater. The crater region shows strong variations of elevation over the whole observation period. Between 2011 and 2016 a decrease of approx. 30-35m can be observed and between 2016 and 2019 parts of the crater show an increase in elevation of approximately the same magnitude. On the southern flank, the strongest variations took place between 2011 and 2016 while hardly any change can be observed between 2016 and 2019. Within the observation period, several lava flows originated, and the origin times of these new lava flows determined from the DEMs fit very well with the times given in literature. Looking at the volume changes in different parts of the southern flank, it can be seen that the lava flows were mostly generated only within a limited time interval. Before that and afterwards, the variations in volume are comparably small. Furthermore, in some parts of the southern flank, the variations in volume indicate that they might not necessarily be caused by lava flows but instead by lahars. The next step will be to investigate the elevation and volume changes in more detail, together with an assessment of the accuracy of the topographic and volumetric change estimates. Our study and results highlight the unique characteristic of single-pass SAR data for DEM generation and its usefulness in advancing our understanding of active volcanoes and highlights the need for future bistatic SAR mission concepts, such as ESA’s EE10 Harmony mission candidate.
Authors: Edrich, Ann-Kathrin (1); Kubanek, Julia (2); Rietbrock, Andreas (3); Gottschämmer, Ellen (3); Heck, Alexandra (3); Kamm, Bettina (3); Westerhaus, Malte (3); Kutterer, Hansjörg (3)One of the most significant discoveries of InSAR geodesy in volcanology is that large silicic volcanoes like Yellowstone, Long Valley and Campi Flegrei undergo transient pulses or cycles of ground uplift followed by periods of either quiescence or ground subsidence. These uplift events have velocities of ~1-10 cm/yr, spatial scales that vary from ~15 km to more than 100 km, and time scales from ~6 months up to at least half a century. These signals have been interpreted as being produced by either magma intrusion, volatile exsolution, fluid flow in the hydrothermal systems that are located in several of these systems, or a combination of these processes. However, inherent ambiguities in the interpretation of geodetic data have not allowed to unravel the physical mechanism of ground uplift for most of them. The role of magma injection has been acknowledged in recent conceptual models of fluid migration beneath Yellowstone, but a proper quantitative understanding is still lacking, particularly in its relation with the near constant state of unrest of the caldera that results in cycles of uplift and subsidence. Further, the large hydrothermal system of Yellowstone complicates the understanding of the driving mechanism of unrest and to what extent ground deformation results from magmatic, hydrothermal or a combination of these processes. The 2004-2009 caldera uplift is the largest instrumentally recorded episode of unrest at Yellowstone caldera. We use GPS and InSAR time series from ENVISAT, ALOS-1, and TerraSAR-X spanning 2004-2015, with a focus in the aforementioned event to understand the mechanisms of unrest. InSAR data recorded∼25 and∼20 cm of uplift at the Sour Creek (SCD) and Mallard Lake (MLD) resurgent domes during 2004-2009, and∼8 cm of subsidence at the Norris Geyser Basin (NGB) during 2004-2008. The SCD/MLD uplift was followed by subsidence across the caldera floor with a maximum at MLD of ∼1.5-2.5 cm/yr and no deformation at NGB. The best-fit source models for the 2004-2009 period are two horizontal sills at depths of ∼8.7 and 10.6 km for the caldera source and NGB, respectively, with volume changes of 0.354 and -0.121 km3, and an overpressure of ∼0.1 MPa. The InSAR and GPS time series record exponentially increasing followed by exponentially decreasing uplift between 2004 and 2009, which is indicative of magma injection into the caldera reservoir, with no need for other mechanisms. However, magma extraction from NGB to the caldera is unable to explain the subsidence coeval with the caldera uplift. Models of magma injection can also explain other episodes of caldera uplift like that in 2014-2015. Distributed sill opening models show that magma is stored across the caldera source with no clear boundary between MLD and SCD. Since the magma overpressure is orders of magnitude below the tensile strength of the encasing rock, historical episodes of unrest like these are very unlikely to trigger an eruption.
Authors: Delgado, Francisco (1,2); Grandin, Raphaël (1)Ground deformation is one of the key parameters to be considered in volcano monitoring. The availability of Synthetic Aperture Radar (SAR) measurements provides, among several, accurate information on the volcano morphology and deformation, particularly when eruptions and lava flows can hamper the use of other remote and in-situ techniques. In this paper, we present the activities relevant to the ground deformation monitoring of the Stromboli volcano (Italy) through SAR sensors, performed by IREA-CNR (Institute for the Electromagnetic Sensing of the Environment) and UNIFI (University of Florence) as Centres of Competence (CoC) for the Italian Civil Protection Department. Stromboli is one of the most well-known volcanoes in the Earth and its persistent activity consists of frequent, small scale, explosions (Strombolian activity). The most hazardous phenomena at Stromboli Island are the tsunamis, induced both by intrusion-related landslides from the Sciara del Fuoco (SdF) unstable flank, and by the entry into the sea of pyroclastic density currents produced during high-intensity explosions, being the last ones occurred on 3rdJuly and 28thAugust of 2019. For these reasons, among other phenomena, Stromboli’s ground deformation is constantly monitored through two GB-InSAR systems, managed by UNIFI, installed on the northern edge of the SdF at two different heights (400 m and 190 m a.s.l.) to monitor the NE portion of the summit crater terrace and the northern portion of the SdF. Moreover, thanks to the Copernicus Sentinel-1 mission, since 2015 an extensive space-borne SAR data set has been systematically acquired from ascending and descending orbits with a revisit time of six days from September 2016. S1 data are then constantly processed by IREA-CNR through the P-SBAS-DInSAR algorithm to obtain displacement time-series on a monthly basis. According to their respective activities, both CoC periodically report to the Italian Department of Civil Protection about the deformation status of the volcano. In this work, we present the results of the monitoring activities carried out by both the CoC over Stromboli Island. In particular, we show the displacement time series obtained with Sentinel-1 data acquired from March 2015 to October 2019 over the whole island from ascending and descending orbits, and the displacement estimated with the GB-InSAR in the same period. Moreover, by combining the displacement measurements retrieved with both monitoring systems, which are characterized by independent acquisition geometries, allowed us to partially reconstruct a 3D deformation field of Sciara del Fuoco area with an unprecedented detail. Furthermore, insights on the volcano crisis occurred from 3rd July 2019, by using DInSAR measurements, is also provided. Finally, we show the preliminary result of a test about an operational monitoring service based on new methodologies for the processing of airborne SAR data, aimed at evaluating its relevance for Civil Protection purposes in emergency context, as during a volcanic eruption. This work has been supported by the 2019-2021 CNR-IREA and Italian Civil Protection Department agreement, the EPOS-IP and EPOS-SP projects of the European Union Horizon 2020 R&I program (grant agreement 676564 and 871121) and the I-AMICA (PONa3_00363) project.
Authors: De Luca, Claudio (1); Di Traglia, Feredico (2); Manzo, Mariarosaria (1); Berardino, Paolo (1); De Novellis, Vincenzo (1); Esposito, Carmen (1); Natale, Antonio (1); Nolesini, Teresa (2); Perna, Stefano (3); Tizzani, Pietro (1); Casagli, Nicola (2); Lanari, Riccardo (1); Casu, Francesco (1)Changbai Mountain Tianchi volcano, located in the southeastern part of Jilin province on the border between China and North Korea, is an active volcano with the most eruptive potential in China. In history, the volcano has experienced several large eruptions, of which the 1000-year-old eruption about 1000 years ago was one of the largest eruptions in the world in the past 2000 years. From 2002 to 2005, the frequency of seismicity in Tianchi volcanic area increased significantly, and the magnitude of the earthquake also became larger than usual, indicating that the volcanic activity entered an active period, and there was a risk of eruption. Since then, the volcano has returned to a quiet period. The latest monitoring results show that on December 22, 2020, a volcanic earthquake cluster suddenly appeared in the Tianchi volcanic area, with 38 volcanic earthquake events of various types occurred. This phenomenon is worthy of attention. Therefore, the SBAS-InSAR combined with Mogi inversion is used to capture the activity of the volcanic area before the occurrence of this phenomenon.The traditional methods of volcanic activity monitoring include GPS, leveling, etc. These methods have the shortcomings of sparse monitoring points and low accuracy, which can not reflect the overall distribution of deformation. Time-series Interferometric Synthetic Aperture Radar (InSAR) can remove the influence of orbit, atmosphere, DEM and other errors in interferometric phase by analyzing long time series SAR data, and obtain large-scale and high-precision deformation monitoring results. Among the commonly used time-series InSAR methods, the Small Baseline Subset time-series InSAR(SBAS-InSAR) has high image utilization, high spatial density and short baseline interference pairs, which effectively reduces the impact of spatio-temporal decorrelation. Because of the vegetation coverage and lack of Permanent Scatterers in Tianchi volcanic area, SBAS-InSAR method is used to obtain the surface deformation of the volcanic area. Mogi model has been applied to the deformation simulation of many volcanic regions at home and abroad because of its strong applicability. In an elastic half-space, when the radius of the pressure source is much smaller than the depth, it can be regarded as a point, and the relationship between the vertical and horizontal displacements caused by explosion or contraction and the parameters of the pressure source is obtained.In this study, based on 19 ALOS images and 63 Sentinel images from November 2018 to December 2020, the surface deformation of Tianchi volcanic area is obtained. The results of these two sets of data can be used as a cross validation for comparison. The results based on ALOS data show that the surface of the area near the crater is obviously uplifted during this period, and the line of sight deformation rate values of the deformation point targets are mostly between -20mm/a and 20mm/a, and the cumulative deformation values are between -50mm and 50 mm, indicating that the underground magma chamber is in an active state of expansion. Then, based on the Mogi model, the cumulative deformation field obtained by SBAS-InSAR is simulated, and the spatial location of magma chamber in Tianchi volcanic area is determined, which is located in the north of the volcano, about 4 kilometers away from the center of Tianchi, with a depth of more than 3 kilometers. The results based on Sentinel data will be presented in a future text.
Authors: Zhang, Jiaqi (1); Wei, Lianhuan (1); Liu, Guoming (2,3); Tolomei, Cristiano (4); Ventura, Guido (4)Spaceborne synthetic aperture radar (SAR) has been providing observations of ice type and conditions, however, sea ice thickness is not currently directly measured from space SAR. This paper presents the detection of surface height changes and deformations over Arctic land fast sea ice using SAR satellite data from ALOS-2 (L-band), Sentinel-1 and Radarsat-2 (C-band) sensors. We investigated the potential of measuring land fast sea ice surface height changes using SAR data, and compared these measurements with that from ICESAT-2 and field measurements of ice draft. Interferometric SAR (InSAR) technique was applied to process ALOS-2, Sentinel-1 and Radarsat-2 data. Small baseline subset (SBAS) approach was explored to process time series SAR observations for retrieval of temporal changes over the winter. Circular fringes were observed over sea ice during January-April and may represent vertical deformation corresponding to steady ice growth. Results indicated that uplift and subsidence patterns observed from SAR sensors varied in different locations. Our InSAR results indicated that a maximum of 57 cm deformation was detected from ALOS-2 data observation, 30 cm from Sentinel-1 and 13 cm from Radarsat-2 data in Cambridge Bay, Nunavut, Canada over 3 months during the winter of 2018-2019. InSAR coherence from L-band ALOS-2 data was high and lasted over 3 months interval during January-April. Compared to ALOS-2 data, high InSAR coherence from C-band Radarsat-2 did not last more than 24 days. Coherence from Sentinel-1 and Radarsat-2 data was good during February-May with 11- and 24-days intervals, but was poor during December-January. Comparison between the ice surface height from ICESAT-2 and the SAR intensity from ALOS2 and Radarsat-2 indicated that SAR intensity fluctuated in response to ice surface height changes, the correlation was not strong. The correlation coefficient between InSAR backscatter from Sentinel-1 and surface height was in the range 0.49-0.83. Comparison between ice surface height from ICESAT-2 and the SAR intensity from ALOS-2 and Radarsat-2 was limited due to the spatial coverage of ICESAT-2. Results indicated that the satellite data are useful to provide surface height information over Arctic land fast sea ice. SAR interferometry shows promise for investigating the Arctic sea ice surface height changes spatially and temporally. However, more data is needed to validate the satellite monitoring of sea ice thickness changes.
Authors: Chen, Zhaohua; Montpetit, Benoit; Banks, Sarah; Behnamian, Amir; White, Lori; Duffe, Jason; Pasher, JonHigh mountain environments are undergoing major changes due to the impact of ongoing climate change. A large variety of geomorphological processes – often showing accelerating magnitudes in the last two decades – have been reshaping high mountain environments in recent years. Multi-temporal differential SAR interferometry (MTI) can deliver surface displacement rates in the order of a fraction of the used wavelength. However, in high mountain environments, several limiting factors constrain the applicability of active remote sensing methods and or strongly affect the accuracy of the achieved results. The study area covers the upper catchment of the Pasterze Glacier area, the largest glacier in Austria. The Pasterze glacier is located directly beneath Austria’s highest mountain, the Grossglockner). The study area is covered by three (2 ascending and one descending orbit) passes of the Sentinel-1 satellites, thus enabling detailed analysis of SAR imaging geometry and its effects on MTI results. Furthermore, the area is characterized by a strong seasonal variability of meteorological parameters. This has a significant impact on the coherence of the interferograms, which showed substantial temporal decorrelation from September to June in the three years of observation. We tested three different approaches for generating deformation maps, one by employing persistent scatterer interferometry (PSI) software and two small baseline subset algorithm (SBAS) implementations. Sentinel-1 processing for PSI was performed utilising a SNAP-StaMPS (Stanford Method for Persistent Scatterers) toolchain. SBAS processing was performed incorporating the P-SBAS (Parallel SBAS) service of the Geohazards Thematic Exploitation Platform (GEP) and using the prototype implementation of the Subsidence and Landslide Monitoring Service of Austria (SuLaMoSA).The quality of the resulting nine products (3 orbits times 3 methods/implementations) was assessed using manually selected, well distributed stable areas (bedrock) in the Pasterze Glacier area and one known landslide right outside the effective glacial area. We assessed the spatial as well as temporal coverages of the products as well as relative and absolute accuracy of the derived deformation rates. Very local atmospheric turbulences, which are not adequately modelled in globally available reanalysis models like ERA5, and the high number of decorrelated interferograms were identified as the most challenging limitations of either MTI analysis in the harsh high mountain environment. Therefore, we finally describe some improvements which were implemented in the SuLaMoSA workflow to increase the applicability of the SBAS method in high mountain areas.
Authors: Gutjahr, Karlheinz (1); Avian, Michael (2); Schlögl, Matthias (2); Widhalm, Barbara (2)The Snow Water Equivalent (SWE) is an essential variable for hydrological and climate models, as well as for flood predictions and water resource management. In-situ measurements of snow parameters can provide precise information, but can only be performed on a limited number of locations. Remote sensing offers the possibility to cover large areas with a high temporal sampling. Especially Synthetic Aperture Radar (SAR) is suitable for snow monitoring, as it can be operated independently from weather and illumination conditions with a spatial resolution on meter scale. Moreover, radar signals up to Ku-Band easily penetrate dry snow, which typically behaves as a non-scattering medium. However, a dry snowpack introduces refraction as well as a signal delay related to its dielectric properties, which results in an interferometric phase term that can be measured with differential interferometry. Studies have shown that this phase difference between two SAR acquisitions of the same scene is directly related to the SWE change [1]. In [2], the model-based retrieval algorithm of [1] was extended to make it applicable for all realistic densities. However, the model relies on the assumption of dry-snow conditions and is tailored to the simple case of open areas (i.e. snow over ground) as the effect of vegetation is not considered. Since the measured phase difference lies in an interval between [-π, π], only a limited range of SWE variations can be estimated unambiguously, while values exceeding this interval lead to wrapping of the interferometric phase signal. This model was employed to retrieve SWE very accurately from ground-based radar data with a short revisit time of 4 hours [2]. However, its application to spaceborne data has been hampered so far by the stringent observation requirements, including the availability of time series data with dense temporal sampling to avoid a loss of coherence, which can result for instance from melting or from a redistribution by wind. However, an increasing number of spaceborne SAR sensors has been launched in the last years, providing denser temporal samplings at different frequencies. This study investigates the potential of multifrequency spaceborne SAR data for SWE retrieval based on the algorithm proposed in [2]. For that purpose, time series data of different satellites are compared. In particular, TerraSAR-X (X-Band), Sentinel-1 (C-Band) and ALOS-2 (L-Band) are chosen for this investigation. The temporal behaviour of the interferometric coherence amplitude and phase terms is analyzed for different locations and the impact of the radar wavelength is evaluated. One study area is the region around the city of Sodankylae, Finland, where ground measurements of snow properties and air temperature are available. These are used to gain a deeper understanding of the temporal changes of the scene between the satellite acquisitions to interpret the coherence behaviour. Furthermore, the influence of different revisit times of the satellites is analyzed. As a next step, the algorithm from [2] is used for the retrieval of SWE. Preliminary results show that, for instance, in the X-Band spaceborne case the main limitation for SWE estimation is represented by phase wrapping that occurs when snow accumulation exceeds a given threshold. This threshold depends on the radar frequency and is calculated for the different cases (X-, C- and L-band) in order to assess the potential and limitations of each dataset. Under the same observation geometry, this threshold is around 7 times higher for L-Band compared to X-Band, allowing a greater SWE change between two observations before phase wrapping occurs. However, radar waves with lower frequencies may be influenced by the ionosphere. The results of these investigations will be presented at the conference. [1] T. Guneriussen, K. A. Hogda, H. Johnsen and I. Lauknes, "InSAR for estimation of changes in snow water equivalent of dry snow," in IEEE Transactions on Geoscience and Remote Sensing, vol. 39, no. 10, pp. 2101-2108, Oct. 2001. [2] Leinss, A. Wiesmann, J. Lemmetyinen and I. Hajnsek, "Snow Water Equivalent of Dry Snow Measured by Differential Interferometry," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 8, no. 8, pp. 3773-3790, Aug. 2015.
Authors: Belinska, Kristina (1,2); Parrella, Giuseppe (1); Hajnsek, Irena (1,2)The grounding line location (GLL) is one of the 4 parameters characterizing the Antarctic Ice Sheet ECV within ESA’s CCI+ programme. The geophysical product was designed within the previous phase (CCI) and since then was derived from TerraSAR-X, Sentinel-1, ALOS-2 and ERS-1/2 SAR data over major ice streams and outlet glaciers around Antarctica through the double difference InSAR technique. In the current stage of the CCI+ project dedicated to the Antarctic Ice Sheet the Sentinel-1 A/B constellation is the main data source used to obtain GLL over the margins of Antarctica. We have been processing time series of double difference interferograms throughout the year aiming at monitoring the short term migration of the dense fringe belt in respect to different tidal and atmospheric conditions. Our GLL production system can be roughly divided into two stages. In a first step DLR’s Integrated Wide Area Processor (IWAP) is used to determine single and double difference interferograms from SAR data. The second step is the mapping of the upper limit of the tidal induced ice shelf flexure. The original GLL production system was based on a manual delineation of the landward limit of the fringe belt which is usually clearly recognizable in the generated double difference interferograms. With increasing availability of SAR data preserving phase coherence the development of an automatic approach for mapping the GLL was required. We present a possible method able to identify and measure the width of the grounding zone in a more automatic fashion. Since the wrapped interferometric phase is inherently a measure of the surface deformation gradient, it can be used as tilt measurement. The advantage is to bypass the phase unwrapping that could harm the robustness of the approach. The algorithm starts from coarse a-priori information about the location of the grounding zone. The gradient information is used both to identify and follow the geometry of the grounding zone and to fit the gradient data with an elastic beam model hence estimating the model parameters. This allows refining the a-priori location - more precisely the position of the hinge line - as well as deriving the shape factor of the ice flexure (which is dependent on the grounding zone width and ice thickness) and the tide difference. The approach can be applied on single and, with some assumptions, on double difference interferograms and is tested in various situations starting with high coherent interferograms and simple shapes to more complex situations (lower coherence, presence of noise, complicated geometry). Validation scenarios include comparison to manually delineated grounding lines and to recently published datasets based on machine learning. Sentinel-1 A/B SAR triplets (acquired during three consecutive repeat pass acquisitions) have been processed as described above over two test areas: Getz Ice Shelf (78.81°S, 127.45°W) and two small ice shelves surrounding the Schirmacher Oasis (70.36°S, 11.38°E). The delineation of dense time series of GLLs should allow to investigate the correlation with the ocean tide level within one tidal cycle. We interpret the short term displacement of the grounding line and the changes in the width and shape of the fringe belt at the grounding zone taking various parameters into account. These include the modelled ocean tide level, atmospheric conditions, bedrock topography and SAR imaging geometry. Our investigations are a step towards the interpretation of grounding line retreat observed over long time periods as an effect of ice thinning and increasing ice shelf instability.
Authors: Floricioiu, Dana (1); Krieger, Lukas (1); Parizzi, Alessandro (1); Ip, Yin Ying (1,2)Mountain glaciers have been identified to be major contributors to sea level rise in the most recent decades. In support, accelerated retreat and surface depletion have been reported by numerous studies from various glaciated regions. The aim of this study is to provide information about the elevation change of globally distributed mountain glaciers between 60° N and 56° S and respective geodetic mass balance. We investigate alpine glaciers and ice caps, a priori excluding the two major ice sheets of Greenland and Antarctica. Due to the aforementioned constraints towards the poles, some Randolph Glacier Inventory (RGI) regions had to be omitted while others were cut to smaller proportions. In total, the satellite observations cover about 148.000 km² of glaciated ground from 12 RGI regions. Different to previous studies, we aim at mass balance results based on one globally consistent geodetic method. In our approach, digital elevation model datasets (DEMs) from two SAR satellite missions are compared to derive elevation change on glaciated areas: 1) Shuttle Radar Topography Mission (SRTM) and 2) Terra Sar-X Add on for Digital Elevation Measurement (TanDEM-X). The elevations of SRTM were chosen as a reference dataset due to its mission procedure. A global DEM was acquired in less than two weeks, enabling to refer to the elevations as a status information of a very distinct time. We assume the mean mission date 16th of Feb as date of acquisition for the whole dataset in later epoch calculations. From various DEM products of SRTM, version 3.0 was chosen due to its enhanced resolution (1 arc sec). We reintroduced all data voids that were filled with third party DEM values in SRTM 3.0. Thus, we could retain the advantage of an exact reference date from the original dataset. As a first and major mission goal TanDEM-X collected satellite imagery for a global DEM. The resulting data archive provides numerous scenes of descending and ascending orbits, from which we chose 4367 to work with. To analyse TANDEM-X elevations, we downloaded archived TanDEM-X CoSSC datasets from the global DEM mission phase For each scene, time stamps were chosen to be in the closest proximity possible to a full year cycle of SRTM reference date. The selected datasets were processed via a highly automated dInSAR processing chain In a first step, data scenes from a consecutive acquisition in along track direction are re-concatenated. For each take, a differential interferogram is calculated with SRTM serving as elevation reference. Following the phase unwrapping, the differential phase is transferred into differential elevations. The reference elevation of SRTM is re-added, to obtain absolute heights for the resulting DEM. Prior to the calculation of elevation differences, a post-processing pipeline is applied to compensate for systematic errors. This includes iterative deramping, horizontal alignment and vertical referencing along the elevation deviations on stable ground. To derive elevation change rates, both DEMs are resampled to the common resolution of 30x30 m and projected to the respective UTM zone. The elevation difference is calculated for each grid cell. A timestamp, referring to the difference of the exact acquisition dates is stored for each pixel correspondingly. As a main result we present a plot of the global dataset as mean aggregated elevation change rates (dh/dt) per 1x1° tile. The overall tendency of elevation change exhibits an overbalance of depletion, with values ranging up to a of -3.14 m/a. Remarkably, the regions of Tien Shan, Kunlun Shan and Tibetan Plateau provide several tiles with average positive values, with a 0.54 m/a maximum. Concerning the non-aggregated, high resolution dh/dt datasets, the peak depletion rates are located in Patagonia, South America. Outlet glaciers of the Southern Patagonia Icefield show surface elevation exceeding -15 m/a at their termini, with outstanding ‑51 m/a at glacier HPS 12. For conversion to mass change values, elevation change datasets are evaluated in subregions, chosen based on topographical and climatic indicators. For South America, High mountain Asia and Europe the subdivisions are aligned with previous publications. Our results show a total glacier mass change rate of -19.43±0.60 Gt a-1 for the South American continent over the observation period. This loss is predominantly driven by the large Patagonian icefields and its outlet glaciers. Nevertheless some of the tropical glaciers show significantly negative specific mass balance in the range of regions in Tierra del Fuego. For High Mountain Asia (RGI regions 13-15), the Himalaya, Nyainqentangla, Hindu Kush and Karakoram ranges results to a mass change rate of -13.3 ± 6.5 Gt a-1. The estimations for bigger remaining regions are: 21.8 ± 2.7 Gt/a for North America (~43,000 km², cut by 60° N) and ‑1.3 ± 0.1 Gt/a for the Alps (2013 km²) and -2.28 ± 0.19 Gt/a for South Georgia. The profound data base of the global DEM mission phase of TanDEM-X serves as an excellent archive to derive this close to global glacier change results. A repeated global acquisition, launched from 2017 and running up to date, will enable us to analyse a follow up timestep for this analysis. On top, the comparison of TanDEM-X datasets from different acquisition periods eliminates the flaw of SRTMs shuttle orbit and thus will enable us to extend our study towards the poles. To provide repeated monitoring, using DEMs derived from X-Band imagery on glaciers an unbroken continuation of respective missions is highly valuable and recommended.
Authors: Malz, Philipp; Seehaus, Thorsten; Sommer, Christian; Farias, David; Braun, MatthiasWe present a generic open-source feature tracking routine that can be applied to any type of imagery to measure sub-pixel displacements between images. The routine builds upon past efforts with the implementation of an efficient feature tracking algorithm for the mass processing of satellite images. With a feature tracking module (autoRIFT) that enhances computational efficiency and a geocoding module (Geogrid) that mitigates existing geocoding issues, the routine is shown to provide two orders of magnitude of runtime improvement with a 20% improvement in accuracy. With regard to satellite imagery, autoRIFT can run on a grid in the native image coordinates (such as radar or map) and, when used in conjunction with the Geogrid module, on a user-defined grid in geographic Cartesian coordinates such as Universal Transverse Mercator or Polar Stereographic. We validate the efficiency and accuracy of this routine by applying it to track ice motion through use of ESA’s Sentinel-1A/B radar and NASA’s Landsat-8 optical data collected over Greenland’s Jakobshavn Isbræ glacier in 2017. Feature-tracked velocity errors are characterized over stable surfaces and different error sources for radar and optical image pairs are investigated, where the seasonal variation and the error dependence on the temporal baseline are analyzed. It is found out that, the best Sentinel-1A/B pair with a 6 day temporal baseline has errors in X/Y of 12 m/year or 39 m/year, compared to 22 m/year or 31 m/year for Landsat-8 with a 16-day time separation. Estimated velocities were compared with reference velocities derived from DLR’s TanDEM-X SAR/InSAR data over the fast-moving glacier outlet. It is observed that the Sentinel-1 results agree with TanDEM-X estimates within 4% compared to 3–7% for Landsat-8. A comprehensive apples-to-apples comparison was made between multiple implementations of the proposed routine and the widely-used “dense ampcor” program from NASA/JPL’s ISCE software. The routine (freely available to public at https://github.com/leiyangleon/autoRIFT) was developed as part of the Inter-mission Time Series of Land Ice Velocity and Elevation (ITS_LIVE; https://its-live.jpl.nasa.gov), a NASA MEaSUREs project, where a standalone version of autoRIFT has been used to generate global land ice displacement velocities from the full archive of Landsat 4/5/7 & 8 imagery. Its integration into the ISCE radar processing software (https://github.com/isce-framework/isce2) and Alaska Satellite Facility’s HyP3 online processing service (https://hyp3.asf.alaska.edu) now allow for the global mapping of ice displacement from both optical and radar imagery as well as multiple spaceborne sensors using a single algorithm on identical output grids. The orders of magnitude reduction in computational cost demonstrated with the new algorithms is particularly relevant for NASA’s upcoming NISAR mission, which will downlink more data than any prior mission ever did. Besides tracking ice displacement, the developed routine and algorithm can also be applied to any other feature tracking applications, such as subsidence, earthquake, land slide and etc.
Authors: Lei, Yang (1); Gardner, Alex (2); Agram, Piyush (1)Permafrost, defined as a combination of ice and various rocks below 0 ℃ over 2 years, is particularly susceptible to ambient interference and climate change. Seasonal variations of hydrothermal conditions lead the top of permafrost to thaw or freeze, thus bringing subsidence or uplift deformation to surface. Permafrost on Qinghai-Tibet Plateau (QTP) with widely distribution has the characteristics of higher temperature and stronger spatial heterogeneity, when it is compared with polar permafrost (Wu et al., 2008). Affected by the increasing mean annual ground temperature over the past several decades, permafrost degradation is generally occurring on QTP (Li et al., 2008). InSAR methods have plenty of advantages in monitoring surface deformation with decent temporal and spatial resolution as well as millimeter precision. In particular, multi-temporal InSAR analysis can remove the disturbances of inaccurate digital elevation model (DEM), atmospheric delays and spatiotemporal decorrelation. We employed two sequential InSAR analysis methods to retrieve surface historical deformation in typical permafrost regions on QTP, and compared possible external conditions such as temperature, precipitation and lake outburst which could lead permafrost degradation. StaMPS-InSAR method was applied on a small permafrost region near Wudaoliang County with short time series of Sentinel-1A images (spanning from March 2017 to June 2018). The deformation results revealed significant seasonal changes of permafrost in a freeze-thaw cycle with lagging days between maximum deformation and temperature. Simultaneously the lagging days about 53 in thawing period is longer than about 34 lagging days in freezing period. In addition, a lake called Salt Lake near the Qinghai-Tibet railway has been expanding dramatically until now after the outburst of Zonag lake in September 2011. We employed Google Earth Engine platform to bulk extract Salt Lake boundary from 2000 to 2019 approximately in monthly scale, and found that Salt Lake had increased 5.35 times in September 2019 than in January 2000. Meanwhile, the distance from the lake boundary to the Qinghai-Tibet railway was significantly reduced from approximately 14 to 8 km. In order to figure out potential relationship between permafrost degradation and lake outburst, surrounding lake surface deformation before and after the outburst of Zonag Lake was retrieved by applying SBAS-InSAR method on temporal Envisat (2003-2010) and Sentinel-1 (2014-2019) datasets. The larger subsidence after lake outburst provides an evidence that the outburst of a headwater lake may significantly accelerate the permafrost degradation surrounding the tailwater lake. Such degradation may be attributed to the thermal alteration of the permafrost thawing-freezing cycle and the melting ground ice, along with the subsequent changes on hydrological connectivity and soil permeability (Rowland et al., 2010; Karlsson et al., 2012). More attention is deserved to pay to global permafrost observation with frequent occurrence of extreme weather in the future. The two successful applications have showed great prospects of sequential InSAR analysis for permafrost degradation monitoring and further thermal hazards detection. References Wu, Q., & Zhang, T. (2008). Recent permafrost warming on the Qinghai-Tibetan Plateau. Journal of Geophysical Research Atmospheres, 113(D13). Li, X., Cheng, G., Jin, H., Kang, E., Che, T., Jin, R., Wu, L., Nan, Z., Wang, J., & Shen, Y. (2008). Cryospheric change in China. GLOBAL AND PLANETARY CHANGE, 62(3-4), 210-218. Rowland, J. C., Jones, C. E., Altmann, G., Bryan, R., Crosby, B. T., Hinzman, L. D., Kane, D. L., Lawrence, D. M., Mancino, A., Marsh, P., McNamara, J. P., Romanvosky, V. E., Toniolo, H., Travis, B. J., Trochim, E., Wilson, C. J., & Geernaert, G. L. (2010). Arctic Landscapes in Transition: Responses to Thawing Permafrost. EOS TRANSACTIONS AMERICAN GEOPHYSICAL UNION, 91(26), 229-230. Karlsson, J. M. R., Lyon, S. W., & Destouni, G. (2012). Thermokarst lake, hydrological flow and water balance indicators of permafrost change in Western Siberia. JOURNAL OF HYDROLOGY, 459-466.
Authors: Han, Jiangping; Lu, PingPresently, glaciers and ice sheets occupy approximately 10% of the Earth’s land surface (Lemke et al., 2007). Since the late 19th century, glaciers and ice sheets have been retreating in most regions of the world (Zemp et al., 2006) at fast rates. It is then of utmost importance to monitor the rates and processes beyond this glacier retreat. Satellite geodetic methods, based on a comparison of topographic data observed at more than two points of time, can measure glacier volume or height changes (Paul et al., 2015). Synthetic Aperture Radar (SAR) data and time-series Differential Interferometric SAR (DInSAR) techniques are potent geodetic tools for this purpose due to their fine-to-medium resolution and all day-night imaging capability. Even though new generations of satellites, such as the Copernicus Sentinel-1 (S1) open up new perspectives for continuous glacier monitoring due to its short revisiting time, coverage, and reliability of service, they have not been exploited thoroughly up till now especially for mass balance estimation of glaciers. In this study, we used SAR data to monitor the dynamics of five glaciers in the Tarfala region, located at 1130 m a.s.l. in the high alpine Kebnekaise Mountains of northern Sweden. We used 135 S1-A/B data in the descending and ascending modes with VV and VH polarizations during the years 2017 to 2019. The small baseline subsets (SBAS) technique was used to create the Line Of Sight (LOS) displacement map. The vertical and horizontal displacement maps were extracted using decomposition of LOS displacement components of the descending and ascending modes. The results showed that all five glaciers experienced a mass loss up to 1.2 m and the movement direction of the glaciers was properly detected with a maximum velocity of 2.3 m per years towards west. We could only observe a limited snow gain in the small glacier located in the southern part. To validate our results we will use the ground mass balance data of one of the glaciers at this location to estimate the accuracy of the mass balance estimated by time-series DInSAR and using 3D ice-flow numerical model to calibrate DInSAR processing and improve the accuracy of mass balance estimation. As earth observations data and modelling approaches are closely interconnected, therefore, data from monitoring can be used to calibrate or validate models, and model simulations help to understand information from monitoring programs and to design focused long-term observations. AcknowledgementsThis research was funded by the Artic Avenue 2019 program between Stockholm University and University of Helsinki. ReferencesLemke, P., Ren, J., Alley, R.B., Allison, I., Carrasco, J., Flato, G., Fujii, Y., Kaser, G., Mote, P., Thomas, R.H., others, 2007. Observations: changes in snow, ice and frozen ground.Paul, F., Bolch, T., Kääb, A., Nagler, T., Nuth, C., Scharrer, K., Shepherd, A., Strozzi, T., Ticconi, F., Bhambri, R., others, 2015. The glaciers climate change initiative: Methods for creating glacier area, elevation change and velocity products. Remote Sens. Environ. 162, 408–426.Zemp, M., Haeberli, W., Hoelzle, M., Paul, F., 2006. Alpine glaciers to disappear within decades? Geophys. Res. Lett. 33.
Authors: Darvishi, Mehdi (1); M. Nia, Mohammad (2); Jaramillo, Fernando (1,3)Permafrost, permanently frozen ground, is prevalent across northern Canada and in many other countries at polar latitudes. Above this permafrost lies an active layer which thaws in the summer and refreezes in the winter. Depending on the water content of this active layer, this seasonal cycle of thawing and refreezing can result in significant ground subsidence and heave as the moisture cycles back and forth between ice and liquid water. Climate change is causing Arctic temperatures to slowly increase, resulting in changes to the extent and depth of this active layer and subsequently leading to variability in the ground subsidence patterns in northern communities. Permafrost degradation has the potential to release carbon that is currently frozen in organic matter, in sufficient quantities to augment global climate change. The loss of permafrost can also lead to long term subsidence trends that can be tremendously damaging to both urban structures and long linear infrastructure projects, such as highways and pipelines. Monitoring the evolution of Arctic permafrost is therefore necessary to understand impacts on climate models, linear infrastructure and Arctic communities. The remote nature of the Canadian Arctic creates significant challenges for wide scale monitoring. Interferometric Synthetic Aperture Radar (InSAR) provides the potential for monitoring conditions across wide swaths of the Arctic, thereby providing regular monitoring of known geohazards and allowing for the identification of thousands of previously unknown hazards. 3vGeomatics (3vG), with the funding support of the Canadian Space Agency, is currently working on an Automated InSAR Monitoring of Arctic Permafrost (AIMAP) project to develop the capacity to efficiently and effectively monitor wide areas of Arctic terrain with an automated processing chain. The project is initially utilizing data from the Sentinel-1 constellation due to the widespread data availability. This is backed up by several footprints of RADARSAT-2 data over 3 locations where a deep temporal archive is available. The automated workflow, however, is being designed to extend to the Radarsat Constellation Mission (RCM) and NiSAR as those data becomes available. The AIMAP project has three primary objectives. First, 3vG has developed an automated snow detection routine utilizing both SAR data and optical Moderate Resolution Imaging Spectroradiometer (MODIS) Snow Cover products. These snow cover maps can be used to identify and remove areas with either reduced coherence from snow cover or areas with added phase signal due to the presence of snow. Secondly, 3vG has deployed a backprojection code (pers. comms. Prof. H. Zebker, Stanford) to enable raw sentinel data to be directly projected into map (ground) geometry, rather than requiring the data to first be processed in range-dopper coordinates. Such a routine allows the inclusion of Sentinel-1 EW data, which is widely available in the polar regions, but only provided by ESA in raw (L0) format. Finally, 3vG is developing strategies to classify and distinguish the seasonally cyclical component of InSAR signals from ongoing secular displacement. Such strategies are critical for characterizing the degradation of permafrost and identifying long term displacement trends that can prove damaging to existing and future infrastructure projects. We illustrate the power of InSAR for permafrost characterisation with results over five sites in the Canadian Arctic: Iqaluit, Tuktoyaktuk, Reindeer Station, Norman Wells, and Alert.
Authors: Pon, Andy Richard (1); Wang, Zheng (1); Mackenzie, David (1); van Heiningen, Jan Adam (1); Sun, Xinyao (1,2); Loader, Daryn James (1); Swart, Hollie Alayna Violet (1); Ghuman, Parwant Singh (1)Permafrost is broadly distributed in Qinghai-Tibet plateau, China. AS climate change heat up the earth, the degradation of permafrost becomes faster and consequently changes the distribution of vegetation and hydrological cycle. Sanjiangyuan, the study site, is located in the hinterland of Qinghai Tibet plateau. It is the fountainhead of yellow river, Yangzi river and Lancang river. Therefore, monitoring the degradation is extraordinary valuable to the environment of China. In this study, we use Persistent Scatterer InSAR (PSI) technique to detect ground settlement in summer periods around Elin lake and Zhaling lake, which are the main parts of Sanjiangyuan region. The subsidence was analyzed by processing 56 Sentinel-1 SAR images and 15 PALSAR-2 images spanning covering 2015-2019 time period. The interferometric processing was done using GAMMA software while the PS analysis was performed using Stanford Method for Persistent Scatterers (StaMPS). The results are then inverted to derive the corresponding active layer thickness over this region. Moreover, for the purpose of investigating the detailed influence of degradation on infrastructures we analyzed 3-m resolution TerraSAR-X images in StripMap mode from May to October 2015 to get the heterogeneous subsidence along the Gonghe-Yushu road. Results indicate mean subsidence rates along the Gonghe-Yushu road exceeding 8cm/yr. There were several pronounced non-uniform settlement section being detected which triggered the embankment collapse of the road.
Authors: Ma, Deying (1,2,4); Motagh, Mahdi (2,3); Liu, Guoxiang (4)Synthetic Aperture RADAR data have proven to be instrumental in providing crucial information about ice sheets like flow speed, grounding line position, as well as ice edge position. Up until the international Polar Year (IPY), only a few dedicated acquisition campaigns were carried out over ice sheets, allowing proof of concept as well as limited scientific studies. IPY fostered international collaboration in planning and execution of data acquisitions, which led to coordinated campaigns and the first Antarctica-wide coverage with interferometric data. Our work with most of the available data showed the value each and every mission provided to ice sheet research as well as to the generation of geoinformation products for ice sheets. Since then, Sentinel-1 has become a reliable resource, with a long-term commitment for ongoing acquisitions in Coastal Antarctica and most of Greenland. With RADARSAT-2 data collected less frequently, but on a continental-scale, C-band data are the dominant information source over ice sheets. X-band data collected using TerraSAR-/TanDEM-X as well as the Cosmo SkyMed constellation provide crucial information over key glaciers; the TanDEM-X coverage and resulting DEM provide crucial elevation information for the region in unparalleled precision. L-band interferometric data are also being collected in some key areas. During IPY, JAXA collected ALOS-PALSAR data in Antarctica on a continental scale and the data became an integral component of the first InSAR phase based ice velocity map of Antarctica. ALOS-2 PALSAR-2 has the advantage of a shorter revisit time (14 days), though large area coverage required multiple tracks with different incidence angles being collected for a specific orbit. We work within the framework of an ALOS-2 PI project and identified tracks covering a limited number of AOI’s to investigate the potential of ALOS-2 PALSAR-2 data for ice sheet research. JAXA kindly collected data from consecutive passes on multiple occasions. While the L-band phase sensitivity to motion is less than that of sensors with shorter wavelengths, the exceptional correlation of L-band data is beneficial for interferogram generation. We show benefits, challenges, and limitations of the high-resolution short repeat data and compare performance to available C-band data. Our findings are directly relevant to the ROSE-L mission which has glacier and ice sheet requirements as well as the upcoming NISAR mission, a dedicated science mission that also has ice sheet science requirements and will provide ongoing left-looking L-band SAR data collection over most of Antarctica. The 12 day revisit time of NISAR is similar to ALOS-2.
Authors: Scheuchl, Bernd (1); Rignot, Eric (1,2); Brancato, Virginia (1,2); Jeong, Seongsu (1); Milillo, Pietro (1); Ehrenfeucht, Shivani (1); Chen, Hanning (1); Mouginot, Jeremie (1,3)Three pass Differential Synthetic Aperture Radar Interferometry (3-pass DInSAR) is the radar remote sensing technique, estimates accurate movement/deformation by nullifying the error/uncertainty component of an external Digital Elevation Model (DEM). The main objective of this study is to measure the glacier movement without using an external DEM. The 3 pass DInSAR was started using at the end of the 20th century with European Remote Sensing (ERS-1/2) Satellite data for earthquake studies and Cryospheric applications. But later this technique was not being used for glacier movement estimation because of data unavailability of 3 successive side looking complex (SLC) images to reach the threshold coherence. The important assumption of the study is that the flow rate of a glacier is constant over the period of observation. Sentinel data builds again to use the 3-pass DInSAR technique to measure glacier movement. For the first time, we are using this technique with sentinel 1A/1B SAR data to generate a glacier velocity map of the Eidembreen glacier, located in the Svalbard region. Three SLC datasets (1st November 2017, 7th November 2017 and 13th November 2017) are selected to generate interferograms. Usually, the topographically induced phase values in 2-pass DInSAR can be eliminated from the interferometric phase with the help of an external DEM. However, in our case, we use the double-difference interferogram (DDI) instead of an external DEM. The DDI is obtained by subtracting the two interferograms (01 and 07 Nov – pair 1; 07 and 13 Nov – pair 2). Under the assumption that the glacier movement is constant over the observation period, i.e., 6 days, the DDI nullifies both 6 days velocity components leaving behind only the topography phase for 1st and 3rd passes of baseline. The differential phase interferogram can be subtracted from anyone of the unwrapped interferogram by multiplying with a corresponding baseline factor. If the baseline of two pairs is same or integer multiple then the interferogram can be directly subtracted from the DDI without using the baseline factor. Finally, this phase difference is converted into path difference to observe the movement of a glacier in LOS direction for 6 days’ time interval. This Eidembreen glacier LOS velocity is observed maximum at the central portion as 1.5 cm/day and reduces along the sides. Three pass DInSAR is more useful for glaciers movement application comparative to the other applications like land subsidence/uplift and earthquake studies. Most of the glaciers are located in undulated terrains and the accuracy of DEM is less over these regions. Hence, the accuracy will be further reduced if we use 2 pass DInSAR for glacier movement studies. But currently, this technique is only possible with Sentinel data (C-band data) to observe the glacier movement. Additionally, a double-difference interferogram also gives information to find the grounding line (i.e., boundary line between grounded and floating ice) and recently these studies are performing by the Sentinel data. In addition to the highly accurate estimation of glacier velocity using 3-pass DInSAR, we are able to identify certain key glacier features which cannot be observed through offset tracking based velocity maps.
Authors: Nela, Bala Raju; Singh, Gulab; Patil, AkshaySnow has high albedo, which is responsible for maintaining Earth’s temperature by reflecting most of the solar radiation into space. Snowmelt is a vital source of fresh water for a billion people across the globe. Hydropower generation and crop production rely on snow meltwater, mainly in the mountainous terrain. The snowmelt rate is also an indicator of climate change and local weather. Despite its importance, the quantitative assessment of the snow is not up to the mark, and we lack the information about the water stored across the mountain ranges on Earth. Snow water equivalent (SWE) represents the total water contained by the snowpack, and it is a product of snow depth (SD) and snow density. Here, we have shown the PolInSAR based algorithm for the retrieval of snow parameters (SD and SWE) and targeted the Indian Himalayan region for the implementation and validation. Two full polarization descending TerraSAR-X bistatic acquisitions with a temporal resolution of 11 days (2016/01/08 and 2016/01/19) is used for the implementation of the algorithm, and to show the temporal changes in the snowpack properties. The field measurements were conducted at the Dhundi observatory located in the Indian Himalaya (at 3000 m above. m. s. l), Manali, Himachal Pradesh. The observatory receives seasonal snowfall between December and March, with peak snowfall occurring in January. Near real-time snow parameters measurement was carried on 2016/01/08 and 2016/01/19 at 6:00 AM IST. The SAR data pre-processing has been applied to master (TerraSAR-X) and slave (TanDEM-X) using standard PolInSAR technique (i.e., calibration, bistatic correction, co-registration, and [T6] matrix generation). Subsequently, the PolInSAR coherences at HH, HV, and VV are generated. The strong positive correlation has been observed between the field measured SD and HV coherence with R2= 0.77 ( on 2016/01/08) and 0.81 ( on 2016/01/19). Subsequently, the linear regression equation has been developed with the field and satellite-derived data, which is used in the SD inversion. The full polarization SAR based algorithm is applied to TerraSAR-X data for snow density retrieval. Finally, the SWE is calculated by multiplying SD and snow density. The SD retrieval validation gives R2=0.61 with RMSE of 6.24 cm. Likewise, the SWE retrieval validation gives R2= 0.76 with RMSE= 7.75 cm. The histogram comparison of SD for the two acquisitions shows similar distribution with reduced SD on 2016/01/19 as expected due to no snowfall between the SAR acquisitions. The performance of the algorithm is promising over rugged terrain surfaces like in the Himalaya.
Authors: Patil, Akshay (1,2,3); Singh, Gulab (2,3); Nela, Bala Raju (2,3)(version with figures included can be found in the attached pdf.) Introduction In this study, we exploited the potential of TerraSAR-X (TSX) for wide area deformation monitoring and advanced post-analysis (Qin et al., 2019). Three frames covering Groningen were processed with the PS/DS technique. The high-resolution InSAR processing has been carried out as part of the subsidence data acquisition scope from the “Study and Data Acquisition Plan Induced Seismicity in Groningen - Winningsplan 2016 (NAM, 2016)”. The main purposes of the project are supporting building deformation analysis using high-resolution InSAR, and aiding geomechanical modeling of surface deformation as a result of compaction in the hydrocarbon reservoirs. Wide Area, Multitrack PS/DS Processing Overview In the 3 stacks of TSX full frames counting 550+ images in total from the past 6 years, the PS-InSAR analysis gives almost 10 million persistent scatterers (PS) and another 6.5 million distributed scatterers (DS) for which the time series were determined. The total area is around 3000 km2, converting to a 5500/km2 average point density. The process was done with the DePSI software (Sousa et al., 2015) originated from Delft University of Technology and maintained by SkyGeo. Validation with GNSS Network: Millimeter Level Accuracy The GNSS network has been used both for referencing and validation. A selected number of GNSS stations have been used, together with PS in a “stable area” (not affected by deep subsurface displacements) to correct for residual effects over a large spatial extent. Subsequently, all GNSS stations have been used in the validation with InSAR within the area of interest (AoI). 33 comparisons between the GNSS records and the InSAR results were made after the GNSS data were converted to line of sight (LoS) direction. For each network station within each TSX frame, the GNSS data was compared with the closest-by PS points. The validation shows an overall accuracy of time series (as represented by the RMSE) at millimeters level, and an overall accuracy of the velocity estimate (the difference between GPS measured velocity and PS measured velocity) within 1 mm/year. The validation is a solid proof of the high accuracy in the millions of point scatterers from the PSI result. Horizontal-Vertical Decomposition and Validation Horizontal-vertical decomposition was performed in areas where a descending frame overlaps with the ascending frame. Small horizontal movements are often observed in subsidence bowls, mainly at the border of the depleting fields. The decomposition can aid the estimation of subsurface parameters. The horizontal/vertical deformation was confirmed through the validation using GNSS stations with good accuracy. Separation of Shallow & Deep Compaction Shallow compaction is present in some areas in the AoI. Shallow compaction can occur as a result of the weight that is imposed on the layers, due to a decreasing groundwater level, or due to peat oxidation. On the other side, the gas extraction inside the AoI would cause a deep compaction. To understand whether the gas extraction activity has caused ground deformation, the shallow compaction needs to be separated from deep compaction. One of the logical assumptions is that houses which are well-founded on a (deep) stable layer are only affected by deep compaction. If this is true, then deep compaction can be measured using scatterers that come directly from houses that are well-founded. Other scatterers could be representative for the total compaction (shallow plus deep). The difference in terms of subsidence between the two is the shallow compaction. This is where an accurate geo-localization of scatterers comes into play. To separate scatterers that come from houses or ground level, the heights of the scatterers should be estimated with sufficient accuracy. By separating the high points from the low points, it can be seen that shallow compaction is present in some areas in our region of interest. For this dataset, the shallow compaction for the observed persistent scatterers in those areas are mostly limited to about 2 mm/year. In addition, the correlation between peat layers and shallow compaction is quite distinct once we overlay the shallow compaction layer to the soil type layer. In Groningen, the spatial distribution of shallow compaction is highly correlated with the distribution of the peat layer. An alternative way to separate shallow and deep compaction is by means of local deformation histogram analysis (see e.g. www.bodemdalingskaart.nl). Building an Automated Infrastructure Monitoring System with Accurate Geo-localization from TerraSAR-X One important aspect of this study is to monitor infrastructure stability. With the dense distribution of TSX PS/DS result, it is possible to build an automated building report that assesses the stability of every single building. In this study, using the building outlines from the BAG (Basisregistratie Adressen en Gebouwen, an open data set), it is possible to aggregate scatterers to each individual building and provide the statistics. Deformation time series are provided both from the high points on the building, as well as from the low point in the direct vicinity. It is worth mentioning that, the characteristics of the TerraSAR- X InSAR results, specifically the high point density, and quality of geolocation and deformation, allow for an analysis of the deformation of individual buildings. This is generally not possible or of significantly lower quality with imagery of lower resolution, such as imagery from Sentinel-1. Conclusions With an accurate 3-D geo-localization estimation to meter level, it is possible to separate high points (i.e., scatterers from well-founded buildings) from low points (i.e., roads, surface layers), and hence aid the discrimination between shallow and deep compaction. In combination with the high resolution, deformation can be assessed at the building level. The deformation time series from multiple tracks are used to decompose horizontal and vertical deformation. Reference [1] Qin, Y., Salzer, J., Maljaars, H., and Leezenberg, P. B.: High resolution InSAR in the Groningen area, project report for NAM, SkyGeo B.V., 2019. [2] NAM: Study and Data Acquisition Plan Induced Seismicity in Groningen, Update Post-Winningsplan 2016, https://www.nam.nl/algemeen/mediatheek-en-downloads/winningsplan-2016/_jcr_content/par/textimage_996696702.stream/1461000509432/99f0b51acc9c013bd2368d2dd1991aef8ecf40d9/study-and-data-acquisition-plan-for-induced-seismicity-for-winningsplan-2016-part1.pdf, Accessed on 2020-03-10. [3] Sousa, J.J., Hooper, A.J., Hanssen, R.F., & Bastos, L.C.: Comparative Study of Two Different PS-InSAR Approaches: DePSI Vs. Stamps., Fringe, 2015.
Authors: Qin, Yuxiao (1); Liu, Shizhuo (2); Maljaars, Hanno (1); Ketelaar, Gini (2); Leezenberg, Pieter Bas (1)Non-dispersive atmospheric phase delay of Synthetic Aperture Radar (SAR) signals depend on air pressure, temperature and moisture in the troposphere. The Interferometric SAR (InSAR) atmospheric signal can be decomposed into a turbulent (dynamic) and stratified (static) tropospheric delay, with the static part being strongly dependent on topography. To obtain reliable ground deformation estimates, the tropospheric delay must be modeled and removed from the InSAR time-series [1,2]. Using a Persistent-scatterer [3,4] InSAR processing chain, we generated deformation time-series for over 50 cities worldwide. Modeling the static atmosphere delay in urban areas requires estimating heights of buildings, which may not be present globally at sufficient resolution in digital elevation models (DEM). We observe buildings with significant static atmosphere delay contributions and a residual spatially-varying thermal-dilation signal. We correct the static atmosphere delay by demodulating a phase contribution per SAR scene that depends linearly on the topographic heights. To model the static atmosphere delay, we maximize the InSAR coherence weighted with the local topographic standard deviation. Two issues limiting this modeling approach are: 1. InSAR processing of cities with minimal topography, which can result in overfitting of the static atmosphere models, and 2. deformation related to topography, which has been observed previously in deformation modeling of volcanoes [5]. We observed these phenomena in several low-lying cities with differential and disparate long-term subsidence of buildings relative to the surrounding land features. This building subsidence depends on their loading and types of foundations, as well as ground-water extraction activities occurring on compressible deposits [6,7]. We improved our static atmosphere delay estimation in such areas by applying a probabilistic approach that penalizes long-term temporal trends in the per-scene coefficients. The resulting regularization is safe to use in general since it depends on the ratio of the uncertainties to the size of the leapfrog coefficients. The heat-island effect affects temperature and convective air flow, which depends on the urban and vegetative characteristics of the landscape [8]. To model the spatially varying static atmosphere delay, we present a patchwise coherence maximization algorithm with validation using results from neighboring areas of interest within the same InSAR footprint. We show that the height-dependent atmospheric phase can vary spatially between urban, sub-urban and rural landscapes [9]. References: [1] Zebker, H.; Rosen, P. A.; Hensley, S.; Atmospheric effects in interferometric synthetic aperture radar surface deformation and topographic maps, J. Geophys. Res.: Solid Earth, vol. 102, pp. 7547-7563, 1997 [2] Hanssen, R. F.; Radar Interferometry: Data Interpretation and Error Analysis, Kluwer Academic Publishers (2001), pp. 148 [3] Ferretti, A.; Prati, C.; Rocca, F.; Permanent Scatterers in SAR Interferometry, IEEE Trans. Geosci. Remote Sens., vol. 39(1), 2001 [4] Hooper, A.; Segall, P.; Zebker, H.; Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcán Alcedo, Galápagos, J. Geophys. Res. Solid Earth, vol. 112(B07407), 2007 [5] Beauducel, F.; Briole, P.; Froger, J. L.; Volcano-wide fringes in ERS synthetic aperture radar interferograms of Etna (1992-1998): Deformation or tropospheric effect, J. Geophys. Res.: Solid Earth, vol. 105, pp. 16391–16402, 2000 [6] Chaussard, E.; Amelung, F.; Abidin, H.; Hong, S.; Sinking cities in Indonesia: ALOS PALSAR detects rapid subsidence due to groundwater and gas extraction, Remote Sens. Environ., vol. 128, pp. 150–161, (2013) [7] Waltham, T.; Sinking Cities, Geology Today, vol 18(3), pp. 95–100 (2002) [8] Pichierri, M.; Bonafoni, S.; Biondi, R.; Satellite air temperature estimation for monitoring the canopy layer heat island of Milan. Remote Sens. Environ., vol. 127, pp. 130–138 (2012) [9] Khutorov, V. E.; Teptin, G. M.; Khutorova, O. G.; Heat island phenomena and it's influence on troposphere Mezo-scale structure measured by set of GPS-GLONASS receivers, 2017 Progress In Electromagnetics Research Symposium - Spring (PIERS), St. Petersburg, pp. 3547-3550 (2017)
Authors: van Heiningen, Jan Adam; Zimmer, Aaron; Pichierri, ManueleMost houses in The Netherlands’ soft soil are built on a foundation so they do not subside. Gas distribution pipelines do not have this solid foundation and thus do subside with the soil. This creates a focal point for strain to accumulate where the gas service pipeline enters the building. Stedin NV manages over 23,000 kilometers of gas distribution pipelines and 1.9 million connections into homes in The Netherlands. Historically, maintenance planning was performed methodologically similar to a brute force search. Frequency of maintenance was based on alerts from customers and the field workforce, and the area of maintenance is based on historic practices; scheduling maintenance per postal code area. This practice assumed both homogeneity of the subsurface and correct/proactive alerts of customers/repairmen. These two factors turn out to be idealized representations of the world. Through application of post-processed PS InSAR, we formed door-by-door subsidence risk estimates to ensure that pipelines with the highest risk are identified and maintained first. This pseudo-strain map has greatly increased deformation interpretability and thus enabled the transition from brute force search to targeted, predictive maintenance in areas of inhomogeneous subsidence and therefore high predicted strain on the subsurface gas network. The main effect of this optimization for Stedin is economical, saving €80.000 per week in unnecessary digging costs. In terms of InSAR processing, the scale of this application is very large, using 7 frames of TerraSAR-X Stripmap data producing 100M+ InSAR time series in mostly urban terrain. In a follow-on investigation, we characterize the stress distribution acting on a network of higher pressure gas distribution pipelines in the same area of interest as the lower pressure service lines. These pipelines are only indirectly connected to stationary objects such as buildings on a foundation. We deploy various types of spatial data science applications - post-InSAR processing, and multivariate taking into account additional variables such as pipeline material, construction type and age. The goal of this multivariate approach is to design a predictive maintenance process for both types of pipelines. In the first project, the predictive capability has proven commercially successful for 1.5M service lines. The second approach resulting in maintenance prediction for high pressure lines is currently being field-tested.
Authors: Visser, Ivo (2); Houtepen, Martijn (1); Leezenberg, Pieter Bas (1); Plain, Morgan (1)The Krechba field is one of several gas fields located in the Algerian Sahara desert, and was set in operation in August 2004 as part of a joint venture with BP, Sonatrach and StatoilHydro. The natural gas in the fields contains up to 10% CO2, which has to be reduced to 0.3% before the gas is sold, resulting in the production of around 1 million tonnes/year CO2. Rather than vent the CO2 to the atmosphere, it is re-injected into the water leg of the Krechba Carboniferous Sandstone gas producing reservoir (20 m thick) via three horizontal wells at a depth of around 1,900metres. CO2 injection started in August 2004 and to date nearly 2.5 million tonnes of CO2 have been injected, amounting to approximately 25% of the gas extracted from the Krechba field over the same period. A number of key technologies to monitor the injection, and the subsurface movement and storage of CO2 have been, and will continue to be deployed to provide long term assurance of sequestration. Time lapse satellite images (using PSInSARTM Technology) which measure ground deformation to assess the movement of CO2 in the subsurface have proven to be much more successful than initially thought, despite the depth of injection and the low void age replacement rate (25%). Satellite images collected since start of injection show clear increases in ground elevation of up to 30mm around the three injectors while subsidence is also apparent in the area of maximum gas production. The images have also confirmed the CO2 is moving in the direction of preferred fracture orientation at reservoir level. Recent down whole pressure measurements in one of the injectors also indicates that the CO2 is being contained within the injection horizon. The In Salah project continues to evaluate different monitoring technologies at this industrial-scale sequestration project. In this context, the acquisition of a 3D seismic survey was integrated in addition to tiltmeter and micro-seismic surveys and observation wells. A key part of the programme aims to work at the geomechanics level which will help to tie the surface observations from satellite to the subsurface movement of CO2 The presentation to the Fringe2020 will focus mainly on the In Salah Gas Project, the structure of the monitoring program, the time evolution satellite imagery acquisition to show more precise details and finally some obtained results. References : 1 Allan Mathieson, Iain Wright1 David Roberts, Philip Ringrose (BP Alternative Energy, Chertsey Road, Sunbury, Middlesex TW16 7LN UK & StatoilHydro, Technology and New Energy, Trondheim, Norway) 2 Vasco D., Ferretti A., Novali F., Reservoir monitoring and characterization using satellite geodetic data: Interferometric synthetic radar observations from the Krechba field, Algeria, LBNL 308-E, Geophysics 3 Onuma, T. and Ohkawa, S., Detection of Surface Deformation related with CO2 Injection by DInSAR at In Salah, Algeria. Proceedings of the 9th International Conference on Greenhouse Gas Control Technologies, 16 – 20. www.sciencedirect.com
Authors: Abed, Mohamed (1); Tayeb, Smain (2); Khaled, Boussedjra (1)ATLAS, SIXENSE’s interferometric processing chain, has been developed around the core software GAMMA to successfully detect and monitor ground motions such as subsidence, heave, building stability and landslides. It has been applied in geotechnical and structural monitoring projects linked to urban construction activities, with particular focus on tunnelling monitoring. Taking advantage of SIXENSE’s experience in geotechnical and automatic surveying, it has been used for measuring vertical ground and structure movements, mainly related to volume loss. Volume loss control is one of the main objectives in big tunnelling and excavation works on densely urbanized areas, where InSAR advantages complement more conventional approaches. The construction and civil engineering world benefits from InSAR technology in different aspects. Ground movements caused by major construction projects, such as tunnelling, have the potential of causing damage to overlying structures. This is even more probable when the project is carried out under or around heavily urbanized areas. In a construction project, both ground and structural behaviours are needed to be considered before the construction works start. The way of controlling it is through monitoring techniques, and it should be carried out while the construction is being done, and after it is finished until ground settlement due to the construction activity has faced out. InSAR technique can overcome the limitations of traditional monitoring techniques by means of providing historical data measurements, wide area measurements without losing resolution over single structures, and providing measurements regularly after the end of the project. In this presentation, ATLAS will be briefly detailed and different application cases of the ATLAS monitoring in different scenarios will be presented in the field of civil engineering: from ATLAS extended areas results to a local asset focused solution to provide ready-to-use friendly-user information in BEYOND Satellite web platform. Various site examples will show how ATLAS has been used in large construction projects around the world to monitor different tunnelling and construction work phases: (i) Access shaft excavation, including dewatering. Dewatering has shown on numerous occasions an impact over long distances, which are covered by InSAR measurements due to the large size of satellite images, which also allows overcoming the problem of loss of references for the ground instruments, (ii) Tunnelling phase, when satellite measurements are used to check any widening of the settlement inside and outside the planned zone of interest (ZOI), as a back-up and verification of the ground instruments, and (iii) Upon termination of the works, when satellite measurements provide a technically and financially efficient way of controlling long term stabilization of movements.
Authors: Devanthery, Nuria; Camafort, Miquel; Ibarrola-Subiza, NereaPersistent Scatterer Interferometry (PSI) is a well-established technique for monitoring millimeter deformation of the Earth’s surface. The availability of free and open SAR data with a repeat cycle of 6-12 days from the Copernicus mission Sentinel-1, allows PSI to be used complementary to traditional surveying techniques. Although the data resolution may not allow an exact determination of the geolocation of the occurring deformation, these deformations can be analyzed with auxiliary data and often indicate the existence of local geophysical processes or correlate with various human activities like mining or groundwater pumping. Traditionally, the PSI-technique is applied for monitoring urban areas or areas containing large man-made objects. Therefore, dams have also been an object of investigation for some years. However, embankment dams or gravity dams have not been included into the observation with PSI yet. This present study investigates deformation dynamics at the Möhne Reservoir in North Rhine-Westphalia, Germany. The PSI-Technique is applied to evaluate Sentinel-1 data from 01/2015 to 11/2020 within the framework of dam stability monitoring and compare the results to in-situ data, such as trigonometric measurements as well as plumb measurements. The applicability and accuracy of the PSI-technique regarding deformation monitoring of dams is analyzed. Furthermore, the effect of the number of observations is evaluated. Hence, one data stack including 37 scenes, one data stack including 67 scenes and one data stack including all 167 scenes during the observation period were created and co-registered to the same reference scene. First results show that movements of the Möhne gravity dam follow a seasonal pattern. Deformations generally reach a maximum of up to 4 mm in horizontal direction in March and April. In the following months, the computed deformations decline until they reach a minimum during August and September, with values usually going down to -4 mm and occasionally even as low as -6 mm. For most observed years, this pattern is similar. However, in 2017 maximum values reached only 1,5 mm in April, which can be related to a comparatively low water level in the reservoir in this drought year. The seasonal pattern of the satellite derived deformations as well as its maxima and minima are in accordance with land surveying activities at this site. The number of Sentinel-1 observations has a decisive influence on the resulting number of computed PS-points. A higher number of PS-points was computed using the data stacks containing 37 Sentinel-1 scenes and 67 Sentinel-1 scenes. Although the processing of all 167 scenes delivers the highest detail in the deformation results, only a few PS-points were identified using this data stack. Additionally, the usage of different data stacks does not only affect the number of PS-points, but also their location. Combining all three data stacks, the highest spatial coverage with PS-points is achieved. Further work will include the analysis of the accuracy of the measured deformations considering the different number of scenes.
Authors: Jänichen, Jannik (1); Dubois, Clémence (1); Baade, Jussi (2); Bettzieche, Volker (3); Schmullius, Christiane (1)During the last two centuries, the development of hydropower and large basin management has increased the construction of dams both in size and number. Nowadays, many of those dams continue to be operational, which makes dam safety and structural monitoring a priority to prevent any potential failure that could cause a major risk. In this work, a case study of a dam monitoring using Sentinel-1 images will be presented. The current study has been carried out in the frame of a large-scale InSAR ground deformation monitoring project covering the full extension of the reservoir. The geological characteristics of the area have already caused slope stability problems in the past, affecting an area near the dam structure. The study has been carried out using the ATLAS interferometric processing chain, implemented by SIXENSE and developed around the core software GAMMA. The dam’s seasonal movements observed in the deformation results of the InSAR historical study have been analyzed and compared with in-situ measurements. A comparison of results has been performed using almost 3 years of continuous high-precision dam monitoring data. The ground instrumentation data consists of several pendulums which monitor horizontal movements in the dam, and a levelling network designed to measure differences in elevation on each gallery inside the structure. Due to differences in the data acquisition geometries, a procedure of transforming data between reference systems has been defined. This transformation allows a direct comparison of InSAR and ground instrumentation data in a common reference system. Finally, a list of considerations that should be taken into account in the monitoring of dams using satellite InSAR technology is provided. For many years, remote sensing techniques have been a useful tool to provide non-invasive information of the land surface. The results obtained on the comparison of InSAR measurements and ground instrumentation data show the effectiveness of InSAR satellite surveying for dam monitoring. InSAR technique brings a cost-effective monitoring system, covering wide areas and providing millimetric precision measurements over the ground and infrastructures. In addition, the availability of archive satellite imagery allows the monitoring of ground deformations in the past. All these features make InSAR a valuable tool for dam monitoring.
Authors: Ibarrola-Subiza, Nerea; Camafort, Miquel; Devanthéry, NúriaInfrastructure monitoring is undergoing a deep process of transformation in recent years. DInSAR technology is leading a new style in monitoring infrastructures and entire cities at unbeatable costs. Open data policies, DInSAR processing automation, AI integration and cloud computing have made this paradigm shift possible. In addition, the number of satellites and the resolution of the images will increase substantially in the upcoming years. In the case of dams, the supervision is regulated by national authorities, due to the high impact of their potential failure. The main goal of the public supervision is to reach a uniform high level of dam safety, and thereby to ensure that these structures are not posing a threat to life, property or the environment. The applicability of satellite radar interferometry for monitoring embankment dams has been tested in several research projects. Thus, DInSAR can be considered a consolidated technique which effectivity for deformation studies has been proved in many cases. With actual SAR data availability, the technique is demonstrated to be of high value at a very low cost as compared with classical geodetic monitoring techniques. In case of other geotechnical or geodetic data are available, they can be successfully complemented. A promising field of application is the early warning of potential failures of critical infrastructures. On May 1, 2020, a large area of the Sardoba reservoir dam in the Sirdaryo region of Uzbekistan collapsed, flooding more than 35,000 hectares in Uzbekistan and Kazakhstan. Six people died and at least 111,000 were evacuated from the Syr Darya river basin. In the present work, deformations of the dam walls prior to failure were assessed using DInSAR (Differential Interferometry Synthetic Aperture Radar) technology. Specifically, SNAP-Stamps PSI analysis was applied to 27 descending images and 31 ascending Sentinel-1A images (May 2019- April 2020). The measurement of the dam movements from the two satellite acquisition geometries allowed the decomposition of movements into vertical (subsidence and uplift) and horizontal on the East / West axis. Results show a consistent pattern of deformation in the East direction of the collapsed section of the dam wall prior to its failure. This opens the way to future early-warning systems implemented in these large critical infrastructures, as the Sardoba dam.
Authors: Sancho, Candela (1); Sánchez, Jaime (1); García-Lanchares, Carlos (1,2); García-Sánchez, Adrián Jesús (1); Fernández-Landa, Alfredo (1); Martínez-Marín, Rubén (1,2); Marchamalo, Miguel (1,2)Many coastal cities are subject to severe natural and anthropogenic disasters such as flooding, storm surges and land subsidence. St. Petersburg is a coastal city with mostly flat topography and elevation ranges between around 1–2 m above the sea level. The city is situated on the shores of the Neva River, at the head of the Gulf of Finland in the Baltic Sea. Therefore, flood risks are a major concern and monitoring surface deformation in St. Petersburg is crucial for planning and reducing disaster risks. Interferometric Synthetic Aperture Radar (InSAR) measurement has the potential to enhance the information available from conventional surveying techniques and increases spatial and temporal coverage at low cost, with potential for ongoing monitoring. InSAR is the most suitable technique for detecting and mapping subsidence of various types of urban areas. The technique has actively developed in using remote sensing data and there are several methods of interferometric processing. These include classical differential interferometry (DInSAR) and serial interferometric methods. The rapid increase of the open radar data amount after the launch of Sentinel-1 satellites suggests the relevance of the research. Its aim was to develop methods for detecting displacements of the earth's surface in cities to secure safe life of a large number of people. The research presents the results of a new monitoring technique based on the method of radar satellite interferometry using open data from the Sentinel-1 radar satellite. The method of multitemporal DInSAR was chosen from the existing methods of satellite radar interferometry for determining the subsidence of the earth's surface in urban areas. This method was applied to St. Petersburg and was based on the Sentenel-1 data received from June 2018 to May 2019. Specified requirements were produced to create subsidence maps using 41 images downloaded for the selected monitoring period. Radar image processing was implemented on SNAP, open source software. As a result, 40 maps of vertical displacements of St. Petersburg were generated. Based on the QGIS geographic information software an analysis and visualization of the results of interferometric processing were performed. Thus, it was possible to identify the systematic subsidence of the soil due to the breakthrough of underground utilities in the Frunze district of St. Petersburg, which caused the accident. An insignificant decrease in soil subsidence was observed over the course of a month. The maximum value recorded by satellite data was 25 cm. The emergency situation that occurred in the time interval corresponding to the monitoring confirms the subsidence detected as a result of interferometric processing. The proposed procedure allows continuous monitoring of land subsidence based on open SAR data and open software. In the future, this method can be used to monitor displacements of the surface and prevent emergencies in urban areas.
Authors: Shirshova, Vera; Baldina, Elena A.Ground water is one of the significant natural resources on the earth. It is utilized as a primary source of fresh water both for human consumption and other uses. With an increase in urban population and infrastructural projects, the requirements of water has been tremendously increased in the urban areas. All these factors are putting pressure on the ground-water levels leading to more and more utilization of the ground waters in the urban areas. This ground water level depletion is highly correlated with the corresponding land subsidence in the urban areas. The deviation in the ground water level results in corresponding deformation in the earth’s surface, hence a large scale ground water assessment can be easily done by monitoring the land subsidence. Time series SAR Interferometry has shown its efficiency in the land deformation monitoring. This study focuses on the utilization of the PsInSAR (Persistent Scatterer Interferometric SAR) technique. Time Series Interferometric Sentinel-1 SAR datasets acquired in the TOPS mode were utilized for this study. The study area of this work is taken as Lucknow, Uttar-Pradesh, India. The corresponding ground water level datasets of both the study areas were taken from the Central Ground water Board, Goverment of India(CGWB) for both pre and post monsoon seasons. The results showed high correlation between ground water level variation and the retrieved deformation. The study was successful in analyzing the performance of the PsInSAR for both urban study area.
Authors: Awasthi, Shubham; Jain, KamalThe storage cavern field at Epe has been brined out of a salt deposit belonging to the lower Rhine salt flat, which extends under the surface of the North German lowlands and part of the Netherlands. The currently 114 caverns are used for brine production and for storage of natural gas, helium and crude oil by in total 8 companies which follow independent operating strategies. Cavern convergence and operational pressure changes cause linear and cyclic surface displacements, respectively. The linear part is monitored regularly by levelling, while we exploit the potential of SAR interferometry for monitoring nonlinear movements over the storage site in our study. Distributed Scatterers (DS) and Persistent Scatterers (PS) were combined for the analysis. The approach comprises DS pre-processing inspired by SqueeSAR followed by a modified version of StaMPS v3.3b, that allows joint processing of pre-filtered DS and PS and supports unwrapping with a phase model composed of linear trend, pressure response and a seasonal component. At this stage of processing, the pressure response is just an unscaled model for the pressure variations at the top of the salt layer. The model was derived using operational data which is indicative for pressure changes in the interior of the caverns. We assume, that cavern pressure propagates through the salt layer according to the theory for visco-elastic behavior of a Kelvin-Voigt body. The model captures well the observed delayed geomechanical response to pressure changes in the caverns. This behavior is different from porous storage media, where the response can be described as elastic and the temporal evolution of displacement is highly correlated with the temporal evolution of reservoir pressure. The seasonal component was introduced to support processing of DS found over a fen, where ground water level changes caused pronounced displacements. The model was extracted from a preliminary InSAR analysis in order to describe well its proper shape. Vertical and East-West movements have been determined based on Sentinel-1 data from ascending and descending orbit. The transformation has been performed separately for the model parameters and the unmodeled residual deformation. Linear subsidence and pressure response are caused by deformation of the caverns situated in a depth of more than 1000m and possess distinct horizontal displacement, while the seasonal deformation over the fen, caused by ground water level changes directly beneath the surface, displays merely vertical displacements. Validation of the findings with ground truth from levelling and groundwater level measurements shows a good agreement. In addition, simple geophysical modeling has been used to support InSAR processing and helps to interpret the observations. It assumes that caverns act as spherical pressure or volume sources embedded in an elastic halfspace, and that the spatial pattern of surface movements results from the superposition of the corresponding displacements. Each cavern is thought to be surrounded by a spherical salt mantle that accounts for the visco-elastic pressure delay. From the outer surface of the salt mantle, pressure is transferred elastically to the surface. This multi-source model is used to describe either the parameters of the linear component of the displacement model or of the pressure response. Linear subsidence is estimated assuming convergence of all caverns. For the pressure response only the gas filled caverns are considered. Compared to a purely geometric orbit combination, the geophysical approach better reproduces the linear deformation measured by levelling, but overestimates the horizontal movement.
Authors: Even, Markus (1); Westerhaus, Malte (1); Simon, Verena (2)Persistent Scatterer Interferometry contains a wealth of information distilled into a handful of encompassing parameters. There is great academic possibility in the right combination & filtering of these parameters, but this further tangles the already complex subject of PS Interferometry. This complexity forms a gap between industry acceptance and implementation and academic ability, even though these new information streams can be of great value.Bridging this gap between academic ability and industry acceptance will produce new economic and social value for companies and communities in diverse fields. It also increases the uptake and acceptance in industry for the value that PS InSAR can add. In this paper we describe our method of creating an easy-to-grasp and intuitive method of visual interaction with these data streams, available to industry.This submission will showcase (2) fields of application where accessibility and interpretability of PS InSAR were successfully improved through case-specific post-processing tools. The first field of industry is housing corporations. An often requested product by housing corporations is a risk map, with different indicators for potentially problematic buildings.While basic PS InSAR can highlight areas that are experiencing unexpected deformation, it does not create a uniform and comprehensive list of risk indicators per building. By filtering, combining and then categorising the data, relevant risk indicators can be created such as the following for building stability:* 25% fastest vs 25% slowest points associated with a building as an indicator for slant instability.* Change in standard deviation of these points over the years. * Change in deformation rates over specific time periods.As a final step, adding case-specific knowledge about the area enables a final product of an interactive, visually intuitive risk map that can also be used for further processing.The second example is underground infrastructure monitoring in built-up areas.First, a subset is made on areas without foundation. These PS points are then interpolated to a plane. Using this plane, lines can be generated to show areas experiencing high internal ground forces. This directional approach allows for a more intuitive assessment of underground infrastructure that is potentially experiencing pressures above their design limits. The transformation of these deformation datasets, which would typically be presented as PS points in a viewer (and require knowledge to interpret correctly), to a post-processing result such as those discussed above enabled end-users to intuitively understand the data provided.
Authors: van der Sleen, Vincent; Kunnen, Jeroen; Plain, MorganNon‐destructive testing (NDT) methods such as Ground Penetrating Radar (GPR), 3D Laser Scanner, accelerometer instruments amongst many others have been used to assess and monitor transport infrastructures in the past few years [1]. However, scientific literature has proven that stand‐alone or integrated use of ground‐based techniques may not represent a definitive solution to some major structural issues, such as scour and differential settlements [2], as these require continuous monitoring and data collection on long‐term basis. To that extent, the use of satellite data‐based Persistent Scatterer Interferometry (PSI) has demonstrated to be effective in measuring displacements of transport infrastructure at the network level [3] [4] and natural terrain [5] over longer periods of observation. This study reports recent results obtained using an innovative integrated method using the Ground Penetrating Radar (GPR) and the Persistent Scatterers Interferometry (PSI) techniques, for the monitoring of railway infrastructures. To this purpose, an experimental campaign was conducted over a railway located in Puglia, Southern Italy.On one hand, GPR was used to obtain structural details of the subsurface (thickness of the ballasted layer, presence of clay and humidity spots, position of the sleepers) of the infrastructure and to identify potential construction‐related issues. Parallel to this, PSI analyses were addressed processing C‐Band Sentinel‐1A SAR products provided by ESA (European Space Agency), and X‐Band COSMO‐Skymed products provided by ASI (Italian Space Agency), to monitor and detect structural displacements with a millimeter accuracy. Outcomes of this investigation outlined the presence of subsidence at both the areas of the transition of the rail‐abutment transition area in a railway truss‐bridge and have proven the proposed integrated approach to be viable to form the basis for achieving a more comprehensive integrated health monitoring mechanism of the structural integrity of railway inspected. Keywords – Persistent Scatterers Interferometry (PSI), Ground Penetrating Radar (GPR), Integrated Health Monitoring, Railway monitoring, Transport Infrastructure Maintenance Acknowledgments: The authors want to acknowledge the Italian Space Agency (ASI) for providing the COSMO-SkyMed Products® (©ASI, 2016-2018). The Sentinel 1A products are provided by ESA (European Space Agency) under the license to use. This research is supported by the Italian Ministry of Education, University and Research (MIUR) under the National Project “EXTRA TN”, PRIN 2017, Prot. 20179BP4SM. In addition, the authors acknowledge funding from the MIUR, in the frame of the “Departments of Excellence Initiative 2018-2022”, attributed to the Department of Engineering of Roma Tre University. References [1] Solla, M., Lorenzo, H., Rial, F.I., Novo, A. (2011). GPR evaluation of the Roman masonry arch bridge of Lugo (Spain), NDT&Int., 44, 8-12. [2] Selvakumaran, S., Plank, S., Geiß, C., Rossi, C., Middleton, C. (2018). Remote monitoring to predict bridge scour failure using Interferometric Synthetic Aperture Radar (InSAR) stacking techniques, Int. J. .Appl. Earth Obs. and Geoinf. 73, 463-470. [3] Tosti, F., Gagliardi, V., D'Amico, F. and Alani, A.M., Transport infrastructure monitoring by data fusion of GPR and SAR imagery information. TIS 2019 International Conference of Rome, 23-24 September 2019. [4] Bianchini Ciampoli, L., Gagliardi, V., Clementini, C. et al. (2019). Transport Infrastructure Monitoring by InSAR and GPR Data Fusion. Surv Geophys. https://doi.org/10.1007/s10712-019-09563-7 [5] Ferretti A, Prati C, Rocca F (2001) Permanent scatters in SAR interferometry. IEEE Trans Geosci Remote Sens 39(1):8–20. https ://doi.org/10.1109/36.89866 1
Authors: Gagliardi, Valerio; Bianchini Ciampoli, Luca; D'Amico, Fabrizio; Benedetto, AndreaBuilt heritage, represented by historical architectures, can be the symbol of nation spirits and the witness of mankind civilizations. Jointly impacted by natural degradation and anthropogenic activities, the sustainable conservation of heritages properties is facing challenges; in particular, for those with large-scale coverage hosted by a complex-diverse environment. As the pilot investigations of the China-Greek bilateral project “SpaCeborne SAR Interferometry as a Noninvasive tool to assess the vulnerability over Cultural hEritage sites, SCIENCE”, in this study, the performance and capability (synoptic observation and quantitative motion inversion) of spaceborne multi-temporal SAR interferometry (MT-InSAR) approaches, including enhanced SBAS and D-TomoSAR, were exploited and assessed in the displacement anomaly detection and risking mapping in the pilot study of Ming Great Wall and city walls. Sentinel-1 A/B (IW mode) SAR data, acquired in the period of 2015-2018, were used for the displacement monitoring along two typical cultural corridors of Ming Great Wall distributed in East and West regions of China, using the enhanced SBAS approach. The correlation analysis with multi-source spatial data (e.g. natural factors of hydrology, wind speed, unstable slope and human activities of mining and relevant construction) shown that: 1) the rammed earth Ming Great Wall in the west region of China (Qingtongxia, Ningxia), being exposed to the alluvial fan-zone of arid desert mountains, its stability can be consequently impacted by the surface runoff erosion and strong wind erosion. A positive correlation between the annual precipitation, wind speed and the InSAR derived motion fields in 2015-2016 and 2017-2018 has been validated. 2) The stability of the cultural corridor along the stone-built Great Wall in the east region of China (Zhangjiakou, Hebei) was impacted by mining and natural unstable slopes. The tropospheric turbulences due to vertical stratification have been minimized, since the corridor is extended along altitudes with high variance. Temporal analysis shown that the proportion of unstable areas for the period of 2015-2016 was 20% versus 15% for the period 2017-2018. For the third study area, the Ming Dynasty City Walls in Nanjing City, China, the extended D-TomoSAR approach (modeling components of linear, nonlinear displacements, height and thermal dilation) was applied, exploiting CSK acquisitions of the period 2015-2017. The results were compared with previous study using TSX acquisitions of the period 2013-2015. Assuming that the displacement is totally vertical, the 180 m wall section the linear uplift rate measured in the period 2013-2015 was no longer existent in the period 2015-2017, implying the provisional impact of anthropogenic effects (e.g. demolition and construction activities) to the stability of wall monuments. The presented case studies demonstrate the crucial role of EO data and relevant information communication technologies (e.g. MT-InSAR for satellite radar images presented in this study) in the preservation of cultural heritage monuments, as boosted by the plurality of the data and the methodologies advocates for the systemization of such activities.
Authors: Chen, Fulong (1,2); Zhou, Wei (1,2); Parcharidis, Issaak (3); Elias, Panagiotis (4)Located on the southwestern shore of Lake Como in northern Italy, the historical centre of Como municipality is built on a thick sequence of Pleistocene glacio-lacustrine, palustrine and alluvial sediments. The city is naturally prone to subside due to the presence of highly compressible unconsolidated Late Holocene silty sediments, and its vicinity to the lake that strongly influences the variations of the groundwater level. Anthropogenic activities (i.e. land reclamation, urban sprawl, deep aquifer exploitation between 1950 and 1975, and the construction of antiflooding facilities along the lakeshore in 2008 - 2010) accelerated the sediment consolidation and altered the groundwater regime, thus amplifying the subsiding rate of the urban area up to 20 mm/year along the coast and causing severe damage to buildings of the city centre. The continuous variation of the groundwater table level due to both natural and anthropogenic factors can result in the occurrence of absolute and/or differential settlements at building foundation depth. When the foundation cannot accommodate differential settlements, damage typically affect the superstructure. The damage severity level depends on either intrinsic building parameters (e.g. construction and geometric characteristics) or external environmental factors (e.g. subsidence intensity, hydro-mechanical soil properties). Interferometric Synthetic Aperture Radar (InSAR) techniques are largely used to profitably investigate and monitor subsiding processes over large areas with millimetric accuracy. In this work we present the first damage analysis of the historical buildings in Como downtown exposed to subsidence using X-band InSAR measurements in combination with damage data collected on site and information on subsoil hydrogeological and stratigraphic setting. A stack of 167 Single Look Complex (SLC) images acquired by the Italian Cosmo-SkyMed (CSK) mission along the descending orbit from 2010 to 2019 was used. The images were processed via multi-baseline Interferometric Point Target Analysis (IPTA) approach that provided a set of georeferenced Point Targets (PT) with the corresponding ground surface displacement time series. These data were used to derive the differential displacements experienced by each building that might have caused damage to the superstructure. During recent field investigations in the historical centre, we assessed the damage severity levels of 600 buildings. The InSAR-derived differential displacements, here used as indicator of the subsidence intensity, were then associated to the building damage severity levels. Based on the collected data, the cause-effect relationships between predisposing (i.e. hydrogeological and stratigraphic features) and triggering factors (i.e. differential displacement) of building damage were investigated. The results reveal the dependency of damage severity on InSAR-derived differential settlements and confirmed the predisposing role played by the local variations of hydrogeological and stratigraphic setting of highly compressible sediments present in the subsoil of Como urban area. These preliminary results pave the way for further building vulnerability analyses that, encompassing a broader range of predisposing/triggering factors and damage records, could be framed within a more efficient management of the inestimable architectural heritage exposed to subsidence risk in Como downtown
Authors: Nappo, Nicoletta (1); Peduto, Dario (2); Polcari, Marco (3); Livio, Franz (1); Ferrario, Maria Francesca (1); Comerci, Valerio (4); Stramondo, Salvatore (3); Michetti, Alessandro Maria (1)The Landers earthquake was the first application of the synthetic aperture radar interferometry (InSAR) technology that mapped a geo-hazard-induced displacement field. Two SAR images from before and after the earthquake were used to calculate the differential phase, which provides information concerning the deformation occurring between acquisitions [1,2]. In the past decades, this technology evolved into more advanced InSAR techniques such as persistent scatterer interferometry (PSI) [3,4], small baseline subset [5], and SqueeSAR™ [6]. In contrast to conventional InSAR techniques, advanced InSAR techniques employ a time series of SAR images instead of only two images. The time series is used to reduce the impact of noise and atmospheric disturbances on the results. Additionally, various studies have showcased the capability of these techniques to record ground deformation with an accuracy down to a millimeter [7,8,9]. Also, their diverse spectrum of applications was presented over the past two decades [10,11,12,13,14]. Most of these studies focused on analyzing the deformation of one specific observation period. However, more recent studies have exploited the continuous stream of new SAR images provided by SAR missions such as the Sentinel-1 mission to update the deformation maps of unstable areas regularly. The intent is to integrate PSI as a fixed component into a permanently operating monitoring system. The information provided by a PSI analysis can be used to automatically identify areas of active deformation, characterize their behavior, and create alerts if any changes in the behavior occur [15,16,17,18,19,20]. The observation target of this study is the Peiros-Parapeiros dam, which is located on the north-western tip of the Peloponnese Peninsula south of the city of Patras. The dam is an embankment dam securing the freshwater supply for the area. The water reservoir is fed by the Peiros and the Parapeiros river [22]. The construction of the dam was finished in early 2019, and the filling started in September 2019. In general, dams can be subject to short-term (i.e., daily, weekly or monthly) as well as long-term (i.e., years or even decades) deformation processes [23]. Some of these deformations, for example, the settlement of the dam's building materials, are expected [26], others are not. Subsidence due to a massive increase in pore pressure, hydraulic fracturing, or soil liquefaction can lead to additional deformation. Also, a varying water level of the reservoir could cause a fluctuating subsidence of the dam [23]. The first filling of a dam, in particular, is a critical time period. The construction is experiencing the full pressure of the water for the first time. Sliding, changes in the dam's geometry, and additional settlement may occur [24]. Monitoring these deformation processes is crucial to ensure the dam's health and functionality as well as the safety of the people living close by [23, 25]. Hence, the challenge in particular is to observe deformation processes on different time scales. First, we want to observe the settlement of the construction, which is supposed to take about three years and started in spring 2019. The second deformation process the dam is experiencing is due to the rising water level during the filling process, which is expected to take three years starting in September 2019. Thirdly, long-term deformations such as a seasonal deformation are due to rising and sinking water levels. Fourthly, since the dam was constructed in a seismically active area, a tectonic shift might also be observed. Some of these phenomena might overlap, others might need years to be observed. This study aims to find a strategy to first observe all these deformation processes and, secondly, deliver deformation maps in a timely fashion so that the information can be used to monitor the dam's state of health. In a previous study, we tested three different strategies to process a continuously growing data set with the PSI algorithm Standford Method for Persistent Scatterer (StaMPS). The first strategy was to regularly update the data set with the newly available images and reprocess the entire data set. For the second and third strategy we decided only to process consecutive subsets of the data set. The subsets consist of a constant number of interferograms. In the case of the second strategy, these subsets had no overlap, and the master scene was chosen individually for each subset. For the third strategy, however, the subsets did overlap, and the master scene was invariable throughout the entire process. We concluded, that the third strategy suits our purposes the most. It can be best described as a sliding window approach. The number of interferograms used for the PSI analysis is invariable. With each update, the oldest interferogram of the data set is exchanged for a new one. The master scene is also invariable as long as the image is still part of the data set. If this is not the case anymore, a new master needs to be selected [27]. With these strategies, most of the interferograms of the subset would be the same as were used in the previous one. This begs the question if it is possible to reuse some of the already derived results to reduce the processing time. In this paper, we are going to examine possible adjustments to the StaMPS algorithm to achieve this. The StaMPS algorithm consists of roughly four steps: persistent scatterer candidate (PSC) identification, persistent scatterer (PS) identification, phase unwrapping, and correcting the interferometric phase. As a first starting point, we will adjust the PSC identification process. This step is done using the amplitude dispersion. The calculation of the amplitude dispersion would be modified so that in the first run, the amplitude dispersion value of each pixel is saved for the next run. Later, the value is only adjusted for the exchanged interferograms and not entirely recalculated in the next run. Adjustments to the PS identification step are based on the idea that since this step is an iterative process, we could shorten the processing time using the old result as a new starting point. Thus, convergence might be reached faster. We plan to compare the results derived using StaMPS with the ones derived with our adjustments to StaMPS and evaluate the time savings being made. Literature [1] Massonnet, D., et al. (1993). "The displacement field of the Landers earthquake mapped by radar interferometry." Nature, vol. 364, no. 6433, pp. 138-142. [2] Zebker, H. A., et al. (1994). "On the derivation of coseismic displacement fields using differential radar interferometry: The Landers earthquake." Journal of Geophysical Research: Solid Earth, vol. 99, no. B10, pp. 19617-19634. [3] Hooper, A., et al. (2004). "A new method for measuring deformation on volcanoes and other natural terrains using InSAR persistent scatterers." Geophysical Research Letters, vol. 31, no. 23, L23611. [4] Ferretti, A., C. Prati, F. Rocca (2001). "Permanent scatterers in SAR interferometry." IEEE Transactions on Geoscience and Remote Sensing, vol. 39, no. 1, pp. 8-20. [5] Berardino, P., et al. (2002). "A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms." IEEE Transactions on Geoscience and Remote Sensing, vol. 40, no. 11, pp. 2375-2383. [6] Ferretti, A., et al. (2011). "A new algorithm for processing interferometric data-stacks: SqueeSAR." IEEE Transactions on Geoscience and Remote Sensing, vol. 49, no. 9, pp. 3460-3470. [7] Marinkovic, P., G. Ketelaar, F. van Leijen, R. Hanssen (2007). "InSAR quality control: Analysis of five years of corner reflector time series". In Proceedings of Fringe 2007 Workshop, pp. 1-8. [8] Adam, N., A. Parizzi, M. Eineder, M. Crosetto (2009). "Practical persistent scatterer processing validation in the course of the Terrafirma project". Journal of Applied Geophysics, vol. 69, no.1, pp. 59-65. [9] Lanari, R., et al. (2007). "An overview of the small baseline subset algorithm: A DInSAR technique for surface deformation analysis". Pure & Applied Geophysics, vol. 164, no. 4, pp. 637-661. [10] Berardino, P., et al. (2003). "Use of differential SAR interferometry in monitoring and modelling large slope instability at Maratea (Basilicata, Italy)." Engineering Geology, vol.68, no. 1, pp. 31-51. [11] Bonì, R., et al. (2015). "PSI-based methodology to land subsidence mechanism recognition." Proceedings of the International Association of Hydrological Sciences, vol. 372, 357, p.4 [12] Bonì, R., et al. (2018). "A methodology to detect and characterize uplift phenomena in urban areas using Sentinel-1 data." Remote Sensing, vol. 10, no. 4, 607, p.23. [13] Tomás, R., et al. (2013). "Monitoring an earthfill dam using differential SAR interferometry: La Pedrera dam, Alicante, Spain." Engineering Geology, vol. 157, pp. 21-32. [14] Sousa, J. J., et al. (2013). "Multi-temporal SAR interferometry reveals acceleration of bridge sinking before collapse." Natural Hazards & Earth System Sciences, vol. 13, no. 3, pp. 659-659. [15] Berti, M., et al. (2013). "Automated classification of Persistent Scatterers Interferometry time series." Natural Hazards and Earth System Sciences, vol. 13, no. 8, pp. 1945-1958. [16] Solari, L., et al. (2018). "Fast detection of ground motions on vulnerable elements using Sentinel-1 InSAR data." Geomatics, Natural Hazards and Risk, vol. 9, no. 1, pp. 152-174. [17] Kalia, A. (2018). "Classification of Landslide Activity on a Regional Scale Using Persistent Scatterer Interferometry at the Moselle Valley (Germany)." Remote Sensing vol. 10, no. 12, 1880, p.16. [18] Navarro, J. A., et al. (2019). "Automating the Detection and Classification of Active Deformation Areas—A Sentinel-Based Toolset." Multidisciplinary Digital Publishing Institute Proceedings, vol. 19, no. 1, 15, p. 4. [19] Tomás, R., et al. (2019). "Semi-automatic identification and pre-screening of geological–geotechnical deformational processes using persistent scatterer interferometry datasets." Remote Sensing, vol. 11, no. 14, 1675, p. 22. [20] Raspini, F., et al. (2018). "Continuous, semi-automatic monitoring of ground deformation using Sentinel-1 satellites." Scientific reports, vol. 8, no. 1, 7253, p. 11. [21] Evers, M., et al. (2019). "A study on recent ground deformation near Patras, Greece." Proc. SPIE 11156, Earth Resources and Environmental Remote Sensing/GIS Applications X, 111560L. [22] Podimata, M. V., P. C. Yannopoulos (2017). "A road map for resolving conflicts in dam's administration: The case of Peiros - Parapeiros dam in Greece". European Water, vol. 60, pp. 415-421. [23] Scaioni, M., et al. (2018). "Geodetic and remote-sensing sensors for dam deformation monitoring." Sensors, vol. 18, no. 11, p. 25. [24] Rashidi, M., S. M. Haeri (2017). "Evaluation of behaviors of earth and rockfill dams during construction and initial impounding using instrumentation data and numerical modeling." Journal of Rock Mechanics and Geotechnical Engineering, vol. 9, no. 4, pp. 709-725. [25] ICOLD, CIGB. (2000). "Automated dam monitoring systems: guidelines and case histories." International Commission on Large Dams, vol. 151. [26] Demetrakopoulos, Α. Κ., et al. (2013). "Dam operator of the water project - Water supply of Patras city, Patras Industrial Zone and Northwest settlements of Achaia prefecture via the Peiros and Parapeiros Rivers". University of Patras, Patras, pp. 136 (in Greek). [27] Evers, M., K. Schulz. (2020). "Strategies for PS Processing of Large Sentinel-1 Datasets". Proc. ISPRS Congress 2020, to be published.
Authors: Evers, Madeline (1,2); Hammer, Horst (1); Thiele, Antje (1,2); Cadario, Erich (1); Schulz, Karsten (1)The failure of a dam can have dire economic and humanitarian consequences for wide swaths of land downstream. Recent failures of both tailings dams and hydroelectric dams have brought significant attention towards monitoring solutions that can be employed to ensure the safety and stability of a dam site. One such remote monitoring solution is satellite Interferometric Synthetic Aperture Radar (InSAR). InSAR monitoring of dams can potentially detect precursor displacement on a dam face weeks to months before a failure, providing critical, actionable warnings to prevent impending disaster. In this study, 3vGeomatics (3vG) presents two analyses of InSAR data over dam failures from 2018 and 2019. The Dam I tailings dam, at the Córrego de Feijão iron ore mine near Brumadinho, Brazil, failed on January 25, 2019, resulting in the deaths of at least 248 people and the release of 12 million cubic meters of tailings. 3vG has analysed four sets of satellite SAR images over this site. These include two at lower resolution (20m), from the Sentinel-1 constellation, and two at higher (3m) resolution, from the TerraSar-X (TSX) and COSMO-SkyMed (CSK) satellites. Subtle signals are observed on the dam wall and significant consolidation of the tailings material is detected behind the dam wall for years before the dam collapse. No accelerations are detected on the dam face at the 2-sigma level in any of the data sets in the two months prior to the dam collapse. The behaviour of the dam does not appear to have been significantly different in the period immediately before the collapse compared to the dam’s behaviour over the past 3 years. For this dam, an InSAR monitoring program operating in real time would not have been able to produce an actionable warning about the imminent collapse of the dam based on the noise level of the available InSAR data and the past behaviour of the dam. Saddle Dam D of the Xe-Pian Xe-Namnoy hydroelectric power project in southeast Laos collapsed on July 23, 2018, resulting in at least 40 deaths and the displacement of over 6000 people due to the release of roughly 400 million cubic meters of water. Sentinel data over the dam, starting in February 2018, were processed by 3vGeomatics. From February to May, the dam appeared stable with no significant displacement detectable. In the middle of May, the dam began to exhibit noticeable displacement, with a total of 2.5 cm of displacement evident by July 17, one week before the dam collapse. Further, the rate of displacement leading up to the collapse appears to have been accelerating. For this dam, InSAR monitoring has clearly identified concerning precursor displacement that could have produced an actionable warning to the dam operators. These two case studies highlight that InSAR monitoring can provide critical early warnings of impending dam collapses in some, but not all, cases of dam collapse.
Authors: Donegan, Stephen Cody; Pon, Andy; Holden, David James; Goldsbury, Ryan Nelson; Loader, Daryn JamesMinerals and metals are essential for generating and supplying «green» energy and they play an increasingly central role in achieving a low carbon future. The total consumption of metals (e.g. Al, Co, Fe, Pb, Li, Mn and Ni) is expected to rise significantly. Mining these resources, however, produces immense amounts of mine tailings that need to be stored locally in ponds impounded by embankment dams, so called tailings storage facilities (TSF). For economic reasons, TSF are usually constructed with local material, that is often not ideal for this purpose, or even from tailings material directly. Failure of a TSF can have catastrophic effects on the immediate surroundings, such as causing loss of life and destroying infrastructure, as well as impacting the environment downstream of the dam. In the past decade, several dam failures of significant scale have occurred around the world. Tailings dams range in size from smaller horseshoe-shaped dams, which seal the pond off against a mountain side, to large ring-shaped dams with crest lengths of several tens of kilometres. Given the size of such dams and their often remote locations, remote sensing is an important tool to monitor their stability. Here, we present two case studies from tailings dams in Europe and in South America, for which ground displacements were mapped with radar interferometry (InSAR): The Zelazny Most tailings dam in Poland and the recently collapsed Feijão Mine tailings dam in Brazil. In both cases, Sentinel-1 stacks were analysed using both PS (Persistent Scatterer) and SBAS (Small BAseline Subset) techniques. Being made of rather soft material with little to no infrastructure installed on the dams, getting coherent backscatter can be challenging. Special focus was given to the crest of the dams, where the slow-moving or stable scatterers on the dams are in close proximity to the much faster moving impoundments. In addition, the effects of the spatial filtering in the SBAS workflow were analysed. The case of the Feijão Mine tailings dam collapse provided an opportunity to investigate whether it was possible to identify precursors in the displacement data or if the collapse occurred near-instantaneously, e.g. by liquefaction due to short-term pore pressure changes. Advantages and disadvantages of both the PS and the SBAS techniques were analysed to assess their suitability for long-term monitoring of tailings dams, which will help to improve risk assessment for tailings dams.
Authors: Vöge, Malte; Frauenfelder, Regula; Salazar, Sean E.; Torgersrud, ØyvindRailway infrastructure includes complex structures such as railway tracks, power lines, embankments, platforms, and bridges. The stability of the railway tracks and the aligned structures is subject to geological characteristics of the earth's surface on which they are built. Several processes affect the stability of the railway infrastructure. The occurrence of surface movements caused by natural (environmental) or anthropological factors, causes instability in the infrastructure. Instability affects the health of the infrastructure by changing its geometry; aging/depreciation can also cause instability. To systematically detect railway instability, we attempt to employ time series InSAR (interferometric synthetic aperture radar) techniques (Chang, 2015; Chang, Dollevoet, & Hanssen, 2017) to generate temporal dynamics of all identified InSAR measurement points (i.e., Persistent Scatterers, PS, for our case). However, interpreting the geolocation of the deformation (deformed PS point) is a challenge due to poor radar estimates—the positioning estimates are in the order of meters (Dheenathayalan, Small, Schubert, & Hanssen, 2016). As a result, associating the PS to the actual ground object is unlikely straightforward. Recently, an innovative approach has been presented to improve the geolocation of PS based on (airborne) laser scanning data (Dheenathayalan, 2019; Hu, Leijen, Chang, Wu, & Hanssen, 2019; Van Natijne, Lindenbergh & Hanssen, 2018). Hereby, the PS are linked to the most likely corresponding laser scanning point. However, the results obtained in previous studies were based on high-resolution SAR data, such as obtained by the TerraSAR-X and RadarSAT-2 (XF-mode). Due to the high-resolution, the precision of the initial PS positions can still be assumed to be relatively high. In this study, we assess the feasibility of applying the technique to medium-resolution SAR data acquired by the Sentinel-1 satellite. Apart from the lower resolution, the Sentinel-1 satellite has a relatively small orbital tube, resulting in a poor relative height estimate of the PS. Both the low resolution and the poor height estimate cause a large uncertainty in the PS position. The height accuracy of Sentinel-1 data is in the order of meters. The error ellipsoid is long and in order of meters in cross-range, range, and azimuth direction. The analysis consists of a number of processing steps. First, the deformation time-series for the area of interest (AOI) is estimated by the integer least-squares estimator method for the Persistent Scatterers (PS) (Hanssen, 2001). The radar estimates are transformed from a 2D radar system to a 3D terrestrial reference frame (Dheenathayalan, Small, & Hanssen, 2014). The unknown height estimates of the PS are improved using a 0.5m2 digital surface model (DSM) as a reference (Yang et al., 2016). The positioning accuracy of the PS is based on the (i) factors influencing all PS systematically; and (ii) factors specific for per pixel (van Leijen, 2014). The association of the radar estimates to the LiDAR point is performed by the nearest point linking method (Van Natijne et al., 2018). We test and evaluate the approach using a stack of Sentinel-1 data over the Netherlands. The dataset consists of 118 images, with a total time span of 708 days. The acquisitions are being obtained from the ascending orbit and in dual-polarized mode (VV/VH). The high-resolution LiDAR data set of Actueel Hoogtebestand Nederland 3 (AHN3) with a point density of 16 points/m2 to improve 3D geolocation of PS. The test area is 15 km long and part of the 172 km long Betuwe freight corridor. The corridor links the Rotterdam port to central Europe and is vital in terms of the regional economy. The freight corridor is operational for heavy loads throughout the year. A buffer of 50 m along the rail track was monitored to detect deformation. Assessment of the results shows that the estimated position of the radar measures is improved considerably, i.e., a shift of -11.1 m. The height correction is implemented per PS point at the sub-pixel level. The snapping is processed at a significance level of 0.25 as a threshold, 97.8 percent of the PS points were linked to the LiDAR data and a standard deviation of 3.9 m in cross-range direction. Whereas the mean of -1.2 m, -1.7 m, and a standard deviation of 2.6 m, 2.4 m, for range and azimuth direction, respectively. We thank the Copernicus services for providing the Sentinel-1 data. SkyGeo, The Netherlands, is acknowledged for performing the PS time-series analysis, which formed the input for our analysis. Reference Chang, L. (2015). Monitoring civil infrastructure using satellite radar interferometry (TU Delft). https://doi.org/10.4233/uuid:f4c6a3a2-73a8-4250-a34f-bc67d1e34516 Chang, L., Dollevoet, R. P. B. J., & Hanssen, R. F. (2017). Nationwide Railway Monitoring Using Satellite SAR Interferometry. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 10(2), 596–604. https://doi.org/10.1109/JSTARS.2016.2584783 Dheenathayalan, P. (2019). Optimizing the exploitation of persistent scatterers in satellite radar interferometry (Delft Universtiy of Technology). https://doi.org/10.4233/uuid:aa1ef96f-4da9-41ff-bff8-30186ef2a541 Dheenathayalan, P., Small, D., & Hanssen, R. F. (2014). {3D} geolocation capability of medium resolution {SAR} sensors. Proceedings of the Joint International Geoscience and Remote Sensing Symposium (IGARSS 2014), 35th Canadian Symposium on Remote Sensing (35th CSRS), 13-18 July 2014, Quebec, Canada, 3–6. Dheenathayalan, P., Small, D., Schubert, A., & Hanssen, R. F. (2016). High-precision positioning of radar scatterers. Journal of Geodesy, 90(5), 403–422. https://doi.org/10.1007/s00190-015-0883-4 Hanssen, R. F. (2001). Radar Interferometry Data Interpretation and Error Analysis (Kluwer Academic Publishers; F. van der Meer, Ed.). https://doi.org/10.1007/0-306-47633-9 Hu, F., Leijen, F. J. van, Chang, L., Wu, J., & Hanssen, R. F. (2019). Monitoring Deformation along Railway Systems Combining Multi-Temporal InSAR and LiDAR Data. Remote Sensing, 11(19), 2298. https://doi.org/10.3390/rs11192298 van Leijen, F. (2014). Persistent Scatterer Interferometry based on geodetic estimation theory (Delft University of Technology). https://doi.org/10.1017/CBO9781107415324.004 Van Natijne, A. L., Lindenbergh, R. C., & Hanssen, R. F. (2018). Massive linking of PS-InSAR deformations to a national airborne laser point cloud. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives, 42(2), 1137–1144. https://doi.org/10.5194/isprs-archives-XLII-2-1137-2018 Yang, M., Dheenathayalan, P., Chang, L., Wang, J., Lindenbergh, R. R. C., Liao, M., & Hanssen, R. F. (2016). High-precision 3D geolocation of persistent scatterers with one single-Epoch GCP and LIDAR DSM data. European Space Agency, (Special Publication) ESA SP, SP-740.
Authors: Sakpal, Nikhil (1); Chang, Ling (1); Oude Elberink, S.J. (1); van Leijen, Freek.J. (2); Hanssen, Ramon F. (2)Open access Sentinel-1 SAR constellation (C-band) launched in 2014 has allowed public managers to develop, maintain and update InSAR monitoring systems at large scale. Tuscany Region deployed in 2017 a semi-automatic ground deformation monitoring system, detecting a new subsidence area over the Pistoia city center (Tuscany Region, central Italy). Since previous subsidence areas related to groundwater extraction were described using InSAR in the sedimentary Florence-Prato-Pistoia (Fi-Po-Pt) basin, the main target of this work is the new detected area located in the Pistoia city area. S-1 results were validated using high resolution Cosmo-SkyMed (X-band) obtaining a good agreement between S-1 and CSK velocities (1.7 and 2.3 cm/year, respectively) and even better comparing the time-series (around 0.7 cm of RMSE). S-1 datasets are composed by 136 ascending and 128 descending images from December 2014 to May 2018, meanwhile CSK datasets are composed by 60 ascending and 34 descending images from January 2015 to March 2018. Possible causes of this phenomenon are still under investigation considering as most plausible hypothesis the subsidence caused by groundwater changes. Although subsidence is described as a main vertical movement, small horizontal displacements have been related. Combining both available acquisition geometries, vertical and east-west components of the movement were calculated, detecting a slow horizontal displacement towards the city center with both satellites (0.5 cm/year in S-1 and 0.7 cm/year in CSK). In order to evaluate the possible affection of subsidence to building stability, two damage surveys were designed inspecting 227 structures. Combining InSAR velocities and observed damages, a damage/no damage fragility curve was calculated. Fragility curves, typically used to calculate damages related to earthquakes and tsunamis, were applied to calculate a continuous damage probability map and transpose the information to a cadastral map, obtaining the damage probability of each building. Moreover, using available information of building market prices from the Italian government OMI database, exposed assets and potential losses were calculated. The proposed methodology achieved the creation of valuable and easy-understandable set of maps about the spatial extension, severity and probability of the studied phenomenon useful for urban planners. The good performance in Pistoia city will lead to apply it in other areas affected by subsidence and other kind of ground displacements like landslides.
Authors: Ezquerro, Pablo (1,5); Del Soldato, Matteo (2); Solari, Lorenzo (3); Tomás, Roberto (4); Raspini, Federico (2); Ceccatelli, Mattia (2); Fernández-Merodo, José Antonio (1); Casagli, Nicola (2); Herrera, Gerardo (1)This contribution discusses how geotechnical risk management in mines and quarries of Sibelco is based on geotechnical design and how local and remote monitoring are developed for complementary use in different geotechnical settings to help improve safety. Sibelco extracts and refines various types of minerals and secondary raw materials with production sites all over the world. Besides high quality industrial silica sands, Sibelco mines many other minerals for industrial applications such as clay & kaolin, olivine, quartz, feldspar, barite, and others. These minerals are extracted by hard rock underground and surface mining, in clay and sand pits, and in dredging operations. Sibelco's focus is on safe mining. In order to mine safely geotechnical risk must be managed as described in Sibelco's safe mining performance standard (HS MGT 2019) and the associated guidelines and technical manuals (e.g. Schmitz 2016, 2018, 2019a, b, c, d; Schmitz and Haggman 2016, Santiago and Schmitz 2017; Santiago 2018; Santiago and Hagen 2018, Schmitz and Berwouts 2019, Schmitz and Habisch 2019) and handbooks (Schmitz, 2018). This approach is important because it shows how Sibelco’s safety culture and our design of and monitoring processes in mines and quarries are intertwined and interbedded with processes such as geotechnics (i.e. the geotechnical management plan) and mine design. For the geotechnical design of e.g. slopes, tunnels and caverns Sibelco follows the principles of Eurocode 7 and uses the concept of the Geotechnical Category described herein. Although not directly aimed at mining, the design philosophy of Eurocode 7 is typical for processes used since forever in mining such as relying on the observational method based on monitoring. During project realisation such monitoring takes place to verify design and to check if specifications are met. Monitoring continues after the project has been realised. The type of monitoring depends on the design, geotechnical category, mining method and geotechnical environment. Monitoring of slopes in hard rock operations is different from monitoring of slopes in clay quarries or monitoring of slopes under water in dredging operations. The parameters that are monitored in mines and quarries of Sibelco vary from local deformation monitoring on the object itself (e.g. over a single rock discontinuity), to large scale monitoring of deformations on the scale of a mine and quarry (e.g. using inSAR), to monitoring of stresses in hard rock operations. As is described in the text monitoring can be done to verify e.g. slope design but can also be used to deduce rock mass properties, used as input parameters for the design process itself. InSAR is a key element in managing geotechnical risk inside and outside the mines and quarries. Looking at slope stability into the actual pit, looking at deformation of on-site tailings storage facilities, but also helps monitoring the infrastructure and housing surrounding the quarry. With InSAR monitoring an up-to-date view of the quarry is always available. One of the benefits of using InSAR is the availability of historic data. With this, it is possible to go back in time and show that the movement of a house occurred prior to the start of the mining operation. This can also be used to discuss the effects of mining with regulators. Another benefit is to have a shared data-source to refer to. Some of the monitored sites have very little deformation: using InSAR data, slope stability can be evaluated without anyone needing to be in the quarry. Apart from being practical, using remote sensing data is also economical. The costs of monitoring a large number of sites remotely using InSAR, are a fraction of the costs of manual measurements. While not being a replacement of manual measurements, InSAR helps - as an automated complementary system - optimizing the use of onsite personnel or instrument resources. By inspecting regularly updated InSAR data, geotechnical management can spot geomechanical risks and direct on-site engineers for further inspection.
Authors: Schmitz, Robrecht (2); Houtepen, Martijn (1); Leezenberg, Pieter Bas (1)With the technological developments of InSAR, much focus has been directed at retroactive analysis of the failures of infrastructure, and yet systematic application of these insights is still considered a distant goal. The implementation of a professionally developed image processing pipeline and the formation of a team of InSAR field application experts has enabled large-scale systematic monitoring of the gamut of large infrastructure projects; bridges, tunnels, hydroelectric reservoirs, railway networks, offshore oil platforms, tailings storage facilities, electricity grids and dams. This submission will discuss the only-now obtainable impact of such a pipeline through a selection of case studies. Upon acquisition of an AOI-applicable image, ‘rapid delivery’ ensures an acquisition-to-ingestion time of under 24 hours, where ingestion is defined as the triggering of our automated image processing pipeline. Furthermore, the acquisition-to-delivery time is under 48 hours, where a delivery contains individualised, quality controlled reports and actionable intelligence on monitoring targets. Core functionality of the automated image pipeline enables the Field Application team to fine-tune the processing of both PS InSAR and DInSAR to individual AOIs and requirements. This fast turnaround is essential for use-cases which require rapid response and has long been a major hurdle on the road to InSAR adoption. Surface events at oil and gas fields is one such rapid-response application. ‘Blisters’ can occur on some oil fields in California due to leakage of corroded injector-wells that are supposed to drive oil to the extraction wells. Leaking injection fluid results in a localised, superficial, inflationary deformational structure. A surface event, or rupture, of such a feature may have large environmental consequences and thus early detection is vital. These features, which prove challenging to detect on-site due to their localised nature, are only now visible through the rapid delivery of high-resolution TerraSAR-X persistent scatterer interferometry. This automation of the ingestion and processing of SAR images was extended to include preliminary blister detection alerts through machine learning, which has reduced the acquisition-to-alert time to 1 hour. Satellite repeat intervals remain the primary source of delay for real-time monitoring. Through the application of this method, the period of undetected injector-leakages has been reduced significantly, which has enabled early shutdown and repair of corroded injector-wells and no further surface events have been observed. Urban tunneling projects is another such application: where consequences of multi-lane tunnel collapse under a major urban centre justify and require a strict monitoring regime. During the tunneling process, grout is applied at pressure into the tunnel walls for ground stabilisation and to expel groundwater. Too little pressure and the tunneling may result in surface subsidence, where inhomogenous subsidence may lead to strain on urban structures and damage, while excessive pressure can lead to surface uplift and the same effect. Systematic, near-real time monitoring of a large urban tunneling project in Sweden has shown the viability of InSAR for such cases and has allowed for optimisation of the tunneling process in response to recorded surface expression data. An active monitoring system is in place to detect any change in subsidence rates of buildings in the affected urban area.
Authors: Maljaars, Hanno; Plain, MorganTailings dam failure has occurred with increasing frequency in recent years, causing loss of life and the release of large volumes of dangerous material damaging to the local ecosystem. Notable examples include Brumadinho, Mariana, Cadia and Mount Polley tailings dam failures. Knowledge of the structural integrity of Tailings Storage Facilities (TSF) form an integral part of mine risk management programs, and InSAR is used increasingly as an accurate and cost-effective means of monitoring the larger area of the TSF, often without need for terrestrial measurements. This is especially relevant as a long-term monitoring solution at sites without year-round access. In the past, InSAR was applied to provide periodical updates or for historical data analyses. Recent developments such as the availability of new satellites constellations with shorter revisit times, combined with increased computing power and processing advances are creating opportunities for new InSAR applications. The longer term goal would be to detect the risk of a TSF collapse and take timely prevention measures. The present work discusses different case studies where InSAR has detected precursor TSF deformation, and how this information was used by geotechnical experts and decision makers to feed into early warning systems. The main concept is to apply InSAR to detect anomalous areas of large or increasing displacement rate in the body of a tailings dam. Practical examples are presented to highlight how InSAR can routinely be used as a support tool to differentiate the cases requiring attention from normal, operational situations where non-critical movements are present. InSAR can also become a valuable tool to support monitoring programs where conventional geotechnical instrumentation is used. Given InSAR interpretation permits identification of regions in a TSF where deformations are higher, sensors installation can be optimised. Mine closure and TSF decommissioning works can also benefit from the use of InSAR measurements in a practical and accurate way to support safety. The conclusions of the study address the benefits and limitations of InSAR as a standalone and complementary technology for TSF, with the aim to contribute to advancements and support the continued efforts of the mining industry to reach safe and sustainable operations.
Authors: Jewell, Richard (1); Salzer, Jacqueline (2); Kormann, Alessander (3); Mahapatra, Pooja (4)The aim of this study was to evaluate millimeter-scale deformations in Tallinn city center in 2016-2020 with the help of multitemporal SAR interferometry (InSAR) technique and ESA Sentinel-1 data. The persistent scattered InSAR (PS-InSAR) was used to analyze Sentinel-1A/B IW SLC images from descending (DESC) and ascending (ASC) orbits: (1) DESC #80 (incidence angle 43.52° & azimuth angle 189.4°) with 193 images, (2) ASC #160 (33.54° & 348.0°) with 203 images. The displacement velocities of PS targets along the line-of-sight (LOS) were estimated for the full period of time (Jun 2016 – Nov 2020), but also for shorter periods to monitor deformation changes in yearly basis. The LOS velocities derived from the ASC and DESC nearly polar orbits provided sufficient information to calculate the horizontal (east-west) component as well as vertical component of the displacement velocity vector. The gridding of point velocities (by using 2D splines) resulted LOS velocity models which then were used to estimate the horizontal and vertical velocity models. Due to the different reference points of LOS velocities (from ASC, DESC orbits) used in PS-InSAR analysis, the unification to a common reference velocity was needed. The 3D velocity of permanent GNSS station MUS2 in Tallinn was applied to fix and unify velocity references. Displacement rates interpolated from the vertical velocity model were compared with the rates based on the repeated levelling of Tallinn height network in 2007/2008 and in 2019. Although similarities between the rates were found, the levelled rates with temporally and spatially sparser data points over different time period did not allow firm conclusions to be drawn. The complexity of the movements in Tallinn is illustrated by the temporal change of PS-InSAR based vertical displacements. The series of vertical velocity models over shorter period of time revealed the subsidence of city center in 2016-2017 and rebound/uplift in 2018-2019 with relatively stable periods in 2017-2018 and 2019-2020 (Fig. 1). Further research would be needed to determine whether the displacements were due to the changes in groundwater level, construction activities, some other anthropogenic/natural factor or combination of them. PS-InSAR analysis through SILLE was ordered and financed by Tallinn City. SILLE (sille.space) is a satellite-based monitoring and risk measurement service developed and maintained by AS Datel.
Authors: Oja, Tõnis; Gruno, AntiThe Millennium Tower (MT) is a 58 story (197 meters), luxury condominium skyscraper located in downtown San Francisco. In 2016, it was publicly reported that MT was sinking and tilting. 3vGeomatics (3vG) was retained as a consultant on behalf of the building’s developer, Mission Street Development LLC, to provide InSAR based historical displacement data of MT. The results were used by consulting experts, in conjunction with other monitoring data for the building, in efforts to resolve litigation related to the MT. The SAR data used for the InSAR analysis comprised 225 descending and 107 ascending TerraSAR-X scenes acquired in StripMap mode at 3 meter resolution from May 2009 to February 2018. 3vG generated over 2000 total data points on MT, covering all four sides and the roof. By modelling the tower as a rigid structure, we were able to use all of these data points to constrain a small number of meaningful fit parameters. The tower is assumed to move vertically, and to tilt from a pivot point near its base. This allows us to fit vertical-displacement, tilt magnitude, and tilt-direction. This is possible with only a single satellite orbit geometry. Using this approach on the ascending and descending data sets separately allowed us to produce two independent sets of results for comparison. Finally, a combined analysis was performed to make the most precise and well sampled determination of the tower’s motion history. 3vG’s InSAR analysis determined that MT moved 20 cm in the vertical (down) direction and that the top of the building moved 45 cm in the southwest direction (tilt) over 9 years. Interestingly, the rate of sinking and tilting, and the direction of tilt, varied over time. In order to produce uncertainties for the results, 3vG performed the same modelling for all nearby skyscrapers with suitable data. Assuming that these buildings were stable, we analyzed the variance of the fit parameters to determine the uncertainties in those parameters for MT. Using this method, we were able to determine the uncertainty in vertical displacement to be 2 cm and the uncertainty in tilt to be 5 cm, with 95% confidence. All large construction projects have some degree of settlement expected by design. The novel approaches developed here open up new opportunities for monitoring future city infrastructure using InSAR based techniques.
Authors: Mackenzie, Todd Phillip; Goldsbury, Ryan NelsonLand subsidence has been occurring in the mega-city of Tehran (with more than 8 million population) in the last decades mainly due to the over extraction of groundwater. Previous studies showed that the maximum subsidence rate could reach to 25 cm/year in some areas in southern Tehran. In addition to land degradation in plains around Tehran, the occurring land subsidence can have severe impacts on the built environment and infrastructure in/around Tehran. Although the large subsidence of southern Tehran has been already investigated and reported in many studies, the details of the spatio-temporal characteristics of the subsidence in this area and also other smaller but important local subsidence mechanisms inside the built environment of Tehran has been overlooked. In this study, we provide new insights into the spatio-temoral pattern of the land subsidence in Tehran from processing of 6 years Sentinel-1 data from 2014-2020. The spatial pattern of subsidence in southern Tehran and its maximum subsidence rate (25 cm/year) are similar to earlier studies. However, the spatio-temporal analysis of the results reveals that the subsidence pattern varies significantly both in time and space. In addition to the large subsidence area of southern Tehran, many local subsidence areas has been also identified and analyzed. The results show that some of these local deformation mechanisms have not been induced by groundwater extraction but by some other anthropogenic mechanisms such as underground activities and tunneling. In summary, the main highlights of this study can be listed as: Highly temporal variations in the pattern of subsidence, especially seen as annual signal at different areas in Tehran, Localized subsidence at/close to important infrastructure (bridges, metro tunnels, pipe-lines, power lines, ….), Highly spatial variation in the spatial pattern of subsidence, depending on geological and hydrological setting of the area, Development of the subsidence towards the main urban area of Tehran, Pre-collapse subsidence pattern at some of the sinkholes in the urban area of Tehran, Significant horizontal deformation (by combining ascending and descending data) at locations with high vertical deformation gradient, Damage to building and infrastructure at locations with horizontal deformation. In addition to the aforementioned points, some InSAR processing challenges regarding phase unwrapping in Tehran area with high deformation rate and geocoding in the dense urban area of Tehran are addressed.
Authors: Ajourlou, Parviz (1,2); Samiei Esfahany, Sami (1)The present paper proposes an advanced observation technology for the airport runway monitoring service (RMS) based on the complementary use of interferometric SAR data from Sentinel-1-A/B satellites, hydrometeorological observations and ground geotechnical surveys. We investigate the suitability and technical feasibility of using satellite dual-sense persistent scatterer interferometry for non-invasive remote control and long-term monitoring of runway surface conditions at Vienna International Airport*. The research focuses on the detection, measurement and interpretation of runway surface distress, pavement deformation and contamination, separately in the touchdown, midpoint, and rollout thirds of the Vienna airstrips on a weekly to monthly time scale, regardless of subjective judgement. We identify system requirements, observation targets, essential conditions and opportunities for the combination of space and ground data in close contact with airport operators and experts on site. Preoperative developments towards practical implementation of the RMS service prototype for planning airport maintenance operations are discussed. The remote sensing technique of persistent scatterer interferometry (PS-InSAR) using Sentinel-1 SAR data is well suited for geotechnical monitoring of pavement distress at airports. However, strict accuracy requirements and usually a small number of permanent scatterers found on runways and taxiways due to the relatively low backscatter of microwaves and random fluctuations in local weather are the main limiting factors for the practical application of PS-InSAR at airports. It appears that we have solved the latter problem on the list using the iterative remove-restore approach to SAR data analysis and weather correction referred to as “better-half” technique, which, unlike other known stratagems, takes into account changes in the state of the target surface in addition to tropospheric corrections. Two Sentinel-1 SAR image sequences with precise orbital data for the periods of 2014-2017 (94 IW scenes) and 2019-2020 (117 IW scenes) were ranked and rearranged according to the hourly precipitation amount and the condition (dry, wet, moist, frozen, snow-covered etc.) of the runway surface recorded at airport weather station No. 110360. These rearranged “all-weather” image sequences were divided into two equally long “dry” and “wet” parts, each including the same master scene, and processed separately using the standard procedure for persistent scatterer interferometric analysis implemented in the SARscape software package. The threshold for interferometric coherence values was set at 0.85. The number of persistent scatterers found on runways and taxiways is 1233 for bad-weather data, 1663 for all-weather data and 3228 PSs for the image time series obtained under good (dry) weather. The resulting displacement time series show a fairly stable state of the runways with an average displacement rate of about 2.5 mm / a around the value zero. In some places and dates, however, we detected outliers with magnitudes up to 9 mm/a that exceed three times the standard deviation criterion. In the “all-weather” displacement time series the outliers constitute approx. 8% of all measurements while in the “better-half” displacement time series the number of outliers is radically reduced. In good-weather displacement time series there are still some outliers with magnitudes exceeding the limit of 6 mm specified for medium-severity distress on runways (see Appendix). Such outliers were subjected to further causal statistical analysis to determine whether they were due to weather-related errors or runway distress. The average magnitude of weather-related errors in the study airfield is between 2 and 4 mm/a, and rainy weather is responsible for 50% of all outliers. Some outliers identified in both the original and better-half displacement signals were related to runway and taxiway renovations in 2016. 29% of all outliers could not be associated with hydrometeorological effects. In close proximity to each suspicious outlier, we have determined several conjugate persistent scatterers, which are scatterers that can be identified as the same target in better and worse halves of SAR datasets. Then we measured the difference in the range coordinates of conjugate “dry” and “wet” PSs with subpixel accuracy. On runways and taxiways, the magnitude of relative range shifts varies between 0.1 and 0.75 pixel and depends on the incidence angle of the radiation. Usually, conjugate “dry” and “wet” PSs are shifted in opposite directions from the position of the reference PS, which is determined with the all-weather dataset. Systematic relative range shifts indicate weather impacts. Wet air has a lower density and refractivity than dry air and causes straightening and shortening of the radar propagation path in the troposphere. The relative range shift of wet PCs is thus towards the satellite (negative). In the deformation maps derived from “wet” data most of the PSs and ground surface appear to be uplifting. In our products they are marked with cyan and blue points. Rapidly varying shift values indicate local deformations on the runway surface. We have accounted for varying incidence angle and mapped relative shifts in the range over the entire airfield on a scale of 20.000. The resulting displacement maps were rasterized with a pixel size of 10 m and laid over optical images from WorldView-3 to identify PSs and to analyse the PS phase variation (see Appendix). The somewhat patchy appearance of the displacement maps is explained by the limited number and the irregular distribution of the conjugate PSs detected on and near the runway surface. The main maximum of the range shift was found near reference point no. 3 on the old tow route with numerous cracks on the surface. Several extremes were found between the runways and in the area of their intersection. Other local maxima coincided with optically recognizable rubber deposits from aircraft tires that accumulated in the take-off, brake-and-turn and touchdown zones of airport runways. In addition, in the south-western part of the airfield, where a third runway is to be built, we discovered a small area with a land subsidence of 1 mm/a. Two outliers at reference points 3 and 2, which were detected on August 1 and 7, 2020, were interpreted as thermal displacements of separate slabs due to rapid daytime changes in surface temperature of 12 ° C and 14 ° C, respectively. The expansion joints built into the runways largely absorb the expansion or compression of separate concrete slabs. However, the proper functioning of the joints is sometimes disrupted by the fact that solid materials fall into the joint space. In hot weather, this leads to a local upward movement of the slab edges or to splinters near the joint. Such emergencies must be reported and repaired immediately because of severe damage potential to aircraft. The resulting deformations of concrete slabs, both horizontal and vertical, are large enough to be recorded and measured with Sentinel PS-InSAR. We conclude that the satellite interferometric RMS controller is useful and feasible. The research was carried out under the contract ID. 881043 (ICARUS) with the Austrian Research Promotion Agency. Sentinel-1 SAR data was downloaded from the Copernicus Open Access Hub at https://scihub.copernicus.eu/ . *) Vienna International Airport serves all types of aircraft without size restrictions and without a curfew. It has two runways, each 60 meters wide, including shoulders, and route indicators on either side of the runways. The capacity of the runway system is given as 68 coordinated flight movements per hour. It is planned to build a third runway in the southern part of the airfield (https://www.viennaairport.com/en).
Authors: Sharov, Aleksey (1); Schardt, Mathias (1); Nikolskiy, Dmitry (2); Gutjahr, Karlheinz (1)Motivation: Sea crossing bridge can cross bays, straits, and deep marine water, which are critical nodes to ensure the smooth flow of traffic arteries on the mainland. Even subtle displacements may affect the performance of bridges and cause high maintenance and repair costs. Thus, the timely and accurate inspection of their structural safety status is essential to guarantee the smooth operation of public transportation, and to avoid personal and property loss[1-2]. However, the traditional bridge deformation monitoring sensors such as the total stations, accelerometers, leveling and strain-meters are only available at sparse discrete locations on a few number of bridges, with a limited spatial extent or low temporal sampling frequency due to the constraints on manpower and financial costs[3-4]. As sea crossing bridge are usually huge in size and complex in structure, the above methods would inevitably miss some security risks and cannot explain the global deformation of the bridge. On the contrary, the Synthetic Aperture Radar Interferometry (InSAR) technique can quickly extract surface deformation over a large area at fine resolution, without the limits of geographical and meteorological conditions[5-7]. It has unique advantages of low labor and material costs, and no effects on bridge normal operation, which allows the timely monitoring of displacement with a dense grid of measurement points, showing huge application potential in bridge structural health monitoring. Current Issues: Although previous studies on InSAR bridge deformation monitoring have achieved certain research results on thermal dilation monitoring and deformation interpretation[8-16], there are still some issues that need to be solved urgently to achieve the structural health monitoring of sea crossing bridge. First of all, the previous research mainly focused on high coherence bridge such as steel bridge. The density and accuracy of point targets (PTs) is low on structures with significant de-coherence effects, leaving how to extract dense and accurate PTs upon partially coherent sea crossing bridge still a problem. Moreover, previous thermal dilation monitoring relies on the linear deformation assumptions of PTs, or a large amount of high-resolution images and calculations, rather than combining the structural characteristics. When the above conditions are not true, it is not appropriate for long-span sea crossing bridge with long thermal dilation propagation distance. Therefore, how to accurately model and separate the thermal dilation of sea crossing bridge is still a challenge. Finally, due to the limit of SAR side-looking imaging geometry, the traditional InSAR results are difficult for non-expert users to comprehend and interpret. The simple analysis of bridge linear deformation map is meaningless for long-span sea crossing bridge with complex deformation patterns, making the InSAR results, currently, still have a certain gap from the engineering applications. Methodology: Aiming at the above issues, the goal of this study is to develop a measurement and analysis method to detect the global deformation of sea crossing bridge quickly and accurately by introducing the structural principle into the conventional time-series InSAR analysis. As for the PTs selection, the PSI and SBAS algorithms have been applied to extract both traditional point targets, i.e., targets on the bridge outline that contains one dominant scatterer, as well as extended targets, i.e., targets on bridge beam that spread over a collection of pixels. Structural semantic information is then introduced for the fine match recognition and posterior screening of the structural PTs. In terms of thermal dilation modelling, the qualitative analysis of the thermal dilation distribution and propagation regulation is firstly implemented through the time-series analysis of interferometric and unwrapping phases. Then, a data-driven regression analysis method weighted by the coherence of interferometry pairs is applied to establish a quantitative relationship between the temperature and the displacement. Finally, the material properties of bridges are utilized to validate the estimated thermal dilation parameters. In order to provide user friendly results, the height of bridges and the local viewing geometry parameters are used to calculate the PTs 3D positions through a local coordinate orthorectification. Then, the structural principle and InSAR observations are integrated to investigate the potential risk sources and security points on the sea crossing bridge based on the 3D visualized results. The risk level of different sections along the sea crossing bridge can be evaluated based on the weighted calculation of the identified risk indicators. Study Objects and Datasets: The East Sea Bridge, connecting the Nanhui New Town in Shanghai and the Yangshan Town in Zhejiang Province, is 25.3 km long. It is a part of the key supporting projects in Yangshan Deepwater Port. As a distinguished sea crossing bridge, the Hangzhou Bay Bridge was built from 2003 to 2008 and connects Jiaxing and Ningbo city, with the total length of 35.7 km. A stack of 31 ascending Sentinel-1 images from 2015 to 2017 is collected for the deformation analysis of these two sea crossing bridges in this study. Preliminary Results: Based on the InSAR and structural principle integration method, the detail deformation along the two sea crossing bridges during the observation time are estimated. After modelling and separating the thermal dilation along the bridge, the structural risk level is evaluated based on the analysis of risk indicators and structural security points. Security sections along the sea crossing bridges are identified, which should be focused on the future maintenance and management. The presented work is part of ongoing research activities in making InSAR techniques applicable for structural health monitoring of sea crossing bridge. The methodologies and outcomes of this study will promote the understanding of the catastrophic mechanisms and evolution of sea crossing bridge, as well as to provide near real-time monitoring results and reliable technical support for sea crossing bridge disaster prevention and daily maintenance.
Authors: Qin, Xiaoqiong (1,2,3); Wang, Chisheng (1,2,3); Xie, Xinfu (1,2,3); Li, Qingquan (1,2,3); Xiong, Siting (1,2,3); Zhang, Bochen (1,2,3)Large-scale man-made linear hydraulic engineering is a powerful way to improve the ability of water resources allocation and water supply guarantee. It has the features of long line, wide areas, complex engineering layout and many types of functional buildings. The South-to-North Water Diversion (SNWD) Project is a great infrastructure of China to alleviate the serious water shortage in North China and optimize the water supply pattern. It consists of three parts of channels: the Eastern Route Project, the Middle Route Project and the Western Route Project. The total length is 3797 km, involving 15 provinces and 500 million residents. To guarantee the secure operation, it is essential to monitor the geo-stability of such large-scale linear hydraulic engineering. However, traditional ground-based techniques are usually limited by the distribution density of sensors and relocation difficulty of monitoring instruments. Large-scale monitoring of channels and buildings also requires huge monitoring funds. The Synthetic Aperture Radar Interferometry (InSAR) has been considered as an effective tool for detecting long-term deformation and identifying possible safety problems in civil engineering. The satellite InSAR has the advantages of large-scale coverage, high-precision deformation measurement with low cost. High-frequency coverage of wide SAR images makes this InSAR more suitable for the deformation monitoring of the large-scale infrastructures, such as the SNWD Project. In this study, a long-term time-series Sentinel-1 IW dataset, including total five frames of Path 11, Path 113 and Path 40, was processed by Persistent Scatters InSAR (PS-InSAR) to monitor the dike of the Middle Route of SNWD Project, as shown in Figure 1. The unstable channel sections were detected and some key sections were analyzed in details. We obtained the distribution map of deformation rate of the channel in the Henan Province with a length of 730 km, as shown in Figure 2. This result indicates that the dike is generally stable, except for partial canal sections. Some deformation of channel sections is caused by the surrounding ground subsidence, while others are deformation of the channel slope itself. The Yuzhou-Changge channel section, according to the deformation rate distribution derived by PS-InSAR, crosses through a subsidence region, and the settlement canal is approximately 2.5 km long, as shown in Figure 3. Its maximum deformation velocity is near -20 mm/yr and the profile of the deformation rate along the channel line presents a very uneven deformation of the channel slope. The result is verified by in-situ leveling measurements and they agree well with each other. The section of the channel in Zhengzhou area also present complex deformation pattern. This channel section passes through deformation areas with uplift and subsidence appearing alternatively. The InSAR deformation time series is consistent with the leveling measurements. The relationship between the special deformation characteristics and the complex distribution of geological faults was discussed. We carried out an analysis of deformation distribution characteristics of Shahe Aqueduct, one of the most technically complex control engineering in the SNWD Project. The InSAR result reveals that this engineering is relatively stable, except for two design sections. For one of them, the deformation rate of its main body is smaller than that at its entrance. This is related to the engineering features of its lower supporting structure and geological conditions. The other one, the left side of which is excavated from the slope while the right side of which is high fill, is undergoing obvious deformation. The time-series InSAR technique with high efficiency, high precision makes it a powerful tool for long-term safety monitoring of the SNWD project. It can greatly reduce the monitoring cost for the SNWD Project. It can not only measure the deformation of the channel slope itself, but also detect the stability of the surrounding area of the channel slope. This is more conductive to identify the root causes of deformation and provide guidance for subsequent channel safety maintenance. The InSAR technique can be extended to other routes of the SNWD Project or other water conservancy projects.
Authors: Wang, Nan (1); Dong, Jie (2); Liao, Mingsheng (1); Zhang, Lu (1)Since the 21st century, the urbanization process of human living environment has been accelerating, and a large number of high-rise buildings and various bridge facilities have appeared. With the use of these building facilities, these building facilities begin to appear different degrees of settlement, deformation, crack and lift, which seriously affect the safety of the daily use of facilities. Therefore, the continuous monitoring of its stability is of great significance to prevent the public casualties and property losses caused by bridge collapse. Compared with the traditional contact monitoring means, a large amount of manpower and material resources are needed to carry out long-term monitoring of buildings and bridges (GPS, level, etc.). InSAR technology belongs to non-contact monitoring means. Monitoring bridges, high-rise buildings and other infrastructure by using InSAR technology has the advantages of all-day detection, high detection accuracy, low detection cost and long time series and fixed period observation. The deformation mechanism of buildings can be better obtained by combining InSAR technology with the structural and mechanical characteristics of buildings. This research project takes Xinlipu bridge over Hunhe river in Shenyang, Liaoning province as the research object. The data of the two satellites are 20 images from August 2015 to October 2016 provided by TerraSAR-X satellite and 55 images from February 7, 2015 to November 15, 2018 provided by Cosmo skymed satellite. The deformation results of Xinlipu bridge across Hunhe river are obtained by means of time series InSAR technique. Then the relationship between the deformation and temperature of each part of the bridge is analyzed, and the temperature model is obtained. The deformation of the bridge is the result of the long-term deflection and the joint action of the thermal dilation, so the separation of the thermal dilation and long-term deflection can help us better study the deformation characteristic mechanism of the bridge. At the same time, the relationship between the deformation and the material and the stress condition is analyzed. Finally, in view of the deformation visualization problem of the complex structure of the bridge, the data of deformation points of different specific parts of the bridge are separated and extracted, so as to better analyze the deformation law of different parts of the bridge. This study plans to use TerraSAR-X and Cosmo Skymed data in the same time period to cross verify the deformation monitoring results of Xinlipu Hunhe bridge. The research results show that the time-series SAR interferometry can effectively extract the long-term deflection data of long-span Bridges with high accuracy, and it can also further predict the future deformation of bridge buildings. Compared with the deformation data of different structural parts of the bridge, it is expected that different structural parts of the bridge have different sizes and change characteristics of trend deformation and temperature deformation, which are related to the specific material and structural characteristics of the bridge. Through the cross-validation of the monitoring data from TerraSAR-X and COSMO-SkyMed, it is estimated that the spatial distribution and magnitude of deformation between them have maintained a good consistency, which can confirm the reliability of the deformation results.
Authors: Wang, Xiaotian (1); Wei, Lianhuan (1); Zhao, Dong (2); Tolomei, Cristiano (3)Subsiding roads, shifting dikes, and bridges under stress: the impact of ground deformation on buildings, roads, and other civil infrastructure is substantial and its associated costs are growing each year. With the use of InSAR displacement time-series (persistent scatterers, PS and distributed scatterers, DS) such land deformation can be measured over a wide-area, in quasi real-time and with a spatial resolution down to millimeters. Man-made structures typically also have a higher likelihood of containing consistently strong radar reflections. Appropriately analyzed, InSAR displacement time-series (PS, DS) may be used to reconstruct a stress history for infrastructure objects such as bridges or dikes, thereby assisting operators in their assessment of the life-cycle of their objects over a wide area. Another use case would be to provide short-term warnings if ground motion or inferred stress levels exceed certain thresholds. Rijkswaterstaat, the Dutch agency for waterways and roads as well as flood protection and prevention, has therefore commissioned a case study to evaluate the use of processed InSAR displacement time-series (PS, DS) for their operations and maintenance planning. The aim of the study was to assess InSAR data quality and explore new ways to improve and exploit using proven methods of statistical analysis and inference. The InSAR data used for the case study covered three areas representing different infrastructure types: a wet environment with drainage sluices and wind turbines, a multi-modal transportation infrastructure (highways, railways, cycle paths), and a highway through peat land. The data originated from Ascending (110, 37, 139) and Descending (161, 88, 15) Sentinel-1 orbits. The conclusions from the study are threefold. First, we found that part of the processed InSAR time-series contains outliers that may complicate further analysis. These were addressed with different techniques for the time domain and the spatial domain. Wide-area outliers were identified using correlation matrices comparing different epochs which in some cases could be correlated with meteorological phenomena during data acquisition. Second, statistical time-series component analysis revealed that ground motion histories exhibit characteristics that exceed simple linear trends or oscillations. We show how statistical clustering techniques can be used to classify these characteristics and interpret them in the context of the particular areas studied. Third, different time-characteristics between Ascending and Descending orbits were observed. Based on preliminary and local analyses we speculate that these differences can be used to assess time-variant horizontal motion. That latter point is important because horizontal stress is at least as important as vertical stress for many pieces of infrastructure.
Authors: Riahi, Nima (1); Meftahi, Reda (1); Arikan, Mahmut (1); Houillon, Gaspard (2); Oyen, Anneleen (3)Transport infrastructure monitoring is an important task for land surveying authorities in Germany. Conventional methods such as levelling or GNSS to detect deformation at roads or bridges are time-consuming and limited in spatial distribution and extent. In our project SAR4Infra, we address these limitations using Interferometric Synthetic Aperture Radar (InSAR) to assess the health and stability of infrastructure over wide areas in Germany. The project's final aim is to develop an automatic InSAR processing chain that takes advantage of regular image acquisitions of Sentinel-1 and provide information on infrastructure health in near-real-time. For that purpose, high computational power is required for processing InSAR time series for large spatial coverage. The DLR offers the cloud computing platform CODE-DE to the German authorities in order to give access to the Copernicus data. It stores the data required for InSAR such as all national Sentinel-1 images, the orbit files and the DEM. Hence, processing on CODE-DE makes huge data transfer unnecessary, which is usually considered as bottleneck in near-real-time monitoring.Monitoring transport infrastructure using InSAR is particularly challenging as linear features are only covered by few pixels in the SAR images. The vegetation surrounding them often mitigates the signal as vegetation is more affected by phase decorrelation. We address these challenges within the Small BAseline Subset (SBAS) time series analysis framework by robust network inversion based on geodetic estimation theory. In a first step, the linear features are masked to reduce the effect of potentially decorrelated vegetation. Next, remaining unwrapping errors are corrected using hypothesis testing of the extended statistical model. Hereby also outliers are detected and discarded to improve the quality of the result. Finally, different measures for reliability and accuracy of the estimated phase time series are evaluated to assess the quality. The procedure is tested on both simulated and real datasets. The first study area is located in northeastern Germany close to Tribsees, where, in October 2017, the Autobahn 20 collapsed due to ground seepage. Its precursor deformation was monitored by levelling over several years. A second study area is located in midwest Germany close to Hambach open pit mine. The clear deformation signals at the surrounding Autobahn, probably due to mining, are used for validating the proposed processing chain.
Authors: Piter, Andreas (1); Haghshenas Haghighi, Mahmud (1); Motagh, Mahdi (2)The city of Recife is the capital of the state of Pernambuco in Brazil. Recife is built on the estuarine area of Capibaribe River, and has been developed during the last centuries by land reclamation from wetlands and mangroves. The rapid urban development in the last fifty years has put pressure on the groundwater resources of the area. To address the necessity of a sustainable management of the groundwater resources the municipality of Recife has enforced regulations on groundwater pumping and the use of old and new wells. The excessive groundwater exploitation especially during drought periods creates both actual and potential problems of lowering groundwater levels, water quality degradation, salinization and risk of land subsidence (Cabral et al., 2008). The rapid urban development on an estuarine area and the lowering of the groundwater table are both factors that can cause land subsidence. Persistent Scatterer Interferometry (PSI) analysis of Sentinel-1 time series was carried out to detect land subsidence in the city of Recife, Brazil. The dataset is composed of sixty-eight Sentinel-1A Interferometric Wide (IW) Single Look Complex (SLC) images of the time period April 2017 – September 2019. The images were acquired in descending orbit in VV (vertical transmitting, vertical receiving) polarization. The results of the PSI analysis show that in the city of Recife occur several ground subsidence areas. The largest land subsidence area occurs between the neighborhoods of Afogados, Torrŏes and Cordeiro. The subsidence rates in this area range from few mm/year up to -15 mm/year. This land subsidence area is a result of groundwater extraction and of subsidence processes in urbanized reclaimed lands. Similar but smaller land subsidence areas occur in few other localities in Recife. In some cases, subsidence with rates of up to -25 mm/year is noted in small zones where new buildings have been constructed in the last decade. This should be due to ground settlement processes, taking a long time due to the particular soils and geology of the locality. The study is a first contribution for further research on the land subsidence hazard in the city of Recife and the surrounding areas by means of satellite radar imagery. References Cabral, J.J.S.P., Farias, V.P., Sobral, M.D.C., de Paiva, A.L., Santos, R.B. (2008). Groundwater management in Recife. Water International, 33, 86-99.
Authors: Bedini, Enton (1,2)Differential Synthetic Aperture Radar (SAR) satellites acquire images of the Earth’s surface by emitting electromagnetic waves and analysing the reflected signal. A number of satellites are continuously circumnavigating the globe, collecting stacks of SAR images which have built up over time going as far back as 1992 over many parts of the world. From these images, changes in the reflected signal (phase change) are captured, and then analysed to identify surface deformation and other changes in the Earth’s surface over time. To overcome the inaccuracy in the measurements provided by conventional DInSAR, multi-image approaches where developed, like PSInSAR, SqueeSAR, SBAS, etc… which requires the availability of a long time series of SAR data. These techniques developed in the last decade are capable of providing millimetre precision, comparable to optical levelling, and a high spatial density of displacement measurements, over long periods of time without need for installing equipment or otherwise accessing the study area. Surface deformation monitoring provides unique data for observing and measuring the performance of hydrocarbon reservoirs, for enhanced oil recovery (EOR) and for carbon dioxide capture and sequestration (CCS). To this aim, radar interferometry (InSAR) and, in particular, have already proven to be valuable and cost-effective tools. For oil/ gas reservoir management, PS measurements make it feasible to examine the temporal and spatial pattern of long-term reservoir response to pumping and extraction, highlighting possible compaction affecting the area surrounding the reservoir and generating potential damage to local infrastructure. To monitor the temporal evolution of the deformation in Krechba gas reservoir, after sequestration of CO2, we used SAR data from two different periods. The first dataset, concerns images from Track 65 of Envisat satellite (period 2003-2010) and the second one is from Track 139 of Sentinel-1 A&B satellites (period 2016-2019). The studied area is located in central Algeria, a desert zone where temporal correlation is guaranteed, especially with the stable platform Sentinel-1. We could notice an important number of interferometric couples with a coherence greater or equal to 0.8, this case is not possible in northern Algeria where the presence of vegetation reduces the coherence. The studied area, is a gas exploitation platform, where the first experiment of CO2 injection in underground s exhausted reservoirs was undertaken. During a period of six years (2004-2010) 3.8 million tonnes of CO2 was injected In three horizontal well; KB-501, KB-502 and KB-503. Numerus studies was conducted using SAR data and GPS to monitor the stability of these wells after the injection operation. Research works using Envisat and Alos/Palsar data confirmed the presence of an important uplift in the three wells, so the injection operation was stopped in 2010. The same results were found in this study using Envisat data. A research work was conducted by our team, to monitor the stability of these wells using Sentinel-1 data. The images used for this purpose cover the period extending from June 2017 to September 2019. We could find that there is an apparent uplift in KB-502 and KB-503 wells but there was no recorded movement at or near KB-501 well. An ongoing research is focused on finding an explanation for these results.
Authors: Hasni, Kamel; Gourine, Bachir; Allal, Saddam-HousseynBig Sochi has always been a famous Russian resort and since the Olympics Sochi-2014 its popularity has been growing constantly. Monitoring of surface deformations which develop due to increasing anthropogenic loading is vital. Our lab in the Schmidt Institute of Physics of the Earth of the Russian Academy of Sciences has been studying different aspects of InSAR application in the Big Sochi region since 2012 (DmitrievP.N. et.al.,2012, Kiseleva et al., 2014; Mikhailov et al., 2014, Smolianinova, 2018,2019,2020). Adler district is one of the most favorable parts of the Big Sochi for InSAR monitoring because of high density of low rise residential development being perfect reflectors of radar signal, moderate and smooth topography (up to 300-400m) and snow free winters. For this region extremely slow surface movements (several cm/Y) are typical. We present surface deformation maps based on processing radar images of ALOS-1 (18 images 2007-2010), ENVISAT (12 images 2010-2012) and Sentinel-1 (more than 300 images 2015-2020 from 2 ascending and 2 descending orbits) SBAS ENVI SARscape software was used for processing. Surface deformations in the Adler district are mainly caused by two processes: subsidence and landsliding. On InSAR maps of surface movements landslides and subsidence areas here can be easily distinguished. Zero topography is a reason to consider a "hot spot" (where surface deformation exceeds a given threshold value) as a subsidence area and alternatively "hot spots" on hill slopes are most likely landslides. The main subsidence areas in Adler are located in the Imereti lowland which is situated on the southern slope of the Greater Caucasus Mountain Range between the Mzymta and Psou rivers. It stretches along the Black Sea coast for about 8 km and 0.8-2.0 km inland. Before construction of the Olympic-2014 facilities started here in 2007, the territory of the lowland was almost undeveloped, occupied by swamps, marshes and regularly floodable agricultural fields. The specific organic, so called “weak soils” are widely spread here, particularly in the central part of the lowland. Being wet these soils may lose their bearing capability and can be squeezed out from under constructions, thus, leading to their subsidence. Special technologies were involved to consolidate soils of basements. We applied InSAR to monitor efficiency of all the stabilizing measures. For the Imereti lowland we present maps of vertical displacements for all the datasets of images. Vertical movements were estimated as vertical projection of the LOS displacements assuming horizontal displacements in this flat area being negligibly small. We revealed seven main areas of subsidence in the Imereti lowland derived from S-1A datasets for 2015-2020. Maximum values of total subsidence since 2007 were fixed in the Central part of the lowland where thick lagoonal deposits with considerable content of “weak soils” are located according to geotechnical mapping. In this area subsidence for the 5 years period 2015-2020 is more than twice as big as for 2007-2012. Total subsidence here for the whole period 2007-2020 is more than 40 cm. Analyzing time-series we pointed out continuous subsidence for all the seven subsiding areas although rates of subsidence vary. There is no noticeable correlation between subsidence and precipitation. We revealed four areas of progressive subsidence and three stabilizing areas. These results help to estimate efficiency of drainage systems and stabilizing actions for the particular areas. Based on S1A 2015-2020 Vlos maps we revealed about 20 areas with active landslides from tens to few hundred meters in size. About half of them were not previously registered by field works because it is difficult to fix small displacements on territories occupied by private housing. It was found out that incorporating acquisitions from different orbits (43A and 145A as well as 123D and 21D) reduces layover and shadow areas and considerably improves detection of active landslides. We selected five places where landslides were clearly visible on Vlos maps to demonstrate detailed investigation of landslides movements. Rate of movements of all these landslides was not uniform during period 2007-2020. To compare activity of the landslides we computed down-the-slope rates (Vsd) where it was possible. We also averaged Vlos values in six month running window. The graphs of the averaged Vlos values demonstrate periodic behavior. Extreme values of the curves for different landslides sometimes are shifted within no more than 1-2 months. On the contrary to subsidence in the Imereti lowland, landslides activity depends upon precipitation. Extreme values of the mean Vlos for six month intervals match peaks of the graph of the total precipitation for six months intervals. Maximum displacement rates correspond to February-April, and minimum - August-October. Usually heavy but short rains do not trigger landslides activity while long lasting moderate rainfalls may be crucial. Our results demonstrate high efficiency of InSAR to detect landslides and subsiding areas and analyze dynamics of these movements in the highly populated Adler district of the Big Sochi, where extremely slow movements can hardly be fixed by field methods. InSAR demonstrated its efficiency to detect many new landslides and subsidence areas as well estimate activity of areas during the last 13 years, i.e. from starting construction of Olympic-2014 facilities up to now. References 1. Dmitriev P. N., Golubev V. I., Isaev Yu. S., Kiseleva E. A., Mikhailov V. O., Smolyaninova E. I. Nekotorye problemy obrabotki i interpretatsii dannykh sputnikovoi radarnoi interferometrii na primere monitoringa opolznevykh protsessov (On processing and interpretation of the SAR interferometry data in the case of the landslide monitoring), Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2012, Vol. 9,No. 2, pp. 130–142. 2. Kiseleva Е., Mikhailov V., Smolyaninova E., Dmitriev P., Golubev V., Timoshkina E., Hooper A., Samiei-Esfahany S., Hanssen R., 2014. PS-InSAR monitoring of landslide activity in the Black Sea coast of the Caucasus. Elsevier, Proceeding Technology, v.16, pp. 404-413. 3. Mikhailov V.O., Kiseleva E.A., Smol’yaninova E.I., Dmitriev P.N., Golubev V.I., Isaev Y.S., Dorokhin K.A., Timoshkina E.P., Khairetdinov S.A., 2014. Some problems of landslide monitoring using satellite radar imagery with different wavelengths: Case study of two landslides in the region of Greater Sochi. Izvestiya - Physics of the Solid Earth, Maik Nauka/Interperiodica Publishing (Russian Federation), v. 50, pp. 576-587. 4.Smolianinova E. I., Kiseleva E. A., Dmitriev P. N., Mikhailov V. O. O vozmozhnosti primeneniya RSA-interferometrii s ispol’zovaniem snimkov so sputnikov Sentinel-1 pri izuchenii opolznevoi aktivnosti v raione gornogo klastera Bol’shogo Sochi (On the possibility of using Sentinel-1 SAR interferometry to study landslide activity in the mountain cluster of the Big Sochi area), Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2018, Vol. 15, No. 4, pp. 103–111. 5. Smolianinova E. I., Kiseleva E. A., Mikhailov V. O., Primenenie RSA-interferometrii snimkov so sputnikov Sentinel-1 pri izuchenii oblastei aktivnykh deformatsii poverkhnosti v pribrezhnom raione Bol’shogo Sochi (Sentinel-1 InSAR for Investigation of Active Deformation Areas: Case Study of the coastal region of the Big Sochi), Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2019, Vol. 16, No. 5,pp. 147–155. 6. E. I. Smolianinova, V. O. Mikhailov, P. N. Dmitriev. Izuchenie i monitoring zon prosedaniya v Imeretinskoj nizmennosti (rajon Bol'shogo Sochi) metodom RSA‑interferometrii s ispol'zovaniem raznochastotnyh sputnikovyh radarnyh snimkov za period 2007–2019 gg. (Subsidence monitoring in the Imereti lowland (the Big Sochi region) using multifrequency InSAR data for 2007–2019) . Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2020, Vol. 17,No. 5, pp. 103-113.
Authors: Smolyaninova, Ekaterina; Mikhailov, ValentinDuring the last decades, radar interferometry techniques have proven to be effective for many environmental applications such as land subsidence monitoring, slope stability, ice sheet motion, earthquake, and volcanic activity studies. In particular, the persistent scatterers interferometry (PSI) technique, can provide the displacement history of coherent targets at mm level of accuracy, useful for land monitoring and hazard monitoring in many different cases. On the contrary, few applications are oriented to reliable structural health monitoring, sometimes prevented by coarse spatial and temporal resolution. The launch of X-band SAR COSMO SkyMed (CSK) constellation with the first next-generation satellite opened new possibilities in this field due to the medium to high spatial resolution, short revisit time and long baselines between satellite passes. In this work, CSK Stripmap HIMAGE data were processed with free and open-source software. SNAP (Sentinel Application Platform), provided by ESA, was used for the coregistration of products and the generation of the stack of interferograms. Successively, StaMPS (Stanford Method for Persistent Scatterers) software package allowed the selection of coherent persistent scatterers and the generation of single displacement histories. A linear tropospheric correction was then performed taking into account the phase and the topographic information with TRAIN (Toolbox for Reducing Atmospheric InSAR Noise) included in StaMPS. Finally, a shapefile containing the displacement history of all the PSs was produced and imported in the open-source QGIS for time series visualization. At the date, in the literature, there are no examples of a SNAP-StaMPS workflow applied to Cosmo SkyMed data. The present work proposes a procedure that combines CSK data with signal analysis techniques for the structural health monitoring of buildings and bridges. The identification of possible foundations settlement was also considered. To estimate the relative deformation of a structure with respect to the ground, a combination of displacement time series of both the ascending and descending orbits of several persistent scatters were used. First, a temporal alignment of time series from the orbits was performed employing a non-uniform resample technique and a multivariate autoregressive model. The latter model allows considering that displacements of a PS are related to those of neighbouring scatters and environmental conditions. The purpose of this step is the retrieval of the displacement value of a multitude of points at a given time and to reduce the noise level of such parameters introduced by measurement uncertainties and errors affecting the SAR processing. Then, under the hypothesis of a rigid body motion of the structure with respect to the ground, first order surface responses were calibrated for the ground and the structure from the predicted PS displacements. The rigid body hypothesis was verified by considering several subsets of PS and checking for possible changes in the response surface estimate. Finally, a numeric evaluation of the uncertainty in the determination of the relative structural motion was performed from initial positional error of targets and uncertainty affecting the displacement values at a given time. A Monte Carlo simulation was carried out adding a statistical scatter to parameters (coordinate of PS and displacement) and performing a statistical analysis of the results in terms of structural motion components.
Authors: Grassi, Francesca; Mancini, Francesco; Vincenzi, LorisBucharest is the largest city of Romania with a continuous growing population, commercial development and surface expansion. New residential centers have been built-up in the peri-urban areas of Bucharest in the last decade, this leading to changes of the infrastructures and of the urban environment. The urban expansion and city transformation could be detected and monitored with Sentinel 1 data by applying advanced differential SAR interferometry techniques. This study relies on multitemporal interferometry techniques (MTI) to characterize vertical and horizontal urban dynamics of the Bucharest city and peri-urban areas. A total of 108 Sentinel 1A and 1B descending data acquired between January 2016 and January 2021 have been used to assess ground deformation and buildings instabilities. The use and re-use of urban space (new building construction, intentional demolition, renovation projects), exploitation of groundwater resources (induced land subsidence), interactions between new foundations, superficial deposits and bedrock geology (settlement of recent buildings), ground and slope instability affecting settled buildings, and geohazard risk areas were investigated. The coherence threshold was set at 0.75 to catch the buildings’ instabilities from the city center. Deformation rates in the intervals -4 mm to -1.5 mm and 1.5 mm to 4 mm mark the transition from stability to instability (subsidence for the negative values and uplift for the positive ones). The negative deformation rates greater than -4 mm indicate the areas affected by subsidence, while the deformation rates higher than 4 mm indicate uplift. The deformation rate ranges from -8.75mm maximum subsidence value, to 5.75mm the maximum uplift value, the most common pattern being of stability, with values between -1.5mm and 1.5mm. The results indicate a slight uplift (up to 4 mm/year) long the Dambovita River while the areas located in N-E of the city center and in the former wetlands the subsidence is up to -8 mm/year. The new residence areas from southern and south-western part built in the monitoring period are quite stable. It is observed land cover changes in the COVID pandemic crisis: more buildings in the peri-urban areas. Local interactions with underlying geology and natural slope instability processes predominate in the central- western and north-eastern sectors. The general trend of the deformation rate is cross-validated through comparison with the results obtained in the previous studies.
Authors: Poenaru, Violeta (1); Radutu, Alina (1,2); Vlad Sandru, Maria Ioana (1); Nedelcu, Ion (1,2)Dam monitoring is an essential component of modern day safety programs in the world. Reports of structural failures that led to collapse of structures, causing significant material and economic losses and large number of fatalities date back from the construction of the first dams highlighting the need of monitoring. As a response to those catastrophic failures, improvements in design and changes in inspection program were developed. Structural Health Monitoring (SHM) is founded on the experimental identification of relevant dynamic properties of large structures, such as dams. In the last decades, civil structures have become even more complex leading to the use of in-built sensors for structures parameters monitoring. Moreover, they undergo phenomena that are only detectable by long-term observation due to the characteristics of parameters behavior, very difficult to observe by means of conventional sensors. The definition and objectives of SHM have been changing through time as technologies evolve, including as main components: real-time monitoring; in-service structures; and array or network of sensors to collect data and represent changes in the condition of a structure over time. Modern space-borne SAR sensors provide spatial resolutions of the order of a meter. Additionally high-resolution SAR data (TerraSAR-X, COSMO-SkyMed, etc.) allow monitoring small dams by a large number of pixels. SAR data are being tested for structural health monitoring (SHM) applications with very satisfactory results. The new generation of high-resolution radar imagery acquired by SAR sensors and the development of advanced Multi-temporal InSAR (MT-InSAR) algorithms such as Persistent Scatterer (PS) and Small BAseline Subsets (SBAS) methods, that retrieve deformation time series and velocity maps from a stack of SAR images acquired in different time over a region, have enhanced our capabilities in recent years in using MT-InSAR techniques for high precision deformation monitoring related to engineering infrastructures. In order to demonstrate the potential and reliability of MT-InSAR technique in dam monitoring, RemoDams Project is testing it in several real case studies in coordination with dam property and managers. The Arenoso dam (Córdoba, S Spain) is an embankment dam located in the Arenoso riverbed, a tributary of the Guadalquivir (Southern Iberian Peninsula) with a capacity of 167 hm3 and a basin of 405 km². The dam has a maximum height of 80 m, a crest length of 1.481 m and a coronation width of 11,30 m with impervious core, filters and rock fill shoulders. Its construction finished in May 2007 being used both for irrigation and energy production in the Guadalquivir basin. No X-band SAR sensor covers the area so the dam is being routinely monitoring with Sentinel-1 SAR data both in ascending and descending orbits providing an excellent source of auscultation. In this work we present a successfully case study of monitoring of a big embankment dam using MT-InSAR with Sentinel-1 data, discussing the integration of MT-InSAR in the monitoring program of this dam.
Authors: Ruiz-Armenteros, Antonio Miguel (1,2,3); Marchamalo-Sacristán, Miguel (4); Bakon, Matus (5,6); Lamas-Fernández, Francisco (7); Delgado Blasco, José Manuel (2); Sánchez-Ballesteros, Vanesa (8); Papco, Juraj (9); González-Rodrigo, Beatriz (10); Lazecky, Milan (11,12); Perissin, Daniele (13,14); Sousa, Joaquim J. (15,16)The scientific community had analysed many case studies which successfully investigated potential and limits of the Multi-Temporal InSAR (MTI) techniques for infrastructure stability monitoring. Despite many applications and improvements offered by new satellites such as Sentinel-1, standardized and operational procedures allowing for the effective use of InSAR data are lacking. An operational InSAR monitoring strategy using Sentinel-1 shall be capable of collecting the physical and dynamic parameters of the monitored areas continuously. Displacement mapping methodologies are still frequently based on the visual inspection of average line-of-sight velocities, as in the case of the ground stability areas detected in the framework of e.g., PanGeo, Terrafirma ESA GMES project or planned European Ground Motion Service (EU-GMS). However, the spatial distribution of velocities as visible from a standard point cloud map is often scattered and affected by noises and offers no reliable information on the probability of movement for intervening areas. Moreover, the extraction of useful information from the MTI techniques is sometimes difficult due to a large number of processed persistent scattering (PS) points, thereby entailing long interpretation times. For a better disentangling between real and false deformation signals, it is necessary to develop an effective risk identification methods and event recognition analysis to adopt appropriate surveillance and risk mitigation strategies. Our post-processing procedure accounts for spatial and statistical dependency among observations in order to extract useful information over low coherent areas. Low values of temporal coherence are usually associated with limitations and constraints of InSAR technology. However, PS with lower temporal coherence might also represent areas where deformation processes progress rapidly, in an unexpected way or in a way that was not assumed in applied functional models within estimation procedure. These would remain undetected, especially over wide areas monitored by Sentinel-1, as their coherence is often weakened and by the rules of standard thresholding procedure they would be discarded. In some cases, reliable estimates over low- coherent areas can be obtained for example by extending the standard linear model, by re-assessing solution space boundaries or by constructing complex models which better describe real deformation scenarios. Wrong model drops the estimate of ensemble coherence, biases the estimate of parameters and increases chances of PS candidates rejection, even though it is in fact a persistent scatterer (e.g. undergoing strong non-linear deformation, seasonal movement, thermal expansion, effect of water level fluctuation, sudden changes like in the case of landslide activation, etc.). After detecting anomalous behaviour, the processing strategy of expert users can be refined and real deformation profiles could be extracted for these areas. On the contrary, if the potential deformation threat has been identified, the non-expert user should be aware of such a situation. The remotIO (Retrieval of Motions and Potential Deformation threats) system has been developed under European Space Agency PECS (Plan for European Cooperating States) project in Slovakia and is currently capable of providing autonomous updates of PSInSAR (Persistent Scatterer InSAR) displacement maps and added-value products for on-time situational awareness and easier interpretation. The results presented in this research are demonstrating the ability of a remotIO system to detect and disentangle between different deformation processes over low-coherent areas and in most recent monitoring periods by triggering a series of alerts, alarms and risk ratings over monitored sites. Many structures and objects inside the areas where deformation processes are confirmed by MTI analysis, would have been left without any information while relying only on classical denoising or filtering procedures such as traditional coherence thresholding. Our systematic approach has proven to be useful in increasing the point densities in final MTI results and in detecting sectors with potential ground deformation threats over low-coherent areas.
Authors: Bakon, Matus (1,2); Czikhardt, Richard (3); Papco, Juraj (3); Barlak, Jan (1); Rovnak, Martin (2); Adamisin, Peter (2); Perissin, Daniele (4); Ruiz-Armenteros, Antonio Miguel (5); Sousa, Joaquim Joao (6,7); Liscak, Pavel (8)The Meinong earthquake struck southwestern Taiwan on 6 February 2016. Serious damages were observed in Tainan City, which is about 35 km west of the epicenter, and the earthquake caused about 117 deaths, 551 injuries, 10 collapsed buildings and 247 seriously damaged buildings. Many geodetic tools, such as precise leveling, GPS, and InSAR, have been applied successfully to measure the coseismic surface deformation. However, to monitor the postseismic surface deformation is difficult due to its slight displacement and the spatiotemporal limitations of the geodetic instruments. It is easy to ignore the potential hazard of surface deformation after earthquake event. This study aims to measure the postseismic surface deformation after the Meinong earthquake. We collect 182 Sentinel-1 images and divide into two datasets. One dataset is before earthquake (26 image pairs in the ascending orbit and 18 image pairs in the descending orbit) between Oct. 2014 and Feb. 2016. another is after earthquake (78 image pairs in the ascending orbit and 58 image pairs in the descending orbit) between Feb. 2016 and Dec. 2018. The SNAP software processes the interferometry and the StaMPS software calculates the line-of-sight (LOS) mean velocity of persistent scatterers. Our results indicate that the local displacement rates vary with the extended linear of the faults and structures after Meinong earthquake. The maximum difference of LOS displacement rate reached near 50 mm/yr before and after the earthquake. To estimate the displacements in the east-west (E-W) and up-down (U-D) directions, we combine the LOS results of both orbits and correct them with the GPS data from the IES, Academia Sinica. The maximum difference of E-W displacement rate reaches near -30 mm/yr crossing the Chishan Fault, and the maximum difference of U-D displacement rate is higher than 30 mm/yr crossing the Houchiali Fault and Hsiaokangshan Fault. The both directional displacements are helpful to assess the safety of the infrastructures such as freeway, railway, and high speed rail (HSR). Our results show that two linear discontinuities, which are located on the structural linears, divid the E-W displacements along the railway and HSR into three segments with different rates after earthquake. The norther part moves toward east, middle part is stable, and southern part moves toward west. The U-D displacements along the infrastructures are more influenced by the active faults than before earthquake event. The obvious differences of deformation behaviors are important to realize the faults mechanism and are beneficial for the postseismic hazard assessment.
Authors: Lu, Chih-Heng (1); Chuang, Ray Y. (2,3); Chiang, Ping-Chen (2); Yen, Jiun-Yee (4)Ground surface deformation measurements related to industrial activity have historically been performed using terrestrial geodetic surveying methods, such as optical levels, theodolites, EDM units and total stations. For the last couple of decades, two satellite-based techniques have significantly increased monitoring capabilities: Interferometric Synthetic Aperture Radar (InSAR) and Global Navigation Satellite System (GNSS). Technical and algorithmic InSAR advances have been considerable – from Differential to Multi-temporal (PS-InSAR, SBAS-InSAR and their combination), improved spatial and temporal resolutions, precise knowledge of satellite orbits, weather models and atmospheric measurements for bias mitigation have improved the reliability of the technique. As for GNSS, increase of the number of satellite constellations, precise knowledge of satellite orbits, atmospheric modelling, multipath reduction, and the advent of wide area networks of actives stations (CORS) have made it possible to achieve very accurate deformation rates in the three dimensions of space and over large areas/regions. Here we present various different industrial contexts where InSAR – standalone or combined with terrestrial geodetic measurements – has significantly improved the outcomes of (industrial) asset management, having leveraged the capabilities of civil, geotechnical and geological engineering. These examples in the Netherlands, Germany and Russia highlight the specific contribution of InSAR to civil and geotechnical engineering, as settlement models and analysis of mechanical strength of infrastructure (e.g. embankments) require dense spatial and temporal deformation measurements. Additionally, geotechnical analyses of infrastructure have extreme spatial localization. InSAR provides a broader spatial overview and allows better contextualization and mitigation of potential hazards. The examples are: (i) A highway extension project in the Netherlands (Almere): InSAR allowed both optimizing the distribution and location of the settlement plates network and densifying the embankment settlement measurements. (ii) A study of the behaviour of the tailings dams of mine waste reservoirs in Germany (Bernburg), where high frequency measurements highlighted the refilling and compaction cycles of this waste. (iii) The measurement of permafrost freeze-thaw cycles on LNG infrastructure, in Siberia (Yamal). Despite adverse conditions (seasonal snow coverage), the use of cleverly selected sub-time series provided clear settlement results. The third example also showcases the shortcomings of InSAR when used in difficult backgrounds. Coherence in time or in space cannot always be guaranteed, and in such cases, the power of InSAR can be unleashed only when combined with terrestrial geodetic measurements. The examples also highlight some additional challenges, such as (1) possible overestimation of uncertainties, especially due atmospheric artifacts, (2) difficulty of comparing measurement techniques from different viewing angles (1D – line of sight for each geometry of InSAR, 1D – vertical for levelling, 3D for GNSS) and (3) different spatial and temporal resolutions. To tackle these issues, Fugro has developed an innovative data integration software (GEOMON) combining InSAR and terrestrial geodetic data through customized weighting, spatiotemporal sampling, and rigorous regression-kriging algorithms, in order to handle various types of data acquired at different locations and periods. Such combination of InSAR and terrestrial geodetic measurements yields a resulting ground surface deformation field with a higher (and better-defined) accuracy and robustness. Through additional examples related to mining-induced subsidence in France, we show how InSAR augmentation can be used for geological engineering in decorrelating areas. Brine extraction and water injection cycles have caused subsidence for decades over two subterranean mines, in eastern France (Saint Nicolas de Port) and southern France (Vauvert).In these cases, multi-temporal InSAR provided dense PS coverage at short intervals in some areas, and very sparse in others – hence the exact deformation magnitude and location were questionable. Precise geodetic levelling campaign was performed once a year and along a few profiles to comply with legal requirements, but inadequate for geological analysis. By combining InSAR and levelling through GEOMON, an improved deformation field was derived, taking advantage of levelling reliability to ascertain the exact ground surface deformation magnitude, and much higher spatial and temporal measurement resolution through InSAR. This methodology allowed (i) constraining (i.e. calibrating) the InSAR near-vertical component resulting from ascending/descending geometries gridded on absolute deformation measurements, (ii) disclosing the actual extent, magnitude and spatial variation of the deformation, notably extending the knowledge of the deformation beyond the profiles derived from levelling, (iii) describing the deformation variation over time, (iv) highlighting not only the annual deformation trend but also transient deformation, notably due to the complex local hydrology, (v) help defining the seismic survey pattern and fill the gap between the seismic survey results (not resolving the top layer) and the surface, and finally (vi) correlating the resulting deformation maps to geophysical data to finetune the geologic model in relation to anthropogenic activity and natural phenomena (groundwater hydrology). Inversion based on Mogi and Okada models further helped in understanding the impact of rheological and geological features on surface measurements and the reservoir dynamics.
Authors: Doucet, Samuel (1); Carme, Jean-Louis (1); Mahapatra, Pooja (2)By processing SAR images using the method of permanent scatterers (PS) we get very accurate data about the close-to-vertical displacements of individual points, in a millimeter range. The result is a time series of point data showing the height changes of the reflector. This method is mainly used to monitor the surface of the Earth, where there could be an increase or decrease of land by anthropogenic or natural impact, such as the monitoring of underground gas or oil reservoirs as in Xuejun et al. (2018), Haghighi et al. (2017), Singhroy et al. (2014), Granda et al. (2012). Also, the method can be used to monitor infrastructure such as tunnel or bridges, Bakon et al. (2014). This PS results are affected by noise making the interpretation difficult. Therefore, smoothing must be applied to the data. In our case, we used a floating average to smooth out the time series for each PS point over a gas reservoir. By smoothing the data, it is possible to determine the trend or pattern (periodicity, sometimes referred to as the seasonality of data). After smoothing the data, we found seasonality in the data, which corresponds to the underground gas storage processes in the monitored area. As our time series data was not regularly sampled, we had to verify the presence of the seasonal changes using the LombScargle periodogram (LSP). LSP has determined that there are points in the data that show annual periodicity. Subsequently, a reference curve was created which corresponded to the gas storage processes. Injection starts in the beginning of May, withdrawal in the beginning of November. This curve was then correlated to the calculated smoothed curves of the individual points. The resulting correlation coefficient divides the points into three categories, namely a positive correlation coefficient (0.3-1) where the points have movement identical to the periodic processes in the underground gas storage, insignificant (0.3-0.3) and the category with a negative correlation coefficient (-0.3 - -1) where the points show the movement opposite to the periodic processes in the underground gas storage. In conclusion, we can divide points into few categories for the next analysis or visualization. In the first step, we can reduce the number of points with LSP on points with annual frequency. In the second step, we get the correlation coefficient of points (time series). We know their moving against the reference curve. We can find an anomaly in moving. For the next work we can model or interpolate missing data in time series and want to use Fast Fourier Transform (FFT) for find new features in time series. The data was prepared by Lazecky (2020). Reference: BAKON, M. et al. (2014) Infrastructure Non-Linear Deformation Monitoring Via Satellite Radar Interferometry, 2014, ScienceDirect, Elsevier GRANDA, J. et al. (2012): Ground motion monitoring with radar technology – Case study of SAGD operations of Leismer: Results of ground motion monitoring during first year of steam injection and preliminary comparison with reservoir performance data, World Heavy oil congress, Aberdeen, Škótsko, 2012 HAGHIGHI, H., et al. (2017): Sentinel-1 InSAR over Germany: Large-Scale Interferometry, Atmospheric Effects, and Ground Deformation Mapping. - ZfV: Zeitschrift für Geodäsie, Geoinformation und Landmanagement, 2017, 4, pp. 245—256. DOI: http://doi.org/10.12902/zfv-0174-2017 Lazecky, M., Hlavacova, I., Hatton, E., Gonzalez, P. J., Martinovic, J. (2020): Displacements Monitoring over Czechia by IT4S1 System for Automatized Interferometric Measurements using Sentinel-1 Data, MDPI Remote Sensing SINGHROY, V., et al. (2014): InSAR monitoring of surface deformation induced by steam injection in the Athabasca oil sands, Canada. In: 2014 IEEE Geoscience and Remote Sensing Symposium. IEEE, 2014, 2014, s. 4796-4799 [cit. 2019-03-03]. DOI: 10.1109/IGARSS.2014.6947567. ISBN 978-1-4799-5775-0. Dostupné z: http://ieeexplore.ieee.org/document/6947567/ Xuejun, Q., et al. (2018): Crustal Deformation in the Hutubi Underground Gas Storage Site in China Observed by GPS and InSAR Measurements. Seismological Research Letters. 89. DOI: 10.1785/0220170221.
Authors: Struhár, Juraj (1); Rapant, Petr (1); Lazecký, Milan (2,3)Many coastal cities in Southeast Asia are increasingly threatened by flooding through sea-level rise, but also extreme rainfall and flash floods from the hinterlands. Surface subsidence additionally increases the flood risks, especially in river delta areas. However, the drivers and consequences of subsidence are little analyzed and understood in many cases. This study investigates the relationship between surface subsidence and land-use in the city of Hue, Central Vietnam. Time-series of Sentinel-1 imagery are utilizing a persistent scatterer approach. This resulted in deformation rates between -25 and +10 millimeters per year. While large parts of the urban area show a stable condition, there are concentrated patterns of high subsidence in the outskirts of the city. In order to identify drivers of these subsidence patterns, the deformation rates are compared to two external datasets: A map of soil types showing the most important soil types in the study area. An analysis of variance (ANOVA) was calculated to confirm a significant difference between the soil types at a 0.001 confidence level, with Plinthic Acrisols as the soil type with the largest negative average surface velocity. This soil type is present in areas of rice cultivation. As the persistent scatterers are retrieved from solid targets, such as roads or buildings, the results indicate that these parts of the city show a higher risk for subsidence because of the subsurface water from the rice fields intrudes the neighboring residential areas. Secondly, a multi-temporal classification of urban areas was conducted based on Landsat imagery to determine the extent of the city for the periods 1980-1999, 2000-2009, and 2010 to 2016. Results indicate higher deformation rates for later stages of urbanization, again at a tested confidence level of 0.001. This means that parts of the city which were built between 2010 and 2016 are more vulnerable to surface subsidence, most likely because they are at the peripheral areas of the city where the ground is not as solid as in the center of the city. Results of this study help to understand drivers and hotspots of surface subsidence for the development of adapted strategies of flood risk reduction.
Authors: Braun, Andreas (1); Bachofer, Felix (2)Geodetic monitoring of surface displacements in urban areas can be used for risk analyses of infrastructural damage and to help understand the underlying geological conditions. The applicability of the Persistent Scatterer Interferometry (PSI) approach for this matter has been demonstrated frequently. The standard PSI approach constrains the displacement analysis to scatterers which are coherent over the whole considered time series (PS). Scatterers which are only temporarily persistent are neglected in the analysis. Various approaches have been developed to also detect such Temporarily Persistent Scatterers (TPS), in order to enhance the PS density and for the purpose of change detection. In this work, we incorporate TPS into an existing PSI algorithm in order to systematically analyze the initial settlement of recently constructed buildings and their surroundings. We built on the TPS detection method by Ansari et al. (2014), which is based on statistical properties of the SAR amplitude time series, and jointly estimate the height and displacement rate of PS and TPS candidates using a multi-stack small baseline PSI approach. After parameter estimation and selection of the final PS and TPS, their single-master phases are jointly unwrapped, using an adapted version of the unwrapping algorithm implemented in the StaMPS software package (Hooper et al., 2012), to receive their displacement time series. We restrict the TPS analysis to appearing TPS, i.e. pixels which change from a distributed scatterer to a PS behaviour, often caused by building constructions in urban areas. The approach can, however, be easily extended to also include other kinds of TPS. We found that the initially estimated emerging date of appearing TPS, defined as the start of the coherent epoch of the pixel, from pure amplitude statistics can be inaccurate. For example, the start of the construction phase of a building is sometimes detected as emerging date of TPS while the true emerging date is the date of construction completion. To overcome such deficiencies, we modified and enhanced the original approach of Ansari et al. (2014) in a way that we iteratively update the emerging date estimation of TPS candidates during the parameter estimation based on a phase coherence estimate in a moving time window. We cluster identified TPS based on their location and emerging date using an unsupervised clustering algorithm, as done in Yang et al. (2017). The purpose is to automatically identify newly constructed or reworked buildings, since we expect these to trigger several TPS which are in close neighborhood and share either a similar emerging date or a continuous sequence of emerging dates. Besides the implicit change detection, this enables a systematic analysis of the displacement time series of these buildings, e.g. to study initial settlement of the buildings and their surrounding. We apply the proposed method to a stack of Sentinel-1 SAR images acquired between 2017 and 2020 over the Vietnamese Cities of Ca Mau and Can Tho, located in the Mekong Delta. For the Mekong Delta, whose mean elevation is less than 2 m above sea level, subsidence rates of up to several centimeters per year have been reported recently. Large differential subsidence rates between infrastructure with pile foundations and the surrounding ground surface have been observed across the considered cities. We show the initial displacement time series of various recently constructed buildings and their surroundings, identified by TPS clustering, in the context of the city-wide deformation patterns. Analyses of possible initial settlement of built-up areas will help in studies of causes of land subsidence, in order to separate initial settlement due to consolidation of surface soil layers from other contributions to subsidence, e.g. groundwater extraction. Ansari, H., Adam, N., & Brcic, R. (2014). Amplitude time series analysis in detection of persistent and temporal coherent scatterers. In 2014 IEEE Geoscience and Remote Sensing Symposium (pp. 2213-2216). IEEE. Hooper A., Bekaert D., Spaans K., Arikan M. (2012). Recent advances in SAR interferometry time series analysis for measuring crustal deformation, Tectonophysics, 514-517, pp.1-13. doi:10.1016/j.tecto.2011.10.013 Yang, C. H., Pang, Y., & Soergel, U. (2017). Monitoring of building construction by 4D change detection using multi-temporal SAR images. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 4, 35.
Authors: Dörr, Nils; Schenk, Andreas; Hinz, StefanHigh-resolution geodetic measurements of crustal deformation from InSAR have the potential to provide crucial constraints on a region’s tectonics, geodynamics and seismic hazard. In COMET, we have extracted a crustal velocity field for the Alpine-Himalayan Seismic Belt (AHSB) from >300,000 Sentinel-1 interferograms and GNSS. The result provides 3-D velocities in low resolution and east-west and sub-vertical velocity fields in high resolution. The sub-vertical velocities, which also include a small component of north-south motion, are dominated by non-tectonic deformation, such as subsidence due to water extraction. The east-west velocity field, however, reveals the tectonics of the AHSB with unprecedented sharpness. The approach described above only provides high-resolution constraints on horizontal displacement in the east-west direction, with the north-south component provided by low-resolution GNSS measurements. Sentinel-1 does, in fact, also have the potential to provide measurements that are sensitive to north-south motion, through exploitation of the burst overlap areas produced by the TOPS acquisition mode. These along-track measurements have lower precision than line-of-sight InSAR and are more effected by ionospheric noise, but have the advantage of being almost insensitive to tropospheric noise. We present a time series approach to tease out the subtle along-track signals associated with interseismic strain. Our approach includes improvements to the mitigation of ionospheric noise and we also investigate different filtering approaches to optimize the reduction of decorrelation noise. In contrast to the relative measurements of line-of-sight InSAR, these along-track measurements are automatically provided in a global reference frame. We present results from five years of data for the West-Lut Fault in eastern Iran and the Chaman Fault in Pakistan and Afghanistan. Our results agree well with independent GNSS measurements; however, the denser coverage of the technique allows us to also detect the variation in slip rate along the faults. Finally, we demonstrate the improvement in the resolution of estimated horizontal strain rates when including these along-track measurements in addition to conventional line-of-sight InSAR measurements.
Authors: Hooper, Andy (1); Piromthong, Pawan (1); Lazecky, Milan (1); Weiss, Jonathan (2); Elliott, John (1); Wright, Tim (1); Maghsoudi, Yasser (1)Inverting seismic and geodetic data for earthquake slip distribution is fundamental for understanding earthquake mechanics and assessing future fault failures. A well-resolved slip model may be used for calculating the stress changes in the crust and determining the probability of earthquake occurrence on adjacent faults. Often, however, the slip inversions are non-unique and/or ill-conditioned. The problem is said to be non-unique when it results in a suite of very different slip models that fit the data similarly well, and ill-conditioned when small changes in the data space cause large changes in the solution space. The non-uniqueness of the problem may be addressed by combining different types of datasets, such as ground displacements, interferometric synthetic-aperture radar (InSAR) and GPS measurements, seismic data (seismicity, strong motion and teleseismic). The ill-conditioning may be addressed by reconditioning the inverse problem in a manner that reduces the solution sensitivity to noise and data processing errors. Implementing both approaches eliminates the need for a-priori regularization, and results in an improved slip model. For a given fault geometry, the objective of slip inversion is to find a slip distribution that minimizes the misfit between the modeled and observed ground displacements. The sensitivity of the solution to small changes in the data may be quantified using the condition number (CN) of the elastic kernel (G), CN = σ1 / σn, with n being the dimension of G, and σ1 > ... > σn > 0 being its singular values. A smaller CN indicates a more stable inverse problem. The stability of the inverse problem is completely determined by the properties of the matrix G, whose structure is determined by the discretization of the data and model spaces. Thus, how the InSAR data is down-sampled and the model is discretized are of great importance. In this study, we will present a new approach for down-sampling InSAR data and discretizing the fault plane which optimizes the inversion stability. First, we examine our new discretization method by inverting the well-studied M 6.3, 6 April 2009 L’Aquila (Italy) earthquake, jointly inverting Envisat InSAR data alongside GPS displacement results. We present numerical Monte Carlo simulations that emphasize the advantage of the new down-sampling method for inverting noisy data and show better agreement of the slip model with a precisely relocated aftershock sequence. Secondly, we invert for the slip distribution of the 2019 Ridgecrest earthquake pair which ruptured a previously unknown orthogonal fault system in the Eastern California Shear Zone. The stronger of the two, an Mw 7.1 earthquake that occurred on July 6, 2019, was preceded by an Mw 6.4 foreshock that occurred 34 hours earlier. We implement our new discretization technique on both Sentinel-1 InSAR data and Planet Labs optical imagery offset tracking. Then, we infer a distinct final slip distribution for the two earthquakes via joint inversion of the InSAR, optical imagery and GPS measurements. The two examples of our new approach show that the fault slip distributions in the inverted models are less affected by changes in the data space and allow us to impose smaller constraints on the inverse problem to get better agreement between different data sets.
Authors: Magen, Yohai (1,2); Baer, Gidon (2); Inbal, Asaf (1); Ziv, Alon (1); Hollingsworth, James (3)The temporal sampling of SAR acquisitions for interferometry is at the order of days for most space-borne SAR missions. For fast processes e.g. at work in earthquake sequences such sampling can be too coarse to image surface effects of individual large earthquakes in a sequence. The consequence is that sequence-spanning interferograms measure the cumulative surface displacement of several earthquakes, which complicates analyses of the individual deformation sources and likely results in some not well constrained source characteristics. A prominent recent example for such a sequence has been the 2019 Ridgecrest earthquake sequence in California where a Mw6.4 earthquake was followed by a Mw7.1 earthquake a day later. Similar situations are known for many other earthquake sequences or earthquake doublets (two similarly large earthquakes). Recordings of seismic waves generated by earthquakes on the other hand are sampled with frequencies well above 1 Hz. Given an inter-earthquake time of some tens of minutes, well separated wave trains of the corresponding large earthquakes can be measured globally. For recordings at regional distances (
Authors: Sudhaus, Henriette (1); Ridderbusch, Jan (2); Steinberg, Andreas (1,3); Donner, Stefanie (3); Ghods, Abdolreza (4)With the advancement of Synthetic Aperture Radar (SAR) satellite missions, Interferometric SAR (InSAR) has become an established method to measure the Earth’s surface deformation caused by earthquakes, greatly improving our ability to observe active tectonic processes. Unfortunately, due to the dominant error sources from decorrelation and atmospheric noise within InSAR data, only earthquakes above a certain size or that are shallow enough can be observed. This circumstance of low signal-to-noise ratio makes InSAR for earthquake studies challenging, especially for small earthquakes (Mw 5.0-6.5) due to their weak signals. To enhance the potential of retrieving InSAR earthquake observations, here we apply InSAR time series analysis of Sentinel-1 data and use three recent earthquakes (Mw 5.6-6.3, ~10 km depth) in south-western Iran as case studies. Whilst the larger of these is more readily identifiable in single interferograms, we use a sequence comprising a range of magnitudes to test the validity and improvement of our approach for decreasing earthquake sizes. We find that the coseismic displacement signals of these earthquakes, which might not be discernible within single interferograms, are better resolved using our approach. Taking the dominant postseismic deformation of the large 2017 Mw 7.3 Darbandikhan earthquake into consideration, we reconstruct the enhanced coseismic interferogram by fitting models for the seismic cycle surface displacement to time series, which reduces random noise and improves the visibility of the earthquake signals under study. In addition, we show a case that two moderate magnitude earthquakes which happened close together in time and space (the Mw 5.6 and Mw 6.3 earthquakes happened at Nov 2018 and Jan 2019 with 42 days and 30 km separation) can be told apart by our approach. Finally, we use seismological observations from the United States Geological Survey (USGS) catalog and recent geodetic solutions to validate the effectiveness of our approach. We find that the reconstructed coseismic interferograms achieve better constrained and seismologically consistent modelling results compared to that using single interferograms. Our work suggests that time series is an effective tool for enhancing the InSAR resolving power for earthquake studies and may also help us study the postseismic processes for large earthquakes.
Authors: Liu, Fei; Elliott, John; Craig, Timothy; Hooper, Andy; Wright, TimEarthquakes are one of the natural hazards that can occur worldwide. One of the key methods for earthquakes’ study is the Differential Synthetic Aperture Radar Interferometry (DInSAR) that in the last 28 years demonstrated to be highly effective for the quantitative measurement of the Earth’s surface deformation, with centimetres to millimetres accuracy [1]. Nowadays, thanks to the advent of the Copernicus Sentinel-1 (S1) [2] mission, the radar Earth Observation (EO) scenario is moving from the historical analysis to operational functionalities. Indeed, the S1 mission, by using the Terrain Observation by Progressive Scans (TOPS) [3] technique, has been designed with the specific aim of natural hazards monitoring via SAR Interferometry, guaranteeing a very large coverage of the illuminated scene (250km of swath). These characteristics sum up with the free & open access data policy, the global scale acquisition plan and the high system reliability thus providing a set of peculiarities that make S1 a game changer in the context of operational EO scenario. By taking benefit of the S1 characteristics, an unsupervised and cloud-based tool for the automatic generation of co-seismic ground displacement maps has been recently proposed [4]. This tool is automatically triggered by the most significant earthquakes (i.e., those with characteristics above selected thresholds) published on the online catalogues (e.g., USGS and INGV) [5-6]. The tool allows the unsupervised generation of pre-event, co-event and post-event interferograms and displacement maps with all available S1 acquisitions taken over the epicentre area, in a time period of one month before and one month after the studied seismic event. The implemented procedure starts from the earthquake information retrieval up to the final displacement map generation and publication, through the download of the needed S1 scenes and the instantiation of the computing resources. Although it was conceived for generating displacement maps as an operational service for the Civil Protection department, the implemented tool has also been applied to study historical events imaged by the S1 data. This allowed us to generate a global archive of DInSAR-based co-seismic displacement maps (Figure 1). The archive has been generated by automatically processing all the Copernicus Sentinel-1 data spanning about 300 significant (at least Mw > 5.5) earthquakes all over the Earth [5]. An empirical relation between magnitude and depth was considered to limit the study to those earthquakes that can likely induce ground deformation. The DInSAR processing has been carried out within the Amazon Web Services Elastic Cloud Compute infrastructure (AWS-EC2) [7]. In particular, the r5.12xlarge instance has been used to also guarantee the processing in parallel of different tracks (relevant to the same earthquake), according to [8]. This strategy allows the processing of a huge amount of data with a significant reduction of the total elapsed time, preserving the precision and accuracy of the generated interferometric results. The obtained results (Interferograms, DInSAR displacement maps, …) are then stored on an AWS S3 bucket for being archived and further exploited. The output data are provided according to the file formats defined within the European Plate Observing System (EPOS) [9] research infrastructure. In particular, the products are provided in GeoTIFF, while metadata follow the ISO 19115, and will be made openly available through the EPOS portal, to be investigated and interpreted by the scientific community. Although the tool has been implemented within an AWS environment, it was conceived to be highly portable and can be integrated within other Cloud Computing environments, such as the Copernicus DIAS platforms. The generated DInSAR products, whose an example is provided in the included Figure, can be used in operational way by the Civil Protection authorities to have a quick response during seismic crisis. Moreover, they can provide the scientific community with a large catalogue of interferograms for investigating the dynamics of surface deformation in the seismic zones around the Earth. Indeed, these products can be further integrated with other parameters that allow understanding the seismic source and the behaviour of the faults interested by the deformation process. Acknowledgements This work has been supported by the 2019-2021 CNR-IREA and Italian Civil Protection Department agreement, the EPOS-IP and EPOS-SP projects of the European Union Horizon 2020 R&I program (grant agreement 676564 and 871121), the I-AMICA (PONa3_00363) project and the EPOS-Italia IREA-INGV agreement. Bibliography [1] D. Massonnet et al., “The displacement field of the Landers earthquake mapped by radar interferometry,” Nature, vol. 364, no.6433, pp. 138– 142, Jul. 1993. [2] R. Torres, P. Snoeij, D. Geudtner, D. Bibby, M. Davidson, E. Attema, P. Potin, B. Rommen, N. Floury, M.Brown, I. Traver, P. Deghaye, B. Duesmann, B. Rosich, N. Miranda, C. Bruno, M. L’Abbate, R. Croci, A. Pietropaolo, M. Huchler, and F. Rostan, 2012. GMES Sentinel-1 mission. Remote Sens. Environ., 120, 9-24, 2012 [3] F. De Zan,A.M. Monti Guarnieri, “TOPSAR: terrain observation by progressive scans,” IEEE Trans. Geosci. Remote Sens. 44 (9), 2352–2360, 2006. [4] F. Monterroso et al., "A Global Archive of Coseismic DInSAR Products Obtained Through Unsupervised Sentinel-1 Data Processing", Remote Sensing 12 (19), 3189, 2020. [5] USGS. United States Geological Survey, Earthquakes hazard program, https://earthquake.usgs.gov/earthquakes/feed [6] INGV, National Institute of Geophysics and Volcanology, http://cnt.rm.ingv.it/feed/atom/all_week [7] Amazon Virtual Private Cloud VPC. [Online]. Available: https://aws.amazon.com/it/documentation/vpc/ [8] M. Manunta et al., "The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment," in IEEE Transactions on Geoscience and Remote Sensing, vol. 57, no. 9, pp. 6259-6281, Sept. 2019. [9] EPOS, European Plate Observing System, [Online]. Available https://www.epos-ip.org/tcs/satellite-data
Authors: Monterroso, Fernando (1,2); de Luca, Claudio (1); Bonano, Manuela (1); Lanari, Riccardo (1); Manunta, Michele (1); Manzo, Mariarosaria (1); De Novellis, Vincenzo (1); Valerio, Emanuela (1); Onorato, Giovanni (1); Zinno, Ivana (1); Casu, Francesco (1)The oblique convergence between the Philippine Sea Plate and the Eurasian Plate results in strain partitioning in the Philippine archipelago, primarily along subduction zones and the 1,200-km long left-lateral strike-slip Philippine Fault. While the latter is known to exhibit both seismic and aseismic slip, detailed analysis of the relationship between these two modes of slip has been missing. Several significant large earthquakes have occurred on the northern and southern segments of the fault, including the 1990 magnitude 7.7 Luzon earthquake, while the central section is known to host smaller events (magnitude 5-6) and aseismic slip. Studies based on GPS surveys have suggested that the fault is creeping up to 3.5 cm/yr on the island of Leyte, similar to long-term slip rate estimates. The lack of apparent slip deficit has led to the understanding that the fault in Leyte is aseismic; however, a significant Mw 6.5 earthquake occurred on the northern segment of the fault on 6 July 2017. With interseismic time-series ALOS data between 2007 and 2011, and coseismic Sentinel-1 and ALOS-2 interferograms, we investigate the behavior of the fault in Leyte at different stages of the seismic cycle. We develop the first probabilistic (Bayesian) slip models of the Philippine Fault, derived from surface displacements measured using InSAR, which allow us to provide details of the relationship between seismic and aseismic slip, and estimate the potential earthquake return time with implications to seismic hazard. Surface velocity maps derived from InSAR time series analysis highlight the shallow aseismic slip right on known traces over 100 km of the fault. This is corroborated by observations of offset cultural features in the field. We also map an elliptical area with up to 2.5 cm/yr subsidence, associated with active geothermal energy production near the fault. With levelling data from prior surveys (from 1997 to 2003), we independently constrain the source of the subsidence to a rectangular sill source model at ~1.4 km depth. This corresponds to the bottom of most extraction wells and can explain the observed subsidence in both InSAR time-series and levelling data. With a fixed subsidence source and vertical fault geometry with variable patch sizes, inverting for slip at depth reveals along-strike variability of aseismic slip rates. In the interseismic model, sections of the fault slip by more than 3 cm/yr from the surface down to 15 km depth in northern and central Leyte, similar to our own inferred long-term tectonic rate (3.3 ± 0.2 cm/yr). Contiguous patches along 20-km of the fault--the Tongonan segment--however, show a significant slip deficit. Modeling slip from coseismic interferograms on the same fault geometry confirms that slip during the 6 July 2017 Mw 6.5 earthquake occurred right on the Tongonan segment, with up to 152 ± 0.2 cm of slip. With intermediate acquisition of Sentinel-1 data a day after the mainshock, our modeling shows that the slip of a Mw 5.8 aftershock four days later, in 10 July, slipped patches of the fault to the south of the mainshock, along a prominent fault bend. Aseismic slip on the Philippine Fault in Leyte may be promoted by the hydrothermal fluids in the volcanic arc, and rock and mineral assemblages conducive to aseismic slip. Finally, the slip deficit and current record of earthquakes in Leyte indicate that the Tongonan segment may be a site of recurring earthquakes.
Authors: Dianala, John Dale (1,2); Jolivet, Romain (3,4); Thomas, Marion Y. (5); Fukushima, Yo (6); Parsons, Barry (2); Walker, Richard Thomas (2)The development of a global tectonic strain rate model with high spatial resolution (~1 km or better) and high accuracy (~10-8 yr-1) was a key 5-10 year goal identified by the Group on Earth Observations (GEO) in the “Santorini Report”, Satellite Earth Observation for Geohazard Risk Management [Bally, 2012]. The demanding requirements for accuracy have driven the Sentinel-1 acquisition strategy in the seismic belts, where the vast majority of deadly earthquakes occur: since Sentinel-1A and -1B became fully operational, the aim has been to acquire ascending and descending Sentinel-1 images for all the seismic belts at least every 12 days. Here we report on efforts to process this vast quantity of data for the largest continental seismic belt, the Alpine-Himalayan Belt (AHB), which covers >10 Million km2 and stretches from Italy through to China. To achieve strain accuracies with InSAR of ~10-8 yr-1, equivalent to relative velocities of 1 mm/yr over length scales of 100 km, requires processing of long, densely sampled time series. In the COMET-LiCS system [Lazecky et al., 2020] we have set out to build a solution capable of automatically processing ~250x250 km “frames” of Sentinel-1 data using a modified SBAS strategy [Morishita et al., 2020] in which interferograms are formed between each new acquisition and three preceding epochs. Details of the system architecture and processing methodology are presented elsewhere in this meeting and by Lazecky et al. [2020]. Here we focus on the results from the AHB, covered by ~600 COMET-LiCS frames, with >150 acquisitions per frame to date. At the time of writing, processing is up to date for around 75% of the frames and we are on track for near complete processing of the AHB by mid-to-late 2021. Around 450,000 interferograms are currently available for download from the COMET-LiCS portal (http://comet.nerc.ac.uk/COMET-LiCS-portal); preliminary average velocities for ascending frames are shown in Figure 1. We use the InSAR data to create regional velocity and strain rate fields by jointly inverting GNSS and InSAR data, solving for the 3D velocity field as well as the reference frame adjustment parameters for the InSAR [Wang and Wright, 2012]. In Turkey the combined velocity field reveals the westward motion of Anatolia relative to Eurasia in unprecedented detail, with localised strain accumulation along the North and East Anatolian Faults [Weiss et al., 2020]. In Tibet, we see focused deformation around some major faults as well as distributed deformation in the interior. In addition, we map rapid vertical signals, many of which are associated with anthropogenic activities. We show that InSAR helps characterize short-wavelength details of the velocity and strain rate fields for large regions. We estimate that the current level of accuracy from InSAR at 100 km length scales in East-West velocities is ~2-3 mm/yr. Future work will improve accuracy and will focus on the use of these data sets in assessing seismic hazard. References Bally, P. E. (2012), Scientific and Technical Memorandum of The International Forum on Satellite EO and Geohazards, 21-23 May 2012, Santorini Greece, doi:10.5270/esa-geo-hzrd-2012Rep. Lazecký M, Spaans K, González PJ, Maghsoudi Y, Morishita Y, Albino F, Elliott J, Greenall N, Hatton E, Hooper A, Juncu D, McDougall A, Walters RJ, Watson CS, Weiss JR, Wright TJ. (2020). LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity. Remote Sensing. 12(15) Morishita, Y., M. Lazecky, T. J. Wright, J. R. Weiss, J. R. Elliott, and A. Hooper (2020), LiCSBAS: An Open-Source InSAR Time Series Analysis Package Integrated with the LiCSAR Automated Sentinel-1 InSAR Processor, Remote Sensing, 12(3), 424. Weiss JR, Walters RJ, Morishita Y, Wright TJ, Lazecky M, Wang H, Hussain E, Hooper AJ, Elliott JR, Rollins C, Yu C, González PJ, Spaans K, Li Z, Parsons B. (2020). High‐resolution surface velocities and strain for Anatolia from Sentinel‐1 InSAR and GNSS data. Geophysical Research Letters. 47(17) Figure 1 Caption: Average line of sight velocities derived from automatic processing of Sentinel-1 InSAR using the COMET LiCSAR/LiCSBAS system. Each frame is locally referenced by removing a planar ramp.
Authors: Wright, Tim J (1); Weiss, Jonathan (1,2); Walters, Richard (3); Morishita, Yu (1,4); Lazecky, Milan (1); Wang, Hua (5); Hussain, Ekbal (6); Shen, Lin (1); Hooper, Andy (1); Elliott, John (1); Rollins, Chris (1); Yu, Chen (7); Gonzalez, Pablo (8); Spaans, Karsten (9); Li, Zhenhong (7); Dodds, Nick (10); Ou, Qi (10); Watson, Andrew (1); Fang, Jin (1); Parsons, Barry (10)The Geohazards Lab, an initiative originated by the European Space Agency (ESA) with the support of several other CEOS space agencies and partners, was developed within the framework of the Committee on Earth Observation Satellites (CEOS) Working Group on Disasters (WG Disasters) to enable a greater use of Earth Observation (EO) data and derived products to assess geohazards and their impact. The Geohazards Lab main pillar is ESA’s Geohazards Exploitation Platform (GEP), a cloud-based environment providing a set of processing tools and services that allow mapping hazard prone land surfaces and monitoring terrain deformation. The Geohazards Exploitation Platform (GEP) is part of the Thematic Exploitation Platforms (TEP) initiative set up by ESA aiming to support the exploitation of satellite EO for geohazards assessment. It follows the Supersites Exploitation Platform (SSEP), originally initiated in the context of the Geohazard Supersites & Natural Laboratories initiative (GSNL). The Geohazards Exploitation Platform has been expanded to address broader objectives of the geohazards community. In particular, it is a contribution to the CEOS Working Group (WG) Disasters to support its Seismic Hazards and Volcano activities. The platform is meant to allow users to easily exploit EO data resources by combining fast data access, processing facilities and flexibility for the user’s own data analysis. The platform provides Data Access services, Data Processing services and PaaS (Platform as a Service) capacities. Data Processing services enable users to process data available in remote or local repositories using a number of well-known tools and on-demand services, and to exploit the results. Conventional InSAR services (e.g. DIAPASON and SNAP S-1 DInSAR service), advanced InSAR services (e.g. InSAR SBAS of CNR-IREA, S1 SNAP-StaMPS of BRGM), optical on-demand services (e.g. MPIC OPT, ALADIM and DSM OPT of the University of Strasbourg/CNRS EOST, S3 Active fire detection) and optical systematic services (e.g. INGV’s STEMP for surface temperature mapping, NOVELTIS and INGV’s VEGAN services for hot spot and vegetation vigor mapping) have been made available or are currently under integration. New platform functionalities have been released in 2019, such as the alerting system for automatic deformation mapping that shall trigger the services available on GEP based on seismic events polled from external systems e.g. USGS pager and Copernicus EMS. In addition, the platform makes available value-added information layers, offered as collections within the Thematic Applications, generated by a set of systematic processing services deployed and running on Cloud Computing resources. In this context, the Geohazards Lab envisages building a collaborative framework, through e-collaboration with expert geoscience centers and users to achieve a greater adoption of EO methods. On GEP, users can promote or share publicly or within closed communities, EO results generated on (or out of) the platform. More and more users from the geohazards community are exploiting satellite EO for the assessment of geohazards, with the advent of large new initiatives and projects of global interest. Starting in 2020, the GEP will apply to be onboarded on the Network of Resources (NoR). The NoR portal will be the single access point for both resource tier providers and platform service providers and will address self-funded and sponsored users looking for broad access to European EO data and processing facilities.
Authors: Papadopoulou, Theodora (1); Bally, Philippe (2); Pacini, Fabrizio (3); Foumelis, Michael (4); Provost, Floriane (2); Barchetta, Francesco (5)The global and systematic coverage of Sentinel-1 radar images allows to characterize, by radar interferometry (InSAR), surface deformations on a continental scale over large active fault systems. This represents considerable progress in fault monitoring and opens new perspectives in seismic hazard assessment. Our study covers the eastern part of the Tibetan plateau and focuses on the Kunlun fault, the Altyn Tagh fault, the Haiyuan fault, and the Xianshuihe fault system, accommodating part of the deformation related to the collision between the Indian and the Eurasian plates. To measure the interseismic deformation across these fault systems, we use an automated Sentinel-1 InSAR processing chain based on the NSBAS approach (Doin et al., 2011, Grandin, 2015), implemented at CNES high-performance computer center in Toulouse in the framework of the FLATSIM project (ForM@Ter LArge-scale multi-Temporal Sentinel-1 Interferometry processing chain in Muscate). We perform a time series analysis of the 2014-2020 Sentinel-1 InSAR data set, acquired along 1200 km-long tracks on seven ascending and seven descending orbits, covering a 1 700 000 km2 area, with a 160 m spatial resolution. We propose a methodology to express InSAR surface velocities in a pseudo-absolute reference frame and decompose the LOS velocity maps into a vertical and a horizontal contribution. InSAR average velocity maps are inverted together with the latest published GPS velocity field using the T-DEFNODE elastic block model to constrain strain partitioning in the study area. In particular, we invert the lateral variations of tectonic loading at depth and the slip deficit rates along these major faults and analyze internal deformation within the blocks. Our results highlight a strong deformation gradient across major faults that we analyze concerning the region's seismic history.
Authors: Lemrabet, Laëtitia (1); Lasserre, Cécile (1); Doin, Marie-Pierre (2); Métois, Marianne (1); Replumaz, Anne (2); Leloup, Philippe-Hervé (1); Sun, Jianbao (3); Chevalier, Marie-Luce (4); ., FLATSIM team (5)Separating different sources of signal in Interferometric Synthetic Aperture Radar (InSAR) studies over large areas is challenging, especially between the long-wavelength changes of atmospheric conditions and tectonic deformations, both correlated to elevation. In this study, we focus on the seismic cycle of the Mexican subduction since 2016. In a first part we present two methods to extract the 2017-2018 Slow Slip Event (SSE) signal in the Guerrero state area (Mexico). The largest SSE in this area occur every 4 years on the subduction interface, and last between 6-12 months. Their equivalent magnitude is about Mw 7.5 and their implications in the seismic cycle and their relation with earthquakes is a key question [1]. The geodetic study of the 2017 - 2018 SSE faces four main difficulties: (1) the permanent GPS network has a low spatial density with uneven distribution (less than 30 stations in an area of 300x300 km); (2) the tropospheric phase delays can be as high as 20cm of apparent ground displacements, with a complex temporal evolution, but the deformation is around 3cm; (3) the tested global weather models fail to correct interferograms with enough accuracy (with residual tropospheric signal higher than the tectonic signal), and (4) the 2017-2018 SSE signal shows complex interplay between 3 major earthquakes (2 in Sept. 2017 and one in Feb. 2018) and their post seismic deformations. To enhance the spatial coverage of the interest area, we use Sentinel-1 InSAR time series, and test two different approaches to extract the SSE signal. The first one, a parametric decomposition, consists in a least-square linear inversion, imposing a functional form for each deformation or tropospheric component. The second approach uses Independent Component Analysis (ICA) of the InSAR time series, this method allows to decompose the data without priori on the signal searched. Both method provide consistant results and allow to separate the atmospheric signal without previous tropospheric corrections. We obtain time series maps of surface displacements along the radar line-of-sight associated with the SSE and validate these results with a comparison to GPS. Combining those two approaches, we propose a method to separate atmospheric delays and tectonic deformation on time series data not corrected from tropospheric delays. From the extracted ground deformation maps, we propose a first-order slip inversion model at the subduction interface during this SSE. In a second part we will present preliminary results about the study of interseismic deformation signal excluding SSE. We search to characterise the lateral variations of the coupling along the subduction plan. To enhance the spatial coverage, our goal is to obtain a coupling map using the velocity extract to the GPS stations, as well as the InSAR velocity map. We use three Sentinel-1 tracks in descending covering the period of 2016 - 2019. This study allow to confirm the low coupling of the Guerrero Gap, and is in agreement to the previous study, using only GPS data [2]. [1] Radiguet, M., Perfettini, H., Cotte, N., Gualandi, A., Valette, B., Kostoglodov, V., Lhomme, T., Walpersdorf, A., Cabral Cano, E. et Campillo, M. (2016). Triggering of the 2014 Mw7.3 Papanoa earthquake by a slow slip event in Guerrero, Mexico. Nature Geoscience, 9(11):829–833. [2] Graham, S., Deme, C., Cabral-Cano, E., Kostoglodov, V., Rousset, B. ,Walpersdorf, A., Cotte, N., Lasserre, C., McCaffrey, R. et Salazar-Tlaczani, L. (2015). Slow sliphistory for the mexico subduction zone : 2005 through 2011. Pure and Applied Geophysics, pages 1–21.
Authors: Maubant, Louise (1); Pathier, Erwan (1); Radiguet, Mathilde (1); Daout, Simon (2); Doin, Marie-Pierre (1); Cotte, Nathalie (1); Kazachkina, Ekatarina (3); Kostoglodov, Vladimir (3)Questions regarding the development of folds, the role of volumetric deformation, and their relationship with earthquakes within a fault and fold system remain unanswered. Estimating fault slip and earthquake hazard using surface observations requires an understanding of how shortening is accommodated along and across tectonic structures during different phases of the seismic cycle. However, measuring the small rates of inter-earthquake deformation with space-geodesy techniques in mountainous areas is still challenging. Here, we construct a 16-year timeline of surface deformation from ESA satellite radar measurements across the North Qaidam thrust system in Tibet, where three Mw 6.3 earthquakes occurred in a relatively short time interval along basement faults underlying folded sediments. We process using NSBAS software the complete Envisat data archive along four overlapping tracks, as well as three Sentinel-1 tracks acquired in interferometric wide-swath mode including Interferograms with both short- and long-temporal baselines to avoid biases arising from the systematic loss of coherence in short-temporal baselines interferograms. Processing steps include correcting interferograms for tropospheric delays before unwrapping based on both empirical phase-elevation estimates and the use of the ERA-5 atmospheric models. After unwrapping and time series analysis, a parametric decomposition of the cumulative surface displacements into linear trends, co-seismic steps, and seasonal functions is performed to map, respectively, the changes of ground velocities, the co-seismic ground motions, and the seasonal ground displacements attributed to the freeze and thaw cycles of the water stored in the shallow layers of the ground. Long-wavelength spatial ramps for all interferograms are also iteratively re-estimated from bedrock pixels surrounding sedimentary basins affected by frost-related processes.The analysis reveals spatio-temporal changes of post-earthquake surface displacement rates and patterns, which continue more than ten years after the seismic events. The decomposition of the Sentinel-1 ascending and descending LOS velocities into vertical and shortening post-earthquake components indicates that long-term transient uplift and shortening is in agreement with the deformation that might be expected from kinematic models of folding. Long-term uplift coincides spatially with young anticlines observed in the geomorphology, with steep gradients in the forelimbs, gentle gradients in the back-limbs, an absence of subsidence in the footwalls, and higher gradients along interpreted bedding planes. Long-term shortening is also different from the surface displacements expected for typical time-varying creep on a narrow dislocation interface and shows rates higher than the average convergence across the whole region. These findings highlight the need to integrate the contribution of non-elastic deformation processes within shortening fault and fold systems for better quantification of slip-on underlying structures and improved understanding of the wide variety of transient ground deformation that can now be detected with geodetic networks.
Authors: Daout, Simon (1,2); Parsons, Barry (1); Walker, Richard (1)The Main Recent Fault (MRF) is a major right-lateral transform fault in the Zagros mountains of Iran. The fault has experienced Mw 6 earthquakes as recently as 2006 (Peyret et al., 2008), driven by 20-30 mm/yr of convergence between the Arabian and Eurasian plates (Khorrami et al., 2019). An accurate estimate of the rate of interseismic strain accumulation is critical both for estimating local seismic hazard, and for developing understanding of the overall distribution and mechanics of continental deformation across Iran, which constitutes one of the widest zones of continental convergence on a global scale. However, there has been major disagreement between previous estimates of the rate of slip on this important fault: previous estimates from regional GNSS observations and geological offsets vary between 1.6-17 mm/yr (Talebian and Jackson, 2002; Hessami et al., 2006; Authemayou et al., 2009; Alipoor et al., 2012; Khorrami et al., 2019). The GNSS data in particular are relatively sparse in the region of the MRF, but whilst the fault is favourably orientated for measurement with InSAR, no previous InSAR velocity estimates have been published to-date. Here we use Sentinel-1 observations to make the first InSAR estimate of the interseismic slip velocity and locking depth for a ~400 km section of the MRF. We use the LiCSAR automated system to process five years (2015-2020) of Sentinel-1 SAR acquisitions for four adjacent and overlapping frames (two descending and two ascending), producing a total of ~1500 interferograms. We apply a NSBAS-type approach using LiCSBAS (Morishita et al., 2019) in order to derive satellite line-of-sight velocities, mitigating the effects of atmospheric noise using GACOS (Yu et al., 2018). We estimate north-south velocities from regional GNSS (Khorrami et al., 2019) and use this to isolate the fault-parallel and vertical velocity components from the overlapping ascending and descending InSAR frames. Finally, in order to estimate the slip-rate and locking depth for the MRF, we fit 1-D screw dislocations models (Savage and Burford, 1973) to fault-perpendicular profiles, using a Bayesian approach as described by Goodman and Weare (2010) and Hussain et al. (2016). Our results show an interseismic slip velocity of 3±2 mm/yr below a locking depth of 20 km. The locking depth is poorly constrained because of the presence of incoherence and large subsidence signals close to the fault trace, and because the slip rate is close to the current sensing limit of Sentinel-1 time series (2-3 mm/yr). These results show that the MRF is an important major crustal structure that shows clear localisation of strain at depth and which accommodates a significant portion of the relative motion between Arabia and Eurasia. However, the overall slip rate is towards the lower end of the range of previous estimates, supporting previous GNSS-derived rates but which is inconsistent with longer-term estimates.
Authors: Watson, Andrew R. (1); Elliott, John R. (1); Walters, Richard J. (2); Houseman, Gregory A. (3)Estimation of coseismic slip distributions constrained by geodetic data is an important step in seismic analyses and earhquake cycle studies. Obtained slip distributions can be exploited to infer various characteristics of earthquakes such as rupture geometry, stress-strain distribution on adjacent faults, and consequently hazard assessment for future tremors. The mathematical model of slip inversion is often ill-conditioned and requires some sort of a priori assumption or regularization to achieve a stable and reliable solution. For this purpose, there are different a priori models and assumptions. The most common assumption is the smoothness of the slip behavior, which is imposed on the solution by the minimization of the second spatial derivative (or Laplacian) of the slip. More recently, some other assumptions have been also introduced such as slip self-similarity behavior that is incorporated in the slip inversion by a self-similar autocorrelation function (e.g., the Von Karman or exponential correlation function). In this regard, there are two challenging issues as follows. The ill-conditioned nature of the slip inversion implies that we can get similar misfit (between data and the model) for different a-priori assumptions. The main question is then which model/assumption is the best given the statistical characteristics of the data? For example, if we apply both Laplacian and Von Karman regularization on a dataset and both provide similar misfit, how can we judge which model is better? Another property of the regularized slip estimates is that they are biased. Consequently, the estimate of each slip patch has a relationship to many (or even all) other patches. This can result in a spurious leakage of the slip to the non-slipping patches. This effect is more significant in the methods that use the self-similarity slip assumption. So to obtain reliable results, it is important to know beforehand what is the real fault rupture area (i.e., which patches have a significant slip) and to avoid the inclusion of the non-slipping patches in the slip inversion. In this study, we try to answer both above questions using the concept of statistical hypothesis testing and by exploiting the knowledge about the statistical properties of geodetic data (in particular InSAR data). For the first problem, we develop a hypothesis-testing scheme based on the probability distribution of the norm of the misfit vector and also the characteristics of the a priori assumption. The model/assumption that have a higher p-value is then selected as the best model. The main property of the proposed approach is that the test not only depends on the misfit and the data statistics but also it accounts for the structure of the a priori assumption. In this way, the proposed test is capable to discriminate between two models with similar misfit. For the second problem, we propose to apply the parameter significance test (PST) on each patch in an iterative manner. We first estimate the slip and its precision (variance) for each patch. Then using the PST we decide whether the estimated slip is significant or not. If not, the patch will be removed from the rupture area and the procedure will be iterated until all the estimated patches have a significant slip. Note that in the both proposed tests, the data noise structure (or covariance matrix) is required. In case of InSAR data, the noise covariace matrix is obtained based on statistical analysis of data over a stable area far from the fault. The performance of the proposed tests is validated by synthetic study, follows by application of the method on the 2015 Mw 8.3 Chile earthquake using Sentinel-1 InSAR data.
Authors: Nazari-Zadeh Mahany, Shima; Samiei Esfahany, Sami; Safari, AbdolrezaWe process Sentinel-1 Synthetic Aperture Radar (SAR) data accumulated in the past ~5 years to study the crustal deformation of the Tibetan Plateau region. In particular, the interseismic behaviors of big faults are considered for strain accumulation or release by creeping activities, such as the Altyn Tagh fault (ATF) at the northern boundary, the Haiyuan fault (HF) at the northeastern boundary, the Kunlun fault (KF) in northern Tibet, and the Xianshuihe-Xiaojiang fault system (XXF) in the eastern or southeastern Tibet regions. As reviewed by Harries (2017), some big faults over the world in different tectonic environments are found to be creeping. Among these creeping faults, the strike-slip, rather than dip-sip faults, are more popular. It is not clear yet the role of creeping faults or fault sections on seismic hazards due to limited samples and related observations. Based on ample SAR data acquired in the past few years, we now have high-quality geodesy data for fault creeping detections. By analyzing the interseismic Spatio-temporal behaviors of Tibetan faults, we expect to find locations of creeping fault sections, analyze its relationship to locked parts with strain accumulation, and reveal its seismic hazards. To accelerate the data processing in such a large region, we utilize a high-performance computation (HPC) facility, named the Sunway TaihuLight, and parallelized computational codes for conventional (ISCE, Gamma, etc.) and time-series (StaMPS) InSAR analysis (Hooper et al., 2007). We adopt the persistent scatterer (PS) method for deformation detection, as the arid condition of the study area is quite suitable for PS identification with high enough density and the processing is easy to be parallelized by patch decomposition. Most of the processing steps work automatically, except the atmospheric filter step, in which we try with different options to minimize the atmospheric phase screen (APS) in time-series data. In particular, we reduce the APS by using the ECMWF ERA-5 model, which is proved to be useful in a statistical framework and could reduce observation uncertainties to some extent depending on InSAR coherence. Finally, we use the surrounding GPS data provided by Wang and Shen (2020) to calibrate InSAR observations in multiple orbits (descending and ascending pass geometry) and invert for high-quality 2D ground deformation and produce fault-parallel motion maps (Shen and Liu, 2020). It is also critical for using time-series GPS observations to calibrate InSAR time-series results, as long-wavelength errors and strong water vaper delays could seriously bias temporal behavior assessments of big faults, leading to unstable and wield motions inconsistent with limited cGPS observations. Besides, we also applied a more conventional while robust SBAS inversion code (Mintpy) to do InSAR time-series analysis on multi-looked SAR data, by evaluating phase closure distributions, temporal coherence, and bootstrap uncertainties of InSAR velocities, which is implemented in parallel by DASK technique. Both PS analysis in full resolution and SBAS processing in reduced resolution are carefully compared for consistency on InSAR velocity estimate. As some consistent results, we obtain the fault creeping signals along the Haiyuan fault (such as Jolivit et al., 2012, 2013) and the Xianshuihe fault (Zhang and Wen, 2014; Allen et al., 1994; Li and Burgmann, 2020), which is already known in previous studies, but with a more clear picture of fault motion. Besides, we also find some creeping signals along different sections (western and middle) of the Altyn Tagh fault system. Though the creeping signals are only localized at a short distance (several 10s of km), compared with the >1500 km ATF, the role of the creeping sections may have some implications for the evolution of the ATF and its seismic hazards at different sections. By invert for interseismic strain accumulation along locked sections and slip distribution on creeping sections along those faults, we infer that the creeping behaviors of big faults could be related to nearby big events within very localized parts, and most parts of those faults are actually still in locked status, and accumulated strains could be released in future earthquakes.
Authors: Sun, Jianbao; Qiu, JiangtaoA surprisingly large portion of large historical earthquakes in South Iceland have occurred in Spring and early Summer, with far less events during the wintertime. Statistical testing shows that this occurrence pattern is very unlikely due to chance, indicating that some seasonal process is influencing the timing of the earthquakes. We use Sentinel-1 time-series analysis to study and quantify the seasonal deformation in the area. Using snow-free images acquired from early Summer to Fall, the results show that the center of Iceland is strongly uplifting over time, most likely in response to reducing load of the major ice caps in Iceland. In addition, we find clear seasonal deformation with fast uplift during Summer, in response to snow melting, while in Winter the deformation reverses to subsidence. These results are in accord with continuous GPS measurements that show similar deformation patterns. Also, the results indicate stronger seasonal variations in the eastern part of the South Iceland seismic zone (~10 mm), closer to significant snow loading, than in its western part (~5 mm). This leads to differential seasonal motions across the seismic area (east-west distance about 60-70 km) that cause spatio-temporal subsurface stress changes that may influence the timing of earthquakes. We assess these seasonal variations by modeling the load-induced stress changes through a single year both in elastic homogenous and in layered half-space models. For this, we account for snow and glacier loading, as well as atmospheric and ocean load variations. Our results show that snow loading puts the seismic zone into compression during the late winter and snow melting during Spring and Summer in turn leads to relaxation of the fault zone compression. The peak in earthquake occurrence is in May and June, or soon after seasonal unloading starts. Therefore, the earthquake rate appears to correlate better with maximum unloading rate rather than the peak of the unloading in the Fall. The periodic load-induced stresses on seismogenic faults in South Iceland are however small (~1 kPa) and appear to only mildly modulate the rapid tectonic stressing rate in the area of ~20 kPa/year. This suggests that other mechanisms may be more influential in controlling the timing of large earthquakes in South Iceland.
Authors: Jónsson, Sigurjón; Cao, YunmengIn spaceborne SAR interferometry Passive Corner Reflectors (PCR) are often used as standard ground points when natural Permanent Scatters (PS) are not available, in areas uncoherent and when the radar backscattering is temporally unstable. Radiometric calibration and phase stability can be checked using their responses. The use of PCR is not always suitable, especially in areas as glaciers, snow covered regions, and mountain slopes where often due to a difficult accessibility installing PCRs can be very time and cost consuming. Furthermore, severe weather conditions can jeopardize their performance. An alternative to PCR can be represented by the installation of Active Reflectors (AR), more compact and lighter. The main limitation of an AR is the need to have available a power source. When the satellite passes overhead the AR receives the signal and retransmit it with a higher strength using the energy of the power source. The stability of its amplitude and phase response is basically affected by the presence of active Radio Frequency (RF) components susceptible to temperature changes. When AR are installed for large periods, from months to years long, the natural variation of the air temperature demands a careful thermal control of the device. According to their use AR can differ a lot in complexity, performance, and costs. Some products have been made available from the market for field campaigns but till now their use does not yet spread. The study here presented reports on the test of an AR designed to operate with Sentinel-1 SAR data, aimed at providing a fair performance/cost benefit and a low power consumption to make feasible the setup of a dense network. These characteristics are obtained through a tradeoff between an elementary RF architecture and the use of off-the-shelf components. In this presentation, after a brief introduction about the device design and implementation, some results concerning the amplitude strength and the phase stability of the AR in two different experimental campaigns are presented. Sentinel-1 images covering almost one year have been processed using a consolidated software chain developed to provide deformation monitoring. The results of two tests are presented; the first was carried out in an agricultural field close to CTTC in a real condition. Phase and amplitude responses were compared to those of a PCR (Rectangular Side Trihedral with a 0.65 m size) close to the AR and to stable buildings, to estimate its long-term stability using Sentinel-1A and -1B signals in the descending orbit geometry. A second test is based on the installation of an AR in a stable point close to a slope affected by landslides located in Slovenija. The analyzed data show a significant influence of the temperature fluctuations during the different periods of the year. To improve the stability of the retrieved displacement a calibration curve was derived using a data acquisition carried out in a controlled environment and using a signal simulated through a Vectorial network Analyzer reproducing a signal similar to that of the Sentinel-1. The results of these tests show that using the implemented prototype, without any active thermal control, the stability of the phase for long time is not compliant with the general requirement to assist deformation monitoring through InSAR, but using a correction formula the stability improves to an equivalent displacement of within ±2 mm. This value could represent a fine trade-off between the performance and the cost when a dense installation with a vast number of sensors is demanded.
Authors: Luzi, Guido; Espín-López, Pedro; Barra, Anna; Monserrat, Oriol; Crosetto, MicheleThe SENTINEL-1 (S-1) constellation of the Copernicus Program has been designed to provide an operational capability for the frequent mapping of the Earth surface, thanks to its two polar-orbiting satellites (SENTINEL-1A and B, launched on April 2014 and April 2016, respectively) performing C-band synthetic aperture radar (SAR) imaging [1]. In particular, with respect to previous C-band space-borne SAR systems (like those on board of the ERS-1/2, ENVISAT and RADARSAT-1/2 missions [2-3]) the S-1 constellation is characterized by enhanced revisit frequency, coverage and reliability for operational services and applications requiring long SAR data time series. Moreover, the S-1 constellation archive is available with a free and open access policy, thus easing the data access and enlarging the scientific community interested in its exploitation. Another key element of the S-1 constellation is its main acquisition mode on land, referred to as Interferometric Wide (IW) swath, which implements the Terrain Observation by Progressive Scans (TOPS) technique [4] that guarantees a very large spatial coverage, with a nominal footprint extending for about 250 km, and is specifically designed for interferometric applications. In addition, the S-1 constellation has stringent requirements based on a high attitude and orbit accuracy and it is intrinsically characterized by small spatial and temporal baselines, with an “orbital tube” of about 200m nominal diameter and a revisit time of 6 days (12 days in the case of only one operating satellites). As already mentioned above, the S-1 data are particularly suitable to be exploited through conventional and advanced interferometric techniques. Among the latter, a widely used advanced DInSAR approach is the Small BAseline Subset (SBAS) technique [5-6], which has already proven its effectiveness to investigate surface displacements with centimeter- to millimeter-level accuracy in different scenarios and is capable to perform analyses at different spatial scales and with multi-sensor data [7-10]. Moreover, a parallel algorithmic solution for the SBAS approach, referred to as Parallel Small BAseline Subset (P-SBAS), has been recently developed [11-12]. This approach permits to generate, in an automatic and unsupervised way, advanced S-1 IW DInSAR products by taking full benefit from the structure of S-1 IW data which are composed by bursts that can be considered as separate acquisitions [4]. Indeed, the processing is intrinsically parallelizable with respect to such independent input data and, therefore, the P-SBAS approach exploits this coarse granularity parallelization strategy in the majority of the steps of the processing chain [12]. Moreover, more sophisticated parallelization approaches are implemented, for the steps that are particularly intensive from the computational viewpoint, by exploiting both multi-node and multi-core programming techniques on parallel computing architectures such as cluster, grid and, above all, cloud computing infrastructures [13-15]. Following a short overview of the implemented S-1 IW P-SBAS processing chain, we will present the results of a DInSAR experiment carried out at the European scale. In particular, we have processed the entire available Sentinel-1 IW dataset collected between March 2015 and September 2018 from descending orbits, along 21 tracks, over a large part of Europe. The overall analysis was carried out by exploiting the S-1 IW P-SBAS processing chain deployed on one of the platforms of the Copernicus Data and Information Access Services (DIAS) and, in particular, on the one referred to as ONDA [16] which has been selected for this experiment through a public tender. The input dataset consists of ~72.000 S-1 images with a covered area of ~4.500.000 km2 (including the overlap among the considered frames). Our DInSAR processing experiment [17] was successfully carried out in ~6 months, finally allowing the retrieval of the deformation time series of the overall investigated area, including ~120.000.000 coherent (multi-looked) SAR pixels, each of these with a spatial dimension of ~80 m. The presented discussion will highlight the main characteristics of the exploited massive S-1 IW data processing solution for such wide area DInSAR experiment. Moreover, the analysis of the achieved results will show the high quality of the retrieved DInSAR time series that can be of interest for the Solid Earth scientific community that promoted this experiment as key study of the EPOSAR service included in the Satellite Data Thematic Core Service of the European Plate Observing System (EPOS) [18] of the European Strategy Forum on Research Infrastructure (ESFRI) [19]. This study, which has permitted to investigate the DIAS capacity to operationally serve systematic and automatic DInSAR processing services such as the one based on the P-SBAS approach, may also have a potentially positive impact for what concerns the future development of the European Ground Motion Service. Acknowledgments This work is supported by: the 2019-2021 IREA-CNR and Italian Civil Protection Department agreement; the I-AMICA (PONa3_00363) project; the H2020 EPOS-SP project (GA 871121), the EPOS-Italia INGV-IREA and the IREA-CNR/DGSUNMIG agreements. References [1] R. Torres et al., “GMES Sentinel-1 mission”, Remote Sens. Environ., vol. 120, pp. 9-24, May 2012. [2] M. Bonano et al.,“From previous C-band to new X-band SAR systems: Assessment of the DInSAR mapping improvement for deformation time-series retrieval in urban areas”, IEEE Trans. Geo. Rem. Sens., vol. 51, pp. 1973-1984, 2013. [3] E. Sansosti et al., “How second generation SAR systems are impacting the analysis of ground deformation”, Int. J. Appl. Earth Obs., vol. 28, pp. 1-11, 2014. [4] F. De Zan and A. M. Monti Guarnieri, "TOPSAR: Terrain Observation by Progressive Scans," IEEE Trans. Geo. Rem. Sens., vol. 44, pp. 2352-2360, 2006. [5] P. Berardino et al., “A new Algorithm for Surface Deformation Monitoring based on Small Baseline Differential SAR Interferograms”, IEEE Trans. Geo. Rem. Sens., 40, 11, 2375-2383, 2002. [6] A. Pepe, et al., “Improved EMCF-SBAS processing chain based on advanced techniques for the noise-filtering and selection of small baseline multi-look DInSAR interferograms,” IEEE Trans. Geosci. Remote Sens., vol. 53, pp. 4394–4417, 2015. [7] R. Lanari et al., “A Small-Baseline Approach for Investigating Deformations on Full-Resolution Differential SAR Interferograms,” IEEE Trans. Geosci. Remote Sens., vol. 42, pp. 1377–1386, 2004. [8] M. Manunta et al., “Two‐scale surface deformation analysis using the SBAS‐DInSAR technique: a case study of the city of Rome, Italy,” Int. J. Remote Sens., vol. 29, pp. 1665–1684, 2008. [9] A. Pepe et al., “On the Generation of ERS/ENVISAT DInSAR Time-Series via the SBAS Technique,” IEEE Geosci. Remote Sens. Lett., vol. 2, pp. 265–269, Jul. 2005. [10] M. Bonano et al., “Long-term ERS/ENVISAT deformation time-series generation at full spatial resolution via the extended SBAS technique,” Int. J. Remote Sens., vol. 33, pp. 4756–4783, 2012. [11] F. Casu et al., “SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 7, pp. 3285–3296, Aug. 2014. [12] M. Manunta et al., “The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment”, IEEE Trans. Geosci. Remote Sens., vol. 57, doi: 10.1109/TGRS.2019.2904912, 2019. [13] I. Zinno et al., “A Cloud Computing Solution for the Efficient Implementation of the P-SBAS DInSAR Approach”, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 10, pp. 802-817, 2017. [14] C. De Luca et al., “Large areas surface deformation analysis through a cloud computing P-SBAS approach for massive processing of DInSAR time series”, Remote Sens. Environ., vol. 202, pp. 3–17, 2017. [15] I. Zinno et al., “National scale surface deformation time series generation through advanced DInSAR processing of sentinel-1 data within a cloud computing environment”, IEEE Trans. Big Data, doi: 10.1109/TBDATA.2018.2863558, 2018. [16] https://www.onda-dias.eu/cms/ (accessed on 10 March 2020). [17] R. Lanari et al., "Automatic generation of sentinel-1 continental scale DInSAR deformation time series through an extended P-SBAS processing pipeline in a cloud computing environment", Remote Sensing, vol.12 (18), 2961, 2020. [18] European Research Infrastructure on Solid Earth. Available online: https://www.epos-ip.org/tcs/satellitedata/ (accessed on 10 March 2020). [19] European Strategy Forum on Research Infrastructure. Available online: http://www.esfri.eu/ (accessed on 10 March 2020).
Authors: Lanari, Riccardo (1); Ali, Zeeshan (1,2); Bonano, Manuela (1); Buonanno, Sabatino (1); Casu, Francesco (1); De Luca, Claudio (1); Fusco, Adele (1); Manunta, Michele (1); Manzo, Mariarosaria (1); Onorato, Giovanni (1); Zeni, Giovanni (1); Zinno, Ivana (1)The Copernicus Land Monitoring Service is developing a Sentinel-1 based Pan-European Ground Motion Service (EGMS). The service will rely on Persistent Scatterer Interferometry (PSI) and Distributed Scatterer Interferometry (DS). Both methods are robust and reliable techniques for ground motion monitoring over wide areas. Thanks to the launch of the Copernicus Sentinel-1 constellation, InSAR has undergone a rapid evolution, leading to the refinement of algorithms and increased industrial awareness. All these factors combined enable the development of national or regional services based on Sentinel-1-derived satellite interferometric products. As such, Norway, Germany, Italy, Denmark and the Netherlands are the first European countries to have implemented or to develop Ground Motion Services (GMS) at regional or national scales. Additionally, in 2017, the Copernicus User Forum and the Copernicus Committee approved the extension of the Copernicus Land Monitoring Service’s product portfolio by the European Ground Motion Service (EGMS), under the responsibility of the European Environment Agency (EEA).The approval followed the White Paper elaborated by the EGMS Task Force (https://land.copernicus.eu/user-corner/technical-library/egms-white-paper). The Task Force was formed as a natural consequence of an increasing discussion within the interferometric community and interest for the launch of a European-scale deformation map. The White Paper outlines the main characteristics of the service, the definition of the service, and the product portfolio. The White Paper was adopted, together with the Service Implementation Plan and Product Specification Document (https://land.copernicus.eu/user-corner/technical-library/egms-specification-and-implementation-plan), for the definition of the technical specifications for the EGMS. The EGMS will provide consistent, regular, standardized, harmonized and reliable information regarding natural and anthropogenic ground motion phenomena over Europe. Both ascending and descending data will be used. The EGMS portfolio is composed of three levels: Level 2a, deformation map with measurements along the radar Line-Of-Sight (LOS). Level 2b, deformation map referenced to observations from the Global Navigation Satellite System (GNSS) network. Level 3, Harmonized East-West and Up-Down deformation maps, obtained by combining the ascending and descending InSAR results from level 2b. The EGMS InSAR processing will include a baseline processing, at full SLC resolution using all available Sentinel-1 acquisitions (currently one repeat-pass every 6 days, for all Europe), followed by product updates every 12 months. The launch of the EGMS baseline is foreseen for Q1 2022. Product reliability is a priority, and as such the production will include appropriate quality control procedures. Data will be distributed to users via an ad-hoc dissemination platform, where visualization, preliminary data analysis, and download will be allowed for everyone, coherently to the Copernicus data policy. Guidelines, user manuals, and reports will be made available to facilitate user uptake. The EGMS will reach a wide range of users with different degrees of familiarity with the interferometric data. Specific user engagement and awareness-raising activities will be organized during the duration of the Service. Validation will be of paramount importance to demonstrate to users that the EGMS products are consistent with the technical specifications and with the expected range of application and use. Validation will be an independent process based on ancillary data not used for production purposes. A dedicated call for tenders will be launched in Q3 2021. The technical specifications for the tender are currently under definition and will be based on the Validation Approach and Plan document prepared by the EGMS Advisory Board (https://land.copernicus.eu/user-corner/technical-library/validation-approach-of-the-egms-product-portfolio). The EGMS Advisory Board, composed of satellite interferometric experts from different national agencies and research centers, has an important role in the frame of the EGMS, helping the European Environment Agency in the implementation of the Service and overseeing production and validation activities.
Authors: Crosetto, Michele (1); Solari, Lorenzo (1); Balasis-Levinsen, Joanna (2); Casagli, Nicola (3); Frei, Michaela (4); Oyen, Anneleen (5); Moldestad, Dag (6)The Norwegian ground motion service (InSAR Norway) has since November 2018 provided InSAR-based ground motion data for more than 4 billion measurement points in Norway. The service uses more than 4000 Copernicus Sentinel-1 images every year and provides government, industry, scientists, and the general public with a free and open InSAR-based ground motion data source. The data are publicly available in a web browser interface, with tools for simple data analysis, as well as download. Catastrophic failures of unstable rock slopes in Norway have occurred several times in the last century with fatal consequences. Therefore, the Geological Survey of Norway (NGU) systematically maps and classifies all unstable slopes in Norway for their hazard and risk, on behalf of the Norwegian Water Resources and Energy Directorate (NVE). InSAR data, based on Radarsat-2 collected from 2009, has played an important role in identifying unstable rock slopes in Norway for many years. Due to practical and financial constraints, Radarsat-2 data were only acquired over parts of Norway, focusing only on areas most prone to unstable rock slopes and catastrophic events, far from the entire country. InSAR Norway has been a game-changer for mapping unstable rock slopes on a national scale. Within a few weeks after the launch, more than 100 new potential unstable slopes were detected. Once an actively deforming slope is identified, its displacement rate provides a first assessment of the hazard level. This allows us to easier and more precisely define priorities for further and more detailed mapping and field investigations. InSAR measurements can be challenging to interpret since they are one-dimensional (along the radar line-of-sight). By combining InSAR data from several satellite geometries, both from ascending and descending orbits, two-dimensional displacements can be estimated. An improved hazard and risk assessment of unstable slopes require a good understanding of the kinematics. Field investigations and comparison with in-situ measurements are still important to validate and interpret the results. We demonstrate how InSAR Norway contributes to a better characterization of the kinematics of unstable slopes and a better understanding of the distribution of unstable rock slopes throughout the country.
Authors: Bredal, Marie Bergvik (1); Dehls, John (1); Hermanns, Reginald (1); Böhme, Martina (1); Penna, Ivanna (1); Nicolet, Pierrick (1); Larsen, Yngvar (2); Marinkovic, Petar (3)In this study, we use a combination of space-borne data from Sentinel-1, GRACE and in-situ measurements to quantitatively map groundwater depletion and assess the sustainability of groundwater consumption at a national scale across Iran. Withdrawal of groundwater in the past few decades boosted the economic development of Iran. A total of 700,000 pumping wells across the country abstract approximately 50 BCM of water each year, up to 90% is consumed in the agricultural sector. As a result of unsustainable groundwater withdrawal over the last three decades, many aquifers across the country have been depleted with an approximate average of 5 m. Gravity observation of GRACE mission from 2002 to 2017 shows a long-term trend of groundwater budget across Iran with approximately 15 Gt of water loss each year. However, the 500-km resolution of GRACE limits its ability to identify the exact location and extent of depleted aquifers. To overcome this limitation, large-scale Sentinel-1 InSAR observation processed using multi-temporal interferometric measurements is used. The results at a spatial resolution of 100 m and temporal resolution of 12 days identify the extent of all subsidence areas across the country and the temporal behavior of land subsidence. The InSAR displacement map suggests that aquifer depletion is widespread across the country, occurring in more than 300 basins. They are mostly located in agricultural areas, but also affecting populated regions. The time series of displacement for different aquifer show long-term trends of subsidence ranging from a few cm/yr to up to 40 cm/yr. In most of the basins, analysis of groundwater measurements indicates a continuous overdraft of groundwater for several decades. As a result, the long-term trend of subsidence is mostly non-elastic and irrecoverable. There are also short-term variations in displacement time series, primarily associated with the seasonal discharge/recharge of the aquifer. This short-term deformation of the aquifer is mainly elastic and recoverable. We analyze the sustainability of groundwater abstraction in different catchments of the country. The results suggest that out of six major watersheds in the country, four experience severe levels of groundwater stress. Our findings can be used for implementing detailed groundwater management strategies on a large scale in order to reduce the vulnerability of the country to the risks induced by groundwater extraction.
Authors: Haghshenas Haghighi, Mahmud (1); Motagh, Mahdi (1,2)The large volumes of Sentinel-1 data produced in Europe are being used to develop a pan-national ground motion service. However, providing useable information to a broad range of stakeholders is a challenge when the number of expert users is limited. Simple thresholding techniques are not always reliable for detecting and classifying complex deformation signals automatically. Therefore, we explore alternative approaches using deep learning. Previous studies have demonstrated the use of Convolutional Neural Networks (CNN) for detecting deformation associated with volcanoes in InSAR data (Anantrasirichai et al, 2018, 2019a, 2019b). Here we focus on the UK, where the main signals are associated with coal-mining, ground water withdrawal, landslides and tunnelling. However, high-resolution UK InSAR product is incompatible with the spatial convolution employed in deep learning due to the sparsity of measurement points. We perform spatial analysis of this high-resolution UK data and found that interferograms show the variance between point velocities increases sharply (the nugget). This behaviour appears as spike noise in the InSAR image, which significantly disturbs gradient calculation in the CNNs. Other challenges are that the number of real samples is insufficient for training and the ground motion is slower and more localised than in the volcano examples. To tackle these problems, we propose three enhancement methods: i) spatial interpolation with modified matrix completion, ii) a synthetic training dataset based on characteristics of real UK interferograms, and iii) enhanced over-wrapping techniques. Spatial interpolation is treated as an inverse problem and solved through a non-convex matrix completion using iterated soft thresholding. This technique achieves the estimation of missing pixels and spike noise reduction simultaneously. For the synthetic training dataset, we combine a synthetic deformation component with turbulent atmospheric delays, then sub-sample it to replicate the sparsity of the UK InSAR data, and add spike noise. Finally, spatial interpolation is applied. To detect slow and localised motion, we apply overwrapping technique, where the phase (φ) is multiplied with a wrap gain (ρ), i.e. φ' = ρφ. We also employ shifting the wrapping boundaries by adding a constant phase offset to the velocity map. This reduces the offset due to different reference points. We report on the ability of the algorithms to detect deformation in the UK, firstly using two test cases of areas of subsidence caused by coalfields and tunnelling. Then we use a UK-wide deformation map. Our framework can detect coal mining areas (e.g. South Wales, Normanton and Castleford, North Staffordshire in Stoke-on-Trent, Northwest Leicester, Northumberland and Durham) and engineering work (e.g. underground constructions in London and slate quarries in North Wales). The detection results from the UK dataset show the potential use of the proposed methods for future automated systems. Refereneces: Anantrasirichai, N., Biggs, J., Albino, F., Hill, P. and Bull, D., 2018. Application of Machine Learning to Classification of Volcanic Deformation in Routinely Generated InSAR Data. Journal of Geophysical Research: Solid Earth, 123(8), 6592-6606. Anantrasirichai, N., Biggs, J., Albino, F., and Bull, D., 2019a. A deep learning approach to detecting volcano deformation from satellite imagery using synthetic datasets. Remote Sensing of the Environment. 230, 111179 Anantrasirichai, N., Biggs, J., Albino, F., and Bull, D., 2019b. The Application of Convolutional Neural Networks to Detect Slow, Sustained Deformation in InSAR Time Series, Geophysical Research Letters.
Authors: Anantrasirichai, Nantheera (1); Biggs, Juliet (1); Kelevitz, Krisztina (2); Sadegh, Zahra (2); Wright, Tim (2); Bull, David (1)Reliable and accurate long term geodetic monitoring with SAR requires the installation of either passive corner reflectors or, alternatively smaller active devices. We report our first results using novel off-the-shelf transponders or electronic corner reflectors (ECRs) for geodetic measurements with Sentinel-1 C-band Synthetic Aperture Radar (SAR) data. For this purpose we set up a triangular arrangement consisting of one trihedral corner reflector and two active ECRs at the campus of German Aerospace Center (DLR) in Oberpfaffenhofen, Germany. We describe the practical aspects of such ECRs as well as first radiometric characteristics. Moreover, we present geometric accuracy numbers derived from imaging geodesy [1], [2], i.e. absolute radargrammetric positioning, as well as from interferometric phase measurements. In the frame of the ESA project SAR-HSU (ESA AO/1-9172/17/I-BG-Baltic+) with the goal of monitoring tide gauges with SAR to connect the height systems of neighbor states (Sweden, Finland, Poland, Estonia,) – some of them are severely influenced by postglacial uplift - a test installation with 11 ECRs surrounding the Baltic Sea is being set up. Due to the long baselines involved and the large waterbody in between, SAR interferometry cannot be used for differential height change measurements. Instead absolute SAR measurements for each single point shall be evaluated. The technique has been demonstrated to achieve cm-level accuracy with high resolution TerraSAR-X data and a ranging accuracy of about 6 cms with Sentinel-1 data. While we have demonstrated this accuracy with CRs, no experience exists so far with active ECRs. To assess the accuracy and stability of the ECRs we set up a small validation setup at DLR Oberpfaffenhofen. Our set-up consists of one mechanical 1.5 m CR and two ECRs placed in a triangle with baselines of approximately 100 to 350 meters. The geolocation of SAR sensors can be evaluated by comparing the reference coordinates of point targets with the measured image data in the 2D SAR image space (range and azimuth). The comparison is performed on an image by image basis, using the precise orbit solution of the SAR satellite as well as corrections for atmospheric path delays, solid Earth tidal deformations, and Sentinel-1 specific system corrections. The details of our methods are summarized in [3] and [4]. At our DLR test site two Sentinel-1 ascending geometries and one descending geometry are usable, each with a temporal sampling of 6 days. First analysis of the passive CR coordinates shows measurement standard deviations of 1 – 5 cm in range, depending on the swath and about 50 cm in azimuth. Especially the range values are excellent values as expected from earlier experiments. The azimuth values are somewhat worse and need to be investigated further. However, the ECRs show a ranging standard deviation of 12 cm, which worse than expected from their radar cross section which is stronger than that of the passive CRs. This effect needs to be investigated closer and compared with interferometric differential phase measurements as soon as more passes are available. In our talk we provide detailed results concerning the accuracy in absolute ranging, in relative interferometric measurements and also practical experiences with the installation and operation of the ECRs. References [1] M. Eineder, C. Minet, P. Steigenberger, X. Cong, and T. Fritz, “Imaging Geodesy - Toward Centimeter-Level Ranging Accuracy with TerraSAR-X,” IEEE Trans. Geosc. Remote Sens., vol. 49, pp. 661-671, 2011. [2] C. Gisinger, U. Balss, R. Pail, X. Zhu, S. Montazeri, S. Gernhardt, and M. Eineder, “Precise Three-Dimensional Stereo Localization of Corner Reflectors and Persistent Scatterers With TerraSAR‑X,” IEEE Trans. Geosc. Remote Sens., vol. 53, no. 4, pp. 1782-1802, 2015. C. Gisinger, U. Balss, H. Breit, A. Schuber, M. Garthwaite, D. Small, T. Gruber, M. Eineder, T. Fritz, N. Miranda, „Recent
Authors: Gisinger, Christoph (1); Eineder, Michael (1); Brcic, Ramon (1); Gruber, Thomas (2); Oikonomidou, Xanthi (2); Heinze, Markus (2)Study of land subsidence in Tehran with Sentinel-1 Data: Geological investigation and GPS analysis Soheil Rajabi Baniani Dr. Ling Chang, Dr. Yasser Maghsoudi Soheil Boulevard 1945 4 Enschede, The Netherlands s.r.baniani@gmail.com s.rajabibaniani@student.utwente.nl Tel: +31684982683 Tehran, as a megacity in Iran, is exposed to a high rate of land deformation. Recent research shows average land deformation speed is up to 39.9 mm/year in southeast plain (from 2014 to 2017)(Foroughnia, Nemati, Maghsoudi, & Perissin, 2019) and groundwater extraction in Tehran plain for agricultural and industrial demands is the most probable driving mechanism (Haghighi & Motagh, 2017). It is undisputed that infrastructure and structure in Tehran are continuously under threat by such rapid land subsidence, and this subsidence may also lead to significant economic losses such as structural damage and high maintenance costs for roads, railways, dikes, pipelines, and buildings. It is believed that the awareness of location and rate of subsidence is essential for future urban development plans. Therefore, when, where, and why the subsidence did/does occur has to be closely monitored and analyzed considering future planning and the importance of infrastructure and structural damage, which has a profound effect on human societies. This study attempts to use Sentinel 1 SAR data to map land subsidence in Tehran and validate the results by using GPS data. With the advent of satellite radar interferometry (InSAR), researchers have utilized this technique considerably for surface elevation modelling and deformation detection. InSAR can measure land displacement at higher spatial resolution and more coverage with comparable accuracy to geodetic observations (Rucci, Ferretti, Monti Guarnieri, & Rocca, 2012) (Galloway et al., 1998) (Rosen, Hensley, Peltzer, & Simons, 2004). On the contrary, the field techniques such as GPS, can provide high accuracy ground measurements at predefined sparse points (i.e., GPS stations) (Moreau, Dauteuil, Bour, & Gavrilenko, 2006) and assist for InSAR results' validation. In this study, we applied the standard time series approach for SAR data processing, that is persistent scatterer interferometry (PSI) (Hooper, Segall, & Zebker, 2007). Note that GPS and PSI observations are inherently not the same, in view of the fact that their Spatio-temporal content is different. GPS offers all measurements in an absolute geographic coordinate system, while PSI offers relative measurements in time and space. Thus, the first interpretable PSI observation is the double–difference, both temporally (between master and slave acquisition time) and spatially (between two points). Therefore, in order to be able to compare the GPS to InSAR observations, they are converted to double–differences along the vertical direction. We implemented the proposed methods for the case of Tehran with an area of about 1600 km2. 52 Sentinel 1A (C-band) dataset acquired between 20018 and 2019 were collected to generate 1,746,317 PS measurement points. Results illustrate that the maximum loss of elevation over the monitoring period amount to 11.7 cm/year. Two GPS stations are deployed in the study area, GPS_m318, GPS-m020 where are located at (35.64 N latitude, 51.29 E longitude), and (35.58 N latitude, 51.42 E longitude) respectively. Land Surveying Unit of Tehran Municipality has read GPS data from 1/1/2006, and the observations have been collected continually within 24 hours of a day. To evaluate PSI results, the GPS observations between 1/1/2018 to 27/10/2019 have been used. Our results show that the maximum and minimum double difference between GPSs and PSs are 0.0536 m, 0.0015 m, respectively; moreover, the histogram shows that most data are in the interval of [-1 1]cm. and RMSE = 0.011 m. Besides, we also apply the velocity comparison of double-differenced GPS and PS. The result shows the PS observations matched well with the GPS observations. The average groundwater level in Tehran decreased by approximately 12m from 1984 to 2011 (Mahmoudpour, Khamehchiyan, Nikudel, & Ghassemi, 2016), the number of exploitation wells in the area has increased from 3906 in 1968 to 26076 in 2003 and 32518 in 2012. Comparing water table variations and land deformation reveal that groundwater withdraw is a major driving mechanism, however, the variation in soil type also plays a crucial role. For instance, in the study area, although groundwater levels (Xutm = 503498, Yutm = 3948916) has decreased by approximately 13m from 2012 to 2017 at the place of Andisheh-Jadid, no subsidence has been detected due to the presence of well grade layers at that location. The analysis result unveil the interaction of land subsidence and piezometric level over time. To sum up, in the southwest of Tehran, land subsidence has been happening where it is often associated with the destruction of aquifers, due to excessive pumping of the wells for the agricultural and industrial activities. Moreover, the correlations among subsidence, soil type, and thickness of soil layers are identified. An InSAR time-series investigation of land displacement for the period of 2018-2019 in the Tehran plain demonstrates that the maximum rates of subsidence is 11.7 mm/yr. By comparing the PSI results with the permanent GPS stations, the actual magnitude and trend of the deformation has been validated and confirmed. Reference: Foroughnia, F., Nemati, S., Maghsoudi, Y., & Perissin, D. (2019). An iterative PS-InSAR method for the analysis of large spatio-temporal baseline data stacks for land subsidence estimation. International Journal of Applied Earth Observation and Geoinformation. https://doi.org/10.1016/j.jag.2018.09.018 Galloway, D. L., Hudnut, K. W., Ingebritsen, S. E., Phillips, S. P., Peltzer, G., Rogez, F., & Rosen, P. A. (1998). Detection of aquifer system compaction and land subsidence using interferometric synthetic aperture radar, Antelope Valley, Mojave Desert, California. Water Resources Research, 34(10), 2573–2585. https://doi.org/10.1029/98WR01285 Haghighi, M. H., & Motagh, M. (2017). Sentinel-1 InSAR over Germany: Large-scale interferometry, atmospheric effects, and ground deformation mapping. ZFV - Zeitschrift Fur Geodasie, Geoinformation Und Landmanagement. https://doi.org/10.12902/zfv-0174-2017 Hooper, A., Segall, P., & Zebker, H. (2007). Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcán Alcedo, Galápagos. Journal of Geophysical Research: Solid Earth, 112(7). https://doi.org/10.1029/2006JB004763 Mahmoudpour, M., Khamehchiyan, M., Nikudel, M. R., & Ghassemi, M. R. (2016). Numerical simulation and prediction of regional land subsidence caused by groundwater exploitation in the southwest plain of Tehran, Iran. Engineering Geology, 201, 6–28. https://doi.org/10.1016/j.enggeo.2015.12.004 Moreau, F., Dauteuil, O., Bour, O., & Gavrilenko, P. (2006). GPS measurements of ground deformation induced by water level variations into a granitic aquifer (French Brittany). Terra Nova, 18(1), 50–54. https://doi.org/10.1111/j.1365-3121.2005.00659.x Rosen, P. A., Hensley, S., Peltzer, G., & Simons, M. (2004, February 3). Updated repeat orbit interferometry package released. Eos, Vol. 85, p. 47. https://doi.org/10.1029/2004EO050004 Rucci, A., Ferretti, A., Monti Guarnieri, A., & Rocca, F. (2012). Sentinel 1 SAR interferometry applications: The outlook for sub millimeter measurements. Remote Sensing of Environment, 120, 156–163. https://doi.org/10.1016/j.rse.2011.09.030
Authors: Rajabi Baniani, Soheil (1); Chang, Ling (2); Maghsoudi, Yasser (3)With the increasing number of SAR satellite missions in recent years and the years ahead, typically multiple InSAR-based ground motion data sets are available for a certain area. In addition, measurements acquired by other techniques, such as levelling, GNSS, or gravity, may have been acquired. These datasets are complementary to each other, due to their spatial density and coverage, temporal density and coverage, and sensitivity (1D versus 3D). Because of this complementary nature, an integration of the datasets to jointly estimate the final parameters and generate an optimal output product is desirable. However, the differences between the techniques make this integration non-trivial. Conventional geodetic processing methodologies require for instance collocated measurements, i.e. at common locations or benchmarks. Therefore, geodetic adjustment and testing procedures are typically applied for each technique/dataset separately, followed by a final integration step. We developed a software package for the integrated adjustment and testing of various geodetic observation types. The software enables the inclusion of levelling data, campaign as well as continuous GNSS measurements, and multiple InSAR datasets, typically acquired from different sensors, different orbits (ascending/descending, incidence angle), and covering different time spans. The methodology consists of a number of processing steps. First, each dataset is pre-processed, to detect and isolate obvious measurement errors and to convert the data into a standardized data format. This data format is based on the space-time matrix concept, and builds upon previous work (van Leijen et al., 2017). Since the spatio-temporal data distribution of the various datasets will be different at this stage, a data reduction approach is applied to the GNSS and InSAR datasets, to create common benchmarks and epochs among the datasets. Then, the actual integrated adjustment and testing is applied. Hereby, the sensitivity direction of the various datasets is accounted for. By incorporating the covariance matrix of each observed dataset, the precision of the different datasets is properly accounted for and an optimal estimate is obtained. Based on the estimated transformation parameters between the different (reduced) datasets, all original datasets can be transformed to the same reference. Since all the (intermediate) products are stored in the same standardized space-time matrix format, standard output products can easily be generated. These output products can range from displacement time series for a single location, to a displacement contour map for a large region. In our contribution we show the developed methodology, and demonstrate the functionality of the software using both simulated and real data. References: Van Leijen, F., Esfahany, S. S., Van Der Marel, H., & Hanssen, R. (2017). A standardized approach for the integration of geodetic data for deformation analysis. In 2017 IEEE International Geoscience and Remote Sensing Symposium: International Cooperation for Global Awareness, IGARSS 2017 - Proceedings (Vol. 2017-July, pp. 957-960). [8127112] Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/IGARSS.2017.8127112
Authors: van Leijen, Freek; van der Marel, Hans; Hanssen, RamonA complex water-related risk scenario characterizes Mexico City, one of the largest metropolises of the world. The city experiences frequent flooding from heavy rains and fast run-off from the hills surrounding the Valley of Mexico basin, which combine with heat waves and water shortage, the latter exacerbated by significant loss due to pipe leaks. Groundwater exploitation for municipal, agricultural and industrial use is in excess of natural recharge, thus inducing depletion of the aquifer system and ground subsidence and, in turn, damage to urban infrastructure, increased surface flooding exposure, and changes in the seismic response of the basin. With a well-marked subsidence pattern and deformation rates as fast as 40 cm/year, Mexico City has frequently being used as testing ground to trial new satellite Synthetic Aperture Radar (SAR) acquisition modes, novel satellite missions, and newly developed Interferometric SAR (InSAR) processing methods. In this paper, we exploit the unprecedented opportunity to generate displacement time series with some hundreds of C-band SAR scenes offered by the Copernicus Sentinel-1 constellation since 2014. We achieve this purpose making specific use of ESA’s Geohazards Exploitation Platform (GEP) [1] as part of the Early Adopters Programme and the Geohazards Lab initiative, the latter developed under the CEOS Working Group on Disasters. This is carried out not only to undertake a scientific study of Mexico City in continuity with previous research, but also to demonstrate what GEP can do for InSAR expert users aiming to process big SAR data and gather a reliable input for further post-processing within the common GIS environment. Multiple SAR viewing geometries were used for the estimation of the 3D land deformation field affecting the city. Sentinel-1 Interferometric Wide (IW) swath mode SAR data were processed in GEP using the Parallel Small BAseline Subset (P-SBAS) on-demand processing service, developed by CNR-IREA [2]. The P-SBAS was integrated on ESA’s G-POD [3] and used for subsidence mapping in central Mexico [4], including in Mexico City [5,6]. The processing workflow was also tailored for Sentinel-1 IW TOPS data, including steps for precise co-registration and interferogram formation at burst level, and subsequent mosaicking of the multi-looked burst interferograms to sub-swaths and then to the complete extent of the IW scene [7]. The P-SBAS analysis reveals peaks of -38.7 cm/year vertical ground deformation velocity in the central sector of Nezahualcóyotl, -32.0 cm/year in the eastern sectors of Gustavo A. Madero and Venustiano Carranza and in the north-western sector of Iztapalapa, and -8.8 cm/year in the area of the metropolitan Cathedral in Cuauhtémoc [8]. The land subsidence pattern retrieved with Sentinel-1 InSAR analysis in the period 2014-2021 is consistent with observations for the last six decades. Comparison of the 3D deformation field with surface geology and geotechnical zoning confirms that the compaction of the lacustrine aquitard is the predominant process [8]. The GEP and its high-performance computing resources and hosted processing chains, open new opportunities for the exploitation of big data from the Sentinel-1 SAR mission to study and monitor geological hazards affecting anthropogenic environments [9]. REFERENCES [1] Foumelis M., Papadopoulou T., Bally P., Pacini F., Provost F., Patruno J. 2019. Monitoring Geohazards Using On-Demand and Systematic Services on Esa’s Geohazards Exploitation Platform. Proceedings of the IEEE International Geoscience and Remote Sensing Symposium - IGARSS 2019, 5457-5460. [2] Casu F., Elefante S., Imperatore P., Zinno I., Manunta M., De Luca C., Lanari R. 2014. SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7, 3285-3296. [3] De Luca C., Cuccu R., Elefante S., Zinno I., Manunta M., Casola V., Rivolta G., Lanari R., Casu F. 2015. An On-Demand Web Tool for the Unsupervised Retrieval of Earth’s Surface Deformation from SAR Data: The P-SBAS Service within the ESA G-POD Environment. Remote Sensing, 7, 15630-15650. [4] Cigna F., Tapete D., Garduño-Monroy V.H., Muñiz-Jauregui J.A., García-Hernández O.H., Jiménez-Haro A. 2019. Wide-area InSAR survey of surface deformation in urban areas and geothermal fields in the eastern Trans-Mexican Volcanic Belt, Mexico. Remote Sensing, 11, 2341. [5] Albano M., Polcari M., Bignami C., Moro M., Saroli M., Stramondo S. 2016. An innovative procedure for monitoring the change in soil seismic response by InSAR data: application to the Mexico City subsidence. International Journal of Applied Earth Observation and Geoinformation, 53, 146-158. [6] Cigna F., Tapete D. 2017. The value of SAR big data for geohazard applications: automated grid processing of ERS-1/2 and ENVISAT data in ESA’s G-POD. In: Soille P., Marchetti P.G. (Eds.), Proc. 2017 Conference on Big Data from Space. BIDS’2017, EUR 28783 EN, JRC108361. Publications Office of the EU, Luxembourg, pp. 165-168. [7] Manunta M., De Luca C., Zinno I., Casu F., Manzo M., Bonano M., Fusco A., Pepe A., Onorato G., Berardino P., De Martino P., Lanari R. 2019. The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment. IEEE Transactions on Geoscience and Remote Sensing, 57, 6259-6281. [8] Cigna F., Tapete D. 2021. Present-day land subsidence rates, surface faulting hazard and risk in Mexico City with 2014-2020 Sentinel-1 IW InSAR. Remote Sensing of Environment, 253, 19 pp. doi:10.1016/j.rse.2020.112161 [9] Cigna F., Tapete D. 2021. Sentinel-1 Big Data Processing with P-SBAS InSAR in the Geohazards Exploitation Platform: An Experiment on Coastal Land Subsidence and Landslides in Italy. Remote Sensing, 13(5), 885. doi:10.3390/rs13050885
Authors: Cigna, Francesca (1,2); Tapete, Deodato (1)Sinkholes have regularly been reported to exhibit distinctive precursory deformations on the Earth’s surface. These sinkhole-related spatio-temporal deformations show patterns, i.e. trends in space and time, that sometimes follow distinct shapes such as an inverted Gaussian function, a cylinder, or a cone [1]. Recent studies showed that these patterns can be detected using Time-series Synthetic Aperture Radar Interferometry (TSInSAR) [2]. A dedicated method to do so, however, is not yet investigated. This study aims to develop and investigate a method coined as 'Sinkhole Scanner' to detect sinkhole-related deformation patterns. The detection of sinkhole-related deformation patterns using TSInSAR entails sufficient InSAR measurement points (IMP) with the associated deformation time series. Here, we propose a 4-dimensional (space and time) moving window-based method, where a 3-dimensional sinkhole model like an inverted Gaussian function, a cylinder or a cone, evolving in time, is fitted to the IMP deformation time series overlapped by the moving window using least-squares estimation. This window has a user-defined size and moves with a user-defined stride over the 2-dimensional horizontal surface. Assessment of the fit is done by the posterior variance and the correlation coefficient. Our method was applied to two datasets: (i) a simulated deformation time series data over IMP locations, (ii) a real deformation time series of 105 Sentinel-1A images (IW mode, VV polarization, descending orbit), acquired between May, 2015 and Jan, 2019, over an area in Monaghan County, Ireland, where a sinkhole had occurred on September 25th, 2018 [3]. The IMP locations were extracted using TSInSAR and were used in both applications. Simulations were performed by configuring both satellite data parameters and sinkhole parameters. The satellite data parameters include spatial resolution, the precision of the deformation estimates and radar wavelength, while sinkhole parameters include the spatial size, maximum depth, and shape of the sinkhole. The deformation time series was taken from the simulated sinkhole deformation signal and modulated with a normally distributed noise in the range of zero and a user-defined precision in the deformation signal. Our results show that the Sinkhole Scanner accurately identifies simulated sinkholes, and detects an evolving Gaussian pattern over the real sinkhole area. The posterior variance for sinkhole detected windows was ~50% lower than that for windows over areas that are unaffected by sinkholes. In the real data application, such detections were also made at several other locations surrounding the sinkhole area, including a location ~700m north of the sinkhole site where another sinkhole was reported in December, 2018 [4]. These detections also suggest other potential brewing sinkholes in the area. Some of these detections coincided with IMPs which were deforming faster than other IMPs in their spatial neighborhood. Additional detections were made in areas where the IMPs showed slow deforming IMPs, which may suggest either brewing sinkholes or false positives. This suggests that the Sinkhole Scanner has the potential to detect sinkhole related deformations which may otherwise get missed when sinkhole related subsidence is identified by just using the deformation velocity maps. Moreover, a trend could be observed in the study area that allowed us to distinguish sinkhole prone and non-sinkhole prone sub-areas. Finally, we observed that smaller sinkholes are identified by reducing the size of the scanning window, however, with a trade-off of missed detections due to a lower number of overlapped IMPs over a smaller window. For example, reducing the window size from 100 × 100 pixels to 50 × 50 pixels reduced the number of scanning windows with no solutions by ~40 percent. The Sinkhole Scanner is able to test various sinkhole related shapes in the InSAR phase deformation time series. The results can be used in addition to the deformation velocity map, to find critical areas where sinkhole related subsidence may occur. The main limitations of the work are that this method may also return false positives, and that so far sinkhole related shapes have been assumed to follow simple geometries. References: [1] Williams, P., 2004. Encyclopedia of Caves and Karst Science, 2nd edition. Fitzroy Dearborn Publ., Ch. Dolines, pp. 304–310 [2] Chang, L., & Hanssen, R. F. (2014). Detection of cavity migration and sinkhole risk using radar interferometric time series. Remote Sensing of Environment, 147, 56–64. https://doi.org/10.1016/j.rse.2014.03.002 [3] RTE. (2018). Fresh sinkholes appear after Monaghan mine collapse. Retrieved March 12, 2021, from RTE website: https://www.rte.ie/news/ulster/2018/0925/998064-monaghan-sink-hole/ [4] BBC News (2018). New sinkhole opens up in Monaghan. Retrieved March 12, 2021, from BBC News website: https://www.bbc.com/news/world-europe-46641209
Authors: Kulshrestha, Anurag; Chang, Ling; Stein, AlfredDuring the past three decades, the Dead Sea (DS) water level has dropped at an average rate of about 1 m/year, resulting in dissolution of a subsurface salt layer by under-saturated groundwater and formation of sinkholes along its coastline. Currently, about 6000 sinkholes are mapped along the Israeli shorelines of the DS, with more than 300 sinkholes formed annually. The sinkholes severely affect the daily life, economy, infrastructure, tourism and industry of the region. Sinkholes are associated with gradual land subsidence, before, during and after their collapse. Systematic high temporal and spatial resolution interferometric synthetic aperture radar (InSAR) observations, augmented by detailed Light Detection and Ranging (LiDAR) measurements enable utilization of interferometric pairs to detect minute precursory subsidence before the sinkholes collapse and generate sinkhole susceptibility maps. By now, InSAR measurements have become fundamental for sinkhole early warning and mitigation along the DS coast in Israel. A semi-automatic processing system has been developed at the Geological Survey of Israel (GSI) using InSAR monitoring of sinkhole precursors for sinkhole early warning along the entire DS shores in Israel. The system allows us to produce an alert for a new sinkhole in less than 24 hours after data acquisition. We continuously monitor displacements along the shore and delineate small decorrelated areas covering all or parts of the subsiding areas as newly formed sinkholes. Final validations of new sinkhole collapses are carried out by field observations, drone photos, aerial photos and LiDAR digital elevation models (DEMs). We present four case studies that illustrate the timelines and effectiveness of our methodology, as well as its limitations and complementary methodologies used for sinkhole monitoring and hazard assessment. Early detection of precursory subsidence along the major highway along the DS, has proved most valuable for timely planning of a temporary alternative route, a few years before a sinkhole collapsed on the highway. The final bypass was planned using InSAR subsidence maps. In an adjacent site, the Mineral Beach spa, InSAR measurements were only partly successful. While a subsidence belt has been observed migrating with time from the west towards the spa, the collapse of sinkholes within the spa has not been foreseen by InSAR due to coherence loss. Knowing this limitation, we conducted a nano-seismic survey in the vicinity of the spa, which revealed precursory subsurface collapses and compensated for the InSAR drawbacks. Following an environmental debate concerning the location of a new pumping station for the DS mineral industry, InSAR subsidence maps were applied as a final decision-making tool, delineating high-risk areas of active subsidence and sinkholes along two of the proposed pipeline alternatives and stable ground along the third route that was finally chosen. On the broader scale, InSAR-derived subsidence maps are introduced into all sinkhole susceptibility maps along the DS, which currently serve as mandatory base maps for planning and licensing of new infrastructures. The case studies presented above, demonstrate the applicability of InSAR monitoring to sinkhole hazard assessment and mitigation along the DS in Israel. The implemented monitoring procedure allows rapid alert and identification of immediate hazard increase within a short time of data acquisition. It also allows coverage of broad areas rather than individual sites as does the nano-seismic monitoring. In addition, InSAR displacement time series have been used to analyze the length of sinkhole precursory times and their relationships to subsurface sediment properties, hydrological conditions, and human activities. Our system has been able to detect precursory subsidence before the collapse of most sinkholes along the DS, and can be applied to the same use in similar geologic settings. However, before implementing this procedure for near-real-time early warning, it is important that the limitations of the InSAR technique are fully understood such as: loss of coherence, the role of accurate DEM and the tradeoffs between subsidence rate, size and the SAR platform properties (revisit time, wavelength, swath etc.).
Authors: Nof, Ran Novitsky (1); Abelson, Meir (1); Raz, Eli (2); Magen, Yochay (1,3); Atzori, Simone (4); Salvi, Stefano (4); Baer, Gidon (1)The Fels Glacier Slide is an actively deforming deep-seated gravitational mass-movement occupying the northern slope of the west-facing Fels Glacier Valley. The slide is situated in the east-central Alaska Range, about 5 km east of the Richardson Highway- and Trans-Alaska Pipeline corridor; about 3 km north of where it crosses the Denali fault. The proximity to critical infrastructure creates a need for pro-active detailed observation and modeling of the slide motion. Ground-based SAR interferometry (InSAR) has been widely used with demonstrated success to monitor landslides and slope instabilities in both man-made and natural terrains [1]. Advantages over spaceborne InSAR include much higher temporal resolution (minutes vs days), as well as the potential to use several line-of-sight (LOS) observation geometries simultaneously, to allow for accurate capturing of the full, three-dimensional motion of landslides. We demonstrate both advantages for the westernmost and most active part of the Fels Glacier Slide by acquiring terrestrial InSAR data with multiple GAMMA Remote Sensing Portable Radar Interferometer (GPRI) units at 5-minute intervals from station locations at the opposing southern slope of the Fels Glacier valley. As the slope deformation is gravity driven [2], surface displacement will be constrained approximately to the falline-up plane, which allows to deduce three-dimensional motion vectors from just two optimally chosen station locations. In this paper we present results derived from data acquired with this two-station configuration for 3-day field campaigns, in July 2017 and July 2018, respectively [3]. Our InSAR analysis of the final displacement field vectors used differential interferograms with three qualitatively different temporal baseline configurations. Namely, interferograms generated from 1) simultaneous acquisitions of the upper and lower receive channels of the GPRI, 2) acquisitions captured at 120-minute intervals from a single channel, and 3) acquisitions captured over a full day for each station. The interferograms derived from simultaneous acquisitions for each station were used [4] to generate a digital elevation model (DEM) that is co-aligned with the radar geometry. The 120-minute and 24-hour interferograms were used to derive the short-term and daily deformation rates in the LOS direction of the sensor (again for each station separately). The InSAR derived DEMs from each station were coregistered and georeferenced to a geodetic reference system using an external LiDAR DEM, captured in 2014. The refined coregistration and geocoding parameters were extracted from a series of key points from the InSAR and reference DEM using the Scale-invariant Feature Transform (SIFT) algorithm. Once matched, the 3D coordinates of the key point pairs were used to retrieve the geocoding/transformation parameters by a Broyden–Fletcher–Goldfarb–Shanno (BFGS) optimization. To ensure reliable deformation rate information for the Fels Glacier Slide, it was essential to estimate and eliminate atmospheric phase contributions from the differential interferograms. This was done by modeling the expected atmospheric phase contributions due to water vapor change in the LOS as a function of range and elevation of each pixel. After compensating for the atmospheric phase, the refined interferograms were characterized by complex high fringe density patterns due to surface displacement (for longer temporal baselines) as well as numerous small regions of incoherence mainly due to local shadow and to a lesser extent surface water in the slope[5]. Particularly, for the full-day differential interferograms this resulted in a failure of standard unwrapping algorithms to produce high-quality unwrapped phase estimates. An alternative demodulation approach was used instead that was based on the up-scaling of highly coherent, short-term (120-minute) interferograms with less than one fringe, to match the temporal baselines of the daily interferograms. As up-scaling also amplifies noise, the scaled short-term interferograms were stacked, averaged and smoothed to create a modeled full-day reference interferogram. The daily differential phase interferograms were then demodulated using this modelled interferogram, resulting in a (reduced fringe density) wrapped residual. The residual was then smoothed in the complex domain, unwrapped using a standard Minimum Cost Flow (MCF) algorithm, and added to the modeled interferogram to produce a new iteration of the modelled daily differential phase. This process was repeated, gradually reducing the smoothing window for the residual, until a residual was recovered at full resolution and within a single fringe. This final unfiltered residual was then added to the modeled phase to produce unwrapped daily differential interferograms (for both stations). Finally, 3D decomposition of the measured deformation from the two LOS directions was carried out under the assumption that the component of the slope displacement parallel to elevation contours is negligible. The resulting deformation maps revealed a discontinuous deformation vector field over the observed active lobe of the slide. The maximum displacement was observed at the exposed toe of the lobe, where average deformation rates of 8 cm/day and 3 cm/day was measured in downslope and emergent direction, respectively. In contrast, deformation was spatially variable and widespread but
Authors: Hosseini, Farnoush (1); Wooster, Nicholas (1); Rabus, Bernhard (1); Meyer, Franz (2)The former coal mining area in South-Limburg, The Netherlands, hosts potentially hundreds of vertical shafts and subsurface voids, of which some tens are known in location, but many are unknown. These voids pose a risk of collapsing, causing local sinkholes, and are a security hazard for the population and infrastructure. It is very difficult to know if, when, where, and how these voids may lead to surface collapse, since the frequency of occurrence is low and the signs at the surface are often very limited. Various in situ methods such as coring, GPR and seismic reflection are able to detect impending sinkholes but are expensive, spatially limited, and do not take the temporal changes into account. A novel technique and method has to be implemented and developed in order systematically look for a signature corresponding to the hazard of interest. Satellite radar interferometry (InSAR) has proven to be able, in ex post facto studies, to detect the minute displacements that appear to occur in the weeks or years preceding a collapse. Yet, until now, there has been no systematic approach of automatic screening and analyzing the vast amount of satellite data that is acquired on a daily basis. Here we demonstrate the feasibility of such a detection procedure and define the relevant metrics to decide on the efficacy of such an impending sinkhole detection system. We systematically analyze the entire area of interest using a predefined kinematic model. The kinematic model is a spatio-temporal simplification of the surface expression of an impending sinkhole, whereby multiple InSAR measurement can be used for the estimation. By implementing a kinematic model, the flexibility is preserved to apply the method on other areas of interest. The method takes the spatial distribution of InSAR measurements from five different satellites into account, and considers the different precision levels of each individual measurement point. The precision levels are important for estimating the minimal detectable deformation parameters, given a selected significance level and detectability power. This leads to a near-daily update of the entire area of interest. The produced maps can be used in an pre-operational way by authorities and stakeholders in Limburg, but could also be produced for any other sinkhole-prone area in the world.
Authors: Felius, Max A.; Hanssen, Ramon F.; Van Leijen, Freek J.The Dead Sea (DS) shore salt karst is one of the most rapidly evolving karst systems worldwide, offering a rare opportunity to observe and study karst processes and geomorphological changes in real time and draw analogies with other, lower-rate karst systems. During the past decade, the DS karst has dramatically developed due to recharge of flash floods into existing and newly formed streambed sinkholes. The process is self-accelerating as salt dissolution below the sinkholes enhances subsidence and additional sinkhole formation, which in turn increase the ponding areas for floodwater and generate additional draining conduits to the subsurface. Using annual Lidar DEMs, InSAR phase and coherence changes, drone photography, and time lapse cameras we document and analyze the onset and progression of this activity in the Ze’elim and Hever fans, southwestern DS. We find a positive correlation between annual subsidence at the Ze’elim fan and the corresponding total floodwater volume measured at a hydrometric station upstream. InSAR phase measurements detect an abrupt increase in subsidence rate around the recharging sinkholes after each flood event followed by a quasi-exponential decay of subsidence rate over periods of several months, and trace subsurface conduits that connect the recharging streambed sinkholes with discharging shoreline sinkholes. The surface paths of flash floods in the braided channels upstream of the recharging sinkholes are detected in the respective coherence pairs, and their relative volumetric flow rates are approximated by the average coherence loss along the flowing channels. Calibration of the coherence with direct hydrometric measurements of flow rates and volumes and with flood simulations enables a semi-quantitative estimation of the flow volumes directly from the coherence maps. Of the 15 flow channels mapped in 2011 across the Ze’elim alluvial fan, only one channel still allows surface flow of floodwater directly into the DS, while all others have already disappeared underground through sinkholes. In Hever fan and in most other alluvial fans and streambeds, some floodwater are also recharged into sinkholes, however, surface runoff is still the dominating mode of flow to the DS. This difference is attributed to the fine-grained sediments and low-gradient stream profiles in Ze’elim, which enable water accumulation and recharge in sinkholes, in contrast with the higher-gradient profiles in other riverbeds, which carry coarser sediments that eventually fill the sinkholes. Further examination of their gradients and lithology onshore and offshore may indicate whether their expected flow mode will be sinking or runoff. This will imply on their future incision patterns and their associated hazards.
Authors: Baer, Gidon (1); Nof, Ran N (1); Swaed, Iyad (1); Ward, Steven N (2); Gavrieli, Ittai (1)It is generally forecasted that as the climate warms, there will be increased frequency and magnitudes of extreme weather events presenting risk to human lives and property. In order to mitigate these risks, it is essential to develop and maintain infrastructure to increase the resilience of communities susceptible to natural disasters. The Dallas-Forth Worth metroplex (DFM), Texas, is one of the largest urban areas in the United States. Founded along the Trinity River, 80% of the city is covered in residual soils and expansive clays. Coupled with these geologic conditions, frequent flooding and development of oil and gas surrounding the city create a dynamic terrain system that affects infrastructure. Persistent Scatterer Interferometry (PSI) is a proven method for the monitoring of time varying deformation rates in urban infrastructure. This project uses 97 ascending orbit Interferometric Wide Swath (IW) scenes acquired by the Sentinel-1 mission from December 2015 to November 2019 for a PSI analysis over the DFM. Over 3 million persistent scatterers were identified across the two subswaths processed, predominantly located within the concentration of buildings in the urban centres and along large infrastructure such as highways and dams. For validation, time series of the vertical positions of 19 GNSS permanent stations operated by the Texas Department of Transportation and the Federal Aviation Administration were acquired from the Nevada Geodetic Laboratory. The regional representativeness of scatterers may vary depending on the type of infrastructure. Two buildings located near each other with similar underlying geology may have different settlement rates depending on foundation design and loading. This may be useful for the monitoring of infrastructure settlement, however for larger scale deformation processes this causes decorrelation between the time series generated from the scatterers and the larger scale trend. Infrastructure such as highways and roads present an opportunity to mitigate this error as they offer targets with relatively consistent foundation design and loading across transects of the metroplex. In order to separate the scatterers according to the type of infrastructure, a LiDAR based ground cover classification along with a 12m resolution TanDEM-X DEM was used to group scatterers into classes. These classes are independently analyzed for local versus regional deformation trends. Results show the potential for identifying increasing flood risk of communities in the DFM and future planning for flood mitigation infrastructure.
Authors: Branson, Andrew Robert; Iribe-Gonzalez, Daniela; Braun, Alexander; Fotopoulos, GeorgiaThe impact of climate change is increasing the frequency and severity of extreme weather events which threatens the integrity and sustainability of transportation infrastructure. Highway bridges are a critical component to Canada’s transportation network and require modern monitoring technologies to provide warning against potential failures and to identify aging infrastructure. Synthetic aperture radar interferometry (InSAR) is a useful tool to monitor bridges since it provides high resolution measurements at regular time intervals over large areas. In this work we evaluated the feasibility of using InSAR to systematically monitor highway bridges. For this, four Canadian bridges with different structural systems, material properties and dimensions were analyzed using 2 to 3 years of SAR data and compared against theoretical predictions. High resolution RADARSAT-2 (Spotlight or Ultra-Fine modes) image stacks from opposing look directions were acquired over the North Channel Bridge (Ontario), Confederation Bridge (PEI), Jacques-Cartier and Victoria Bridges (Quebec) to measure height, displacement rate and thermal sensitivity. Stacks from opposite viewing directions allowed us to compute 2D deformation maps (longitudinal + vertical components) and validate measurements derived from each stack independently. We found that a detailed backscatter analysis of the bridge based on known geometry information from the bridge construction drawings is essential to understand which bridge elements are represented by given coherent targets during the InSAR analysis. Additionally, alternative processing strategies were investigated to improve bridge geo-localization and to remove baseline and atmospheric phase residuals at scenes covered by large water bodies. InSAR results have shown promise to detect seasonal deformation and any potential hazardous long-term deformation trend. In general, agreement between InSAR observations and theoretical models was good for bridges that contain multiple bridge elements with stable phase response. However integration of ascending and descending InSAR measurements at bridges with complex structures is challenging, leading to residual measurements that can’t be explained by theoretical models. Results for the Confederation Bridge were unsatisfactory due to several factors such as bridge dimensions (extended length + relatively narrow cross section), limited number of bridge elements giving coherent targets, limited land coverage in the satellite image, and the fact that the bridge signal is in layover with that of the ocean water. As a follow on, we conducted a two-phase study to assess the feasibility of using small corner reflectors to improve monitoring over the Confederation and North Channel bridges. First, we investigated potential locations to install corner reflectors of different sizes based in multiple SAR parameters such as signal to clutter ratio, viewing geometry, resolution and contextual structural information derived from bridge construction drawings. Then, two corner reflectors were installed at optimal locations at each bridge and evaluated its performance using a stack of images acquired over ~6 months. Results showed good performance of Radar Cross Section, high spectral diversity and stable phase during the entire monitoring period, with minimal impact during winter and icy conditions. It can be concluded that bridges with few coherent/phase-stable targets like the Confederation Bridge could substantially benefit from the use of corner reflectors for more accurate monitoring of bridge deformation.
Authors: Greene Gondi, Fernando (1); Cusson, Daniel (2); Eppler, Jayson (1)The 4 year-long Recovery Observatory project (https://www.recovery-observatory.org) was triggered by the Committee on Earth Observation Satellites (CEOS) to define a sustainable vision for increased use of satellite EO in support of recovery after 2016 Hurricane Matthew struck southwestern Haiti. ESA’s Geohazards Exploitation Platform (GEP) [1] was used to develop a workflow based on big Synthetic Aperture Radar (SAR) data to access, process and generate value-added products that Haitian end-users can use to support their decision-making processes and recovery progress monitoring in the most affected departments, i.e. Grand’Anse, Sud and Nippes. Copernicus Sentinel-1 IW scenes were processed with SNAP and SNAC tools to generate change detection products, such as Interferometric SAR (InSAR) coherence and amplitude change maps, allowing the identification of flooded areas and landscape changes due to urbanization and other surface processes. Advanced InSAR ground deformation products generated with the Persistent Scatterers InSAR (PS-InSAR) FASTVEL tool [2] developed by TRE-ALTAMIRA and the Parallel Small BAseline Subset (P-SBAS) tool [3] by CNR-IREA provided input to map unstable areas at new housing built on unstable ground west of the town of Jérémie and along its western coastline, which highlight potential concern for urban development and reconstruction. The use of Sentinel-1 data combined with tailored monitoring campaigns at high and very high resolution with DLR’s 3 m resolution TerraSAR-X StripMap and ASI’s 1 m resolution COSMO-SkyMed Enhanced SpotLight SAR imagery allowed national to local scale coverage of the Recovery Observatory project area, with medium to very high spatial resolution, and up to weekly site revisit. Multi-sensor and multi-scale geohazard information derived from SAR data were analyzed to enhance the knowledge on surface processes occurring in Haiti, in the context of the field evidence and ground truth gathered by ASI, CNES and CNIGS during the technical surveys carried out in 2019. While the whole SAR image processing workflow could have been done on local processors, one of the crucial goals of the Recovery Observatory project team was to design a data analysis chain working on the GEP with hosted processing services and resources, and therefore with the characteristic of being fully transferrable to the Haitian partners [4]. To this aim, CNES also coordinated a capacity building plan, including effective knowledge transfer of SAR and optical image handling and interpretation skills to the Haitian partners. REFERENCES [1] Foumelis M., Papadopoulou T., Bally P., Pacini F., Provost F., Patruno J. 2019. Monitoring Geohazards Using On-Demand and Systematic Services on Esa’s Geohazards Exploitation Platform. Proceedings of the IEEE International Geoscience and Remote Sensing Symposium - IGARSS 2019, 5457-5460. [2] Ferretti A., Prati C., Rocca F. 2001. Permanent scatterers in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 39, 8-20. [3] Casu F., Elefante S., Imperatore P., Zinno I., Manunta M., De Luca C., Lanari R. 2014. SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 7, 3285-3296. [4] Cigna F., Tapete D., Danzeglocke J., Bally Ph., Cuccu R., Papadopoulou T., Caumont H., Collet A., de Boissezon H., Eddy A., Piard B.E. 2020. Supporting recovery after 2016 Hurricane Matthew in Haiti with big SAR data processing in the Geohazards Exploitation Platform (GEP). Proc. IGARSS 2020, 26 Sept - 2 Oct 2020, Waikoloa, Hawaii, USA. pp. 6867-6870. doi:10.1109/IGARSS39084.2020.9323231
Authors: Cigna, Francesca (1); Tapete, Deodato (1); Danzeglocke, Jens (2); Bally, Philippe (3); Cuccu, Roberto (4,5); Papadopoulou, Theodora (6); Caumont, Hervé (7); Collet, Agwilh (8); de Boissezon, Helene (8); Eddy, Andrew (9); Piard, Boby E. (10)Geotechnical engineering and construction in permafrost environment can be challenging. Especially, when the thermal regime of the building/soil interface crosses the freezing point multiple times per year, thus imposing important changes in the strength and bearing capacity of the soil. Soils in permafrost environments might act unfavourably towards the structures. The important processes affecting soil-structure interaction are (1) frost heave and thaw settlement, (2) reduction/increase of bearing capacity of thawed/frozen state. Those effects and their dependency on soil parameters such as granulometry, water content, thermal regime etc. are well known on the scale of soil samples from thorough laboratory testing. Although the physical properties of the geotechnical factors are known to a level onto which stable and durable construction is possible as well in permafrost environments, in reality stability problems of constructions in permafrost are wide-spread. There are many examples of affected buildings, linear infrastructure (power lines, pipelines, roads) and airports in the Arctic that show beginning signs of structure weakening or even complete failure due to an increased frost action or a drastic reduction of the bearing capacity of the soil. Some of the damages result of self-inflicted changes of the in-situ thermal regime by e.g. an insufficient thermal insulation of the heated construction against the frozen ground. Another factor results from the difficulty in the proper dimensioning of the geotechnical parameters in Arctic environment is often caused by the complicated geotechnical survey situation. Complex logistics and bad weather (storms, snowfalls) during the ideal time for the soil exploration as well as the overall costs and timing for the exploration are not beneficial for the proper dimensioning. Moreover, long term stable constructions might now enter a stage where frost actions get severe due to a climatically driven increase of the Active Layer Thickness (ALT). Foundations formerly embedded underneath the permafrost table might now lie within the active layer that suffers from seasonal thaw/refreezing and thus the soil suffers from thaw weakening and a reduced bearing capacity. As a result (differential) settlements are acting on the construction. To generate a regional overview over the current frost action situation, remote sensing techniques would be most efficient. With satellite InSAR offering new perspectives with the currently large amount of data from different sensors and a worldwide acquisition scenario (e.g. Sentinel-1) available, it is possible to track the seasonal heave/settlement of freezing/thawing soil to a level, where an estimation of useful factors for the geotechnical engineering might be deduced. However, from the remote sensing perspective, there exist a still relevant number of open questions: e.g. is a coverage of the complete freeze-thaw cycle possible even due to snow/meltwater cover? How can the measured ground movement amplitudes and rates be translated in terms of soil conditions and compositions? To what level of detail must an envisaged “frost susceptibility map” be brought in order to show its usefulness in the daily planning a construction business in the Arctic community. In the ongoing ESA AALM4INFRAM project, that we elaborate jointly with important Greenlandic stakeholders, a methodological approach combining satellite based remote sensing together with in-situ data from field surveys is used to: (1) map from Sentinel-1 InSAR data the current amplitude of seasonal frost/thaw cycle induced ground movements, (2) model ALT over specific area from optical/multispectral imagery and ground information data (3) possibly derive the frost susceptibility maps for this regions. With widespread permafrost degradation likely to occur in the future, we aim to estimate as well the degradation of the permafrost soils and potential hazards to the ground and bearing structures related to frost susceptibility. In our contribution we will outline the ESA AALM4INFRAM project, the questions already answered and the questions that are still open after first results from the Ilulissat (W Greenland) sample test-site.
Authors: Caduff, Rafael (1); Ingeman-Nielsen, Thomas (2); Mätzler, Eva (3); Scheer, Johanna (2); Strozzi, Tazio (1); How, Penelope (3)Offshore platforms are critical infrastructure components of the oil and gas industry. Over time, the operational safety and longevity of offshore installations may become compromised, due to extraction activities and natural processes, such as weathering and self-weight-induced settlements. The suitability of Sentinel-1 data for monitoring of permanent topside facilities was evaluated as a potential cost-effective add-on or alternative to conventional geodetic monitoring techniques, such as differential GNSS measurements. The first offshore case study that was selected was the Adriatic LNG Terminal, the world's first gravity-based offshore liquid natural gas production facility, located approximately 15 km from the Veneto coastline. To test the viability of an InSAR-based monitoring method, a Persistent Scatterers approach was implemented for stacks of Sentinel-1 scenes acquired from both ascending and descending orbit geometries over the same nine-month period. The Adriatic LNG Terminal is an isolated target, surrounded entirely by water, which presented technical challenges for the analyses, especially for co-registration of the SLCs and for geocoding of the results. While preserving only those scatterers with a temporal coherence threshold exceeding 0.9, the analyses detected between 44 and 109 individual scatterers with mean vertical settlement velocities between 9 mm and 28 mm per year and a mean vertical precision of less than 2 mm per year. The estimated settlement values varied between analyses performed with ascending and those performed with descending orbit datasets but were generally in agreement with measured GNSS trends. A similar set of multi-orbit analyses were performed for two double-platform installations located in the Azeri-Chirag-Gunashli oil field complex, approximately 120 km off the coast of Azerbaijan in the Caspian Sea. A stack of approximately 100 images each, for ascending and descending orbit paths, were processed for the period between 2015 and 2020. The results revealed that scatterer coverage was highly dependent on the geometry of the structures, while the resolution did not allow pinpointing of individual scatterers on the substructures. Moreover, in both cases, the westerly substructure cast radarshadow onto the easterly substructure in the ascending orbit. Without reference points, the geocoding of results was not precise enough for decomposition of horizontal and vertical deformation components. Furthermore, GNSS measurements were not available for either of the installations. While the described PS-InSAR technique should be further tested to assess the performance and to determine the precision and the accuracy of the results, the applicability of the methodology was demonstrated. In the future, the technique may be applied to monitor other fixed offshore platforms and gravity-based structures for which long-term subsidence monitoring is desired.
Authors: Salazar, Sean E.; Vöge, Malte; Frauenfelder, RegulaIn fast growing cities, tunnels are increasingly adopted solutions to meet the demand for more effective transportation. As settlements caused by tunnel excavations can damage buildings along the tunnel alignment, a large portion of investments in underground construction projects is typically devoted to the assessment of settlement-induced damage to buildings. To contain the project costs, only a limited number of buildings is usually included in the monitoring scheme, and therefore damage assessment procedures are traditionally based on highly conservative assumptions. Modern space-borne Synthetic Aperture Radar (SAR) missions can provide monitoring data over large areas, guaranteeing high spatial resolutions and short revisit times. Persistent Scatterer Interferometry (PSI) [1,2] can be used to extract building deformations over time from long temporal series of InSAR images, providing measurements with an accuracy comparable to traditional in-situ monitoring, i.e. of the order of millimetre, and at a much lower cost. However, without an integration with structural models, PS-InSAR data cannot provide meaningful information on the building conditions. This integration is particularly demanding for large excavation projects, where hundreds of buildings need to be assessed. In this research, we present a new methodology for the integration of PS-InSAR-based building deformations within damage assessment procedures to estimate the level of vulnerability of buildings adjacent to tunnel excavations. The methodology combines in an automated workflow PS-InSAR data, GIS (Geographical Information System)-building databases and semi-empirical models of the building response to tunnelling, to provide a more accurate estimate of each structure damage level. We tested the proposed methodology on the Crossrail tunnel alignment in London, UK. Crossrail tunnelling activities started in May 2012, and resulted in the excavation of 21 km twin tunnels below central London. We used as an input historical PS-InSAR data obtained by processing 72 COSMO-SkyMed descending images from 2011 to 2015 [3]. The processing led to the identification of 228,000 PSs over the monitored area, which correspond to an average density of about 9000 PS/km2. The map in Figure 1 shows the distribution of cumulative displacements along the Crossrail tunnel alignment, revealing the settlement caused by the excavation. In the region above the tunnels, line of sight (LOS) displacements between -2 cm and -3.5 cm were observed. PS points were automatically associated to the buildings along the tunnel route, and for each building, the corresponding PS-InSAR-based displacements were used to estimate the actual building settlement profile, using the fitting model described in Giardina et al., 2019 [4]. Figure 2 shows an example of a specific building, for which the PS-InSAR measurements were used to reconstruct the settlement below the structure. Then, the actual building settlement curves were analysed through a semi-empirical model of the building response to tunnelling [5] to estimate the maximum building strains. On the basis of its maximum strain, a level of damage was assigned to each building, and damage maps showing the distribution of building damage levels were the output of the proposed methodology (Figure 3). The developed algorithm enabled the identification of the structural damage of 858 buildings, highlighting its capability as a city-scale assessment tool. Additionally, the application of the proposed algorithm made available for the first time a large dataset of field observations of the building response to tunnelling. This allowed the identification of relationships between building construction materials, foundation typologies and global building behaviour. The findings can help improving current damage assessment procedures and advance the understanding of building response to tunnelling, with an impact on future excavation projects all over the world. References: Ferretti, A., Prati, C. and Rocca, F., 2000. Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry. IEEE Transactions on geoscience and remote sensing, 38(5), pp.2202-2212. Ferretti, A., Prati, C. and Rocca, F., 2001. Permanent scatterers in SAR interferometry. IEEE Transactions on geoscience and remote sensing, 39(1), pp.8-20. Milillo, P., Giardina, G., DeJong, M.J., Perissin, D. and Milillo, G., 2018. Multi-temporal InSAR structural damage assessment: The London crossrail case study. Remote Sensing, 10(2), p.287. doi: 10.3390/rs10020287 Giardina, G., Milillo, P., DeJong, M.J., Perissin, D. and Milillo, G., 2019. Evaluation of InSAR monitoring data for post‐tunnelling settlement damage assessment. Structural Control and Health Monitoring, 26(2), p.e2285. doi: 10.1002/stc.2285 Burland, J.B., Mair, R.J. and Standing, R.N., 2004. Ground performance and building response due to tunnelling. In: Jardine, R.J., Potts, D.M., Higgins, K.G. (Eds.), Conference on Advances in Geotechnical Engineering, vol. 1. Institution of Civil Engineers, pp.291–342.
Authors: Macchiarulo, Valentina (1); Milillo, Pietro (2,3); DeJong, Matthew J. (4); Giardina, Giorgia (5)In April 2019, large parts of Khuzestan province in Iran were affected by intense record rainfall in the Zagros mountains. Persian Gulf catchment received approximately 30% of its long-term average rainfall over the course of a few days. As several dams along Karkheh and Dez rivers, two major rivers in this catchment, reached their storage limits the water had to be released from the reservoirs, which resulted in flooding downstream of the dams. Several cities and more than 200 villages were flooded and many people had to be evacuated. Many of the dams affected by the 2019 flood were embankment dams, previously reported to exhibit post-construction settlements, at places reaching 13 cm/yr. Therefore, during and after the flood, significant concerns were raised about their health and stability because a dam failure poses a significant hazard with tremendous loss of life and properties. In this study, we use Sentinel-1 InSAR to monitor the response of embankment dams in Khuzestan to the 2019 flood event. We process the full archive of Sentinel-1 using the Small Baseline Subset (SBAS) approach and estimate the time series of displacement for three different embankment dams in Khuzestan province. The first two analyzed cases are Karkheh and Gotvand, which have the largest capacities in the country and became operational in 2001 and 2012, respectively. The third studied dam is the Masjed-Soleyman dam, previously reported to sustain a high rate of displacement since its operation in 2002. The Sentinel-1 InSAR displacement results indicate that all observed dams exhibit long-term cm-scale post-construction settlement before the flood. The time series of displacement for Karkheh and Gotvand dams show gentle changes of displacement in response to the increase in water level following the flood. For the Masjed-Soleyman dam, however, the movement accelerates sharply after the flood with more than 2 cm of displacement on the crest in only two months. For the Masjed-Soleyman dam that experiences the most severe effect of the flood, we also analyzed high-resolution data from TerraSAR-X and COSMO-SkyMed. The results provide a detailed picture of the displacement pattern over the crest and the body of the dam before and after the flood.
Authors: Haghshenas Haghighi, Mahmud (1); Motagh, Mahdi (1,2)Key words: Sea crossing bridge; Structural health monitoring; Time-series InSAR; Structural principle; Thermal dilation Motivation: Sea crossing bridge can cross bays, straits, and deep marine water, which are critical nodes to ensure the smooth flow of traffic arteries on the mainland. Even subtle displacements may affect the performance of bridges and cause high maintenance and repair costs. Thus, the timely and accurate inspection of their structural safety status is essential to guarantee the smooth operation of public transportation, and to avoid personal and property loss[1-2]. However, the traditional bridge deformation monitoring sensors such as the total stations, accelerometers, leveling and strain-meters are only available at sparse discrete locations on a few number of bridges, with a limited spatial extent or low temporal sampling frequency due to the constraints on manpower and financial costs[3-4]. As sea crossing bridge are usually huge in size and complex in structure, the above methods would inevitably miss some security risks and cannot explain the global deformation of the bridge. On the contrary, the Synthetic Aperture Radar Interferometry (InSAR) technique can quickly extract surface deformation over a large area at fine resolution, without the limits of geographical and meteorological conditions[5-7]. It has unique advantages of low labor and material costs, and no effects on bridge normal operation, which allows the timely monitoring of displacement with a dense grid of measurement points, showing huge application potential in bridge structural health monitoring. Current Issues: Although previous studies on InSAR bridge deformation monitoring have achieved certain research results on thermal dilation monitoring and deformation interpretation[8-16], there are still some issues that need to be solved urgently to achieve the structural health monitoring of sea crossing bridge. First of all, the previous research mainly focused on high coherence bridge such as steel bridge. The density and accuracy of point targets (PTs) is low on structures with significant de-coherence effects, leaving how to extract dense and accurate PTs upon partially coherent sea crossing bridge still a problem. Moreover, previous thermal dilation monitoring relies on the linear deformation assumptions of PTs, or a large amount of high-resolution images and calculations, rather than combining the structural characteristics. When the above conditions are not true, it is not appropriate for long-span sea crossing bridge with long thermal dilation propagation distance. Therefore, how to accurately model and separate the thermal dilation of sea crossing bridge is still a challenge. Finally, due to the limit of SAR side-looking imaging geometry, the traditional InSAR results are difficult for non-expert users to comprehend and interpret. The simple analysis of bridge linear deformation map is meaningless for long-span sea crossing bridge with complex deformation patterns, making the InSAR results, currently, still have a certain gap from the engineering applications. Methodology: Aiming at the above issues, the goal of this study is to develop a measurement and analysis method to detect the global deformation of sea crossing bridge quickly and accurately by introducing the structural principle into the conventional time-series InSAR analysis. As for the PTs selection, the PSI and SBAS algorithms have been applied to extract both traditional point targets, i.e., targets on the bridge outline that contains one dominant scatterer, as well as extended targets, i.e., targets on bridge beam that spread over a collection of pixels. Structural semantic information is then introduced for the fine match recognition and posterior screening of the structural PTs. In terms of thermal dilation modelling, the qualitative analysis of the thermal dilation distribution and propagation regulation is firstly implemented through the time-series analysis of interferometric and unwrapping phases. Then, a data-driven regression analysis method weighted by the coherence of interferometry pairs is applied to establish a quantitative relationship between the temperature and the displacement. Finally, the material properties of bridges are utilized to validate the estimated thermal dilation parameters. In order to provide user friendly results, the height of bridges and the local viewing geometry parameters are used to calculate the PTs 3D positions through a local coordinate orthorectification. Then, the structural principle and InSAR observations are integrated to investigate the potential risk sources and security points on the sea crossing bridge based on the 3D visualized results. The risk level of different sections along the sea crossing bridge can be evaluated based on the weighted calculation of the identified risk indicators. Study Objects and Datasets: The East Sea Bridge, connecting the Nanhui New Town in Shanghai and the Yangshan Town in Zhejiang Province, is 25.3 km long. It is a part of the key supporting projects in Yangshan Deepwater Port. As a distinguished sea crossing bridge, the Hangzhou Bay Bridge was built from 2003 to 2008 and connects Jiaxing and Ningbo city, with the total length of 35.7 km. A stack of 31 ascending Sentinel-1 images from 2015 to 2017 is collected for the deformation analysis of these two sea crossing bridges in this study. TerraSAR-X images in Shanghai are also used for the cross validation of Sentinel-1 results. Preliminary Results: Based on the InSAR and structural principle integration method, the detail deformation along the two sea crossing bridges during the observation time are estimated. After modelling and separating the thermal dilation along the bridge, the structural risk level is evaluated based on the analysis of risk indicators and structural security points. Security sections along the sea crossing bridges are identified, which should be focused on the future maintenance and management. The presented work is part of ongoing research activities in making InSAR techniques applicable for structural health monitoring of sea crossing bridge. The methodologies and outcomes of this study will promote the understanding of the catastrophic mechanisms and evolution of sea crossing bridge, as well as to provide near real-time monitoring results and reliable technical support for sea crossing bridge disaster prevention and daily maintenance. References: [1] Ismail Z, Ibrahim Z, Chao A, et al. 2012. Approach to reduce the limitations of model identification in damage detection using limited field data for nondestructive structural health monitoring of a cable-stayed bridge [J]. Journal of Bridge Engineering, 17(17): 867-875. [2] Fornaro, G., Reale, D., Verde, S. 2013. Bridge thermal dilation monitoring with millimeter sensitivity via multidimensional SAR imaging [J]. IEEE Geoscience and Remote Sensing Letters, 10(4): 677-681. [3] Osmanoğlu, B., Sunar, F., Wdowinski, S., Cabral-Cano, E. 2016. Time series analysis of InSAR data: Methods and trends [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 115: 90-102. [4] Qin X., Liao M., Zhang L., Yang M. 2017. Structural health and stability assessment of high-speed railways via thermal dilation mapping with time-series InSAR analysis [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 10(6): 2999-3010. [5] Zhang, L., Sun, Q., Hu, J. 2018. Potential of TCPInSAR in monitoring linear infrastructure with a small dataset of SAR images: application of the Donghai Bridge, China [J]. Applied Sciences, 8(3): 425. [6] Karimzadeh, S., Matsuoka, M., Ogushi, F. 2018. Spatiotemporal deformation patterns of the Lake Urmia Causeway as characterized by multisensory InSAR analysis [J]. Scientific Reports, 8(1): 5357. [7] Chang, L., Hanssen, R. 2014. Detection of cavity migration and sinkhole risk using radar interferometric time series [J]. Remote Sensing of Environment, 187: 49-61. [8] Crosetto, M., Monserrat, O., Cuevas-González, M., Devanthery, N., Luzi, G., Crippa, B. 2015. Measuring thermal expansion using X-band persistent scatterer interferometry [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 100: 84-91. [9] Lazecky, M., Hlavacova, I., Bakon, M., Sousa, J., Perissin, D., Patricio, G. 2017. Bridge displacements monitoring using space-borne X-Band SAR interferometry [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 10(1): 205-210. [10] Milillo, P., Giardina, G., Perissin, D., Milillo, G., Coletta, A., Terranova, C. 2019. Pre-collapse space geodetic observations of critical infrastructure: the Morandi Bridge, Genoa, Italy [J]. Remote Sensing, 11: 1-14. [11] Salvakumaran, S., Plank, S., Geiß, C., Rossi, C., Middleton, C. 2018. Remote monitoring to predict bridge scour failure using Interferometric Synthetic Aperture Radar (InSAR) stacking techniques [J]. International Journal of Applied Earth Observation and Geoinformation, 73: 463-470. [12] Jung, J., Kim, D., Vadivel, S., Yun, S. 2019. Long-term deflection monitoring for bridges using X and C-band time-series SAR interferometry [J]. Remote Sensing, 11: 1258. [13] Huang, Q., Crosetto, M., Monserrat, O., Crippa, B. 2017. Displacement monitoring and modelling of a high-speed railway bridge using C-band Sentinel-1 data [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 128: 204-211. [14] Qin, X., Zhang, L., Yang, M., Luo, H., Liao, M., Ding, X. 2018. Mapping surface deformation and thermal dilation of arch bridges by structure-driven multi-temporal DInSAR analysis [J]. Remote Sensing of Environment, 216: 71-90. [15] Qin, X., Ding, X., Liao, M., Zhang, L., Wang, C. 2019. A bridge-tailored multi-temporal DInSAR approach for remote exploration of deformation characteristics and mechanisms of complexly structured bridges [J]. ISPRS Journal of Photogrammetry and Remote Sensing. 156: 27-50. [16] Ma, P., Li, T., Fang, C., Lin, H. 2019. A tentative test for measuring the sub-millimeter settlement and uplift of a high-speed railway bridge using COSMO-SkyMed images [J]. ISPRS Journal of Photogrammetry and Remote Sensing, 155: 1-12.
Authors: Qin, Xiaoqiong (1,2,3); Wang, Chisheng (1,2,3); Li, Qingquan (1,2,3); Xiong, Siting (1,2,3)Sensing technologies with the ability to take frequent measurements provide repeatable data to monitor stability of infrastructures and manage risky conditions (Bennetts et al. 2016). InSAR is a remote sensing technique measuring land deformation regularly with millimeter accuracy and provides insights into infrastructure assets that are difficult to access or frequently monitor with conventional approaches. It can also be used for baseline assessment studies prior to construction by exploiting archived satellite imagery. Sentinel-1A and -1B, a two-satellite imaging radar constellation operated by European Space Agency (ESA), acquires 6-day/12-day images over Europe/Globe and tackles the limitation of cost by providing free data. Therefore, it increases opportunities to collect regular data. Validation of InSAR methods improves acceptability of the technique and can provide valuable input for InSAR development activities. Results of validation depend on various aspects including number and temporal sampling of the images, density and quality of the measurement points and distance from the reference point. Most of the validations are based on comparison of the InSAR products (deformation velocities and time series) with independent measurements of the same quantities (Crosetto et al., 2015). Several projects have validated InSAR velocities and time series. ESA conducted two major validation projects using ERS/Envisat data sets: PSIC4 (Raucoules et al., 2009), and the Terrafirma Validation Project, part of the EU/ESA Global Monitoring for Environment and Security (GMES) programme (Adam et al., 2009). The PSIC4 test area was a coal mining area in the South of France, which was undergoing rapid subsidence and did not include stable features. The Terrafirma validation project had two aims: result validation via comparison with ground truth levelling, and inter-comparison of the results of different InSAR providers. The case study was an urban area with zero or moderate rates of deformation. Many different InSAR time series approaches have been developed to extract deformation signals (Osmanoglu et al., 2016). Recently, we compared four different InSAR time series methods to process Sentinel-1 data (Sadeghi et al,. 2021): a modified RapidSAR algorithm (Spaans and Hooper, 2016), the PS-only option of StaMPS (Stanford Method for Persistent Scatterers) (Hooper et al., 2007), the SqueeSAR algorithm (Ferretti et al., 2011) and GAMMA-IPTA (Werner et al., 2003). The results showed that Sentinel-1 InSAR provides comparable results that are independent of processing approaches. However, there are considerable differences in some aspects of the results. The most substantial differences between the algorithms were related to the measurement density (varying from approximately 500 pixels/km2 to approximately 5600 pixels/km2 in an urban area). In this research we focus on the validation of Sentinel-1 InSAR products processed by the modified RapidSAR and conventional StaMPS algorithms to monitor infrastructure stability. The first case study is Hammersmith Flyover in London, UK. From 2010 to present, the bridge has been fitted with a number of different monitoring systems that test whether bridge bearings are able to move freely. Using structural modelling and understanding of the bridge behavior, the primary cause of deformation of Hammersmith Flyover was found to be thermal loading which causes expansion and contraction in the longitudinal direction of the bridge (Webb et al., 2014). For the purpose of this study, only those measurements providing direct information on the thermal expansion of the bridge are considered: specifically, we use data from temperature sensors embedded at 4 midspan locations, and displacement transducers (Linear Variable Differential Transformer or LVDT), whichmeasure linear displacement at bearing locations at the base of each of piers. To compare InSAR measurements with LVDT data, we use Sentinel-1 InSAR time series produced using two different approaches, modified RapidSAR and conventional StaMPS, using both ascending and descending geometries. We focus on one pier and select the corresponding selected InSAR pixels. The comparison of ascending and descending data in LOS direction, which have opposite signs, confirms that the ground mainly deforms horizontally. We convert the InSAR LOS deformation time series to the bridge parallel orientation and compare them with LVDT movement time series. We define the temporal and spatial references as the first common date and a stable area close to the bridge. The results confirm an excellent agreement between the LVDT and InSAR, even with different geometry of satellite imagery (ascending and descending) and independent InSAR time series algorithms (StaMPS and RapidSAR). The RMSE of the LVDT deformation data and the converted InSAR deformation data estimated by RapidSAR using ascending data, RapidSAR using descending data and StaMPS using ascending data are 3.2 mm, 3.09 mm and 4 mm, respectively. This confirms a high level of similarity between the deformation histories of the time series data and shows that all data sets are detecting similar deformation signature. The thermal dilation signature measured by ground movement shows a linear correlation with the temperature of the bridge at the time of measurements. Therefore, we define a thermal dilation term in the InSAR phase equation and estimate this from the InSAR time series. The estimated thermal dilation parameter is 1.07 mm/°C for LVDT data, 1.04 mm/°C for InSAR, which agree with each other very well. The second case study is at Bank station in London, UK. 1.3 km of tunnels have been constructed since 2017 as part of the modernization and expansion of Bank underground station. The movements induced by excavation activity had the potential to cause damage to the important structures including historic St Mary Abchurch and Mansion House. The structural response of the buildings was monitored by various independent approaches including 3D prisms during major tunneling works by Geocisa under the guidance of London Underground. The 3D prism data recorded for Mansion House building show significant deformation in both horizontal and vertical direction at the time of tunneling activity. In order to compare RapidSAR InSAR data with 3D prisms data, we convert the 3D data to the LOS direction of the ascending and descending geometry of processed Sentinel-1 data using the corresponding heading and incidence angles. Prior to the comparison, we corrected the geocoding errors in the InSAR data, which can be divided to the individual shifts and gross shifts. The low resolution DEM used in the InSAR processing (SRTM 30 m) failed to estimate the correct heights of the SAR pixels covering the buildings and the surrounding roads and introduced individual shifts in range direction in the geocoding step of InSAR processing. We solve the individual shifts for the measurement points using extracting the corresponding heights from LiDAR DSM as a high resolution elevation model in an iterative approach. The gross shift is assumed to be constant in range and azimuth direction and can be solved after individual shift correction. The results of the comparison confirm excellent matches between the InSAR data and 3D prisms installed on the Mansion House building. Using comparison with ground truth data collected by installed sensors in the structures, we showed the capability of InSAR techniques, which are able to provide a dense network of measurement points using Sentinel-1 data, to extract deformation signal accurately. For the next steps, we plan to compare the validation outputs with the results of the same InSAR techniques processed using high-resolution data sets such as TerraSAR-X. We also plan to investigate the agreement between the expected deformation from structural models and our InSAR deformation products. References: Bennetts, J., Vardanega, P.J., Taylor, C.A. and Denton, S.R., 2016. Bridge data - What do we collect and how do we use it?, Proceedings of the International Conference on Smart Infrastructure and Construction, 531–536. https://doi.org/10.1680/tfitsi.61279.531 Crosetto, M., Montserrat, O., Cuevas-González, M., Devanthéry, N., and Crippa, B., 2015. Persistent Scatterer Interferometry: A Review.” ISPRS Journal of Photogrammetry and Remote Sensing 115, 78–89. Raucoules, D., Bourgine, B., de Michele, M., Le Cozannet, G., Closset, L., Bremmer, C., Veldkamp, H., Tragheim, D., Bateson, L., Crosetto, M., Agudo, M., Engdahl, M., 2009. Validation and intercomparison of persistent Scatterers interferometry: PSIC4 project results. J. Appl. Geophys. 68 (3), 335–347. https://doi.org/10.1016/j Adam, N., Parizzi, A., Eineder, M., Crosetto, M., 2009. Practical persistent scatterer processing validation in the course of the Terrafirma project. J. Appl. Geophys. 69, 59–65. https://doi.org/10.1016/j.jappgeo.2009.07.002. Osmanoglu, ˘ B., Sunar, F., Wdowinski, S., Cabral-Cano, E., 2016. Time series analysis of InSAR data: methods and trends. ISPRS J. Photogramm. Remote Sens. 115, 90–102. https://doi.org/10.1016/j.isprsjprs.2015.10.003. Spaans, K., Hooper, A., 2016. InSAR processing for volcano monitoring and other nearreal time applications. J. Geophys. Res. Solid Earth, 121, 2947–2960. https://doi. org/10.1002/2015JB012752. Hooper, A., Segall, P., Zebker, H., 2007. Persistent scatterer interferometric synthetic aperture radar for crustal deformation analysis, with application to Volcan ´ Alcedo, Galapagos. ´ J. Geophys. Res. Solid Earth, 112, B7. https://doi.org/10.1029/ 2006JB004763 Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F., Rucci, A., 2011. A new algorithm for processing Interferometric data-stacks: SqueeSAR. IEEE Trans. Geosci. Remote Sens. 49, 3460–3470. https://doi.org/10.1109/TGRS.2011.2124465. Sadeghi, Z., Wright, T.J., Hooper, A.J., Jordan, C., Novellino, A., Bateson, L., Biggs, J., 2021, Benchmarking and inter-comparison of Sentinel-1 InSAR velocities and time series, Remote Sensing of Environment, 256, 112306, https://doi.org/10.1016/j.rse.2021.112306 Webb, G.T., Vardanega, P.J., Fidler, P.R.A., and Middleton C.R. 2014. Analysis of Structural Health Monitoring Data from Hammersmith Flyover, Journal of Bridge Engineering (ASCE), 19 (6), 1–11, https://doi.org/10.1061/(ASCE)BE.1943-5592.0000587. Werner, C., Wegmuller, U., Strozzi, T., and Wiesmann, A., "Interferometric point target analysis for deformation mapping," IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477), Toulouse, France, 2003, pp. 4362-4364 vol.7, doi: 10.1109/IGARSS.2003.1295516
Authors: Sadeghi, Zahra (1); Wright, Tim.J (1); Hooper, Andrew.J (1); Selvakumaran, Sivasakthy (2); Acikgoz, Sinan (3)InSAR principles and the underlying SAR technology were developed about forty years ago and swiftly found a home in academia where hundreds of PhDs and MS theses have steadily improved the data acquisition and processing as well as the data interpretation. Much of the academic interpretive work focuses on studies either of slow, long-term changes in ground elevation caused by changes in the shallow subsurface down to around 300 metres (1,000 ft) or on instabilities of man-made infrastructure such as roads, buildings and bridges. Meanwhile, the general public is becoming increasingly aware of InSAR because of the graphics seen almost instantly on-line or in newspaper articles that explain newsworthy natural disasters like landslides or earthquakes using InSAR displays of the fine details. The vast majority of these InSAR studies are of events or phenomena that occurred in the past and are accompanied by interpretations and explanations of past events. However, the real value of InSAR – the economic value –lies in its use in the commercial world where an understanding of past and present surface motion can be used to improve current operational work and to improve future outcomes. Commercial customers use InSAR across a wide range of industries from petroleum (upstream production to midstream production facilities and pipelines, to downstream refineries), to construction (stability of new buildings), and to infra-structure (roads, bridges, etc.). As the revisit times for satellites get reduced, PS pixel sizes get smaller and more numerous, data storage and computing in the cloud gets cheaper, the quality and quantity of what the customer receives increases and so does the reliability of the interpreted results. The result is a dramatic improvement in the ability to forecast future outcomes and adjust operational strategies. Three examples from the upstream petroleum industry are used to illustrate the use of InSAR and its commercial value to their operations. The first example is how InSAR-derived surface deformation is used to monitor the performance of oil reservoirs in several oil fields in the San Joaquin Valley of central California. In a nutshell, good petroleum industry practice is to inject the same volume of liquid into a reservoir as is being produced in order to maintain pressure in the reservoir and thus maintain production. When less is injected than produced, the reservoirs tend to compact by an amount that depends on their mechanical strength. When more is injected than produced, the reservoirs expand. These volume changes will get transmitted to the ground surface where they are seen by InSAR as minute amounts of surface uplift or subsidence. The small changes in elevation, in the range of millimetres/month and covering several to hundreds of hectares, are used by operators to manage their waterflood operations They get updated subsidence maps every 11 or 24 days, depending on the choice of satellite, and then adjust the flow rates of their injection wells in areas where the InSAR shows uplift (need less injection) or subsidence (need more injection). If subsidence is particularly severe and increasing injection is not feasible then production is cut back but this has an immediate negative financial effect. The second example shows how the same InSAR surface deformation work in the San Joaquin Valley is also used for EHS purposes. Here, InSAR is used to find precursors to ‘surface expressions’ which are caused when water or steam is injected under pressure into the reservoirs but breaks through natural geologic barriers and comes to the surface where it causes a variety of EHS problems. Although all injection of water and steam is intended to go into the reservoirs, there are many reasons for its escape towards the surface and eventual flow at the surface. Luckily, as the water or steam is moving through unconsolidated layers of sand and towards the surface it inflates the ground surface very slightly in a characteristic blister-like shape covering a small circular area of only a few hectares. Left unmitigated, the water, steam or oil causing the blister will quickly work its way to the surface so early identification of the blister is essential so that the causes of the liquid’s escape can be cured. Because the overburden gets pushed up relatively quickly, identification of a blister needs to be done in the processing stages so that the customer can be alerted as soon as possible. This gives the customer the maximum amount of time to find out where, why, and how the liquids are escaping so that they can fix the problem before it gets out of hand. The third example illustrates the use of InSAR to identify potential sink holes in the Permian basin of west Texas. Sink holes are problematic wherever they occur, we often read about sink holes in karstic areas where they suddenly swallow cars, houses and roads. However, the ground surface above a future sink hole typically subsides slowly for some time before the ground collapses catastrophically. This subsidence can be detected and measured by InSAR. The Permian basin of west Texas is the heart of the unconventional oil boom in North America but much of the area has a shallow zone of karstic evaporites just below the surface which are easily dissolved by moving groundwater. The sink holes develop somewhat randomly and before the current oil boom they generally affected only the road network. As pad drilling became the style for efficient development of the unconventional oil and gas, huge amounts of money got tied up in relatively small areas that had up to several dozen oil wells. Having a sink hole develop on the pad requires costly remedial measures and might also have severe EHS consequences. The result is that detecting a potential sink hole before clearing the pad to drill the wells and also finding areas for pads that are unlikely to develop sink holes has huge economic value. We are currently using InSAR to find sink holes and are characterizing their deformation style throughout their life cycle. The results show that they can generally be detected while they are still deep enough to cause no surface problems and early enough for drilling pads and access roads to be relocated before there are problems. The three examples for the upstream petroleum industry show the power of using past and present InSAR to look into the future and make operational decisions that have significant monetary and EHS-mitigation value. Other industries are also using InSAR to improve how they operate. As SAR satellites get more common, plus revisit time and data resolution improve, the more InSAR will move from academia into being a mainstream commercial technology.
Authors: Allan, Malcolm E.; Leezenberg, Pieter BasTime-series of geodetic data can be exploited to estimate the surface deformation due to natural and anthropogenic phenomena over long periods of time. In particular, the use of Synthetic Aperture Radar Interferometry (InSAR) allows to detect and measure slow tectonic signals such as interseismic strain accumulation over a specific area of interest. As strain accumulation can occur in response to active tectonic processes, e.g. when the seismogenic faults are locked during the interseismic period, the estimation of surface deformation and the long-term strain rate can be considered a useful approach for seismic hazard investigations. In this study, we used InSAR time series to estimate the ground deformation in Northeastern Italy. The area is tectonically active and characterized by a convergent regime due to the motion of the Adria plate in NNW direction towards Euroasia at a rate of ~ 2mm/yr. The Southeastern Alps and the Dinarides, located on the western and the eastern sector of the region respectively, are mainly characterized by the presence of active thrusts and strike-slip faults. According to the historical and the modern catalogs, the seismogenic faults located in the area of interest are able to generate Mw 5.5-6 earthquakes. The last one was the 1976 Mw 6.5 earthquake that occurred in Friuli, followed by a strong aftershock sequence. We used SAR images acquired by the Sentinel 1A/B satellites from ascending and descending orbits, collected in the 2015-2019 temporal interval. A single-master stack of interferograms was generated with the ESA SNAP software by using several scripts provided by Snap2StaMPS. The interferograms have been used as input to the StaMPS (Stanford Method for Persistent Scatterers) multi-temporal InSAR processing chain. Different processing settings were tested to ensure consistency of the results, in particular concerning the phase unwrapping step. After the application of a spatial-temporal filter, the resulting Line-of-Sight (LoS) deformation time series have been calibrated with GNSS measurements in Adria-fixed reference frame. The information provided by GNSS stations, homogenously distributed in the area of interest, has been collected in the same time interval (2015-2019). After the calibration, the ascending and descending LOS datasets were combined to obtain the vertical and the horizontal (east-west) deformation components, comparing the results with GNSS data. Geodetic data have been then analyzed to identify large-scale trends compatible with regional tectonics. Besides the meaningful negative vertical signal located on the plain and the on coastal zones due to the subsidence, a positive vertical signal can be observed toward the Alps, in the northern region of Veneto and Friuli-Venezia Giulia, with a rate of ~ 2mm/yr. Moreover, horizontal velocities with a rate of 1-2 mm/yr are observed close to main tectonic structures, especially in the eastern and the northwestern sector of the study area.
Authors: Areggi, Giulia (1); Merryman Boncori, John Peter (2); Pezzo, Giuseppe (3); Serpelloni, Enrico (4); Bonini, Lorenzo (1,3)ABSTRACT CNN, March 2018 reported that “A large crack of 15m deep and 15m wide, stretching several kilometres, appeared suddenly in Suswa area of south-western Kenya”. The news was widely reported as evidence of opening of African rift system and the progressive growth in geologic future (MNM, 2018; USA today, 2018; PBS, 2018; Forbes, 2018; The weather network, 2018). We studied the chasm and the surrounding region using differential interferometry technique to understand the ground deformation during this period. Sentinel S1A images acquired on March 03, 2018 to April 20, 2018 (4 Images) were processed to generate interferograms using SNAP Desktop software. The interferograms generated did not show displacement or ground deformation over the chasm area along the line of site. Further, no fault escarpments were observed in the optical data as the land is flat on either side of the crack. The Norok area received 200% more rainfall which eroded the ash mixed soil, which is visible in the post event optical image. A further analysis of satellite optical imageries acquired from 2010 to 2016 indicates that similar erosional features were created in a short span of along the Norok–Mai Mahiu highway suggesting that these flash flood events have caused erosion before. We interpreted that the chasm was formed by sudden subsurface erosion and subsidence, not by tectonic pull-apart along the active fault in East African graben system. Interestingly, we also observed a deformation fringe of ~5-7.5cm along the line of sight along the NW-SE fault during between 15th - 29th March and 29th March - 04th April, 2018 suggesting tectonic deformation in the region. We extended the study towards Eastern part of East African rift zone (EEAR) extending from the Afro-Arabian Rift System (AFAR) triple junction to the Kenyan rift valley to understand ground deformation pattern vis a vis precipitation or decadal scale tectonic deformation, where there are some interesting studies were carried out using GPS and INSAR. We observed ground deformation at many locations from southern part of EEAR - Arusha County, Tanzania to northern part of EEAR- Turkana in Kenya as under, i) The interferograms generated shows deformation of 5-7.5cm (LOS) along NE-SW direction between 15th of March 29th March and 29th March to 04th April, 2018 in Suswa rift. ii) A deformation of ~4 cm (LOS) is observed in circular pattern at several locations between 27 March to 08 April, 2018 in Arusha County, Tanzania. iii) A deformation of ~4 cm (LOS) is observed in circular pattern between 22 March to 03th April, 2018 at Mt. Kenya. iv) A deformation of ~2-4 cm (LOS) is observed along NE-SW direction between 22 March to 03th April, 2018 at Samburu County, Kenya. The observed deformations are unique as it does not accompany any seismic or volcanic activity in the region in spite of the occurrence of Suswa and Longonot volcanos and several rift parallel faults in theregion. The ground deformation can occur also due to ground water extraction, oil and hydrocarbons recovery, coal mining, landslide etc.; however, we think it is possibly due to the heavy precipitation during March, 2018 over EEAR region coupled with an episodic long-term aseismic ground relaxation associated with the stress build up from past earthquakes or the rift related tectonics in the region. The present study points towards the necessity of rigorous monitoring of ground deformation including DInSAR techniques to identify the zones of active deformation in the active rift region. Reference Sahadevan, D. K., Pandey, A. K., Malik, K., & Maisnam, D. (2019). Chasm at East African Suswa Rift: Possible Explanations. Journal of the Indian Society of Remote Sensing, 47(10), 1773-1780.
Authors: Sahadevan, Dinesh Kumar; Pandey, Anand KumarHow does fault slip follow an earthquake rupture front propagating faster than the local shear-wave velocity (i.e., at supershear speed)? How does a supershear rupture front pass through a geometrically complex fault system? Resolving the evolution of such complex earthquake ruptures is fundamental to our understanding of earthquake-source physics, but these events have not been well captured by conventional waveform inversions of observational data. Resolving earthquake source evolution that possibly involves supershear rupture in a geometrically complex fault system requires finite-fault inversion that is more flexible than conventional inversion schemes. Conventional inverse solutions have been stabilized by limiting the model space and decreasing the degree of freedom for slip vectors. However, these limitations are not necessarily physical requirements for representing source processes. Moreover, inappropriate assumptions about the fault geometry can increase modeling errors, produce non-unique final solutions, and make it difficult to interpret those solutions (Shimizu et al., 2020). We applied a new framework of finite-fault inversion to globally observed teleseismic waveforms and resolved both the spatiotemporal evolution of slip and the fault geometry of the 2018 Palu earthquake (moment magnitude 7.6) in Sulawesi, Indonesia. In our inversion formulation, we used five basis double-couple components of the potency-density tensor (Ampuero and Dahlen, 2005) to represent slip (Shimizu et al., 2020), where a priori assumptions of fault geometry for each subfault in the model space are not required; instead, fault geometry is resolved by our inversion. That is, we simultaneously resolved both the spatiotemporal evolution of slip and the fault geometry of the 2018 Palu earthquake. We then compared our source model with the Interferometric Synthetic Aperture Radar (InSAR) map of the surface rupture trace (Bao et al., 2019; Socquet et al., 2019), which shows bends near the epicenter and south of Palu Bay. We show that supershear rupture propagation for this event was sustained by transient slip stagnation and advancement as the rupture front passed through the geometrically complex fault system. This peculiar inchworm-like slip evolution was caused by the rupture front encountering fault bends with favorable and unfavorable orientations for rupture propagation. Our analysis also identified the possible existence of a fault junction beneath Palu Bay connecting an unmapped primary fault in northern Sulawesi with the Palu-Koro fault in the south. We propose that the overall persistence of supershear rupture propagation during the 2018 Palu earthquake was a response to the geometric complexity of the fault system, which was the key driver of the transient and episodic acceleration and deceleration of slip evolution. [References] Ampuero, J.-P. and Dahlen, F. A., 2005. Ambiguity of the Moment Tensor, Bull. Seism. Soc. Am., 95(2), 390, doi: 10.1785/0120040103. Bao, H., Ampuero, J.-P., Meng, L., Fielding, E. J., Liang, C., Milliner, C. W. D., Feng, T., and Huang, H., 2019. Early and persistent supershear rupture of the 2018 magnitude 7.5 Palu earthquake, Nat. Geosci., 12(3), 200–205, doi: 10.1038/s41561-018-0297-z. Shimizu, K., Yagi, Y., Okuwaki, R., and Fukahata, Y., 2020. Development of an inversion method to extract information on fault geometry from teleseismic data, Geophys. J. Int., 220(2), 1055–1065, doi: 10.1093/gji/ggz496. Socquet, A., Hollingsworth, J., Pathier, E., and Bouchon, M., 2019. Evidence of supershear during the 2018 magnitude 7.5 Palu earthquake from space geodesy, Nat. Geosci., 12(3), 192–199, doi: 10.1038/s41561-018-0296-0.
Authors: Okuwaki, Ryo (1,3,5); Hirano, Shiro (2); Yagi, Yuji (3); Shimizu, Kousuke (4)On 26 November 2019, at UTC 02:54:11 a.m., a Mw 6.2 earthquake shocked north-central Albania: the epicentre has been located 20 km North of the coastal city of Durrës, but it has been easily felt up to 400 km away. Moreover, it was the strongest affecting the Adriatic coast since the Mw 7.1 1971 Montenegro earthquake, causing more than 50 deaths and up to 3000 people injured. Interferometric Synthetic Aperture Radar (InSAR) can be considered as one of the most important techniques for imaging large surface deformation occurred during earthquakes. Furthermore, in addition to the estimation of surface displacement, results coming from InSAR processing are also preciously useful for the retrieval of relevant information such as seismogenic source parameters. This event was already analyzed by using Differential SAR Interferometry (DInSAR), but still some pending doubts remain concerning the deformation source characteristics. The retrieval of the best source parameters can be important in order to understand the mechanism that triggered the earthquake. In this work we will take advantage of the above mentioned InSAR potentials, providing three types of products: 1) differential interferograms giving a general overview of the surface deformation; 2) decomposed maps to best determine the sense of the fault slip; 3) modeling of coseismic slip through inversion. Thus, first of all, we provide maps of co-seismic displacement due to this event. The analyzed data come from Sentinel-1A/B satellites (in both Ascending and Descending tracks) which work in the C-band in 12-day revisit cycle, 6 days apart from one another, providing the best identification for the detection of surface co-seismic displacement with cm-scale accuracy. In particular, for the Ascending track (175) we used the data acquired on 20 November and 26 November (2019), while for the Descending track (153) those acquired on 25 November and 1 December (2019). Data processing was performed through the Sentinel Application Platform (SNAP) software which is freely provided by the European Space Agency (ESA). The co-seismic displacement maps obtained in the Line of Sight (LoS) of the satellite are then decomposed into horizontal (east-west) and vertical components. We then further investigate the resulting maps to establish the sense of the motion along the whole area showing larger deformation. Finally, we performed an inversion of the interferometric data using a finite rectangular fault plane in an uniform half-space and an uniform slip. We used the Geodetic Bayesian Inversion Software (GBIS) developed and freely provided by the University of Leeds. After the sampling of the input data and the definition of model parameters, we estimate the parameters of the seismogenic source.
Authors: Orlandi, Diana; Areggi, Giulia; Bonini, Lorenzoin the recent years, the interferometric synthetic aperture radar (InSAR) scientific community has primarily been focused on adapting the available multi-temporal Interferometric SAR (MT-InSAR) codes to process large sequences of SAR data collected by the new constellations of SAR satellites. In particular, a significant role has been played by the Sentinel-1A/B (S-1) twin satellites of the European Union Copernicus initiative. The free and open access policy of the Sentinel-1 data [1] and the weekly repetition frequency of the observations contributed to SAR/InSAR technology’s evolution to afford the challenges of the present-day, big-data era. New developments are required to adapt the existing processing codes to handle Interferometric Wide (IW) Swath S-1 SAR data. These were collected through the Terrain Observation with Progressive Scans (TOPS) mode, which is the principal acquisition mode of S-1 over lands. In this context, the enhanced capabilities of the S-1 system revealed useful also to perform change detection analyses. Change Detection (CD) using remotely sensed data represents a well-established technique for generating damage maps on a global scale after a natural hazard [2]. Synthetic aperture radar (SAR) microwaves can penetrate forest canopies and obtain structural information about the underlying surface, especially at longer wavelengths. Many features can be extracted from sets of SAR images, and the most commonly used factor is the backscattering coefficient, which is a fundamental indicator of ground conditions due to its sensitivity to surface roughness concerning polarization. Moreover, the polarimetric decomposition through eigenanalysis of the complex data allows one to identify the characteristics describing the type and contribution of scattering mechanisms. An alternative to the exclusive use of SAR backscattering signature for the generation of Land-Use-Land-Change (LULC) products is using the interferometric SAR approach. In this work, coherent and incoherent CD methods are applied in an area affected by an earthquake. In particular, a set of S-1 images over the area of Durazzo, Albania, were used to test the feasibility of some recent coherent change detection methods based on the analysis of sets of SAR interferograms ecompassing the main seismic event that struck the region on November 26, 2019. The work of Monti Guarnieri [3], based on the identification and analysis of a group of DS pixels, is employed. However, with respect to [3], in this preliminarily work, we tested the feasibility of the algorithm to work when only a reduced set of small baseline, high coherent, interferograms are used for the identification of the changes elated to distributed scatterers. Further theoretrical analyses are required to identify which is the most suitable set of small interferograms to be exploited for change detection purposes. Experiments are carried out both at single-look and multi-look scales. [1] Desnos, Y.L.; Foumelis, M.; Engdahl, M. Sentinel-1 Mission Scientific Exploitation Activities. IEEE Int. Geosci. Remote Sens. Symp. 2017, 19, 19364. [2] Jung, J., Kim, D., Lavalle, M., Yun, S.-H., 2016. Coherent Change Detection Using InSAR Temporal Decorrelation Model: A Case Study for Volcanic Ash Detection. IEEE Trans. Geosci. Remote Sens. 54, 5765–5775. https://doi.org/10.1109/TGRS.2016.2572166 [3] A.V. Monti-Guarnieri, M. A. Brovelli, M. Manzoni, M. Mariotti d’Alessandro, M. E. Molinari, and D. Oxoli, “Coherent Change Detection for Multipass SAR” , IEEE Trans. Geosci. Remote Sens., vol. 56, no. 11, Nov. 2018.
Authors: Pepe, Antonio (1); Albano, Felice (2)The Xianshuihe fault is one of the most tectonically active faults in the eastern Tibetan plateau. Understanding its slip behaviour and frictional properties is essential for determining the seismic potential. We use Sentinel-1A/B data from ascending and descending tracks to estimate the deformation field and retrieve three-dimensional velocities during the period of 2014-2019. We find that shallow creep is widely distributed along the entire Xianshuihe fault. Elastic dislocation modelling based on fault-parallel velocities reveals that the tectonic loading rate along the Xianshuihe fault is in the range of 8.8-17.9 mm/yr with a locking depth of 7.6-18.9 km. The shallow aseismic slip rate ranges from 3.3-7.8 mm/yr on the Daofu and Qianning segments, to 16.3-19.8 mm/yr on the western Kangding segment; the high slip rate on the western Kangding segment is likely due to the postseismic relaxation of the Kangding 2014 Mw 5.6-5.9 earthquakes. The seismic moment accumulated on the Qianning segment since the last major event is equivalent to a Mw 6.6 earthquake. Analysis of the evolution of aseismic slip in the framework of rate-and-state friction shows that the the Kangding segment has a much smaller (a-b) (0.0016) than the Daofu segment, (a-b) = 0.0117. The difference is potentially controlled by different rock types between the Kangding segment (granite) and other segments of the Xianshuihe fault (quartz-rich sandstone).
Authors: Qiao, Xin; Zhou, YuOn 21 September 1999, the Mw 7.6 Chi-Chi earthquake, one of the largest inland earthquakes in Taiwan struck the Taipei Basin, in the Central western part of the island, killing more than 2400 people and damaging 100 000 structures. The rupture was complex with several dislocations along the 100-km long Chelungpu thrust fault. Revisiting this earthquake is a challenge as the precision and coverage of the datasets available are quite poor. Furthermore, the complexity of the area (steep topography and dense vegetation) is preventing coherent outputs from C-band interferometric synthetic aperture radar (SAR). However, taking advantage of the differences and complementaries of the variety of datasets (radar and optical images, GNSS, levelling and benchmarks) to build a methodology to realise a joint inversion is restraining the model and bringing more precision. In the first part of this study, several earth observations were processed to investigate the co-seismic deformation. Generalized Akaike's Bayesian Information Criterion (gABIC), a relative weighting package and the Particle Swarm Optimization and Okada Inversion package (PSOKINV), a geodetic inversion package, were used to determine the relative weight between the datasets and to determine more precisely the slip distribution. We compared the results run with and without gABIC and using each dataset alone and then jointly. By using gABIC prior to the slip distribution estimation, differences in the slip pattern are observable and the gABIC-joint model show a smoother distribution. The results were compared with previous models from the literature and are compatible. The second part of the study focuses on the post-seismic deformation of this event. 20 years of GPS and InSAR data are available to analyse the current deformation, detect if a modification of the post-seismic deformation is visible over these two decades and get more knowledge on the afterslip. The modeling is done using a burger's rheology in the purpose of analysing the interplay between afterslip and viscoelastic relaxation. Most of the energy was released at the north part of the fault during the co-seismic phase, where it is curving toward east and mainly for depths less than 4km while the majority of the afterslip is happening in the south part of the Chelungpu fault. The approaches used have shown to be a viable ones for the study of complicated case such as the Chi-Chi earthquake and can significantly beneficiate from the weight determination. This study emphasizes the necessity of realistic fault geometry and the need of consistency between datasets and precise weighting.
Authors: Roger, Marine (1,2); Li, Zhenhong (1,2); Clarke, Peter (1,2)Systematically characterizing slip behaviours on active faults is key to unraveling the physics of tectonic faulting and the interplay between slow and fast earthquakes. Interferometric Synthetic Aperture Radar (InSAR), by enabling measurement of ground deformation at a global scale every few days, may hold the key to those interactions. However, atmospheric propagation delays often exceed ground deformation of interest despite state-of-the art processing, and thus InSAR analysis requires expert interpretation and a prioriknowledge of fault systems, precluding global investigations of deformation dynamics. Here we show that a deep auto-encoder architecture tailored to untangle ground deformation from noise in InSAR time series autonomously extracts deformation signals, without prior knowledge of a fault's location or slip behaviour.Applied to InSAR data over the North Anatolian Fault, our method reaches 2 mm detection, revealing a slow earthquake twice as extensive as previously recognized.We further explore the generalization of our approach to inflation/deflation-induced deformation, applying the same methodology to the geothermal field of Coso, California. As the properties of the atmosphere cannot be measured at the same spatial and temporal resolution as InSAR acquisitions, InSAR time series still contain large amplitude atmospheric delays, on the order of centimeters, in spite of recent marked improvements in atmospheric correction and processing strategies. For this reason, expert processing and analysis is required to interpret InSAR data. Furthermore, since the onset of the Sentinel 1 mission, the amount of available InSAR data has grown at a pace that is already challenging the ability of the community to process and analyze it, and the upcoming NISAR mission will increase the amount of available InSAR data several fold. Therefore, significant effort has been put into developing strategies to build time series with such vast data sets. Nonetheless automatic, autonomous InSAR interpretation methods are poised to become essential, if just to leverage the increasing spatial and temporal resolution of the data. In this presentation, we will first introduce the notion of auto-encoders before describing the architecture of our neural network. We then describe our training set and perform preliminary tests on synthetic data. We finally highlight the efficiency of our algorithm on two reconstructed time series of ground deformation, the first one derived from COSMO-SkyMed acquisitions and the second one derived from Sentinel 1A-B SAR acquisitions The initial application of our method on InSAR time series enables the direct observation of a slow earthquake, refining previous estimates, autonomously and without prior knowledge. In particular, we expect that the ability to systematically observe fault and pressure source deformation at a global scale will further the understanding of hydrologic, volcanic and tectonic processes, and may bring us closer to bridging the observational gap that exists for transient surface deformation.
Authors: Rouet-Leduc, Bertrand (1); Jolivet, Romain (2); Dalaison, Manon (2); Hulbert, Claudia (2)Crustal earthquakes are events of sudden stress release through rock failure, for example as a consequence of continuous and long-term stress buildup at tectonic faults that eventually exceeds the strength of rock. Before failure, under increasing stress at a fault, the surrounding crust is slowly deforming. The amount and pattern of crustal deformation carries information about location and potential magnitude of future earthquakes.Time series of space-borne interferometric Synthetic Aperture Radar (InSAR) data can be used to precisely measure the surface motion, which corresponds to the crustal deformation, in the radar line-of-sight. These observations open the opportunity to study fault loading in terms of location, size of locked parts at faults and their slip deficit. Here we study the North Anatolian Fault (NAF), a major right-lateral strike-slip fault zone of about 1500 km length in the north of Turkey and we create its first large-scale 3D finite-fault model based on InSAR data.We use the InSAR time series of data recorded by ESA’s Envisat SAR satellite between 2002 and 2010 (Hussain et al. 2018, Walters et al. In 2014). We represent the fault with several vertical, planar fault segments that trace the NAF with reasonable resolution. The medium model is a layered half space with a viscoelastic lower crust and mantle. We use the plate motion difference calculated through an Euler pole to set up a back-slip finite-fault model. We then optimize the back-slip as the slip deficit, the width and the depth of the locked fault zone at each segment to achieve a good fit to the measured surface motion.We find shallow locking depths and small slip deficits in the eastern and westernmost regions of the NAF, while the central part shows both deeper locking depths and larger slip deficits for the observation period. The trends of both parameters are in an overall agreement to earlier studies. There, InSAR-time series data have been used to calculate slip deficits at the North Anatolian fault with 2D models and/or assuming a homogeneous and purely elastic medium. Local modeled differences therefore might be connected to differences in the modeling approaches but also remain subject to further investigations and discussions.Our model provides a very suitable basis for future time-dependent modeling of the slip deficit at the NAF that includes also more recent InSAR time series based on data from the Sentinel-1 radar satellite mission of ESA.
Authors: Seidel, Alison Larissa; Sudhaus, HenrietteSmall or large earthquakes occur constantly and are commonly recognized as one of the unpredictable natural disasters. This is the reason to deliver reliable information on the distribution of the destructions caused, the locations of the places and infrastructure objects that are mostly affected and need urgent actions from the authorities responsible for emergency situations. In the recent years the advances in remote sensing techniques and especially synthetic aperture radar (SAR) instruments operationally provide data that after thematic processing are able to provide information for fast impact assessment shortly after the event. The enormous advantage of using SAR is that they deliver data regardless of the weather conditions or presence of external illumination. It is to emphasize that this manner the size of the areas that can be explored is much larger than by in-situ investigations. Additional benefit in using SAR is that it is capable to detect deformations in range of centimeters. The new opportunities given by the Sentinel-1 mission and its data archive maintained by ESA provide scientists with possibilities to extend their research and thus to offer to the institutions and the society accurate information on the ground motions soon after an earthquake event. In this paper we present the results obtained after processing SAR data for three seismic events targeting surface deformations in the affected areas. In the first one used were SAR data to determine the deformation parameters of the Aegean earthquake affecting the Greek island of Kos and the town of Bodrum located in the southwest of Turkey on 20th July 2017. The second one is the event that occurred on 25th October 2018 on the island of Zakintos in Ionian Sea having magnitude of Mw=6.4. Third studied earthquake took place close to the town of Durres, Albania on 26th November 2019 into which 50 people died and many left homeless. In this study reported are also the specific aspects during the DInSAR data processing that needs to be accounted in order to increase the robustness of results when studying this type of events. Emphasized is the need of better integration of the information from SAR and GNSS data which will deliver reliable information to the local and national authorities. The authors expect that this research can significantly contribute for revealing the mechanisms of the ongoing tectonic processes in the region.
Authors: Nikolov, Hristo Stoianov (1); Atanasova, Mila (2); Protopopova, Valentina (2)The 1600 km-long Altyn Tagh Fault (ATF) is a major intra-continental strike-slip fault along the Northern Tibetan Plateau, the slip rate of which has significant implications for our understanding of the present-day tectonic processes of the Tibetan Plateau region. Previous studies of interseismic deformation from geodetic measurements over the ATF have only focused on specific portions and may not provide an overall picture of the variation of localised strain accumulation along the fault. In this research, we present an interseismic velocity field along ~1500 km length of the ATF, derived from Sentinel-1 interferograms spanning the period between late 2014 and 2019, which is the first time such a large-scale analysis has been carried out for this fault with Interferometric Synthetic Aperture Radar (InSAR). As the interseismic deformation signals correlate strongly with the 6 km topographic relief across the ATF, we apply a developed spatially varying scaling method for InSAR tropospheric corrections to reduce the tropospheric effects and derive a clearer deformation signal over the fault. To derive a consistent velocity field over the extensive spatial scale, we present a new scheme for stitching InSAR LOS velocities estimated from multiple tracks. Using a modified elastic half-space model, we find significant strain accumulation along the 1500 km length of the ATF, at a relatively fast rate of ~10 mm/yr and quite localised along the fault. The results indicate an eastward decrease of the slip rate along the fault from 11.6 ± 1.0 mm/yr to 7.5 ± 1.2 mm/yr over the western portion to the central portion, whereas it increases again to 11.1 ± 1.1 mm/yr over the eastern portion. We find the locking depths are generally shallower than 20 km when the fault is shown as a single strand over the western and central portions, which supports the small thickness of the elastic layer in the lithosphere suggested by previous depth distribution of earthquakes in this region. However, we find locking depths of ~40 km over the central and eastern portions where the fault breaks into three strands. Furthermore, the results suggest that no significant creeping occurs along the fault. To investigate the pattern of strain localisation along the ATF, we fit a shear zone model to the derived long-term InSAR velocity field. Inverting for shear zone width reveals two broad shear zones along the ATF, where the strain is distributed over multiple strands rather than concentrated on a single narrow strand. The broad shear zones explain the high estimates of the locking depth found when using the elastic half-space model and also off-fault seismic activity on the strands away from the ATF in these areas. The results also show a relatively wider shear zone from the central portion eastward, where the ATF breaks into three parallel strands. We find a high slip rate of 11.5 ± 1.0 mm/yr along the south-western segment of the ATF, a region not typically covered by previous studies, is transferred to the structurally linked left-lateral strike-slip Longmu-Gozha Co Fault. It demonstrates that the generation of the NS-trending normal faulting events in this region, such as the 2008 Mw 7.2 Yutian earthquake, is ascribed to the EW-trending extensional stress at the Ashikule step-over zone between the two left-lateral faults. We also find a high shear strain rate greater than 0.4 μstrain/yr in this region, which could be caused by the stress loading effects of the recent seismic activities, implying that there may be a relatively higher earthquake potential. This study suggests that a slip deficit of around 1 m has been accumulated along the ATF since the last major earthquake events (Mw > 7) in 1924, and indicates that the fault is capable of rupturing with the potential for a magnitude 7.5 or larger earthquake.
Authors: Shen, Lin; Hooper, Andy; Elliott, John; Wright, TimOn February 13, 2021, an earthquake of magnitude 7.3 with a ~55 km depth hypocenter occurred off the coast of Fukushima Prefecture, Japan. This earthquake caused damage to lifelines and transportation systems in a wide area of eastern Japan. In general, such large earthquakes cause surface displacement. The Geospatial Information Authority of Japan (GSI) estimated the horizontal surface displacements using global navigation satellite system (GNSS) data. As a result, it was found that small surface displacements occurred in a wide area of eastern Japan. Particularly, in the eastern part of Fukushima Prefecture, surface displacement of about 2 cm occurred in the west or southwest direction. The GSI used the GNSS results as preliminary information to estimate the earthquake source fault model. Specifically, they used the Markov Chain Monte Carlo (MCMC) method to estimate the model parameters of the source fault on a single rectangular fault. However, the source fault model was not uniquely determined. The reason for this may be that the GNSS surface displacement data is not spatially dense and therefore insufficient as preliminary information. In this study, we report the analysis results of the earthquake source fault model using the surface displacement data from PSInSAR analysis in addition to the spatially insufficient GNSS surface displacement data. To create these data from PSInSAR analysis, we use the data from January 2019 to February 2021 for Sentinel-1 Ascending and Descending. The PSInSAR analysis can estimate the surface displacement over a wider area than GNSS. On the other hand, the temporal resolution of PSInSAR analysis is lower than that of GNSS. By combining these two data sets, it enables to estimate surface displacement data with both higher spatial and temporal resolutions. This data can be used to estimate the source fault model as preliminary information. In addition, we report the analysis results of the source fault model using the surface displacement data obtained by PSInSAR analysis only, and calculate the amount of change in the estimated parameters compared to the case using GNSS surface displacement data. If the amount of change is small, it may be possible to estimate the earthquake source fault model using surface displacement data from the PSInSAR analysis instead of GNSS data. In fact, in areas where GNSS surface displacement data are not available, surface displacement data have been obtained only from PSInSAR analysis. Therefore, it is important for such areas to estimate the source fault model using the surface displacement data from PSInSAR analysis.
Authors: Shigemitsu, Yutaro; Ishitsuka, Kazuya; Lin, WeirenEarthquakes occurrence mechanisms have been known since the early 20th Century, when scientists understood they are the result of a sudden, catastrophic release of strain energy that has slowly accumulated around tectonic faults. Since the first practical demonstration of Space-borne interferometric synthetic aperture radar (InSAR), in 1992 a ERS-1 interferogram was generated showing the surface deformation caused by the Landers earthquake, a long way was traveled in terms of InSAR developments. The rapid development of InSAR techniques and mostly the surprising results in detecting surface deformations inspired a generation of scientists, which drastically increased the InSAR community. This success achieved by InSAR-based monitoring techniques, mainly reached through ESA’s datasets exploitation (ERS-1/2, Envisat and more recently Sentinel-1), has aroused the interest of many national space agencies. This was the case of the Canadian, German, Italian, Japanese or South Korean space agencies. Moreover, some collaborations between space agencies from different countries led to the development of new SAR missions. This is the case of the Spanish PAZ mission, which follows the German’s TerraSAR-X/TanDEM-X satellite orbits, and the images are fully compatible. Recently, private companies are also entering into the InSAR competition, which is contributing to the advance in this field of Earth observation. The fast growth of SAR achieved, as a result of different bands and resolutions has greatly facilitated InSAR data source accessibility. InSAR has become an essential technique to detect surface variations due to earthquakes, volcanoes, landslides, glaciers, aquifers and other natural or man-made hazards. The launch of C-band Sentinel-1 mission represented a milestone in InSAR technology. This medium resolution sensor is characterized by its 250 km-wide swath, global coverage, and short revisit time (6 or 12 days). Thus, these abundant SAR images have greatly promoted the InSAR ground deformation monitoring from the regional to national scale. However, effectively exploiting the massive volume of Sentinel-1 data over wide areas has posed a great challenge for existing InSAR tools. Currently, Sentinel-1 provides approximately 10 TB of SAR products per day. Moreover, the total volume of national-scale SAR archives has reached the PB level. InSAR data is nowadays regarded as remote sensing big data. Thus, automated systems for detecting deformation in InSAR imagery is mandatory to fully explore the archives. With machine learning (ML), the earthquake science and InSAR communities have a new suite of tools to apply to detect deformations in wide areas; In this ongoing work, we explore the LICS database (https://comet.nerc.ac.uk/comet-lics-portal/), both wrapped and unwrapped interferograms without atmospheric correction, to automatically detect earthquake fringes. After preparing the dataset, some CNN have been trained in three ways: i) with wrapped interferograms; ii) with unwrapped interferograms; and iii) with both. In this way, it was possible to identify the best performant CNN in the automatic detection of earthquake fringes. Finally, the pre-trained CNN was used to segment the area affected by the earthquake so that it can be used with GIS-based software. These preliminary results show that ML-based tools are the future for global interferogram processing and analysis.
Authors: Silva, Bruno Miguel (1); Sousa, Joaquim João (2,3); Cunha, António Manuel (2,3)The high spatial and temporal coverage of the Sentinel-1 constellation enables us to measure deformation in active continental rifts including those in middle- and low-income countries, such as the East African Rift. We focus here on the Main Ethiopian Rift, which is an active narrow rift, 300 km long by 80 km wide, with active volcanic centres and magmatic segments located along the ~20 km wide rift axis. The age and eruptive histories of these volcanoes are poorly known, and little ground-based monitoring equipment existed until recently. We processed and analysed Sentinel-1 data from October 2014-2019 and identified sustained and episodic surface deformation at several volcanic centres, including Tulu Moye and Corbetti. Previous deformation at these volcanoes had already been identified using the Envisat data from 2004-2010 (e.g. Biggs et al., 2011), however, the temporal coverage was limited. In 2015, we observed a dyke intrusion to the north of Fentale volcano. The dyke opening began at the time of a seismic swarm and continued for over 6 months. The deformation rate decayed exponentially with a characteristics time of ~3 months. We use a Bayesian inversion technique to estimate the first-order source mechanism based on an elastic half-space model and estimated a 6 km long dyke at a depth of 4-6 km, sub-parallel to the rift-floor fault system. This is the first dyke intrusion to have been directly detected in the Main Ethiopian Rift, and the long timescale is more consistent with a high viscosity magma of rhyolitic composition. We combine the Sentinel-1 InSAR data with available GNSS data to calculate a three-dimensional velocity field and strain rate map. These observations are key to understanding the development of extensional plate boundaries. The velocity field across the rift shows a high gradient over the rift floor and flank. The Nubia-Somalia separation rate is 4-6 mm/yr with decreasing rate from north to south. The highest strain rates (3x10-8/yr – 4x108/yr) are observed on the rift-floor and decreases towards the rift boundary. The spatially variable distribution of the strain rate demonstrates the importance of geodetic data in understanding the current state of continental deformation that leads to continental breakup and seismic hazard. Together, these observations provide evidence for magmatic and tectonic activities in the Main Ethiopian Rift and demonstrate the potential for InSAR to contribute to mapping the geohazard and potential geothermal resources in the region.
Authors: Tessema, Tesfaye Temtime (1); Biggs, Juliet (1); Weiss, Jonathan (2); Lazecky, Milan (3); Lewi, Elias (4)The Western North America InSAR (WInSAR) consortium was originally established to facilitate collaboration in and advancement of Earth science research using radar remote sensing, with a major emphasis on differential interferometric synthetic aperture (DInSAR) studies. Its members are a consortium of non-commercial scientists at more than 300 international institutions, including universities, research laboratories, and public agencies. WInSAR also advocates for the open exchange of SAR data by seeking to enlarge the number of member organizations. Hosted and supported by UNAVCO, both organizations are non-profit, membership-governed groups funded by the National Aeronautics and Space Administration (NASA), the National Science Foundation (NSF), and the U.S. Geological Survey (USGS). With the support of UNAVCO, WInSAR has an important education mission, providing courses and workshops in DInSAR processing at both the introductory level for non-specialists, and week-long courses in DInSAR processing software, including both GMTSAR and ISCE. Both members and non-members can attend the annual short courses hosted by UNAVCO. In 2020 these courses were held entirely online, a format which enabled significantly increased participation from a large number of institutions around the world. Here we review the successes and lessons learned from this new format and discuss our plans to expand our education mission through a digital format moving forward. To help maximize the scientific return of radar data, WInSAR provides password-protected data access with partners that include the Geohazard Supersite and Natural Laboratories initiative of the Group on Earth Observations (GEO), the Alaska Satellite Facility, and various space agencies. Data access is in accordance with the regulations of the space agency that created and provided the data to the archive. All members have access to the ISCE (InSAR Scientific Computing Environment) open-source software developed at Caltech/JPL/Stanford. WInSAR also enables member institutions to archive user-derived products and provide digital object identifier (DOI) numbers for future access and publication in accordance with FAIR (findable, accessible, interoperable, and reusable) data principles. We present a summary of these activities and a survey of recent results by WInSAR member scientists from around the world based on Sentinel-1, ALOS, ALOS-2, TerraSAR-X, Envisat and ERS-1/-2 SAR data, with additional SAR data from other satellites and the NASA airborne UAVSAR SAR system. Studies address a wide variety of DInSAR topics and themes, including technique development and surface change related to earthquake and tectonic processes, volcanic activity, groundwater extraction and subsidence, glaciers, landslides, permafrost change, infrastructure monitoring, and subsurface energy production from around the world.
Authors: Lindsey, Eric (1); Chaussard, Estelle (2); Bekaert, David (3); Tymofyeyeva, Ekaterina (3); Chen, Ann Jingyi (4); Tiampo, Kristy (5); Crosby, Christopher (6)The Tien Shan orogenic belt accommodates about half (15-20 mm/yr) of the rate of convergence of India relative to Eurasia. The way the deformation is distributed is poorly known but key to understanding lithospheric deformation deep within the continental interiors, as well as the hazards posed by intraplate faults. The Dzhungarian region in the northern Tien Shan is a tectonic transition zone between the active orogenic belt and the stable Kazakh platform. There are several structures in this transition zone, with the longest fault being the NW-SE strike-slip Dzhungarian fault (DZF). Clear fault traces, as well as scarps, are visible from the optical images; however, its slip-rate is known from only one preliminary field-based result of 2.2 mm/yr, limiting our understanding of the role of this major intracontinental fault zone. In this study, we provide the first multi-temporal interferometric synthetic aperture radar (InSAR) observations in this remote area of the northern Tien Shan using about 5 years of ESA Sentinel-1 satellite radar measurements. Despite the large data set and short return periods available from the Sentinel-1A and Sentinel-1B satellites, the extreme seasonal variations in climate and abrupt changes in soil moisture component produce severe unwrapping difficulties in both short- and long-baseline interferograms of the study area. We focus on a series of corrections to the wrapped interferograms in order to reduce the variability of the phase and thus enable unwrapping over large areas. The interferograms are first corrected for tropospheric effects using either the ERA-5 atmospheric model from ECMWF or additional quadratic relationships of phase and elevation with a relationship that varies with azimuth in order to capture the north-south variability of the residual atmospheric delays. In addition, a specific focus on moisture-related seasonal deformation signals and its extraction with a stacking or principal component analysis approach allow us to correctly unwrap the interferograms, in particular across agricultural areas and the edge of the sedimentary basins, and isolate bedrock pixels that are not affected by the seasonal signal for further tectonic analysis. The method consists of removing the spatial template of the deformation extracted from a series of well-unwrapped interferograms to assist the unwrapping of the noisier interferograms across sedimentary basins, where we observe strong and asymmetric phase gradients. After unwrapping, the moisture template and the empirical atmospheric corrections previously removed are re-introduced to reconstruct the fully unwrapped phase signal. We also tried another method which is to use the extracted moisture template as a guidance for unwrapping path to lower the unwrapping errors and their propagation. The final Line of Sight velocity maps, acquired both in ascending and descending views, reach the necessary millimeter per year precision to access the slow strain rate accumulation in this region. We are thus able to provide a first estimation of the present-day on-going movement of the Dzhungarian fault at the western edge of the Dzhungarian Basin.
Authors: Tsai, Chia-Hsin {Wendy} (1); Daout, Simon (2); Walker, Richard Thomas (1); Parsons, Barry (1)Dushanbe, Tajikistan’s capital city, lies on or close to a number of active faults. The city sits on the northern boundary of the Tajik basin, or Afghan-Tajik depression, a sedimentary basin bounded by the Tian Shan, Pamir, Hindu Kush and Afghan platform. A set of north-south trending folds and associated thrust faults sweep up the basin interior and curve to the east at the northern boundary. Here, they merge into the right-lateral Ilyak fault, which runs to the south of Dushanbe and the surrounding towns. ~5mm/yr of shortening and 8-15mm/yr of right-lateral motion across the Ilyak fault were inferred from GPS data by Metzger et al., 2020. However, GPS point data alone are insufficient to determine the extent to which the Ilyak fault or basin thrusts are locked or slipping aseismically. The 1989 MLH5.5 Gissar earthquake (and others in 1953 and 1968) occurred close to the western end of the Ilyak fault, but the exact faults they ruptured are unknown. Additionally, it is not clear whether the measured shortening is accommodated by the Ilyak fault or is partitioned onto other structures. In order to produce a continuous map of the deformation, we performed a multi-temporal InSAR time series analysis of the Tajik basin using the New Small Baselines Subset (NSBAS) processing chain. This method allows us to image small amplitude tectonic signals. We use data from ESA’s Sentinel-1 radar satellites in interferometric wide swath mode from descending track 151 and ascending track 71 between 36 and 39 degrees latitude. We process interferogram networks spanning a ~5 year period, with interferogram intervals between 12 days and ~1 year. We use both long and short-interval interferograms in order to check for bias linked to systematic loss of coherence in regions of high moisture or snow. To aid the unwrapping of longer baseline interferograms, we first empirically correct for topography related delays in the wrapped interferometric phase, and then add back the correction after unwrapping. We compare predictive atmospheric corrections based on weather models from ECMWF and empirical corrections to assess their validity. We produce high-resolution ascending and descending track LOS velocity maps. We observe a strong subsidence signal corresponding to agricultural land use, which may be due to groundwater extraction or soil moisture processes. We mask these areas in order to study the tectonic deformation. A sharp step in LOS velocity across the Ilyak fault indicates aseismic slip at shallow depths. We also observe shallow aseismic slip at the base of a basin anticline, directly to the southwest of Dushanbe (the Bobotag thrust). We use our LOS velocity measurements to test simple models of faulting. We combine our InSAR study with geomorphology analysis using Pleiades-derived high-resolution DEMs, mapping out additional active faults beneath the urban area from their geomorphological expression.
Authors: Wilkinson, Roberta (1); Daout, Simon (1,2); Parsons, Barry (1); Walker, Richard (1); Pierce, Ian (1); Ishchuk, Anatoly (3)Weihe Basin is located in Shaanxi Province, central China, bordering the south of Ordos block in the Weibei uplift belt, the north of North Qinling orogenic belt, the east of the southwestern margin of the Ordos arc structural belt, and the west of the Shanxi uplift belt, which plays an important role in the well-known Fen-Wei seismic belt, spanning 350km from east to west. Hundreds of active faults were discovered over the basin in history, and most of these deeply buried thrust faults criss-crossed the basin (Fig.1). The main faults that control the sedimentary basin formation include Beishan Piedmont fault, Guanshan fault, Qinling North Piedmont fault, Chang'an-Lintong fault, and Lishan Piedmont fault. Weihe Basin is characterized by intense crustal activity, and more than 25 earthquakes have been recorded with magnitude larger than 5.0 since 1177. One MS 8.0 earthquake occurred in Huaxian was recorded in 1556, however, there are no documented cases greater than MS 5.5 since the 20th century. In order to understand the structure and slip rate of the known fault zones, identify potential active faults as well as their activity intensities, and compensate for the existing discrete located GPS and leveling measurements in Weihe Basin, 92 scenes ascending Sentinel-1 (06/20/2015-05/30/2019) and 6 scenes descending ALOS-2 ScanSAR (04/01/2015-12/19/2018) datasets were utilized to derive LOS deformation velocity respectively. The geoid offset correction between EGM96 datum and WGS84 datum was applied to the 1-arc-second (~30 m) Shuttle Radar Topography Mission (SRTM) DEM, that is used as an external DEM to remove the topographic phase from the interferograms for both the Sentinel-1 and ALOS-2 ScanSAR processing. GACOS atmospheric products are employed to reduce the effects of tropospheric delay. Then the deformation components in the east-west and vertical directions are estimated using the descending and ascending InSAR images together over the entire basin (Fig.1). The standard deviation of vertical deformation between InSAR and leveling measurement is around 2 mm/yr. InSAR measurements allow us to position not only the previously known faults, but also new active fault not previously revealed, which indicates that the complex internal motion within the Weihe Basin could be controlled by multiple faults. Finally, we calculate the slip rates and blocking depths of the Longxian-Mazhao fault, Puyang-Lantian, the southern part of Weihe fault, and the Kouzhen-Guanshan fault, and evaluate their potentials to arise large earthquakes.
Authors: Niu, Yufen (1); Qu, Feifei (1); Zhu, Wu (1); Zhang, Qin (1); Lu, Zhong (2); Zhao, Chaoying (1); Qu, Wei (1); Hu, Yaxuan (3)Salt structures have been one of the considerable interest subjects over the last few decades due to its significant utilization in hydrocarbons exploration, gas storage as well as waste disposal. For instance, the recent activity of some salt domes should be considered when planning repositories for gas and nuclear waste. Moreover, understanding the geometries, kinematics, and dynamics of diapirs would be useful to all geoscientists concerned with crustal deformations. Interferometric Synthetic Aperture Radar (InSAR) technique is a valuable modern technology that can be used to map land deformation over the salt structure with high spatial and temporal resolutions. Iran is known for having a unique abundance of salt structures in terms of evaporite deposits. Central basin of Iran is one of the notable regions that contain different evaporate provinces like Kalut basin in the north of Yazd province, which provides an exceptional opportunity to investigate at the surface the style and mechanics of diapiric intrusion. Kalut basin of investigation interest for three reasons. The first, Kalut basin includes elongated, elliptical, as well as circular domes with unusually pure, white, and massive diapirs. Second, to our knowledge, no attempt has been made so far to study on activation of the salt structures in the Kalut Basin. The third, the unique condition of this area in terms of having different salt structures and semi-arid weather condition without any vegetation have provided an appropriate natural laboratory for applying the InSAR method in this unusual area. Kalut basin consists of salt extrusions, seasonal lakes, and seasonal rivers. In this study, we try to explain the displacement pattern for each of them. This research is the first-of-its-kind study that has considered rivers displacement alongside diapirs. Salt extrusions in Kalut basin are generally classified in terms of geometry into three types: Salt-wall canopy, Salt-stock canopy, and the Salt massif. We used remote sensing data collected from Google Earth to characterize salt structures properties in the Kalut basin. This technique enabled us to map the geological features like salt extrusions precisely. We simplified the geological setting of the Kalut basin using optical images of Landsat and geological maps of the area to characterize the distribution of surficial deposits in the Kalut basin. This method led to determine several salt welds between salt extrusions to help for reconstruction and interpretation of salt development in this basin. In this research, the surficial deformation of nine salt diapirs in the Kalut Basin was monitored using synthetic aperture radar (InSAR) technique. We used 75 Single Look Complex (SLC) IW scenes acquired on the ascending (Path: 57 & Frame: 101-103) collected during November 2014 through November 2018 and 62 SLC images on descending tracks (path: 64 & Frame: 481-483) collected during October 2014 through January 2018, both tracks are covering the same period. We produced 257 and 258 interferograms in ascending and descending tracks, respectively. The Sentinel-1A images processed by the Small BAseline Subset (SBAS) approaches. We combined ascending and descending results from different oblique geometries to decompose the vertical and horizontal components of displacement. The results suggest that the mean velocity map of Sentinel-1A has higher accuracy respect to previous radar satellites, and makes it possible to study small-scale salt structure with less than 1000-m dimensions. Monitoring deformation of salt structures showed that the Kalut Basin is an active diapiric basin, and its seven salt diapirs are rising, as well as none of them showed subsidence. As such, vertical displacements appeared with the same sign in both ascending and descending results, and the vertical uplift rate of diapirs varies between 2 and 6 mm/yr respect to the local references while the rate of displacement in seasonal rivers is higher than salt structures. For obtaining salt kinematics, we mapped six profiles on the vertical deformation result and showed phase change across salt structures to topography (SRTM 30m). For better visualization, we compare the Landsat image of google earth with the phase changes graph. It should be noted; we used local weather datasets from the nearest weather station to compare the correlation between diapir motions and weather conditions. The comparison of displacement with precipitation and temperature data did not show a meaningful relationship.
Authors: Mohammadnia, MohammadHossein; Mousavi, Zahra; Najafi, Mahdi; Shahbazi, SaeedehOn December 29, 2020, a Mw 6.4 earthquake struck Sisak-Moslavina county, in central Croatia. The mainshock epicenter was located near the city of Petrinja (45.422°N 16.255°E), approximately 50 km SE of Zagreb. Several M > 4 aftershocks followed the main event in the same area. Some foreshocks also occurred the day before, as the Mw 4.8 event located 8 km W of Petrinja. The active tectonic structures in this area are strike-slip and thrust fault systems that accommodate the deformation due to the Adria-Eurasia plates collision. Specifically, the earthquake occurred along the Popusko-Petrinja strike-slip fault at a depth of 10 km. In this study, we focused on the estimation of the co-seismic deformation by using Differential Interferometry (DInSAR) technique. We processed co-seismic interferometric single pairs acquired by the European Sentinel-1 A/B satellites, obtaining displacement maps over the epicentral area. For the ascending track (n° 146) we used the data acquired on 18 and 30 December (2020), and for the descending track (n° 124) those acquired on 29 December 2020 and 4 January 2021 We exploited the functionalities of SNAP (SentiNel Application Platform), software provided by the European Space Agency (ESA), to obtain the wrapped and the unwrapped maps of the study area. Starting from the Single-Look Complex (SLC) images, we obtained the interferograms of the ascending and descending datasets. After the removal of the topographic phase by using the 30m Shuttle Radar Topography Mission Digital Elevation Model (SRTM-1sec DEM), we applied the Goldstein filter and multilooking to enhance the signal and strongly reduce the noise related to decorrelation effects. Specifically, for multilook operation, we chose 6 and 2 looks in range and in azimuth, respectively, obtaining a grid size of about 30x30m. The filtered maps have been then unwrapped by using SNAPHU algorithm, considering Minimum Cost Flow (MCF) and Smooth as Initial Method and Statistical Cost Mode respectively. Finally, after the phase to displacement conversion and the terrain correction, we obtained two displacement maps showing the coseismic deformation with respect to the Line-Of-Sight (LoS) of the satellites. Ascending and descending pairs reveal a displacement field in agreement with the strike-slip focal mechanism. Subsequent non-linear and linear data inversions allowed us to estimate source parameters and slip distribution along the fault plane. Our results give us the opportunity to discuss the seismogenic behaviors of the active tectonic structures in a complex transpressive tectonic context.
Authors: Bevilacqua, Claudio (1); Areggi, Giulia (1); Orlandi, Diana (1); Pezzo, Giuseppe (2); Bonini, Lorenzo (1,2)The main active tectonic structure in the western part of Central Sulawesi (Insonesia) is the left-lateral Palu-Koro strike-slip fault. Its offshore section was thought not to have broken during the Mw 7.5 Palu Earthquake on 28 September 2018, challenging the established knowledge of the tectonic setting at this location. Here, we use Sentinel-1 SAR interferometry to produce a map of the ground velocities in the area of the Mw 7.5 earthquake for the seven months following the 2018 earthquake. We show evidences of surface deformation along the western coast of the Palu bay indicating that the Palu Koro offshore fault section might have contribute, or been affected, by the earthquake. As the possibility of multi-segment ruptures is a high concern in the area because of the high seismic and tsunami hazard, we present here, a fault model that includes the offshore section of the Palu Koro fault. Thanks to four independents space-based geodetics measurements of the co-seismic displacement (Sentinel-1 and Sentinel-2 correlograms) we constrain the 3D co-seismic ground displacements. The modeling of these displacements allows us to estimate the co-seismic fault slip amplitude and geometry at depth. At the end, we consider the multi segment faulting scenario, including the offshore section of the Palu-Koro fault, as a plausible model to explain the submarine landslides and the tsunamis. It also gives the opportunity to observe a super-shear earthquake in the context of complex fault network and implies an increase of the probability of submarine landslides within the bay in the forthcoming years.
Authors: Guillaume, Bacques (1); De Michele, Marcello (2); Foumelis, Michael (2); Raucoules, Daniel (2); Lemoine, Anne (2); Briole, Pierre (3)Western Slovenia and the northern part of Croatia are characterized by the presence of active strike-slip and thrust fault systems. The tectonic activity of the area is related to the counterclockwise rotation of the Adria microplate and its motion toward Eurasia with a rate of convergence of few mm/yr. The tectonic activity of the region is also testified by the occurrence of several significant earthquakes in the past, as the 1511 Idrija earthquake (Mw 6.8). Considering the geological features and the current and historical seismicity of the region, the surface deformation estimation could provide useful information about the active faults located in the study area. In the present work, we estimated the surface deformation in western Slovenia and northern Croatia considering a period of about 5 years by using SAR and GNSS data. The use of Persistent Scatterers Interferometry (PS-InSAR) enables the estimation of the deformation of the area of interest, by exploiting the presence of pixels characterized by high reflectivity and high phase stability. We used this technique to process synthetic aperture radar (SAR) images acquired by the European satellites Sentinel 1A/B in ascending and descending orbit tracks spanning March 2015 to December 2019. Here, we processed the more than four hundred SAR acquisitions by using the ESA SNAP (SentiNel Application Platform) software and the scripts provided by snap2StaMPS. A stack of single-master interferograms was generated and then used as input data in StaMPS (Stanford Methods for Persistent Scatterers) software. After the PS processing, we obtained mean ground velocity maps, showing the displacement rate for each pixel with respect to the satellite Line-of-Sight (Los). Finally, the calibration of the post-processed ascending and descending LoS maps was performed in order to remove orbital ramps and residual atmospheric contributions. In this case, we used the velocity provided by GNSS stations located in the area of interest, considering the same period of time (2015-2019) in Adria-fixed reference frame. Considering the final results, our study confirms a low strain rate in the region. However, the presence of some anomalies in the velocity field close to presumed active faults can suggest aseismic slip of some of these structures.
Authors: Maurizio, Gerardo (1); Areggi, Giulia (1); Pezzo, Giuseppe (2); Serpelloni, Enrico (3); Bonini, Lorenzo (1,2)An earthquake of magnitude 6.8 Mw occurred on January 24 at 20.55, where the epicenter was in Sivrice district of Elazig in Turkey. It is known that the earthquake, which occurred 8.06 km deep, is 800 meters away from the closest settlement, the village of Çevtaş. It was recorded that the apparent duration of the earthquake was calculated as 20.4 seconds according to the first determinations. AFAD has been declared as the four settlements after the closest cycle center to the epicenter of the earthquake, Doğanbağı Village, 1.38 kilometers away, Kalaba Village, 2.86 kilometers away, Kılıçkaya Village, 3.24 kilometers away, and Ilıncak Village, 3.98 kilometers away. When the focal mechanism solutions made after the earthquake are evaluated together, it is considered that the 6.8 magnitude earthquake developed on the Sivrice Pötürge segment of the Eastern Anatolian Fault, which is a left-lateral strike-slip fault, and the tear developed in an area of 50-55 kilometers. In the region, from 1900 until today, 299 earthquakes with the magnitude of 6.8 have occurred, the largest of which is 6.8. After the 6.8 magnitude Elazığ earthquake on January 24, a total of 2,795 aftershocks occurred, 24 of which were 4 and over. In the earthquake where 45 people survived the wreckage, 41 people died. More than one thousand people who were injured in the earthquake applied to the hospital. Immediately after the earthquake, a total of 7,524 personnel from AFAD, Gendarmerie, UMKE, Fire Brigade and non-governmental organizations participated in the studies. Disaster Coordination Center Liaison Offices were established at 28 points in the earthquake region, 310 vehicles with various features, 5 mobile coordination trucks, 24 search and rescue dogs and diving teams were sent to the dam's relatives for precautionary purposes. Besides, 122 vehicles from non-governmental organizations such as Kızılay, AKUT, and IHH participated in the works. After the earthquake, 132 accommodation areas in Elazig, including school buildings, public guesthouses and sports facilities, including mass tent areas, were made available to citizens affected by the earthquake. In the tent areas, 3 846 tents were established, 19 thousand 409 tents were distributed individually. There are still 12.618 citizens in the accommodation areas, of which 775 are in schools and public guesthouses. In Malatya, 5,948 tents were distributed and installed individually. It was decided to dispatch 5,851 containers to Elazig, 4 thousand 790 of them were planned to be placed in temporary shelter areas in Sivrice district in 6 different regions. In Elazığ, 1995 living containers, 102 public toilet containers, 43 shower containers, and 77 office containers were installed. In Malatya, 700 single-storey containers were planned to be placed in 6 different temporary accommodation areas planned to be established in Pötürge, Doğanyol, Battalgazi and Kale districts. Scope of “Aid and Distribution Activities”, “Psychosocial Support Studies” and “Cash Aids” services, it is very significant to know all these supplies done by AFAD and the other organizations in time and adequate in terms of disaster crises management system. For this purpose in this paper; all the studies carried out by AFAD to measure the effectiveness of disaster management were compared with the deformation areas obtained by InSAR observations. In this way, the effectiveness of disaster management and aid to damaged areas were evaluated in terms of time and intensity. With this aim, We processed both ascending and descending track of Sentinel-1 A/B interferograms for January 24 Mw6.8 coseismic event, Sivrice/Elazığ, Turkey. The rupture was detected over more than 30 km along the Pütürge segment of East Anatolian Fault. We applied bayesian joined inversion using InSAR geodetic and seismic observations to estimate the finite fault geometry and fault slip distribution with reasonable uncertainty. The results show a joint inversion of InSAR and seismic waves can improve locations of earthquakes and fault rupture location to have a better risk assessment. As a result; crisis management (disaster management) gaps in the field have been compared with the deformation maps derived from InSAR observations and the problems evaluated before which areas were more under the effect of deformation. In light of these evaluations; It has been concluded that the disaster management studies carried out by AFAD have successfully reached all regions affected by the disaster.
Authors: Canaslan Comut, Fatma (1); Mohseni Aref, Mohammad (2)On 4th July 2019, a sequence of damaging earthquakes began near the city of Ridgecrest, California. It included a foreshock with a moment magnitude (Mw) of 6.4, followed by a Mw 7.1 mainshock almost 34 hours later. The earthquakes were mainly felt in areas of California Nevada and Arizona, being Ridgecrest and Trona the most affected towns. These two events are part of a long-lived cluster historical earthquakes which started in 1872, including the 1993 Mw 7.3 Landers and the 1999 Mw 7.1 Hector Mine ruptures. The earthquakes occurred in an arid area with almost no vegetation, which makes it suitable for observing the ground displacement from radar and optical sensing techniques. The objective of this study is to provide a methodology to retrieve a complete 3-dimensional displacement field of the earthquake by combining radar and optical observations obtained from Sentinel-1 and Sentinel-2 missions, respectively. The radar data were used to generate displacement maps from phase interferometric processing, while the displacement maps for amplitude data and optical observations (band 4) were obtained from a pixel offset tracking by applying the cross-correlation technique. The interferometric processing was done using the SNAP toolbox, amplitude offset track was generated from GAMMA software and optical offset tracking was produced using MicMac software. All generated displacement maps were integrated using weighted least squares (WLS) by utilizing various weighting scenarios for the observations. The resulting 3D displacement field were then validated using high accuracy GNSS observations from both continuous and campaign stations located in the near field of the study area. The results show a displacement of around -1.5 to 1.5 m for the East and North component, and -0.5 to 0.5 m for the Up component. Additionally, our findings show the importance of appropriate weighting of the observations to give proper estimates of the 3D displacement field. The best agreement with GNSS observations was obtained when combining phase interferometry and amplitude offset tracking performed on Sentinel-1 observations. The RMS of difference between the results obtained by this combination and 10 GNSS stations was approx. 10 cm. Combining all observations together with just equal weighting increased the RMS by 51%.
Authors: Canizares, Carolina (1); Motagh, Mahdi (2,3); Haghshenas Haghighi, Mahmud (3)The left-lateral Xianshuihe Fault lies at the eastern edge of the Tibetan Plateau. Crust to the west of the fault rotates clockwise around the eastern Himalayan syntaxis. The Xianshuihe fault is tectonically active and considered to have substantial earthquake potential. Creeping behaviour of the fault has been reported along some sections of the fault. However, the spatiotemporal details of creep are not well characterised. The last three-decade period has seen a dramatic improvement in our ability to measure earthquake cycle deformation using space-based geodesy. Although there are currently thousands of Global Navigation Satellite System (GNSS) instruments installed permanently at sites on Earth, the main weakness of GNSS data is that the station spacing is still sparse, making it challenging to resolve high velocity gradients (i.e. regions of high strain rates) which usually occur close to active faults. With the increasing volume of Interferometric Synthetic Aperture Radar (InSAR) data and improvements in data quality and processing techniques, full exploitation of this abundant data has become possible, so that we can measure crustal deformation using space-based observations with high resolution and accuracy. The InSAR technique is becoming a powerful tool for measuring large-scale continental tectonic deformation. LiCSAR is an automated processor which produces Sentinel-1 interferograms and derived products for tectonic and volcanic areas globally. LiCSBAS, an open-source InSAR time series analysis package that integrates with LiCSAR products, enables us to obtain displacement time series efficiently. In this study, we use 6 descending frames (acquired between 2014 and 2020) to obtain preliminary results of the velocity field of the Xianshuihe fault from LiCSAR/LiCSBAS processing chain. During the LiCSAR processing, the main workflow includes Single Look Complex (SLC) merging, coregistration, interferogram formation, phase unwrapping, and geocoding. The batch processing jobs can fail due to errors related to generations of specific products at various steps. In some cases, gaps exist in the interferometric networks due to no acquisitions, severe decorrelation (e.g. caused by snow cover), and/or unclosed loop phases. We fill these gaps by creating or repairing specific interferograms so that the networks are well connected. In the LiCSBAS processing procedure, there are five steps for data preparation and six steps for time series processing. We first downsample the data by a factor of 10 to a pixel size of ~1 km. GACOS data are used for tropospheric correction. Low coherent areas (average coherences ≤ 0.1) are masked, which improves the loop closure phase. Before time series analysis, incorrectly unwrapped data are identified by checking the coherence and coverage of unwrapped data and loop closure, and these are removed. We obtain Line-Of-Sight (LOS) time series results with this refined stack of data and masking of noisy pixels based on several noise indices, followed by spatiotemporal filtering. Then the LOS rates are transformed into a Eurasia-fixed reference frame by using GNSS velocities. We first average the velocities of InSAR pixels within 1 km radius around each GNSS station. We project the GNSS rates into the LOS and calculate the differences from the averaged InSAR rates. We determine and remove the best fit plane according to the residual rates, during which we also add additional constraints of minimising velocity differences within common bursts that appear in adjacent frames. This procedure is carried out separately for each track. Finally, we obtain a robust and consistent velocity field, with a sharp gradient across the fault. We will also present the slip rate, explore the strain accumulation and characterise the details of shallow creep along this fault and examine deformation associated with other faults in the region.
Authors: Fang, Jin; Wright, Tim; Lazecky, Milan; Maghsoudi, Yasser; Elliott, John; Craig, Tim; Hooper, AndySpace-born Synthetic Aperture Radar Interferometry (InSAR) plays an important role in a variety of Earth Science disciplines. On one hand, open data policy, e.g. used for Sentinel-1A/B satellites, has significantly improved the availability of InSAR data for addressing natural hazard issues. On the other hand, ongoing acquisitions of SAR imagery with intervals of a few days (e.g. 6 days for Sentinel-1AB) have imposed greater challenges for researchers to handle increased data volume of SAR observations. In this study, we present an enhanced InSAR automation system gInSAR, which was originally developed at the Canada Center for Mapping and Earth Observations Natural Resources Canada (CCMEO NRCAN) in support of InSAR processing of the RADARSAT Constellation Mission (RCM) SAR data. gInSAR provides rich facilities in metadata management, pixel-based analysis, time series (TS) modelling and data visualization. Major characteristics of the TS module of gInSAR package include: 1) support sequential data processing under framework of the Small BAseline Subset (SBAS) InSAR; 2) being equipped with several advanced TS analysis algorithms, e.g. MTInSAR (Multi-temporal InSAR), independent Atmospheric Phase Screen (APS) reducing algorithm by using common scenes and automated unwrapping error correction; 3) flexible processing strategies, in which co-, post-seismic deformation signals and APS are separated in a single processing chain at a pixel level. Finally, we show two years of deformation history associated with the 2017 Mw7.3 Sarpol Zahab earthquake, which was located on the border of Iran and Iraq. The two satellites of Sentinel-1A&B were operated in the full power to acquire data covering the earthquake area since the mainshock. Four tracks of Sentinel-1 TOPS data cover the earthquake area. Continuing surface deformation after the mainshock has been explicitly mapped using InSAR techniques. Over 100 SAR acquisitions from a single track have been made available to the public. In combination with ALOS2 L-band SAR data, the Iran case is one of the best observed events so far, which offers unique opportunities to explore earthquake physics, seismic hazard assessment and new algorithms for ingesting multiple datasets. APS is critical in small deformation analysis using InSAR techniques, particularly in Iran. It is the key step to mitigate effects of APS on the retrieval of long-term deformation rates, e.g. a few mm per year. Thanks to the dry conditions in the region, interferometric SAR pairs with temporal baselines greater than one year can still maintain excellent interferometric coherence in SAR interferometry with Terrain Observation with Progressive Scans SAR (TOPS) SAR data in this region. We processed all potential interferometric pairs, and generated 1757 interferograms for an ascending orbital track 72. The maximum accumulated postseismic deformation of > 10 cm in the direction of Light of Sight (LOS) has been revealed. The preliminary deformation time series shows that aseismic deformation processes occurred logarithmically with time since the occurrence of the 2017 mainshock. Several large aftershocks of M>6 have been recorded in the InSAR results, showing surrounded by the continuous aseismic postseismic deformation. The complete InSAR observations obtained in this study suggest that the “quiet” afterslip triggered by the 2017 Iran earthquake is fully responsible for the later-on aftershock sequence.
Authors: Feng, Wanpeng (1); Samsonov, Sergey (2); Chang, Ling (3)We analyzed synthetic aperture radar (SAR) images from Copernicus Sentinel-1A and -1B satellites operated by the European Space Agency and the Advanced Land Observation Satellite-2 (ALOS-2) satellite operated by Japanese Aerospace Exploration Agency for the 4July 2019 Mw 6.4 and 5 July Mw 7.1 Ridgecrest Earthquakes. We integrate geodetic measurements for the three-dimensional vector field of coseismic surface deformation for the two events and measure the early postseismic deformation, using SAR data from Sentinel-1 and ALOS-2 satellites. We combine less precise large-scale displacements from SAR images by pixel offset tracking or matching, including the along-track component, with the more precise SAR interferometry (InSAR) measurements in the radar line-of-sight direction and intermediate-precision along-track InSAR to estimate all three components of the surface displacement for the two events together. InSAR coherence and coherence change maps the surface disruptions due to fault ruptures reaching the surface. The combined measurements reveal large-scale deformation due to slip at depth and near-fault deformation. The NW-striking fault that was the main rupture in the Mw 7.1 earthquake has variations on the amount of slip reaching the surface. The Garlock fault had triggered slip of about 15 mm along a short section directly south of the main rupture. About 3 km NW of the Mw 7.1 epicenter, the surface fault separates into two strands that form a pull-apart with about 1 meter of down-drop. We image postseismic deformation with InSAR data. Initial analysis of the first months InSAR measurements indicates the pull-apart continued dropping. The main fault had substantial afterslip close to the epicenter just north of where the largest coseismic slip occurred. The area including the fault zone north of the epicenter also had large postseismic rebound, possibly due to poroelastic effects. The northwest corner between the Mw 6.4 fault rupture and the Mw 7.1 fault rupture had measurable postseismic rebound that is very likely poroelastic. Sentinel-1 6-day repeats enable postseismic time-series analysis to reduce atmospheric effects. Slip on a NE-striking fault near the northern end of the main rupture in the first weeks, in the same zone as large and numerous aftershocks along NE-striking and NW-striking trends shows complex deformation.
Authors: Fielding, Eric Jameson (1); Stevenson, Oliver (2); Liu, Zhen (1); Zhong, Minyan (2); Sangha, Simran (1); Liang, Cunren (2); Yun, Sang-Ho (1); Simons, Mark (2); Wang, Kang (3); Burgmann, Roland (3)Using the ESA SNAP software, we made Sentinel_1 interferograms over the area of the November 26, 2019 Mw = 6.4 Durrës earthquake (Albania), with data from the ascending 175 and descending 153 orbits. In both views there are three clear fringes oriented NW-SE are visible, with a peak line of sight length shortening of ~8.4 cm near the village Hamallaj, 15 km NE of Durrës. Most of the deformation pattern is onshore. In both orbits, especially in the descending one, a secondary lobe with opposite sign is visible east of the main one. We picked eighty four values on the positive fringes of the two interferograms to feed an inversion model of ground deformation. We did not pick the negative lobes so as to use them later for control. In order to assign an exact reference to the fringes, and thus an absolute value to the picked points, we calculated, using the Gipsy-Oasis 6.4 software, the co-seismic motion at the GNSS station DUR2 (Durrës) which is located within the fringes. This motion is the following: East -13 ± 2 mm, North -21 ± 2 mm, Up 14 ± 4 mm. Assuming in our model a half-space elastic medium and uniform slip along a rectangular fault surface, we invert the 84 picked measurements to retrieve the parameters of the fault. After a series of inversions made with an exhaustive range of input parameters, they converge towards a stable and robust solution with root mean square residual of 3.6 mm thus ~1/8 of a fringe. The inferred model demonstrate that the earthquake occurred deep in the crust on a low angle fault (23°) dipping towards east with centroid at 18 km depth. The best fitting length and width of the fault are 22 and 11 km and the reverse slip 0.73 m. The seismic moment deduced from our model agrees with those of the published seismic moment tensors. The parametric solutions of the mainshock (such as location and the depth of the epicenter) between the seismological centers varies significantly however the geodetic solution seems to be robust. This result demonstrates the increased significance of InSAR where the seismic data aren’t conclusive. The fault involved in this earthquake is therefore a blind thrust fault having its root on the main basal thrust i.e. along the main Ionian thrust that separates Adria-Apulia from Eurasia.
Authors: Elias, Panagiotis (1); Ganas, Athanassios (2); Tsironi, Varvara (2,3); Cannavo, Flavio (4); Briole, Pierre (5); Valkaniotis, Sotiris (6); Koukouvelas, Ioannis (3); Sokos, Efthimios (3)With the advent of the space-borne synthetic aperture interferometry (InSAR), the complexity (and sometimes simplicity) of earthquake faulting has been revealed. Except for the earthquakes along well-developed mature faults, the surface ruptures caused by an earthquake, as captured by InSAR, are rarely characterized by straight lines, but rather by multiple curved traces. It has also been recognized from InSAR studies that small ruptures are often present around the main rupture traces. These ruptures are considered to be triggered by the static and/or dynamic stress change due to the main rupture, and they can thus be called “secondary ruptures”. The 16 April 2016 Mw 7.0 Kumamoto earthquake in western Japan caused numerous secondary ruptures around the main seismogenic faults. Among them, a set of two secondary faults that run parallel to each other appeared around the northeastern edge of Futagawa Fault, which caused the main rupture. The surface offsets along the secondary faults were well recognized in the interferograms computed from ALOS-2/PALSAR-2 data, exhibiting approximately 1.5 and 3.2 km of rupture lengths, and vertical and dextral offsets of up to 19 cm (Fukushima and Ishimura, 2020). By conducting fault slip inversion, the maximum slip was estimated to be 19 cm at a depth of around 500 meters. The amounts of offsets and the maximum slip are exceptionally large for their lengths, which implies the peculiarity of secondary ruptures compared with seismogenic ruptures. Furthermore, paleoseismic trenching identified a penultimate event with vertical displacements almost similar to the 2016 event, indicating repeated nature of secondary faultings (Ishimura et al., 2021). These results indicate that the rupture physics of the secondary faults are different from that of seismogenic ruptures, but that the secondary faults can provide information on the nearby seismogenic faults. The potential abundance of such secondary ruptures may explain the enigma of the apparent scarcity of low-slip-rate active faults in Japan. Reference: Fukushima, Y., Ishimura, D. Characteristics of secondary-ruptured faults in the Aso Caldera triggered by the 2016 Mw 7.0 Kumamoto earthquake. Earth Planets Space 72, 175 (2020). Ishimura, D., Tsutsumi, H., Toda, S. et al. Repeated triggered ruptures on a distributed secondary fault system: an example from the 2016 Kumamoto earthquake, southwest Japan. Earth Planets Space 73, 39 (2021).
Authors: Fukushima, Yo (1); Ishimura, Daisuke (2)Embedded between the South Tian Shan in the north, the Pamir in the east, and the Hindu Kush in the south, the Tajik basin is a remnant of the Mesozoic-Miocene Tajik-Tarim basin. Since ~12 Ma, ~E-W shortening has been dominating due to the westward collapse of the north-advancing Pamir plateau, inverting the basin into a thin-skinned, W-convex fold-and-thrust belt detached on Upper Jurassic evaporites. The detachment depth is ~6-8 km b.s.l. under most of the basin, shallowing north towards the Tian Shan. Geologic cross sections yield a maximum of 150 km of E-W shortening, distributed between foreland- and hinterland-vergent fold and thrusts. From the eastern to the western rim of the basin, sparse global positioning (GNSS) rates decay from ~15 mm/yr WNW to 2 mm/yr NNW. Seismicity highlights dextral shear along the ~E-striking Ilyak fault – bounding the basin in the north –, and distributed E-W shortening in the central and eastern Tajik basin and in the foothills of the Hindu Kush. The majority of seismic events occurs below the evaporitic detachment. We present rate maps of the region obtained from Sentinel-1 radar interferometric (InSAR) time-series. The underlying data-base comprises 900+ radar scenes, acquired over 2-4.5 years in two view angles (LOS) on 13 frames. The initial LiCSAR interferograms1) and tropospheric delay maps2) were created automatically. The LOS rate maps resulting from a small-baseline inversion (LiCSBAS) were Gaussian-filtered both in space and time. Before decomposition to east and vertical rates, the maps were tied to a Eurasian-stable GNSS reference frame. The final products span from the western basin to the eastern Pamir, and from the southern edge of the Tian Shan to the northern Hindu Kush, covering an area of 270'000 km2 with a spatial sampling of ~400 m. The most reliable results were obtained in the Tajik basin, where the rate maps unveil a combination of basin-scale tectonics, localized halokinesis, effects of extensive irrigation, and seasonal precipitation. Our key findings are: (1) The Tajik basin infill is largely being displaced west as a result of the western collapse of the Pamir. The westward rates decrease away from the Pamir, reflecting dissipated shortening on thin-skinned structures. (2) A bulk of E-W shortening of ~6 mm/yr is absorbed by the most external Babatag (back)thrust with >20 km of past displacement evidenced by borehole data. (3) The Ilyak fault accommodates ~5-8 mm/yr of dextral slip with eastward increasing values; sharply decaying rates suggest a locking depth of ≤1 km or creep. (4) A strong (>10 mm/yr) uplift and westward motion is associated with the sinistral-transpressive Darvaz fault, bounding the basin against the western Pamir. (5) The highest displacement rates >300 mm/yr are demonstrated over the Hoja Mumin salt fountain. 1) See LiCSAR data portal: https://comet.nerc.ac.uk/comet-lics-portal/2) See Generic Atmospheric Correction Online Service for InSAR: http://www.gacos.net/ Figure 1: A) East and B) vertical rates tied to horizontal GNSS data (color-coded circles in A, where red circles were excluded). Intermediate-depth seismicity at >200 km, inlcuding the 2015 Mw7.5 earthquake is marked in purple. Crustal Mw>6.4 earthquakes are marked in green; co- and post-seismic (~1.5 yr) radar scenes of these events were excluded from the InSAR analysis.
Authors: Metzger, Sabrina (1); Gagala, Lukasz (2); Ratschbacher, Lothar (3); Schurr, Bernd (1)We present the first comprehensive mapping of surface fault creep along the Rodgers Creek-Maacama fault zone in northern California, using Sentinel-1 InSAR data. This fault zone is challenging to study with InSAR, due to dense boreal vegetation along much of its length. However, the advent of frequent repeat imaging from the Sentinel-1 mission (12-day repeat coverage from mid-2016 onwards) allows the possibility of coherent InSAR coverage along the whole fault zone. Using the stack processor within the JPL ISCE software, we produce over 400 descending track interferograms, and 300 from an ascending track, with baselines < 100 m, and time spans < 1 year, with overlapping coverage over the fault zone of interest. We analyse these data sets using the small baseline subset algorithm in the StaMPS/MTI code, producing a dense set of displacement time series and velocity estimates in a zone ~10 km wide, centred on the fault trace. By measuring profile offsets from 79 profiles at 2.5 km spacing along strike, we estimate cross-fault line-of-sight velocity changes in both ascending and descending viewing geometries, and decompose these into horizontal and vertical velocity changes, assuming that any horizontal deformation occurs in the local fault-parallel direction. We propagate errors from the scatter in the data along each profile through the same estimation process to obtain uncertainties in those estimated rates. We identify significant (i.e. more than 2 standard deviations above zero) right-lateral horizontal velocity changes along more than 50% of the total length of the Maacama fault. In the vicinity of the cities of Ukiah and Willits, where creep had been inferred from offset cultural features, we estimate creep rates of 3–6 mm/yr, consistent with earlier estimates, but can extend the zone of mapped creep significantly outside of those two cities. The segment of the Maacama fault between the towns of Hopland and Cloverdale peaks at 7–11 mm/yr of fault-parallel creep, and is adjacent to The Geysers, a major geothermal power producing area, raising the intriguing possibility of a causal link. Along the Rodgers Creek fault, we reproduce the results of our earlier InSAR studies, with creep rates of ~2 mm/yr along much of the 30 km segment linking the cities of Santa Rosa and Healdsburg. This improved map of surface fault creep over almost 200 km of a major active fault zone will allow us to improve seismic hazard estimates; areas with surface creep are likely to be barriers to seismic rupture.
Authors: Funning, Gareth (1); Swiatlowski, Jerlyn (1); Bekaert, David (2)On June 17, 2019, an Ms6.0 earthquake occurred in Changning, Sichuan, China (Changning event), which is the largest earthquake within 50 km of the area since records. It attracts great attention as the area has the largest shale gas production and large mineral salt production in China. Based on Interferometric Synthetic Aperture Radar (InSAR), we measure the coseismic deformation and build the fault models of the Changning event and two earlier Ms>5.0 earthquakes (P1:2018/12/16 Ms5.7 and P2:2019/1/3 Ms5.3) using Sentinel-1 and ALOS2 satellite data. The surface deformation caused by the Changning event is mainly uplift with a maximum of 17.2 cm (towards the satellite). Because of the significant non-Double-Couple character of the earthquake, we obtain a double-fault model (FMB) for the event. The final model shows that the Changning event was caused by a small fault (FMB2) and a big fault (FMB1) with a left-lateral strike and thrust slip. The strike of the main fault is 128° with a dip angle of 46°. The total seismic moment obtained by inversion is 6.68×1017 Nm, corresponding to Mw 5.85. The model is roughly consistent with the double slip model provided by seismology. This provides the geodetic evidence for the double slip of the Changning event. Based on the fault model, we analyze the cause of the Changning earthquake from the local tectonic setting, Coulomb stress change, mining, and fluid injection. The results show that the Changning earthquake is likely to be affected by external stimulation. However, the Changning earthquake does not show the characteristics induced by hydraulic fracturing. The stress change on the main fault after the two earlier Ms>5.0 earthquakes is main positive, with a maximum of 0.09 MPa and the water-flooding position of the salt mine is highly coincident with the main slip area of the fault. Therefore, there is no direct evidence showing that the Changning earthquake was related to hydraulic fracturing. We consider that the event may be related to salt mining. P1 and P2 may also play an important role in advancing the Changning earthquake. The characteristics of aftershock distribution indicate that seismic activities may be controlled by crust heterogeneity and structural complexity.
Authors: Gao, Hua (1); Liao, Mingsheng (1); Xu, Wenbin (2); Liu, Xiaoge (2); Fang, Nan (2)In April 2016, a sequence of large earthquakes including Mw7.0 hit the central part of Kyushu island, Japan. A nation-wide GNSS network operated by GSI (GEONET) and SAR sensors such as Sentinel-1 and ALOS-2 detected large coseismic deformation. According to preceding studies, the source of this earthquake sequence mainly consists of two faults: i.e. the ENE-WSW trending Futagawa and NE-SW trending Hinagu faults. Many surface ruptures were also found along these two faults. Sentinel-1 and ALOS-2 detected several spots of deformation on and off the source faults in the Kumamoto plain. GEONET also detected large postseismic deformation following this earthquake sequence, but its spatial resolution is not enough to reveal deformation especially in the vicinity of source faults. In order to detect detailed spatio-temporal distribution of postseismic deformation, we analyzed SAR images acquired by Sentinel-1 and ALOS-2. We used LiCSBAS developed by Morishita et al. (2020) to the analysis of Sentinel-1 interferograms during April 20, 2016 and April 28, 2018 in the COMET-LiCS Sentinel-1 SAR portal. We analyzed ALOS-2 images of three paths with Gamma(R) software. We fixed the first image that was acquired after April 16, the date of the largest shock (Mw7.0) in the sequence, for each path. Interferograms of ALOS-2 sometimes suffers from large ionospheric disturbances. We apply ionospheric correction to interferograms of ALOS-2 using the strategy of Wegmueller et al. (2018). Then we unwrapped interferograms and stack them. In the Sentinel-1 interferograms (Fig. 1), we obtained deformation only in the Kumamoto plain due to low coherence in mountainous region. LOS increase is prevailing on the northwestern side of the Futagawa and Hinagu faults in the descending interferograms. In ascending interferograms, a sharp boundary of positive and negative LOS changes appears along the Hinagu fault. We also found narrow zones of significant LOS changes, which are sandwiched by the Futagawa fault and another parallel fault, and have average rate that exceeds 30 mm/year. Notable off-fault LOS change extends northwestward from the edge of the Futagawa fault, was found in the Kumamoto plain on the northwestern side of source faults. Its average rate amounts to 15 mm/yr. Another sharp change is seen north off the Futagawa fault, where no active faults were found. In the Aso caldera, two spots of significant LOS increases were found in both ascending and descending interferograms: one is located in the vicinity of northern rim, the other is on the northern frank of central cone. On the other hand, LOS decrease was observed around the western rim of the caldera. ALOS-2 interferograms showed us that a westward movement is dominant on the southeast side of the Futagawa-Hinagu faults. Dominant changes along the Hinagu fault can be explained by afterslip on the shallow part of faults. Off fault LOS change in the Kumamoto plane can be interpreted by afterslip of normal faults trending in NW-SE direction. Other deformations, however, are not necessarily understandable with afterslip of faults. There may be spatial correlation of LOS changes with geological structures such as pyroclastic flow deposits from Aso volcano and alluvial deposits. This area is famous for abundant groundwater, whose flow might have affected postseismic deformation. ALOS-2/PALSAR-2 images were provided by JAXA through the activity of the Earthquake SAR WG (GSI) and Pixel (ERI, Univ. Tokyo). Copyright of ALOS-2/PALSAR-2 images belongs to JAXA. Sentinel-1 interferograms were provided by COMET.
Authors: Hashimoto, ManabuTitle of paper: Neotectonic Study Of The Sibi Re-entrant Name: Sukru Onur Karaca Affiliations: MSc Student in Geology, Department of Earth and Atmospheric Sciences / University of Houston Address: 2724 Munger St, Houston, Texas 77023, E-mail: onurkaraca87@hotmail.com / sokaraca@uh.edu, Phone: +1 312 6190732 Proposed topic for the paper Advances in InSAR theory & methodological innovations Earthquakes & tectonics / seismic hazards Presentation preference: Poster Abstract The Suleiman Fold-Thrust Belt represents an active deformational front at the western margin of the Indian plate. It is sub-divided into the eastern, central, and western parts. This study focuses on the western part of the Suleiman Fold-Thrust Belt that comprises two parallel NW-SE oriented faults; Harnai Fault and Karahi Fault. These faults have known thrust components; however, there remains uncertainty about the lateral components of motion. I used 2D seismic data to constrain the subsurface structural geometry and the Small Baseline Subset (SBAS)-Interferometric Synthetic Aperture Radar (InSAR) technique using Sentinel-1A datasets to decompose displacement into the vertical and horizontal components employing ascending and descending path geometries. Tectonics of this area were assessed using two geomorphic indices, i.e., Hypsometric Integral (HI) and Valley Floor Width to Height Ratio (Vf). For InSAR datasets, 45 ascending and 45 descending orbit Sentinel 1A images were processed and used to calculate the vertical and horizontal displacement components. InSAR results show that the Karahi Fault has a dextral (right-lateral) movement, with a rate of movement of ~15 mm/year for descending and ~10 mm/year for ascending path geometries. The Harnai Fault does not show any lateral movement. Seismic data support InSAR results that the Harnai Fault is a blind thrust. This suggests that the block between these two faults displays a clockwise rotation that creates the "bookshelf model". Hypsometric Integral (0.44) and Valley Floor Width to Height Ratio (0.53) indices results also indicate that the northern part of the Karahi Fault is tectonically more active than the southern part of the Harnai Fault.
Authors: Karaca, Sukru OnurOver the past 30 years, Interferometric Synthetic Aperture Radar (InSAR) has established itself as a methodology capable of monitoring co-seismic surface deformations. The main advantages of InSAR for the study of co-seismic deformation are (i) the low cost (actually Sentinel 1 SAR scenes are free as well as archive ERS1 & 2, Envisat), (ii) the high spatial resolution of the observations, and (iii) the existing SAR data that covers most of the regions of the world with relative high temporal resolution (in many earthquake events InSAR measurements are the only available). The earthquake deformation field measured by InSAR is the projection of ground displacement associated with the seismogenic fault in the line of sight (LOS) of the satellite but there is the capability of displacement decomposition in order to calculate the vertical, east-west components of the displacement. Minimum input data requirements in order to calculate the 2D measurements are to obtained from different acquisition geometries InSAR measurements (in LOS) over the same study area. This paper describes the capability of the InSAR to detect and measure co-seismic deformation of a light seismic event with magnitude
Authors: Karavias, Andreas (1); Krassakis, Pavlos (2); Lymperopoulos, Konstantinos (3); Gatsios, Theodoros (4); Parcharidis, Issaak (1)Monitoring slow and large spatial-scale deformation with InSAR has increasingly been successful since the 1990s, allowing for resolving the near-east and vertical displacement components. However, north displacements have been difficult to retrieve due to the limited sensitivity of standard InSAR observations in that direction. The TOPS acquisition mode of Sentinel-1 is achieved by rotating the antenna beam in azimuth direction. Each sub-swath is acquired quasi-simultaneously by switching from burst to burst. In burst-overlap areas between neighboring bursts, the ground is imaged twice from two different view directions (forward-looking and backward-looking), with a large (maximum 1°) squint angle difference. Sentinel-1 TOPS data thus provide an increased squint angle diversity within the burst overlap areas, allowing for a better resolution of the along-track horizontal motion than conventional along-track split-beam interferometry. We applied time-series analysis to a large number of along-track burst-overlap interferometry (BOI) data of the southern Dead Sea fault, from both ascending and descending orbits, to retrieve the near north-south interseismic deformation near the fault. Within the burst overlap areas, the along-track component of displacement is mostly free of tropospheric and topographic residuals, while ionospheric disturbances remain. Averaging in time can reduce most of the ionospheric effects, significantly enhancing the performance of the surface displacement estimation. Our derived along-track velocities from both ascending and descending orbits show a clear and consistent velocity change across the Dead Sea fault, indicating a fault-parallel velocity of 4.5-5 mm/yr, which is consistent with GPS observations. The results demonstrate the applicability of BOI time-series analysis for measuring slow-moving (millimeters per year) north-south deformation by combing hundreds of images.
Authors: Li, Xing; Jónsson, SigurjónThe South Island of New Zealand is formed in the continental transpression between the Australian and Pacific plates. Oblique convergence of ~ 40 mm/yr has resulted in the formation of the Alpine Fault and the ongoing uplift of the Southern Alps. Existing horizontal velocity fields created across New Zealand have been based on comprehensive GNSS campaigns, with data points spaced on average 10-20 km apart, though this can decrease to as little as 2 km in some areas, such as Arthur’s Pass. However, the launch of the Sentinel constellation will allow the measurement of displacements, and therefore the production of velocity products, at potentially much higher temporal and spatial resolutions, in addition to providing a sensitivity to vertical motion that is often absent in GNSS studies. The Southern Alps are a particularly difficult target for InSAR time series analysis, due to the many sources of incoherence that they contain. Seasonal variation in the coherence prevents the use of a simple small-baseline network. Rather, we use a coherence matrix approach to design our network, where every epoch combination with a mean coherence over the Southern Alps higher than a threshold of 0.15 is used. The StaMPS processor is then used, with an iterative unwrapping technique reducing unwrapping errors. We generate LOS velocity maps for each of the two ascending and descending tracks that cover the central Southern Alps. By carrying out a combined inversion of all these velocities using a Best Linear Unbiased Estimator, we can generate a horizontal and vertical velocity field. Insensitivity to N-S motion is dealt with by first removing the north component for the LOS velocity using a GNSS north velocity field. All LOS velocities are placed into an Australian-fixed plate reference system by removing a plane fitted through the residual between the LOS and regional GNSS velocities. However, as the GNSS consists only of a horizontal component, then part of this residual plane must be caused by vertical motion. We therefore try using two iteration inversion, where the vertical velocities from the first inversion are removed from the LOS, the LOS is placed into the GNSS reference frame, and the vertical component added back before the second inversion. We find that InSAR-derived uplift rates vary spatially both along and perpendicular to the Alpine Fault strike, comparable to uplift rates measured by existing GNSS transects, and exhumation rates calculated from thermo-seismic models. Slip-rates and locking depth for the fault is also found to be similar to that measured geologically and geodetically.
Authors: McGrath, Jack (1); Elliott, John (1); Wright, Tim (1); Hamling, Ian (2)The 2016 Mw 7.8 Kaikōura earthquake represents an extremely complex process involving over ten major crustal faults, altering many conventional understandings of multi-fault ruptures. One of the most striking unsolved mysteries relates to the Hikurangi subduction slab, which has long been considered as technically inactive, given the fact that most of the plate motions have been accommodated along the crustal Marlborough Fault System (MFS). However, combined geodetic and seismologic datasets did not rule out potential slip along the southern Hikurangi subduction interface, whose slip magnitude, if any, remains debatable and is hard to determine based on the existing co-seismic observations. Here we present clear evidence of the triggered afterslips on the Hikurangi subduction slab beneath south-western Marlborough using 13 months of GPS and InSAR data. The spatially and temporally correlated atmospheric errors in SAR interferograms are problematic, and hence a new InSAR time series approach, combining the Generic Atmospheric Correction Online Service (GACOS) with a spatial-temporal Atmospheric Phase Screen (APS) filter to facilitate the InSAR time series analysis, is developed. The APS filter reduces the short-wavelength residuals substantially but fails to remove the long-wavelength error even after the ramp removal, revealing that the GACOS correction has played a key role in mitigating long-wavelength atmospheric effects. The resultant InSAR time series enables us to show clear evidence of a triggered Hikurangi subduction slab through time-dependent afterslip modelling, which suggests the Hikurangi subduction slab has an existing but low co-seismic moment release and accommodates up to 0.8 m plate motion during 13 months of post-seismic period. This enables the investigation of the spatial relationship between the afterslip and co-seismic slip distributions on the Hikurangi megathrust, with the interface moved during the mainshock at a depth between 25 and 30 km and the following afterslip occurred at ~32 km. In addition, we also found substantial shallow slip sources on the Needles fault. This implies a much more complex event that undergoes slips along numerous faults with diverse orientations and directions, propagating from the hypocentre both horizontally to the adjacent fault segments and vertically to the underlying deep thrust. Furthermore, the 2016 Kaikōura earthquake turns out to be a giant engine that triggers nearly the entire Hikurangi subduction slab, including a slow slip event on the active northern part, and afterslips on the less active central offshore part plus, most significantly, on the previously inactive south-western end beneath the MFS.
Authors: Yu, Chen (1); Li, Zhenhong (2,1)Land subsidence is a global problem and a morphological phenomenon. The subsidence is a morphology phenomenon that is influenced by the upward motion of the earth. The cause of this phenomenon may be influenced by natural and anthropogenic factors. This phenomenon can cause irreparable damage to the affected areas if not properly managed. Identifying subsidence areas and estimating its rate will certainly play a key role in managing the phenomenon. The prevailing arid climate in most of the interior of Iran and the concentration of increasing industrial, agricultural and drinking water consumption on groundwater resources, has provided a suitable infrastructure for the event of this phenomenon. Differential SAR interferometry (D-In SAR) is capable of imaging Surface deformation over wide areas with centimeter to millimeter accuracy and with high spatial resolution as well as low cost per measurement. One of the most suitable methods for detecting subsidence phenomena is the use of differential Interferometric Synthetic Aperture Radar (D- In SAR) technique. In Kerman province, many areas have faced the phenomenon of subsidence, including plains such as Bardsir or Rfsanjan and this province has a high rate of subsidence in Iran. In this study, an area of Faryab plain in Kerman province was studied. Initially, satellite images in the period 2015 to 2017 were downloaded and the required processors were performed on them. After processing, the results showed that in the selected time intervals of 2015, 2016, and 2017, the amount of land displacement in the whole region is as follow, which is 7 cm subsidence on 5/5/15 until 08/08/2015, with a 5 cm rise and fall and On 08/01/2016 to 22/10/2016, nearly 9 cm subsided and nearly 8 cm uplifted. During the period 22/02/2017 to 31/12/2017, there was a 7 centimeter subsidence and land recession and nearly 6 centimeter elevation. But by cutting the area, separating the main part of the plain, the displacement rate is from -3 cm to + 4 as shown in Figure 8, the final map of subsidence and uplift of the Faryab plain is outlined. Keywords: Subsidence, Radar Interferometry, Sentinel-1, Faryab
Authors: Almodaresi, S.Ali; Keshavarz, S.Razieh; Batayi, JalalRetreat of the Barry Glacier within Prince William Sound, Alaska, has led to the destabilization of landslides flanking the glacier. The largest of these landslides, on the order of 500 M m3, could generate a tsunami that would impact much of the Prince William Sound if it failed and rapidly entered the water of the fjord. This landslide and tsunami threat initiated a state and federal multi-agency response, and has led to ground-based, aerial, and satellite monitoring efforts, community stakeholder involvement, and public outreach. In 2020, the U.S. Geological Survey began monitoring the landslide movement using synthetic aperture radar (SAR) data from the Sentinel-1 and RADARSAT-2 satellites. Monthly interferometric SAR (InSAR) acquisitions show that between May and September 2020, the landslide experienced little to no movement. In October 2020, InSAR revealed downslope movement of over 17 cm, with different parts of the landslide experiencing different spatial and temporal movement. These variations correlate well to recent structural mapping at the landslide using high resolution lidar data, and allow for the preliminary definition of landslide kinematic elements and potential source areas for future scenario-based tsunami modeling. Since late October 2020, snowfall has prevented the use of InSAR for monitoring, although RADARSAT-2 amplitude data is continuing to be monitored until bare ground conditions return in spring 2021. InSAR applications to landslides such as these can improve emergency management efforts, landslide failure forecasts, and slope stability and tsunami modeling.
Authors: Schaefer, Lauren N. (1); Coe, Jeffrey A. (1); Godt, Jonathan W. (1); Wolken, Gabriel J. (2)Here, we present the national deformation map of Romania, along with an analysis of the deformation patterns over one satellite track. To our knowledge, this is the first national deformation map for Romania. The deformation map was created using the persistent scatterer technique (PSI), using more than 1500 Sentinel-1 A/B SLC images. The deformation map can be accessed through the PSTool web application, available at http://pstool.terrasigna.com/. Upon request, the user receives the credentials to access the data. We searched for two deformation patterns, for track 36. The first type are periodic patterns associated with thermal dilation due to yearly seasonal temperature variations. These patterns were detected using spectral analysis, i.e., comparing the power of the spectral components associated with seasonal variations with the rest of the spectral components, in a uniformly resampled deformation profile. The second type represents patterns where the deviation from the local linear trend exceeds a certain threshold. The detector employed here processes a time series sequentially, it computes at every step the deviation from the prediction of the current model (the novelty) and updates the model by integrating the new data. When this deviation exceeds a threshold, the deformation profile is considered of the second type. The results, i.e., detected patterns, were compared with results obtained using a supervised machine learning approach. We created two maps using the pattern detectors described in the previous section. In this work, we experimented with a different approach than our previous publications, which was based on machine learning. This approach has the advantage of being easier to apply to new data sets because it does not require a training set, and it needs less computational resources. However, we implemented a different detector for each type of deformation, instead of a general approach when using machine learning.
Authors: Toma, Stefan-Adrian (1,2); Teleaga, Delia (1); Poncos, Valentin (1); Grozea, Cristian (3)Nowadays, SAR Interferometry (InSAR) has a wide spectrum of applications. In recent six years, by the advent of Sentinel-1 satellite as well as being open access, Sentinel-1 data plays a fundamental role in microwave remote sensing, in particular InSAR. Even though it has an affordable temporal baseline resolution, in some specific situations, users are still confronted with decorrelation issue because of changing in the behavior of scatterers in master and slave images acquisition time. Hence, to deal with this drawback, it is of vital necessity to turn to persistent scatterer interferometry (PSI) techniques. Fortunately, the latest version of StaMPS, a well-known algorithm to detect persistent scatterers (PS) is strengthened with new functions and scripts working on pre-processed Sentinel-1 data. This technique selects such scatterers based on two substantial factors. At first, it uses the amplitude dispersion, an index of scatterer stability over a time-series of measurements, to recognize a large number of points as candidate PSs. In the next step to be considered as the advantage of this method, among pixels selected by the former index, those having the lower amount of phase diversity over the period will be selected as final PSs. The PS pixels are then unwrapped using a two-step approach carried out in the time and space spaces, separately. In addition to StaMPS, GAMMA Remote Sensing and Consulting AG (GAMMA) has presented another algorithm called interferometric point target analysis (IPTA) for selection and processing PS points. In IPTA, interferometric phases are only calculated for the selected points. Thus, based on the co-registered single look complex (SLC) images, a candidate list of point targets is identified using both amplitude dispersion and spectral diversity. After identification of the PS candidates, the stack of differential interferograms is analyzed in temporal domain. This step which is done for all points is to estimate both height and deformation rates by performing a least-squares regression on the differential phases. The ability to step by step improvement of the results namely consideration of height correction, deformation rate, baseline refinement as well as atmospheric correction are the advantage of this method. Two steps are critical in each PSI approach: i) PS selection and ii) phase unwrapping. In what follows, the performance of two StaMPS and IPTA methods for landslide monitoring in Kahroud located in the north of Iran is evaluated. 28 Sentinel-1 SLC images acquired between 2014 and 2016 during which the landslide is active, are applied. To have a fair appraisal of their functionality, the pre-processing step including the co-registration and generation of the differential interferograms which is required in both methods is performed in GAMMA., The results obtained from both methods are compared in detail in terms of the number of PS selected in the landslide area and unwrapping errors. The results extracted from the methods are further verified with a comprehensive conventional InSAR time series analysis which tries to estimate the deformation rate, residual topographic effects as well as atmospheric signal. To make this comparison more tangible, a number of assessment criterion are calculated. Detailed investigations reveal some merits of StaMPS for PS detection in higher density while some other superiority of IPTA in deformation rate estimation.
Authors: Tavakkoliestahbanati, Amin; Yarmohammadtouski, Milad; Veiskarami, Mehdi; Dehghani, MaryamABSTRACT India has the fifth largest coal reserves and is the second-largest producer of coal in the world. The coal in India is mostly produced from opencast mining methods due to less production cost, high mechanization and low wastage. Both open cast and underground mining activities are reasons for environmental damage and pollution. Underground mining methods are suitable for the extraction of deeper coal seams, which may lead to land subsidence. Opencast mining is a surface mining technique suitable for shallow coal seams. Opencast methods face challenges due to the instability of bench slopes and the deformation of overburden dumps. The main problem in an underground mine is to provide adequate support to the roof. In contrast, in an opencast mine, the stability of benches and overburden dumps has to be appropriately maintained. Slope and overburden failures occur when they lose self-retainability due to seasonal changes, induced fracture planes, and weathering of rock. They tend to move in a downward direction under the gravity of their weight. Due to the sudden and unexpected collapse nature of these slopes, the events of coalmine failures were a disastrous situation leading to loss of life, costly equipment, and production. Hence, it is often necessary to monitor mine slopes, and overburden dumps continuously to understand their behaviour and give early warning. The results from these studies also assist in taking preventive measures to safeguard mines and mine workers against sudden failures. In this regard, the spaceborne SAR interferometry (InSAR) technique is advantageous over conventional ground-based survey methods for measuring the ground deformation efficiently over the mine slopes and overburden dumps. Synthetic Aperture Radar Interferometry (InSAR) utilizes the phase difference between two image acquisitions to obtain Digital Elevation Model (DEM) of a terrain. Differential SAR interferometry (DInSAR) is an extension of the conventional InSAR technique, which is carried out by subtracting the estimated topographic phase from a digital elevation model (DEM) to obtain phase component due to deformation. However, problems like temporal decorrelation, atmospheric phase delay, and baseline restriction have put some limitations on operational studies. Persistent Scatterer Interferometry technique (PSI) is an advanced InSAR technique, which can achieve sub-centimeter level accuracy in mapping time series deformation by reducing the effect of atmospheric phase delay and geometric decorrelation up to a large extent by exploiting a series of interferograms. Coherence is preserved for those pixels whose backscattering response is dominated by persistent scatterers (PS). In this study, several SAR images of the European Space Agency’s Sentinel-1 satellite with VV polarization are collected for two major coalfields in India viz. Jharia (SECL), Singruali (NCL) and one lignite field (Neyveli lignite corporation) from 2018 to 2019. The PSI process involved selecting one image as a master to form interferograms with all other images. Differential interferograms are formed by removing the topographic phase component estimated using a 30 m SRTM DEM. Then StaMPS (Stanford Method for Persistent Scatterers) software is used to analyze the PS pixels. Initially, based on the amplitude dispersion value < 0.42, some pixels are selected as permanent scatterers. Phase analysis is carried out further to weed out some of the initially selected PS pixels by using spatial correlation of the interferometric phase. The DEM error is estimated and removed from the interferometric phase using an iterative regression approach. In the next step, 3D phase unwrapping is performed in time and space to estimate the unwrapped phase of the PS points. Phase delay due to the atmospheric phase screen is estimated by applying a high pass filter in the temporal domain and low pass filter in the spatial domain to remove it from the total phase component. After PSI analysis using StaMPS software, it is observed that the majority of the PS points are spotted on the slopes facing in the line of sight (LOS) direction of the radar, overburden dumpsites and surrounding urban areas. Jharia coal mines are located on the east side of India in Jharia, Jharkhand, which consists of nine large opencast mines near Damodar river valley. Due to the unscientific methods adopted by erstwhile private owners, large areas in coal mines were subjected to mine fires and subsidence. The mine slope benches of Dobari, Nayadh, Gondhudih, and Angarpathar opencast mines show deformation up to 4.2, 3.4, 5.2, and 3.7 mm/year away from the radar from 2018 to 2019. The overburden dumps sites of Alkusa, Talgaria, Suranga, Lodna, Bera, Keshalpur, and Barmasia areas show subsidence up to 6.4, 16.1, 11.6, 17.2, 9.5, 6.1 and 8.2 mm/year from 2018 to 2019. Apart from mining areas, subsidence is also observed in Dhanbad, Bhelatand colony, and Santaldih power plant areas up to 4, 5.7, and 5.5 mm/year. Singruali coalfield is located in the Singruali district of Madhya Pradesh and consists of 10 large, highly mechanized open cast mines. The observations of the Singruali coalfield indicate that the mine slopes from Amlohri, Gorbi Block-B, Nigahi, Jayant, and Chilkadand mines undergo deformation up to 22, 17.9, 19.6, 20.1 and 22.7 mm/year away from the radar. Some PS points with subsidence rate up to 19.1 and 15.6 mm/year are detected on the overburden dumps of Chilkadand and Kakri mines. The bench slope adjacent to the road in Basi mine is also deforming at the rate of 16 mm/year. For mining, one tonne of lignite in Neyveli lignite open cast mines, 13 tons of water has to be pumped out due to the presence of underground water aquifer below the lignite seams. A maximum subsidence rate of 38.7 mm/year on the overburden dumpsite situated in Mine -II and upliftment up to 22.8 mm/year in adjacent urban areas (MK Colony and GP Nagar) are observed in Neyveli area. There are two slope benches, one from the central region of Mine-1A, and the other from the southside of Mine-II showing deformation up to 21 to 29 mm/year away from the radar. The results in this study emphasize the deforming zones in the coalfields viz. bench slopes, overburden dump sites and surrounding urban areas. A high density of PS points on the deforming bench slopes and overburden sites can help in greater visualization of the deformation phenomena. Due to the presence of vegetation, less number of PS points are observed on some of the overburden dumpsites.
Authors: Sriramoju, Manoj Kumar (1); Vaka, Divya Sekhar (1); Rao, Y. S. (1); Kumar, Vineet (2)Main aim of this study is to investigate surface displacement occurrence in Bangkok, Thailand from 2017 to 2019 using Persistent Scatterers Interferometric Synthetic Aperture Radar (PS-InSAR) time-series analysis. Bangkok, the capital city of Thailand, sits on the flood plain of the Chao Phraya River, which traverses the Lower Chao Phraya Basin towards the Gulf of Thailand, with an average elevation of 1–2 meters above mean sea level. Soft to stiff dark gray to black clay, also known as “Bangkok Clay,” makes up the top layer of Bangkok. It ranges in thickness from 20–30 m. Beneath the Bangkok Clay layer are several confined aquifers, primarily characterized by sand and gravel intercalated by clay layers. The most heavily exploited aquifers are the three more superficial such as Phra Pradaeng, Nakhon Luang, and Nonthaburi aquifers, however, deeper aquifers have also been pumped, mainly by industries and the commercial sector. Due to over extraction of groundwater, Bangkok city and its peripheral areas faced land subsidence problems since 1970s. However, since 1998 the rates of subsidence have been gradually decreasing since the Thai Government introduced restrictions on the withdrawal and use of underground water over the whole the metropolitan area. Government of Thailand since the years 1978 monitored the evolution of subsidence in the whole metropolitan region by establishing and monitoring an extensive network of leveling benchmarks. This study is using 67 images in ascending and 113 images in descending geometries acquired between July 2017 and December 2019 by Sentinel-1 A and B satellites in Interferometric Wide Swath (IW) mode. We generated and stacked the differential interferograms within ESA SNAP application. We then processed PS-InSAR time-series in StaMPS Software. The PS candidates were selected by setting the amplitude dispersion at 0.35 to reduce the signal noise. We applied the APS GACOS method to take into account the atmospheric influence. Both ascending and descending geometries showed comparable results in term of density of PS as well as displacement patterns. Uplifting areas can be clearly identified along the Chaophaya river in the central areas of Bangkok. While high rates of subsidence can be tracked around Bang Na district and Prawet District at the Eastern Bangkok and Bang Khun Thian district at the Western Bangkok.
Authors: Soontorn, Jittrapon (1); Pattela, Taraka Venkatadripathi (2); Ninsawat, Sarawut (1); Giao, Pham Huy (1); Virdis, Salvatore G.P. (1)Bolivia is prone to landslides due to the rugged topography of the Andes in western Bolivia and the complex regional hydrogeological conditions with high erosion rates. Our study area, Villa de la Independencia, the capital of the Ayopaya Province, Department of Cochabamba, Bolivia, is in such an environment. Motivated by the observed cracks in downtown walls and the wall collapse of the eastern bus station, we used satellite radar observations to fully trace the landslide movements and systematically assess the failure risk in this area. Landslide monitoring with Sentinel-1 data on such a local scale can provide continuous deformation information in short revisit cycles, which is beneficial to characterise the spatial pattern and temporal behaviour of landslide motions. Further analysis on the deformation time series and environmental changes can assist in identifying potential failure triggers of a landslide. We used Sentinel-1 data spanning five years to monitor the landslide activity in Villa de la Independencia where the city lies at the foot of a slope. 162 Sentinel-1 descending images from 16 October 2014 to 31 December 2019 and 135 ascending images from 3 November 2014 to 25 December 2019 with a minimum time interval of six days were collected and processed. To keep the spatial resolution of Sentinel-1 and reduce atmospheric disturbance in the study area, we applied a single-look Small Baseline Subset (SBAS) method integrated with tropospheric delay corrections. The multiple-master interferograms were imported into StaMPS after applying corrections from the Generic Atmospheric Correction Online Service (GACOS). Displacement velocity maps in the Line of Sight (LOS) show that there is a landslide risky area 2.5 km southeast of the city, with a velocity of about 30 mm/yr. The risky slope deserves attention because it is below Highway 25 and its instability would threaten the public transit. The city centre and the slope above it have different east-west movements due to the different aspect. Combing the cumulative displacements of the descending and ascending tracks, we estimated the optimal slope angle and moving direction of the sliding plane. The city centre is moving 13.3° east-south along a slope of 3.4° while the upper block is sliding 48.1° west-south along a slope of 8.0°. LOS displacement time series and 30-day cumulative rainfall data show that both the city centre and upper block have begun to accelerate after 2018 in response to the increased precipitation. The deformation projected onto the slope has accumulated by about 100 mm during the rainy seasons of 2018 and 2019. To conclude, the landslide risk in the area of Villa de la Independencia is relatively high since 2018 due to the possible triggering by precipitation.
Authors: Song, Chuang (1); Li, Zhenhong (2,1); Utili, Stefano (1); Yu, Chen (1)Land subsidence or gradual depression of the ground surface occurs due to groundwater over-exploitation. The pressure drop caused by groundwater pumping results in an increase in the effective stress applied to the aquifer skeleton and the aquifer consequently consolidates due to the increased stress. The subsidence phenomenon, in addition to the destruction of buildings, change in the height and slope of drains and stream channels, damages the screen and casing of pumping wells through consolidation of layers adjacent to wells and reduces their productivity. Some areas are more prone to subsidence due to climatic, geological, and humidity conditions. Significant land subsidence has been reported as a result of pumping wells in many areas of the world, such as California and Texas in the United States, Bangkok, Mexico, London and Tokyo cities and China, Greece, and Nigeria. Estimating the land subsidence rate in an area can play a major role in the proper management of water and soil resources. For this purpose, the capability of DInSAR technique to measure land subsidence at low cost and large scale compared to conventional methods such as slope inclinometer, accelerometer, GPS survey, tiltmeters and strain gauges has been well proven. Ground surface displacements such as subsidence can be measured with high spatial and temporal resolution and millimeter accuracy. This technology is based on the analysis of pair radar images acquired at two different times to determine changes in the line of sight (LOS) of features. Nowadays, land subsidence is considered a major hazard in many plains of Iran, especially during drought periods due to over-exploitation of groundwater resources. The aim of this study is to determine the rate of land subsidence in the Najafabad, study area in Isfahan, Iran using the radar data. Parts of this region have long been considered an agricultural pole, but unfortunately in recent decades drilling of deep and semi-deep wells and consequently unbalance groundwater exploitation have resulted in a significant drop in water table of the aquifer. Therefore, the occurrence of land subsidence is not unexpected, and more susceptible parts of the region to this hazard should be identified and monitored for management and preventive measures, since subsidence is controlled by several variables. Therefore, causes of land subsidence in the Najafabad aquifer have been investigated using differential radar interferometry (DInSAR) technique. The ground subsidence estimated by processing of the ASAR (acquired between 2003 and 2009), PALSAR (acquired between 2006 and 2010) and SENTINEL (acquired between 2016 and 2017) radar data was verified by comparing with the data of differential leveling performed between 1984 to 2004 and the hydrogeological analysis of groundwater level changes during 2002 and 2014 and aerial extent of land subsidence in the study was determined. Results of the radar data processing indicates that the land subsidence in the Najafabad aquifer is 41 cm during a 6 years period with an average annual rate of 6.7 cm based on data of the ASAR sensor and 30 cm for a 4 years period and an annual rate of 7.7 cm on the basis of the PALSAR sensor data. However, the land subsidence was measured to be 7.8 centimeters between 2016 and 2017 based on the SENTINEL data processing, indicating an annual increase in the land subsidence rate. Accordingly, most of the displacements have occurred in the Tiranchi, Kushk, Ghahedrijan, Goldasht and Falavarjan cities. Results of the analysis beside suggesting spatial occurrence of the land subsidence, show 41 cm land subsidence in the Tiranchi area. The hydrogeological analysis of simultaneous data of 46 observation wells in the aquifer with the radar data acquied between 2002 and 2014 between 5 and 55 m below ground level in south, east to south North of the Najafabad city corresponds to the subsidence zones of the earth. The spatial extent of groundwater depletion well agrees with the extent of the land subsidence obtained from the processing of radar satellite data indicating a significant relationship between the groundwater level drawdown and land subsidence. Among the radar satellite data used in this study, PALSAR and SENTINEL sensors were identified by DInSAR method with the best spatial accuracy and quality. Results of the radar data processing can be employed for the hazard zonation directly utilized for management and planning of control and preventive measures.
Authors: Shirani, Kourosh (1); Pasandi, Mehrdad (2)The paper presents the preliminary results of an ongoing international multidisciplinary project funded by the United Nations Development Programme (UNDP) over the Polog Region in R.N. Macedonia. The territory of R.N. Macedonia is severely and frequently hit by landslides, responsible for direct and indirect impacts on the structures, infrastructure and population. Northwest part of the country, more precisely the Polog Region, is the most prone to landsliding processes, primarily triggered by heavy rainfalls and favored by its geological, morphological and tectonic setting. Over the study area, extending for approximately 1000 km2, preliminary landslide susceptibility studies were performed; however, the extent of landslide hazard in the region needs to be understood to a higher level of detail. The necessity for appropriate risk management for this part of the country is urging and landslide susceptibility, hazard and risk assessments are deemed necessary. This means that a landslide inventory of appropriate quality for the region is necessary, in order to perform the proper selection and validation of the susceptibility (or hazard) models and enable systematic approach in the management of the landslide hazard in general terms. The aim of the project is to derive a landslide inventory via the detection and characterization of landslides through i) conventional methods (geological and geomorphological analyses supported by field checks), ii) satellite remote sensing DInSAR techniques, iii) simplified numerical analyses at regional scale. The results, together with the analysis of environmental factors, triggering factors and elements at risk will represent the necessary background for more detailed landslide susceptibility, hazard and risk assessment. Conventional geotechnical monitoring of ground displacements can be valuably complemented by the processing of images acquired by spaceborne Synthetic Aperture Radar (SAR) sensors and processed with advanced multipass Differential Interferometric (DInSAR) techniques. For the purpose of the project, a set of 237 multipass SAR images acquired over ascending orbits by the Sentinel-1A and -B satellites of the European Space Agency, on relative orbit 175 spanning a period from April 2015 to December 2019, was exploited. Based on the extent of the Polog Region with respect to the wide coverage of a single Sentinel-1 slice acquired in standard Interferometric Wide Swath (IW) acquisition mode, the processing has been limited to four bursts of the IW2 subswath. Data have been processed at the full available resolution via the SAR Tomography technique for the detection of single Persistent Scatterers (PS). The analysis of the resulting surface deformation measurements pursued two main goals. The first one concerned the check of the inventoried landslides in terms of boundaries and state of activity: because of the severe decorrelation induced by the large presence of mountainous areas; 24 landslides out of a total of 136 resulted to be covered. Interestingly, only nine of the covered landslides were fully classified in the inventory in terms of type and state of activity. Accordingly, the analysis of DInSAR data could be useful to provide some insights into the kinematics of the above mentioned phenomena. More interestingly, DInSAR deformation measurements have also highlighted the presence of “hotspot” characterized by the presence of moving PS that can be exploited, jointly with geomorphological/geological criteria and photointerpretation, for the detection of unmapped unstable sloping areas and/or to address more detailed studies aimed at landslide characterization that should be addressed in the subsequent stage of the project.
Authors: Reale, Diego (1); Sansosti, Eugenio (1); Simona, Verde (1); Gianfranco, Fornaro (1); Gianfranco, Nicodemo (2); Dario, Peduto (2); Milorad, Jovanovski (3); Igor, Peshevski (3); Gjorgi, Gjorgiev (3); Natasha, Nedelkovska (4)Over the last years, there is great interest in the detection and monitoring of ground deformation phenomena. Highlighting the areas that suffer from such phenomena is crucial for both civil protection and the optimization of urban development plan and preservation. In addition, deformation data are considered key information for the assessment of the anthropogenic impact on the natural environment, stemming from e.g., groundwater over–exploitation and industrialization. Recent advances in spaceborne SAR interferometry (InSAR), especially after the launch of the Sentinel-1 constellation, have enhanced our ability to measure remotely large-scale ground deformation, with mm-level accuracy. The main objective of this study was to use this technique to investigate the temporal patterns of ground deformation in the greater area of Thessaloniki. Thessaloniki is located in the northern part of Greece, between Thermaikos Gulf and Vertiskos Mountain, and over the years has become a highly populated and heavily industrialized region. Due to its geographic location and geophysical seating, the city is dominated by almost continuous, but low in magnitude, tectonic activity, suggesting that the area is subject to small but important displacements. A large set of ascending and descending C-band Sentinel1-SAR images, covering the period between 2017 and 2019, was used for the investigation of these phenomena in the area under study. The interferometric deformation measurements were enhanced using the SBAS (Small Baseline Subsets) algorithm, an advanced multi-pass DInSAR technique that combines unwrapped interferograms with small spatial and temporal baselines to minimize the topographic and atmospheric artifacts. The analysis was generated using the EO processing services provided by the Geohazards Exploitation Platform (GEP) of the European Space Agency (ESA). High subsidence rates of about -2 cm/year were observed at both the western and the eastern part of the area, where Axios and Anthemountas basins are located, respectively. On the contrary, at Mygdonia fault, which is resides northwards from the city of Thessaloniki, deformation rates were about +1 cm/year, suggesting that the site is dominated by an uplift trend. The results of the study concluded that both geophysical and human activities have an intertemporal impact in the area, which must be taken into account by local authorities and the scientific community. Moreover, the results made clear that advanced interferometric techniques should be considered as both a valuable and low-cost tool for validating ground deformation and an alternative to ground-based measurements.
Authors: Triantafyllou, Anastasia; Vergos, GeorgeLandslide is one of the major and most frequently occurring geo-hazards around the world. After the 2005 Kashmir Earthquake in Pakistan, a large number of landslides were triggered or influenced. Hence, complete landslide inventory mapping, which systematically records the time, quantity, type, location, scale and other information of existing landslides, is of great significance and importance for the investigation and evaluation of landslide hazards at regional scale. Moreover, the landslide inventory updates of Hunza River watershed in Pakistan are recognized as important accomplishments for the key project “Investigation on the Geohazards Prevention and Mitigation along China-Pakistan Economic Corridor under Climate Change” operated by research institutions from China and Pakistan collaboratively. Traditional landslide inventory mapping mainly relies on the geological field surveys, optical remote sensing images, UAV photogrammetry and digital elevation mode analysis. Most of them are based on the investigation and inventory mapping of existing landslides, but lack of analysis of historical landslides and the spatio-temporal evolutions of developing landslides. The differential interferometric synthetic aperture radar technique (D-InSAR) has the advantage of high spatial resolutions, large simultaneous observation areas, and competitive accuracies, without geographical or man-made restrictions, thus could realize the macro dynamic monitoring of landslide hazards at large scale, and exhibits an important supplement to the traditional methods for landslide inventory mapping. However, due to the inherent side-look imaging mode of radar satellites, radar observations are seriously affected by the geometric distortions in mountainous areas. Therefore, it is inevitable that single track radar observation cannot cover various landslides with different slopes and aspects in a large scale, resulting in the blind area of landslide identification and the omission of landslide inventory mapping. This study employed time series InSAR method with multi-platform and multi-orbit radar observations, to carry out the multi-temporal inventory updates and long time-series deformation monitoring for historical and existing landslides in Hunza River watershed of Pakistan. A large number of SAR datasets acquired by L-band ascending ALOS/PALSAR, C-band descending and ascending Sentinel-1A/B images are involved in the time series InSAR processing chain. In the experimental area of this study, the dominant process that shaped the morphology of Hunza River main channel should be mass wasting. Various types of slides have been recognized in this region, such as rotational and retrogressive slumps, debris flows, earthflows, translational bedrock slides, complex and composite landslides. As Hunza River watershed has undulating terrains and abundant water vapor on valley slopes, which resulted in the severe problems of atmospheric disturbance in time series InSAR processing, TSInSAR-AEM method proposed by Zhenhong Li is deployed with integration of atmospheric phase estimation, which accurately separated the atmospheric disturbance and improved the reliability of time series deformation results. The application of TSInSAR-AEM model not only helps to separate atmospheric artefacts from deformation signals more accurately, but also accelerates the convergence achieved during iterations. Time series InSAR processing is carried out for independent SAR dataset, while the multi-temporal landslide inventory updates could be achieved by mosaic combinations of suspected landslides identified by multi-platform and multi-orbit radar datasets. Moreover, the DEM gradient maps and SAR intensity maps are also used to verify the identified landslides in Hunza River watershed. The reliability and accuracy of InSAR recognition results are further verified by published literatures and field investigations. To summarize, the fusion of multi-platform and multi-orbit radar remote sensing observations for landslide inventory updates can highlight the following advantages: 1) It could provide multiple observations of different incidence angles for the same ground object, so as to make up the inherent defects of radar side-look imaging, and achieve more effective coverage of radar observations; 2) Independent time-series deformation observations obtained from multiple SAR datasets could be cross verified with each other; 3) On the basis of reasonable hypothesis or combined with prior knowledge, the 2D/3D deformation fields of landslides could be retrieved by using the geometric differences of orbital observations from diversified radar datasets, which provides the analysis basis for understanding the evolution mechanism of various landslides, especially for scientists who do not know much about radar remote sensing techniques.
Authors: Qu, Tengteng (1,4); Su, Lijun (2,3,4,5); Li, Zhenhong (6,7)The “Pas de l’Ours” landslide is a paleo-landslide located in the Queyras valley (Southeast French Alps). The landslide has been reactivated in Spring 2017 when first signs of intense deformation appeared on the road located at the foot of the slope. The total moving mass is estimated at 17 million cubic meters, with a width of 1 km and a length of 600 m, which makes it currently one of the largest active landslides in the France. In addition to the large deforming mass, numerous rockfalls and mudflows have occurred and have severely damaged the road. Since 2018, the road is closed due to its destruction by the landslide. The landslide is still active and alternates between periods of strong acceleration (m.day-1) mostly occurring in late winter and early spring time, and slower motion (mm to cm. day-1) occurring in late spring and summer time. Since 2017, the landslide has been instrumented to monitor several parameters such as the displacement rate (measured by a ground-based-SAR, four GNSS receivers and a photo camera), the micro-seismic activity of the slope and the water levels in boreholes. Beside in-situ measurements, space-borne observations from the Copernicus Sentinel-1 and 2 sensors provided regular acquisitions. These datasets are processed separately with different techniques to retrieve the displacement of the landslide over time. The Sentinel-2 dataset is processed on the Geohazards Exploitation Platform (GEP) by the Multiple-Pair Image Correlation (MPIC), an offset tracking algorithm; the Sentinel-1 data are processed by the SqueeSAR® algorithm providing displacement time series for Distributed and Persistent Scatterers. The resulting time series are cross-validated with the in-situ measures. The offset tracking technique is sensitive to ground motion of metric magnitude in the horizontal compoments; the SqueeSAR® technique is sensitive to smaller displacement of few millimetre to several centimetres. The complementary of these techniques is fused to analyse the activity of the landslide in the periods of acceleration and deceleration. The 2014-2019 yearly accelerations/decelerations are compared and correlated to the meteorological conditions in order to understand the driving mechanisms of the instability through time.
Authors: Provost, Floriane (1); Philippe, Bally (1); Ferretti, Alessandro (2); Malet, Jean-Philippe (3)Deformation monitoring by means of SAR interferometry in areas under the influence of mining exploitation is very challenging due to quick non-linear deformation rates. Various interferometric methods can be used for this purpose, but all of them have their advantages and limitations. Persistent Scatterers Interferometry (PSInSAR) is probably the most reliable and accurate method for deformation measurement in urbanized areas with relatively small linear deformation rates because this method mange quite well to reduce deformation components due to artefacts related to the changing atmosphere. However, the PSInSAR method fails in big subsidence troughs that correspond directly to underground exploitation. On the other hand, the classical approach of Differential Interferometry Synthetic Aperture Radar (DInSAR) is able to measure deformation in troughs with big subsidence values but the deformation phase component is strongly contaminated by environmental artefacts. Therefore, to measure effectively subsidence in the mining areas we propose an approach being a combination of the PSInSAR and DInSAR methods. We use the DInSAR method in the subsidence trough and the PSInSAR method in surrounding areas. The differences between the PSInSAR Line of Sight (LOS) values and DInSAR LOS values calculated in the stable and small deformation areas allow for the trend modelling over the study area. We assume, that this trend represents mainly the atmospheric phase component not subtracted in the conventional DInSAR approach. Therefore, we transform DInSAR results into the PSI domain by means of the trend removal. This proceeding is applied separately to the ascending and descending Sentinel-1A/B data. Then, values of the PSInSAR and DInSAR results are fused by completely regularised splines functions, and LOS deformations from ascending and descending orbit were used for 3D decomposition (setting north component equal to zero). We apply the proposed approach in the study area located in the south-western part of Upper Silesian Coal Basin over the area of the oldest mining in the Poland-Rydułtowy mine. Six days Sentinel-1A/B data covering the period between 04.03.2018 and 13.03.2019 has been used for this study. Levelling data for the same time span has been used as the reference data. The integrated PSInSAR-DInSAR approach allowed us to detect maximal subsidence in the study area equal to 1.05 m. The root means squares error calculated based on the reference data is equal to 22 mm. The archived results indicates that InSAR can be effectively used form monitoring mining subsidence. The proposed approach can be utilized for the effective estimation of large gradients and nonlinear deformations as well as the small displacements in the area surrounding the subsidence trough.
Authors: Pawluszek-Filipiak, Kamila; Borkowski, AndrzejIn this study, we investigate the occurrence of seasonal ground displacement in the coastal area of the Albegna river plain (central Italy) for the past four years (2017 to 2020) by using the Persistent Scatterer Interferometry (PSI) technique. Land use/cover in the study area is characterized primarily by agriculture, urban towns, and industrial clusters. By some years, damages to buildings and infrastructure have been reported to be connected to ground displacement which may be related to groundwater exploitation. However, the extent and amount of ground displacements are only qualitatively known in the area. So, as a first step, we applied the single master PSI approach from October 2017 to September 2020 using Sentinel-1 SAR data to map the spatial and temporal extent of the displacements by the Stanford Method for Persistent Scatterers (StaMPS). The mean velocity data estimated is a long-term linear trend of the time-displacement function; hence they did not allow to characterize the nonlinear seasonal and the local behavior of the process. Moreover, the density of the PS around the buildings where damages have been reported are deficient due to the loss of coherence related to long temporal baselines. To overcome these limitations, we implemented the single master PSI approach splitting the Sentinel-1 data into two seasonal subsets per year. Each seasonal subset is composed of images according to the winter (“wet” conditions) and summer (“dry” conditions) seasons and co-registered to a unique master image (individual for each seasonal dataset) to minimize the dispersion of geometrical and temporal baseline values. A highly accurate Lidar DSM was used for the interferogram formation between all SAR images and the master. Finally, the PSI technique was performed using the StaMPS, and the mean velocity for each season was estimated. The estimated velocity was evaluated as a linear trend, so representing measured displacements into six months. The displacement maps for each season allowed us to highlight the seasonal ground displacement trends with maximum magnitude in the order of 20 mm in six months, but most of the PS was observed to move less than 5 mm. High-accuracy levelling data, coeval with SAR acquisitions, were acquired within the study area, providing the vertical ground displacement values measured similar to those obtained by the StaMPS PSI. While the seasonal approach decreased the temporal decorrelation thus improving the coherence in the area, the density of the PS is not consistent among seasons. Also, the low magnitude of seasonal displacements with higher standard deviation does not allow us to spatially map the displacement field, even though local significant displacement values are recognized. We analyse both the groundwater level change data and the geotechnical properties of sub-surface soils in the study area, and we infer that local soil swelling-shrinkage processes may be regarded as possible causes of the seasonal ground displacements detected by the PSInSAR. Keywords: PSInSAR, Seasonal Ground displacement, soil swelling, and shrinkage
Authors: Pattela, Taraka Venkatadripathi (1); Disperati, Leonardo (1); Virdis, Salvatore G.P. (2); Pignatiello, Marco (1); Fantozzi, Pier Lorenzo (1); Morelli, Alberto (1)Global mean sea level rise (GMSL) is a serious threat to the low-lying coastal areas. GMSL rise reconstructed from relative sea-level changes are observed by tide gauges that installed around the world. The relative sea-level changes measured with reference to the local datum established on the adjacent land. For reliable RSL measurements, the tide gauge records are corrected for several components including GIA. However, vertical land motion (VLM) which is a local component that significantly affects the tide gauge measurements along the low-lying coastal areas. Existing methods such as periodic leveling and GNSS approaches are often inadequate due to the lack the GNSS stations. Permanent GNSS stations are used to measure the VLM trend at tide gauges, however, only few tide gauges are equipped with collocated GNSS stations. ue to lack of permanent collocated GNSS stations, other techniques for instance Time-series InSAR techniques were provide multi-temporal relative land motion estimates over the large areas. Persistent Scatterers InSAR (PS-InSAR), Small Baseline Subset (SBAS) are widely accepted for estimating the land displacement over time-scale. In this study, we applied the space-borne Interferometric SAR technique to measure the local ground motion using Sentinel-1 SAR data. The Korean peninsula is surrounded by the East Sea/Sea of Japan, the Yellow Sea and the East China Sea have continuously monitoring tide gauges with a record length of more than 30 years. We acquire C-band Sentinel-1 SAR data (both ascending and descending mode) over the Korean Peninsula during 2014/11 and 2019/04. We estimate the high-resolution (~ 10 m) land motion at tide gauges (mm-level accuracy) over these 21 tide gauges and, compared with available collocated GNSS observations. 2D displacements (vertical and horizontal) are derived from ascending and descending mode InSAR displacements. The linear trend of VLM observed from our InSAR estimates is used to compensate for the relative velocity of sea-level changes observed from tide gauges. However, the inadequate PS pixels density often introduces error sources such as unwrapping error. Stanford Method for Persistent Scatterers (StaMPS) algorithm employs both multi-temporal PS-InSAR and SBAS techniques. StaMPS-SBAS techniques have been developed to address various decorrelation problems affecting deformations. However, the applicability of both StaMPS PS-InSAR and SBAS algorithms over coastal and non-urban regions are still remain challenging. To overcome this issue, we present the sequential pair selection for StaMPS-SBAS algorithm to measure the ground motion at Pohang and Yeosu tide gauges. According to the Korean Hydrographic and Oceanographic Agency (KHOA), several tide gauges installed on the coastlines of the Korean peninsula has showed inconsistent sea level changes in the recent years. Particularly, the stations installed at Pohang and Jeju has showed rapid sea level rises, since 2013. In this study, we aim to measure the surface displacements on the coastal areas of southeastern coast of the Korean peninsula and to assess the impact of the local land subsidence on the sea level rise estimation by the time-series sequential-SBAS approach.
Authors: Palanisamy Vadivel, Suresh Krishnan (1); Kim, Duk-jin (1); Jung, Jungkyo (2); Cho, Yang-Ki (1)Analysis of Mining-Induced Terrain Deformation Phenomena Using a Direct Integration Persistent Scatterer Interferometry Approach Riccardo Palamàa*, Michele Crosettoa, Oriol Monserrata, Bruno Crippab, Marek Mrózc, Magdalena Mleczkoc, Natalia Ostrowskac, a Centre Tecnològic de Telecomunicacions de Catalunya (CTTC/CERCA), Division of Geomatics, Castelldefels, Barcelona, Spain b Department of Earth Sciences, Section of Geophysics, University of Milan, Milan, Italy c Institute of Geodesy, University of Warmia and Mazury in Olsztyn, Poland *corresponding author: riccardo.palama@cttc.cat Underground mining activity often triggers substantial ground surface displacements both in vertical, i.e. subsidences, and horizontal directions. This work is devoted to illustrate a method that employs satellite radar interferometry to extract the vertical and horizontal components of the ground surface deformations along the radar line-of-sight (LOS). Interferometric Synthetic Aperture Radar (InSAR) has been employed to study terrain deformation for decades and has seen relevant developments in terms of accuracy and coverage with the introduction of techniques based on persistent scatterers (Persistent Scatterer Interferometry, PSI). The availability of SAR datasets with increasing spatial and temporal coverage, with decreasing temporal intervals between two subsequent acquisitions, such as the ones collected by the Copernicus Sentinel-1 constellation, gives the opportunity to analyse and monitor ground-surface deformation phenomena, of natural origin (such as landslides) or man-induced. In this work, we employ a satellite SAR dataset collected by the Sentinel-1 A&B sensors, using the interferometric wide swath (IW) acquisition mode and both descending and ascending pass. Vertically polarized transmit-receive (VV) data were analysed, as they are expected to provide a higher PS coverage with respect to other polarisations. The area of interest (AOI) is located around the town of Polkowice, south-western Poland, and is delimited by a rectangle whose North-West and South-East vertices have coordinates (51° 30' 25.49801" N, 16° 1' 7.62359"E) and (51° 28' 51.15166" N, 16° 8' 23.57851" E), respectively. The presence of copper mines within the AOI causes ground-surface deformations, with both a vertical component (subsidence) and horizontal component, detectable in the East-West direction. The main processing steps leading to the formation of the PS grid consist of (i) image coregistration; (ii) interferogram generation, achieved using a direct integration approach, i.e. each interferogram is generated from two subsequent images; (iii) multilooking, using a window whose azimuth and range lengths are 2 and 10, respectively, yielding a pixel size of about 28 m in azimuth and 23 m in range; (iv) PS selection, based on a coherence threshold of 0.4; (v) phase unwrapping; (vi) time series generation; (vii) removal of the Atmospheric Phase Screen, which is previously estimated using a combination of spatial low-pass and temporal high-pass filters; (viii) geocoding. This procedure was applied to both Ascending and Descending data, thus enabling the extraction of time series of the East-West (EW) and Up-Down (UD) displacements. The analysed datasets were collected by the Sentinel-1 constellation in two different blocks, the first one from September 19, 2019 to February 5, 2020 and the second block from February 5 to June 28, 2020, with an interval of six days between two subsequent acquisitions.. Within the AOI, several trihedral corner reflectors were located in order to increase the number of detected PSs with very high coherence. The results show the presence of three main areas that are affected by ground surface deformation. The UD component of the linear velocity reaches maximum values of about 40-50 mm/year, common to all the affected areas, whereas the EW horizontal component is maximum for the area at the south of Polkowice (about 30 mm/year eastward) and almost null for the other areas. In addition to the geocoded time series, a quality-index map was calculated as the standard deviation of the residuals of the velocity estimation, which was previously extracted as the linear fit that best approximates the deformation time series. Such quality index usually presents high values if the phase noise of the PS point is high or if the deformation is non-linear. References: R. Bamler,P. Hartl, Synthetic aperture radar interferometry, Inverse Problems, 14, R1-R54, 1998 D. Massonnet, K.L. Feigl, Radar interferometry and its application to changes in the Earth's surface, Reviews of Geophysics, 36(4), 441-500, 1998. A. Ferretti, C. Prati, F. Rocca, Permanent scatterers in SAR interferometry, IEEE Transactions on Geoscience and Remote Sensing, 39(1), pp. 8-20, 2001. F. van Leijen, Persistent Scatterer Interferometry based on geodetic estimation theory, PhD Thesis, Delft University of Technology, 2014. Crosetto, M., Monserrat, O., Cuevas-González, M., Devanthéry, N., & Crippa, B. (2016). Persistent Scatterer Interferometry: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 115, 78–89. Presentation preference: poster
Authors: Palamà, Riccardo (1); Crosetto, Michele (1); Monserrat, Oriol (1); Crippa, Bruno (2); Mróz, Marek (3); Mleczko, Magdalena (3); Ostrowska, Natalia (3)We employ multiple synthetic aperture radar (SAR) satellites including ALOS, COSMOS-SkyMed (CSK) and Sentinel-1 to derive the ground deformation of the Hanoi metropolitan in the period of 2007-2018. The three datasets have separate acquisition specifications, so we adopt different time-series analysis approaches accordingly. For ALOS and CSK images, combination of single-master persistent scatter (PS) approach and multi-master distributed scatter (DS) approach yields high coherence pixels and improves spatial coverage. For Sentinel-1 images, we implement an adaptive pairing strategy in the small baseline subset (SBAS) approach to deal with the issue of fading signals and to maximize the spatial coverage. In post processing, we remove the contribution of horizontal components from the line-of-site (LOS) displacement time-series by using horizontal velocity fields interpolated from 6 GPS stations, and estimate the vertical displacement thereof. The resultant vertical deformation rates agree with leveling measurements at R2 ~0.8 and ~0.9 for ALOS and CSK respectively. Sentinel-1 time-series agrees with the continuous GPS records in the long-term trend, with GPS showing higher seasonal variations. We also analyze 22 years (1997-2018) of groundwater levels and compute the pumping rates thereby in conjunction with Global Precipitation Measurement (GPM) data. In central Hanoi, the decreasing rate of the main Pleistocene hydraulic level changes from 0.5 m/yr in year 1997-2006 to nearly zero in year 2007-2018. The ground displacement time-series, however, still reveals ongoing subsidence at 1-2 cm/yr during the latter period. This subsidence magnitude is similar to the compaction of layer between the surface and 44m depth where begins of Pleistocene aquifer. Hence, this remnant subsidence may be the result of hydrodynamic lag due to delayed compression in fine-grained layers. One-dimensional aquitard compaction modeling suggests another 70 years before the land subsidence completes. In suburban Hanoi, the decrease of Pleistocene hydraulic level exacerbates during the last 3.5 years (2015-2018), leading to significant land subsidence at 3-4 cm/yr in the same period. Pumping rate estimates also reveal transition of pumping center from central Hanoi to the new urban area after 2014. The subsidence patches in Ha Dong (3 cm/yr) and Hoai Duc (>4 cm/yr) may require further attention for future urban planning. In Hoang Mai, the sharp discontinuity of subsidence rate and borehole records suggest a potential hydrogeological boundary with significantly higher rapid subsidence rate to the south of the boundary during 2007-2011. In addition, compaction modeling shows a remarkable difference in hydrogeological parameters for the compacting layers between central Hanoi and the surrounding districts. All these findings should facilitate better water resource management in the Hanoi metropolitan.
Authors: Nguyen, Minh (1,2); Lin, Yunung Nina (2); Tran, Quoc Cuong (3); Chan, Yu Chang (2); Tseng, Kuo Hsin (4); Chang, Chung Pai (4)The European Plate Observing System (EPOS) integrates existing and newly created multidisciplinary infrastructures and products in the field of Earth sciences in Europe. One of the major goals of the EPOS project is to develop an integrated service of research infrastructure at the local and national levels. In this paper, the authors demonstrate the results of deformation research carried out for the integration of GNSS and InSAR data realized as part of the EPOS project. In Poland, the most exposed region on the effects of deformation is the area of Upper Silesia Coal Basin (USCB) in the south of the country. Polish coal deposit is one of the largest in Europe. Hence, the exploitation of resources has been carried out there for the last 200 years. The mining works cause continuous subsidence of the most populated area in Poland, therefore the long-term deformation research on this area is necessary. To conduct GNSS and InSAR research on the USCB area, eight high-frequency GNSS receivers and six InSAR persistent scatterers have been purchased in the EPOS project. With regarding the new GNSS stations, the multi-GNSS post-processing service has been created. The multi-GNSS service is a Double-Difference network solution, based on GPS, Glonass and Galileo daily observations, CODE Final products with a reference to IGS and EPN stations. The daily coordinates estimation allows detecting the long-term vertical and horizontal displacements of the surface. The InSAR computations were conducted using PSInSAR and DInSAR techniques in a 6-day exact repeat cycle. In the research, the Sentinel-1 satellite images have been used in ascending and descending orbits. In general, the InSAR data provide larger spatial coverage in comparison to a single point of GNSS station, however only in the line of sight direction (LOS). To be able to compare InSAR and GNSS data, the topocentric GNSS positions of the station have been converted to LOS direction. In this paper, we depict the results obtained in the research of multi-GNSS and InSAR deformations in the USCB area. The main goal of the paper is to determine the correlation between GNSS and InSAR deformation results. The determined time series of integrated use of GNSS and InSAR measurements enables an extension of the concept of safety monitoring in spatio-temporal mode.
Authors: Tondaś, Damian; Pawluszek-Filipiak, Kamila; Ilieva, Maya; Rohm, Witold; Kapłon, JanMonitoring land subsidence due to groundwater exploitation, natural oil and gas extraction, improper building foundations close to coastal areas, and tectonic movements are crucial in understanding the behavior of Earth’s surface in urban areas. In the context of urban monitoring, uncontrolled constructions and irregular development activities are currently major factors causing land subsidence. Nowadays, several coastal cities, due to subsidence, are prone to seawater ingression, risk of sinkhole formation, and affected by frequent flooding during high tide and heavy rainfall. The city of Mumbai (earlier known as Bombay) in India comprises of seven main islands, which are merged to form one large island by several land reclamation projects in the 18th century. As per 2011 Census of India report, Mumbai is India’s densely populous city with 12.44 million population. The city is consisting of deep natural harbor, a major seaport on the Arabian Sea, three main hill ranges, six major lakes and numerous creeks. Except for the hill ranges, most of the city falls close to or below the mean sea level making it more vulnerable to submergence during geomorphological changes such as high tide and heavy rainfall. The soil cover in the city is predominantly sandy due to its proximity to the Arabian Sea. Alluvial and loamy soils are also present in the suburban areas. Mumbai lies in an active seismic zone (Seismic Zone 3) with several faults across the city and outskirts. Therefore, land subsidence in Mumbai, in any form, due to geomorphological changes, tectonic or anthropogenic activities has to be studied continuously to understand the behavior of Earth’s surface. The results from these studies will help to take necessary or preventive measures before the occurrence of any unknown event. In this work, we use spaceborne remote sensing satellite images to map and monitor land subsidence in Mumbai city from 2016 to 2019. Interferometric Synthetic Aperture Radar (InSAR) is an efficient tool monitoring ground surface displacement with millimeter level accuracy. The technique is widely applied to monitor land subsidence, tectonic deformations, volcanic eruptions, and glacier movements. The Persistent Scatterer Interferometry (PSI) technique utilizes multiple SAR images to estimate deformation over a region. The process involves analyzing stable phase pixels known as persistent scatterer (PS) points from differential interferograms formed with respect to a single master. In this study, we use 79 images obtained on descending track of Sentinel-1A from 2016 to 2019 analyzing the subsidence of Mumbai city using the PSI technique. An image acquired on November 16, 2017 is used as the master image for coregistration and interferogram formation. Approximately 85 sq.km. urban area covering two subswaths (IW1, IW2) of the Sentinel-1 interferometric wide swath mode data are used for the displacement analysis. We processed and compared the PSI technique implemented in SARscape and Wavelet based InSAR (WabInSAR) using Gamma software for the displacement analysis over Mumbai city. The PSI approach is suitable for urban areas where a large number of human made targets such as buildings, bridges, electric poles and monuments are extensively available. Amplitude dispersion and coherence are used as a PS point selection criterion in SARscape software. In SARscpe, a linear displacement model is used in which the deformation rate and the height of the PS points are estimated by removing the unwanted phase components. The number of images (79) and the regular interval between the image acquisitions are beneficial in reliable coherence estimation of the PS points. A single coherent target usually dominates point targets, which have low temporal variability of the backscatter. In the WabInSAR technique, a wavelet-based analysis is used to examine the interferometric phase noise and identify high coherent pixels. Wavelet-based filters are applied to reduce the effect of topographically correlated atmospheric phase delay and spatially uncorrelated DEM error correction (Shirzaei, 2013). Using a reweighted least squares estimation, the set of interferograms are inverted to generate deformation time-series and velocities along the line-of-sight (LOS) direction. The PSI analysis using SARscape revealed two major subsidence areas (Wadala west and Virar) and an area (Adaigoan) showing significant uplift in Mumbai city. The Wadala west area, which is close to the mangrove plantations in the west shore of the Thane creek, indicate subsidence of 80 mm from 2016 to 2019. We also observed some extent of an expressway (Sion-Panvel expressway) leading to a bridge over the Thane creek exhibiting subsidence. The extent of the expressway undergoing subsidence lies close to the mangrove plantation zone surrounded by saltpans along the road. A densely populated and irregularly constructed slum area along the Ghatkopar-Mankhurd Link Road merging with the Sion-Panvel expressway also show subsidence in the range of 60 mm over the same period. On the other hand, the Virar subsidence region comprises of a planned built-up area indicating subsidence of approximately 40 mm in three years. Uplift up to 80 mm is observed in the Adaigoan area from 2016 to 2019. The reason for the observed uplift is to be analyzed in detail. Most of these areas indicate a linear displacement trend. A few areas away from the main city and the remaining study area indicate minor subsidence and uplift values between -25 mm to + 25 mm. The results from WabInSAR analysis of Mumbai city using Gamma software show similar displacement trends. A few points marked in some areas along the west coast of the city indicate uplift up to 30 mm, whereas points on the eastern shore (near the Thane creek) indicate subsidence up to 40 mm from 2016 to 2019. The major displacement areas are similar in both the results. We observed a new region (Vasai), close to the Virar subsidence area indicating subsidence in the order of 30 mm/year. The regions around the Thane creek are showing subsidence of 0 to 10 mm/year. In some regions, there are few variations in the result from the software due to the difference in initial PS point selection criteria, different atmospheric correction window sizes and the deformation rate estimation approach. These differences are to be studied in detail in future work. Overall, the results highlight the presence of subsidence and uplift zones in Mumbai and its surroundings. The correlation between groundwater extractions, proximity to the mangrove plantations, coastline, and soil type are to be studied in detail to understand the driving mechanisms behind the observed displacement.
Authors: Vaka, Divya Sekhar (1); Rao, Y. S. (1); Ojha, Chandrakanta (2); Kumar, Vineet (3)The Salmas Plain, with an area of about 445 km2, is located in West Azerbaijan province of Iran and 20 km west of Urmia Lake. The Salmas aquifer is one of the coastal aquifers flows into the Urmia Lake. The over-extraction from groundwater for agriculture and domestic use along with the recent draught occurrences has led to sever groundwater level decline and shrinkage of the lake. Owing to severe aquifer depletion, the subsidence of the ground surface has occurred over a broad areas worldwide. Interferometric Synthetic Aperture Radar (InSAR), a remote sensing technique for measuring surface deformation, can serve as a tool to measure the spatiotemporal land deformation across aquifers. Here, we first identified the mean velocity map from time series analysis of Sentinel-1 images, and then decomposed it into both vertical (inelastic subsidence rate) and lateral components of land deformation. Subsequently, using a joint well-InSAR data, we (1) estimated aquifer storage parameters, (2) reconstructed the hydraulic head, and (3) determined the total, recoverable, and irreversible components of groundwater storage (GWS) variations in the Salmas confined aquifer. In this work, the mean velocity map is obtained from InSAR time series analysis of Sentinel-1 SAR data archives from March 2015 to February 2018. First, we produced 135 and 130 differential interferograms for the descending and ascending images, respectively. Temporal and spatial baselines of interferograms are less than 150 days and 130 meters, respectively. A 30-m digital elevation model (DEM) is used as a reference topography model for the topography-related phase corrections and geocoding of all the interferograms. SNAPHU and precise orbital coordinates were used to correct flattern phase component of the interferograms (Chen and Zebker 2001). Finally, we applied a pixel-based InSAR SBAS time-series analysis on the corrected and filtered interferograms to estimate the mean velocity map in both radar and geography coordinate systems. Given the InSAR-derived land deformations across the Salmas Plain, we used some equations to relate the deformations to the hydraulic head in the confined aquifer to estimate elastic skeletal storativity, reconstruct head, and determine the GWS variations. The mean velocity map shows a significant subsidence rate ~60±3 mm per year in the plain. Decompose of Line of sight (LOS) deformation from ascending and descending map presents 80 mm/year and 3.5 mm/year in vertical and lateral deformation, respectively. Given the high correlation between the deformation and groundwater level indicates that the variations in the aquifer water storage is the main factor controlling subsidence across the Salmas Plain. We estimated skeletal elastic storativity, time delay between the head change and land deformation at 15 piezometer locations, through a joint InSAR-well data analysis. The average skeletal elastic storativity is calculated to be ~0.035 (ranges from 0.0387 to 0.144). The maximum time delay is estimated to be 150 days that happened in middle to east part of plain with thicker fined-grained units. InSAR-derived vertical deformations from the piezometer locations were transformed to head changes with reasonable precision. Also, we identified spatiotemporal variations in the total GWS, recoverable GWS, and irreversible GWS. Interestingly, the average total GWS variation of the aquifer system is estimated to be ~18 million cubic meters over the recent last 3 years because of overexploitation from groundwater. Also, it is founded that the irreversible and recoverable GWSs are account for 30% and 70% of the total annual GWS depletion occurred across the Salmas Plain. The findings show that most of the groundwater pumping from the confined aquifer comes from dewatering of compacted aquitards. This study indicates that quantification of irreversible component of GWS in those aquifers of significant subsidence is essential for sustainable water resources management.
Authors: Shahbazi, Saeedeh; Mousavi, Zahra; Rezaei, Abolfazl; Gholami, Roozbeh; Mohamadnia, MohamadhosseinMulti-baseline Synthetic Aperture Radar (SAR) interferometry technique utilizes multi-pass acquisitions in order to measure and monitor the displacements of the Earth’s surface over time. In the framework of multi-temporal data acquisitions, multi-look multi-baseline SAR interferometry (MLMB InSAR) exploits amplitude and phase of the received data and performs multi-looking through the estimation of covariance matrix in order to increase the interferometric phase quality. Unlike to the conventional interferometric SAR (InSAR) techniques that may show some deficiency to cope with distributed coherent scatterers, the MLMB InSAR framework is able to deal with distributed targets with moderate coherence. One important issue in MLMB InSAR framework is the proper estimation of covariance matrix. The biased estimation is crucial for satisfactory results achievements, while improper estimation of covariance matrix can ruins image structures and confuse the targets contributions specifically in non-homogeneous region of urban images. Additionally, discovering optimal window size and number of look is another issues may affect the estimation of covariance matrix and consequently the displacement results of MLMB InSAR technique. This paper aims at assessing the role of covariance matrix in MLMB SAR interferometry. Most recently, non-local estimation approach of covariance matrix estimation such as NLSAR has shown great performance in terms of resolution preserving and biased estimation of covariance matrix. In this paper, the efficiency of NLSAR based covariance matrix estimation is compared with the traditional boxcar approach for height and deformation mapping using MLMB InSAR framework. The procedure implemented in maximum likelihood approach with ability to reduce variations occur due to the noise of data and estimate maximum likelihood of height and displacement simultaneously. Experimental results using Sentinel-1 Images over Tehran, has shown that the displacement is more accurately estimated by NLSAR based framework. In addition, the outcomes of study showed that the sensitivity of MLMB InSAR framework to the displacement estimation using NLSAR is less than the traditional boxcar approach. The experiments with aim of height and deformation mapping over north-west of Tehran, Iran, are shown. The pros and cons of NLSAR as a non-local processing by considering similarities of pixel’s statistical distributions, over Boxcar as local technique which does not notice the homogeneity of area, is presented by giving their results and their performances. The analyses cover 4D processing of study area with multi-baseline stack of 15 C-band SAR images in VV polarization channel and acquired by the Sentinel-1A spaceborne radar system from February 18, 2017 to January 15, 2019. Moreover, for evaluation of results obtained from multi-baseline InSAR processing, the measurements of a GPS station have been considered and weighted parametric least square adjustment and A-posteriori variance factor test applied to GPS time series measurements.
Authors: Zamani, RoghayehOn 8th August 2017, a magnitude Mw 6.5 earthquake occurred in the County of Jiuzhaigou in Sichuan Province, China (USGS, https://earthquake.usgs.gov/). Because of its high magnitude and shallow epicenter, the earthquake caused grave casualties and property losses. Furthermore, the earthquake triggered numerous secondary mountain disasters such as rockfall and landslide. Shortly after the Jiuzhaigou earthquake, some studies had assessed the number, distribution and other characteristics of the coseismic landslides based on pre- and post-event remote sensing data. However, slow-moving post-earthquake landslides don’t show visible change at short notice and are often neglected by the common change detection method. These slow-moving landslides will pose a long-term potential threat to people’s life and property. Therefore, a detailed monitoring of the slow-moving landslides is crucial to post-earthquake recovery and reconstruction. Synthetic aperture radar interferometry (InSAR) technology, which has large ground coverage and high spatial resolution, has great advantages for geological disaster observation under all weather conditions. It has been utilized in ground surface deformation measuring of landslide for disaster detection and assessment. In this research, we adopt an optimized strategy that combines D-InSAR technique with SBAS InSAR to accurately detect and monitor the slow-moving post-earthquake landslides in Jiuzhaigou area. Firstly, we carry out a quick detection across wide area using differential InSAR (DInSAR) technique with 6 ALOS‐2 PALSAR‐2 ascending images. To retrieve the temporal evolution of these landslides, a detailed monitoring of specific landslides is carried out using the short baseline subset InSAR (SBAS-InSAR) during the period from 2007 to 2019. Combining with other multi-source data (including field investigation, LiDAR and optical image), we perform an in-depth analysis of the impact of the 2017 Jiuzhaigou earthquake on slow-moving post-earthquake landslides. The results show that there are 16 slow-moving post-earthquake landslides which can be detected by InSAR analyses in Jiuzhaigou area, including 8 landslides close to residential areas. These landslides are mainly distributed in the NE plate (active plate), and the slide directions mainly are east (including 9 landslides) and southeast (include 5 landslides). Secondly, we found that the earthquake’s impact on the slow-moving landslides can mainly be classified into three categories: (a) accelerating the active historical landslides that were already sliding before the earthquake; (b) reactivating the stable historical landslides which were undisturbed before the earthquake; (c) directly triggering the landslides that had stable geological environment before the earthquake. For each case, we demonstrate some case studies of landslide affected by the earthquake, and perform a detailed analysis combining with multi-source data.
Authors: Cai, Jiehua (1); Zhang, Lu (1); Liao, Mingsheng (1); Dong, Jie (2)The launch of the European Copernicus satellite Sentinel-1 has sparked a significant growth in the use of InSAR data. The growth is highly motivated by the continuous acquisition of uniform radar images from 2014 and until 2030 at the earliest. Thanks to the European Commission’s policy on free and open data as well as the six-day repeat cycle over Europe, a unique dataset is made available, which, for the first time, enables deformation monitoring on regional, national and continental-wide scales. The above is exploited by several countries working to establish operational services for nationwide deformation monitoring based on Sentinel-1 imagery. The generated data will be available under a free and open data policy. Norway and Germany launched their services in 2018 and 2019, respectively, while e.g. the Netherlands and Denmark are working towards this goal. Furthermore, the European Commission has started the establishment of a European Ground Motion Service (EGMS); an initiative led by the European Environmental Agency, which will result in an addition to Copernicus Land Monitoring Service portfolio. In Denmark, the work towards a nationwide service is led by the Agency for Data Supply and Efficiency. The motivation is two-fold: As a governmental institution, we are responsible for producing, maintaining, and distributing nationwide data sets, which can serve as input for an improved basis of decision and support the digital growth. Furthermore, as a national mapping agency, another responsibility concerns the development and maintenance of the national geodetic infrastructure. We see several potentials with including Sentinel-1 data in this work and thus have identified a number of investigations to clarify the potentials. The current presentation will elaborate on the motivation. The first part will focus on the steps towards an operational service, including the distribution of nationwide maps in 2018 and 2019 and the efforts to support end-users in the public and private sectors in case-studies related to climate change adaptation, monitoring of infrastructure, etc. The second part will focus on the geodetic aspect and more specifically the efforts to datum reference the Sentinel-1 data using time series from our nationwide network of permanent GNSS stations. Most stations are located in areas with few persistent scatterers, or scatterers whose movements are uncorrelated to that of the GNSS station. As such, one step in the process involves the co-location of the GNSS stations with active or passive reflectors. Active transponders have the advantage of being small and thus to be installed directly on the foundation of a GNSS station. In order to ensure that we have a well-functioning set-up for the datum referencing, we have built a test site including a GNSS station, one corner reflector and three Compact Active Transponders. The first results from the above efforts will be addressed as well.
Authors: Balasis-Levinsen, Joanna; Nissen, Martin; Keller, Kristian; Meister, AslakPeatlands are vital in the global carbon cycle because they store around 30% of global soil carbon. Deforestation over peatlands has been increasing leading to peat drainage. As it is drained, the land depression occurs as the result of biological oxidation and the loss of carbon. Indonesia has the highest coverage of peatlands globally and it also releases huge carbon due to peatland degradation. Conventionally, the peat height change is quantified by using differential GPS or measuring the change in the groundwater level. However, these techniques are spatially limited. Therefore, the use of Synthetic Aperture Radar (SAR) data based Interferometric SAR (InSAR) technique is more efficient. The limitation is the vegetation coverage in tropical peatland can lead to the decorrelation because of the canopy change, but it can be minimized by processing time series InSAR data. In this study, we applied Small BAseline Subset (SBAS) InSAR technique to estimate the peatland subsidence rate using Sentinel-1 data in Bengkalis Island, Indonesia, where deforestation has occurred massively. The estimation was validated using the approach of groundwater level data. Since mostly peatlands in Bengkalis Island have been drained, the whole study area encountered subsidence during the observation period. Our results show the mean subsidence in the study area is -0.968 cm/year with the mean total subsidence (2017-2019) is -3.368 cm. Each land cover type was observed to have different rates. The variability of rate is caused by the difference of land management related to the drainage. The use of C-band SAR Sentinel-1A and B data can provide multiple scenes with 6 days revisit which is the highest among other SAR data, however only 12 days revisit is available during this study conducted. The higher temporal resolution enables the feasibility of peat vertical change monitoring continuously. The results from this study will be useful to support Indonesia’s Program on Peat Restoration, which is committed to restoring degraded peatlands. Therefore, time series InSAR of Sentinel-1 can help investigate the progress of the restoration activities spatially and temporally.
Authors: Umarhadi, Deha (1); Avtar, Ram (1); Tutubalina, Olga (2)Land subsidence associated with groundwater extraction in the city of Beijing, China, has been a problem for decades. Remote sensing has been used extensively in prior studies to monitor subsidence in Beijing. However, given recent changes in precipitation and groundwater management, there is an urgent need to update the subsidence record and to evaluate whether the long-term spatiotemporal patterns of subsidence have changed. This study, therefore, investigates the recent spatiotemporal patterns of land subsidence in Beijing by tomography-based persistent scatterer interferometry SAR (Tomo-PSInSAR) technology, using 39 RadarSAT-2 images from 2012 to 2015 and 33 Sentinel-1 images from 2016 to May-2018, and drawing upon Geographic Information System (GIS) spatial analysis methods. Vertical ground deformation rates in Beijing were found to range from -176.2 to +12.3 mm year-1 from 2012 to 2015 but subsequently decreased to -119 to +8 mm year-1 from 2016 to May 2018. Three spatial scales of subsidence are evident: At the metropolitan regional-scale, the total area of subsidence area is about 1235.2 km2, and comprises four main subsiding regions, located in the northern and eastern parts of the city. More than 85% of the subsiding area is located between the Fifth and Sixth Ring Roads. At a more local scale, eight main subsidence bowls are characterized by different patterns of subsidence. Some of the subsidence bowls are separated by active faults. Time-series data of the displacement shows that the decreasing subsidence rate after 2016 could be due to the 1 m rise in mean groundwater level from the end of 2014 to mid-2018. This change in groundwater level is likely due to an increase in precipitation since 2016, and water transfers, which reached 2.3 × 109 m3 by 2017 from the South-North Water Transfer Project. At the scale of individual infrastructure projects, the Beijing subway, main roads, and the Capital Airport all show severe uneven subsidence, which is a cause for concern. To our knowledge, this research is the first study using satellite SAR remote sensing methods to document the change in the land subsidence rate of Beijing. Starting in 2016, the rate notably declined, suggesting that subsidence mitigation strategies are beginning to have an effect.
Authors: Zhou, ChaodongAbstract: With the exploitation of coal resources, its reserves have dropped sharply, and most of the mines around the world, such as Netherlands, Germany, Belgium, France, the United Kingdom, Japan and so on have been closed or abandoned. Especially in China, due to the coal de-capacity policy implemented in recent years, a number of mines and open pits have followed the same fate. According to statistics, at least 19,106 mines were closed in 2000-2016 in China, and the numbers are growing. After a mine closure, the decommissioning of the mine drainage equipment makes the groundwater level to rise, which may cause serious environmental and geological disasters. The renovation, redevelopment and ecological environment management of closed mines have become one primary issue for economic and social sustainable development. Therefore, it is important to monitor the ground deformation of closed mines and study their laws and mechanisms. Persistent Interferometric (PSI) techniques are a perfect tool for such purpose. After the mine closure, its surface deformation is usually long-term and hidden by new crops and vegetation. As a result, it is difficult to select sufficient persistent scatterers for long-term time-series PSI analysis, which leads to the inability to effectively detect the spatial location and evolution process of surface deformation. To increase the number of highly coherent points, permanent (PS) and distributed scatterers (DS) can be jointly combined in time-series PSI analysis. In this experiment, a new processing chain of StaMPS is provided. The DS preparation consists of the identification of SHP by KStest, selection of DS and the estimation of their optimal phases through phase decomposition, while the PS preparation concentrates on the selection of PS base on the amplitude dispersion index. The selected DS and PS are jointly processed to estimate the deformation rates in StaMPS. A total of 88 ascending Sentinel-1A images covering Xuzhou western coal field from October 4th, 2016 to October 25th, 2019 have been collected and processed. In order to verify the reliability of DSInSAR, the results of PSInSAR and SBAS have been obtained simultaneously by StaMPS/SBAS. From the monitoring results, it can be seen that there are seven deformation bowls in the study area, including three uplift zones, which are located in the middle of the Jiahe Mine, the northeast of the Pangzhuang Mine, and the south of the study area. The surface uplift may have been caused by the rise in groundwater level after the mine closure, which increased the pore pressure and reduced the effective stresses of the overburden, thus causing the surface to uplift. Compared with the results of PSInSAR and SBAS, the deformation area obtained by DSInSAR is more complete than PSInSAR, and its reliability is higher than SBAS results. The number of high coherence points obtained by DSInSAR is 124,193, which are 13.3 and 1.7 times that of PSInSAR and SBAS, respectively. Due to the large area of farmland, seasonal changes in crops have resulted in sparse high coherence points selected by PSI in the farmland. In order to evaluate the accuracy of the results obtained by the three methods, 32 levelling points located in the study area were selected. The highly coherent points within 50m of the levelling point were selected for comparison. Due to the difference in the number and density of high coherence points selected by the three methods, DSInSAR, PSInSAR, and SBAS selected 23, 11, and 21 levelling points for accuracy verification, respectively. The standard deviation of DSInSAR, PSInSAR and SBAS results are 2.40mm 2.79mm and 3.48mm respectively; the root mean square error of DSInSAR, PSInSAR and SBAS results are 3.81mm, 3.84mm and 6.44mm respectively. It can be seen from the accuracy comparison that DSInSAR monitoring results are more accurate than the other two methods. DSInSAR combined with DS and PS improves the number and density of high coherence points, reduces the phase unwrapping error, and improves the accuracy of the deformation. In addition, DSInSAR improves the spatial continuity and integrity of surface deformation and helps to detect the spatial location and evolution of surface deformation after the mine closure. Therefore, the combination of DS and PS to obtain the surface deformation of closed mines can effectively overcome the difficulty of insufficient high coherence points caused by long-term deformation monitoring, and provide data support for further research on the laws and mechanisms of surface deformation of closed mines. In the next stage, we will still use Sentinel-1A data to obtain the deformation of the Xuzhou western coal field based on PSI time series combining DS and PS.
Authors: Zheng, Meinan (1); Deng, Kazhong (1); Mallorqui, Jordi J. (2)At present, the application of time series InSAR surface subsidence monitoring technology is mainly concentrated in urban areas. For mining areas, especially in Open-pit mine, there are usually more vegetation coverage, low coherence, and large deformation gradient, which makes it difficult for time series InSAR technology to obtain high-density mining area surface deformation information. In fact, it is difficult for traditional linear deformation observation to fully reflect the temporal and spatial evolution of the three-dimensional deformation of the whole mining area. How to obtain high-precision and high-density deformation information is the main difficulty of InSAR monitoring in mining area. Making full use of the ground-based geosensor network monitoring system to carry out satellite-ground collaborative InSAR monitoring computing can realize early identification and discovery of problems in large-scale macro monitoring, and accurate monitoring of local areas in real-time warning. On the other hand, in addition to the vertical deformation, the mining area is usually accompanied by more obvious deformation in the horizontal direction; while the conventional InSAR can only obtain the deformation of the radar line of sight. How to jointly use multi-platform and multi-angle SAR images to obtain high-precision three-dimensional deformation information is an important technology for ground deformation monitoring in mining areas. Research on collaborative observation methods that combine multiple vertical observations such as Geosensor Network, InSAR, LiDAR, and inversion techniques where multiple remote sensing observation results are mutually additional information or constraint conditions, and build to the point, line and surface integrated monitoring methods for the ground disaster of open-pit joint mining area. From point-line collaboration, point-surface collaboration, and surface-surface collaboration, a satellite-ground coordinated data fusion calculation system is constructed, and GNSS, georobots, ground laser scanning and satellite remote sensing data can be used to achieve loose combination and tight combination level fusion calculations. (1) Collaborating satellite InSAR of large-extends surface subsidence monitoring with geosensor network monitoring system, such as ground-based SAR of high dynamic local area monitoring and GNSS precise monitoring of specific surface feature points, satellite-ground collaborative monitoring are constructed and the effectiveness of monitoring are evaluated. (2) Collaborating geosensor network with satellite remote sensing, and the fusing computation of the observation results, additional information or constraints relationship between the geosensor network and satellite remote sensing, the integrated monitoring and analysing system is built for mining geohazards monitoring. (3) By fusing high-resolution SAR image and high-resolution optical image, the location, size, boundary and other characteristics of the mining area geohazards feature object are determined, and the typical mining area geohazards inventory map is established. (4) Taking the horizontal deformation rate calculated by the geosensor network as a constraint, the slant range deformation rate obtained by InSAR is decomposed into vertical deformation rate and horizontal deformation rate enhanced by the joint computing of satellite-ground collaborative monitoring. (5) The early identification knowledge database for abnormal data and real danger are established, and early warning services for ground disaster monitoring are implemented based on GIS services. The study area is Pingshuo mining area which located in the northern part of Shanxi Province, China. The mining methods of the Pingshuo mining area include jointly open-pit and well work. The deformation-related areas include dumps, industrial squares, railway lines, power supply towers, etc. The disasters involved include cracks, subsidence, collapse, landslides, etc. The range of deformation is large and the situation is complicated. The geosensor network system consisting of GeoRobot, GNSS and ground-based SAR and environment sensors have been deployed which can automatically monitor and analyse surface deformation in local and special area. The high-resolution SAR data of surface deformation monitoring in Pingshuo mining area were processed and analyzed, and the displacement and settlement maps of the monitored area were obtained. The processing results of the InSAR and the GNSS were compared and analyzed. Acknowledgments： This work was supported by National Key Research and Development Program of China (China, 2018YFB0505402) and National Natural Science Foundation of China (41771443).
Authors: Zhang, JinThis study aims to design and demonstrate a method to concatenate Radarsat-2 satellite datasets with operational modes and map subsidence dynamics. Studies in the literature have shown the potential of land subsidence monitoring using Radarsat-2 [1,2,3]. Yet, integrating Radarsat-2 SAR data with two different operational modes, i.e., Standard Mode (SM) and Extra-Fine Mode (XFM) has been scarcely discussed. We consider a Radarsat-2 dataset collected between 18/10/2010 and 03/11/2016. For this dataset, SAR data were acquired in SM. The spatial resolution is 13.5m and 7m in range and azimuth direction, respectively. Between 27/11/2016 and 16/12/2017, SM was permanently substituted by XFM with a 3.1m × 4.6m (range × azimuth) resolution. The operational model switch makes the straightforward integration of SAR data with SM and XFM pragmatically unfeasible. Therefore, we propose a pragmatic approach to 1) identify and pair coherent points (tie-points) that exist in both SM and XFM data stacks; 2) concatenate the associated time series of every tie point pair. For this purpose, we design a 3-step processing procedure: The first step generates Persistent Scatterers (PS) and the associated deformation time series by standard PSI approach for SM and XFM datasets, respectively. Note that here we also use lidar-derived DEM and DSM to correct PS geolocation and screen out plausible PS. The second step integrates data in space by tie-point pairs. Each tie-point pair is composed of two PS which are separately generated from the SM and XFM Radarsat-2 datasets, but represent a common ground target. Considering systematic geolocation bias due to distinct reference points between the two datasets and geolocation errors, we detect tie-points by the following process: i) uniform the reference point. We do so by normalizing all PS amplitude time series and then select a new reference point with the minimum Amplitude Dispersion Index (ADI) value. All PS geolocation and time series reference points are aligned to this new reference point. ii) estimate geolocation quality per PS. We use a geolocation error ellipsoid to describe the geolocation quality in range, azimuth, and cross-range direction [4]. iii) select tie-point pairs. We do so by calculating the intersection areas of any two PS points’ geolocation error ellipsoid. We use a Monte Carlo method to estimate the intersection area. Two points which have maximum intersection areas are identified as a tie-point pair. The third step integrates data in time by concatenating the deformation time series of each tie-point pair. For this purpose, We use a multiple hypothesis testing (MHT) method [5] to model the (long-term) deformation time series and identify deformation anomalies. To demonstrate the applicability of this approach, we use 84 Standard operational mode and 17 Extra-Fine mode SAR data, both in descending orbital mode, to monitor the subsidence over Rotterdam and Delft, in the Netherlands. We identified 18128 tie-point pairs, 5 intersection types of error ellipsoids, 5 deformation models, and constructed their long-term deformation time series. The specific procedures and results are displayed in [6]. We conclude that our method can remove the limitation of different modes Radarsat-2 data to conduct data integration and thereby monitor long-term surface deformation. Reference [1] HOWELL, Stephen EL, et al. Estimating melt onset over Arctic sea ice from time series multi-sensor Sentinel-1 and RADARSAT-2 backscatter. Remote Sensing of Environment, 2019, 229: 48-59. [2] SHARMA, Amit Kumar, et al. Evaluation of Radarsat-2 quad-pol SAR time-series images for monitoring groundwater irrigation. International Journal of Digital Earth, 2019, 12.10: 1177-1197. [3] SAMSONOV, Sergey. Three-dimensional deformation time series of glacier motion from multiple-aperture DInSAR observation. Journal of Geodesy, 2019, 93.12: 2651-2660. [4] Prabu Dheenatheyalan, David Small, Adrian Schubert, and Ramon F. Hanssen. High-precision positioning of radar scatterers. Journal of Geodesy, 2016, 90.5: 403-422. [5] Ling Chang and Ramon F. Hanssen. A probabilistic approach for InSAR time-series postprocessing. IEEE Transactions on Geoscience and Remote Sensing, 2015, 54.1: 421-430. [6] Bin Zhang, Ling Chang, and Alfred Stein. Spatio-temporal linking for medium and high resolution SAR satellite data. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, under review.
Authors: Zhang, Bin; Chang, Ling; Stein, AlfredLand subsidence defined as the gradual vertical movement of the earth's surface caused by the movement of the earth's subsurface material. Land subsidence problems can be caused by natural and anthropogenic problems or a combination of both. Land subsidence is a smooth process, affecting large areas and progressing at a slow rate. Many people do not feel this happening until the impact is felt, such as increased frequency of flooding, high sea water disturbance, cracks and damage to buildings and infrastructure. Land subsidence also occurred on the island of Java, which is also the desest island in Indonesia, where 60% of the population lives on the island of Java. In the past 50 years, industrialization and urbanization have transformed the agricultural and rural areas of Java into a mixture of rural-urban to mega-urban areas. Some strategic cities such as the capital cities of Jakarta, Semarang, Pekalongan, and Surabaya and one of the important infrastructures that connect the western and eastern parts of the island of Java are also on the northern side of the island of Java. The estimated economic costs of the impact of land subsidence are of course, very large. RADAR / SAR satellite images do have the ability to detect and monitor soil deformations with centimeter to millimeter accuracy over large, detailed areas and can reach remote areas that are difficult to reach by direct measurement. One application that can use to conduct land subsidence analysis is the Geohazard Exploitation Platform (GEP), a platform with data from various combined sources for accessing, processing and publishing satellite data for earth observation and its derived products. The use of GEP analysis is proven to be able to produce land subsidence information quickly with low resource use. From this monitoring, it is known quantitatively that from 2015-2019 land subsidence in Jakarta was 1-10 cm/year, but in certain areas, it could reach 25-28 cm/year. In the city of Semarang, it is known that in 2015-2019, the average land subsidence rate is 1-10 cm/year. This value must still be tested with data from field measurements, to obtain the accuracy of the information generated.
Authors: Dipo, Yudhatama; Mohammad, Ardha; Rya, Mukhoriyah; Esthi, Kurnia Dewi; Samsul, ArifinThe process of coastal erosion can be defined as the retreat of beaches, dunes and cliffs, including the habitats of sand, earth and rocks that have formed along the edges of these. On the beaches, this displacement is an unwanted movement of sediments, which frequently modifies the coastline, said process affects a variety of ecosystems, infrastructure and the lifestyle of the local population. In Costa Rica, specifically in the South Caribbean, in areas such as Cieneguita, the mouth of the Estero Negro river, the Bananito river and in the Cahuita National Park, an accelerated erosion process has been reported when compared to other nearby areas. Due to the above, a research project was developed at Universidad Nacional through which it was possible to identify the sites that present the largest receding in the coastline, besides the ones previously mentioned, Westfalia, Puerto Vargas and Refugio Nacional de Vida Silvestre Gandoca – Manzanillo. From a socioeconomic point of view, the South Caribbean region is at a rather vulnerable condition to coastal risks, such as coastal flooding, beach erosion, the retreat of the coastal platform and cliffs, and the increase of waves and extreme winds, among other aspects that have an impact on their local dynamics as well as on the socioeconomic status of the inhabitants. In this context of vulnerability, in the year 2020 a research project was started seeking to assess the relationship between the local geodynamic and the coastline erosion processes and diagnose the governance conditions amid the challenges that these processes pose to the adaptation of the exposed population. In pursue of the previously mentioned objective several recollection and information generation techniques are being used: Using SLC images from the Sentinel 1 mission, the method Persistent Scatter InSAR (PSI) is being applied for the study of the regional tectonic and the data processing from GNSS stations located in the site of study. For the local study, a measure of topographic profiles in the points of interest is being performed, which provides detailed information of the local changes occurring the shoreline. This method is complemented with the sand granulometry analysis. The historical data of the tide gauges in Limón, Costa Rica and Cristóbal, Panama are also being analysed in order to have information about the change in sea level in these locations. With this data, the analysis of how these phenomena and the variables are contributing to the erosive and identified process can be carried out and modifications can be made in order to generate adaptation actions.
Authors: Valverde, Jose; Paniagua, Diana; Barrantes, GustavoTianjin is one of the large urban regions in China that has been suffering from severe land subsidence induced by excessive groundwater withdrawal for many decades as many other economic centers worldwide. Thus Tianjin (China) becomes such a sinking city and has been faced with enhanced flood and infrastructure destruction risks. However, continuously monitoring urban subsidence cross a large area has always been a challenge for both research community and urban administration. This article applied a method of combining the public-available Sentinel-1A/1B (Track149 and 69) Synthetic Aperture Radar and GPS datasets for precisely mapping urban subsidence. Numerous Interferometric Synthetic Aperture Radar (InSAR) techniques have been developed to derive land-surface-deformation from SAR images at a regular temporal interval of dozens of days, which provided enough large dataset for subsidence research. However, the accuracy of InSAR is degraded by unmodeled atmospheric phase delay and other errors involved in interferometric processing. In this study, we used SAR images from Sentinel-1A and Sentinel-1B from October 2014 to December 2019, along with the Stacking approach. We initially extracted the surface deformation information from 110 Sentinel-1A and 93 Sentinel-1B TOPS (Terrain Observation by Progressive Scans) images using GMTSAR software package. We used data from GPS displacements to calibrate the initial InSAR results to achieve a final InSAR velocity field. The calibration by removing an atmospheric plane and GPS deleted systematic errors in the InSAR measurements due to orbit errors, long-wavelength atmospheric effects, and other sources. Our results mapped the spatial and temporal land deformation spanning 2014 to 2019. We also reported subsidence extracted from extensometer and compared it with the InSAR results. The results show that ongoing subsidence in northern and central part of Tianjin is minor (< 1cm/year), while rapid subsidence (3~14cm/year) is ongoing in western districts, particularly in urban areas of Wuqing, Beichen, and Xiqing districts. Moderate subsidence (2~4 cm/year) is ongoing in southern disticts (Jinghai District and Binhai New District) and adjacent regions in Hebei Province. Continuous GPS recorded steady subsidence of 4.7 cm/year in Wuqin district (TJWQ), and 2 cm/year in Binhai New District since 2010 (TJBH). GPS measurements also indicate that subsidence in Cangzhou (HECX) accelerated since 2016 from previous 2.5 cm/year (2010-2015) to current 3.7 cm/year (2016-2019). At last, the deformation velocity map across Tianjin and time series of selected points were compared with groundwater data to illustrate their relationship. And Principal component analysis (PCA) is conducted to figure out the main reason resulting in the subsidence. This study provides mm-level accuracy long-term subsidence rate in Tianjin area and the first-hand information for local governments to evaluate and avoid subsidence risks so that the official can make more effective plans for future urban development.
Authors: Yu, Xiao; Wang, Guoquan; Liu, YuhaoLandslides are one of the most common and dangerous threats in the world that generate considerable damage and economic losses. Landslides identification and monitoring are two significant research aspects for landslides analysis. Also, landslide mode deduction is key for the prevention of future hazards. Nowadays, an efficient landslide monitoring tool is the Differential Synthetic Aperture Radar Interferometry (DInSAR). Persistent Scatter Interferometry (PSI) is the advanced InSAR technique that has significantly improved upon traditional InSAR methods by increasing the accuracy of results (millimeter-scale precision). The main objective of this project was to detect and monitoring of Ahar Chai sub-basin landslides based on the PSI technique. The data used include 22 ENVISAT-ASAR satellite images for the period 2003 to 2010 in the descending orbit. The results showed that the surface of some of the old landslides is still active. The ground surface displacement velocity is estimated between 58 to -22.5 mm/year in the Ahar Chai sub-basin. The spatial distribution of landslides is mainly located on the southern slope of the Ahar Chai river. Due to the time mismatch of the used satellite images with the date of the Ahar-Varzaghan earthquake, it was not possible to investigate its effects in the region of the region's unsteady exacerbation. Recent unstable slopes of this region should be examined using new processing techniques and updated & available radar images. The accuracy of the results is necessary to be evaluated due to accurate measurement
Authors: Yarahmadi, Jamshid (1); Shadfar, Samad (2)Nanjing, located in the western gateway of the Yangtze River Delta, is the provincial capital of Jiangsu province and the second-largest city in the East China region. Since adopting a strategy of developing its economy across the Yangtze River, Nanjing has accelerated its pace of development to foster new industries and towns. Hexi New Town, located in the southwest of Nanjing's main city zone, is one of the large scale residential and commercial regions. Due to the rapid urban construction and impact of the geological condition, the Hexi area has been suffering from land subsidence. By utilizing remotely sensed images obtained from various sensors viz. Landsat MSS, TM, ETM+ and OLI/TRIS, the urban Sprawl and its spatial and temporal characteristics have been studied over 35 years (1984-2018). The results indicate that urban spatial expansion experienced three stages. However, most of the studies focus on land subsidence monitoring of the Hexi area during the rapid external expansion stage (from 2004 to 2014). In this study, the Small Baseline InSAR (SBAS-InSAR) technique has been employed to process Sentinel-1A images acquired between 2015 and 2018 to investigate land subsidence in the Hexi area of Nanjing city. The SBAS-InSAR results are validated by the ground leveling measurements and the derived subsidence rates show an RMS difference of 3.67 millimeters per year (mm/yr) with a mean of 3.11 ± 1.08 mm/yr compared with the leveling data in the same period. Results show that four major areas of subsidence are detected: two of them are in the north of Hexi New Town, one is in the southeast and the rest is in the south. These demonstrate that the SBAS-InSAR technique with Sentinel-1 Data is a powerful tool for longtime monitoring of land subsidence. Besides, we investigate the influence of natural conditions and human activities on land subsidence by analyzing InSAR derived subsidence rate maps. The three main areas have suffered subsidence problems before 2014 and the main triggering factor of land subsidence is the consolidation and compression of shallow strata. According to our previous research, the maximum annual subsidence rate reached more than 50 mm/yr from 2007 to 2011. But in this study, the subsidence rate varies from−31.16 to 10.20 mm/yr, which indicates that the subsidence problem of these areas has been relieved in recent years. The rest area is located in the south of Hexi New town and land subsidence in this area is not apparent before 2014. Comparisons between high-resolution images acquired in March 2014 and November 2018 show that many new buildings are constructed in this area and the construction of traffic has been speeding up in recent years.
Authors: Xu, Jia; Yang, Zhen; He, XiufengJiangsu is located in the lower reaches of the Yangtze River and the northern part of the Yangtze River Delta, with a total area of 107,200 square kilometres. It is one of the most developed provinces and accounts for 10% of China’s Gross Domestic Product. Beginning in the 1990s, Jiangsu was the region most affected by the ground subsidence in China, and the settling funnel areas with 400 mm subsidence covered more than 2,000 square kilometres. In the regions of Suzhou, Wuxi and Changzhou, groundwater mining is completely prohibited from late 2005. After more than ten years of strict management, the local subsidence rate has slowed down significantly, the hydrogeological environment has improved sustainably, and the groundwater level has risen across the board. Continuous monitoring of ground subsidence can provide decision support for water resources management and groundwater extraction planning. Interferometric synthetic aperture radar (InSAR), especially with the high data availability of the new generation of satellites, such as Copernicus Sentinel-1, has undergone rapid development in monitoring deformation over a wide spatial coverage. Nowadays, InSAR mainly faces the accuracy limitation due to atmospheric effects and the constraints of computing resources brought by big data. Here, we propose a new approach to combine conventional InSAR stacking with Generic Atmospheric Correction Online Service for InSAR (GACOS) based atmospheric correction to generate displacement time series as well as the mean displacement rate. Firstly, the atmospheric signals of the InSAR interferograms are removed using GACOS products, so that the residuals are more likely to be random. Then, the mean rate of each pixel is estimated in a temporal window by least squares. The displacement time series can also be obtained based on the integration of the mean rate. This process is algorithmic efficiency and could be processed in parallel. To further accelerate the InSAR processing of massive data, a Graphical Processing Units (GPU) based processor is implemented. The conventional InSAR data processing time of an interferogram covering a single frame could be reduced from around 1 hour to less than 10 minutes in the same computational environment. We will present the ground subsidence map over the entire Jiangsu Province based on the proposed approach. In particular, we will process the available Sentinel-1 TOPS data acquired in 2019. External data, such as Levelling and Global Navigation Satellite System (GNSS) data from Continuously Operating Reference Station (CORS), will be collected and used for validation. Moreover, assessment of GACOS performance will also be included in this work.
Authors: Xiao, Ruya (1); Li, Yongsheng (2); Yu, Chen (3); Li, Zhenhong (4,3); Xu, Jia (1); He, Xiufeng (1)Landslides are a major cause of damages or losses of property and life around the world. The detection and mapping of active slopes is a critical task for landslide disaster prevention and early warnings development. SAR Interferometry (InSAR) has been proven to be effective for landslides deformation measurement. However there are blind areas, at where the landslides cannot be detected by InSAR. These areas are generally caused by geometric distortion and vegetation decorrelation. In engineering applications, identification of blind areas can not only provide a reference for the selection of SAR data, but also provide guidance for other technical tools such as field geological surveys to improve the identification of potential landslides. This study divides the wide-area InSAR landslide detection into two parts: Prior identification of undetectable areas and early identification of landslide. First，we use DEM data, landcover map, and SAR acquisition parameters (including satellite heading angle and radar incidence angle) to conduct visibility estimation, sensitivity estimation, and InSAR applicability estimation, then forming a comprehensive evaluation index. The visibility estimation is used to identify layover and shadow areas caused by SAR observation geometric distortion. The sensitivity estimation is used to identify areas that are not sensitive to landslide deformation detection. The applicability estimation is used to identify low-coherence areas due to vegetation coverage. Their comprehensive evaluation indicators are provided to provide the final applicability map of landslide detection. Then, time series InSAR is used to detect the wide-area landslides using the StaMPS/SBAS tool. A huge landslide occurred on 23rd July, 2019 in Shuicheng County, Guizhou Province, China, causing huge damage. Taking Shuicheng County as an example (Figure 1), this study focuses on blind areas of landslide detection and early identification of landslides. This place is in the area of high mountains and canyons, with elevation ranging from 700 to 3000 meters above sea level. The landform landscape is dominated by mountains and hills, and there are landforms such as mountain plains, plateaus, and terraces. And the central Shuicheng County is violently eroded by the severe cutting and erosion of the Beipan River, hence the geometric distortion is very serious. The L-band ALOS-2 PALSAR-2 and C-band Sentinel-1 datasets were used. Among them, the ALOS-2 PALSAR-2 data set is in FBD mode, with a total of 16 scenes; the Sentinel-1 data set is in IW wide mode, with 67 ascending scenes and 69 descending scenes, respectively. The result blind area detection is shown Figure 2. From the visibility of landslide detection (Figure 2(a)), the shadow area is distributed on both banks of the river valley and the edge area of the elevated terrain. Most of the shadow areas of the ascending orbit overlap with the layover areas of the descending orbit. Therefore, some areas are still undetectable even combining both ascending and descending orbits, which needs additional technical means for supplementary investigation. From the sensitivity of landslide detection (Figure 2(b)), the sensitivity to the north-south slope is the lowest, and the east-west slope is highly sensitive. Therefore, special attention should be paid to landslides with low sensitivity. Judging from the applicability of InSAR (Figure 2(c)), , the applicability is poor in most areas due to dense forest cover, which needs long-wavelength SAR data to obtain high coherence. We detected six active landslides from the deformation rate maps (Figure 3), covering more than 500 square kilometers, which were interactively verified by the three SAR datasets. Most of the landslides are distributed on the edge of the high-level terrain, presenting huge threat to the local residents at the valleys. There are two landslide areas caused by mining deformation (Figure 4). Continuous monitoring of the safety of buildings around the mine is required, although the danger is relatively small. It should be pointed out that the Jichangzhen Landslide was not identified from the wide-area detection result. The Jichangzhen landslide is located four kilometers west of the Beipanjiang River, on the southern slope of Jichang Town. The landslide descended from south to north, destroying the buildings that on the slope and foothills, causing huge casualties. From the time series deformation of point P3 (Figure 5), no obvious deformation were found. However, there are landslides called Fanazu and Xujiaying landslides that have been detected on the opposite mountain, whose deformation degree exceeds 60mm/y. The time series deformation of points P1 and P2 shows seasonal pattern. The deformation in winter is small, and the deformation in summer is large. Considering that the geological conditions here as same as those of the two nearby landslides, we speculate that the landslide may be a sudden landslide caused by rainfall. Therefore, the rainfall threshold method should be adopted in the region to carry out meteorological and geological disaster monitoring and early warning.
Authors: Wang, Yian (1); Do, Jie (1); Gong, Jianya (1); Liao, Mingsheng (2); Zhang, Lu (2)In recent years very slow moving landslides has been monitored in Tyrol using very precise geodetic surveys. From these measurements it is possible to reconstruct the deformation patterns of the mass movements in vertical, north–south and east-west direction. Thanks to the availability of Sentinel-1 data and the application of the MSBAS-2D method (Samsonov et al., 2020) it is now possible to combine ascending and descending data in order to obtain prolonged time series which can be compared with the GPS measurement along the east-west and vertical directions. At the Geological Survey of Austria, we adopt a reliable and incremental workflow applied to Sentinel-1, which can be used for landslide monitoring. This workflow is made of four components: The “Centre spatial de Liege“ InSAR Suite (CIS) software (Deerauw et al., 2019, Deerauw et al., 2020); A series of scripts for mass processing (d´Oreye et al., 2019); The MSBAS algorithm (Samsonov et al., 2020); A new module based on GACOS (Yu et al., 2018) data for tropospheric delay removal. We will present three case studies, which are located in the province of Tirol and were investigated within the scope of the Vigilans project founded by the FFG Austrian Space Applications Program. Two of these mountain slopes are active since 2016 and a creep behaviour of a deep-seated landslide is threatening a small village since 2012. This behaviour is clearly visible by a previous PSI and SBAS analysis performed by the Geological Survey of Austria with Envisat data. Moreover, as it was already mentioned in Hormes et al., 2020, a considerable number of geodetic points, which were thought to be stable are situated on unstable terrain. Thanks to the cooperation of our institution with the Federal Office of Metrology and Surveying (BEV) this could be taken into account, in order to proof the reliability of the InSAR-data derived from deformation maps in a large scale. In fact, however for the three case studies the time series, derived from Sentinel-1, has being used as benchmarks for reliability, the geodetic points, which bear only the mean annual displacement information, allow us to evaluate the quality of the velocity maps for east-west and up-down components. Those preliminary results show good agreement with the geodetic unstable points and more interestingly, we were able to reliable assess the vertical and lateral displacement trends for the three sites. Literature: Derauw, D., d’Oreye, N., Jaspard, M., Caselli, A., & Samsonov, S, (2020): Ongoing automated ground deformation monitoring of Domuyo - Laguna del Maule area (Argentina) using Sentinel-1 MSBAS time series: Methodology description and first observations for the period 2015–2020, Journal of South American Earth Sciences, Volume 104, December 2020, 102850. Derauw D., Libert L., Barbier C., Orban A., Kervyn F., Samsonov S., d’Oreye N. (2019). The CSL InSAR Suite processor: specificities of a command line InSAR processing software specifically adapted for automated time series processing. Abstract, 13-17 May 2019, ESA Living Planet Symposium 2019, Milano, Italy. d’Oreye N., Derauw D., Libert L., Samsonov S., Dille A., Nobile A., Monsieiurs E., Dewitte O., Kervyn F. (2019). Automatization of InSAR mass processing using CSL InSAR Suite (CIS) software for Multidimensional Small Baseline Subset (MSBAS) analysis: example combining Sentinel-1 and Cosmo-SkyMed SAR data for landslides monitoring in South Kivu, DR Congo. Abstract, 13-17 May 2019, ESA Living Planet Symposium 2019, Milano, Italy. Hormes A., Adams M, Amabile, A, Blauensteiner F., Demmler C., Fey C., Ostermann, M., Rechberger C., Sausgruber T., Vecchiotti F., Vick M.,L, and Zangerl C., (2020) Innovative methods to monitor rock and mountain slope deformation. Geomechanics and Tunneling, Volume 13, February 2020. Samsonov S., Dille A., Dewitte O., Kervyn F., and d'Oreye N. (2020) Satellite interferometry for mapping surface deformation time series in one, two and three dimensions: A new method illustrated on a slow-moving landslide. Engineering Geology 266 (2020). Yu, C., Li, Z., Penna, N. T., & Crippa, P. (2018). Generic atmospheric correction model for Interferometric Synthetic Aperture Radar observations. Journal of Geophysical Research: Solid Earth, 123.
Authors: Vecchiotti, Filippo (1); Amabile, Annasara (1); Ostermann, Marc (1); Adams, Marc (2); Demmler, Christian (3); Fey, Christine (4); Hormes, Anne (3); Rechberger, Christina (4); Sausgruber, Thomas (5); Zangerl, Christian (4)Ongoing geological instability associated with mass movements and sinkhole formations is jeopardizing some of the densely populated neighbourhoods of Maceió in Brazil, where fractures and damages on infrastructures have been intensified since early 2018, forcing the authorities to relocate the inhabitants and change the urbanistic asset of the neighbourhoods. Deep geomechanical alterations related to salt dissolution due to mining activities which were further aggravated by severe precipitation events have been indicated as potential causes of the geohazard. Here we present a complete history of the temporal and spatial evolution of subsidence in the last two decades using a full archive of Synthetic Aperture Radar (SAR) from past and currently operational satellite missions: C-band ASAR ENVISAT, the L-band ALOS-1 POLSAR, L-band ALOS-2 POLSAR and C-band Sentinel-1 missions. The historic missions cover the time period between October 2003 and January 2011. There is a data gap between January 2011 and March 2015. The currently operational ALOS-2 covers the time interval between March 2015 and September 2019 in ascending mood and the period between April 2015 May 2017 in the descending mood. Sentinel-1A covers the period from October 2016 until now. Our findings show precursory subsidence already in 2004-2005 at rates 4 cm/year near the Mundaú lagoon coast, which increased more than six-fold (up to 25 cm/year) in 2017-2018. Maximum cumulative subsidence of approx. 2m was detected for the whole period between July 2004 and November 2020. Angular distortions due to surface subsidence suggest the extent of geological hazard over an area of 1 square km, with most of the cracks surveyed in 2018 falling in the region classified as high and very high levels of hazard, which mostly coincide with the transition zone where the ground gradually goes from stable to an unstable condition. Geophysical inversion and a 2D discrete element modelling (DEM), suggest an upward cavity migration related to cavity roof collapses of a shallow source ( depth < 1 km and > 500 m) and annual volume change of approximately E+05m3, comparable to that of a single salt cavity. Moreover, the 2D DEM modelling highlights the generation of extended fractured zones related to the cavities instabilities that foster underground water percolation and erosion, aggravating furthermore the geological instability in the region.
Authors: Vassileva, Magdalena Stefanova (1,2); Al-Halbouni, Djamil (1); Motagh, Mahdi (1,2); Walter, Thomas (1); Dahm, Torsten (1,3); Wetze, Hans-Ulrich (1)The Tishinskoye field is located on the territory of the Ore Altai in the East Kazakhstan region of Kazakhstan in a mountain basin at the foot of the Ivanovo ridge, in the upper reaches of the Ulba river (a tributary of the Irtysh). The height above sea level is about 800 meters. The fields are being developed by KAZZINC LLP. The subsoil of East Kazakhstan is rich in minerals. Lead, zinc, silver, gold, copper, titanium, tantalum, magnesium, cadmium, tellurium, and other metals are mined and processed here. The terrain of the work area is mountainous, forests and taiga, and Alpine meadows. Mountain systems are divided by wide inter-mountain depressions. A significant part of the mountains is covered by taiga. Mainly dominated by coniferous species: spruce and fir. Huge arrays of shrubs grow here. The ore bodies of Tishinsky deposit were explored to a depth of 1200÷1300 m. The field was Worked out in a joint way-open and underground mining. At the moment, there is only mining of ore by the mine method-under the quarry, which has been completed. In turn, there are active vertical mines and structures located near the quarry area. In this regard, the development of the Tishinsky field and the resulting underground voids cause deformation of the massif, and as a result, the earth's surface is shifted at the quarry. Therefore, monitoring of deformation processes on the territory of the Tishinsky quarry is an important task that ensures safe and continuous development of the Deposit. The object of research is a quarry with joint underground mining. The expected subsidence at the beginning of observations in 2017 is up to 700 mm / year. Data from the Sentinel-1a and-1b spacecraft were used. Due to the expected high settling rates, a time limit of 30 days was imposed on the construction of interferograms. The spatial basis was not limited.The radar data was processed in the GAMMA software. Taking into account the large height differences, as well as the involvement of horizontal displacement deformations in the processes, positive results were achieved when estimating subsidence using averaging results from two opposite orbits of the Sentinel-1 satellite. A small square of 200*200 pixels was used. The mask was not used for the phase scan. In this case, despite the low spatial resolution of the Sentinel-1 satellite data, it was possible to determine subsidence up to a value of 0.7 m/year. Subsidence maps based on multidirectional data allowed us to estimate horizontal shifts in the quarry. The serif vectors were the displacement vectors from the two orbits of the Sentinel-1 satellites, and the third serif vector was the normal to the side of the quarry. The orbits of the Sentinel-1a and Sentinel-1b satellites are multidirectional: Sentinel-1a takes pictures in an easterly oblique direction, moving North, and Sentinel-1b in a westerly direction, moving South (except for the angle of inclination of the orbit). Potentially, two remote sensing vectors can solve a serif in the Z-X plane and determine shifts in this plane, i.e. subsidence, and horizontal displacements along the x-axis. next, we consider the geometric parameters with respect to the WGS84 (m) coordinate system in the UTM – 44 projection of the zone at the Tishinsky quarry. To determine the horizontal shifts in the Z-X plane, it was necessary to determine the projections of the angles of inclination on the z-x plane. Next, it was necessary to solve the serif on the plane using trigonometric dependencies. Strict determination of horizontal offsets in the Z-Y plane is not possible, since this configuration uses only two vectors located approximately coaxially in the z-x plane.a minimum of three vectors are required for spatial serif. To determine offsets in the Z-Y plane in a non-strict way, it is assumed that as a result of monitoring vectors of this configuration, projections of planned offsets were measured, the vectors of which were directed radially to the center of the quarry (or from the center). Finding projections of y-axis offsets requires in this case simply multiplying the x-axis offsets by the tangent of the azimuth angle of the selected point relative to the center of the quarry. The configuration of the location of the sensing vectors of satellites relative to the coordinate system and configuration career (as a spatial geometric shapes) has required the transition of the azimuthal system with an inclination relative to the existing 8.5°, and having a non-linear change of angles in a circle. Due to the quarry configuration, corners are compressed in the range 270°-90°, and stretched in the range 90°-270°. The distortion of the angles around the circle is based on a sinusoidal relationship. If this configuration is allowed, points located outside the expected displacement zone were eliminated. That is, for points located outside the quarry, the assumption of their radial displacement was not applied. In addition, points located at azimuth angles 80-100 and 260-280 were eliminated, since at azimuths close to 90 and 270, determining horizontal shifts based on the configuration of the observation data vectors becomes impossible due to the tendency of the tangent of the azimuth angle to infinity. Thus, spatial displacements of points on the Tishinsky quarry were obtained. All displacements were calculated over three years from 2017 to 2019. the Maximum horizontal displacements were recorded on the Northern slope of the quarry, reaching a value of up to 390 mm in 3 years. It should be noted that the inclination of spacecraft orbits led to the formation of "blind spots" on the southern side of the quarry, which may not have allowed to identify the maximum horizontal shifts. Also, the fact of the influence of the orbit inclination on the detected values is confirmed by the detected horizontal shifts along the x axis, where the maximum values are not fixed symmetrically on the left and right sides, but are shifted to the North of the Western and Eastern sides. These regions are directed normally to the sensing vectors. In this work, a model of absolute shifts (horizontal and vertical) was obtained based on Sentinel-1 data on the territory of the Tishinsky quarry of the Republic of Kazakhstan. Space data was used for three monitoring seasons – from 2017 to 2019. The season used 17-18 scenes of each satellite. A positive result was achieved at this site due to good data coherence due to the exposed sides of the quarry and their geometric shape.
Authors: Musikhin, Vasiliy; Solomennikov, Mikhail; Anastasiya, Khvostanceva; Nina, KharinaAs already widely shown, landslide mapping and monitoring of slope instabilities have been greatly benefiting from the use of remote sensing technologies in the recent past. Earth observation (EO) optical systems are mainly used for post-failure mapping and recognition of landslide features, supporting and even exceeding traditional field surveys by the possibility of creating systematic multi-temporal landslide inventories. Advances in EO radar remote sensing contribute to detailed quantification of the state of activity and measurement of superficial and/or previously undocumented motions within unstable slopes, complementing the information obtained by optical remote sensing systems for landslide hazard assessment. The current availability and recent advances in both optical and radar satellite technologies, e.g. thanks to the development of EU Copernicus program and other commercial and national satellite programs, have made the use of EO data ever more effective for landslide investigations. This paper describes the outcome of our research conducted using both radar and optical satellite remote sensing to characterize spatiotemporal patterns of landslide occurrence that happened in response to the March-April 2019 heavy rainfall events in the Golestan province, of Northern Iran. The 2019 heavy rainfall events in Iran occurred from mid-March to April 2019 and have caused widespread flooding in 25 of Iran’s 31 provinces. Several parts of the country have experienced extensive landsliding as a consequence of the March-April rainfall events, whereas the provinces of Golestan and Lorestan were among the hardest-hit areas. At first a semi-automated approach using time series of optical Planet Scope and Sentinel-2A/B data was used for large-area landslide detection. Several catastrophic failures were identified and classified according to their mechanism, areal extent and material involved. The obtained results were subsequently evaluated by field surveys conducted in September 2019 in cooperation between the GFZ Potsdam and the Forest, Range and Watershed Management Organization of Iran (FRWM). Moreover, C-band Sentine-1 and L-band ALOS-2 data were analysed using differential interferometric synthetic aperture radar (DInSAR) to quantify the state of slope instability in both precursory and post-disaster phases at a number of places. Our findings show that an integrated approach exploring satellite imagery, in-situ measurements and field survey allows not only the determination of landslide and slope instability processes in Golestan following the 2019 extreme rainfall events, but also helps to better understand the mechanism of mass movements and the role of meteorological and anthropogenic influencing factors in controlling the onset of slope failures.
Authors: Motagh, Mahdi (1,2); Stefanova Vassileva, Magdalena (1,2); Roessner, Sigrid (1); Behling, Robert (1); Akbari, Bahman (3); Abbasi, Mohammad (3); Haghshenas Haghighi, Mahmud (1,2)The purpose of this study is regular monitoring of an active landslide movements in the area golf club ”Thracian Cliffs” located on the Northern Black Sea coast of Bulgaria by combined use of data from global navigation satellite systems and interferometric images from Synthetic-aperture radar (SAR).The geological feature of the area is landslide strip formed by the old landslides having an average width of 400–500 m exhibiting steep slopes 40–50 m at certain locations. Usually in this region landslide bodies are formed by 3–4 visible linearly oriented steps and hills (landslide packages) having different heights. Most often behind them are located enclosed negative ground forms with permanent or temporary swamps. The toe of the studied landslide is found below today's sea level, where 1-2 invisible on the surface landslide packages are formed. The depth of the old landslides reaches up to 40-60 m and more at certain places. In addition to the ancient landslides recent active local landslide processes emerge. Their depth and range are usually less than that of the ancient landslides, but they cause considerable damages to the population and the economy. In recent years such landslides are: the landslide just outside the golf club “Thracian Cliffs” and separate plots inside the area of the same golf club. These structurally complex geological formations situated in this area give rise to shallow landslides, which can be monitored and studied with InSAR-based techniques.In this research the primary source of SAR data is the Sentinel-1 data hub. The first step to achieve the main objective was to create a local archive with Sentinel-1A/B images from 58th ascending and 36th descending orbits for the region of Northeastern Bulgaria who at the moment consists of about 400 SLC images. For mapping the deformations in the region of interest interferometric images (IFIs) at intervals 4 and 8 months were produced. Those time intervals were used since one of the main factors affecting the quality of the IFIs is the vegetation and for this reason IFIs were created only from autumn and spring scenes. Other factor supporting this is that most of the landslides activations occur in those seasons. One more factor that should be considered before producing IFI is the presence of snow – used data are from days with no snow coverage. The ground displacements are calculated from the phase signal after unwrapping and are color-coded. In the final image those pixels from the IFI that had corresponding coherence values below 0.3 were removed since they were considered unreliable. For this study a control geodynamic network covering the landslide area before golf club” Thracian Cliffs” was established. It consists of a total of 10 GNSS points, stabilized with metal pipes 35 cm long and used to monitor deformations along the road leading to the studied area. The first GNSS measurement cycle of the geodynamic network was carried out in June 2019 and second cycle is made in July 2020.The results presented provide solid base to affirm that the IFIs produced from satellite SAR data are suitable for studying the ground displacements that occurs after landslide activation. For the specific period the displacements calculated range from few centimeters to decimeters which is in line with the expected yearly values for this area. This information can be considered as the only source for sites, when the terrain is difficult to reach and impassable. It can be said that the implemented approach can be used for exploration and monitoring of the whole landslide area. The authors established strong relationship between geodetic and satellite derived information concerning ongoing landslide processes. It needs to be underlined that the results confirm the fact that the monitoring of landslides by InSAR techniques can save a lot of time and efforts of handling the damages caused by them .
Authors: Atanasova-Zlatareva, Mila Stoyanova (1); Nikolov, Hristo (2)Land subsidence related to groundwater withdrawal is a potentially dangerous and underrated geohazard due to its slow displacement rate. However, many areas around the World suffer the strong impact of land subsidence effects like reduction of aquifer-system storage capacity, land fissures, damage on buildings and civil infrastructures, and increased flood susceptibility and risk. In the current context of climate change, urban sprawl and population growth, the use of strategic groundwater resources is expected to increase, fostering the development of land subsidence. In order to raise awareness, inform decision-makers and help in formulation of effective land-subsidence policies, we have developed a global map of potential subsidence due to groundwater withdrawal (Herrera et al., 2021). For this purpose, we have used information from 200 land subsidence reported areas to perform a statistical analysis of the main variables influencing subsidence to identify environmental settings favoring the development of this phenomenon and the anthropogenic factors leading to groundwater depletion. Lithology, slope, land cover and climate were selected for the susceptibility model. Recurrence of the different classes of each layer showed that most of the records were located over flat sedimentary basins and coasts, near urban and agricultural areas with cool/arid climate and dry seasons. Last, groundwater depletion information was introduced to compute the potential model. The results of this work showed that 1.6% of global surface is threaten by land subsidence, affecting 19% of global population and 12% of global GDP. InSAR is currently one of the most important techniques for detecting and monitoring land subsidence. In this work, 74 of the 182 publications used InSAR data for land subsidence monitoring, which emphasizes the significance of this technique for the study of this phenomenon, especially in recent years. InSAR results from previous works over six locations have been used to perform a spatial validation of the global map results. Good correlation between subsidence rates and high potential areas detected confirms that InSAR is an appropriate technique for new studies on the identified high potential subsidence areas. In view of new upcoming projects geared towards the provision of massive InSAR data like the Copernicus European Ground Motion Service, the potential land subsidence map can help InSAR community to interpret and identify possible triggers of detected displacements. Reference: Herrera-García, G., Ezquerro, P., Tomás, R., Béjar-Pizarro, M., López-Vinielles, J., Rossi, M., Mateos, R.M., Carreón-Freyre, D., Lambert, J., Teatini, P., Cabral-Cano, E., Erkens, G., Galloway, D., Hung, W.-C., Kakar, N., Sneed, M., Tosi, L., Wang, H. & Ye, S. 2021. Mapping the global threat of land subsidence. Science, 371, 34-36.
Authors: Ezquerro, Pablo (1,2); Herrera-García, Gerardo (1); Tomás, Roberto (3); Béjar-Pizarro, Marta (1); López-Vinielles, Juan (1,4); Mateos, Rosa M. (1)It is well known that, the Synthetic Aperture Radar Interferometry technique (InSAR), which measures surface deformation from satellites, can provide insights about pressure changes in the subsurface. At active geothermal fields, pressure variations are usually induced by both fluid extraction and fluid injection. Consequently, the use of InSAR across geothermal fields can help monitoring production and recharge, in addition to help monitoring hazardous deformation. Here, we present examples of InSAR application at four different geothermal sites. The first example is at the Olkaria power station (Kenya), which produces about 727 MW from five power plants. Between summer 2018 and summer 2019, we observe regional and more localised signals related to fluid extraction and fluid injection. The second example is at the Cerro Prieto geothermal complex (Mexico), which produces about 820 MW from four power plants. Over the entire year 2019, we observe a complex mixture of deformation signals, related to two independent factors: the geothermal production and the regional tectonic activity. Additionally, a subsidence signal is observed over the recharge area. This confirms that the lack of geothermal fluid extracted at the power plants is due to a lack of geothermal fluid replenishment in the reservoir. The third example is at the Coso geothermal complex (California), which is located in a tectonically active area and which produces about 270 MW from four power plants. Analyzing the degree of correlation between pumping data and surface deformation provides indications on the tectonic and/or geothermal origin of the deformation signals. Finally, the fourth example is at the Svartsengi power station (Iceland, near the Blue Lagoon), which produces about 75 MW from six power plants. The area has undergone strong underground volcanic and tectonic activity in early 2020 and early 2021. We analyse the impact of these natural events on the geothermal site. These four examples illustrate well the potential of InSAR in (1) assisting geothermal production by providing insights of fluid flow dynamics, and (2) investigating the anthropogenic and natural contributions of the deformation signal. Additionally, regular InSAR monitoring can contribute to forecast any hazardous deformation that would put the exploitation at risk.
Authors: Giniaux, Jeanne; Magnall, Nathan; Lloyd, Amy; Carvalho, LucasStudies of slope stability, form and process in geomorphology increasingly consider the mechanical characteristics of earth materials. Riprap is the most widely used solution for protecting earth dams. Considerable progress on predicting riprap stability has been made over the last 30 years, based mainly on Los Angeles abrasion tests. Various stability formulas that relate the median weight (W50) of the riprap to the slope have been proposed. The purpose of this paper is to present the results of the field investigations carried out in the Taleqan riprap dam in the summer of 2016 and of monitoring of its surface displacement. In this study, we aimed to assess the practicality of the InSAR method for monitoring surface deformation through the analysis of the surface deformation of the Taleqan dam northwest of Tehran, Iran. The analysis was carried out using 102 (Ascending) Sentinel-1A images and TerraSAR-X (TSX) images from 1.01.2014 to 1.08.2019 and from 15.12.2017 to 30.9.2018, respectively. Deformation determined through the satellite imagery was compared to the deformation observed by Los Angeles abrasion tests . We show that high-resolution X-band SAR data provides a much more detailed view of dam deformation that is not possible to infer from field measurements. High-resolution TSX data reveal that the dam has a maximum deformation rate of approximately 80 mm/y in the radar line-of-sight (LOS). The downstream part of the dam has a deformation rate of 120 mm/y. Sentinel-1A images from 2014 to 2019 show the velocity of -4 mm/yr. However, the survey results indicate relative stability of the embankment mass for the time period of March 2014 to December 2015. During the field surveys, however, drastic weathering and erosion of the dam downstream riprap materials were observed. Crushing and changing the dimensions of riprap rocks reduce the friction angle of the layer. The comparison of the results at the dam’s body permits us to draw certain conclusions on the mechanisms, weathering and erosion of the riprap rocks based on the principal causes of damages. Finally, one rock type among five used in riprap was more easily eroded, thus caused displacement of the riprap.
Authors: Ghadimi, Mehrnoosh (1,3); Hooper, Andrew (2); Whipp, David (3)The groundwater exploitation induced land subsidence is a well-known phenomenon and has been documented in places like Tokyo, Jakarta, Tehran, and Mexico. Delhi- the capital of India, is the fifth most populous city in the world, with a population density of nearly 30,000 people per square mile. Like other global megacities, Delhi is also facing the looming groundwater crisis due to urbanization and illegal pumping and is considered as a critical zone by the government of India. The rapid extraction of groundwater and supporting physiography and hydrogeology of this region makes it prone to land subsidence. In this study, we demonstrate the results of InSAR time series analysis of the NCR Delhi region using Permanent Scatterer Interferometry (PSInSAR). By constructing single-master interferograms derived from Sentinel-1 ascending and descending SAR datasets acquired between 2014 and 2020, we computed the time series of land subsidence in the region. The interferometric processing was done using SNAP followed by time-series analysis using Stanford Method for Persistent Scatterers (StaMPS). To remove the tropospheric error, a phase-based linear correction was applied using the TRAIN toolbox. LOS velocities are then decomposed into the vertical and east-west direction to visualize the uplift and deformation more clearly. Finally, the results were compared with in-situ groundwater observation wells to study the correlation between the two. Our analysis identifies three distinct subsidence features in NCR Region. The maximum deformation exceeding 15 cm/year is found in the immediate vicinity of Indra Gandhi International Airport. The other two regions, i.e., Faridabad and Gurugram show a deformation rate of 4 and 5cm/year, respectively. Interestingly, an uplift of the rate of 2cm/year is identified in Dwarka, just adjacent to the deformation zone. The in-situ groundwater well data provided by Central Ground Water Board (CGWB) India is not consistent and has many gaps. However, the careful data analysis of some of the wells followed by spatio-temporal interpolation provides an estimate of the depth to the groundwater level. The groundwater depletion rates show an agreement with the trend of land subsidence. Furthermore, the results signify that the areas with high subsidence rates show a high groundwater depth in Delhi NCR during 2014-2017.
Authors: Garg, Shagun (1,2,3); Karanam, Vamshi (2,3,4); Motagh, Mahdi (2,3)Sea transgression observed in Albanian shores of Adriatic Sea has been for two decades in experts attention, trying to understand real factors behind this phenomenon that has destroyed economic activities in two touristic beaches – Semani in south and Patoku in north. Satellite imagery was used to delineate changes of waterlines and two main hypotheses are considered as the cause of sea transgression: probably natural abrasion of sandy shores from weaves and sea currents versus ground subsidence caused by human activities and neotectonic activity in Preadriatic Depression (lowland part of the country bordering Adriatic Sea). In this work the analysis of Sentinel-1 SAR images for identification of vertical (line of sight) ground movement is presented. Data were obtained from ESA repository https://scihub.copernicus.eu/ and processed with Sentinel Application Platform [http://step.esa.int/main/toolboxes/snap/]. A virtual machine in the ESA RSS Cloud Toolbox platform [https://eogrid.esrin.esa.int/cloudtoolbox/] was offered by ESA to download Sentinel images and run the SNAP software. Holocene evolution of Adriatic Sea shores is found in (Fuache et al, 2010). An overview of actual Preadriatic Depression is found in (Simeoni et al, 1997), (Pano et al, 2006) and (Frasheri et al, 2011), indicating Adriatic shoreline changes and Preadriatic Depression tectonic faults. Photos of seashore status, testimonial of sea transgression and related damages, are included in (Frasheri et al, 2014); methodology for combination of NIR bands to identify water bodies changes is presented, illustrated with traces of river deltas and sand shift, arguments for the abrasion hypothesis. An analysis of human activities impact in Adriatic Sea shores is found in (De Leo et al, 2016). Tentative of using SAR differential interferometry for tracing eventual ground subsidence in Preadriatic Depression presented in (Lamaj et al, 2015). Using persistent scatterers for a small section of central Depression, low values of subsidence were observed in urban areas and the delta of Shkumbini River. Analysis of persistent scatterers was done for the city of Tirana is found in (Kuçaj, 2016). In both cases calculations for SAR analysis using persistent scatterers were done by partners in Italy. Further SAR analysis is presented in three papers of 2017 (Frasheri et al, 2017). While vegetation coverage is supposed to generate enough noise in interferograms, fringes are persistent in some hilly ranges but not enough to consider the subsidence. Situation in northern part of Depression with fringes persistent in Renci and Kakariqi mountains while missing in the narrow valley separating them is considered as a proof that other environmental factors may be the cause. Subsidence traces associated with high values of coherence were observed in urban areas and highways. A suite of SAR images was obtained spanning the period of 2015-2018 with 3-4 months step for both northern and southern areas of Preadriatic Depression. Terrain elevation was generated from interferogram based on images from years 2015 and 2018. Northern part of Preadriatic Depression is characterized by mostly rugged terrain and minor flat areas of Durresi – Lalzi, FusheKruja – Lezha, Zadrima – Shkodra, with the Shkodra Lake at the northwest. In southern part of Depression the flatlands of Myzeqe are extended from Vlora up to Durresi Bay with a low hills range in between. With particular interest are two parallel long mountains of Renci and Kakarriqi situated at north of Shengjini Bay, separated by the narrow long valley of Torovica. While heights of these mountains are characterized by presence of fringes while not the valley in between. For each area interferograms were generated using images for dates as follows, with master image from the year 2015. Average was calculated from suites of coherence and phase of each interferogram. The average coherence from the suite of interferograms is presented in Fig.3. Coherence values are classified with red for values under 05. and cyan for values over 0.5. It is apparent that higher values of coherence correlate with highlands and urbanized areas. Main cities are clearly visible, including the highway Tirana – Durresi. Average phase from the same suite of interferograms with master image from the year 2015 is characterized by persistent fringes situated in slopes around Shkodra Lake, Renci and Kakariqi mountains and heights in the east bordering Zadrima flatland, also Tirana city and top of Mountain with Holes in the east. In southern part of Depression there are few weak traces of fringes, except mountains of Ionian rivera south of Vlora city. Scatterers were identified from coherence values higher than 98% and to reduce their number the second order phase trend was calculated for them, keeping only pixels with error less than 25%. Resulted that such pixels were grouped in a small number of clusters concentrated mostly in inhabited areas and highways with buildings around. There were less clusters in hilly areas where fringes were usually obtained. The variation of wrapped phases for such “persistent” clusters seems quite random. A tentative was done to unwrap phases, and a clearer variation of phases was obtained. The sign of first wrapped phase difference was used for orientation increase versus decrease unwrapping. The variation of unwrapped phases resulted 20 – 50 rad implying unrealistic line of sight displacement. The first order coefficients of temporal second order trend for unwrapped phases of persistent clusters displayed a different situation for the alternative hypothesis of increasing versus decreasing of line of sight displacements, compared with wrapped phase case. Considering that excessive variations of wrapped phases were subject of noises, a second test was done calculating temporal polynomial trend analysis for 1200 days period of wrapped phases for all temporal suites of pixel in interferograms images. Significant coherence and phase characterize urban areas and highways, especially the segment Tirana – Durresi. Subsidence in Tirana city, due to intensive building during last twenty years, is already verified in (Kucaj 2016). The highways are subject of intense car traffic and new buildings along its sides, creating premises for subsidence. Variation of displacement orientation is supposed from combination of subsidence and increase of buildings heights. Combination of negative & positive trends in different areas probably from environmental changes as erosion, development of permanent vegetation, and underground water exploitation. Scattered small habitations are built everywhere during last thirty years, and in (Lamaj et al. 2015) persistent scatterers characterized by slow subsidence is observed in areas with concentration of habitations. Presence of negative trend of wrapped phase in Semani area seashore supports the hypothesis of subsidence phenomenon in this area, as one of causes for sea transgression and loss of sandy beaches (Fig.9b). In mountainous areas there are no significant scatterers characterized by high coherence values, oscillations observed in wrapped phases are hypothesized to be due to seasonal variations of vegetation and slopes erosion. In mountainous area Renci – Kakariqi and Torovica valley in between there are few scatterers with high coherence values are obtained, and oscillations observed in wrapped phases are hypothesized to be due to seasonal variations of vegetation and probably slopes erosion. Missing of fringes in the valley does not favor the subsidence hypothesis.
Authors: Frasheri, NekiRising groundwater in abandoned mines may result in ground movements at the surface overlying underground works. Feedbacks on several abandoned coalfields (e.g. France, Belgium and UK) show that significant uplift can occur, reaching in some cases up to several tens of centimetres. In the framework of its missions, the Post-mining Department of the French Geological Survey (BRGM) has the mandate of surveying French abandoned coalfields for potential ground movements. The most commonly used method for this mission is levelling, performed at a given frequency. Mine water recovery is a process that can last several tens of years. With the goal to optimize the long-term survey, the levelling technique has been re-examined by considering: 1) the phenomenology and kinetics of ground movements expected, in relation to water recovery rates; 2) alternative geodetic technologies, specifically spaceborne Interferometric Synthetic Aperture Radar (InSAR). An InSAR retro-analysis of ground movements, covering more than 20 years after mine activity ceasing, has thus been realized on a French abandoned coalfield aiming to: 1) extend ground movements detection capabilities in areas not covered by levelling, 2) compare InSAR analysis to levelling data to evaluate robustness of displacement measurements. This study was carried out using the entire archive of ERS, ENVISAT and Copernicus Sentinel-1 SAR data. The comparison of InSAR results with levelling measurements shows a significant correlation. The InSAR analysis was able to highlight ground movements of a few millimetres per year, with the same order of precision than classical levelling methods. The observed ground displacement patterns expand well beyond the coalfield, indicating the influence of non-mining-induced phenomena as well. It emphasizes also the importance to verify the stability of levelling benchmarks, as they might be affected by a combination of mining and non-mining induced displacements. These results point out the necessity of being cautious when interpreting the origin of damages caused by ground movements. Our findings offer perspectives for the operational use of satellite InSAR for long-term survey of ground movements on abandoned coalfields especially when the expected movements are relatively slow. The ability to cover wide area, the frequency of data availability, the accuracy of the measured displacement rates and the opportunity to perform retro-analysis are strong assets in post-mining monitoring context.
Authors: Morel, Jacques (1); Foumelis, Michael (2); Raucoules, Daniel (1); Lemal, Sandrine (1)Since the launch of the first Sentinel satellite, Sentinel-1A in April 2014 more than six years of Synthetic Aperture Radar (SAR) observations are now available and the Sentinel-1 tandem ensures a reliable, growing dataset for various applications in the future. Sentinel-1 is designed to be particularly effective in wide area persistent scatterer (PS) analysis for long-term stability and deformation analysis. In recent years, a few countrywide PSI motion maps have already been created by national agencies, large companies and consortia in European countries like Italy, the Netherlands and Norway. These projects prove that there is a great demand for deformation information from Sentinel-1 data and also show that the construction of a national PSI deformation map is feasible. Compared to these, our approach is not based on large-scale, semi-automatized, robust processing of the incoming data flow, but high resolution, precise, individual processing of relatively smaller regions. The persistent scatterer SAR data processing is entirely done by the Gamma software from Single Look Complex (SLC) data to displacement time series, using a single master approach. The dataset includes all Sentinel-1 observations since 2014, but some individual patches cover different timeframes from 2.5 to more than 5 years. Each processed block consists of several bursts of a single sub-swath. An important issue is the combination of these independently processed areas to have a relatively homogeneous, common referenced, consistent large-scale PSI database. We investigated a few approaches including the matching of overlapping regions as well as using a multi-technique (InSAR, GNSS, levelling) reference point network. Several different validation works have also been performed across the country. This work is also meant to be a demonstration that the technology is adequate to create a PSI velocity map of the whole country with very modest resources. Scaling it up in terms of automatic processing algorithms, improved data management, providing higher-level products and better dissemination are some of the next technical steps and we also plan to complement PSI method to have even better coverage outside urban areas, better homogenization of the dataset and regular reprocessing of selected areas. Hungary has a total area of 93 030 km2 and except a few mountain ranges it is mostly characterized by large plains and its area is dominantly covered by soft sediment, its topography is gentle and the tectonic activity is moderate. Thus, our results are generally reflecting ground motions associated with anthropogenic activities, related to mining, oil, gas, and water exploitation and civil engineering, etc.
Authors: Farkas, Péter; Frey, Sándor; Grenerczy, GyulaThe rise in sea level is expected to considerably aggravate the impact of coastal hazards caused in the coming years. Potential impacts are expected to be significant, for example the increase in the frequency of floods, the acceleration of coastal erosion, as well as the imminent slump of coastal areas from salty waters. The impact depends not only on the intensity and extent of coastal changes but also on human response and adaptability to hazards, including the type of exposure. Coastal areas with very low altitudes are particularly vulnerable to the Relative Sea Level change. In particular, for coastal urban centers the risk is increased due to high exposure. Coastal protected areas are also of great interest due to the high conservation and protection of the special characteristics of the natural environment. Land settlement on a local scale can make a significant contribution to the relative sea level rise, so it is necessary to understand the rhythm and spatial distribution of potential surface deformation in coastal areas. The present study aims at the combination of the Multi-Temporal Interferometric Synthetic Aperture Radar methodology exploiting Copernicus Sentinel-1 data and sea surface height measurements from altimetry data in coastal urban and special protected areas in Greece. GNSS data are used to anchor the vertical velocities to a common reference frame. The contribution of this research to scientific knowledge is primarily the determination of the balance between land deformation (from 2014 till today) and the sea level change to extract the relative sea level change in three pilot cases: the city of Alexandroupolis and the Evros Delta, the coastal zone of Thermaikos Gulf and the coastal area of Killini – Araxos (Patras Gulf) in NW Peloponnese which are areas with special characteristics. In these three case studies the relative Sea Level rise is varied between 2.5cm and 30cm in 50 years. Considering the availability of ground motion services the scaling–up of relative Sea Level rise products in coastal areas in continental and progressively in global scale is mandatory for the design of mitigation measures and adaptation especially in the most exposed regions as the North Africa. The weak points that affect the precision and the scaling up procedures should be subjected to further improvement are considered.
Authors: Elias, Panagiotis (1); Benekos, George (2); Perrou, Theodora (1,2); Fylaktos, Minos (2); Bantouvaki, Konstantina (2); Parcharidis, Isaak (2)Currently, the use of Satellite Synthetic Aperture Radar Interferometry (InSAR) technology for ground surface displacements detection knows a fast development within the remote sensing field. Indeed, in the last years, the monitoring techniques of satellite borne SAR, such as Small BAseline Subset (SBAS), have become wider spread, thanks to great accessibility of SAR data and of new missions, for research purposes and commercial projects. The SBAS technique is based on a proper combination of unwrapped interferograms characterized by a small spatial separation (perpendicular baseline) between the acquisition orbits and a short time interval (temporal baseline) between the acquisition epochs. This approach permits to generate mean deformation velocity maps of the investigated area as well as to detect the temporal evolution of the ongoing displacements. The present paper deals with the use of Sentinel-1A satellite radar images for deformation monitoring of Dam region. In this frame, a collaboration project of geodetic auscultation of Kramis’s Dam (State of Mostaganem, North West of Algeria) using GNSS and InSAR techniques was launched by Centre des Techniques Spatiales (CTS, Arzew) and Agence Nationale des Barrages et Transferts (ANBT, Algiers). It aims to assess Kramis’s Dam deformations zones (settlement, displacement and landslide phenomena) using GNSS and InSAR techniques. The interferometric processing of 20 Sentinel-1A radar images, ascending pass of Track30, was performed by P-SBAS (CNR-IREA, Italy) software under GEP (Geohazard theamtic Exploration Platform). The preliminary results concern the mean deformation velocity map and time series of vertical displacements, over more than one year period (August 2018 - October 2019). The mean velocity of deformation is around 1cm/yr, for dam region. The displacements of the studied points, on dike of the dam and on upstream reservoir side, have two trends: slight and strong. The first one concerns the period of September 2018 until June 2019, of about of -2.0 mm to 7.5 mm, and second one is for the period of June to early October 2019 (summer period), of about of 7.5 mm to 15.0 mm. The increase of displacements is probably due to water diminution of dam reservoir causing by rainfall decreasing and irrigation use. Indeed, the water level has lost of about of 3m from July to September 2019. Therefore the hydrostatic pressure, on the dike and on the sides of reservoir, decreases causing swelling of these zones as consequence of elasticity effect on the dam region.
Authors: Gourine, Bachir; Hasni, KamelLandslides are a major threat in all mountainous regions throughout the world, causing huge losses of life and property every year. The measurement of landslide surface deformation can facilitate prevention and mitigation of landslide disasters. The potential landslides can be detected through capturing the pre-failure deformation. The short-term emergency monitoring can be carried out to ensure the rescue security after the landslide collapse using ground based InSAR. The satellite InSAR is more suitable for long-term post-failure landslide risk assessment. This study takes the Xinmo landslide as an example to demonstrate the pre- and post- failure landslide analysis using both satellite and ground based InSAR. The Xinmo landslide happened in the early morning of 24 June 2017 at about 5:38 am local time, and this catastrophic event caused enormous casualties in Xinmo Village, Mao County, Sichuan Province in China (Fig. 1). We use the Sentinel-1 data to capture the pre-failure movement. Both the ascending and descending Sentinel-1 were processed using StaMPS/SBAS. The sliding source area can be identified from their deformation rate map (Fig. 2). The accelerated movement of the source area just before the collapse was measured by both the ascending and descending Sentinel-1 data (Fig. 3). The IBIS FL system was used to assess the post-failure stability of the Xinmo landslide, starting at 20:45:41 in 29 June and ending at 08:52:10 in 05 July. We obtained the averaged linear deformation rate (Fig. 4) from the 927 scenes for the Xinmo landslide. The averaged linear deformation rate map presents two unstable areas moving toward the GBSAR instrument. One is the debris mass on the west flank of the landslide. The whole east side moved with maximum linear deformation rate larger than 40 mm/d. Another is located at the deposit area. The spatial pattern is slender along the elevation direction, indicating the broken rocks slipping along the slope. The post-failure ascending and descending Sentinel-1 datasets acquired from June 2017 to December 2019 were processed using StaMPS/SBAS method. The ascending (Fig. 5(a)) and descending (Fig. 5(b)) orbits present similar deformation distribution. The north, south, and east sides of the debris mass on the west flank are unstable after the landsldie failure as indicated by the Sentinel-1 results. The maximum LOS deformation rate reached -80 mm/yr. In addition, a large number of rocks and gravels loosely stacked at the bottom of the landslide, i.e. the deposit area. These loose deposits are compacted under the force of gravety. The compaction process behaved as subsidence, which can be captured by the Sentinel-1 satellites. The Sentinel-1 results reveal that the subsidence mainly happenedd in the east side of the deposit area (Fig. 5). Note: The figures are included in the attached PDF file.
Authors: Dong, Jie; Liao, Mingsheng; Zhang, LuLandslide early warning remains a grand challenge due to the high human cost of catastrophic landslides globally and the difficulty of identifying a diverse range of landslide triggering factors. There have been only a very limited number of success stories to date. However, recent advances in earth observation (EO) from ground, aircraft and space have dramatically improved our ability to detect and monitor active landslides and a growing body of geotechnical theory suggests that pre-failure behavior can provide clues to the location and timing of impending catastrophic failures. On June 24th 2017, a landslide of 13 million m3 suddenly buried the Xinmo village, Sichuan province, China, causing 10 deaths and 73 missing. Interferometric Synthetic Aperture Radar (InSAR) analysis with ESA’s Sentinel-1 imagery suggests that satellite radar observations cannot only assist with mapping the landslide source area and boundary, but also be used to identify landslide precursors. Another successful case on early warning was in Heifangtai landslide, China. The early warning was released 8 hours before the landslide, prompting a government led emergency response and evacuation. This case clearly demonstrates the potential importance of real-time displacement measurements and the role that in-situ sensors could play in early warning systems. These two recent landslides in China as case studies, to demonstrate that (i) satellite radar observations can be used to detect deformation precursors to catastrophic landslide occurrence, and (ii) early warning can be achieved with real-time in-situ observations. Bearing in mind the complementary features of InSAR and GNSS in terms of deformation monitoring, we discuss the feasibility of earth observation (EO) for landslide early warning and a novel and exciting framework is then proposed to employ EO technologies to build an operational landslide early warning system (EWS). The proposed EWS enables to early detect potential landslides with a wide coverage and gives real-time warning to improve preparedness and resilience of local communities, suggesting that, nowadays, EO provides a unique opportunity for landslides EWS. In the final, the outlook on the three big questions, (Q1) where are potential landslides, (Q2) when will landslides occur, and (Q3) how to best reduce landslide disaster risk, were discussed and presented.
Authors: Dai, Keren (1); Li, Zhenhong (2,3); Xu, Qiang (1)Knowing the devastating aftermath of deep-seated landslide (DSL) back to 2009 during Typhoon Morakot, the authorities of Taiwan have invested enormous resources to enhance the country’s disaster prevention and response capabilities. Currently, LiDAR derived DEM and field survey are used to identify and interpret the failure mechanism and susceptibility level of DSL, whereas ground surface deformation systems are adopted to formulate the country’s disaster prevention and alert policies. To monitor the surface deformation of deep-seated landslide, the Taiwanese government has targeted the deep-seated landslides located in Lantai area of Ilan to launch a unique landslide monitoring project integrating multidisciplinary geophysics experiments, including UAV-LiDAR, single-frequency GPS, drilling, electric resistivity survey, Multi-temporal InSAR (MTInSAR) and fiber optics sensors, in 2014. This experimental multi-temporal monitoring project, which is based on the framework of MT-InSAR proposed by Zhang et al. (2014), can estimate the average long-term ground deformation rate of deep-seated landslide by optimizing the number of temporarily coherent points (TCPs) in the interferometric ALOS (2007-2011), ALOS2 (2014-2018) and Sentinel (2014-2018) data. Also, as single-frequency L1 GPS provides high-precision positioning, 12 single-frequency GPS installed in potential moving blocks were selected to validate the InSAR data and to analyze the kinematic behavior of Lantai area. In the system, all of received data, which are processed into the standard receiver independent exchange format (RINEX), are transferred to a data processing server via the 4G wireless network. The positioning results, which can be used as future reference for system configuration and installation, are stored in an online database and displayed via a webpage. Therefore, the adoption of several survey techniques, time-series observation on ground surface deformation, history of collapses and single-frequency GPS displacement data can help us to identify the failure mechanism of deep-seated landslides and further locate the failure planes and sub-sliding areas. Our study shows the efficiency and economics of each monitoring technique, on the other hand, can be evaluated by investment comparisons and data collection quality. For large scaled landslides in different geologic and geomorphologic terrains, different combinations of monitoring techniques will be proposed.
Authors: Chen, Rou-Fei (1); Liang, Hong-Yu (2); Hsu, Ya-Ju (3); Wang, Kuo-Lung (4); Yin, Hsiao-Yuan (5); Feng, Mei-Chen (5); Zhang, Lei (2); Lin, Ching-Weei (6)Synthetic aperture radar (SAR) is a coherent active microwave imaging method. In remote sensing it is used for mapping the scattering properties of the Earth's surface in the respective wavelength domain. Many physical and geometric parameters of the imaged scene contribute to the grey value of a SAR image pixel. Scene inversion suffers from this high ambiguity and requires SAR data taken at : different wavelength, polarization, time and incidence angle. Interferometric SAR (InSAR) exploits the phase differences of at least two complex-valued SAR images acquired from different orbit positions and/or at different times. The information derived from these interferometric data sets can be used to measure several geophysical quantities, such as topography, deformations (volcanoes, earthquakes, glacier flows, ocean currents, vegetation properties..etc). Our study tends to demonstrate the effectiveness of radar interferometry to detect different deformation modes. The goal of this study is to employ this powerful technique to measure the evolution displacements related to different sources.
Authors: Chaib, Idir (1); Aguemoune, Samir (2)Triggered landslides disrupt the emergency response effort following an earthquake, and in addition present a significant secondary hazard in their own right. Information on their spatial distribution is required within days of an earthquake by emergency response coordinators but is often delayed or left incomplete due to cloud cover in optical satellite imagery. Currently, in the absence of optical satellite imagery, the spatial distribution of triggered landslides is modelled using multi-factor regression analysis of datasets such as digital topography and predicted or measured ground accelerations. This allows a predicted landslide density surface to be generated within hours of an event, although its accuracy improves if the surface is generated later using higher quality ground shaking data. Satellite radar has the potential to detect landslides even when optical imagery cannot, as it can be acquired through cloud and is available globally within days of any earthquake on land. Identifying landslides in radar data, however, is challenging. To-date, methods based on radar coherence have been successfully applied to urban damage detection, but with more mixed success to landslide detection. Here, we examine how InSAR coherence data can be used for rapid earthquake-triggered landslide detection. We do so using Sentinel-1 and ALOS-2 imagery, covering the 2015 Gorkha, Nepal; 2018 Hokkaido, Japan and 2018 Lombok, Indonesia earthquakes. First, we develop and test new methods of landslide detection using radar coherence. We initially test existing methods, but find them unreliable due to background spatial and temporal variability of coherence. We therefore propose two new approaches, using additional pre- and post- seismic imagery to mitigate this effect. In the first, we calculate the difference between a traditional 'boxcar' coherence estimated from small groups of adjacent pixels and a 'sibling' coherence estimated from small groups of pixels selected from within a wider window. In the second, we compare pre-, co- and post-seismic coherence estimates, as landslides are expected to initially decrease and then increase in coherence. Our new methods show significant potential for rapid post-earthquake landslide mapping, and improve on existing coherence methods. For example, in Nepal, our boxcar-sibling method outperforms existing methods by up to 13% when using Sentinel-1 data. We find that performance of the different classification methods depends on: 1) the desired resolution of the landslide information; 2) the radar wavelength; and 3) the landslide properties (size, shape and spatial distribution). Finally, we make recommendations for which method is most suitable for different regions and different scenarios. Second, we combine new and existing radar coherence methods with the existing empirical regression-based framework, using Random Forest regression to generate fast and accurate post-disaster landslide information, even under cloudy weather conditions. We aim to test whether the incorporation of radar data improves modelled landslide density, and if so, which data and derived classification lead to the largest improvement for different events. We test both: 1) locally calibrated empirical models, trained on a small area of each event and used to predict the landslide distribution in the rest of the study area; 2) a preliminary “global” model, trained on two of our case study events and used to predict the third. For both model types, we begin with a model based on topography, ground shaking estimates, lithology and land cover. We then progressively add Sentinel-1 and ALOS-2 radar data in the chronological order it became available following each earthquake, and test scenarios in which only Sentinel-1, only ALOS-2 and in which data from both satellite sensors are available. We find that incorporating radar data into the models significantly improves predictive skill. We see the greatest improvement upon incorporating the first ALOS-2 image and find the best model performance when both ALOS-2 and Sentinel-1 are used. We find that radar data improves model performance: 1) over timescales relevant for emergency-response, (i.e. within 2 weeks of each earthquake); and 2) such that it has a consistently high degree of skill (increasing ROC AUC by up to 0.2 and resulting in ROC AUC > 0.8 at all sites for global models). Overall, we find that satellite radar data has major potential for use in rapid post-disaster mapping of landslides, both as an independent technique and as a supplement to better constrain existing empirical models.
Authors: Burrows, Katy (1); Walters, Richard (2); Milledge, David (3); Bellugi, Dino (4); Densmore, Alexander (2)The interferometric analysis of multi-temporal radar imagery is an important source of information for the detection and monitoring of ground subsidence in the urban areas. The advent of the Sentinel-1 Synthetic Aperture Radar (SAR) imagery from the European Space Agency (ESA) has opened new opportunities to the remote sensing community for the detection and monitoring of the ground subsidence hazard in urban areas. In this contribution are shown results of the application of Sentinel-1 radar imagery to ground subsidence detection and monitoring in the cities of Tirana and Durrës in central Albania. The multi-temporal Sentinel-1A IW interferometric stack of fifty-eight Sentinel-1 images in ascending orbit of the time period January 2017 – December 2018 was analyzed using the Persistent Scatterer Interferometry (PSI) technique (Ferreti et al., 2001). Tirana is the largest city and the capital of Albania. The city of Tirana has had a considerable urban expansion in the last thirty years. Based on the analysis of Sentinel-1A radar imagery of 2017-2018 the city of Tirana is relatively stable for what regards the ground subsidence hazard. However, The PSI analysis of the 2017-2018 Sentinel-1A for the city of Tirana detected a ground subsidence zone with a vertical displacement velocity of about -7 mm/year at Yzberish, along a segment of the Tirana Outer Ring road. This ground subsidence zone is also noted by PSI analysis of multi-temporal COSMO-SkyMed X-band radar data of the years 2011-2014 (Wasowski et al., 2015). At Yzberish a rapid urbanization with the construction of a number of tall buildings has taken place in the last twenty years. In addition, a relatively sparsely populated area in the northwestern part of the study area also shows ground subsidence of about -7 mm/year. The cause of the subsidence is the lowering of the water table and prolonged ground settlement processes due to particular geology of the area. The study represents the first reported analysis of the Sentinel-1 data for ground motion detection in the city of Tirana, and can serve as a basis for further monitoring of the ground subsidence hazard in this urban area. The coastal city of Durrës is an important economic, social, cultural, and historical center of Albania. In the last twenty years the city of Durrës has experienced rapid urban expansion. A large part of this urban expansion is in the form of urban sprawl and informal settlements. In addition to the increased density of the built-up area in the city, large areas of agricultural lands have been urbanized. Especially notable is the case of the informal settlement constructions in the reclaimed lands of Durrës marsh in the last twenty years. The marsh of Durrës was completely reclaimed by works carried out during 1962-1967. The reclaimed lands are very susceptible to ground settlement due to soil compaction. The results show a considerable zone of ground subsidence with values of up to -30 mm/year that occur in the reclaimed lands of the Durrës marsh. Analysis of descending orbit multi-temporal Sentinel-1 imagery of the same period 2017-2018 produced similar results. References Ferretti, A., Prati, C., Rocca, F. (2001). Permanent scatterers in SAR interferometry. IEEE Transactions in Geosciences and Remote Sensing, 39, 8–20. Wasowski J., Bovenga F., Nutricato R., Nitti D.O., Chiaradia M.T., Kucaj, S., Strati B. (2015). High resolution satellite multi-temporal interferometry for detecting and monitoring landslide and subsidence hazards. 10 p. Proceedings 10th Asian Regional Conference of IAEG, Kyoto, 26-27 Sept., 2015.
Authors: Bedini, Enton (1,2)It is extremely crucial to identify and monitor landslides in the urban areas, which results in a significant number of disasters on land having high topographic variability. Landslides monitoring also requires comparing different sources of parameters (areal width, movement speed, surface topography, and precipitation amount or soil moisture) with each other to determine the current state of the landslide and its time evolution. In this study, it was aimed to determine the landslides occurring in Beylikdüzü and Esenyurt region in Istanbul with the help of interferometric synthetic aperture radar (InSAR) technique and to compare the results with Global Navigation Satellite System (GNSS) measurements. In the study, Sentinel-1 satellite sensor and periodically performed GNSS measurements were used, and the results were compared. The study area is prone to multi-hazards affecting millions of resident. The region area is affected by two major hazard types that have been seen frequently are earthquakes and slow-moving active landslides. In the study, landslide displacement in Istanbul is evaluated with time series analysis using a Persistent Scatterer Interferometry (PSI) method. We used multi-temporal C-band (5.6 cm wavelength) new generation single look complex (SLC) data from ESA in single polarization (VV). Landslides were monitored with the PSI technique for the whole peninsula between Büyükçekmece and Küçükçekmece lakes and, the results were evaluated. The process of determining the deformation movements was carried out with the images acquired in both ascending (140 images) and descending (131 images) orbits. The Sentinel-1 images acquired in a period of approximately 5 years between January 2015 and December 2019 for both orbits. In order to determine the landslide behaviour, PSI technique was used due to the characteristic of the areas where building density is high. As expected, the Permanent Scatter (PS) points achieved in the analyses executed with Sentinel-1 images were mostly obtained in urban areas on the peninsula. Pre-processing steps include the TOPSAR split, apply orbit file, back geocoding, TOPSAR deburst, interferogram formation, topographic correction using SRTM 1 sec HGT data and StaMPS export, in which all processing was performed with open source tools of Sentinel Application Platform (SNAP) software. After the pre-processing steps the processes in turn of phase noise estimation, PS selection, PS weeding, phase correction, unwrapping of interferograms were performed with open-source StaMPS (Stanford Method for Persistent Scatterers) software. The amount of deformation was obtained in line of sight (LOS) direction for both orbits. To compare these results with GNSS measurements, it should be obtained in the north-south, east-west (horizontal) and up-down (vertical) direction. The proportion of displacement amounts in these three directions can be obtained by using ascending and descending SAR images. In this study, a 2D approach is conducted due to the polar orbit movement of the satellite, since the displacements determined in the north-south direction cannot be obtained precisely enough. When we investigate the vertical displacement, it is seen that the displacement in this direction is smaller than the displacement on the horizontal direction. The highest vertical displacement was determined between -9 mm/yr and 6 mm/yr across the work area. Similar to horizontal displacements, there is displacement between -10 mm/yr and 10 mm/yr. In addition, GNSS measurements were carried out to determine the surface displacements and to validate the PSI results, with 3 survey campaigns at 12 GNSS sites previously determined in the study region. These measurements were carried out in April, July and October 2018, respectively. GNSS data were collected with sufficient accuracy (≥5 hours), 10° elevation angle and 30 seconds interval by using dual-frequency GNSS receivers. When GNSS and InSAR results are compared, it is observed that some points move in a similar direction while the others show a various directions. The most likely reason for this is that in the regions where GNSS points are present, points cannot be obtained in InSAR results. It is also thought that the PS and GNSS points are not in the same position and that landslides due to local movements may occur. Since the landslide takes place due to effective parameters such as geological formation, precipitation and slope, it shows different behaviours in local areas. To fulfil the main purposes of this study, landslides were determined and monitored by InSAR, and the determined results were demonstrated the capabilities of multi-temporal InSAR techniques. According to the results of this study, it is necessary to be prepared for natural disasters that may occur in Beylikdüzü and Esenyurt districts of Istanbul, which is constantly growing and has become an area of interest for people with its current industrial development. For this purpose, it is necessary to constantly and effectively monitor the behaviour of the landslide, which is one of the natural disasters that can cause great damage. Moreover it is important not only to identify landslides, but also to investigate their geological and geomorphological aspects in terms of preventing disasters.
Authors: Bayik, Caglar (1); Abdikan, Saygin (2); Ozdemir, Alpay (3); Arikan, Mahmut (4); Balik Sanli, Fusun (3); Dogan, Ugur (3)Groundwater in volcanic islands is usually the main source of freshwater, and it is essential for sustainable development. In Tenerife Island, groundwater extraction occurs by drilling horizontal water tunnels, called water galleries, as well as numerous coastal wells. Since around 1900, but especially since the 1960s, hundreds of water tunnels have been drilled for agriculture and freshwater supply. This has resulted in a sustained extraction of groundwater larger than the natural recharge, leading to a general water table decline, locally up to 200 m of down drop. Since 2000, satellite radar interferometry (InSAR) applied to measure surface deformation has located several subsidence bowls (e.g., Fernandez et al., 2009). The localized surface deformation patterns have been correlated with water table changes and hence aquifer compaction. However, no further investigations have been carried out to confirm which characteristics (chemical composition, texture, porous network, alterations, etc.) of the volcanic materials can control compaction process, and to which extent porous volcanic units, the most abundant material in Tenerife, can compact to explain the observed surface deformation. This lack of knowledge might affect the effectiveness of water management policies. To investigate the compaction processes affecting the volcanic aquifer, we propose to set up a passive hydrogeophysical monitoring network composed of geodetic and seismological instruments. However, considering logistic constrains it is desirable to have as low as possible number of observation sites, whist maximizing the detection and characterization of the aquifer dynamics. Here, we explore different network configurations to maximize the spatial and temporal characterization of the compaction processes using machine learning methods (low-rank matrix techniques). We pose the network design as an optimization process with the aim to parsimoniously have as fewer as possible ground station sites, and have a low error on reconstructing spatiotemporal land subsidence observations. Land subsidence rates were estimated using Sentinel-1 radar interferometric observations from October 2014 to December 2020. This method allows for an optimal network configuration, with respect to the dual penalty function, which facilitate the decision making. Nevertheless, this type of network design should be regarded as proposals because some station site conditions are a priori unknown. Although, one could modify the penalty function to optimize the network considering additional types of information, e.g., geological materials, groundwater table time series, etc. References: Fernandez, J., et al. (2009), Gravity-driven deformation of Tenerife measured by InSAR time series analysis, Geophys. Res. Lett., 36, L04306, doi:10.1029/2008GL036920.
Authors: Gonzalez, Pablo J. (1,2); Charco, Maria (3); Eff-Darwich, Antonio (4); Lamur, Anthony (2); Marrero, Rayco (5); De Angelis, Silvio (2)Estimation of surface displacement is important for understanding groundwater activity, exploitation of natural resources and monitoring of strain accumulation. SAR interferometry is one of the most attractive ways to estimate the surface deformation. In 2014, Sentinel-1 launched by ESA and it has IWS mode which can monitor wide area, about 250 km swath with high spatial resolution. Sentinel-1 is superior to the former satellites so it can estimate surface deformation wider and more frequently.In our study, we focus on Taiwan island. In Taiwan, the main cause of surface deformation has been reported as plate tectonics and massive groundwater pumping. Taiwan is one of the most active tectonic regions in the world, and earthquakes frequently occur. And recently, groundwater pumping has been a growing concern in western Taiwan. It causes land subsidence due to dropping groundwater level. In our study, we revealed the recent deformation in Taiwan island using SAR data from Sentinel-1 with IWS mode, and divided two main causes; long-term and seasonal deformation. And then, compared with previous researches, we evaluated the effectiveness of our results.In this study, we processed 62 SAR data of ascending orbit from Sentinel-1 between October 2017 and October 2019. We used 115 interferograms from InSAR analysis to generate time-series displacement based on the SBAS method. Baseline of all interferograms was less than 200 m. Then, using time series surface deformation, we estimated short-term seasonal deformation and long-term deformation. Specifically, we applied a least-square fitting of sinusoidal and linear function to time-series displacements. In the result of time-series analysis, over 3 cm/year displacement was estimated at areas in the western Taiwan and the eastern Taiwan. Seasonal result revealed that seasonal deformation was significant in the western, whereas no seasonality was found in the eastern. According to Huang et al. (2016), the region faced the anthropogenic groundwater pumping. And in the eastern area, there were few effects of groundwater activity, thus we interpreted tectonics activity causes the estimated deformation.For validation of these results, we compared some previous researches (Huang et al.,2016; Su et al.,2017) using time series analysis with our results. Both of them revealed the trend of deformation in western and northeast Taiwan is significantly similar. Then, we compared with previous research about seasonality (Huang et al., 2016), the trend in western Taiwan has also similarity. Our results showed the Sentinel-1 wide-swath SAR data is effective in the study of surface displacement.
Authors: Hiranaka, Shoki; Ishitsuka, Kazuya; Lin, WeirenWater level fluctuations in the largest inland water body, the Caspian Sea, cause a regular loading on the basin, which result in a regular land subsidence and uplift. This topic is not already studied, however, if it is significant, it will affect coastal based measurements and surveying. According to literature, water level fluctuations in the Caspian Sea include a seasonal cycle of about 35 cm, as well as, a variant interannual fluctuation. Vertical land movements of the Caspian Sea bottom can also be revealed beyond the shorelines. Using Sentinel-1A InSAR images, in this study we have investigated vertical land deformations in five selected regions around the Caspian Sea over the years 2019 and 2020, in order to examine the water level fluctuations effect on land deformations. Five regions of tens km2 area, located in four sides of the Caspian Sea are selected as study regions. In order to ensure receiving high effects from water loading, the study areas are specified close to the shorelines. For the same reason, four of the five regions are defined in peninsulas, where the land is maximally surrounded by the sea. For preventing the land cover and land use related noises, we choose low populated areas and as much as possible free from vegetation covers (Except the southern study region which includes inevitably vegetations). We also preferred to select an areas not having complicated topography. The study regions include: one region in Absheron Peninsula of Azerbaijan, two regions in Krasnovodsk Peninsula of Turkmenistan, one region in Mangyshlak Peninsula of Kazakhstan, and one region in the Rasht plain in the coasts of Iran (Fig.1). For all the study regions six single look complex (SLC) Sentinel-1A images on descending path and Interferometric Wide swath (IW) mode, span from 8 January 2019 to 13 July 2020 are provided. With assigning different Master-Slave combinations, 15 inteferograms are generated for each study region. After applying necessary process in SNAP software, maps of vertical land deformations are produced. Accordingly, based on the Master-Slave local times, atmospheric and ionospheric delays are calculated. The maps of total delays are converted to equivalent vertical displacement and subtracted from the InSAR derived vertical displacement in order to correct for atmospheric and ionospheric effects. We calculate dry and wet atmospheric delays from vertical temperature and humidity profiles provided from hourly ERA5 reanalysis. Using reanalysis of specific cloud, rain and snow water contents we also accounted for delays caused by liquid particles within could, rain and snow. The specific water contents are interpolated for the exact times of image acquisitions (Fig.2 and Fig. 3). They are also compared to half-hourly precipitation data from the Global Precipitation Model (GPM). Ionospheric delay is calculated by analyzing hourly IONEX data from the GNSS data center (Fig.4). For evaluating the results, we compared our estimations of total delay with GPS delay records located in the vicinity. We used GPS records of total zenith path delay (ZPD) obtained by GNSS data center. This GPS network provides very high resolution records (5 minutes interval), but it is limited to very sparsely distributed ground stations which are located relatively far from our study regions. Although, GPS delay data provide low resolution for application in InSAR derived measures; however, in the absence of other insitu measurements we benefited using GNSS ZPD records. The data from three GPS stations of Atyrau, Tehran and Aruch are compared to the total delay derived from REA-5 profiles, for the same time and locations. After corrections for atmospheric and ionospheric total delays, the displacement maps derived from InSAR consist of two sorts of displacements: A global diagonal displacement with different magnitudes (depending on the temporal baselines) is detected in most of displacement maps (Fig.5). The second sort of displacement is local displacements. Ignoring the residuals in a form of single pixels which are spatially filtered, one concentration of subsidence pixels (12 X 4 km) is detected in eastern part of the sea. This deformation is caused by an earthquake of 4.9 Richter degree magnitudes on 6 Jan 2020 (Fig.6). Having five study areas processed and 15 displacement maps produced for each study area, we obtained a 4-Dimension network consist of 75 displacement nodes. Based on interferogram mean coherence values, displacement maps produced from low coherent data are removed from the network (6 maps are removed). Applying a set of outlier filter and smoothing filter on the network resulted in detection of seasonal and interannual land displacement signals. The time of seasonal minimum and maximum are mostly convenient with those of water level fluctuations. The amplitude of seasonal and interannual land displacement, however, depends directly on the smoothing applied to the network. Thus, it needs to sophisticate carefully the filters and smoothing applied to the data before any result extension (we are currently doing the last analysis for the best filters to be selected). A Pearson correlation test between the deformation values in five study regions show that, high correlations exist between the displacement values derived from the study areas located on the peninsulas (more than 0.71); but, the Rasht plain in the south shows the lowest correlation with other samples (0.54 in average). At last, we compared between the land displacement estimated from InSAR and displacement expected from elastic models derived from altimetry time series. The water loading effect on land movement is complicated as it consists on interannual trend, annual cycle and possibly local movements. The integration of these components may induce heterogeneous land movements depending on land rheological structure, geological structure and distance to center of the load.
Authors: Moradi, Ayoubwater scarcity in the Middle East, and the high frequency of conflict that emerges over what few resources do exist, is well established. The recent drought that began in 2007 has further strained the limited water resources in the region. Iran is experiencing a serious water crisis. The government blames the current crisis on the changing climate, frequent droughts, and unauthorized well extraction. In this regard, to investigate the amount of subsidence in the area, we used InSAR method for identification unauthorized wells in the south of Tehran (capital of Iran). We used Sentinel1-A from 2014 to 2019. The analysis is carried out using 89(Descending), 84(Ascending) Sentinel1-A images from 1.1.2014 to 31.12.2019. Results of the multi temporal analysis of the ps point and SBAS show that 8 unauthorized wells observed in the region. The maximum rate of subsidence is 128 mm/y. Therefore, field work results confirmed the presence of unauthorized wells. unauthorized wells, subsidence, SBAS, PS, Sentinel1-A, South of Tehran
Authors: Keyani, MohammadaliA tenth of the world’s population lives in coastal zones (Evans, 2004), where human activity and rising global sea levels intensify the erosion and flooding. The predicted impact of climate change will only increase the risks. Understanding the coastal processes driving cliff instability can help coastal engineers establish how, when and where the cliff collapses will occur. The United Kingdom has a high population density, much of it concentrated around coastal zones and therefore impacted by coastal erosion. In just England and Wales, almost 28% of the coastal zone is eroding more than 100 m a-1 (Nicholls and Cazenave, 2010). Coastal cliff instability is one of the most intense forms of the coastal erosion and it is often considered stochastic in nature and very difficult to predict. Cliff instability, leading to coastal erosion can threaten national infrastructure, creating risks to pipelines, buried cables, beaches, roads and other property. In southern England, coastal defences exist, but need to be constantly managed to remediate the risk of the cliff collapses. A variety of methods are available for monitoring of the coastal zones, including laser scanning and instrumentation, as well as field work and laboratory testing. Unfortunately all of them require specialist equipment and, due to the harsh coastal environment, like salt damaging the equipment or measurements not being possible to be taken due to bad weather or tides, is often not very practical or cost effective. Space-borne Interferometric Synthetic Aperture Radar (InSAR) is increasingly used to monitor wide areas for local scale ground motions related to natural and anthropogenic geohazards, with up to millimetre accuracy. Since 2015, the European Space Agency (ESA) has launched two satellites, Sentinel-1A and Sentinel-1B, which scan the Earth surface every 6 days, providing frequent and detailed data. These satellites are an exceptional source of information that can be used for land motion monitoring, geological and topographic mapping to understand the evolution of the coastal cliff zone. InSAR data, which are freely available, with potential for millimetre scale accuracy gives a fantastic opportunity to assess and monitor dangers of coastal erosion and cliff collapses. The aim of this research is to understand the processes which drive coastal cliff instability and then to consider if and how these are being observed with a use of remote sensing data like InSAR. It is necessary to investigate how this data can be used to monitor processes affecting coastal zones and how can this be implemented for predicting coastal changes to support safety of the communities living in these areas. The project’s main focus is on the analysis of coastal landslides and their monitoring in close to real time. Data from Sentinel-1A and 1B are processed to produce PS (persistent scatterer) time series to allow analysis of the movement of the landslide areas over time. Two case studies are presented, both focused on coastal chalk in the UK, as it produces a great analogue for other rocks types, being a soft rock and prone to failure. The methodologies being developed during the time of the project are to be universal and applicable to any coastal zones all around the world, not limited to rapidly eroding and failing chalk cliffs, but useful for all rock types, be it resistant rocks taking centuries to fail or weak rocks taking several years. REFERENCES: EVANS, E. 2004. Foresight: future flooding: scientific summary: volume I: future risks and their drivers. NICHOLLS, R. J. & CAZENAVE, A. 2010. Sea-level rise and its impact on coastal zones. science, 328, 1517-1520.
Authors: Mider, Gosia (1); Lawrence, James (1); Mason, Philippa (1); Ghail, Richard (2)Residents of Kabul, Afghanistan have historically relied on groundwater delivered from unconfined aquifers. However, recent drought events and population growth have resulted in withdrawing groundwater beyond sustainable rates and water level decrease. Groundwater overexploitation might have induced various magnitudes of ground subsidence, however, the issue had been overlooked. In this study, we assessed the spatial and temporal evolution of ground deformation in Kabul and analysed the main governing processes. Deformation rates were extracted using InSAR SBAS time-series analysis of C-Band Sentinel-1 products from both ascending and descending orbits acquired from 2014 to 2019. Interferometric processing was implemented using GAMMA software. Following the unwrapping of the differential interferograms, the time-series analysis was performed using InSAR TS+AEM, which incorporates the Atmospheric Phase Screen (APS) estimation. In order to overcome the inherent limitation of InSAR- measurements being made in Line-of-Sight (LOS) direction only, multi-geometry data fusion was employed by combining two independent LOS measurements to estimate vertical and horizontal ground surface displacement field. InSAR analysis delineated four subsidence bowls with highly variable extent and deformation magnitudes in four aquifers present within the city boundaries, the largest of which is detected in the Upper Kabul basin area with the vertical component accounting for -53 mm/year and negligible horizontal motion. The InSAR-derived subsidence time-series results were interpreted in relation to the spatio-temporal variations of hydrological and geotechnical conditions of the study area to identify the triggering factors of subsidence. The trend observed in deformation time-series and groundwater level change shows a good agreement, suggesting that groundwater drawdown is largely a function of groundwater level change. Also, the susceptibility of the aquifers to consolidation is largely caused by the prevalence of fine-grained sediments in the subsurface. Differential consolidation is a function of variation in thickness and types of the lithology of the base sediment. In the context of climate change and increasing urban sprawl in Kabul, monitoring ground deformation using the satellite InSAR is an invaluable tool that provides new opportunities to inform not only subsidence hazard mitigation plans, but also approaches for sustainable management of groundwater resources and for improved planning of operations such as Managed Artificial Recharge (MAR) of aquifers.
Authors: Meldebekova, Gauhar (1); Chen, Yu (1); Li, Zhenhong (2,1)The evolution of slope instabilities towards catastrophic failure events is accompanied by the progressive increase of ground displacements. For this reason, accurate data of surface deformation in space and time is important for the analysis and the interpretation of the associated hazard and risk potential. Remote sensing techniques have demonstrated to be a valid complement to standard in-situ monitoring. In particular, Differential Synthetic Aperture Radar Interferometry (DInSAR) from satellite-based imagery as well as from ground-based platforms allowed great advances in the identification and monitoring of surface deformation processes. However, the results obtained with currently available DInSAR methods might be hindered by insufficient spatial or temporal resolutions and/or due to intrinsic limitations of the methods used. For example, satellite based DInSAR provides great advantages to cover large areas, identify and map landslide displacements, and to monitor their spatial and temporal evolution in periods ranging from days to years. Despite, when the scenario evolves towards a potential failure, revisit times in the order of several days are not sufficient anymore. Phase aliasing, decorrelation, geometrical distortions, and layover/shadowing effects typical of satellite acquisitions may additionally hamper the possibility of accurate interpretations. On the other end, terrestrial based radar systems provide higher sensitivity to very small movements and are generally used when the landslide scenario requires continuous monitoring. These devices can obtain representative information over the area of interest, however, their spatial coverage is often limited and, since they exploit Ku- or X-band wavelengths, may suffer when large surface displacements occur. For this reason, comprehensive analyses of surface displacements should imply the use of multiple interferometric radar datasets. Here we present the results of an integrated investigation performed in Brienz/Brinzauls, GR, Switzerland. There, an area of 3 km2 is affected by a large and complex compound landslide with surface velocities locally exceeding values of 1 m/year. The scenario poses high concerns to the community, first for the village directly affected by extensive damage to the buildings (some of them already evacuated for safety), and more in general for the transportation network lines that would be affected in case ofcatastrophic failure. We selected analyzed SAR data acquired from: (i) satellite platforms, Sentinel-1 (C-Band) and ALOS-2 (L-Band); (ii) ground-based GAMMA Portable Radar Interferometer (GPRI, Ku-Band) from different locations; (iii) Car-borne SAR L-Band imagery, acquired from similar locations as the GPRI. We explore the results of interferometric analyses considering different observation periods and weather conditions.
Authors: Manconi, Andrea (1,2); Caduff, Rafael (1); Strozzi, Tazio (1); Frey, Othmar (1,2); Werner, Charles (1); Wegmüller, Urs (1)In the frame of the EU funded GEOSES* project the monitoring of dangerous natural and anthropogenic geo-processes is being built up using space geodetic technologies and concentrating on the Hungary-Slovakia-Romania-Ukraine cross-border region. Additional objective of the project is the prevention and monitoring of natural hazards and emergency situations (e.g. landslides, sinkholes or river erosion) and integrate advanced techniques in new, coordinated and innovative ways in order to improve our understanding of land deformation and its impact on the environment. Besides the ground-based and airborne techniques involved in the project, our study focuses one of the fastest developing space-borne remote sensing technology, namely InSAR, which is an outstanding tool to perform large scale ground deformation observation and monitoring. To perform such monitoring, we utilized Sentinel-1 Level-1 SLC acquisitions since 2014 until 2021 over the indicated cross-border region. We have set up an automated processing chain of Sentinel-1 acquisitions to generate long-term ground deformation data based on the analysis of the respective time-series for the monitoring of ground displacements. The pre-processing part of the workflow includes the migration of the input data from the Alaska Satellite Facility (ASF), the integration of precise orbits from S1QC, the corresponding radiometric calibration and mosaicing of the TOPS mode data, as well as the geocoding of the geometrical reference. As next step, all slave acquisition have be co-registered to the geometrical reference using iterative intensity matching and spectral diversity methods, as well as subsequent deramping has been also performed. To retrieve deformation time series from co-registered SLCs stack, we have performed multi-reference Interferometric Point Target Analysis (IPTA) using singe-look and multi-look phases with GAMMA Software. The proposed pipeline also includes an automatic phase unwrapping error detection method, such aims to detect layers in the multi-reference stack which are significantly affected by unwrapping errors. The applied unwrapping error detection method implements gradient edge detection technique on magnitude of gradient of unwrapped differential interferometric phase. After identifying the unwrapping error affected layers, the unwrapping error can be fixed with iterative subsequent re-wrapping, filtering and spatial unwrapping of the selected layers. As part of the IPTA processing, both topographical and orbit-related phase component, as well as the atmospheric phase, height-dependent atmospheric phase and linear phase term supplemented with the deformation phase are modelled and refined through iterative steps. To retrieve recent deformations, SVD LSQ optimization has been utilized to transform the multi-reference stack to single-reference phase time-series such could be converted to LOS displacements within the IPTA processing chain. The derived results are interpreted in regional scale and through local examples in the introduced cross-border region as well. * Hungary-Slovakia-Romania-Ukraine (HU-SK-RO-UA) ENI Cross-border Cooperation Programme (2014-2020) “GeoSES” - Extension of the operational "Space Emergency System"
Authors: Magyar, Balint (1,2); Horvath, Roland (1); Toth, Sandor (1); Hajdu, Istvan (1); Kenyeres, Ambrus (1)With regard to the increasing frequency of extreme weather events, the risks natural hazards pose have never been as relevant as today. One of the world’s most widespread hazards are landslides, affecting mountainous regions all over the globe. As there is a variety of landslide categories each affecting different areas and triggered by different factors, this project will focus on sliding and resulting flowing events. With strong precipitation events as a common trigger, these landslides are directly affected by changes in the global weather systems. Providing reliable, fast, and accessible monitoring tools is key for minimizing or even preventing human losses and major damage to infrastructure. The increasing numbers of earth observation satellite missions and easier access to data for scientific purposes leads to an abundance of data ready for processing. While most of the current research aims at refining products based on one single sensor type, this project instead focuses on combining different sensor types using machine learning approaches in order to utilize all their advantages. To achieve a fast product, manual work must be kept as little as possible. We focus on products that can be downloaded by either the Alaska Satellite Facility or the Google Earth Engine which can be accessed by their APIs. On these two platforms active and passive remote sensing images, as well as map products about possible triggers and topographical data, are easily accessible. The project can be divided into four main processing steps. First, a susceptibility map is created based on the topographical data as well as first remote sensing products such as precipitation and landcover maps. The aim of this step is to reduce the area which has to be observed into smaller patches to lower the amount of computation power needed. Second, in each patch active and passive remote sensing methods1 are used for change detection. The poster will be focusing on the synthetic aperture radar (SAR) data. Here, we conduct a common differential interferometric SAR (DInSAR) approach2 and create an interferogram stack for acquisitions in both ascending as well as descending orbit. Regarding the acquisition geometry, we continue by determining pixelwise which orbit has the better visibility for our analysis. In the following step, we use all products (phase, coherence, and intensity). To create these products, we use the python package snappy from ESA, which gives us access to the processing function of the SNAP software. On the passive remote sensing side, we will use the Google Earth Engine (EE) to acquire Sentinel 2 and Landsat images. The EE offers a simple way of creating a change detection analysis3, which can be integrated into our analysis. Third, for each patch, a graph network based on the topography is created. This graph network is used to store all different data types in the nodes. As landslides are gravitationally induced movements, the edges are based on the flow direction. With the possibility to look at the connected nodes within a specific distance instead of neighboring pixels, more precise neighborhood operations can be calculated. Possible operations are the mean and the standard deviation of the remote sensing data stored in each node in or against the flow direction. With this approach, we try to reduce the necessary minimal width of an event to be detected. The fourth part is the training of classification algorithms4. This step is still in process. To train a machine learning algorithm we will make use of the open global landslide data from NASA as well as various open access landslide inventories (e.g. from GNS) as ground truth data. The entries will be preprocessed and classified into different groups to enhance the training. The poster will give an overview of the whole framework (flow-chart), with the main focus on the DInSAR processing and the effect of the graph network on the neighborhood operations. This will be shown for a test area, which will later be used in the training process. References G. Metternicht, L. Hurni, R. Gogu, Remote sensing of landslides: An analysis of the potential contribution to geo-spatial systems for hazard assessment in mountainous environments. Remote Sensing of Environment. 98, 284–303 (2005). Barra et al., First insights on the potential of Sentinel-1 for landslides detection. Geomatics, Natural Hazards and Risk. 7, 1874–1883 (2016). J. V. Fayne, A. Ahamed, J. Roberts-Pierel, A. C. Rumsey, D. Kirschbaum, Automated satellite-based landslide identification product for Nepal. Earth Interactions. 23, 1–21 (2019). D. J. Lary, A. H. Alavi, A. H. Gandomi, A. L. Walker, Machine learning in geosciences and remote sensing. Geoscience Frontiers. 7, 3–10 (2016).
Authors: Luck, Manuel Andreas (1); Hajnsek, Irena (1,2)Multi-temporal SAR interferometry (InSAR) estimates the displacement time series of coherent radar scatterers. Current InSAR processing approaches often assume the same deformation model for all scatterers within the area of interest. However, this assumption is often wrong, and time series need to be approached individually [1], [2]. Individual, point-wise approach for large InSAR datasets is limited by high computational demands. The additional problem is imposed by the presence of outliers and phase unwrapping errors, which directly affect the estimation quality. This work describes the algorithm for (i) estimating and selecting the best displacement model for individual point time series and (ii) detecting outlying measurements in the time series. The InSAR measurement quality of individual scatterers varies, which affects the estimation methods. Therefore, our approach uses a priori variances obtained by the variance components estimation within geodetic InSAR processing. We present two different approaches for outlier detection and correction in the InSAR displacement time series. The first approach uses the conventional statistical methods for individual point-wise outlier detection, such as median absolute deviation confidence intervals around the displacement model. The second approach uses machine learning principles to cluster points based on their displacement behavior as well as the temporal occurrence of outliers. Using clusters instead of individual points allows more efficient analysis of average time series per cluster and consequent cluster-wise outlier detection, correction, and time-series filtering. We show the results of our approach on the Sentinel-1 InSAR time series of a case study from Slovakia. The area of interest is affected by landslides and undermining subsidence, with a specific, non-linear character of the movement. Our post-processing procedure parameterized the displacement time series despite the presence of a non-linear motion, thus enabling reliable outlier detection and unwrapping error correction. The accuracy of the obtained results was validated using corner reflectors. [1] Ling Chang and Ramon F. Hanssen, „A Probabilistic Approach for InSAR Time-Series Postprocessing“, IEEE Trans. Geosci. Remote Sens., vol. 54, no. 1, Jan. 2016. [2] Bas van de Kerkhof, Victor Pankratius, Ling Chang, Rob van Swol and Ramon F. Hanssen, „Individual Scatterer Model Learning for Satellite Interferometry“, IEEE Trans. Geosci. Remote Sens., vol. 58, no. 2, Feb. 2020.
Authors: Kubica, Lukas; Czikhardt, Richard; Papco, JurajPersistent Scatterer interferometry (PSI) is a group of advanced interferometric SAR techniques used to measure and monitor terrain deformation. Sentinel-1 has improved the data acquisition throughout and, compared to previous sensors, increased considerably the DInSAR and PSI deformation monitoring potential. Two of the critical problems with InSAR have been the effect of the refractive atmosphere on the interferometric phase and the phase unwrapping ambiguity. The low density of PS in non-urban areas have inspired the development of alternative approaches and refinement of the PSIG chain. Along with the efforts to develop methods to mitigate the above mentioned problems, this paper also investigates the presence of a new signal in multilooked interferograms which cannot be explained by noise, atmospheric or earth surface topography changes. This signal is very prominent in the short temporal interferograms and is contributing significantly to the bias in estimation of Earth deformations. The paper uses two data driven procedures to obtain a terrain deformation map with a set of selected points (Persistent/Dispersed Scatterer (PS/DS)) containing the information of the estimated annual velocity and the accumulated deformation at each image acquisition time, with minimized errors due to the effects of atmosphere and phase unwrapping and working exceptionally well in small areas. The general methodology estimates the linear and nonlinear components of the displacement, the error of the digital elevation model (DEM) used to cancel the topographic terms, and the atmospheric artefacts from a network of interferograms of small temporal baseline. Followed by phase unwrapping, where we are providing an automatic solution for detecting and correcting unwrapping errors by unwrapping spatially first and then temporally, which is called 2+1D phase unwrapping. Then comes the atmospheric phase removal using a low pass filter and derivation of the accumulated phase. In the last step, the final deformation and velocity maps are derived from the accumulated phase. The approaches discussed in this paper has the advantage of having a higher density of points, also is direct fast and flexible. The paper also attempts to study the characteristics of the incomprehensible signal pattern in the short temporal interferograms. The effectiveness of the approaches is illustrated through application on different test sites.
Authors: Krishnakumar, Vrinda; Qiu, Zhiwei; Monserrat, Oriol; Crosetto, MicheleWith the advances of ESA’s Sentinel-1 InSAR (Interferometric Synthetic Aperture Radar) mission there are freely available remote sensing ground deformation observations all over the globe that allow for continuous monitoring of natural hazards and structural instabilities. The Digital Environment initiative in the UK aims to include these remote sensing data in the effort of forecasting and mitigating hazards across the UK. However, analyses of low coherence areas (e.g. forested and vegetated areas) with conventional InSAR methodologies are difficult to perform due to the limiting factor of temporal decorrelation. Even the application of the permanent scatterer (PS) technique may not be successful when there is a low density of stable radar targets. Using artificial reflectors with high radar cross section (RCS) can be a way to overcome this limitation and achieve measurements with a good signal-to-clutter ratio (SCR). In order to be able to include Sentinel-1 data in the UK’s Digital Environment it is important to understand the advantages and limitations of these observations and interpret them appropriately. The Hollin Hill landslide observatory in North Yorkshire is used by the British Geological Survey in their efforts to understand landslide processes, and to test new technologies and methodologies for slope stability characterisation and monitoring. We present InSAR results for the Hollin Hill landslide where a variety of ground-based geophysica, geodetic and climatic measurements (e.g. GPS, Electric resistivity tomography, meteorological observations) are available for comparison with InSAR data. We use Sentinel-1 InSAR data acquired between Oct 2015 and Jan 2020 to study the behaviour of this landslide. We find that the line-of-sight component of the down-slope movement is 2.7 mm/yr in the descending track, and 7.5-7.7 mm/yr in the ascending track. The InSAR measurements also highlight the seasonal behaviour of this landslide. In July 2019 six corner reflectors were installed to improve the coherence of the InSAR measurements, especially in the ascending acquisition mode. We present comparison with ground-based measurements such as the movement recorded by the GPS measurements of the pegs of the ERT survey and the moisture recorded by the various instruments at the site, and show the improvement introduced by the corner reflectors. We assess the applicability and utility of corner reflectors for active land sliding processes in the UK. There are over 17000 landslides in the UK (Pennington et al., 2015) and yearly financial losses are likely to be considerably in excess of £10 million (Gibson et al., 2013). Monitoring this large number of landslides is a challenge, and remote sensing techniques can play a great role in providing costly effective information. As we overcome limitations, such as temporal coverage with the Sentinel-1 mission, and vegetation cover with the installation of corner reflectors, it will be more and more feasible to use InSAR technology for continuous observation of a large number of landslides.
Authors: Kelevitz, Krisztina (1); Chambers, Jonathan E (2); Boyd, James P (2); Novellino, Alessandro (2); Jordan, Colm (2); Biggs, Juliet (3); Hooper, Andrew (1); Wright, Tim J (1)The Mekong delta is inhabited by more than 20 million people in Vietnam and is among the most biologically diverse waterscapes and agriculturally productive of the world. However, most of the Delta is below 2 m of the sea level and hence is highly vulnerable to the additive effects of land subsidence and sea-level rise due to global climate change. Critically, these two factors directly impact a variety of hazards which can be associated with subsurface saline intrusion, increases in the depth and duration of annual flooding, and naturally occurring arsenic contamination. Large scale land subsidence can be measured using satellite-based SAR imagery processed by interferometry with high accuracy from space. Recently, the Copernicus Earth observation program headed by the European commission in partnership with the ESA led to the launching of the Sentinel-1 mission, providing global open access C-band SAR coverage of Earth surface. For example, to cover Mekong Delta-wide, a full SAR Sentinel-1 image, characterized by 3 swaths and 10 bursts (e.g. 25200 pixels in azimuth and 67500 pixels in range direction, covering about 300 km x 250 km), is stored in complex values with about 14 GB memory data. From 2015-2020, there are 250 multi-temporal SAR images. To do interferometry processing, we need to be able to intelligently handle ensemble 250 x 14 GB data. Thus, although Sentinel-1 Big Data offers the best opportunity for land subsidence monitoring, it is challenge due to an unprecedented big volume of multi-temporal InSAR dataset. In this paper, we aim to provide a perspective on feasible methods to handle the Delta-wide land subsidence. To make a better understanding the challenge, we provide an example on the random access memory (RAM) requirement over the low Mekong Delta area in Vietnam with respect to Small Baseline Subsets (SBAS) and persistent/distributed scatterers (PSDS) approaches. From 2015-2019, we selected 55 Sentinel-1 images to test the performance, resulting in 750 GB. We then extract the information only at the PSDS measure points (e.g. 14x10^6), whereas we do multilook (e.g. 3 in azimuth and 15 in range direction, resulting in 38x10^6 pixels) in the SBAS processing to increase signal-to-noise ratio. For SBAS, we use MintPy, whereas the PSDS analysis is carried out by the TomoSAR. We assume the same requirement of memory (e.g. four times) to process signal in both PSDS and SBAS methods. With PSDS method, it requires 24 GB RAM to process, whereas SBAS needs 67 GB (e.g. 4x750/45 GB). Hence, with the capacity of RAM of 128/256 GB (in 2019), we can handle such 55 images amounts of data. However, the problem is that we have more than 250 Sentinel-1 images now (February 2020) and of course there is much more in near future. Is there a possibility to solve this problem? With SBAS (i.e., MintPy), it always loads all the data and store at once, so that the best solution is to reduce the number of images to meet the memory requirement. With PSDS approach, it needs future works, e.g. maturity of Sequential Estimator, to develop novel methods to process such Big InSAR Data.
Authors: Ho Tong Minh, Dinh (1); Ngo, Yen-Nhi (2); Vuong, Quoc Viet (3); Le, Trung Chon (4); Le Toan, Thuy (5)Jharia Coalfields, the largest and one of the oldest coalfields in India, are critically affected by coal fires due to the poor management of coal mines in the past. Apart from causing loss of non-renewable and precious coking coal, coal fires also contaminate the environment, thus substantially contributing to global warming. The subsidence induced by surface and subsurface fires in Jharia coalfields is a prominent concern that requires immediate attention. Coal fires burn the underground coal leading to land subsidence, which in turn aggravates the coal fires by creating cracks and fissures that serve as inlets for oxygen. This phenomenon leads to the creation of sinkholes in the area resulting in severe loss of lives and infrastructure. The current study analyses the relationship between Land subsidence and coal fires in the Jharia Coalfields, India. For the study, satellite imagery from thermal and microwave regions of the electromagnetic spectrum is used to deduce the effect of coal fires on mining-induced subsidence. 60 Sentinel-1, C-band images captured in ascending and descending direction during the dry season of 2017-19 are used for the Persistent Scatterer Interferometry analysis. The Satellite data acquired from the Landsat-8 thermal band (band 10) are used to derive the Temperature anomaly information of the study area. The Sentinel-1 data are pre-processed using SNAP. Further, the PSI analysis is carried out using the Stanford Method for Persistent Scatterers (StaMPS) method. The results are corrected for atmospheric errors and decomposed into vertical subsidence information to quantify the extent and intensity of subsidence due to subsurface coal fires. The results show a maximum subsidence rate of approximately 20 cm/yr in the Jharia coalfields. Just a few meters away from the settlements, the largest deformation zone covering about 1.5 sq. Km is identified in the Kusunda underground mine. With thermal imagery, surface and subsurface coal fires are detected and compared to the land subsidence information. The findings exhibit a positive correlation between the subsidence velocity and Land Surface Temperature in the study area. It is noted that the regions with a subsidence rate above 10cm/yr show maximum temperature anomaly nearly at +20oC. The study demonstrates the potential of the combination of SAR data and thermal imagery for effective monitoring of coal fire-induced land subsidence.
Authors: Karanam, Vamshi (1); Motagh, Mahdi (2,3); Garg, Shagun (3); Jain, Kamal (1)IntroductionThe unprecedented availability of open SAR (Synthetic Aperture Radar) data, open source interferometric software packages as well as the ready-to-use interferometric products such as ground displacement maps, resulted in their wide use by a range of stakeholders such as scientists, institutional organizations and commercial sector. Furthermore, in recent years, many algorithmic approaches and improvements able to exploit the large amount of multi-temporal interferometric SAR data and provide high accuracy ground deformation results have been developed. However, the repeatability and the reliability of multi-temporal interferometric processing still remains an open issue, mainly due to the ongoing and continuous algorithmic improvements (Garthwaite et al., 2019). Different processing methodologies may produce inconsistent ground displacement results. Moreover, the high complexity of the interferometric SAR signals makes interpretation a challenging task (Pritchard et al., 2018).In order to improve the interpretation and usability of the interferometric SAR signals we propose an approach to identify the driving factors of the ground deformation. The measured total ground deformation can be considered as a sum of several deformation components induced by several driving factors. Driving factors can be related with natural and anthropogenic processes such as tectonic activity, groundwater extraction etc. Burnol et al., 2019 successfully identified connections of natural driving factors (temperature and soil moisture) in the ground deformation. In this study, we present a wavelet-based approach to explore the connections of each driving factor with ground deformation results from interferometric SAR.Area of Interest (AOI)The study area covers a part of the Korisos basin, in Kastoria region, which is located in the northwestern Greece. The main deformation source of the study area is expected to be related to groundwater extraction activities for irrigation and domestic use purposes during the dry season from May to September (Gianneli et al., 2011). The area of the region of interest is about 95 square kilometers. The land cover of the study area consists mainly of agricultural, mountainous regions and some villages.Data and MethodsFirstly, the ground deformation information time series, using a well-known PS (Persistent Scatterer) approach implemented in STaMPS toolbox (Hooper et al., 2012), were extracted. A dataset of 193 single look complex C-band SAR images acquired by Sentinel-1A/B, from September of 2016 till December of 2019, have been processed. The preprocessing of the SAR dataset in order to be further processed by the PS (STaMPS) approaches was performed using the ISCE software package (Rosen et al., 2012). As a reference stable point, a high coherent point located in Kastoria city outside of the AOI was selected. The first image of the SAR dataset was used as temporal reference. The extraction of the topographic term from the interferograms was done using a photogrammetrically produced digital elevation model with ground pixel size of 5m created in 2009, provided by KTIMATOLOGIO S.A. The calculated displacement time series one from each approach was sampled every 6 days. The displacement values correspond to the direction between the sensor and the target (line of sight). Only the points with high temporal coherence (>0.90) were used in this study.At a second stage, the clustering of the ground deformation time series is performed. For dimensionality reduction reasons, the t-distributed stochastic neighbor embedding (T-SNE) transformation was applied in the ground deformation for each point. Then using elbow and silhouette methods the optimal number of clusters was calculated. Finally, a density-based spatial clustering of applications with noise (HDBSCAN) was applied to the transformed ground deformation data.At a third stage, the preprocessing of the information related with the driving forces of the ground deformation is implemented. The driving forces that were considered consist of a) vegetation index calculated from a stack of 250 Sentinel-2 acquisitions, b) the variations of the area of the spatial extent of the lake water calculated from the intensity of the Sentinel-1 dataset, c) the temperature from ERA5 reanalysis dataset, d) the soil moisture from ERA5 reanalysis dataset and e) the total precipitation from ERA5 reanalysis dataset. It is important to state that our AOI covers a single ERA5 pixel, which means we have only a single value for the whole AOI in time.At a fourth and final stage, cross wavelet analysis (XWT) of the calculated ground deformation time series in conjunction to each one of the considered driving factors for each cluster was performed. The XWT analysis was performed to identify patterns at different time intervals and consists of two steps. The first step is the calculation of the continuous wavelet transform (CWT) of each time series to identify localized intermittent periodicities at time/frequency space. The second step is the multiplication of the CWT of the first time series with the complex conjugate of the CTW of each driven factor. Each XWT result is a 2-D representation of the absolute value and the phase of the complex number in the time-frequency space. The XWT analysis was performed using the freely available MATLAB wavelet tool (Grinsted et al., 2004).ResultsThe main results of our case study are: A subsidence phenomenon with seasonal behaviour is identified over Mavrochori and Polikarpi villages. Clustering of ground deformation provided spatially well-separated clusters. From XWT analysis it was found that the deformation information of the clusters over agricultural regions presented the highest agreement (in-phase) with no time lag with the variations of water cycle. The water cycle variations are captured via the driving factors: a) lake water variations, b) precipitation and c) Soil moisture. From XWT analysis, ground deformation and NDVI was found to have an anti-phase relationship for a 6-month period. Over agricultural clusters, ground deformation leads around 0.5-1 month the NDVI. This time lag can be potentially related with the different agricultural actions on each field over the agricultural regions. Over the regions with natural vegetation no time lags are identified. From XWT analysis , ground deformation and temperature was found to have an anti-phase relationship for a 1-year period. Over agricultural regions, deformation leads temperature by about one month, in comparison with all the other regions that no time lag was identified. We believe that the observed time lag over agricultural regions can be related with the crop calendar and agricultural practices. Ιt is worth noting that in the current work, we used spatially coarse information from ERA5 for half of the considered driving factors. The study indicated the need of more accurate data related to the identified driving factors. ConclusionsIn this study, we presented a new wavelet-based approach for revealing the different interactions of several driving factors to the ground deformation. The methodology can be used as an exploration tool to identify the driving factors of the ground deformation. We consider the exploration of the driving factors of the ground deformations a critical step towards applying new methodologies such as data-driven approaches. We believe that the proposed methodology is a robust and efficient way to explore and analyse the impact of multiple and co-existing potential driving factors to a massive volume of ground deformation information.AcknowledgementsWe would like to thank the European Space Agency (ESA) for supplying the Sentinel-1 images. The authors would like to acknowledge KTIMATOLOGIO S.A. for kindly providing the DEM for the study area. We would also like to thank A. Grinsted for his freely available MATLAB wavelet toolbox (http://www.glaciology.net/wavelet-coherence). Finally, we would like to thank the organizations and people that contributed to open source interferometric software packages.ReferencesBurnol, A., Aochi, H., Raucoules, D., Veloso, F. M., Koudogbo, F. N., Fumagalli, A., ... & Maisons, C. (2019). Wavelet-based analysis of ground deformation coupling satellite acquisitions (Sentinel-1, SMOS) and data from shallow and deep wells in Southwestern France. Scientific reports, 9(1), 1-12.Garthwaite, M. C., Miller, V. L., Saunders, S., Parks, M. M., Hu, G., & Parker, A. L. (2019). A Simplified Approach to Operational InSAR Monitoring of Volcano Deformation in Low-and Middle-Income Countries: Case Study of Rabaul Caldera, Papua New Guinea. Frontiers in Earth Science, 6, 240.Gianneli, C., Voudouris, K., & Soulios, G. (2011). Water resources in Korisos basin, NW Greece: interactions between surface and ground water. International Journal of Water, 6(1-2), 57-73.Grinsted, A., Moore, J. C., & Jevrejeva, S. (2004). Application of the cross wavelet transform and wavelet coherence to geophysical time series. Nonlinear Processes in Geophysics, 11, 561-566.Hooper, A., Bekaert, D., Spaans, K., & Arıkan, M. (2012). Recent advances in SAR interferometry time series analysis for measuring crustal deformation. Tectonophysics, 514, 1-13.Pritchard, M. E., Biggs, J., Wauthier, C., Sansosti, E., Arnold, D. W., Delgado, F., ... & Zoffoli, S. (2018). Towards coordinated regional multi-satellite InSAR volcano observations: results from the Latin America pilot project. Journal of Applied Volcanology, 7(1), 1-28.
Authors: Karamvasis, Kleanthis; Karathanassi, VassiliaLandslides are among the major disasters of large scale that present various impacts and may affect the natural environment as well as urban areas. Therefore monitoring landslide activity over extensive areas occurs as an important aspect at the management of natural or man-made hazards and risk assessment. Several events need to be detected immediately after their occurrence, in order to detect the extent of the phenomenon and make important decisions concerning the surrounding environment. Detection and further monitoring of landslides could be accomplished by several methods. Among the available disciplines, remote sensing could offer valuable tools as it provides wide coverage of the affected areas, high frequency data and information about non-visible spectral regions. In addition, advances in technology offer a wealth of satellite data, methods and techniques that are constantly increasing over time giving the ability to exploit various potentials. This study includes the detection of a mine landslide in Greece using SAR and optical data. It concerns a large scale landslide phenomenon that took place in a lignite mine in the north-western part of Greece. The event occurred in the 10th of June 2017 leading to multiple economic and social impacts such as the destruction of a large part of the mine, cancelation of mining activities, as well as the partial destruction and evacuation of an adjacent village. Sentinel-1 and Sentinel-2 satellite data acquired before and after the event are used to detect the landslide after its occurrence. These different open access data sources are combined implementing digital image processing techniques, in order to make optimal use of their complementary information content. The processing is done using the open source software SNAP and includes InSAR processing to exploit the phase difference between the data sets derived from Sentinel-1 in order to capture the initial deformation caused by the landslide. Furthermore, the optical data derived from Sentinel-2 are processed for change detection using ERDAS Imagine software and the results are further combined with the corresponding products derived from SAR data. In addition, visual interpretation of the affected area is used as an effective tool in the overall study, while relevant geological data provide information about the geological background of the area. The resulted images are suitable for detecting changes in the mine area after the event, taking the advantage of multimodal approaches such as the synergy of various data types. Synthetic aperture radar data are suitable for rapid, successful and operational monitoring of the landslide and appear valuable for decision making and risk management. The contribution of Sentinel-2 high resolution optical imagery to the detection is evident, offering additional information.
Authors: Karagianni, AikateriniIn the mountainous areas of western China, the gradual change landslide as hidden dangers are widespread, and their formation conditions are often highly concealed and complex, which greatly increases the difficulty of early identification and monitoring of regional landslide hazards. The spaceborne InSAR technique is a powerful tool to detect the surface deformation and has obvious advantages in the identification and monitoring of potentially unstable slopes at a catchment scale. However, in areas with significant topographic relief, the phase error caused by the vertical stratification of the atmospheric and the residual phase of the undulating terrain can not be ignored, which has not been considered and removed in the conventional time series InSAR analysis. Besides, decorrelations caused by complex terrain and vegetation coverage in southwest mountainous areas of China, make it impossible for conventional time series InSAR methods to identify sufficient measurement points to get desired monitoring results. Considering the above-mentioned problems, in this paper, we are intended to deeply analyze the temporal and spatial distribution of the information, such as amplitude, phase and coherence of SAR images to improve the error-signal model of InSAR time series analysis; besides, we will make a joint exploitation of persistent scatterers (PS) and distributed scatterers (DS) to depict the landslide surface displacement in detail to improve the InSAR ability in mapping landslide deformation feature by combining multi-source and multi-orbit SAR dataset. In our study, archived ascending C-band EnviSAR ASAR, L-band ALOS PALSAR and C-band Sentinel IW datasets will be analyzed to detect the potentially unstable slopes. Case study of Danba County using both L-band ALOS PALSAR images (2006–2011) and C-band Sentinel IW images (2017-2018) have demonstrated the effectiveness of our propose method in monitoring the activity of potential landslide in China's mountainous west. Later, a joint analysis of the existing terrestrial observations, the InSAR measurements, and the field surveys will be conducted to give validation and geological interpretation of the InSAR derived results. In order to better explore the relations between the InSAR deformation results and the internal failure mechanism, the deformation evolution characteristic, the triggering factors, and the deformation failure modes of typical landslide will be analyzed and discussed.
Authors: Jiang, Yanan (1); Xu, Qiang (1); Lu, Zhong (2)The European Space Agency’s Copernicus Sentinel-1A/B twin-satellite constellation has greatly advanced the Earth-imaging capabilities of SAR remote sensing with improved spatiotemporal resolution, which fits into the scope of monitoring slow-moving landslides at rates of millimeters to meters per year. The contemporary polar-orbiting SAR satellites map the ground surface from the oblique line-of-sight with a horizontal component dominated by east-west displacements. Therefore, the present-day spaceborne InSAR method is insensitive to north-south motions, and unstable slopes may not face in favorable directions of SAR satellites. A complete 3D displacement solution requires at least three inputs from distinct perspectives. SAR satellites share similar trajectories, and thus strictly speaking, only two distinct InSAR LOS measurements from ascending and descending satellite orbits are available. Arbitrary assumptions such as simply no north-south motions or constraints imposed by topographic gradients can provide a quasi-3D displacement estimate, yet this is subject to large bias. As a complement to InSAR LOS measurements, pixel offset tracking provides additional azimuth offset measurements; however, its utility largely depends on the signal-to-noise ratio, and not all Earth mass movement processes have sufficiently large displacements and available SAR imagery of high spatial resolution. The Uninhabited Aerial Vehicle SAR (UAVSAR) is an airborne SAR system deployed by NASA/JPL, and it collects data at meter to submeter resolution along user-specified flight paths. Here we rely on the airborne UAVSAR and the spaceborne Sentinel-1 systems to characterize the Slumgullion landslide in Colorado, USA that moves at a couple of centimeters per day, and present additional spaceborne ALOS-2 system data for a geological complex of landslides and creeping Hayward fault over the San Francisco East Bay Hills in California, USA that moves at a few centimeters per year. The flexible trajectory of the aircraft and the hybrid InSAR and pixel offset tracking methods allow for an optimal 3D displacement solution, which can be further used to investigate the formation of morphological structures, landslide-fault interactions, non-Newtonian mass movements, slope channel modulation, and the spatiotemporal-dependent sensitivity to hydroclimatic variabilities.
Authors: Hu, Xie (1); Bürgmann, Roland (2); Fielding, Eric (3)ABSTRACT: Wuhan, the capital of Hubei Province and a megacity in China, has a population of more than 10 million, and has experienced significant development over the past decades. The city is located in a karst ecosystem that comprises extensive carbonate rocks and soft soils. Intensive municipal and industrial construction along with localized karst suffusion has led to Wuhan being one of the most sinkhole-prone metropolises in China, which poses a serious threat to the urban infrastructure and environment. In this study, we evaluate the evolution and current status of subsidence in Wuhan using a multi-sensor approach. Analysis of 43 C-band Sentinel-1A images (May 5, 2015 to July 5, 2017) and 120 X-band TerraSAR-X images (April 16, 2013 to September 8, 2017) enabled the extraction of the deformation temporal evolution. Multi-sensors interferometric synthetic aperture radar (InSAR) analysis for a 4-year periods revealed three distinct subsidence patterns, mostly concentrated in the i) Hankou commercial area (HK) (maximum rates exceeding -6 cm/yr), ii) Metallurgical Avenue in Qingshan and Yangluo industrial areas (-3 cm/yr), and iii) Baishazhou karst area (-3 cm/yr). Accuracy assessment of the InSAR results with 106 leveling benchmarks and cross-validation between the TerraSAR-X and Sentinel-1A images indicated overall root-mean-square errors of 2.5 mm/yr, and 3.1 mm/yr, respectively. The analysis of the derived displacements coupled with information from hydrogeological data, karst cave drilling analysis, groundwater and the Yangtze River water level records, and the mean IBI extraction over Wuhan suggest the following. i) Surface subsidence in the northern part of Wuhan (comprising Zones 1, 2, 3 and 6) have been seriously affected by anthropogenic activities. This was confirmed by the fact that most of the subsidence clusters observed in this study occurred along the subway under construction, construction sites, new ports, and heavy industry production bases. Moreover, the rapid decline of the groundwater level was almost synchronized with land subsidence. ii) In the highly developed karst areas (Zones 4 and 5), the typical structure of "upper cohesive soil and lower sand" provides a good material basis for subsurface suffosion and vacuum suction. The existence of a large number of karst caves within a depth of 30 m along with the intensive municipal engineering excavation in this area will inevitably increase the lateral and vertical hydraulic gradients between the pore water, karst water and the Yangtze River water, and accelerate the suffosion effect, thus leading to sinkhole subsidence. iii) The building load density generally increased land subsidence, but it was only limited to areas with large subsidence rates. Furthermore, the contribution of building load density was not the dominant factor overall as indicated by the maximum correlation coefficient being just 50% in areas with significant subsidence rates. Keywords: Sinkhole subsidence; InSAR analysis; Natural suffosion; Urban development; Building load
Authors: Hu, Jiyuan (1); Motagh, Mahdi (2); Guo, Jiming (1); Haghshenas Haghighi, Mahmud (2); Li, Tao (3)The near-surface geology of northern Germany is characterised by glacial deposits, which are deformed and penetrated by rising permian and upper triassic salt structures. The salt structures mainly rise along tectonic fault zones and today partly trace the structures of the Glückstadt Graben. The observation of ground motion potentially associated to these processes poses a special challenge for geophysics. It requires the measurement of motion rates with an accuracy of only a few millimetres per year, a sufficient spatial coverage of tenth of square kilometres, and a spatial density of the measurement points of less than one per square kilometre. To measure ground motion, we use radar interferometric time series data that are based on SAR (Synthetic Aperture Radar) images acquired by ESA satellites ERS-1 and ERS-2 between 1992 and 2000 over Schleswig-Holstein and Hamburg. Such radar interferometric time series analyses are possible for temporally stable backscattering objects (persistent scatterers) on the ground. Generally, this results in spatially dense observations over built-up areas and less dense observations in rural areas. We use a set of geostatistical and statistical methods to analyse these time series data. We detect signals of large-scale surface-deforming processes such as the subsidence of the marshes along the Elbe or the use of large gas caverns, as well as signals of small-scale processes such as the swelling of anhydrite at the Segeberger "Kalkberg" and subsidence processes at the edges of the historic old town of Lübeck. We are introducing techniques that allow us to derive ground motion even in areas with low scatterer density. Finally, we specify quality criteria to keep even such data consistent, which suffer from large time gaps in the acquisitions. Furthermore, we show how our methods can be used to link ERS data with the newer Sentinel-1 data.Our work extends the area of application of the PS-InSAR technique, from areas with high motion rates to regions with particularly low motion rates. It shows limitations of the current practices, particularly resulting from the missing decomposition of long term and short term effects in the PS-InSAR time series, a geographical resolution too coarse for precise monitoring and the low density of observations over important installations like the dikes along the west coast of Schleswig-Holstein.
Authors: Hoogestraat, Dieter Hidde (1); Sudhaus, Dr. Henriette (1); Omlin, Andreas (2)Intensity rainfall with a long-term period plays a principal role in triggering debris flow, shallow landslide, and deep-seated landslide around the mountainous area in Taiwan. To reduce the damage and to prevent loss of life resulting from the catastrophic landslide, this study proposed a new procedure aimed to interpret both new and ancient landslides and recent reactivation. The adaptive time series InSAR (ATS-InSAR) analysis approach we adopted combing the fusion data of persistent scatterer (PS) and distributed scatterer (DS) points, which improves the density of ground deformation points. This algorithm consists of three processing module; (i) APS estimation; (ii) PS selection, focused on the stable amplitude data on radar image; (iii) DS selection for the area of temporal and time geometrical decorrelation. For the long-term and steady data supply, the Sentinel-1 SAR image is the primary dataset we choose to develop the approach. This study selected a Butanbunasi basin were located in Kaoshiung City, southern Taiwan, as the study area. After Typhoon Morakot (2009), Butanbunasi catchment became the largest bare area characterized by numerous of the deep-seated landslide, debris slide, and sediment transportation in Taiwan. The sediment deposition, which came from the landslide zonation, formed an alluvial fan in the downstream estuary. It frequently affected the traffic safety of Highway Route 20 during the torrent rainfall season. In this paper, we mapped and classified different landslide types by adopting optical image and high-resolution LiDAR DTM. In general, the landslide inventory may provide past hotspot information on the vulnerable area, but the recent reactivation and topographic evolution of landslide are still insufficient. Therefore, the multiple-stage displacements derived from ATS-InSAR can be utilized as a sequence of ground movements on the hillslope. It presented 2-D recent landslide activity instead of traditional single-point monitoring while integrating the spatial and temporal information on the past landslide. To precisely define the possible landslide-prone hotspot, the concept of active deformation area(ADA) was considered. This approach enhances the various pattern of ground displacement trends by statistical test and spatial analysis. We highlighted serval large-scale landslides with a dangerous degree in Butanbunasi catchment during Jul. 2018-Nov. 2019. More field validation, such as in-situ investigation and UAV 3D terrain mapping, also confirmed the model accuracy. Additionally, the time-series displacement helps us recognizing which slope is more dangerous after critical events, including earthquakes or heavy rainfall. For instance, a gentle displacement curve transforms into a steep trend on the hillslope in a short-term period means may indicate a higher landslide susceptibility soon. Although the approach was developed to assess potential landslide hazards, it may contribute to updating the national-wide environmental geologic map and provide competent authority to make better decisions reducing the disaster impact.
Authors: Lee, Ching-Fang (1); Lin, Shih-Yuan (2); Lin, Yu-Ching (3); Lan, Chen-Wei (1)Across New Zealand, the Australian and Pacific plates obliquely converge a rate of 38-49 mm/yr. Variations in the relative plate motions from north to south has led to a transition from the subduction of the Pacific plate beneath the North Island to dextral transpression and strike-slip faulting through the Marlborough Sounds and central South Island. Despite vast improvements in mapping the location of active faults, there have been numerous large damaging earthquakes across New Zealand over the last 10-15 years with many occurring along previously unknown faults including the Mw7.8 Kaikoura earthquake in 2016. Our current estimates of interseismic strain and long-term fault slip rates have largely been based on modelling of GPS velocities and active fault studies. The campaign GPS network in New Zealand, which was established in the mid-1990's (with some expansion since then), has benchmarks which are typically located 10-20 km apart (although as dense as 5-10 km in places) and is currently re-measured approximately every eight years. A relatively comprehensive continuous GPS network has operated in much of the North Island (~20-30 km average spacing) for the last 10-15 years, although continuous GPS sites are much sparser in the South Island. Here, we combine ~10 years of InSAR observations from Envisat with interseismic campaign and continuous GPS velocities to build a high-resolution velocity field of New Zealand. Despite limited data acquisitions, strong atmospheric noise and contamination of non-tectonic signals, we successfully form the first InSAR derived velocity field of New Zealand. In addition to the full LOS velocity field, by exploiting the horizontal GPS observations we are able to estimate the vertical component of the deformation. In the North Island, we detect wide spread subsidence of the east coast as well as through the Taupo Volcanic zone. Localised subsidence associated with anthropogenic activities is also observed from mining, geothermal and hydrocarbon extraction. In the South Island, interseismic subsidence is observed in the Kaikoura region supporting previous observations of at least partial locking of the southern Hikurangi subduction interface. Despite data challenges in the mountainous regions from landslides, sediment compaction and glaciers, InSAR data shows localised uplift of the Southern Alps.
Authors: Hamling, Ian (1); Wright, Tim (2); Hrensdottir, Sigrun (1); Wallace, Laura (1,3)The northeast Tibetan Plateau, encompassing the Qilian Shan, Haiyuan Fault and East Kunlun Fault, is a tectonically active region prone to large earthquakes, including in the past century the 1920 MW 7.9 Haiyuan earthquake which resulted in over 230,000 deaths. Due to the long recurrence interval and patchy earthquake catalogues, geodetic estimates of crustal strain accumulation provide essential constraints on seismic hazard assessment. Here we present continuous velocity fields across the NE Tibetan Plateau, covering an area of ~440,000 km2 at
Authors: Ou, Qi (1); Daout, Simon (1,4); Weiss, Jonathan (2,3); Lazecky, Milan (2); Shen, Lin (2); Parsons, Barry (1)Geodetic measurements of crustal deformation provide crucial constraints on a region’s tectonics, geodynamics and seismic hazard. However, such geodetic constraints have traditionally been hampered by poor spatial and/or temporal sampling, which can result in ambiguities about how the lithosphere accommodates strain in space and time, and therefore where and how often earthquakes might occur. High-resolution surface deformation maps address this limitation by imaging (rather than presuming or modelling) where and how deformation takes place. These maps are now within reach for the Alpine-Himalayan Belt thanks to the COMET-LiCSAR InSAR processing system, which performs large-scale automated processing and time-series analysis of Sentinel-1 InSAR data. Expanding from our work focused on Anatolia, we are combining LiCSAR products with GNSS data to generate high-resolution maps of tectonic strain rates across the central Alpine-Himalayan Belt. Then, assuming that the buildup rate of seismic moment (deficit) from this geodetically-derived strain is balanced over the long term by the rate of moment release in earthquakes, we pair these strain rate maps with seismic catalogs to estimate the recurrence intervals of large, moderate and small earthquakes throughout the region. We also use arguments from dislocation modeling to identify two key signatures of a locked fault in a strain rate field, allowing us to convert the strain maps to “effective fault maps” and assess the contribution of individual fault systems to crustal deformation and seismic hazard. Finally, we address how to expand these approaches to the Alpine-Himalaya Belt as a whole.
Authors: Rollins, Chris (1); Wright, Tim (1); Hooper, Andy (1); Weiss, Jonathan R. (2); Maghsoudi, Yasser (1); Lazecky, Milan (1)Interferometric Synthetic Aperture Radar (InSAR) phase unwrapping error is a major limiting factor on the InSAR-derived tectonic deformation velocity. This is particularly the case when atmospheric turbulence, large deformation gradient and strong phase noise exist. To address limitations of previous phase unwrapping error correction methods which are not supported for multi-looked InSAR data, here we present a new algorithm that integrates decorrelation phase correction, triplet phase closure (TPC) test and integer linear programming (ILP) to overcome this limit. The rationale behind is that we mitigate the phase inconsistence using decorrelation correction and then detect the phase unwrapping error magnitude using TPC. Next we borrow the ILP from Compressed Sensing that converts the phase unwrapping error correction to a sparse signal recovery problem. We demonstrated the validity of our method by using synthetic data and 5-years Sentinel-1 real data covering the Central San Andreas Fault creep section, where exists obvious tectonic deformation, strong atmospheric disturbance and decorrelated scatterers, and the inverted long-term creep model constrained by InSAR velocity after correction shows lower uncertainty than that constrained by the uncorrected one.
Authors: Ma, ZhangfengHow earthquakes initiate and run-away into major ruptures is still a challenging research topic, that will benefit from increasing our capability to observe processes from the seismogenic source regions. In recent years, two models for earthquake nucleation have been proposed to explain earthquakes sequences, a slow-slipping model [Tape et al., 2018] and a cascade model [Ellsworth and Faith, 2018]. Moreover, those models are based mostly on analysing seismic data. Corroboration and validation with independent datasets could advance our understanding. Within this context, we use geodetic data to contribute to the study of seismogenic source regions during earthquake sequences. Particularly interesting are earthquake swarms as they do not obey typical earthquake empirical laws, e.g., Gutemberg-Richter law. This deviation might be due to a disproportioned contribution of aseismic processes, and hence provide an opportunity to investigate the role of aseismic behaviour in the nucleation and propagation of earthquakes. In this communication, we study a shallow seismic swarm in Hawthorne (Nevada, USA), in March-September 2011. We identify three stages with respect to the time when the most energetic event (M4.6) occurred: a nucleation phase (before-M4.6) from 15 March to 19 April, the most energetic phase (spanning the M4.6 event), and the post-energetic phase (after-M4.6) until 17 September. We process Envisat and Radarsat-2 data and generate 8 SAR interferograms to quantify surface displacements. Interestingly, during the “nucleation” phase, 2 Radarsat-2 interferograms indicate clear surface displacement signals (~4 cm away from satellite line-of-sight motion). We also find that, in interferograms covering the most seismically energetic, surface displacement signals are larger in magnitude and located further north with respect to the nucleation phase. During the post-energetic phase, surface displacements are detected along a very narrow spatial band with clear phase discontinuities suggesting surface ruptures. The analysis of the interferograms suggests that fault slip might have occurred along a fault system with a geometry consisting of two fault planes. To develop a fault kinematic model, firstly we apply an art-of-the-state inversion method solving for uniform distribution in a Bayesian approach directly from the interferometric wrapped phase [Jiang and González, 2020]. The modelling of selected interferograms covering the different phases confirms that ground motion could be caused by a fault geometry with two distinct fault planes. During the nucleation phase, the observed ground motion would be consistent with slip along a N-S striking normal fault. While modelling of interferograms covering the most seismically energetic and the post-energetic phases points towards a different fault segment on a NE-SW-trending normal fault. Based on the fault geometry modelling, together with ground motion discontinuities, we construct a smooth fault plane with uniformly discretized triangular. Then, we estimate the fault slip distribution model with associated uncertainties. Here, we develop a new distributed fault slip inversion method based on a prescribed regularization based on an experimentally validated physics-based crack model. To further investigate the temporal evolution of fault slip with a higher temporal resolution, we invert for fault slip time-series using all available interferograms [González et al., 2013]. The inversion results suggest that: (1) there were three areas with different spatio-temporal slipping behaviour: a narrow (5km2) slip area on the southern fault with a high slip rate (lower boundary: 1.5cm/day, or 1.7x10-7m/s) occurring in the nucleation phase; a wider (15km2) slip area on the northern fault which ruptured during the energetic phase and with lower average slip (10cm); and a shallow slip area (depth=1km) just above the second area in the post-energetic phase and with minor average slip rate (lower boundary: 0.2cm/day, or 2.3x10-8m/s); (2) our results are more consistent with a cascade model of discrete slip patches, rather than a slow-slipping model thought as a growing elliptical crack; (3) the geodetic moment is 50 times of all combined seismic event in the nucleation phase, and it implies the aseismic moment ratio is variable from 100%. The ratio reduces over time during the swarm, but remains larger than 60% after the most energetic phase. Our study of the 2011 Hawthorne shallow crust seismic swarm allows us to illuminate fault slip in much greater spatial detail than usually possible using GNSS networks. We conclude that this seismic swarm there was characterized by significant aseismic fault processes, most likely slow-slip or localized fluid-enhanced fault slip, along with discrete segments of the fault plane active before and after the largest earthquakes. This study contributes to highlighting the importance of using geodetic data to understand the role of aseismic processes in the nucleation and propagation of earthquakes. References Tape, Carl, et al. "Earthquake nucleation and fault slip complexity in the lower crust of central Alaska." Nature Geoscience 11.7 (2018): 536-541. Ellsworth, William L., and Fatih Bulut. "Nucleation of the 1999 Izmit earthquake by a triggered cascade of foreshocks." Nature Geoscience 11.7 (2018): 531-535. Jiang, Yu, and González, Pablo J. "Bayesian Inversion of Wrapped Satellite Interferometric Phase to Estimate Fault and Volcano Surface Ground Deformation Models." Journal of Geophysical Research: Solid Earth 125.5 (2020): e2019JB018313. González, Pablo J., et al. "Magma storage and migration associated with the 2011–2012 El Hierro eruption: implications for crustal magmatic systems at oceanic island volcanoes." Journal of Geophysical Research: Solid Earth 118.8 (2013): 4361-4377.
Authors: Jiang, Yu (1); González, Pablo J. (1,2)SAR interferometry has stepped in the big-data era, particularly with the acquisition capability and open-data policy of ESA’s Sentinel-1 SAR mission. Large amount of Sentinel-1 SAR images has been acquired and archived, allowing for generating thousands of interferograms, covering millions of square kilometers. In such a large-scale interferometry scenario, many applications still focus on monitoring kilometer-scale local deformation, sparsely distributed in a large area. It is thus not effective to apply the time-series InSAR analysis to the whole image stack, but to focus on areas with deformation. Aiming at this, we present our recent work built upon deep neural networks to firstly detect localized deformation and then carry on the time-series analysis on small interferogram patches with deformation signals. For local deformation detection, we propose a method combining phase-gradient stacking and the widely-used neural network for tiny object detection: You Only Look Once (YOLOv3) to detect slow-moving landslides from large-scale interferograms. The stacked phase gradients clearly present the characteristics of localized surface deformation, mainly caused by slow-moving landslides, avoiding from the impact of phase unwrapping and atmospheric effects. We then train the YOLOv3 network with the stacked phase-gradient maps with known landslides to achieve the quick and automatic moving landslide detection. We design a Phase Unwrapping Network (PUNet) to unwrap the cropped interferogram patches centered on the detected local deformation. To train the network, we develop interferogram simulation strategies to generate various training samples based on distorted 2D Gaussian surface and fractal Perlin noises. The PUNet is proposed to learn a direct mapping from interferograms to unwrapped phases by treating phase unwrapping as a regression problem. The PUNet structure is inspired by a classical denoising DnCNN model, residual networks, and dilated convolution. The DnCNN has been shown to be effective in image denoising, avoiding the downsampling procedure. As an infrastructure, it can effectively counter SAR interferogram with dense fringes and decorrelation noises. Meanwhile, the dilated convolution can dramatically increase the receptive field of a network, while the residual module can accelerate the training procedure. We apply our networks for mining subsidence and landslides monitoring. Comparing with traditional time-series InSAR analysis, the presented strategy not only reduces the computation time, but also avoids the influence of large-scale tropospheric delays and the propagation of possible unwrapping errors. The presented methods introduce artificial intelligence to the time-series InSAR processing chain and make the mission of regularly monitoring localized deformation sparsely distributed in large scale feasible and more efficient. As future work, we can further improve the temporal resolution of InSAR based local deformation monitoring by training networks combining interferograms from C-band and L-band SAR images, which will be available soon from future SAR missions such as NiSAR and LuTan-1.
Authors: Wang, Teng (1); Wu, Zhipeng (2); Fu, Lv (1)The assessment of the natural hazards related to slow slope gravitational mass movements and to permafrost deformation is essential for the proactive management of the risk in the Alpine environment and for the relevant information on climate changes that can be derived from the study of the evolution of such phenomena. In this context, in the Alpine region, many kinds of landforms like rock and debris-covered glaciers or landslides can be affected by these phenomena. To this aim in-situ, proximal and remote sensing technologies can be used. In particular, Synthetic Aperture Radar (SAR) data and techniques have the surplus-value respect to other geodetic methods of providing information over distributed areas even at night-time and through the cloud cover. Many new generation SAR techniques and data are available to monitor superficial deformations, but especially in the challenging mountain environments, the best solution for all these phenomena does not exist. This is due to the different characteristics of techniques (such as amplitude tracking, SAR simple- or multi-interferometry), data (in terms of wavelength, spatial and temporal resolution) and superficial deformations (which are characterized by different ranges, temporal evolutions, and seasonality behaviours, while the related areas of interest differ for size, steepness, spatial heterogeneity and soil properties). This work considers two very different test sites characterized by superficial deformation phenomena located in South Tyrol (Italy): the Lazaun rock glacier and the Corvara landslide. However, in terms of analysis with SAR data, both present problems typical of mountainous areas. This makes possible to define and test criteria for the choice of SAR data and techniques. The Lazaun rock glacier is situated in the southern Ötztal Alps (upper Schnals Valley, South Tyrol) and covers an area of 0.12 km2. The medium-sized active rock glacier extends from 2480 to 2700 meters a.s.l. and it moves downstream because of the deformation of internal ice. Very high-resolution TerraSAR-X images were used for Differential SAR Interferometry and amplitude tracking processing to monitor its displacement. Moreover, Sentinel-1 data were processed through the Small Baseline Subset technique and from the combination of both ascending and descending acquisition geometries it was possible to estimate the vertical displacement rate in the area of interest, identifying mass movements and other displacements related to the permafrost creep. The Corvara landslide is located in Badia Valley with a surface area of 2.5 km2. Being a vegetated area, the installation of artificial scatterers allowed to measure the displacements occurred from 2016 to 2018 through the analysis of the amplitude of very high-resolution TerraSAR-X data. Moreover, the application of a multi-temporal technique of Differential SAR Interferometry to Sentinel-1 data permitted to monitor the landslide with extensive spatial coverage and to retrieve the displacement evolution of the area during the snow-free period. GNSS, UAV and Ground-Based SAR campaigns in the first case and GNSS in the second one were used for comparison and validation of the results obtained from SAR satellite data. Processing and analysis were conducted within the ALPSMOTION project funded by the Province of Bolzano Italy. It will also be described an unsupervised method to update the regional rock glacier inventory exploiting interferometric SAR coherence and based on Sentinel-1 data, to discriminate between moving and no-moving landforms over large spatial areas developed in this project. Acknowledgements This work was conducted within the project ALPSMOTION (ALPine Slow slope Movement moniTorIng and detectiON with remote and proximal sensing), coordinated by Eurac Research-Institute for Earth Observation and funded by the Autonomous Province of Bolzano, Alto Adige, “Ripartizione Diritto allo Studio, Università e Ricerca Scientifica.” The Sentinel-1 and TSX data were processed with SARscape software (SARMAP). GB-SAR instrument has been made available by the University of Pavia. GPS data were provided by the University of Innsbruck. TSX Data were provided by the European Space Agency, Project Proposal id 34722, © DLR, distribution Airbus DS Geo GmbH, all rights reserved.
Authors: Cuozzo, Giovanni (1); Callegari, Mattia (1); Marin, Carlo (1); Beccaro, Lisa (1); Bertone, Aldo (2); Steger, Stefan (1); Seppi, Roberto (2); Zucca, Francesco (2); Dematteis, Niccolò (3); Riccardi, Paolo (4); Notarnicola, Claudia (1)In the last six years, following the launch and commisioning of Sentinel-1, opportunities to use InSAR have increased dramatically, especially outside of academia. Open access data acquired using consistent, frequent and reliable scheduling allow for the potential application of InSAR in a variety of commercial settings. Data quality, measurement coverage and quanitification of confidence levels are important, but to realise its full potential and make InSAR generally accepted and used in industry, we need to make data access more rapid and convenient, and the data itself easier to interpret by those without a background in InSAR or remote sensing. At SatSense, we have released a country wide ground movement product for the United Kingdom, based on full resolution Sentinel-1 data. We provide complete coverage of the United Kingdom in both ascending and descending geometries and can update our time series in near-real time, resulting in two to six measurements every six days everywhere in the country. The instant availability of up-to-date ground movement data, combined with its inherent ability to measure large areas cost-effectively, means InSAR can be used as the go-to first overview of ground movements. This opens up a range of industry applications, such as the property market, infrastructure monitoring and geotechnical applications. We provide a range of convenient data delivery options, including a web interface for data visualisation and ordering, API access and a tileserver which can be directly accessed by GIS software. Here, we discuss how our processing and data delivery ecosystem works and provide a range of examples of ground movement across the United Kingdom, covering all our target markets. We will broadly cover how we process the large volumes of data acquired over the United Kingdom, and demonstrate our data visualisation portal and tile server. Finally, we will show case studies covering ground movements due to landslides, sinkholes, building work and mining, both current and historical. Our UK dataset contains a wealth of ground deformation signals over a range of spatial and temporal scales, demonstrating that InSAR ground movement data can provide important information and add significant value to industry, government agencies and the general public.
Authors: Spaans, Karsten (1); Ingleby, Tom (1); West, Michael (1); Hooper, Andy (2); Wright, Tim (2)Sentinel-1 satellite allows to monitor infrastructure in almost 7 years long datasets using about than 300 images by now. Such a long (and dense) dataset brings better accuracy and allows to reduce ambiguity problems, but, on the other hand, non-linear movements lead to the decrease of temporal coherence and therefore to possible exclusion of the point. Coherence is defined as a property of a point (resolution cell), representing stability of the reflection from that point. Lower values of coherence should be attributed to noisy points, caused e.g. by low vegetation, temporary movement or movement of only part of the resolution cell. But, as temporal coherence is calculated from residues with regard to the (usually) linear model, if a point is subject to non-linear movement, its coherence decreases and does no more represent the noise of the point. Also, non-linear movements often lead to incorrect ambiguity estimation and “jumps” in the time series. These are searched for and corrected, based on local trends. Also, in such a long dataset, some points may be noisy in a temporary period. In order to distinguish points moving in a non-linear way from points with higher noise, we utilize time series classification to estimate the right “model” for each point, and calculate a new value of coherence, called model coherence, which should serve as a better estimation of noise. At the same time, we recognise which points are subject to subsidence, uplift, or if the trend changes during the monitoring period (or a period set by a user). We approximate the time series with distinct intervals with constant velocity and noise level, and we also evaluate one-shot movements. Processing is performed statistically. For each time interval, we estimate also the velocity precision, velocity accuracy (based on the processing of an area supposed to be stable) and reliability (with regard to ambiguities). Our method allows also to work with points that are temporarily very noisy (i.e. irreliable). The overall reliability of the InSAR results is then improved by combining results from all satellite tracks available. The described technique was used for railway monitoring in Czech republic where non-linear movements are frequent due to railway works or due to slow landslides.
Authors: Hlavacova, Ivana; Kolomaznik, Jan; Struhar, JurajRadar interferometric processing, such as Persistent Scatterer Interferometry (PSI), with their potential for high spatial coverage and high accuracy, has been increasingly used to obtain ground movements. However, the resulting one-dimensional side-looking measurements in line of sight (LoS) require further analysis in order to derive three-dimensional ground movement components such as height changes. One of these methods is the decomposition for ground movement components by combining cross-heading LoS-datasets (ascending and descending), which can due to their opposite East-West looking directions differentiate the vertical (height change) and lateral displacements in eastern direction. Thanks to the open access of the SAR datasets from Sentinel-1 mission it is possible to apply such decomposition widely, e.g. for the whole European area. Although, this method and its application has been discussed commonly in publications, the gained measurement quality, especially the systematic error due to the neglected displacements in the northern direction, has not been sufficiently documented. In this paper, by using ascending and descending Sentinel-1 data, we discuss the quality assessment of the determined time series of ground movement components, which is carried out by quantifying their statistical uncertainties via a polynomial trend analysis and by estimating their systematic errors due to the neglected displacements in the northern direction. The systematic errors for the ground movement components calculated from the decomposition of two cross-heading datasets can be expressed by the multiplication of the true northern displacement and a so called ε-factor, which depends on the incidence angle and the azimuth looking direction of the respective radar acquisitions. In this work we introduce these ε-factors for the vertical and the East-West horizontal components, which make it possible to evaluate the magnitude of systematic errors even prior to performing the decomposition. In addition, similar evaluations were performed for the vertical movement component calculated from a single LoS-dataset. A comparison of these results shows that the potential systematic error for the vertical component from decomposition is significantly lower than from single LoS-dataset. Moreover, based on the ε-factor, the magnitude of the systematic error for the calculated eastern displacement tends towards zero, which means it is negligible. In terms of the statistical uncertainties of the displacement time series after the decomposition, it can be stated that they depend on the uncertainty of the used PSI-results and the respective constellation. Once the incidence angle is smaller than 45°, the uncertainty of the eastern displacement will be larger than the vertical one.
Authors: Yin, Xiaoxuan; Busch, Wolfgang