We document one of several methodologies used to validate the NASA-ISRO Synthetic Aperture Radar (NISAR) mission requirements for solid earth deformation. NISAR’s deformation requirements cover steady-state, coseismic, and transient deformation processes and were designed to confirm that the mission is able to meet its solid earth science goals. We use independent observations of earth surface deformation from continuous Global Navigation Satellite System (GNSS) stations as ground truth for NISAR-observed deformation, and we provide a statistical framework to assess the quality of the associated NISAR data products. Our validation workflows have been developed as Jupyter Notebooks and are publicly available via GitHub/GitLab.

The NASA ISRO Synthetic Aperture Radar (NISAR) is scheduled for launch early in 2024 from the Satish Dhawan Space Centre (SDSC), at Sriharikota, near Chennai, India. This mission is the result of a collaboration between NASA and Indian Space Research Organization (ISRO), where NASA has contributed elements of the mission such as an L-band SAR, and ISRO has contributed other elements, such as an S-band SAR. After successful launch, the NISAR mission will collect left-looking L-band SAR data over most of the Earth’s land areas twice during every 12-day exact repeat orbit. (once while in an ascending orbit direction and once while in a descending orbit direction). NASA and ISRO have individual and joint requirements on the mission that include the performance of the imaging radars onboard the spacecraft. For example, NASA must demonstrate that this L-band SAR will achieve a set of identified science measurement accuracy requirements that span Ecosystem science, Solid Earth science, and Cryosphere science disciplines. Likewise, ISRO has several applications objectives on both the L-band and S-band data from NISAR that the ISRO science team and project will be developing and testing. Pre-launch and post-launch activities have been planned to validate that these requirements are met. Here, we will discuss how the NASA plans are being executed and will present any initial results at the conference.

Systematic phase errors induced by processes such as soil and vegetation moisture dynamics impair interferometric synthetic aperture radar (InSAR) displacement measurements. They can manifest as closure errors and as discrepancies between phase history estimates obtained with contrasting algorithms. To advance InSAR deformation estimation algorithms and understand the biases, we need to quantify the temporal characteristics of closure errors. Here, we introduce temporal closure signatures, complete and nonredundant descriptions of the closure errors as a function of time and timescale. Temporal closure signatures are extracted using a suitable basis of the closure covector space, the annihilator of the vector space of consistent phases. Observed Sentinel-1 signatures from two such bases exhibited contrasting seasonal and long-timescale characteristics that covaried with land and vegetation cover. Strong spatial associations of closure signatures with secular and seasonal discrepancies in phase histories estimated from a subset or all interferograms were observed. To interpret these observations, we made predictions using simple interferometric scattering models. Model predictions for seasonally changing vegetation with a long-term drying trend resembled observed closure signatures over scrublands. The model further captured the observed subsidence in the nearest-neighbor relative to the all-interferogram estimate. The simulated deformation estimate deteriorated with the inclusion of long-term interferograms, as was also predicted for subresolution differential motion, but these predictions require validation with empirical data. The observations and simulations illustrate the potential of closure signatures for identifying sources of systematic closure errors, informing algorithm selection, and enabling model-driven bias correction.

The fading signal is a term coined by [1] and describes InSAR deformation velocity biases that become apparent between different Small Baseline Subset (SBAS) time series solutions. As the maximum temporal baseline, or bandwidth, is increased, the observed velocity estimate of distributed scatters converged to that of persistent scatters. In turn, they also converged to that of full-network-based solutions estimated from the InSAR covariance matrix [1] . Velocity deviations between SBAS (with only few connections) and full-network estimators can be on the scale of multiple cm / yr [1] , [2] . These biases are related to non-zero closure phase [2] , a measure of phase inconsistency between triplets of interferograms. Based on these closure phases and the observed temporal baseline dependence, [2] offers a means of correcting this bias using these closure phases and under the assumption that longer-baseline interferograms are less error-prone.

Among the most rapid and dramatic changes related to Arctic permafrost thaw are retrogressive thaw slumps. These slumps evolve by a retreat of the slump headwall during the summer months, making their change visible by comparing digital elevation models over time or by identifying the induced vegetation changes in time-series of optical satellite images. In this study we use digital elevation models generated from TanDEM-X observations to detect and derive volume and area change rates for RTSs on the Taymyr Peninsula. The available data takes allows to compare two time periods: from 2010/11/12 to 2016/17 and from 2017/18 to 2020/21. From 2016 onwards optical Sentinel-2 observations are available for which we manually map thaw slumps for each year in a small sub-area and compare the results to the TanDEM-X mapped RTSs. We found a strong, non-linear increase in the second time-period of the TanDEM-X period and by using the optical mapped RTSs we could attribute this increase to the Siberian heatwave in 2020.

Retrieving a consistent phase history from a multi-looked InSAR stack depends critically on the accuracy of the (coherence) magnitude estimates. To estimate the magnitudes more reliably, I propose three regularization methods: Hadamard, spectral, and Hadamard–spectral regularization. All three are computationally cheap and parameterized such that they are guaranteed to yield valid magnitude matrices. These regularizers achieved relative improvements in the phase history accuracy of up to 40% in simulations. The improvements were greatest for low long-term coherences. All three methods performed similarly in the simulations and in a Sentinel-1 stack, for which the local phase dispersion decreased with regularization. Implementation of the regularizers into operational processing chains is expected to improve deformation and uncertainty estimates, especially for local movements over decorrelating terrain.

Ionospheric Faraday rotation distorts satellite radar observations of the Earth’s surface. While its impact on radiometric observables is well understood, the errors in repeat-pass interferometric synthetic aperture radar (InSAR) observations and hence in deformation analysis are largely unknown. Because Faraday rotation cannot rigorously be compensated for in nonquad-pol systems, it is imperative to determine the magnitude and nature of the deformation errors. Focusing on distributed targets at L-band, we assess the errors for a range of land covers using airborne observations with simulated Faraday rotation. We find that the deformation error may reach 2 mm in the copol channels over a solar cycle. It can exceed 5 mm for intense solar maxima. The cross-pol channel is more susceptible to severe errors. We identify the leakage of polarimetric phase contributions into the interferometric phase as a dominant error source. The polarimetric scattering characteristics induce a systematic dependence of the Faraday-induced deformation errors on land cover and topography. Also, their temporal characteristics, with pronounced seasonal and quasi-decadal variability, predispose these systematic errors to be misinterpreted as deformation. While the relatively small magnitude of 1–2 mm is of limited concern in many applications, the persistence on semiannual to multiannual time scales compels attention when long-term deformation is to be estimated with millimetric accuracy. Phase errors induced by uncompensated Faraday rotation constitute a nonnegligible source of bias in interferometric deformation measurements.

Deformation estimation from radar interferometric stacks has to confront speckle over decorrelating distributed targets. Inferring the speckle-induced uncertainty in the estimated phase history is challenging. Previously published estimates based on Fisher information (FI) can underestimate the errors by an order of magnitude. Here, we introduce three improvements to mitigate the bias. We: 1) account for uncertainty in the magnitudes of the interferometric covariance matrix elements; 2) penalize the likelihood to reduce the impact of coherence biases on the phase history uncertainty estimates; and 3) constrain the covariance magnitudes to stabilize the estimation. In simulations, these improvements substantially reduced the bias in the uncertainty estimates. Bias reduction was due to an increase in the predicted uncertainty (improvements 1–3) and a decrease in the actual error (improvements 2 and 3). Temporal correlations–crucial for model fitting and testing–were also estimated more accurately. In observations, the underestimation relative to the observed spatial variability was largely eliminated. In contrast to the alternative estimates based on spatial variability, the improved FI uncertainty estimates are applicable to small-scale phenomena such as sinkholes. They can serve as foundation for reliable uncertainty estimates of the deformation derived in subsequent interferometric processing steps, thus bolstering model testing and data fusion.

Vast areas of the Arctic host ice-rich permafrost, which is becoming increasingly vulnerable to rapid thaw in a warming climate. When ice-rich permafrost soil thaws it becomes unstable, leading to changes in the topography. The most rapid and dramatic changes occur due to retrogressive thaw slumps. These slumps evolve by a retreat of the slump headwall during the summer months, making them detectable by comparing digital elevation models over time. Fort his study we used digital elevation models derived from bistatic single-pass radar observation from the TanDEM-X satellite pair over the time span from 2011 to 2017. Here we present area and volumetric change rates for thaw slumps from two study areas in Northern Canada (Mackenzie River Delta uplands and Banks Island). The RTSs in the two study area show large differences in terms of RTS size and growth rates. The computation of typical terrain controls like aspect, slope and RTS location (hillslope, shoreline) did not correlate with RTS activity, suggesting that other factors especially soil properties (e.g. ground ice content) play a larger role for RTS evolution.

The Gaussian speckle model for homogeneously distributed targets is commonly assumed to apply in repeat-pass radar interferometric analyses, for instance, in deformation estimation. This is despite widespread evidence from snapshot intensity observations indicating deviations from Gaussianity, as many natural land surfaces are intrinsically heterogeneous. The concern is that neglecting heterogeneity will deteriorate the phase estimates and induce underestimation of the uncertainty. Here, we introduce and theoretically characterize compound models that extend the Gaussian speckle model for repeat-pass stacks by representing heterogeneity in intensity and phase. In two L-band repeat-pass data sets, we find pervasive deviations from Gaussianity. Our estimates suggest that the heterogeneity in intensity is largely due to time-invariant, rather than dynamic, texture. Deviations from Gaussianity associated with phase heterogeneity are generally less pronounced. One notable exception with large estimated phase heterogeneity occurs over a permafrost wetland, where degrading ice wedges induce subsidence that is variable on the resolution scale. For deformation analyses, accounting for heterogeneity has, on average, a moderate impact on the phase estimates and the estimated phase uncertainty, which increases by 10% on average. However, in intrinsically heterogeneous areas, such as the permafrost wetland, the accuracy of the phase estimate can realistically improve by up to 20%, and the predicted phase uncertainty increases by 30%. The improvements in phase estimation accuracy and in the quality of the uncertainty estimates when accounting for heterogeneous speckle can, on occasion, make a notable difference for subtle or small-scale deformation.

Vast areas of the Arctic host ice-rich permafrost, which is becoming increasingly vulnerable to rapid thaw in a warming climate. When ice-rich permafrost soil thaws it becomes unstable, leading to changes in the topography. The most rapid and dramatic changes occur due to retrogressive thaw slumps. These slumps evolve by a retreat of the slump headwall during the summer months, making them detectable by comparing digital elevation models over time. Here we present first results of a retrogressive thaw slump detection method using bistatic single-pass radar observation from the TanDEM-X satellite pair over the time span from 2011 to 2017. Our processing chain include the digital elevation generation process, elevation model differencing, water body mask generation and a method for the detection of significant height changes. We detected 157 active RTS validated by high- to medium-resolution optical data in our study region (3700 km2) located in the Mackenzie River Delta, Northwest Territories, Canada.

In the subarctic tundra, soil moisture information can benefit permafrost monitoring and ecological studies, but fine-scale remote-sensing approaches are lacking. We explore the suitability of C-band SAR, paying attention to two challenges soil moisture retrieval faces. First, the microtopography and the heterogeneous organic soils impart unique microwave scattering properties, even in absence of noteworthy shrub cover. Empirically, we find the polarimetric response is highly random (entropies >0.7). The randomness limits the applicability of purely polarimetric approaches to soil moisture estimation, as it causes a tailor-made decomposition to break down. For comparison, the L-band scattering response is more surfacelike, also in terms of its angular characteristics. The second challenge concerns the large spatial but small temporal variability of soil moisture observed at our site. Accordingly, the Radarsat-2 C-band backscatter has a limited dynamic range (~2 dB). However, contrary to polarimetric indicators, it shows a clear surface soil moisture signal. To account for the small dynamic range while retaining a 100-m spatial resolution, we embed an empirical time-series model in a Bayesian framework. This framework adaptively pools information from neighboring grid cells, thus increasing the precision. The retrieved soil moisture index achieves correlations of 0.3-0.5 with in situ data at 5 cm depth and, upon calibration, root-mean-square errors of3m-3. As this approach is applicable to Sentinel-1 data, it can potentially provide frequent soil moisture estimates across large regions. In the long term, L-band data hold greater promise for operational retrievals.

There was a typographical error in[1, eq. (18)]. Instead of

This paper presents the first experimental demonstration of the virtual bandwidth synthetic aperture radar (VB-SAR) imaging scheme. VB-SAR is a newly developed subsurface imaging technique which, in stark contrast to traditional close proximity ground penetrating radar schemes, promises imaging from remote standoff platforms such as aircraft and satellites. It specifically exploits the differential interferometric SAR (DInSAR) phase history of a radar wave within a drying soil volume to generate high-resolution vertical maps of the scattering through the soil volume. For this study, a stack of C-band vertically polarized DInSAR images of a sandy soil containing a buried target was collected in the laboratory while the soil moisture was varied-first during controlled water addition, and then during subsequent drying. The wetting image set established the moisture-phase relationship for the soil, which was then applied to the drying DInSAR image set using the VB-SAR scheme. This allowed retrieval of high-resolution VB-SAR imagery with a vertical discrimination of 0.04 m from a stack of 1-m vertical resolution DInSAR images. This paper unequivocally shows that the basic principles of the VB-SAR technique are valid and opens the door to further investigation of this promising technique.

Differential interferometric synthetic aperture radar (DInSAR) measurements are sensitive to displacements, but also to soil moisture mνchanges. Here, we analyze whether soil moisture can be estimated from three DInSAR observables without making any assumptions about its complex spatio-temporal dynamics, with the goal of removing its contribution from the displacement estimates. We find that the referenced DInSAR phase can be a suitable means to estimate mνtime series up to an overall offset, as indicated by correlations with in situ measurements of 0.75-0.90 in two campaigns. However, the phase can only be referenced when no displacements (and atmospheric delays) occur or when they can be estimated reliably. We study the separability of displacements and mνusing two additional DInSAR observables (closure phase and coherence magnitude) that are sensitive to mνbut insensitive to displacements. However, our analyses show that neither contains enough information for this purpose, i.e., it is not possible to estimate mνuniquely. The soil moisture correction of the displacement estimates is hence ambiguous too. Their applicability is furthermore limited by their proneness to model misspecifications and decorrelation. Consequently, the separation of soil moisture changes and displacements using DInSAR observations alone is difficult in practice, and-like for mitigating tropospheric errors-additional data (e.g., external mνestimates) or assumptions (e.g., spatiotemporal patterns) are required when the mνeffects on the displacement estimates are comparable to the magnitude of the movements. This will be critical when soil moisture changes are correlated with the actual displacements.

The first experimental demonstration of the Virtual Bandwidth SAR (VB-SAR) scheme is provided. VB-SAR is a new technique that promises subsurface imaging of soils at ultra-high, centimeter-scale resolution at large stand-off distances applicable to aircraft and spacecraft. This paper reports on how a stack of C-band images were used to retrieve high resolution vertical profiles of the backscattering through a soil in the laboratory. The VB-SAR scheme captures the phase behavior of a soil across a stack of DInSAR images as the soil dries. The real frequency of the interrogating radar behaves as a higher, virtual frequency within the soil by virtue of its higher-than-air dielectric. As the dielectric changes with time, the DInSAR stack captures a virtual bandwidth. Using this scheme, it was possible to produce a vertical slice of the backscatter through a soil at 10cm resolution, much improved on the formal 1m resolution offered by the real 150MHz bandwidth.

Permafrost soils have been shown to respond rapidly to warming temperatures. When ice-rich permafrost soils thaw, the melting ground ice reduces the volume and stability of the soils, inducing changes in the topography. We monitor surface elevation changes in three test sites in Northern Eurasia using single-pass TanDEM-X Science Phase data with submetre vertical precision. The results indicate the suitability of single-pass InSAR data for monitoring thaw-induced topographic changes (e.g. coastal erosion) but they also reveal the spurious impact of late-lying snow packs and water bodies, both of which are common in lowland permafrost areas. Furthermore, the coherence and hence the precision with which elevation changes can be estimated is found to be limited by the noise level in certain cases. As some of these influences could be mitigated using appropriate mission and acquisition designs, we conclude that single-pass interferometry has considerable potential for monitoring thaw-induced surface elevation changes in permafrost areas, which in turn could contribute to assessing their vulnerability, fate, and climate system feedback in a warming climate.

Depth-resolved radar imaging at L- to X-band has barely been applied to soils owing to limitations imposed by wave absorption within the soil and the resolutions attainable from air- or spaceborne platforms. Rather, soils are commonly studied using radar systems that cannot resolve the depth component. In this study, we adapt tomographic profiling to image the wetting and drying of sandy soil using a ground-based radar with a depth resolution of about 10 cm. The depth-resolving capabilities are achieved using synthetic aperture processing of the measurements obtained with downward pointing antennas operating at C-band with 2 GHz bandwidth. The observed subsurface backscatter appears to be governed by the local soil moisture content and the soil moisture content above (absorption). When the soil moisture content changes, the observed differential interferometric phase and coherence are consistent with the notion that the total depth-averaged interferometric return is governed by volume scattering and wave propagation within the soil. However, existing models of the depth-averaged interferometric coherence do not include variations in the volume scattering power induced by soil moisture changes, which the backscatter observations indicate exist. Besides improving our understanding of the radar backscatter from heterogeneous soils, depth-resolved imaging may in future also provide direct information about the spatial variability of soil properties and soil moisture dynamics.

The polarization diversity of the phase causes ambiguities in the estimation of displacements using differential interferometry. Over natural surfaces such as vegetated areas, the magnitude of these ambiguities is potentially related to complex dynamic processes such as vegetation growth. As the properties and possible origins of such diversity (besides noiselike influences) over changing vegetation canopies are virtually unknown, we propose to investigate them empirically using an L-band zero-baseline data set covering one growing season over different agricultural crops. We frequently observe HH-VV phase differences exceeding 0.5π, corresponding to a displacement discrepancy of 3 cm. The HH-VV phase difference and other properties of the polarimetric coherence regions (e.g., the shape and the relation to the in situ observed biomass) vary with the crop type. The observations over wheat and barley and, to a lesser extent, rape suggest the presence of birefringence within the canopy. By contrast, those over maize and sugar beet, while also showing phase diversity, cannot be explained by birefringence or similarly simple models. Irrespective of the origin of this dependence, its presence and systematic nature indicate the potential importance of vegetation effects in differential interferometry, which may limit the accuracy of the estimated deformations over vegetated areas.

Displacements of the Earth's surface can be estimated using differential interferometric synthetic aperture radar. The estimates are derived from the phase difference between two radar acquisitions. When at least three such acquisitions are available, one can compute the displacement between the first and the third acquisition and compare it with the sum of the two intermediate displacements. These two are expected to be equal for a piston-like spatially uniform deformation. However, this is not necessarily the case in measured data. Such lack of phase closure can be due to decorrelation noise alone. It has also been attributed to complex scattering processes such as soil moisture changes or multiple scattering sources. However, the nature of these nonrandom effects is only poorly understood in cold regions, as the role of snow and freeze/thaw processes has not been studied to date. To distinguish the noise-like and the systematic effects, an asymptotic Wald significance test is proposed. It detects situations when the observed closure error cannot solely be explained by noise. Such situations with p25%) in the X-band observations of ice-rich permafrost regions in the Lena Delta, Russia, indicating the presence of processes that can have systematic and deleterious impacts on the estimation of surface movements. Satellite-based monitoring of these displacements is thus possibly subject to complex error sources in high-latitude regions.

The polarization diversity of the phase in differential interferometry causes ambiguities in the estimation of displacements: it might be related to complex dynamic processes such as vegetation growth. As the properties and possible origins besides noise-like influences of such diversity over changing vegetation canopies are virtually unknown, we propose to investigate them empirically using an L band data set covering one growing season over different agricultural crops. The polarimetric coherence regions (i.e. the shape and relation to biomass) over wheat, barley and, to a lesser extent, rape suggest the presence of birefringence within the canopy, which can yield phase differences between HH and VV channels exceeding 0.5π. The remaining fields (maize and sugar beet) - while also showing such a dependence on polarization - cannot be explained by birefringence. Irrespective of the origin of this dependence, its presence and magnitude indicate the importance of vegetation effects in differential interferometry, with potentially deleterious influences on the estimation of deformations.

Changes in soil moisture between the two radar acquisitions can impact the observed coherence 7 in differential interferometry: both correlation |γ| and phase φ are affected. The influence on the latter potentially biases the estimation of deformations. These effects have been found to be variable in magnitude and dependent on polarization, as opposed to predictions by existing models. Such diversity can be explained when the soil is modelled as a half-space with spatially varying dielectric properties and a rough interface. The first-order perturbative solution achieves - upon calibration with L band data - median correlations ρ at HH of 0.77 for the phase Φ and of 0.50 for |γ|. The depth distribution of the scattering heterogeneities within the soil impacts the sensitvity of the observables to soil moisture changes in a way similar to a changing relative importance of the surface and the volume components, thus leading to similar qualities of fit when the latter is estimated. The first-order expansion does not predict any impact on the HV coherence, which is however empirically found to display similar sensitivities to soil moisture as the co-pol channels. These results indicate that the first-order solution, while not able to reproduce all observed phenomena, can capture some of the more salient patterns of the effect of soil moisture changes on the HH and VV DInSAR signals.

The second-order statistics are among the most important observables in synthetic aperture radar (SAR) polarimetry and are usually reported as covariance or coherence matrices. They are restricted to particular forms provided the target exhibits a certain kind of symmetry. As these constraints are not exactly fulfilled in real data, statistical tests are proposed for checking the validity of an invariance hypothesis. The application of these likelihood-ratio tests to airborne L-band data reveals a strong dependence of the test statistics on both the land cover and the number of looks; furthermore, temporal changes such as vegetation growth are evident: for example, lack of reflection symmetry for mature rape fields. This finding is at odds with commonly employed models, which assume (and predict) reflection invariance. For the Freeman-Durden decomposition, which relies on reflection symmetry, the connection between this invariance and negative powers (unphysical result) is found to be weak.

Different mechanisms for the impact of soil moisture on interferometric radar data have been proposed, but its magnitude, sign and even presence have barely been studied empirically and thus remain poorly understood. In this study the dependence of the phase and coherence magnitude on soil moisture was inferred empirically with regression techniques: this was done for two airborne data sets at L band. The phase dependence was significant (α = 0.05) for more than 70% of the fields at HH polarization, its sign corresponding to an increase in optical path upon wetting. This trend was similar in both campaigns, whereas the prevalence of soil moisture-related decorrelation differs. These results are only consistent with a dielectric origin of the soil moisture effects, and not with soil swelling or the penetration depth hypothesis.

The triple collocation technique, which retrieves the error variances of three sets of measurements of the same parameter, is applied to soil moisture records in central Spain: ASCAT remote sensing observations, REMEDHUS in-situ probes, and the ERA Interim model. The objective is the estimation of the temporal variability of the error of ASCAT. The three data sets have to be calibrated with respect to each other as they show different mean values and dynamic ranges. The time-variant estimation of both the error and the calibration parameters is shown to be very sensitive to the extents of the temporal windows used and the calibration procedure. Due to the temporal fluctuations of the calibration constants, artefacts such as seasonal variations and extreme values are introduced. This case study shows that the temporal analysis of the errors using the collocation technique can lead to spurious results when the data sets have to be referenced with respect to one another.

A novel sensor fusion algorithm for retrieving the freeze/thaw (f/t) state from scatterometer data is presented: It is based on a probabilistic model, which is a variant of the Hidden Markov model, and it computes the probability that the landscape is frozen, thawed, or thawing for each day. By combining Ku- and C-band scatterometer data, the distinct backscattering properties of snow, soil, and vegetation at the two radar bands are exploited. The parameters that are necessary for inferring the f/t state are estimated in an unsupervised fashion, i.e., no training data are required. Comparison with model and in situ temperature data in a test area in Siberia/northern China indicates that the approach yields promising results (typical accuracies exceeding 90%); difficulties are encountered over bare rock and areas where large fluctuations in soil moisture are common. These limitations turn out to be closely linked to the inherent assumptions of the probabilistic model.