Conference Agenda

Standard InSAR methods use a parameterization that focuses on past behavior to describe how a particle or scatterer has moved over time up to the present. The archetypal parameter associated with this approach is the average velocity, potentially combined with higher-order terms, such as polynomial or seasonal components. For diagnostic or forensic applications, this parameterization may be adequate. However, it is clearly not optimal for describing the motion of a particle or scatterer whose behavior changes after a certain moment in time—for example, suddenly accelerating landslides, impending sinkholes, or infrastructure instabilities.

It is precisely in such situations that satellite-based InSAR can provide significant value, since near-continuous acquisition updates enable the detection and identification of changes in displacement behavior. For these applications, the standard parameterization is therefore suboptimal.

Here, we propose an alternative parameterization based on the instantaneous state vector of the particle, with instantaneous position and instantaneous velocity as the key parameters of interest, alongside time-invariant parameters such as cross-range position, thermal coefficient, and nuisance parameters related to atmospheric delay. This parameterization enables a recursive Kalman filter approach, consisting of a continuous cycle of prediction and update steps. The “Instantaneous State” methodology exhibits specific characteristics, benefits, and challenges depending on the application at hand. Consequently, the method is not directly suitable for so-called application-agnostic, or double-A, InSAR products, which are typically produced without a clearly and uniquely defined end user or objective. However, it is well suited for application-aware and application-aligned (triple-A) InSAR products.

In its most elementary form, Instantaneous State InSAR operates on arcs, where the motion of one vertex is described relative to another (base) vertex. After an initial batch initialization, the state transition is performed to predict the state vector for the new epoch. Uncertainty in this transition is modeled by introducing a difference vector that incorporates information about the expected smoothness of the point’s behavior. Whereas the traditional parameterization implicitly assumes “infinite smoothness”—for example, by assuming steady-state linear behavior—this approach relaxes that assumption and allows for varying degrees of smoothness through the adoption of a smoothness doublet.

This smoothness doublet must be determined based on contextual information, reinforcing its suitability for triple-A products. It is used to characterize an Ornstein–Uhlenbeck process, in which the mean-reverting velocity at a given time is statistically related to its past values, with a dependence that decays exponentially over time.

As soon as a new observation is acquired, the updated instantaneous state is computed as a weighted combination of the predicted state and the new observation. An important characteristic of the approach is the estimation of the appropriate integer phase ambiguity at each measurement update.

In this presentation, we will outline the method, which is available online (https://eartharxiv.org/repository/view/9445/), describe its characteristics, and demonstrate its efficacy.

Two global scale radar satellite systems, the NASA/ISRO NISAR and ESA Sentinel-1 satellites, are currently collecting comprehensive InSAR-quality observations of most of Earth’s land and ice surfaces. Because these radars operate at rather different wavelengths, the interactions of each with the surface will differ. We test the hypothesis that the primary difference in radar scattering at the two wavelengths is the penetration of each signal into the solid Earth surface, vegetation canopies, and icy terrains, resulting in very different levels of subsurface components in the radar echoes. The longer wavelength L-band waves might be expected to penetrate ~4x deeper into the volume scattering medium, promising more sensitivity to soil moisture variations, the influence of a ground component over vegetated areas, and structure in the topmost 100 m of dry snow zones in polar areas.

Here we present example side by side comparisons of interferogram time series from both systems, and identify phenomenological differences between them. If contrast between the two data sets corresponds to places where we expect more surface penetration, this supports the hypothesis. In these cases, we present simple models of the scattering process at each wavelength using the same physical model but varying the radar wavelength. More specifically, we can solve for i) the volume decorrelation component by comparing acquisitions at several spatial baselines at each frequency, and ii) the magnitude of the subsurface component from phase closure of interferogram triples. For ice in particular, given the volume scattering component we can infer the depth of penetration, which determines what layers of any moving ice are contributing to the observed velocities, and perhaps constraints on layer thicknesses.

It is worth noting that the selection of operating wavelength for each radar was primarily driven by technical factors, most specifically for NISAR to decrease fringe density and minimize temporal decorrelation. Nonetheless the wavelength diversity permits a more insightful description of the surface than is possible with a single wavelength, and opens up opportunities for the observation of a wider variety of subsurface and subsurface conditions.

Conventional InSAR methods are primarily sensitive to east-west and vertical components of ground motion and have very limited sensitivity to the north-south component of the three-dimensional motion vector. TOPS burst overlaps in azimuth (De Zan et al, 2014, Prats-Iraola et al, 2012) and range (Nergizci et al, 2025) provide sensitivity to along-track motion using Burst Overlap Interferometry (BOI). This technique has reduced sensitivity compared to standard InSAR, but have been shown to provide more precise measurements of along-track motion than conventional methods based on offset-tracking in cases when the coherence is properly preserved.

In standard Interferometric Wideswath (IW) SLC products from the Sentinel-1 mission, these overlap zones are limited to about 2 km, both in azimuth and range. These products contain only pixels of uniform resolution, corresponding to fixed processing bandwidth in range and azimuth. However, by relaxing this requirement and keep the full spatial extent contained in the Level-0 SAR data, we show that is possible to extend the burst overlaps significantly. In the range direction, each burst is extended by one pulse length, enlarging the swath overlap zone by up to 15 km. In the azimuth direction, the burst coverage is extended by the length of the processed synthetic aperture, corresponding to an increase of the overlap of up to 8 km. The extra coverage is characterized by a gradual loss of spatial resolution from the edge of the original overlap zone to the edge of the extended overlap region due to the reduction of effective bandwidth. However, for mapping of large-scale motion, the loss of resolution in the extended regions is usually not an issue.

In this contribution, we elaborate on the necessary adaptations to standard SAR focusing methodology in order to produce Sentinel-1 SLC data with extended spatial overlap between bursts, and we describe the spectral and spatial characteristics of the resulting extended SLCs. Then we describe how such SLCs can be interferometrically combined in the extended overlap zones. Furthermore, we analyze the statistical performance of the resulting along-track InSAR measurements.

We apply the methodology to different types of significant along-track motion, where the extended burst overlap coverage provides a significant improvement over standard methodology. Examples include recent major strike-slip earthquakes; the 2023 Kahramanmaraş events (Mw 7.8 and Mw 7.5) and the 2025 Myanmar earthquake (Mw 7.7), and ice sheet motion in East Antarctica.

De Zan, F., et al. "Interferometry with TOPS: Coregistration and azimuth shifts." EUSAR 2014; 10th European Conference on Synthetic Aperture Radar. VDE, 2014.

Nergizci, M., Lazecky, M., Wright, T. J., Hooper, A., Ou, Q., Magnard, C., & Çakir, Z. (2025). Refining 3D Displacement Fields and Coseismic Slip Models of the 2023 Kahramanmaraş Earthquakes Using Subswath and Burst Overlap Interferometry (SBOI). Journal of Geophysical Research.

Prats-Iraola, P., Scheiber, R., Marotti, L., Wollstadt, S., & Reigber, A. (2012). TOPS Interferometry With TerraSAR-X. IEEE Transactions on Geoscience and Remote Sensing, 50(8), 3179–3188. https://doi.org/10.1109/TGRS.2011.2178247

Polarimetric information has been shown to greatly enhance interferometric synthetic aperture radar (InSAR), in particular by enabling optimization approaches that maximize interferometric coherence resulting in decreased phase noise. While InSAR is routinely used for surface deformation monitoring in various applications, polarimetric InSAR (PolInSAR) techniques are not commonly implemented operationally. Many InSAR applications, such as critical infrastructure monitoring and slope stability assessment, require high resolution SAR images with short turnaround time to analytic products. The lower resolution and swath widths associated with fully polarimetric, or quadrature polarimetric (QP), SAR beam modes therefore often preclude their use in these applications. Moreover, PolInSAR coherence optimization methods can be computationally demanding, increasing operational complexity and time taken to produce deformation measurements. Despite the significant improvement offered by PolInSAR coherence optimization, operational constraints have limited its use outside of proof-of-concept demonstrations.

Dual polarimetric (DP) and compact polarimetric (CP) beam modes can acquire SAR images at high resolution with large swath widths, as opposed to QP modes which necessitate transmission of additional pulses to capture the full scattering matrix. A result of this trade-off is the reduced polarimetric information content in DP and CP imagery. PolInSAR coherence optimization approaches are more operationally viable with DP and CP beam modes due to the improved spatial resolution, swath width, and reduction in computational requirements from the simplified polarimetric parameter space. Indeed, PolInSAR coherence optimization has been shown to improve InSAR deformation monitoring with DP imagery. However, the DP mode is inherently biased to a subset of all potential scattering processes resulting from transmission of a single linear polarization, reducing the improvement gained by PolInSAR coherence optimization. The use of DP PolInSAR coherence optimization is currently not widely adopted in InSAR deformation monitoring programs. Through transmission of a circularly polarized signal, the CP mode offers a more balanced representation of scattering processes that increases polarimetric information content over DP, in some cases approaching that of QP. As a result, CP polarimetry offers an intriguing solution to operationalizing PolInSAR coherence optimization techniques for deformation monitoring.

In this work, we demonstrate an implementation of PolInSAR coherence optimization that operates on SAR imagery from CP beam modes that greatly enhances deformation monitoring with minimal impact on processing pipelines. We obviate the numerical optimization problem by instead focusing on identifying the dominant, temporally stable polarimetric component for each target. Using high resolution CP imagery from the RADARSAT Constellation Mission, we show how this approach results in increased target density and reduced phase noise with only modest increases in processing time compared with single-channel InSAR deformation analyses.

With the upcoming launch of MDA Space’s CHORUS SAR constellation mission at the end of 2026, the C-band CHORUS-C satellite will soon enable routine access for the first time to high resolution imaging in CP Spotlight and Stripmap modes for InSAR monitoring. With PolInSAR coherence optimization techniques such as that presented here, these data offer new ways in which InSAR monitoring can meet key requirements for different applications. For example, monitoring of critical infrastructure, such as bridges, transportation corridors, and dams, requires dense target coverage to capture localized deformation. High quality InSAR measurements of these structures, such as those resulting from CP mode PolInSAR coherence optimization, are crucial for informing effective decision making by stakeholders.

The Sentinel-1 satellite mission has been key to the achievement of interferometric synthetic aperture radar (InSAR)-based displacement rates that approach mm/yr precision, particularly in regions without significant vegetation and where long time series of observations exist. However, for more subtle displacement signals, separating the effects of surface processes from deformation due to deeper sources is still challenging. Here, we present a new method based on combinations of co-polarized (VV) and cross-polarized (VH) InSAR data. Cross-polarized data is typically noisier than the co-polarized data and is not widely used for InSAR. However, comparisons of co- and cross-polarized phase data can allow separation of the contributions from different processes. Signals due to deeper sources, such as slip along faults, should appear the same in both data types, while differences can be due to changes in surface characteristics. We examine full-resolution, unfiltered, VV and VH Sentinel-1 data covering Death Valley, California between January 2017 and March 2025 (Figure 1). We find that displacement rates derived from VV and VH data differ by several mm/yr in some areas, particularly at three alluvial fans on the west side of our focus region. We define a metric called the “cumulative unwrapped phase difference rate,” which helps us determine if a pixel has consistently similar, or significantly diverging VV and VH displacement histories. Small values of this metric indicate that the pixel is less likely to be affected by shallow processes in the soil, whereas large values of this metric indicate pixels that are likely influenced by these processes. We propose that these shallow soil processes include transient changes in surface properties (e.g., soil moisture) and changes in surface geometry (e.g., salt crystal growth and swelling clays). Pixel behavior may also depend on its grain size distribution (e.g., a large rock serves as stable reflector and the pixel containing it maintains consistency between VV and VH over time). We threshold the cumulative unwrapped phase difference rate metric to define multiple populations of interspersed pixels, and we compare rate differences between pixel populations over short spatial scales. We show that rates based only on the VV imagery differ by a few mm/yr between subsets of pixels where the VV-VH differences are large or small. This suggests that leveraging an underutilized dataset (VH), in combination with VV data, can help researchers reliably identify pixels that are the least impacted by surface processes, and therefore provide the most reliable estimate of long-term surface deformation. While our work focuses on Death Valley, similar mm/yr-scale biases could impact endorheic basins around the world and influence analyses of interseismic motion, hazard estimates, and groundwater studies.