Conference Agenda

Preliminary evidence indicates the presence of significant slope instabilities within multiple fjord systems across western and eastern Greenland. In some cases, these instabilities could generate tsunamis if large-scale failures were to occur. To identify potentially hazardous sites and to support the development of an early warning framework, our study evaluates the feasibility of monitoring such mass movements using satellite Differential Interferometric Synthetic Aperture Radar (DInSAR). Due to the limited understanding of the deformation processes operating at the target sites—including their driving mechanisms, spatial extent, and temporal characteristics—a multi-stage analytical approach is necessary. This progressive strategy allows us to first establish a regional overview before focusing on site-specific deformation patterns requiring more sophisticated techniques. Our primary aims are therefore to characterize the current activity levels of the slope instabilities and reconstruct the temporal evolution of slope displacements from 2015 onward.

Greenland presents significant challenges for DInSAR applications. While 80-85% of the island is covered by a permanent ice sheet, the ice-free coastal zones experience extensive winter snow cover lasting several months. The terrain itself is predominantly mountainous and rocky, characterized by a narrow, rugged coastline incised by deep fjords and flanked by towering mountains and tidewater glaciers. As a consequence, there are considerable areas not visible to the satellites or with poor line-of-sight sensitivity. Additionally, the ionospheric effects are more severe in polar regions compared to mid-latitudes. These combined factors make Greenland a particularly challenging environment for DInSAR analysis. As an initial step toward detecting and potentially quantifying very slow slope movements, we therefore applied a Sentinel-1 Persistent Scatterer Interferometry (PSI) analysis to three small test areas - each several hundred square kilometers (e.g., ~20km x 20km) - in the Uummannaq and Nuuk Fjord areas.

Our preliminary analyses demonstrate that the temporal evolution of movements in these areas between 2015 and 2025 can be reliably captured with Sentinel-1 PSI. We identified a well-defined snow-free period extending from approximately May/June through October/November, characterized by reduced out-of-season snow cover precipitation and consistent height coverage throughout. Therefore, a simple coherence threshold allows for the selection of snow-free acquisitions. Tropospheric effects are primarily limited to a height-dependent component, which can be estimated and removed using a phase-to-height relationship, while turbulence is notably restricted due to the dry, cold air. Ionospheric effects are generally small for Sentinel-1 data on a local scale and can be effectively removed using large-scale filters. Sentinel-1 acquisitions are available at regular intervals of 6 days (2017-2021 and 2025-onward) and 12 days (2015-2016 and 2022-2024). These favorable conditions enable robust processing and allowed us to extend our analysis to larger areas on the order of several tens of thousands of square kilometers in the Uummannaq and Nuuk fjord systems in western Greenland and the Tasiilaq Fjord in eastern Greenland using both ascending and descending data.

In this contribution, we will first discuss the main processing challenges and steps. We then present selected results from our study of slope instabilities. The large-scale overview of ongoing movements helped to prioritise relevant locations for detailed analyses and field visits. In the next phase of our work, we will conduct more detailed investigations to refine our understanding of the observed deformation patterns. Specifically, fast motions on the order of several cm/year—which cannot be detected with persistent scatterer interferometry—will be analyzed using multi-temporal, multi-looked interferograms. Future work also aims to progressively expand the analyzed area to additional ice-free regions of Greenland, currently encompassing approximately 338,000 to 410,000 km², or about 15-20% of the island's total area.

Land subsidence along the shores of the Dead Sea occurs in two primary modes: above dissolution cavities in association with sinkholes, and a continuous belt that follows the retreating shoreline. While the sinkhole mechanism is well established and forms the basis of the Geological Survey of Israel’s early-warning system, the origin, spatial extent, and temporal behavior of shoreline subsidence remain insufficiently understood. Clarifying its governing processes is essential for hazard assessment and infrastructure planning along the rapidly retreating shoreline.

Since 2018, systematic TerraSAR-X acquisitions have enabled the generation of 11-day repeat InSAR time series along the entire western shoreline of the northern Dead Sea basin. This dense spatial and temporal sampling allows, for the first time, separation between sinkhole-related deformation and shoreline-controlled subsidence and enables characterization of both seasonal and multi-year responses to water-level variations.

We analyzed 30 shore-perpendicular deformation profiles along the ~50-km shoreline of the northern DS basin for the period January 2019 to December 2024. A narrow subsidence zone, 100-400 m wide, is consistently observed landward of the shoreline. Line-of-sight displacement is maximal adjacent to the waterline and decreases westward. Subsidence rates reach up to 3 cm/month during summer, coinciding with peak evaporation and rapid lake-level decline, and decrease to 0-0.5 cm/month in winter when the water-level drop slows or temporarily reverses.

The deformation time series correlates closely with Dead Sea level variations, with a delay shorter than 10 days, within the temporal resolution of both SAR and level measurements. A flash flood in December 2025 produced a rapid rise followed by a drop in water level in a semi-detached southern basin and triggered measurable shoreline subsidence within a single 11-day acquisition cycle, indicating an almost instantaneous mechanical response to hydraulic forcing. Concurrently, the location of peak deformation migrates eastward together with shoreline retreat. Points initially located at the shoreline continue to subside for several years after exposure, following a quasi-exponential decay with a characteristic time scale of a few years.

Previous studies based on C-band InSAR and GNSS measurements attributed shoreline subsidence to regional aquifer-system consolidation driven by groundwater-level decline and increased effective stress. However, our observations challenge this interpretation: (1) the deformation is confined to a narrow shore-parallel belt rather than a broad zone of groundwater decline, and (2) seasonal subsidence amplitudes show no systematic dependence on shallow lithology derived from nearby boreholes.

We therefore propose a two-scale consolidation mechanism. At the shoreline, newly exposed sediments undergo rapid subaerial dewatering and consolidation, producing deformation that tracks short-term water-level fluctuations with minimal delay. Simultaneously, increased effective stress propagates downward through the sediment column over longer timescales, generating multi-year post-exposure subsidence decay that may depend on deeper lithological properties. This combined hydromechanical adjustment explains both the immediate response to seasonal and event-scale water-level changes and the prolonged subsidence observed after shoreline retreat. Our results provide a physical framework for interpreting InSAR observations along retreating shorelines and improve hazard assessment in rapidly declining terminal lakes worldwide.

Coastal regions across Java- particularly low‑elevation coastal zones (LECZs)- are experiencing rapid land subsidence driven by groundwater extraction, urbanisation, and underlying geological conditions, with rates in many locations far exceeding global sea-level rise. This produces a compounding multi‑hazard in which climate‑driven sea‑level rise (SLR) and vertical land motion (VLM) interact to accelerate relative sea‑level change (SLC). Java, home to more than half of Indonesia’s population and undergoing rapid demographic and economic expansion, is already highly exposed to coastal flooding. Yet most global and national assessments rely primarily on climate‑driven SLR projections, often neglecting subsidence and therefore underestimating local risk. Understanding the spatial variability of VLM and its drivers is essential for effective coastal planning, hazard mitigation, and long‑term adaptation. To address this gap and provide a more realistic estimate of population exposure, we integrate InSAR‑derived ground deformation with physical datasets describing land cover, geology, and elevation, alongside socioeconomic information on buildings and population density, and projections of climate‑driven sea‑level rise.

We generate a new island‑wide, high-resolution map of ground deformation for Java using Sentinel‑1 InSAR data from nine satellite tracks (2016-2023). Interferograms were processed using the LiCSAR (Lazecky et al., 2020) and LiCSBAS (Morishita et al., 2020) SBAS workflow, with atmospheric corrections from GACOS (Yu et al., 2018) and coherence‑based multilooking to ~500 m resolution. To ensure consistency across tracks, we referenced all line-of-sight (LOS) velocities to GNSS stations maintained by Indonesia’s Geospatial Information Agency, following a plane‑fitting approach to align InSAR and GNSS frames. By combining ascending and descending geometries and constraining the north component with GNSS, we decomposed LOS velocities into vertical and east-west components. All ancillary datasets- including land cover, geology, elevation, and population projections- were resampled to the InSAR grid to enable integrated analysis across Java’s coastal zone. We define the low elevation coastal zone in this study as areas ≤50 m elevation.

To assess future exposure, we combined present‑day subsidence rates with high‑emission SLR projections (RCP8.5; ~0.73 m SLR by 2100) and 1 km population projections from Shared Socioeconomic Pathways (Wang et al. 2022). We assume current subsidence patterns persist spatially and temporally, providing a reasonable worst‑case scenario for long‑term coastal risk.

Our results reveal widespread and substantial subsidence across Java’s northern coastal plain where unconsolidated alluvium and clay‑rich deposits dominate, and in many locations, subsidence rates are up to ten times higher than SLR, underscoring the dominance of local geological and anthropogenic processes in shaping coastal risk. Specifically, 7 million people and 10% of urban areas are subsiding faster than 14 mm/yr, and 5% exceed 28 mm/yr, and several cities exhibit extreme rates- Pekalongan and Demak show median subsidence of 87 mm/yr and 43 mm/yr, respectively.

When combined with projected SLR, subsidence dramatically increases future exposure- we estimate that by 2100 and under the worst-case socioeconomic pathway, if we do not consider VLM, half a million people will be below sea level. However, when considering the relative sea level change due to SLR and VLM, this number increases by six times, to over 3 million people.

This study demonstrates that failing to incorporate VLM leads to systematic underestimation of populations and infrastructure at risk from rising seas. Our integrated geodetic-geological-socioeconomic framework highlights the urgent need for locally tailored adaptation strategies, including groundwater regulation, sustainable urban development, and targeted coastal protection. As coastal megacities continue to grow, multi‑hazard assessments that capture the interaction between climate‑driven and geological processes will be essential for safeguarding livelihoods, infrastructure, and cultural heritage across Indonesia and other subsiding LECZs worldwide.

Monitoring surface motion in agricultural landscapes remains crucial for understanding subsidence and groundwater-related processes that directly affect the persistence of these regions. Recent studies on peat pastures demonstrate that integrating long-term C-band InSAR data from Sentinel-1 with contextual information and knowledge of surface motion characteristics enables the modeling, parameterization, and estimation of displacements, as well as the estimation of integer phase ambiguities. Within this framework, our main objective is to estimate InSAR-derived displacement parameters using the coherent periods of InSAR time series to characterize ground surface motion, including both seasonal uplift and subsidence.

In this study, we upscale these estimates across croplands, where land cover changes, particularly during the vegetation growth cycle, shorten the duration of coherent InSAR data and lead to discontinuous time series. We investigate the potential of integrating multi-frequency InSAR data, combining C-band (Sentinel-1) and L-band (SAOCOM), to improve coherence persistence and observation stability in agricultural regions. We use SAR data acquired from August 2022 to December 2025 over an agricultural area in eastern Groningen, the Netherlands. We select two test sites, each containing at least five parcels with various land cover types and a corner reflector on an integrated geodetic reference station (IGRS) serving as the local reference point.

By analyzing multi-temporal InSAR data across selected agricultural and peatland parcels, we evaluate temporal coherence statistics as both a quality metric and a physical descriptor of land surface dynamics. Preliminary findings indicate that L-band data maintain substantially sufficient coherence over vegetated parcels, yielding longer coherent periods. In permanent grasslands, the C-band provides a coherent InSAR phase time series for 49% of the observation period. In contrast, L-band provides 97%, improving continuous temporal coverage in the time series and reducing the unknown 'loss-of-lock' offset between coherent periods. Similarly, within cropped parcels, we can use 90% of the L-band and 37% of the C-band time series data. Note that, along a single orbit track, Sentinel-1 acquires images 2.5 times more frequently than SAOCOM. When combined, this multi-frequency framework increases the temporal density of reliable measurements, potentially enabling the upscaling of ground motion parameters across more heterogeneous vegetated regions.

Using observations from coherent periods in both C-band and L-band, we will estimate InSAR-based displacement parameters to describe surface motion in relation to precipitation and evaporation recorded at a nearby meteorological station. This parameterization is preferred over estimating the absolute phase because (i) it minimizes errors introduced by the unknown offset between coherent segments, and (ii) it allows the estimated parameters to model surface displacement during incoherent periods and outside acquisition times, assuming that soil components remain unchanged.

The results highlight the capability of multi-frequency InSAR to overcome decorrelation limitations in vegetated landscapes and to advance regional subsidence mapping strategies. The integration of ongoing C-band Sentinel-1 data with L-band SAOCOM and NISAR will create new opportunities for characterizing surface motion processes in agricultural and peatland environments.

In this study, a modified version of the approach proposed in [1] was applied to measure the spatio-temporal ground deformation pattern of the Store Vildmose area, in northern Jutland, Denmark. This area, once the largest raised bog in Denmark, now includes grasslands, cultivated fields, a peat extraction area, as well as remains of the old raised bog. Several restoration projects have been carried out and are ongoing on this site, lead by the Danish Nature Agency, as part of a national initiative to reduce greenhouse gas emissions.

Sentinel-1 Persistent Scatterer InSAR was applied to 5 corner reflectors deployed by Geopartner Inspections in December 2021, revealing seasonal deformation patterns, with amplitudes up to several centimeters. Furthermore, a Distributed Scatterer (DS) multitemporal DInSAR approach was carried out each year within the most coherent season, spanning autumn through winter, revealing a significant spatial variability of the ground deformation, which correlates very strongly with in situ peat thickness measurements.

In order to characterize, both the spatial and the temporal variability of the deformation, the approach proposed in [1] was applied, exploiting the slightly different periods in which neighbouring fields retain a sufficient level of InSAR coherence. Interferograms with a short temporal baseline were formed within each coherent time-window of a specified minimum duration, and disconnected deformation time-series were generated after 2D phase unwrapping and a weighted least-squares inversion. To increase the coverage of the measurements, the coherent time-window, and thus the interferogram network, were computed on a per-pixel basis. Precipitation and evapotranspiration data from the Danish Meterological Institute were then used as an input to the SPAMS model [2], which was in turn used to connect the disjoint time-series, using the approach proposed in [1].

The results were validated against the corner reflector PSInSAR timeseries, and demonstrate that DSInSAR, applied to temporally disconnected coherent windows, can provide valuable information on the spatial variability of the ground deformations, for areas which do not feature any PSs, and are thus not covered by the EGMS or by national PSInSAR-based deformation products. Recovery of the full deformation time-series for these areas, using a model-based approach, requires prior knowledge of the land cover, and a complementarity between the coherent time-windows of different portions of the area of interest.

References:

[1] P. Conroy, S. A. N. van Diepen, F. J. van Leijen and R. F. Hanssen, "Bridging Loss-of-Lock in InSAR Time Series of Distributed Scatterers," in IEEE Transactins on Geoscience and Remote Sensing, vol. 61, pp. 1-11, 2023, Art no. 5220911.

[2] P. Conroy, S.A.N. van Diepen, R.F. Hanssen, ”SPAMS: A new empirical model for soft soil surface displacement based on meteorological input data,” Vol. 440, 116699, 2023.