Sensors
- Sentinel-1 A/B (C-band SAR, ESA): 5.6 cm wavelength; 6-day repeat at mid-latitudes with both satellites operational. Interferometric Wide Swath mode gives 5 × 20 m ground range resolution. Phase sensitivity to line-of-sight displacement is roughly 2.8 mm per 2π cycle, making it well suited to the 1–10 mm/yr deformation rates typical of halokinetic settings. Free archive from 2014.
- ERS-1/2 and Envisat ASAR (C-band SAR, ESA archive): ERS data run from 1992; Envisat ASAR from 2002 to 2012. Together they provide a 30-year C-band archive enabling long-baseline time-series analysis and detection of trend changes. ERS tandem mode offered 1-day repeat, useful for high-coherence interferograms over vegetated terrain. ESA archive access is free for scientific use.
- COSMO-SkyMed (X-band SAR, ASI): 3.1 cm wavelength improves phase sensitivity relative to C-band and reduces the minimum detectable deformation rate. Spotlight mode reaches 1 m resolution, resolving dissolution features only tens of metres across. Revisit can be as short as 1 day with constellation tasking. Commercial licence required.
- TerraSAR-X / TanDEM-X (X-band SAR, DLR): Stripmap mode at 3 m resolution; Spotlight at sub-metre. The TanDEM-X global DEM (12 m posting, absolute vertical accuracy better than 10 m, relative better than 2 m) provides a baseline topographic model for separating topographic from deformation phase. Useful for generating precise reference DEMs over diapir crests.
Why salt refuses to stay still
Halite has a density of roughly 2,160 kg/m³ and a viscosity that, under geological stress, allows it to flow on timescales of thousands to millions of years. Where a salt layer is buried beneath denser clastic sediments, the density inversion drives Rayleigh-Taylor instability: the salt rises as diapirs, pillows or walls, deforming the overburden above. The Zechstein evaporite sequence beneath the southern North Sea and northern Germany represents one of the world's best-documented halokinetic provinces, with diapirs reaching the near-surface across the Netherlands, Germany and Denmark. The Dead Sea basin hosts a different but equally active system, where Lisan and Sedom salt formations produce both diapir rise and, critically, dissolution subsidence where freshwater contact dissolves salt at shallow depth.
Two surface signals emerge from these processes. Diapir rise produces broad domal uplift, typically a few millimetres per year, extending over kilometres. Dissolution produces localised subsidence bowls, sometimes only hundreds of metres wide, where the rate can exceed 10 mm/yr and where sudden collapse into sinkholes is a genuine endpoint. The two signals can coexist on the same diapir: the crest rises while the flanks, exposed to groundwater, dissolve and subside. Distinguishing them matters enormously for infrastructure siting, because the hazard profiles differ by an order of magnitude.
What InSAR actually measures, and where it struggles
Interferometric SAR measures the change in two-way travel time of microwave pulses between passes. A full 2π phase cycle corresponds to half the radar wavelength in line-of-sight displacement: 2.8 mm for C-band Sentinel-1, 1.55 mm for X-band COSMO-SkyMed. Persistent Scatterer InSAR (PS-InSAR) and Small Baseline Subset (SBAS) processing over long time-series can resolve mean velocities to 0.1–0.5 mm/yr in favourable conditions, which is well below the typical halokinetic signal. Published studies over the Zechstein basin, including work using ERS and Envisat stacks, have mapped diapir-related deformation at rates of 1–5 mm/yr with spatial patterns consistent with structural geological models.
Coherence is the practical constraint. Agricultural land loses coherence rapidly in C-band between 6-day passes; PS-InSAR depends on finding stable point scatterers such as buildings, roads or bare rock outcrops. Over the Dead Sea shoreline, where dissolution sinkholes are actively forming, the sparse built environment and salt-encrusted flats actually provide good coherence. Over forested or heavily farmed terrain in the Zechstein province, distributed scatterer methods (DS-InSAR) or X-band data with shorter decorrelation times become necessary. Atmospheric water vapour is the other limit: a 1 km column of tropospheric anomaly can introduce several centimetres of apparent phase delay, which must be corrected using ERA5 reanalysis or GACOS tropospheric correction before slow halokinetic signals are interpretable.
The Dead Sea as a field laboratory
The western shore of the Dead Sea has become one of the most intensively studied dissolution-subsidence environments on Earth, for reasons that are partly scientific and partly urgent. The lake level has dropped more than 35 m since the 1960s, exposing shallow Holocene salt layers to freshwater infiltration from the surrounding aquifer system. The result is a proliferation of sinkholes, with thousands documented along the western shore. Published InSAR studies using ERS-1/2 and Sentinel-1 data have detected precursory subsidence bowls 50–200 m wide and 10–50 mm/yr in amplitude appearing months to years before visible sinkhole collapse. The spatial pattern is not random: subsidence clusters follow the boundary between the freshwater-saltwater interface in the shallow subsurface.
This precursory signal is the analytically important finding. A subsidence bowl that deepens at 20 mm/yr over two years and then accelerates sharply is a recognisable collapse precursor in the InSAR record, even if the surface shows no visible crack. The detection limit is set by coherence and atmospheric noise rather than by the physics of the deformation itself. Over the Dead Sea salt flats, coherence is high enough that Sentinel-1 6-day pairs produce usable interferograms, and time-series analysis over 18–24 months can resolve the acceleration phase. The same methodology applies, in principle, to any shallow evaporite system with adequate radar coherence.
Zechstein basin: slower signals, longer archives
The Zechstein diapirs of northern Germany and the Netherlands present a different analytical challenge. Deformation rates are generally slower, 1–3 mm/yr of uplift over the diapir crest being typical, and the signal must be separated from other sources of ground motion including groundwater extraction, peat compaction and post-glacial isostatic adjustment. The 30-year ERS-to-Sentinel C-band archive is the critical asset here. By combining ERS (1992–2000), Envisat (2003–2010) and Sentinel-1 (2014–present) stacks, it is possible to construct a multi-decade velocity field and test whether halokinetic rates have been stationary or have changed in response to industrial activity such as gas extraction from overlying reservoirs.
Several published studies have used this archive to map deformation over specific Zechstein structures in Germany and the Netherlands, identifying both uplift over active diapir crests and subsidence in rim synclines where salt has been evacuated. The spatial resolution of Sentinel-1 wide-swath InSAR, at 5 × 20 m in range and azimuth, is sufficient to resolve features at the kilometre scale but will miss dissolution features narrower than roughly 50–100 m unless X-band data are added. For regulatory purposes, where a gas storage facility or waste repository is sited above a Zechstein structure, the combination of multi-decade velocity trend and current Sentinel-1 monitoring provides a defensible evidence base.
Honest limits of the method
InSAR measures displacement in the satellite's line of sight, not in three dimensions. Separating vertical from horizontal motion requires at least two look directions, typically ascending and descending passes, and even then the north-south component is poorly constrained by C-band SAR. For a diapir producing purely vertical uplift this is manageable, but where lateral salt flow drives horizontal surface strain the single-geometry interpretation is ambiguous.
Vegetation is a persistent problem. C-band coherence over agricultural fields in the Zechstein province can drop below 0.3 within a single 6-day interval, making PS-InSAR dependent on sparse built infrastructure. L-band SAR, such as ALOS-2 PALSAR-2, penetrates vegetation canopy better and maintains coherence over longer baselines, but its 46-day repeat limits temporal resolution of fast-evolving dissolution features. Cloud cover does not affect SAR directly, but the atmospheric correction quality depends on auxiliary meteorological data that may itself carry uncertainty. Finally, InSAR detects surface displacement; it does not image the salt body directly. Structural interpretation requires integration with seismic reflection data, borehole logs or gravity surveys to constrain the subsurface geometry that explains the surface signal.
From interferogram to decision
The workflow from raw SAR acquisition to a usable deformation product involves several steps with non-trivial choices: orbit baseline selection, co-registration, phase unwrapping, atmospheric correction and time-series inversion. Each step introduces uncertainty that should be propagated to the final velocity estimate. A credible halokinetic monitoring product reports not just a velocity map but confidence intervals, coherence masks and the sensitivity floor below which signals cannot be distinguished from noise.
Satellize processes Sentinel-1 and archive SAR stacks for clients requiring ground motion monitoring over evaporite-prone geology, applying SBAS and PS workflows with GACOS atmospheric correction. For a client assessing infrastructure risk above a Zechstein structure, the concrete output is a georeferenced velocity field, an anomaly report flagging any pixel cluster exceeding a client-defined threshold, and a time-series plot for selected points of interest. If you have a specific structure or licence area in mind, the right next step is to share its coordinates so we can assess archive coherence and recommend the appropriate sensor combination before any processing commitment is made.
Typical figures
| Spatial resolution (Sentinel-1 IW mode) | 5 m (range) × 20 m (azimuth) ground range detected; PS-InSAR velocity products typically gridded at 20–100 m |
| Spatial resolution (COSMO-SkyMed Spotlight) | ~1 m; resolves dissolution features as small as 20–50 m across |
| Revisit period | 6 days (Sentinel-1 A+B at mid-latitudes); 1–4 days (COSMO-SkyMed constellation tasking); 35 days (ERS-1/2 standard) |
| Minimum detectable mean velocity | 0.1–0.5 mm/yr (PS-InSAR over stable coherent targets, long time-series); ~1–2 mm/yr (SBAS over distributed scatterers) |
| Radar frequency / wavelength | C-band: 5.405 GHz / 5.6 cm (Sentinel-1, ERS, Envisat); X-band: 9.6 GHz / 3.1 cm (COSMO-SkyMed, TerraSAR-X) |
| Archive depth | C-band: 1992–present (ERS → Envisat → Sentinel-1); X-band commercial: ~2007–present |
| Swath width | 250 km (Sentinel-1 IW); 40 km (COSMO-SkyMed Stripmap); ~10 km (Spotlight modes) |
| Atmospheric correction | ERA5 reanalysis or GACOS tropospheric delay maps; residual error typically 2–5 mm per scene |
| Deliverable formats | GeoTIFF velocity rasters, shapefiles of anomaly polygons, CSV time-series per point of interest, PDF anomaly reports |
| Typical processing latency (operational monitoring) | 24–72 hours after SAR acquisition for near-real-time interferogram; full time-series update monthly or quarterly |
Analytics Satellize can run
| Multi-decade mean velocity field | SBAS time-series inversion combining ERS, Envisat and Sentinel-1 stacks; GACOS atmospheric correction applied per scene | GeoTIFF velocity map (mm/yr, line-of-sight) with per-pixel uncertainty; PDF structural interpretation report |
| Persistent Scatterer network over built infrastructure | PS-InSAR (Ferretti et al. approach) applied to Sentinel-1 or COSMO-SkyMed stack; phase history modelled as linear velocity plus seasonal and atmospheric terms | Shapefile of PS points with velocity, standard deviation and time-series; flagged points exceeding client-defined threshold |
| Dissolution sinkhole precursor detection | Spatial clustering of accelerating subsidence bowls in SBAS time-series; acceleration onset flagged when rate exceeds 2× background in a 6-month window | Alert GIS layer updated on each new Sentinel-1 acquisition; email notification with coordinates and time-series plot |
| Ascending/descending decomposition into vertical and east-west components | Geometric decomposition using incidence angles of two independent InSAR geometries; north-south component acknowledged as unresolved | Separate vertical and horizontal velocity GeoTIFFs; uncertainty ellipses per grid cell reported in metadata |
| Diapir crest delineation and structural boundary mapping | Spatial gradient analysis of velocity field to locate hinge zones between uplift and subsidence; comparison with available gravity and seismic horizon data where provided by client | Shapefile of inferred structural boundaries; annotated cross-section diagram |
| Baseline coherence assessment for new areas of interest | Archive coherence map computed from existing Sentinel-1 pairs over client's area of interest; identification of PS-dense zones and vegetated low-coherence zones | One-page feasibility memo with coherence map, recommended sensor and processing strategy, and indicative detection limits |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.