Land subsidence from groundwater and hydrocarbon extraction
InSAR time series from Sentinel-1 and decades of ERS/Envisat archive data can measure broad subsidence bowls caused by aquifer compaction and hydrocarbon withdrawal at millimetre-per-year precision, giving regulators the attribution evidence that piezometers alone cannot provide.
Sensors
- Sentinel-1 A/B (C-band SAR): 5.6 cm wavelength; Interferometric Wide Swath mode covers 250 km at 5 x 20 m resolution with 6-day revisit (single satellite, 12-day with one satellite operational). The workhorse for current subsidence monitoring; free archive from 2014.
- ERS-1/2 and Envisat ASAR (C-band SAR, archive): ERS data from 1992, Envisat to 2012. Together they provide a 20-year pre-Sentinel baseline essential for distinguishing long-term inelastic compaction trends from shorter elastic cycles. Resolution 25–30 m, 35-day repeat.
- ALOS-2 PALSAR-2 (L-band SAR): 23.6 cm wavelength penetrates vegetation and maintains coherence over agricultural land where C-band loses it. 3–46 m resolution depending on mode; 14-day repeat. Particularly useful in the Po Delta and other vegetated deltas where Sentinel-1 coherence degrades seasonally.
- ALOS PALSAR (L-band SAR, archive): 2006–2011 archive fills the Envisat-to-Sentinel gap at L-band. 10–100 m resolution. Useful for constructing continuous decadal time series in coherence-challenged terrain.
What a subsidence bowl actually tells you
Groundwater and hydrocarbon extraction produce broad, low-gradient deformation bowls, sometimes tens of kilometres wide but only a few metres deep over decades. That geometry is the signature. Tectonic signals are sharper; mining voids produce tighter, steeper-sided troughs. A wide bowl with a spatial pattern that mirrors the footprint of an abstraction wellfield is strong circumstantial evidence of poroelastic or inelastic compaction.
The distinction between elastic and inelastic compaction matters enormously for regulators. Elastic compaction reverses when heads recover; inelastic compaction, driven by irreversible collapse of fine-grained aquitard material, does not. InSAR time series can separate these behaviours by correlating the deformation signal with seasonal and multi-year piezometric head fluctuations. Where subsidence tracks head drawdown and recovers with head rise, the aquifer skeleton is still in the elastic range. Where cumulative subsidence outpaces head recovery, inelastic compaction is occurring. Published work on the Santa Clara Valley and the San Joaquin Valley has used exactly this comparison to estimate the fraction of total compaction that is permanent.
The archive is the argument
A single Sentinel-1 interferogram shows displacement between two passes. A time series built from hundreds of interferograms, processed by Persistent Scatterer InSAR (PS-InSAR) or Small Baseline Subset (SBAS) methods, resolves velocities to roughly 1–2 mm per year in stable urban areas, though agricultural land with low coherence can degrade that to 5–10 mm per year or worse. The ERS archive back to 1992 means that for well-studied basins, it is now possible to construct 30-year velocity maps.
The documented cases are striking. Mexico City's lakebed districts have subsided more than 9 metres since the mid-20th century, with InSAR studies using ERS and Envisat recording rates exceeding 300 mm per year in some zones. Tehran's southern districts showed rates of 200–250 mm per year in Envisat-era studies, driven by over-abstraction from the Tehran aquifer. The Central Valley of California saw localised subsidence exceeding 600 mm in a single year during the 2012–2016 drought, captured by Sentinel-1 and ALOS-2 and subsequently used by the California Department of Water Resources to prioritise Sustainable Groundwater Management Act interventions. The Po Delta has accumulated more than 3 metres of subsidence since the 1950s, a combination of natural sediment compaction and historical gas extraction, with Sentinel-1 now tracking residual rates of 10–30 mm per year in the most affected areas.
Honest limits of the method
InSAR measures displacement in the satellite line-of-sight direction, not vertical displacement directly. Converting to vertical requires either ascending and descending pass combination, which Sentinel-1's acquisition strategy supports over most land areas, or an assumption that horizontal motion is negligible. Over pure compaction bowls the horizontal assumption is reasonable at the bowl centre; at the margins it is not, and horizontal strain can crack infrastructure even where vertical rates look modest.
Coherence loss is the practical ceiling on what can be measured. Dense vegetation, ploughed fields, and flooding all destroy the phase relationship between passes. Agricultural basins in the Central Valley and Po Delta present seasonal coherence gaps that SBAS processing can partially bridge but cannot eliminate. L-band from ALOS-2 helps, but even L-band loses coherence over rice paddies. Atmospheric water vapour introduces apparent deformation signals of 10–30 mm per interferogram; time series averaging suppresses this, but a single interferogram over a basin should never be interpreted without atmospheric correction. Finally, InSAR gives surface displacement. Attributing that signal to a specific aquifer unit at depth requires integration with borehole stratigraphy, extensometer data where it exists, and a poroelastic model. The satellite data alone cannot tell you which layer is compacting.
From displacement map to abstraction licence
The regulatory application is the reason this capability is commercially interesting. Groundwater regulators in California, Iran, Italy, and Mexico have all used InSAR-derived subsidence maps as evidence in abstraction licensing and enforcement decisions. The logic is straightforward: if a wellfield's spatial footprint correlates with a subsidence bowl, and the time evolution of that bowl correlates with pumping volumes, the burden of proof shifts onto the operator to demonstrate otherwise.
Poroelastic modelling formalises this. The Terzaghi consolidation framework, extended to three dimensions, relates piezometric head change to compaction through the skeletal specific storage coefficient. InSAR provides the compaction; piezometric head data provides the forcing. Fitting the model gives an estimate of specific storage, which separates elastic and inelastic components and can project future compaction under alternative abstraction scenarios. This is the analysis that turns a displacement map into a number a water authority can act on.
Satellize runs this class of InSAR time series analytics on open Sentinel-1 data and can extend it into the ERS/Envisat archive for basins where decadal context is needed.
Hydrocarbon fields: a different geometry, the same physics
Reservoir compaction above depleting hydrocarbon fields follows the same poroelastic principles but with some differences worth noting. The deforming layer is typically deeper, the bowl is often more symmetric, and the signal is frequently smaller in magnitude but more spatially coherent than groundwater subsidence, which tends to be patchy because abstraction is distributed across many wells. The Groningen gas field in the Netherlands is the most extensively documented case: TNO and KNMI have used both levelling and InSAR to track subsidence of up to 36 cm at the bowl centre since production began, and the deformation signal has been linked to induced seismicity through stress modelling. The Wilmington oil field in Long Beach, California, produced more than 9 metres of subsidence before pressure maintenance by water injection arrested it in the 1960s, a case documented entirely by levelling but now used as a calibration benchmark for InSAR retrospective studies.
For active fields, Sentinel-1 provides the monitoring cadence. For older fields where production has ceased, the archive allows reconstruction of the compaction history and assessment of rebound, which is relevant for CO2 storage site selection and for understanding residual seismic hazard.
Typical figures
| Spatial resolution (Sentinel-1 IW mode) | 5 x 20 m (single look); PS/SBAS point density in urban areas typically 100–1000 points per km² |
| Revisit period | 6 days (Sentinel-1 A+B combined, where both operational); 12 days single satellite; 35 days ERS/Envisat archive |
| Minimum detectable velocity (urban, PS-InSAR) | 1–2 mm per year with sufficient time series length (>2 years); 5–10 mm per year over low-coherence agricultural land |
| Minimum detectable single-epoch displacement | ~5–10 mm per interferogram after atmospheric correction; smaller signals require time series averaging |
| Radar frequency / wavelength | C-band 5.405 GHz / 5.6 cm (Sentinel-1, ERS, Envisat); L-band 1.27 GHz / 23.6 cm (ALOS-2 PALSAR-2) |
| Archive depth | ERS-1 from 1992; continuous C-band archive to present; approximately 30 years of usable data over well-covered basins |
| Swath width | 250 km (Sentinel-1 IW); 70 km (ALOS-2 stripmap); 100 km (ALOS-2 ScanSAR) |
| Line-of-sight ambiguity | Ascending + descending combination required to separate vertical and east-west horizontal; north-south horizontal component not recoverable from polar-orbiting SAR |
| Typical processing latency (operational monitoring) | 3–7 days after acquisition for interferogram generation; time series updates monthly or per campaign depending on specification |
| Delivery formats | GeoTIFF velocity maps, shapefiles of PS/SBAS point clouds with velocity and time series attributes, CSV time series per point, PDF technical report |
Analytics Satellize can run
| Mean velocity map | PS-InSAR or SBAS time series inversion over Sentinel-1 or ALOS-2 stack | GeoTIFF and point-cloud shapefile showing mm/year line-of-sight and vertical velocity across the basin of interest |
| Displacement time series per monitoring point | PS or coherent pixel time series extraction with atmospheric phase screen correction | CSV or GeoPackage of cumulative displacement vs. date per point; suitable for import into piezometric head comparison workflows |
| Elastic vs. inelastic compaction partitioning | Correlation of InSAR displacement time series with client-supplied piezometric head records; residual after elastic fit attributed to inelastic storage | Technical report with scatter plots, estimated skeletal specific storage coefficients, and fraction of permanent compaction by sub-region |
| Decadal subsidence history | Multi-sensor archive chain: ERS (1992–2000), Envisat (2002–2012), ALOS (2006–2011), Sentinel-1 (2014–present), cross-calibrated by SBAS | Unified velocity and cumulative displacement maps spanning up to 30 years, with uncertainty estimates at epoch boundaries |
| Subsidence bowl attribution overlay | Spatial correlation of deformation footprint with wellfield or reservoir geometry using published poroelastic bowl models (e.g. Geertsma nucleus-of-strain) | GIS layer overlaying deformation contours on wellfield cadastre; PDF summary for regulatory submission |
| Infrastructure strain hotspot identification | Gradient analysis of velocity field to locate zones of differential settlement exceeding user-defined thresholds | Alert layer flagging linear infrastructure (roads, pipelines, canals) crossing high-gradient zones; updated per Sentinel-1 acquisition cycle |
| Future compaction scenario projection | Poroelastic forward model (Terzaghi/Biot framework) calibrated to InSAR-derived storage coefficients and client abstraction scenarios | Tabular and mapped projections of subsidence under continued, reduced, and ceased abstraction; uncertainty bounds included |
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.