Groundwater depletion monitoring for agricultural and rural land value
GRACE-FO gravity anomalies, downscaled with GLDAS hydrology, reveal aquifer systems under sustained depletion stress years before that stress appears in a soil survey or a water-rights register. Irrigated agricultural land dependent on a declining aquifer carries a forward-looking discount that conventional valuation ignores.
Sensors
- GRACE-FO (Gravity Recovery and Climate Experiment Follow-On): Measures changes in terrestrial water storage (TWS) by detecting gravitational anomalies caused by mass redistribution. Spatial resolution is approximately 300–400 km at the native spherical-harmonic level (degree-60 solutions), with monthly composites. Detects TWS changes of roughly 1–2 cm equivalent water height at basin scale. Operated jointly by NASA and DLR; data publicly available.
- GLDAS (Global Land Data Assimilation System): Land-surface model outputs (soil moisture, canopy water, snow water equivalent) at 0.25-degree (~25 km) resolution, updated monthly. Used to isolate the groundwater storage component from GRACE-FO TWS by subtracting surface and soil water contributions. Produced by NASA GSFC and NCEP; freely accessible via NASA Earthdata.
- Sentinel-1 (C-band SAR, InSAR mode): Interferometric SAR detects millimetre-to-centimetre surface subsidence caused by compaction of over-extracted aquifer sediments. 5 m × 20 m resolution (IW mode), 6-day revisit at mid-latitudes with two satellites. Subsidence rates as low as a few millimetres per year are detectable over multi-year stacks. A useful physical corroboration of the gravity-derived depletion signal.
- MODIS (Terra/Aqua, evapotranspiration products): MOD16 evapotranspiration product at 500 m resolution, 8-day composites. Sustained high ET over an irrigation district during drought years is a direct indicator of groundwater-fed irrigation stress on the aquifer. Provides the demand-side context that gravity anomalies alone cannot disaggregate.
What a gravity satellite actually measures, and why it matters for land value
GRACE and its successor GRACE-FO do not image the ground. They measure the gravitational pull of mass. When an aquifer loses water, the crust above it becomes slightly lighter, and the two satellites flying in tandem detect the change in their separation to within a few microns. The result is a monthly map of terrestrial water storage anomalies across the entire planet.
For a property analyst, the implication is straightforward. A parcel of irrigated farmland is worth what it can produce. What it can produce depends, in large part, on whether the water that feeds its pivots will still be there in ten, twenty or thirty years. A static soil survey tells you nothing about that. A GRACE-FO time series going back to 2002 (combining the original GRACE mission) tells you quite a lot. Aquifers in the Central Valley of California, the North China Plain, the Arabian Peninsula and the Northwest Sahara have all shown sustained negative TWS trends that are well-documented in peer-reviewed literature. Land overlying those systems carries a risk that is rarely priced into comparable sales.
Turning a 300-kilometre gravity blob into a parcel-level signal
The principal limitation of GRACE-FO is spatial resolution. A single pixel covers an area roughly the size of France. That is not directly useful for valuing a 500-hectare farm. Downscaling is the necessary step, and the published method is well-established: subtract GLDAS model estimates of soil moisture, snow and surface water from the GRACE-FO TWS anomaly to isolate the groundwater storage residual, then spatially disaggregate using ancillary hydrogeological data such as aquifer extent maps, well-log records and land-use layers.
The result is not a precise measurement of water depth at a single bore. It is a basin-scale depletion trend, expressed in centimetres of equivalent water height per year, attributed to the hydrogeological unit beneath a given property. That is enough to answer the question a buyer or lender actually needs answered: is this parcel sitting above an aquifer in structural decline, and at what rate?
Sentinel-1 InSAR adds a physical check. Aquifer compaction causes the land surface to subside, sometimes by several centimetres per year in heavily extracted basins. The San Joaquin Valley has shown subsidence exceeding 30 cm per year in some locations, documented by NASA JPL using Sentinel-1 data. Where subsidence and a negative GRACE-FO trend coincide, the depletion signal is corroborated by two independent physical mechanisms.
Honest limits: what the method cannot do
GRACE-FO cannot resolve individual aquifers within a basin where multiple hydrogeological units are stacked vertically. It measures total column water storage change. If a shallow, rapidly recharged alluvial layer sits above a deep fossil aquifer, the two signals mix. Separating them requires local well data or published hydrogeological surveys.
Monthly composites mean the method is not suited to detecting seasonal fluctuations as a valuation signal. The relevant signal is the multi-year trend, ideally over a decade or more of combined GRACE and GRACE-FO data. Short time windows produce noisy results dominated by wet and dry year variability.
InSAR subsidence mapping has its own constraints. Subsidence is irreversible compaction of fine-grained sediments; not all aquifer depletion produces measurable subsidence, particularly in consolidated rock aquifers. Dense vegetation and steep terrain also degrade InSAR coherence. MODIS ET is cloud-affected and represents a demand proxy, not a supply measurement. None of these sensors replaces a hydrogeologist's site assessment for a high-value transaction. They narrow the field of concern and flag parcels that warrant that deeper investigation.
The discount the market has not yet applied
Water-rights regimes vary enormously. In prior-appropriation states and jurisdictions, senior rights holders may continue pumping while junior rights are curtailed. In unregulated basins, the aquifer is a commons being depleted by collective action with no price signal attached. In either case, the satellite record often reveals a trend that has been running for fifteen to twenty years while land prices in the same basin have continued to reflect current productivity rather than forward water availability.
This is the core of the valuation argument. A GRACE-FO depletion trend, expressed as a rate of storage loss in km³ per year and compared against published estimates of aquifer storage capacity and recharge rates, supports a quantified scenario: at current extraction rates, how many decades of irrigation remain before pumping costs become prohibitive or yields collapse? That scenario feeds directly into a discounted cash-flow model for agricultural land.
Satellize applies this method as part of its broader satellite-data analytics work. The approach is directly analogous to the data-integration logic used in the Tonga crop-estimation programme, where multiple open sensors are combined to produce an output no single sensor could deliver alone.
From basin trend to investment decision
The practical workflow runs in three stages. First, screen a portfolio or target region against the GRACE-FO TWS anomaly archive to identify basins showing statistically significant negative trends over the full available record (2002 to present, with a gap in 2017–2018 between missions). Second, downscale using GLDAS and available hydrogeological data to assign depletion rates to specific aquifer units beneath target parcels. Third, cross-check with Sentinel-1 subsidence maps and MODIS ET anomalies to assess whether extraction stress is accelerating.
The output is not a single number but a risk tier: parcels where the gravity, subsidence and ET signals all point in the same direction sit in a different risk category from parcels where only one signal is elevated. That tiering informs due diligence scope, lender covenant terms and, in some cases, the decision not to proceed. The method is most powerful as a screening tool applied across a large portfolio before detailed site work begins, not as a replacement for it.
Typical figures
| GRACE-FO native spatial resolution | ~300–400 km (degree-60 spherical harmonic solutions); basin-scale analysis only at native resolution |
| GRACE-FO revisit / temporal resolution | Monthly composites; combined GRACE + GRACE-FO archive from April 2002 to present (gap mid-2017 to mid-2018) |
| Minimum detectable TWS change | ~1–2 cm equivalent water height at basin scale; smaller signals require multi-month averaging |
| GLDAS spatial resolution | 0.25 degrees (~25 km); monthly outputs for soil moisture, snow and canopy water partitioning |
| Sentinel-1 InSAR resolution | 5 m × 20 m (IW mode); 6-day revisit at mid-latitudes; subsidence detection to ~2–5 mm/year over multi-year stacks |
| MODIS ET product resolution and cadence | 500 m, 8-day composites (MOD16); affected by cloud cover; archive from 2000 to present |
| Downscaled groundwater anomaly output resolution | Aquifer-unit level, typically 1–25 km depending on hydrogeological data availability; not parcel-precise |
| Archive depth | GRACE + GRACE-FO: ~22 years; Sentinel-1: from 2014; MODIS: from 2000 |
| Delivery format | GeoTIFF depletion-rate rasters, GIS polygon overlays by aquifer unit, PDF risk-tier report, CSV trend tables |
Analytics Satellize can run
| Basin-scale TWS depletion trend map | GRACE-FO mascon or spherical-harmonic solutions differenced over user-defined period; linear trend fitting with seasonal decomposition | GeoTIFF raster of cm/year TWS change; PDF summary with statistical significance by basin |
| Groundwater storage anomaly (GWSA) time series by aquifer unit | GRACE-FO TWS minus GLDAS soil moisture, snow and canopy water; residual attributed to hydrogeological units from published aquifer maps | CSV time series per aquifer unit; GIS polygon layer with mean depletion rate and trend confidence interval |
| Surface subsidence map (InSAR-derived) | Sentinel-1 Persistent Scatterer or Small Baseline Subset (SBAS) InSAR over multi-year stack; displacement time series extraction | GeoTIFF subsidence-rate map (mm/year); overlay with aquifer unit boundaries for corroboration scoring |
| Irrigation demand stress indicator | MODIS MOD16 ET anomaly relative to long-term baseline; elevated ET during low-precipitation years flagged as groundwater-fed extraction signal | Annual ET anomaly raster; time-series chart per land parcel or irrigation district |
| Aquifer depletion risk tier by parcel | Composite scoring of GWSA trend, InSAR subsidence rate and ET anomaly; parcels binned into low / moderate / high / critical depletion exposure tiers | GIS polygon layer with risk tier attribute; due-diligence summary report per target parcel or portfolio |
| Forward water-availability scenario | Current GWSA depletion rate extrapolated against published aquifer storage capacity and recharge estimates; scenario expressed as years to critical pumping-cost threshold under current extraction | Scenario table and chart per aquifer unit; input-ready parameters for DCF land-valuation model |
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.