Satellite Gravity Data Yields First Consistent Global Map of Groundwater Recharge

Beneath the world’s deserts, farmlands and rainforests lies a resource that sustains billions of people: groundwater. Yet despite its importance, scientists have never agreed on how much of it the planet’s aquifers actually receive each year. A new study published in Hydrogeology Journal by Vagner Ferreira of Hohai University and Christopher Ndehedehe of Griffith University takes a major step toward resolving that uncertainty. By combining satellite measurements of Earth’s shifting gravity field with a state-of-the-art land surface model, the researchers have produced a globally consistent estimate of groundwater recharge for the period 2004 to 2024, arriving at a figure of roughly 16,081.5 cubic kilometres per year. The result sits comfortably within the range of previous multi-study averages of 14,806 plus or minus 1,721 cubic kilometres per year, but it was achieved with a method that is far simpler, faster and more uniform than anything attempted before.

The innovation at the heart of the study is what the authors call the groundwater storage fluctuation, or GSF, method. Traditional approaches to estimating recharge are notoriously fragmented. Some rely on chloride mass balance calculations that require detailed chemical sampling of rainfall and soil water. Others depend on numerical models that must be calibrated against local field data, or on water-table fluctuation techniques that demand dense networks of monitoring wells. Because each technique carries its own assumptions, biases and data requirements, global syntheses have struggled to reconcile regional estimates into a coherent planetary picture. The GSF method sidesteps much of this complexity by working directly with monthly fluxes of groundwater storage anomalies, meaning it requires no seasonal templates, no pre-defined recharge events and no region-specific calibration.

The raw material for the method comes from an unusual but powerful source: two satellites flying in formation hundreds of kilometres above Earth. The Gravity Recovery and Climate Experiment, known as GRACE, and its successor GRACE Follow-On detect minute changes in the distance between the two spacecraft as they pass over regions of varying mass. When aquifers fill, the added water tugs slightly more strongly on the satellites; when aquifers are drained by pumping or drought, the gravitational pull weakens. These gravity signals, processed into estimates of terrestrial water storage, are then fed into the Catchment Land Surface Model of NASA’s Global Land Data Assimilation System, version 2.2. The assimilation constrains the model’s simulated groundwater storage with real satellite observations, producing a monthly record of groundwater storage anomalies across the entire globe at a quarter-degree resolution.

From these storage anomalies, the GSF method derives three quantities that water managers care deeply about: gross recharge, total discharge and net recharge. Gross recharge represents the total inflow of water into aquifers, while total discharge captures all outflows, both natural processes such as baseflow to rivers and springs, and human-related withdrawals. Without additional data, the two types of discharge cannot be separated, a limitation the authors acknowledge explicitly. Net recharge, the difference between the two, indicates whether an aquifer is gaining or losing water overall. Because the method operates on month-to-month changes in storage rather than absolute values, it is insensitive to many of the systematic errors that plague model-based estimates, and it can be applied identically in every basin on Earth, from the humid tropics to the driest deserts.

The spatial pattern the method reveals is striking. Recharge rates climb above 300 millimetres per year in humid tropical regions where abundant rainfall percolates through permeable soils, while falling below 100 millimetres per year across arid zones where evaporation competes fiercely with infiltration. Globally, the study finds a trend of 0.34 millimetres per year over the two decades examined, but this trend is not statistically significant, suggesting that on average the planet’s total recharge has remained remarkably stable even as its distribution shifts. That stability, however, masks dramatic swings at the regional scale. Arid basins show the highest interannual variability of any climate zone, a finding with profound implications for the communities that depend on them.

Three basins illustrate the point vividly. Australia’s Murray-Darling Basin, the agricultural heartland of the continent, exhibits a coefficient of variation of 48 percent, meaning its annual recharge nearly doubles or halves from one year to the next. Brazil’s São Francisco basin follows at 40 percent, and the Colorado basin in the United States at 36 percent. In these water-stressed regions, a single wet year can replenish aquifers substantially, while consecutive dry years can leave them depleted for extended periods. For farmers, city planners and environmental regulators, such volatility means that average recharge figures are dangerously misleading; sustainable pumping limits must account for the boom-and-bust rhythm of arid-zone hydrology rather than assuming a steady annual supply.

Of course, a new method is only as good as its validation, and the authors subjected their estimates to rigorous testing against independent data. Across Australia, they compared their satellite-derived recharge map with a high-resolution dataset built from the chloride mass balance technique, an entirely independent approach that estimates recharge from the accumulation of atmospheric chloride in soil profiles. The spatial agreement between the two was strong, with a rank correlation of 0.82. Zooming into the Murray-Darling Basin, the agreement held firm at 0.83, with a standard deviation of 32.1 millimetres per year. These numbers indicate that a method relying solely on satellite gravity and model assimilation can reproduce the spatial texture of recharge that laborious field campaigns have documented.

The underlying groundwater storage estimates were also checked against reality, using records from 259 monitoring boreholes across the Murray-Darling Basin. The correlation between the satellite-constrained storage anomalies and the in-situ observations was 0.47, with a standard deviation of 33.9 millimetres, a value that falls within the bootstrap uncertainty limit of 38.2 millimetres. While a correlation of 0.47 is moderate rather than spectacular, it is respectable given the enormous difference in scale between a point measurement in a borehole and a satellite footprint spanning hundreds of kilometres. The comparison confirms that the GRACE-constrained product captures genuine storage dynamics rather than artefacts of the modelling chain, even in a basin with complex geology and intense irrigation.

The authors are careful to spell out the limits of their approach. GRACE’s effective spatial resolution is approximately 300 kilometres, which means that grid-cell estimates in data-sparse regions or areas with low signal-to-noise ratios should be interpreted with caution. Small aquifers, localized recharge hotspots and narrow alluvial corridors simply fall below the satellites’ resolving power. The method also inherits uncertainties from the land surface model itself, particularly in regions where soil properties, vegetation and irrigation practices are poorly mapped. Nevertheless, the consistency of the global total with independent model ensembles, combined with strong regional validation, suggests the GSF framework offers a robust new baseline for planetary-scale groundwater accounting.

The timing of this work could hardly be more significant. Global assessments have documented rapid groundwater declines in many of the world’s major aquifers, alongside isolated cases of recovery driven by deliberate management intervention. As climate change intensifies droughts and alters precipitation patterns, the balance between recharge and withdrawal in arid and semi-arid basins will become ever more precarious. A globally consistent, satellite-based recharge estimate gives hydrogeologists, governments and international agencies a common yardstick for the first time, one that can be updated as new satellite data arrive. If the world is to manage its hidden water reserves wisely, it must first know how fast those reserves are being replenished. This study brings that knowledge within reach, transforming a patchwork of incompatible local estimates into a single, coherent view of the water flowing silently into the aquifers beneath our feet.

Subject of Research: Global estimation of groundwater recharge using the groundwater storage fluctuation method applied to GRACE-constrained GLDAS-2.2 CLSM data

Article Title: Towards global groundwater recharge estimation from GRACE-constrained GLDAS-CLSM data

Article References: Ferreira, V., & Ndehedehe, C. (2026). Towards global groundwater recharge estimation from GRACE-constrained GLDAS-CLSM data. Hydrogeology Journal. https://doi.org/10.1007/s10040-026-03169-1

Image Credits: AI Generated

DOI: 10.1007/s10040-026-03169-1

Keywords: groundwater recharge, GRACE, GLDAS, CLSM, groundwater storage fluctuation, satellite gravimetry, hydrogeology, Murray-Darling Basin, chloride mass balance, water resources, arid basins, remote sensing