Around 30% of aquifers in the United States are already in critical condition, with early signs also appearing in several French basins. This deterioration, fueled by pumping that far exceeds the natural recharge rate, threatens to drive some underground reservoirs to a point of no return, where their physical structure collapses and cannot recover.
In response to this challenge, an international team has developed a pioneering method that combines satellite data from the Copernicus program, advanced statistical models, and neural networks to map aquifers with unprecedented resolution. This approach opens the way for more proactive and sustainable management of groundwater resources, both in the United States and Europe.
What the study reveals: Precise results and key figures
Researchers have identified more than 1,200 critical aquifer zones across North America, using a predictive model that integrates over 15,000 measurements of water levels, precipitation, and temperatures. Among these zones, 450 are at risk of 'point of no return' by 2035, according to simulations based on CMIP6 climate scenarios.
In France, application of the same algorithm has revealed more than 200 vulnerable basins, mainly in the Southeast and Paris Basin regions, where agricultural and urban demand is highest. Estimates indicate an average decrease in stored water volume of 12% to 25% by 2040 if current practices persist.
These results, published in the journal Nature Water, underscore the importance of continuous monitoring and rapid adaptation of water withdrawal policies. They also demonstrate that atmospheric data (precipitation, evapotranspiration) integrated at the local scale is essential for anticipating aquifer level drops.