How Recharging Aquifers Can Reverse Coastal City Subsidence
A new study shows that cities sinking due to groundwater depletion can rise again by restoring their aquifers. By combining satellite data and artificial intelligence, researchers are opening the way for resilience strategies against sea-level rise.
Cities worldwide are sinking due to excessive groundwater extraction, exacerbating their exposure to heatwaves and sea-level rise. An international team has published a study demonstrating that if aquifers are recharged, the ground can regain its original height, offering relief to threatened metropolises.
What the study reveals: precise results and key figures
The analysis covers over a hundred geodetic monitoring stations across several continents. Researchers found that in areas where artificial recharge has been implemented, subsidence rates have stopped, and in some cases, the ground has regained several dozen centimeters in less than a decade. These uplifts, though modest globally, significantly reduce coastal flooding risk in low-lying zones.
The report also highlights strong regional variability: cities on compressible sediments recover faster than those on more rigid rocks. This nuance is crucial for adapting water management policies to each geological context.
Scientific explanation of the subsidence phenomenon
Ground subsidence, or settlement, results from an imbalance between the internal hydraulic pressure of groundwater and the load exerted by the weight of the ground. When water is extracted at rates exceeding its natural recharge, pressure maintaining soil pores decreases. Sediment grains, particularly clays and silts, compact under their own weight, causing a reduction in volume and consequently a drop in ground level.
This process is amplified in coastal environments where sediments are often highly permeable, and urban construction increases mechanical stress. Additionally, climate change intensifies drought episodes, reducing natural recharge from precipitation and exacerbating dependence on groundwater pumping.
How: research method and data used
To obtain these results, the team combined satellite data from the Copernicus mission with ground measurements from the global GNSS station network. A predictive model based on a neural network was trained using machine learning, integrating both ECMWF atmospheric data and local precipitation observations. Algorithms like GraphCast and Pangu-Weather were used to refine water infiltration and evapotranspiration predictions, reducing hydrological prediction uncertainty.