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

Accurate estimation of groundwater recharge at regional to global scales is essential for sustainable water resource management, yet current methods lack consistency. This study introduces the groundwater storage fluctuation (GSF) method to derive gross recharge, total discharge (including both natural and human-related outflows that cannot be distinguished without additional data), and net recharge from GRACE-constrained GLDAS-2.2 CLSM groundwater storage (GWS) anomalies. Based on monthly GWS fluxes, the GSF method requires no seasonal templates or specific recharge-event identification. Global mean annual recharge for 2004-2024 is estimated at 16,081.5 km $$^{3}$$ /year, consistent with existing global hydrological models and multi-study averages (14,806 ± 1,721 km $$^{3}$$ /year). Recharge ranges from <100 mm/year in arid zones to >300 mm/year in humid tropical areas, with a non-significant global trend of 0.34 mm/year. Arid basins show the highest interannual variability, with coefficients of variation of 48% for the Murray-Darling Basin (Australia), 40% for São Francisco (Brazil), and 36% for Colorado (USA). Validation against an Australian chloride mass balance (CMB) dataset shows strong spatial agreement ( $$\\varvec{\\rho }$$ = 0.82). In the Murray-Darling Basin, CMB validation confirms this performance ( $$\\varvec{\\rho }$$ = 0.83 and a standard deviation, STD, of 32.1 mm/year). The underlying GWS, validated against 259 boreholes in the basin, correlates at 0.47 with an STD of 33.9 mm, within the bootstrap uncertainty limit of 38.2 mm. Results should be interpreted at GRACE’s effective resolution ( $$\\sim $$ 300 km), and grid-cell estimates in data-sparse or low signal-to-noise ratio areas should be used with caution.