Résumé
Crystalline aquifers are among the most complex groundwater systems, requiring adequate methods for realistic characterization and suitable techniques for improving the long-term management of groundwater resources. A tool is needed that can assess the aquifer hydrodynamic parameters cost-effectively. A model is presented, based on a groundwater- budget equation and water-table fluctuation method, which combines the upscaling and the regionalization of aquifer parameters, in particular specific yield (S y) in three dimensions (3D) and the recharge in two dimensions (2-D) from rainfall at watershed scale. The tool was tested and validated on the 53-km 2 Maheshwaram watershed, southern India, at a 685 m × 685 m cell scale, and was calibrated on seasonal groundwater levels from 2011 to 2016. Comparison between computed and observed levels shows an absolute residual mean and a root mean square error of 1.17 m and 1.8 m, respectively, showing the robustness of the model. S y ranges from 0.3 to 5% (mean 1.4%), which is in good agreement with previous studies. The annual recharge from rainfall is also in good agreement with earlier studies and, despite its strong annual variability (16 to 199 mm/y) at watershed scale, it shows that spatial recharge is clearly controlled by spatial structure, from one year to another. Groundwater levels were also forecasted from 2020 to 2039 based on the climate and groundwater abstraction scenarios. The results show severe water-level depletion around 2024-2026 but it would be more stable in the future (after 2030) because of a lower frequency of low-rainfall monsoons.