Improving the representation of hydrological processes for the next generation of earth system models

Earth system models (ESMs) have significant uncertainties when simulating the terrestrial water cycle in water-limited regions such as Mediterranean climate regions (MCRs), where groundwater is crucial for sustaining vegetation and regulating land-atmosphere interactions. Current ESMs rely on simplified free-drainage lower boundary conditions, which neglect dynamic groundwater processes and limit their ability to accurately represent water storage and availability. This thesis addresses this issue through three work streams. First, it assesses water resource changes in Northern Hemisphere MCRs (2002-2023) using GRACE/GRACE-FO satellite gravity measurements. A multiple linear regression framework identified climatic drivers of total water storage (TWS) variability. Results reveal negative TWS trends in the Euro-Mediterranean region and heterogeneous patterns in the western United States. Temperature is the primary driver of long-term decline, while precipitation dominates interannual variability. Projecting these sensitivities onto future warming scenarios suggests severe reductions in accessible freshwater and potential transitions toward semi-arid conditions in vulnerable subregions. Second, a novel dynamic groundwater parameterization is developed in ECLand (ECMWF Land Surface Model) replacing free drainage with explicit water table depth coupled to the unsaturated soil column. This parameterization enables bidirectional water exchange between saturated and unsaturated zones, calibrated using observational estimates of water table depth. Third, the effects of including dynamic groundwater are analyzed through sensitivity experiments. The dynamic scheme yields consistent improvements in all evaluated variables. Surface energy fluxes benefit most in shallow water table regions through increased root zone moisture and latent-sensible heat modulation. Subsurface hydrology improves, producing realistic runoff magnitudes. Improvement magnitude depends on water table depth, reflecting different groundwater-land interaction mechanisms. This study shows that explicitly modeling groundwater dynamics is essential for accurately projecting water availability in water-limited regions under future climate change.