Abstract Freshwater systems in Central Europe are increasingly vulnerable to climate change and anthropogenic pressures. This study investigates future dynamics of Lake Groß Glienicke, a groundwater‐fed lake experiencing rapid water‐level decline in recent decades. A fully integrated hydrologic model (HydroGeoSphere) was calibrated and validated to simulate groundwater‐surface water interactions. The model was forced with 43 high‐resolution EURO‐CORDEX climate projections. Lake evaporation was estimated using an energy‐balance approach, and machine‐learning models forecast groundwater abstractions and Havel River levels as boundary conditions. A meta‐modeling framework generated large ensembles to characterize uncertainty and test sensitivity to climate inputs. During 2008–2023, only about 5 percent of rainfall became groundwater recharge. Future projections diverge: wet scenarios maintain stable recharge and lake levels, the median shows gradual decline, and the driest scenario projects near‐complete recharge suppression after 2030 and more than 2 m lake‐level decline by 2100. Observed lake and groundwater declines since 2015 follow the lower ensemble envelope, suggesting an emerging dry regime. In the driest scenario, annual lake evaporation exceeds 1,000 mm by late century, whereas median and wet scenarios show no significant evaporation trend. Seasonal dynamics shift, with peak water levels moving from April to June and prolonged autumn low‐water periods. End‐century lake‐level differences exceed 4 m, indicating considerable uncertainty. Radiation uncertainties, particularly incoming global radiation, propagate into evaporation estimates. Overall, changing precipitation dynamics and persistent droughts emerge as key external drivers, while the integrated hydrologic and machine‐learning framework reveals how these drivers propagate through recharge, groundwater‐lake exchange, and abstraction‐related boundary conditions to shape lake resilience.
Mahmoodi et al. (2026) studied this question.