Low-flow resilience and anthropogenic influences in mid-sized catchments: Regionalised hydrological modelling in the Creuse basin (France)

Study region Accurate low-flow prediction in small, ungauged catchments is essential for sustainable water resource management, particularly in regions vulnerable to increasing drought severity. This study examines the heterogeneous river basins of the Creuse Department, France. Study focus The GR6J hydrological model was calibrated using a bi-objective function (log-NSE and percent bias) targeting summer low-flows at 12 gauged stations. The calibrated model was then forced by 100 stochastic hydroclimatic trajectories (1995–2024) and regionalised across 170 ungauged basins. New hydrological insights for the region Low-flow regimes show pronounced spatial heterogeneity. Specific discharges (QMNA) range from near-zero in many small headwater basins (< 20 km²), which frequently approach flow cessation, to more sustained flows in larger basins. Some medium‑sized basins also display unexpectedly low specific discharges due to limited storage, high evaporative demand and geological constraints. Basin area influences absolute discharge but fails to fully explain specific flow variability. Anthropogenic regulation further modulates resilience. Reservoir operations distort model parameters (particularly the slow‑storage component X6) and reduce transferability. Small hydraulic structures emerge as critical, yet under‑quantified, contributors to water retention. Empirical (QMNA P20) and probabilistic (QMNA5 GEV and log‑normal) indicators show systematic offsets, with log‑normal estimates proving more conservative, affecting abstraction thresholds. This is the first spatially explicit probabilistic assessment of low‑flow hazards and anthropogenic modifications in the Creuse. It supports adaptive water management under increasing climatic pressure.