Multi-model assessment of present and future mean and extreme river flows in major West African basins

Hydrological extremes pose increasing risks to rapidly growing and climate-sensitive regions of West Africa. This study presents a multi-model assessment of historical and future mean and extreme river discharge across the Senegal, Niger, Gambia, Volta, and Ouémé basins. Five distributed hydrological models (CHyM, HMF-WA, CWatM, MIROC-INTEG-LAND, and WaterGAP2.2e) are evaluated against observed discharge (1998–2007) and forced with Global and Regional Climate Models under CMIP5 (RCP2.6/8.5) and CMIP6 (SSP1–2.6/5–8.5) scenarios for 2015–2099. All models reproduce the unimodal seasonal cycle, with CHyM, HMF-WA, and WaterGAP2.2e showing the best overall performance. RCM-driven simulations improve seasonal flow representation compared to GCM forcing, although uncertainties remain for extreme flows. Ensemble spread analysis highlights substantial structural uncertainty, with coefficients of variation ranging from 0.36 to 0.52 and wide discharge ranges across models. In addition, variance decomposition shows that hydrological model structure is the dominant source of uncertainty (59–94%), while climate forcing contributes 6–41%. Future projections reveal strong regional contrasts. Near-term changes (2025–2049) indicate moderate increases in peak flows (∼20%). By the end of the century, mean discharge declines in the Senegal, Niger, and Gambia basins under high-emission scenarios, while the Volta and Ouémé basins show sustained increases. Notably, extreme flows intensify by 60–80% under SSP5–8.5, even where mean flows decrease. These findings demonstrate that extreme discharge can intensify independently of mean trends and emphasize the dominant role of model structure in uncertainty, with important implications for flood risk management and regional adaptation strategies.