Effect of hydrological model selection in climate change impact assessment in the Dano tropical catchment (Burkina-Faso)

Applying a validated hydrological model is a common approach in climate change impact studies. The current study used an ensemble of five regional climate models and two hydrological simulation models (HBV-light and GR4J). Both models were successfully calibrated and validated with coefficients of determination-R2 and Nash and Sutcliffe efficiencies-NSE ranging between 0.64 and 0.88. Compared to the reference period (1976-2005), the projected temperature shows an increase for the future periods 2021-2100, whereas for the projected precipitation change, a mixed trend is expected. The projected discharge change is very similar to the precipitation signal. The results further indicate for some climate datasets that, the climate-induced discharge change is lower than 50 mm per year for both HBV-light and GR4J, while the inter-comparison of discharges between the two hydrological models indicates differences exceeding 100 mm per year. Therefore, the choice of the hydrological model overscores the impact of the projected climate change.