Evaluating the impact of satellite precipitation data uncertainties on streamflow and actual evapotranspiration simulations

Abstract

The uncertainty associated with satellite precipitation data poses a challenge for hydrological modelling in West Africa, where access to observed data remains difficult. This study aims to analyse these uncertainties in the performance of GR4J, in terms of its simulation of discharge (Q) and actual evapotranspiration (AET), through parameter transfer, recalibration and multi-objective calibration strategies. The analysis is carried out on four West African catchment areas using IMERG, GPCP, MSWEP and TAMSAT. GR4J is evaluated using the Kling-Gupta Efficiency and Percent Bias metrics. The use of parameters calibrated against observed rainfall reduces performance when the model is forced with these products. Recalibration improves the reproduction of discharge data, but involves parametric compensation. Multi-objective calibration reduces the effects of compensation and improves the consistency of the AET, whilst maintaining satisfactory performance in terms of discharge. These results show that multi-objective calibration enhances the reliability of hydrological simulations in West Africa.