Résumé
Climatic and environmental change is a major political and scientific concern, even more for western Africa, already weakened by three decades of drought. This work aims at an optimal consideration of the spatiotemporal variability of land-use changes and rain in the monthly rainfall-runoff modelling of water resources of the upper Niger River. A detailed sensitivity analysis of the semi-lumped version of the GR2M model was first carried out, and then long-term land-use changes were mapped using Landsat images to estimate some spatiotemporal variations of one of the model's parameter. Lastly, rainfall data from Meteosat in 2004 and 2005 were used to get a better estimation of the spatial variability of the rain. This study allowed the identification of the possible ways of improving water resources modelling using the semi-lumped version of the GR2M model.