Abstract:
The aim of the study was to test the possible adequacy of an ensemble model to reproduce the observed flows in the
Mékrou basin and in what measure this ensemble simulation could be used instead of a unique model. Single model applications
showed shortcomings in terms of simulating both high and low flows at the same time. Thereby, the models were calibrated
according to two different modes (high and low flows) and they were tested further through the elaboration of three various
ensembles. The observed hydrographs were separated in three parts each in order to evaluate with much precision which model
or ensemble fits best the hydrographs. On the basis of the Root Mean Squared Error (RMSE), the models and derived ensembles
were assessed using discharge data. In this paper, the comparison is made between models, mean and median ensembles. Good
results were obtained for all models and ensembles but the best ones were achieved by the mean-based ensemble.
