Résumé
The objective of this study is the elaboration of a rainfall-runoff model capable of predicting future floods resulting from extreme rainfall events in a rugged Moroccan Mediterranean basin with marly lands. The integration of the deterministic hydrological modelling system, Hydrologic Engineering Center-Hydrologic Modelling System, with a Geographic Information System was employed to construct the hydrological model. The calibration and validation of the model were performed using gridded daily occurrences from the period of 2003 to 2010. The model’s performance was assessed using various statistical indices, including R-squared, root-mean-square error, percent bias, and Nash-Sutcliffe performance parameter. The results indicated a good agreement between observed and simulated flow, with an average R-squared of 0.81. The model exhibited low residual flow variation, averaging at 0.51, and achieved favorable Nash-Sutcliffe performance values, with an average of 0.69. However, in extreme events that include multiple flood hydrographs, the model tends to underestimate the initial one. Despite this limitation, the overall performance of the model was considered good for 50% of the statistical indices and excellent for an additional 25% of them. However, 25% of the indices indicated unsatisfactory performance. Future research could focus on improving the model’s accuracy by investigating indicators for predicting unscheduled releases from upstream dams, which could contribute to potential variations in observed flow. Also, it’s important to implement a reservoir control policy that optimizes storage capacity and effectively mitigates floods.