Comparison of modeling approaches for flood forecasting in the High Atlas Mountains of Morocco

In the Mediterranean region, floods are causing extended damages to the population and infrastructures. In Morocco, only a few studies have been conducted to understand flood processes while the vulnerability to floods is high. The goals of this study are to compare two modeling approaches for floods using either lumped or spatial rainfall and also to evaluate hydrological forecast capabilities. The Rheraya research catchment is characterized with steep slopes, altitudes ranging from 1027 to 4167 m, and a strong variability of rainfall. The lumped and distributed models provided similar results and reproduced well a sample of six flood events recorded in 2014. However, the distributed model provided the best estimation of the initial conditions, estimated from the ESA-CCI satellite soil moisture product and the Antecedent Precipitation Index. The validation of the lumped and the distributed models, using ESA-CCI to initialize the models, provides a Nash coefficient of 0.61 and 0.63 respectively. Then, two meteorological forecasts provided by the AROME and ALADIN models were evaluated against observed precipitation to provide a hydrological forecast. The AROME forecast performed better but still with a strong bias compared to observed precipitation. Further research is needed to link quantitative precipitation forecasts with hydrological models in this type of catchment.