Résumé
The catchments located in the High-Atlas Mountains of Morocco are providing most of the surface water resources for the Tensift basin around Marrakech, but due to their steep slopes and limited vegetation cover they are also subjected to severe flood events. This study is focusing on the Ourika (502km²) and the Rheraya (225km²) catchments, with altitudes ranging between 1000m and 4000m. In august 1995, a thunderstorm with a maximum intensity of 100mm/h caused a peak discharge of 1070m3.s-1 causing several deaths and damages. Following this event, from 2002 to 2012 a real-time telemetered monitoring network has been setup in the Ourika and the Rheraya catchments, to alert the population based on precipitation and discharge thresholds. In November 2014, several high-intensity rainfall events struck the southern regions of Morocco, including the Ourika and Rheraya catchments, and the data collected during these events is considered in the present work. The goal of this study is to move beyond the alert system towards the real-time modeling of floods, in order to provide short-term hydrological previsions based on measurements and meteorological forecasts, provided by Maroc-Météo with the Aladin-Bachir and Arome models. The first results are here presented, trying to answer three research questions: i) Is the magnitude or the occurrence of extreme floods changing with time? ii) Is the precipitation station network representative of the spatial variability of rainfall in these mountainous catchments? iii) Can hydrological models reproduce floods in an adequate manner in these complex catchments? These questions are addressed using the long-term (1962-present) discharge measurements available, together with hourly precipitation and discharge measured from the telemetered network for the year 2014 as inputs of global or distributed hydrological models. The results indicate no significant trends for the magnitude or the occurrence of floods in the catchments. Rainfall-runoff models are able to reproduce the floods that occurred in 2014 with a good accuracy, but strongly dependent on the structure of the models. A first evaluation of the meteorological forecasts is also presented.