Résumé
Urban catchments are very complex systems due to urban development. In these contexts, the intenseurban expansion induces local changes in the hydrological cycle and increases flood risk by extending impervious surfaces and altering the natural watercourses. Therefore, modelling them ischallenging task specially in data-sparse regions. Indeed, a compromise has to be reached to take into account the heterogeneity while avoiding over parameterisation. The Oued Fez catchment located in the northern part of Morocco is a typical example of urban catchments: its land cover is highly heterogeneous and has channelised water courses. In this work, the ATHYS modelling platform is used to simulate 59 selected flood events monitored on the urban and peri-urban parts of the Oued Fez catchment over the 2008-2018 periods. Two production functions, SCS (Soil Conservation Service) and a linear reservoir are combined with the lag and route transfer function. The drainage network is derived based on a high-resolution Digital Elevation Model (Source Dem from ALOS PALSAR RTC, https://asf,alaska,edu/datasets/sar-data-sets/alos-palsar/, 12,5m resolution). The road network is used as a proxy of the stormwater network to force the drainage directions along the motorways and three types of land use classes are used to parameterize the two production functions. Tests are carried out at hourly and sub-hourly time steps using both the natural and modified drainage networks. The main results show that using the road network and minimal land use classes improves model performance. The field data also shows that despite lower rainfall values, the highest peak flow values are recorded in 2017/2018. This is due to stream channelisation and increased stormwater network coverage.