Improved floodplain modelling with FABDEM and Sentinel-2 earth observations in the middle valley of the Senegal River

Résumé fourni par la source

In the alluvial plains of large rivers, the study of flood dynamics is essential to appreciate water resource variations and preserve associated ecosystem services, in particular biodiversity, groundwater recharge and flood-recession agriculture. Hydraulic modelling provides valuable opportunities to simulate the dynamics of surface water flows but are challenged by the very flat topography and the sparse field observations, especially in Africa. By combining advances in earth observations (Digital Elevation Models and Sentinel-2 surface water areas), field observations (stage, flow gauging, river profiles) and hydraulic modelling (HEC-RAS), we aim to improve the understanding of surface water dynamics in the Senegal River floodplain. In this region, flood-recession agriculture is a complementary activity to irrigated agriculture and plays an important role in the subsistence of local populations.Recent, open-access DEMs (AW3D, COPDEM, FABDEM, NASADEM, SRTM, TanDEM) were compared against field observations revealing the superior performance of FABDEM (RMSE = 0.58). FABDEM was then pre-processed to recondition the elevations of the river bed based on field river profiles. The HEC-RAS model was calibrated to simulate the flow propagation from the Bakel to Diama over the period 2017-2020 and to accurately map flood-prone areas detected on Sentinel-2 imagery at the scale of individual depressions and the whole floodplain. Results show that the model reproduces flood dynamics with good accuracy, with KGE on water levels reaching 0.78 at Bakel and 0.65 at Matam gauging stations. The model also enabled the 2D representation of flooded areas, providing the first accurate representation of inundated areas in this floodplain, and their variations under climatic and dam construction scenarios. The excellent performance obtained with FABDEM highlights the enhanced opportunities it extends to develop hydraulic models of complex, poorly gauged floodplains, and support the management of water resources.Key words: Floodplain, HEC-RAS, remote sensing, hydraulic modelling, Senegal River, Middle valley.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.