Résumé
<p>The estimation of spatialized Evapotranspiration (ET) is of great interest to tackle various scientific issues in hydrology. It is particularly relevant in Sahelian Regions where there is a spatial scarcity of in-situ ET measurements, despite having key issues in water management and food safety. Nowadays, many ET products are available with different temporal and spatial resolutions. These products use a wide range of methods: empirical ET equations, Land Surface Models (LSM), energy balance models, in-situ measurements upscaling, neural networks or data fusion approaches. In this study, a quite exhaustive review of available ET products has been performed over Sahel. We have evaluated the potential of 18 ET products, both at local scale, using eddy-covariance measurements and a LSM calibrated on sahelian ecosystems, and at mesoscale, using the ALMIP2 models ensemble (20 LSM). Results show that only few products succeed at representing realistic temporal and spatial repartition of ET in sahelian context. The wide variety of temporal and spatial resolutions is also limiting the applications of such products, especially in eco-hydrology or agricultural management. Alternative methods are thus proposed to provide more reliable ET estimation over sahelian region by combining multi-sources ET estimations and satellite data. First, derived at kilometer scale using the EVASPA S-SEBI Sahel (E3S) contextual approach and MODIS Thermal Infrared (TIR) data on clear sky days. Then days with clouds have been filled with a data fusion algorithm. This algorithm uses ET estimates from the Amsterdam Methodology (GLEAM); which proved to have one of the best temporal representation of ET in our study area; and/or a simple parametric model (PAMEAS). Annual and seasonal ET estimates have been compared with the existing products on sahelian areas in Niger and Mali, and will be soon in Senegal too in the frame of the EVAP&#8217;EAU project (ICIREWARD Unesco Center). The E3S model and temporal interpolation techniques for daily ET estimations are also investigated in the mission group of the TRISHNA satellite mission (CNES-ISRO) which will provide TIR data and products with high spatial (60m) and temporal (3 days) resolution.</p>