Model, satellite and ground-based estimates of evapotranspiration. A comparison in sub-humid tropical West Africa (Benin) within the framework of the ALMIP2 project

Evapotranspiration (ET) is a major sink term of the land surface water budget in tropical areas. However, accurate estimations of this term remain difficult as it can be measured directly only at the local scale, using complex experimental systems and data processing procedures (such as the eddy covariance method). Spatially-distributed satellite-based ET products or estimations from land surface models (LSM) are available, but their evaluation remains an issue, especially over those regions with poor ground measurement networks.

Under the auspices of the international AMMA (African Monsoon Multidisciplinary Analysis) project, comprehensive experimental monitoring systems have been set-up over three meso-scale domains in West Africa as a part of the AMMA-CATCH long term observatory. The ALMIP2 project (AMMA Land Surface Model Inter-comparison Project, phase 2) focuses on the evaluation of the water and energy cycles in current state-of-the-art LSMs for these 3 domains which sample large meridional climatic and ecological gradients. This talk focuses on the comparison of ET as computed by ALMIP2 models, current satellite-based products (e.g., ALEXI and MODIS) and eddy covariance data on the Upper Oueme watershed (Benin). The uncertainties on the observations and the multi-model spread are accounted for in the analyses. ALMIP-simulated ET generally matches the other estimates (satellite and ground measurements) in the rainy season when ET is close to the potential evaporation rate, but is generally underestimated in the dry season, especially just before the monsoon. The spatial pattern of ET is highly variable among these estimates. The eddy covariance measurements available at the local scale, and prior knowledge of the hydrological processes in the area (especially the below-ground components of the water cycle) are used to highlight the possible causes of the discrepancies between the various estimations of ET.