Résumé
The widespread usage of the surface balance energy models coupled with remote sensing techniques proved their effectiveness in terms of accurately assessing evapotranspiration rates, crop water requirements as well as crop water usage/productivities. Still, such models require diverse inputs/elements, including climatic, topo-graphic, soil, and remote sensing datasets. The availability of these inputs is questionable and outdated in many developing countries. Thus, an Improved Surface Energy Balance Algorithm for Land (SEBAL), also known as SEBALI, was proposed and adapted for regions lacking soil-related datasets. Still, the usage of a standard calibration method in the surface energy balance models, yielding constant value throughout the cropping season, requires further revision and improvement. In this context, this study proposed a novel dynamic calibration approach to be included within SEBALI, related to the actual wind speed and relative humidity conditions. It was followed by the calibrations and validations of monthly evapotranspiration SEBALI values. Datasets were retrieved randomly from four countries (i.e. Belgium, Germany, Italy and United States) representing different climatic zones, between 2013 and 2014, and based on Eddy Covariance flux towers' outputs. When calibrated, results showed a Root Mean Square Error (RMSE) of 21.32 mm and an Average of Mean Error (AME) of 15.4 mm between monthly SEBALI outputs and flux towers' datasets, with an R-squared value of 88.4%. When investigating SEBALI outputs on different buffered zones around the flux towers along a changing Normalized Difference Vegetation Index (NDVI), ET values were poorly correlated (i.e. R-squared lower than 60%) in any buffer zone outside the parcels' boundary. RMSE showed values larger than 40 mm/month, even at a buffer zone of 250 m. This was related to the diverse Land Use Land Cover (LULC) classes, generating different evapo-transpiration rates, found at the boundary of the selected parcels. With the proposed dynamic calibration, the enhanced SEBALI could be then implemented in any agricultural region missing soil-related datasets with high accuracy.