Monitoring evapotranspiration over the Alpilles test site by introducing remote sensing data at various spatial resolutions into a dynamic SVAT model

Abstract : Remote sensing estimation of evapotranspiration (ET) was done by combining remote sensing data and the ISBA soil‐vegetation‐atmosphere transfer model over the Alpilles test site. We tested the possible use of low resolution data (∼1km) to derive leaf area index (LAI) at the field scale using a disaggregation method. Disaggregated LAI were then used as inputs of ISBA for monitoring ET for 9 months. Estimation of LAI and ET were first performed at high resolution for being used as reference for evaluating the use of low resolution data. Estimations of LAI at high spatial resolution using an artificial neural network (ANN) algorithm were in very good agreement with ground measurements. At low resolution, we found that it was possible to estimate accurately LAI for the most frequent types of vegetation cover, wheat and sunflower, but not for the other types. However, the estimation of ET from disaggregated low resolution data was found to be quite accurate for any type of vegetation cover (the comparison to high resolution estimation was good). ISBA simulations were eventually compared to independent estimates of ET using thermal infrared and a simplified energy balance equation showing large discrepancies in some areas or for some crop types: these corresponded to area with soil characteristics being different from those used in the simulation and to crops which were irrigated (irrigation inputs were not accounted in the simulations). This study enlightened the possible use of low resolution data for monitoring crop evapotranspiration at the field scale and the possibility of identifying areas with soil having contrasted water behaviour and irrigated crops.