The utility and applicability of vegetation index-based models for the spatial disaggregation of evapotranspiration

Estimating evapotranspiration (ET) at finer spatial resolution is key for agricultural water management. Spatial disaggregation is often used to improve the resolution of ET estimated from satellite data. Two vegetation index-based ET disaggregation models: Disaggregation of solar radiation factor (DiSoRa) and Multi-Sensor data fusion-ET (MSDF-ET) were used for disaggregating ET estimated from Landsat at 30 m to a resolution of 3 m using images from PlanetScope constellation of satellites. Two vineyards, one in California, USA, and the other in Maharashtra, India, were chosen for the study. ET at 30 m was estimated using the two-source Soil Plant Atmosphere and Remote Sensing Evapotranspiration (SPARSE) model, and the validation was carried out using eddy covariance observations. The sensitivity of the disaggregation models to NDVI was tested. It was found that using NDVI from different sensors affected the accuracy of the disaggregated ET. The disaggregation models performed well consistently during the crop growing season. However, in the presence of any inherent stresses in the crop without getting reflected in the NDVI, the disaggregation models led to marginally increased errors than the coarse resolution ET. The results from this study indicate that finer-resolution ET estimates are feasible from disaggregation and may provide useful information for precision irrigation scheduling.