Mapping super high resolution evapotranspiration in oasis-desert areas using UAV multi-sensor data

High spatial resolution maps of evapotranspiration (ET) for precision agricultural irrigation, water resource management are increasingly important in the context of climate change. Here, we conducted extensive unmanned aerial vehicle (UAV) experiments in oasis-desert areas with three 2 km x 1 km regions during two growing seasons. The spatially distributed land surface temperature (LST), albedo and leaf area index (LAI) were retrieved from thermal infrared, multispectral and LiDAR (light detection and ranging) multi-sensor data, and the surface energy balance system (SEBS) model with an optimized surface roughness length for heat transfer was utilized to estimate ET with a super high resolution (SHR) of 0.2 m. The model's outputs were validated using the measurements from the eddy covariance (EC) with a source area of hundred meters and optical-microwave scintillometer (OMS) with a source area of approximately 2 km x 1 km. They yielded mean root mean square error (RMSE) values of 53 and 45 W/m2 for sensible heat flux (H) and 70 and 65 W/m2 for latent heat flux (LE), respectively. The seasonal characteristics of LE distributions indicated the difference of LE in oasis-desert areas varied with the growing period. These sub-pixel differences in the cropland, wetland, and desert experimental areas were caused mainly by the agricultural activities, the vegetation coverage, and the topographic peaks and valleys, respectively. This study can potentially provide spatially sub-pixel information on agricultural and ecological water management and genuinely bridge the spatio-temporal scale gap between field observations and satellite remote sensing.