Comparative assessment of two evapotranspiration models from earth observation data in arid agricultural regions

Abstract Accurate estimation of evapotranspiration (ET) and its components is crucial for sustainable water management in agriculture, particularly in semi-arid regions. This study presents the first direct comparison of ET products from WaPOR version 3.0 Level 3 (WaPORv3), which utilizes a thermal-optical fusion approach, and the Shuttleworth–Wallace Sentinel-2 (SW-S2) model. Unlike WaPORv3, which relies on Land Surface Temperature (LST) to derive moisture stress, SW-S2 is an original, fully optical approach that constrains substrate and canopy resistances using a Shortwave Infrared (SWIR)-derived water index (OPTRAM). By using single-sensor Sentinel-2 inputs at 20 m, SW-S2 offers key operational advantages over thermal-based systems, including simplified data harmonization, reduced latency, and the elimination of spatial noise often associated with thermal sharpening. In the absence of direct ET measurements and using Deming regression with WaPORv3 as methodological reference, both models show high agreement for total ET and canopy transpiration (Tc) across daily, dekadal, monthly, and seasonal scales at regional extents. Although the presence of systematic biases prevents a conclusion of strict statistical interchangeability, both products demonstrate strong operational consistency for regional irrigation management. Field-scale performance remains comparable for maize, potato, vineyards, and dense orchards, with seasonal ET values aligning with literature benchmarks from lysimeters, flux towers, and energy balance models. Soil evaporation (Es) exhibits lower agreement at regional scale due to differences in soil resistance parameterization and LST downscaling effects, though crop-specific field-level estimates improve. These findings suggest that SW-S2 is a promising alternative to thermal-based models for irrigation monitoring, showing consistency with WaPORv3 under the analyzed conditions, and potential utility in fragmented agricultural landscapes, where high-resolution, low-latency optical ET products can support operational water management.