Thermodynamically constrained closed-form surface energy balance using medium-resolution remote sensing for efficient evapotranspiration mapping

Medium-resolution (10-100 m) satellite evapotranspiration (ET) products are rapidly advancing agricultural water resources research and management, however, underperformance across non-agricultural land cover limits research and application potentials more broadly. These inconsistencies are the result of multiple factors, including model structure and representation of ET dynamics across space and time. In regionally expansive land covers such as forests and shrublands, ET is primarily governed by equilibrium radiative energy exchange, whereas in croplands it is often amplified by advected heat from adjacent water-limited areas. While select models represent these processes, opportunities for improved conceptual and numerical representation are clear based on recent satellite ET model intercomparison studies. Here, we introduce a thermodynamic constraint in which ET is independent of aerodynamic conductance, enabling a closed-form analytical solution to the two-source surface energy balance under advection-free conditions. To account for advection, we conditionally incorporate an aerodynamic term where and when advection is significant. Landsat thermal, optical, and land cover data are used in combination with gridded meteorological data within the presented Radiation Advection Diffusivity-independent ET (RADET) modeling framework to predict ET. Performance is evaluated using in situ flux observations at daily and monthly scales across the contiguous United States (CONUS) along with intercomparisons to the widely used operational OpenET and MODIS products. Results indicate that RADET has superior performance across all land cover classes, with substantial improvements in forests and shrublands. Application of Landsat data with novel analytical solutions of the surface energy balance enables computationally efficient generation of medium-resolution ET products at scale with good performance across all land cover, advancing research and application potentials across many disciplines.