A fully remote sensing-based implementation of the two-source energy balance model: an application over Mediterranean crops

Abstract

Applications of the two-source energy balance (TSEB) scheme require either in-situ meteorological data to characterize the upper boundary conditions or the implementation of complex multi-scale approaches (ALEXI/DisALEXI). Over remote areas, detailed meteorological forcing (i.e., air temperature and wind speed) are often missing, limiting the quality of the simulated fluxes. To compute surface energy fluxes, the use of wet and dry boundary conditions, commonly referred to as hot and cold pixels, is a widely adopted strategy in thermal-based, single-source surface energy balance models for defining the relationship between satellite land-surface temperature (LST) and the surface-atmosphere temperature gradient. This contextual scaling approach reduces model sensitivity to biases in LST retrievals, but it has been previously tested within the TSEB modelling framework only in limited capacity. An automatic procedure for retrieving the two boundary temperatures is here proposed, removing the need for external meteorological data and leading to temperature values that are unbiased compared to ideal estimations (from in-situ observations) and characterized by deviations on the order of 1.5 and 4.5 °C for cold and hot conditions, respectively. Despite the lower accuracy in the hot pixel temperature, this does not seem to significantly affect the overall capability of the model to reproduce observed fluxes, with errors in instantaneous sensible and latent heat fluxes in the order of 60 W m−2 (slightly above 1 mm d−1 on daily evapotranspiration) over a set of 16 sites in the US and Italy, characterized by typical Mediterranean crops. The proposed TSEB implementation is fully remote sensing based, meaning satellite-consistent retrievals of air temperature and wind speed are obtained directly from information available within the satellite scene itself. This approach represents a suitable alternative to accurately model evapotranspiration and other surface energy fluxes in the absence of reliable meteorological data.