Field scale partitioning of Landsat land surface temperature into soil and canopy components for evapotranspiration assessment using a two-source energy balance model

Application of the two-source energy balance (TSEB) model to satellite data requires the definition of canopy (Tc) and soil (Ts) temperatures for the partitioning of the latent heat flux into transpiration (T) and evaporation (E) components. In this study, we evaluate the possibility of directly separating the satellite land-surface temperature (LST) into soil and canopy components, removing the need to adopt the iterative solution currently used in the TSEB approach. The method exploits contextual information at field scale, under the assumption that the field is homogeneous and characterized by spatially uniform Tc and Ts values. The approach was tested on a set of fields in California with typical woody perennial Mediterranean crops, and it was compared against the outcomes of two other standard versions of TSEB, as well as in-situ flux measurements. Overall, the proposed partitioning approach performs similarly to the standard TSEB models in reproducing instantaneous surface energy fluxes (mean absolute difference, MAD, on the order of 60 W m− 2 for turbulent fluxes), with only a slight tendency to overestimate transpiration. Differences in transpiration are mostly driven by divergences in modelled Tc (MAD between 2 and 3 °C), whereas differences in Ts are more limited (MAD between 1 and 2 °C). The main discrepancies were observed over low canopy coverage conditions (fractional cover lower than 35%), where small changes in Ts resulted in large differences in Tc. Despite this drawback, the results show that the method is a suitable alternative for a more straightforward operational field-scale application of the TSEB under suitable surface conditions.