Spatial sensitivity of surface energy balance algorithms to meteorological data in a heterogeneous environment

Increasing demand for water security requires improved accuracy in water accounting. Quantification of actual evapotranspiration is an essential part of this accounting and is frequently derived from surface energy balance (SEB) models that combine satellite remote sensing and on-ground measurements of meteorological data. However, many of the world's major water supply catchments are highly heterogeneous with land types, vegetation communities, and topography varying spatially. We compared the performance of seven meteorological interpolation methods and three SEB algorithms (the Simplified Surface Energy Balance Index (S-SEBI), the Hybrid Dual-Source Scheme and Trapezoid Framework-Based Evapotranspiration Model (HTEM), and the Surface Energy Balance System (SEBS)), testing their sensitivity to meteorological and remotely sensed inputs in a heterogeneous environment. Under a two dimensional framework, accuracy of interpolation methods varied among SEB meteorological inputs, suggesting that combining methods could improve overall accuracy. SEB algorithms were influenced by the density, type, and variability of meteorological inputs and sensitivity analysis showed that wind speed and air temperature were almost as influential as surface temperature for HTEM and SEBS. SEBS was the most sensitive to meteorological variability caused by choice of interpolation method when analysed globally, while HTEM was the most sensitive to local meteorological variation at flux towers. S-SEBI's simple structure made it the least sensitive to meteorological inputs and interpolation methods. Continued improvement in spatially explicit interpolation methods, combined with increased densities of meteorological stations, will increase accuracy and confidence in remotely sensed SEB fluxes, contributing to improved water accounting in heterogeneous catchments