Evaluating one- and two-source energy balance models in estimating surface evapotranspiration from Landsat-derived surface temperature and field measurements

Abstract

This article compares one- and two-source energy balance (OSEB and TSEB) models in the estimates of surface energy components using Landsat imagery and surface measurements acquired from an experimental field at Yucheng Station in Northern China. Compared to surface measurements, similar performance between the TSEB and OSEB models has been observed for estimated surface net radiation and soil heat flux. The root mean square difference (RMSD) is within 14–39 W m−2 in both the TSEB and OSEB models. The residual energy (E R) correction method yields the best agreement in comparisons of the sensible (H) and latent (LE) heat fluxes estimated using both the TSEB and OSEB models to the eddy covariance (EC) system measurements. The TSEB model is shown to greatly outperform the OSEB model in reproducing surface H and LE measurements. Cirrus clouds are likely responsible for the surface temperature retrieved from the enhanced thematic mapper plus (ETM+) sensor being lower than air temperature on days of the year (DOYs) 178 and 218 of 2009. This atmospheric stability is contrary to the unstable atmosphere that the EC measurements observe. If data on these two days are excluded and the E R correction method is applied, when comparing the estimated H and LE to the EC measurements, RMSD is within 55 W m−2 in the TSEB model and is larger than 97 W m−2 in the OSEB model.