Satellite-based method for estimating the spatial distribution of crop evapotranspiration: Sensitivity to the priestley-taylor coefficient

Abstract

This work discusses an operational method for actual evapotranspiration (ET) retrieval from remote sensing, considering a minimum quantity of ancillary data. The method consists in a graphical approach based on the Priestley-Taylor (PT) equation, where the dry soil and non-limiting water conditions are defined by land surface temperature (LST) and vegetation index (VI) space, both retrieved from remote sensing. Using ET tower flux measurements and Landsat 5 TM images of an irrigation scheme in southeast Spain, a sensitivity analysis of ET spatial distribution was performed for the period 2009-2011 with respect to: (i) the shape (trapezoidal or rectangular) of the LST-VI space; and (ii) the value of the PT coefficient, α. The results from ground truth validation were satisfactory, both shapes providing similar performances in estimating ET, with root mean square error ~30 W/m2 and relative difference ~10% with respect to tower-based measurements. Importantly, the best fit with ground data was found for a close to 1, a somewhat different value from the commonly used value of 1.27, indicating that substantial error might arise when using the latter value. Overall, our study underlines the importance of a more precise knowledge of the actual value of a coefficient when using ET retrieval methods based on the LST-VI space.