Abstract
Evapotranspiration (ET) is a fundamental phenomenon within terrestrial system processes that involves water on land and in the atmosphere. ET is one of the main components of energy production on Earth, and it controls the water balance. The determination of ET is very complex and requires an experienced modeler and a substantial number of working hours. Thus, we present a novel algorithm called the automated surface energy balance algorithm for land (ASEBAL) to automate the application of all stages of the surface energy balance algorithm for land (SEBAL) to estimate ET from a long time series of MODIS orbital images. In this study, ASEBAL was applied to automate the endmember pixel selection and estimation of ET for 282 images of the Ipanema River Basin, which is located in a semiarid area of Brazil. The mean execution time was 3 min for each image, with the mean daily ET ranging from 1.63 mm day(-1) to 6.22 nm day(-1), presenting a mean ET of 3.86 mm day(-1) and a standard deviation of 0.97 mm day(-1). The comparison between the manual and automated selections of the endmember pixels performed in 37 images showed an average difference of 7% in the values of the selected pixels. A mean difference of 0.28 mm between the manual and automatic ET estimates was observed. In addition, a pixel-by-pixel comparison yielded an average R-2 of 0.82. The proposed automated pixel selection procedure remarkably reduced the execution time and eliminated the possibility of human error. Thus, ASEBAL was shown to be an efficient tool for applications with many images and was also shown to be a less complex task, with lower error incidences and much shorter execution time than the traditional method used to apply the algorithm.
NSTL
Elsevier