Automation of choosing hot and cold pixels process in the estimate of the sensitive heat flow and evapotranspiration

Abstract

Characterize accurately and accurate evapotranspiration (ET) on a global scale has been one of the major determinant challenges in terrestrial ecosystem studies, climate dynamics and hydrological cycle. The goal in this study was to evaluate methodologies presented in the literature proposing a new methodology and automation in the process of selection of hot and cold pixels to need the sensible heat flux H and ET. The study was conducted in four different locations in the state of Mato Grosso, Brazil. And two areas of forest and two formed by pasture. For the H automation process we used the proposed Bastiaanssen et al. (1998a), Gao et al. (2011), and other vegetation using different levels of the literature suggests. The automation of the selection of hot and cold pixels process was successful, showing correlations above 0.83 for the estimates of H and 0.75 for estimates of ET. The identification of hot and cold pixels using the methodology by similarity of neighboring pixels using the average of pixels showed the highest correlations to the H in the study areas. LAI had the best indicator in the automated identification of hot and cold pixels to estimate the daily evapotranspiration (ET24h).