Evapotranspiration Analysis using a Zonal Temperature-Vegetation Remote Sensing Approach

Abstract:

Remote sensing methods based on temperature-vegetation relationships have been widely used to analyze regional evapotranspiration (ET) patterns. These methods essentially rely on the identification of extreme soil moisture conditions (i.e., cold and hot pixels) as determined from the outer boundaries a triangle or trapezoid-shaped scatterplot of radiometric surface temperature and vegetation fraction. However, the validity of these pixels has been increasingly questioned in studies that involve heterogeneous climatic and land cover conditions, compromising the performance of remote sensing-based ET estimates in large areas. Here we developed a geospatial analysis scheme to discretize a targeted area into climate-land zones based on precipitation, radiation, wind, humidity, soil, and vegetation factors. Zonal ET estimates based on temperature-vegetation indexes were spatially aggregated to generate regional ET estimates over different time scales. We applied this approach to Jordan using Landsat and MODIS images over the period 2005-2014 and compared the ET estimates to MOD16 ET products and pan evaporation measurements. We explored contrasting hydrological conditions across a range of bioclimatic regions. Results show that our approach is able to capture regional ET heterogeneity in both water-limited and energy-limited environments and identify the hydroclimatic controls. Our comparisons of different methods shed light on the efficient estimation of agricultural water use in data-scarce, arid/semi-arid regions based on remote sensing.