Spatially Disaggregating Satellite Land Surface Temperature With a Nonlinear Model Across Agricultural Areas

Abstract

AbstractAccurate remotely sensed land surface temperature (LST) is a promising tool for predicting surface evapotranspiration (ET). The spatial resolution of commonly existing daily satellite products (i.e., Moderate Resolution Imaging Spectroradiometer [MODIS] LST) is ~1 km, which remains relatively low for used in estimating ET. This paper developed a model that disaggregates ~1‐km spatial resolution MODIS‐derived LST data to fine spatial resolutions of 250 m. The proposed model was achieved by using a spatial and temporal nonlinear strategy that contains the predictor variables of the Bowen ratio, the photochemical reflectance index, and the normalized difference vegetation index. The proposed disaggregation model was assessed mainly at two agriculture sites, including the Heihe River Basin in China and the Walnut Creek Watershed in the United States, during the growing seasons. The assessment procedure was conducted at both the field scale and the image scale in terms of disaggregated LST and ET. The statistical results demonstrated that the proposed model produced 250‐m LST and ET that matched better with the observed values and achieved more accurate LST and ET relative to other reference ones. Our study shows that surface moisture status and vegetation physiological dynamic are important factors in improving the LST disaggregation over the agriculture region. The results of this study have the potential to improve water resource management and sustainable water use.