Daily Ten-ST-GEE: An open access and fully automated 10-m LST downscaling system

Abstract

In remote sensing applications, data fusion is a combination of satellite images from different sources, aimed to improve the spatial and/or temporal resolution of the final output. This process also named spatial sharpening or spatial downscaling is required in several Land Surface Temperature-based (LST) studies, ranging from water table estimations and urban heating assessments to volcano activity monitoring. In this study, we propose a Google Earth Engine-based (GEE) daily 10-m LST retrieval system, named daily Ten-ST-GEE. It combines both MODIS and Sentinel-2 satellite products and uses the robust least squares statistical approach for data fusion. We validate the daily Ten-ST-GEE against two airborne TIR images over the Hat Creek region, in California, USA with a MAE of 2.27 °C. The cross-evaluation over the 1-km MODIS LST and the inter-comparison to the 30-m L8 LST in six different sites across the globe showed very promising results (i.e., average MAE less than 1 °C). As the daily Ten-ST-GEE is fully-automated, open-source, user-friendly and freely-accessible, it can be portable to other regions with diverse climatic regimes. This would greatly improve the downscaling initiatives and provide the scientific community with much-needed downscaled LST information. © 2022 Elsevier Ltd