Assessment of High-Resolution LST Derived From the Synergy of Sentinel-2 and Sentinel-3 in Agricultural Areas

Abstract

This work explores the potential of obtaining high-resolution thermal infrared (TIR) data provided by the Sentinel-2 (S2) & Sentinel-3 (S3) constellation in a typical semiarid agricultural environment. Maps of land surface temperature (LST) with 10–20 m spatial resolution were obtained from the synergy S2–S3 in the Barrax test site in Spain, for a set of 14 different dates in the summers of 2018–2019. Ground measurements of LST transects covering a variety of croplands and surface conditions were used for a ground validation of the disaggregation approaches. A cross validation of the LST products was also conducted using Landsat-8/TIRS images. Two recent approaches exploiting the linkages between shortwave and thermal data were adapted and tested, with differences in the inputs, the physical-mathematical framework, or the treatment of the LST residuals, and two options for the original 1 km S3 LST data were considered. Despite the large range of temperatures registered (295–330 K), differences with observed values resulted in an average RMSE < 3.0 K and a negligible systematic deviation, showing good results even in small fields ∼1 ha. Results confirm the need for appropriate adjustment techniques of the LST residuals obtained to better capture the low temperature conditions. The systematic overestimations introduced by the use of the operational sea and land surface temperature radiometer L2 LST product, and the limitations associated with certain irrigation management are discussed. Results in this work offer a solution to the lack of high-resolution satellite TIR data, and provide new opportunities for LST applications in agricultural areas.

Keywords

• Copernicus
• disaggregation
• downscaling
• land surface temperature (LST)
• thermal infrared (TIR)