Downscaling of Thermal Images Over the Gaza Strip Using the Land Surface Temperature—Spectral Indices Relation: Case Study; Hot, Arid, and Semi-Arid Areas

Abstract

Recent research shows that the spectral characteristics of land covers and their responses to LST are different based on their climatic type. Resultantly, spectral indices (SIs) show different evaluations to be successfully used in sharpening techniques like the DisTrad, to sharpen LST over several land covers, especially in urban areas. The main objectives of this study are; 1)- to evaluate the spatial relationship between LST and a number of most commonly used urban's remote sensing SIs (21 SIs described in table 2) over the Gaza Strip in two land cover scenarios: all land covers "All" and the urban mask “urban”; 2)- to downscale aggregated low-resolution Landsat 8\LST image at 1000 m to a higher resolution of 100 m. Spectral indices and land surface temperature are calculated using the Landsat 8 image of summer 2017. Spatial regression analysis between SIs and LST within the "All" land cover class and at 1000 m resolutions show the best SIs that have the highest correlation (R2) with LST are DBSI (0.66) and ABEI (0.59). While in the "urban" class, the same indices shows also the highest correlation; BAEI (0.57) and DBSI (0.64). Moreover, statistical validation with LST observation at 100m resolution (Landsat 8\LST), DisTrad was found successful to downscale LST to 100 m resolution over UCI areas using the indices ABEI and DBSI with the highest correlation (R2) over the "All" class (0.77 and 0.73 respectively) and over the "urban" mask using DBSI (0.59) and BAEI (0.58).