Geospatial Modeling of the Drivers of Land Surface Temperature in Salyan District: Application of GWR

Abstract

Land Surface Temperature (LST) is an important component of surface energy balance, microclimate
modulation, hydrological cycles, vegetation productivity, and human wellbeing. Thus, a spatial knowledge of
its variation and drivers is essential in mountainous country like Nepal, where topographic complexity and heterogeneous land use create strong thermal contrasts. This paper explored the geospatial determinants of LST in Salyan District, Nepal, using Geographically Weighted Regression (GWR). The dependent variable of the study, LST, was exported from the MODIS through Google Earth Engine platform, while elevation, slope, solar radiation, and proximity to streams were derived from SRTM DEM. Building point density was extracted from OpenStreetMap. The results showed that the GWR substantially enhanced the explanation of spatial heterogeneity, with R² = 0.939 and Adjusted R² = 0.915, outperforming global regression models. Moran's I Spatial residual analysis (Index=0.092, Z=2.98, p=0.0028) also indicated a statistically significant clustering but within an acceptable level for spatial factors, and radiation dynamics on LST, and it shows the applicability of GWR for climate-sensitive planning and resource management in Nepal's mid-hill districts.