Unravelling environmental influences on land surface temperature in the Kaduna River Basin: a novel integration of machine learning and GeoDetector analysis

Land surface temperature is a key indicator of environmental conditions, and it is affected by both natural and anthropogenic activities. This study assessed the environmental determinants of land surface temperature (LST) within the Kaduna River Basin (KRB) using an integrated framework of machine learning and GeoDetector analysis. Using satellite-derived datasets, including MODIS (LST and NDVI), Sentinel-2 (LULC), and SRTM (elevation), this study quantifies regional and seasonal LST drivers. The results demonstrate significant seasonal shifts: the mean NDVI increased from 0.34 in the dry season to 0.51 in the wet season, whereas the mean LST decreased from 33.57°C to 31.99°C. Stacked ensemble machine learning (R2 = 0.72 dry; 0.66 wet) and GeoDetector analysis identified vegetation (NDVI) and elevation as primary controls on the LST. Notably, the interaction between the NDVI and elevation was found to be more influential than individual factors, highlighting the synergistic role of topography and green cover in regulating basin-scale thermal dynamics. This study provides critical policy insights by stressing the importance of sustaining plant cover, particularly in specific topographical zones, in managing basin-scale thermal dynamics in the face of growing regional warming.