Study the spatiotemporal variation characteristics and driving forces differences of land surface temperature in the Chishui River Basin based on the Bayesian Structural equation model

Clarifying the driving processes of the thermal environment of a basin is important for maintaining the Water-land ecological security of the basin. The main focus in current basin Land surface temperature (LST) driver research is on the direct effects of the drivers, but there is still insufficient understanding of their interactions and indirect effects. This study based on Bayesian Structural Equation Model (BSEM) to study the driving force influencing LST in the Chishui River Basin from three perspectives: the spatial distribution of long-term LST at a 1 km resolution, changes, and the spatial distribution of instantaneous LST at a 100 m resolution. The results indicate that: (1) Over the 20 years in the Chishui River Basin, daytime LST (DLST) and day-night LST difference (LSTd-LSTn) exhibited a cooling trend of 0.039 K/year and 0.0312 K/year, respectively, while nighttime LST (NLST) showed a warming trend of 0.0238 K/year. (2) The spatial distribution of 1 km long time-series LST is mainly influenced by terrain and climate, while the change of LST is primarily influenced by climate and the degree of vegetation cover. The 100 m instantaneous LST is mainly influenced by terrain and human activities. The differences in dominant driving factors of LST at different spatial scales stem from the scale dependence of driving factors. (3) Terrain directly affects LST and indirectly through climate, human activities, and biophysical parameters; human activities and climate directly affect LST and indirectly through biophysical parameters. NDVI directly affects LST and plays an important intermediate regulatory role. This study has certain theoretical guidance significance in the aspects of ecological civilization construction and surface water thermal environment in the basin.