Significant roles of snow and vegetation cover in modulating altitudinal gradients of land surface temperature over Asia high mountains

The land surface temperature gradient (LSTG) serves as a key indicator of mountain thermal patterns that critically influences hydrological and ecological processes in mountainous regions. However, our understanding of LSTG across the Asian high mountains (AHM) is limited due to sparse observations and unexplored influences of surface characteristic heterogeneity within the grid used for LSTG calculation. Using a novel gridded dataset of monthly LSTG with improved local reliability, this study firstly uncovers significant spatiotemporal variations in AHM LSTGs. Then, employing an explainable machine learning approach coupled with surface energy balance analyses, we investigate the effects of ten environmental factors, including sub-grid gradients of snow, cloud, and vegetation cover. Our results reveal that spatial variations of annual LSTGs are predominantly driven by downward longwave radiation and sub-grid snow cover gradient index (SGI), with SGI dominating during cold months (April‒November) due to the substantial cooling effects of snow at higher elevations. Seasonally, three distinct LSTG patterns are detected: Spring-unimodal (21.0% of the study area), characterized by a single peak in spring; Summer-unimodal (25.2%), with a single peak in summer; and Spring-Autumn-bimodal (34.6%), showing two peaks in spring and autumn. These patterns are closely linked to local snow cover dynamics, with SGI playing a critical role in shaping the Spring-unimodal and Spring-Autumn-bimodal patterns through snow-albedo feedback, while surface net radiation primarily drives the Summer-unimodal pattern. Interannually, most of the significant LSTG trends are decreasing, mainly attributed to accelerated snow cover depletion at higher elevations, while increased vegetation cover gradient is the main contributor to increasing LSTGs. These findings highlight the importance of considering fine-scale surface heterogeneity in understanding mountain climate dynamics. The results also inform future research on integrating LSTG into ecological, agricultural, and hydrological models to better predict climate change impacts on high-mountain ecosystems.