Long-term evolution and climatic associations of dry and warm snow droughts across the Tibetan plateau

Seasonal snowpack on the Tibetan Plateau (TP) is an important store of water, yet long-term changes in snow drought types and their value for monitoring snow-water deficits remain unclear. Using multi-source data for 1982–2025, we classified dry (SWEI-3 ≤ −0.5; SPI-3 ≤ −0.5) and warm (SWEI-3 ≤ −0.5; SPI-3 > −0.5) snow droughts and assessed their trends, regional patterns, drought burden, and statistical climatic associations. Here, “warm” identifies an SWE deficit without a concurrent precipitation deficit, not a temperature-defined event. Total snow drought area increased by 0.595 percentage points yr −1 ( p < 0.001), mainly reflecting warm snow drought expansion (0.687 percentage points yr −1 ). Warm snow droughts accounted for 72.8% of snow drought occurrence, including 69%, 75%, and 72% in the western, south-central, and southeastern TP, respectively; the south-central TP showed the fastest historical increase. The positive warm-drought trend persisted when the product-transition years were excluded and under two alternative fixed seasonal windows, although trend magnitudes varied. Mean drought intensity changed little, and the increasing snow drought burden index largely reflected spatial expansion. XGBoost–SHAP identified precipitation as an important statistical discriminator (30.9% for dry and 23.1% for warm snow droughts), but the warm-drought model showed limited temporal discrimination (ROC-AUC = 0.548). No lagged climate-mode correlations survived false discovery rate correction, and composite NDVI differences were not significant. These results support monitoring seasonal SWE deficits without establishing specific physical mechanisms or downstream hydrological and ecological impacts.