Stable snowmelt supply dominates enhanced runoff stability in the Tianshan Mountains of Central Asia

Study region This study focuses on 89 basins in the Tianshan Mountains (TMs). Study focus The strength of seasonal cycle (R) from recently developed Directional Statistics method was employed to quantify the changes in runoff stability (R runoff ) in the TMs during 1980–2019. ERA5-Land daily and monthly runoff data were validated against station observations and the Tianshan Watershed Streamflow dataset (TSWS) at basins-scale, respectively. Random Forest model (RF) coupled with SHapley Additive Explanations (SHAP) was applied to identify the dominant drivers of R runoff changes. New hydrological insights for the region Our results revealed a widespread increase in runoff stability, indicated by a decreasing trend of R runoff , across 82% of the basins, with approximately 24% showing significant ( p < 0.05) trends. Enhanced runoff stability was characterized by a more uniform intra‑annual runoff distribution and reduced intensity and frequency of runoff extremes. A more even seasonal distribution of snowmelt dominated enhanced runoff stability, as evidenced by decreasing R sm across 68% of the basins. Rain-to-runoff conversion efficiency declined in 97% of basins, 39% of which exhibited significant declines. This decline was attributed to enhanced actual evapotranspiration associated with vegetation greening. These changes reduced runoff extremes, highlighting the buffering capacity of vegetation in hydrological regulation in arid regions and offering important implications for improving water resources management in the arid TMs.