Spatiotemporal dynamics and attribution of meteorological-hydrological drought propagation in the central section of the northern Tianshan Mountains

An in-depth understanding of the complex link between meteorological drought and streamflow deficits, and the underlying influencing factors, is critical for managing water resources in high-altitude, arid regions influenced by the cryosphere. Quantitatively attributing drought propagation dynamics in such areas, a key aspect of characterizing drought hazards, remains challenging. This study employs a Copula-Bayesian framework, using the Standardized Precipitation Evapotranspiration Index (SPEI) and Standardized Runoff Index (SRI), to quantify the spatiotemporal characteristics (propagation time, probability, threshold) of drought propagation. A Random Forest model is utilized for driver attribution. Results show: (1) Propagation time exhibits a distinct west–east spatial gradient reflecting catchment buffering capacity, ranging from 6–9 months in western and central basins to 3–5 months in eastern basins; (2) The propagation process shows significant seasonal dynamics, characterized by a two-stage evolution culminating in a late-summer ‘vulnerability window’ with amplified sensitivity (probability peak ≈ 0.70), demonstrating a long-term intensifying trend; (3) Potential evapotranspiration (PET, 20.82%), elevation (17.98%), and soil moisture (17.83%) are the primary drivers influencing propagation threshold spatial patterns. These findings provide a quantitative basis for drought early warning and are essential for assessing evolving hazards and mitigating risks in this region.