Synoptic Background and Atmospheric Conditions for Hailstorms in Weining (Yunnan‐Guizhou Plateau): Multi‐Source Data Support and Machine Learning‐Based Parameter Importance Ranking

Abstract Hailstorms over the Yunnan‐Guizhou Plateau cause significant losses in key cash crops such as tobacco, inflicting substantial economic impacts. Weining, an eastern plateau region prone to spring‐summer hailstorms, requires further research on hailstorm environmental conditions to improve forecasting and disaster mitigation. This study investigates these aspects using multi‐source data, objective classification, vorticity diagnostics, and machine learning (ML) to derive robust conclusions. Results show hailstorms predominantly occur in afternoon–evening producing hailstones mostly with diameters <20 mm, under five distinct low‐level synoptic circulation patterns, but all sharing a common feature: a trough from the southeastern Tibetan Plateau and cyclonic vorticity associated with the Southwest Vortex near the Sichuan Basin. Spring hailstorms have stronger dynamical forcing (the plateau trough coupled with intense westerly/southwesterly inducing strong convergence), while summer hailstorms have weaker forcing but enhanced moisture transport, with 42.3% linked to tropical cyclones. XGBoost analysis of 17 environmental parameters reveals consistent enhancements in instability and hail growth environments across seasons. Key parameters, listed in order of importance, are most unstable convective available potential energy (MUCAPE), hail growth zone (HGZ), mixed layer height (MLH) and PW in spring; MLH, HGZ, MUCAPE and freezing level height (FLH) in summer, indicating seasonal thermodynamic differences. Additionally, ERA5‐derived thermodynamic parameters (e.g., equilibrium level, FLH, and MUCAPE) have significant biases compared to radiosonde data, impairing ML‐based importance assessment. This study overcomes limitations of single‐data‐source and traditional methods, providing valuable scientific insights for improving hailstorm forecasting in complex terrain. Large samples and ML ensure reliable identification of seasonal differences and objective parameter importance quantification, laying a foundation for enhancing hailstorm predictability with significance for disaster mitigation.