Abstract:Based on Doppler weather radar data, ERA5 reanalysis data and ground automatic station observation data, WRF (The Weather Research and Forecasting) model, radar radial wind quality control and GSI cycling assimilation are used to conduct numerical simulation research on a thunderstorm gale process in Nantong, Jiangsu Province on April 30, 2021, and compare and analyze the spatial-temporal evolution and structural characteristics of radar reflectivity, wind field and thermal dynamics simulated by different test schemes. The results show that the cycling assimilation scheme (Exp3) with assimilation intervals of 30 minutes has significantly improved the simulation results compared to the non-assimilated and non-adjacent cycling assimilation scheme, indicating that cycling assimilation and increasing the frequency effectively improve the initial field. The UNRAVEL dealiasing algorithm effectively removes velocity ambiguities and improves the initial field in the cycling assimilation scheme (Exp4) after processing radar data with dealiasing, leading to significant adjustments in the simulated radar reflectivity and surface wind fields for this thunderstorm wind event.The corresponding features and evolving trends are consistent with observations, demonstrating that the cycling assimilation scheme with UNRAVEL dealiasing is more effective in improving the initial field. From the results of the thermodynamic field, the dynamic and thermal structure of Exp4 scheme is improved more obviously, with upper-level divergence and lower-level convergence, and a "cold-warm-cold" thermodynamic structure. This structure is accompanied by strong upward motion, a north-high/south-low pressure distribution, and strong vertical wind shear, which helps to transfer horizontal momentum from the mid-high level to the near-surface layer through the sinking airflow, thereby triggering this thunderstorm wind event.