Abstract:Based on the EAR5 reanalysis data in June from 1979 to 2016, the moist thermodynamic advection parameter, thermal helicity, divergence flux, moisture divergence flux and the thermodynamic wave-activity density are selected as five comprehensive factors. The probability prediction model of regional persistent rainstorm in Hunan is constructed by the means of nuclear density estimation and based on the optimal factor and weight combination which is established with the best TS score as the test standard. The results show that the average TS of independent samples from 2017 to 2019 reaches 29.9%, which is a positive skill relative to the European Centre for Medium-Range Weather Forecasts (ECMWF) fine grid forecast (with an average TS score of 22.4%). During the two regional persistent rainstorm operational experiments in the 2021 and 2022 flood seasons, the rainstorm forecast with a 24 h leadtime by the Hunan regional perisistent rainstorm probability prediction model is superior to the forecasts of ECMWF and CMA-GFS large-scale model as well as CMA-SH and CMA-GD regional mesoscale model. Therefore, the Hunan regional persistent rainstorm prbability prediction model has a strong ability to forecast the regional persistent rainstorm in Hunan.