Abstract:Using the precipitable water vapor (PWV) data from Global Navigation Satellite System (GNSS) meteorological network and the time series of hourly precipitation, temperature, pressure and relative humidity from the adjacent meteorological stations in Hubei Province from June to July in 2019 and 2020, we build a new threshold prediction model of short-time severe precipitation on the basis of PWV, 6 h PWV tendency (PWV*) and pseudo-equivalent temperature anomaly (θse*) in this study. The critical success index (CSI) and probability of detection (POD) are used to determine and validate the thresholds for the three predictors. This new model is tested by the precipitation data in June and July, 2021. The results show that CSI and POD scores are 0.167 and 0.593, respectively, which are higher than the objective forecast scores for short-time severe precipitation by conventional operational methods. About 48% of the short-time severe precipitation events are successfully predicted with a lead time of 24 h, and 78% occur within 48 h. The analysis results show that 78.6% of the severe precipitation occur under the condition of the 15 h cumulative value of PWV* (∑PWV*) ≥75 mm and the 24 h cumulative value of θse* (∑θse*) ≥ 30 K. This finding could be positively indicative of severe rainfall in Meiyu season and torrential rains with the PWV high-value zone superimposed to the high-value zones of ∑PWV* and ∑θse*.