基于客观分型的冰雹概率预报方法研究

Study on Hail Probability Forecast Method Based on Objective Classification

利用NCEP GFS资料,对2004-2015年湖北省25次冰雹天气过程54个大冰雹样本进行天气形势分类和环境场物理量统计。基于物理量百分位统计特征,利用连续概率方法,选用升、降半岭函数建立槽前型、槽后型冰雹发生概率预报模型,利用二分法计算东风气流型冰雹发生概率。在业务中,利用ECMWF数值预报模式,在实现各类天气形势客观识别的基础上进行概率预报,并在2017-2020年开展实时业务运行。四次典型冰雹过程的检验结果表明,三类天气形势下冰雹的预报效果均较好,取60%概率作为预报冰雹有无的临界阈值,命中率达91.2%,漏报率8.8%,空报率51.5%,TS评分为46.3%,预报时间提前量为12~24 h。

Based on NCEP(National Centers for Environmental Prediction)GFS(Global Forecast System)data,the weather situation was classified and the environmental parameters were analyzed by using 54 large hail samples from 25 hail weather processes in Hubei Province during 2004-2015.Based on the percentile statistical characteristics of environmental parameters,by selecting the ascending and descending half ridge functions,the continuous probability method is used to establish the probability prediction models of hail occurrence in weather situation of front trough and behind the trough.The hail occurrence probability in weather situation of easterly flow type is calculated by dichotomy.In the practical forecast,ECMWF(European Centre for Medium-Range Weather Forecasts)numerical forecast model is used to carry out the hail probability forecast based on objective classification of three types of weather situation,and real-time forecasting is carried out during 2017-2020.The evaluation results of four typical hail processes in three types of weather situation show that the hail forecast effect is good.Taking 60%probability as the critical threshold for hail prediction,the POD(probability of detection)is 91.2%,the MR(missing rate)is 8.8%,the FAR(false alarm rate)is 51.5%,and the TS(threat score)is 46.3%,The forecast time advance can be from 12 to 24 hours.