基于动力相似方法的台风极端降水概率预报研究

A RESEARCH ON THE PREDICTION OF TYPHOON EXTREME PRECIPITATION BASED ON DYNAMIC SIMILITUDE METHODS

采用动力与统计相结合的方法,对台风引起的极端降水事件进行概率预报方法的研究,回算并评估了25个在2012-2015年间登陆中国的台风。此外,还对该预报方法的相似成员数和有效概率阈值进行了敏感性试验。试验结果显示,该台风极端降水概率预报方法具有一定的预报技巧,能准确预报多数产生极端降水的站点,但预报范围略偏大,空报率较高。相似成员数的敏感性试验表明,100%的相似成员数比率具有更优的预报效果;在相似成员数比率为100%的基础上,24 h预报的有效概率阈值在0%~2%之间的预报效果最好,48 h的有效概率阈值在0%~1.5%之间的预报效果最好。最后对"苏迪罗"和"海鸥"两个台风进行个例预报试验,结果表明对不同个例,预报效果存在较大差异。

This paper uses a combination of dynamic similitude methods to study the method of probabilistic prediction of extreme precipitation events caused by tropical cyclones,and evaluates 25 tropical cyclones which landed in China from 2012 to 2015.In addition,we conduct two sensitivity experiments on optimal similar number of members and effective probability threshold.Experiment results show that the typhoon extreme precipitation probability forecasting method has certain forecasting skills,and can accurately predict many sites occurring extreme precipitation,but the forecast range is larger than the actual situation,and the FAR is higher.The sensitivity experiment of the optimal similar number of members shows that the ratio of similar members with 100%has a better forecasting effect.Based on the optimal similarity of 100%,the effective probability threshold has a better forecasting effect between 0%and 2%of24 h and between 0%and 1.5%of 48 h.Finally,a case prediction experiment is carried out for two typhoons,i.e.,Soudelor and Kalmaegi.The results show that there are significant differences in the forecast results of different cases.