对流参数主分量旋转法在西昌地区雷暴预报中的应用

为研究对流参数主分量旋转方案对西昌地区雷暴预报的影响,利用2006~2008年雨季(6~9月)T213模式输出产品,采用F-分值法和主分量旋转法,分别初选和精选因子,并结合二元Logistic回归分析,建立了西昌北郊某站的雷暴释用预报的模型,并对该站2010年雨季雷暴进行预报试验。试验结果表明:精选因子的KMO检验值在0.65~0.85,且Bartlett球形度检验显著,说明该站的雷暴对流参数适用于因子分析;极大方差旋转能够较好地分离因子载荷,对雷暴信息的提取能力较强;试报的TS评分在0.3左右,24小时预报的Hedike评分可达0.337,模型的预报准确率较高,持续性较好。综上所述,采用主分量旋转法对雷暴的MOS预报有一定的改善效果。

Application of Rotating Principal Component Method of Convective Parameter in Thunderstorm Forecast at Xichang,Sichuan

The purpose of this paper is to find out the effect of the Rotating Principal Component Method of Convective Parameter in thunderstorm？ forecast at Xichang,Sichuan.Based on the output product of T213 Model in the rainy season from 2006 to 2008,F-score Method and Rotated Principal Component Method were used to filtrate factors preliminarily and carefully respectively,in addition,combined with the binary Logistic regression analyzing,a forecasting model was constructed for the forecasting of interpreting and applying for the thunderstorm for certain site in north suburb of Xichang,and prediction tests were done for the thunderstorms during the rainy season in 2010.The results are as follows：Firstly,the KMO-testing value of selecting factor is between 0.65 and 0.85,and the Bartlett Sphericity test is significantly,which showed that convective parameters of thunderstorms at this site are suitable for factor analyzing.Secondly,factor loading could be separated well by the maximum variance rotation and the capability of extracting information of thunderstorms improved markedly.Thirdly,TS score of forecasting is around 0.3,especially,Hedike score of 24 hours forecasting can be 0.337,which indicated that the forecasting accuracy is high and sustainability is better.In conclusion,MOS-forecasting of thunderstorms can be improved by using Rotated Principal Component method.