Lightning prediction method based on class-weighted dual v-support vector machine

A lightning prediction model is established to predict the lightning in 24h in Chongqing, the model is based on the character of lightning weather, the advantages of the support vector machine (SVM) method in solving learning problems of nonlinear and high dimensional samples, and the class-weighted dual v-SVM (WDv-SVM) -an improved algorithm of SVM. According to high-altitude and surface data during 1998 to 2008 provided by the Micaps system in China Meteorological Administration and the lightning observation data collected from 35 ground stations all over the city, the predictors related to lightning occurred are calculated. Compared with c-SVM and v-SVM, WDv-SVM is provided with superior classification accuracies and prediction accuracies. Consequently, the lightning prediction system in operational application is developed on the basis of the model referred.