Study on Warning Index of Rainstorm in Sichuan Based on Ensemble Prediction Products

Based on construited sample data in Sichuan, with the ensemble prediction products in flood season over 2005-2007 from Chengdu Regional Meteorological Center as well as numerical rainfall products from Germany and Japan, a warning-rainstorm model is developed by using of Support Vector Machines method (SVM).The model was applied to predict daily rainstorm in flood season in 2008.The application result shows that the warning index as 0.2, the model is good at indicating fixed-point rainstorm prediction relatively.