基于MM5模式的站点暴雨预报方法研究

基于对MM5模式输出产品的解释应用，建立了站点定量的暴雨MOS预报方法。从临沂中尺度数值预报模式产品中挑选x1（模式预报降水），…，x12（沙氏指数）等12个模式产品和物理诊断因子为基本可能预报因子，又添加它们自身的非线性型xi^2xi^3,xi^1/2，1nxi，e^xi（i=1，…，12）和各因子间非线性交互作xixj（i=1，…，11，j=i＋1，…，12）型的126个可能预报因子。用非线性逐步判别做晴、雨二级判别，然后对雨型样本做逐步回归，得到暴雨的MOS预报方法。检验结果表明：该模型对暴雨有较强的预报能力，可应用于实际预报业务中。

Study of Station Heavy Rainfall Forecasting Method Based on the MM5 Model Result

MOS forecasting method on heavy rainfall is developed based on the MM5 model in Linyi Meteorological Bureau. 12 basic possible predictors Xl （model forecasting precipitation） ...x12（ Si index ）which are predicts and diagnostic factors are selected from numerical model. Meanwhile, non-linear items xi^2xi^3,xi^1/2,1nxi,e^xi（i=1,…,12） and the corresponding non-linear cross items xixj are accepted as the other 126 possible predictors（i=1,2…,11,j=i＋1,…,12）.Non-linear successive discrimination is adopted in statistical mathematical model of MOS forecast on heavy rainfall to distinguish the clear and rainy type. Then successive regression is performed on the rainy type. The results show that this model has great capabilities in forecasting heavy rainfall, and the model can be applied in real-time heavy rainfall forecast.