均生函数场预报模型对降水预报的试验研究

THE TEST STUDIES ON THE MEAN GENERATING FUNCTION FIELD MODEL OF PRECIPITATION FORECAST

基于均生函数的处理原则 ,运用经验正交分解方法 ,将正交展开的时间系数作为预报因子 ,并引入逐步算法进行因子筛选建立预报模型。将均生函数场预报模型用于云南汛期雨量及 5月雨量预报试验。结果表明 :用这种模型作区域性降水预报有一定的优越性 ,并可作多步预测 ,是一种滚动客观预报方法。

Based on mean generating function principle, the predict and field are analysed by Empirical Orthogonal Function(EOF) method, time coefficient is given to be accepted as predictor and predictive model is set up using stepwise sieving to choose predictors. The mean generating function field model is applied to the precipitation prediction in rainy season and rainfall in May of Yunnan province. The predictive result shows the model has a certain virtue in the forecasting of regional rain and multi step prediction capacity.