用均生函数预测模型预报降水的若干试验

SOME EXPERIMENTS AND RESEARCHES ON PRECIPITATION FORECAST WITH MEAN GENERATING FUNCTION MODEL

关键词:  均生函数  主成分模型  逐步筛选  汛期雨量

Key words:  Mean generating function  EOF model  Stepwise sieving  Rainy season precipitation

摘要:

基于均生函数的处理原则,对目前常用的均生函数逐步筛选模型、主成分模型进行预报试验,提出带逐步筛选的主成分模型,并用变化辐度较大的降水序列对三种模型进行若干预测试验。结果表明:逐步筛选的主成分模型优于主成分模型,在对不同预报步长的预测试验中也表明,逐步筛选的主成分模型在预测效果以及模型的稳定性方面都是较好的。

Abstract:

In this paper based on mean generating function principle,the experimental forecast with stepwise sieving model and Empirical Orthogonal Function (EOF) model of mean generating function have been made. We put forward a new kind of EOF model with stepwise sieving,and all of the three forecasting models have been experimented on precipitation sequence with great variation amplitude. The result shows the EOF model with stepwise sie ving is better than EOF model alone,and another result which forecasts have been experimented by different forecast step-length also shows EOF model with stepwise sieving is very good in the respect of forecast effect and model operation stability.