模糊神经网络方法在定点定量降水预报中的应用研究

以单站24小时降水量作为预报对象,采用模糊神经网络方法进行了新的数值预报产品释用预报方法研究.首先通过对T 213、ECMW F预报因子场以及高空气象探测资料进行处理,有效浓缩多种物理量因子场的实况及预报信息,并进一步建立了南宁、桂林、河池、百色4站的降水模糊神经网络释用预报模型.运用与实际业务相同的预报方法对2006年6—8月进行逐日的降水量预报试验,并与相同时次的T 213降水预报产品进行对比分析.结果表明,4个单站的定点、定量模糊神经网络降水预报模型,在预报性能上明显优于同期T 213数值预报模式的降水预报结果.

Research on Fuzzy Neural Networks and Application in Forecasting Precipitation of Quantifying and Locating

Taking the precipitation at single station as the forecast object,the new interpretation forecasting method of NWP with fuzzy neural networks(FNN)is studied.The sounding data and the forecast factors from T213 and ECMWF is preprocessed to concentrate forecasting information of various physical factors field and the observations.The precipitation forecast model based on the FNN is built on the four stations of Nanning,Guilin,Hechi and Baise.The forecast model is tested to forecast the precipitation day by day during the period from June to August in 2006 and its product is compared with the output from T213.Results show that the forecast function of the precipitation of quantifying and locating forecast model based on the FNN is obvious better than the T213 at the same time.