摘要
选取河南省5个代表站,分别代表河南省5个片,将气候预测中常用的74项环流特征量资料进行归一化处理,分别将其与5个代表站的冬季温度进行相关普查,在筛选预测因子的基础上,利用SVM两类分类和回归方法,建立河南各代表站冬季温度预测推理模型,用2000/2001~2004/2005年4年进行试报,结果显示SVM方法是处理非线性分类和回归等问题的有效方法,做分类和回归预测时,各代表站对应的SVM推理模型均具有良好的预报能力,且对温度预测SVM回归优于SVM分类。
To represent five regions of Henan Province,five stations are selected.The normalization is conducted on the 74 characteristic quantities of circulation frequently used in climate prediction,with which the winter temperature correlative analysis of five representative stations is made.The winter temperature forecast model is built based on the selected predicting factors by using the Support Vector Machine(SVM) regression method.The performance of the model is evaluated by using the data of 2000,2001,2004,a...
出处
《气象科技》
2005年第S1期100-104,共5页
Meteorological Science and Technology
关键词
支持向量机
归一化
分类
回归
温度预测
support vector machine
normalization
classification
regression method
temperature prediction
