期刊文献+

CAR和SVM方法在郑州冬半年大雾气候趋势预测中的试用 被引量:21

Application of CAR and SVM to Prediction of Climatic Trend of Fog in Winter Half Year
下载PDF
导出
摘要 以郑州冬半年大雾日数为对象,在分析其气候特征的基础上,尝试大雾日数的气候趋势预测。首先选择气候预测中常用的环流特征量作为因子群,通过相关筛选,选取与预测对象相关系数较大的环流特征量作为预测因子,然后分别采用多变量自回归(CAR)和支持向量基(SVM)回归两种方法,建立郑州冬半年大雾日数预测模型。CAR方法回报正确率为88%,SVM方法回报正确率为82.4%;经2002/2003-2005/2006年4 a的独立样本试报,两种方法平均预测准确率(Ts)均为75%。 The number of heavy fog days in Zhengzhou in winter half year is analyzed, and on the basis of the climatic characteristics, this article attempts to predict the climatic trend of the number of heavy fog days. First, the circulation characteristic variables, which are common in climate predicting, are chosen as part of the prognostic factor group. Then, the predict model for the number of heavy fog days in Zhengzhou in winter half year is set up, using CAR and SVM methods. The results show that the accuracy of CAR is 88% , and SVM 82.4% ; the independent samples from 2002/2003 to 2005/2006 are used for predicting, and the forecast effect is good, the average forecast accuracy of these two methods are 75 % respectively.
出处 《气象与环境科学》 2008年第1期16-19,共4页 Meteorological and Environmental Sciences
关键词 气候趋势预测 多变量自回归 支持向量基回归 最小二乘法 fog prediction of climatic trend CAR SVM regression least square method
  • 相关文献

参考文献7

二级参考文献19

共引文献251

同被引文献246

引证文献21

二级引证文献154

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部