预报因子选择的条件数方法及其在台风强度预报中的应用  

Selection of Factors Based on Condition Number and its Application for Typhoon Intensity Prediction

作　　者：农吉夫[1] 

机构地区：[1]广西民族大学理学院,广西南宁530006

出　　处：《数学的实践与认识》2014年第23期146-152,共7页Mathematics in Practice and Theory

基　　金：国家自然科学基金(11061005);广西教育厅科研项目(201204LX083)

摘　　要：选择合适的变量是建立多元线性回归方程的重要问题.以减弱诸多变量之间的复共线性为目标,采用条件数分析方法选择多元回归模型的自变量.最后以西北太平洋海域2001-2011年5-7月的台风强度为研究对象,利用条件数方法建立预报方程进行预报试验,并进一步将该预报方法与逐步回归方法进行对比分析.The important thing is to choose suitable independent variable in building multi- variate linear regression equation. In this paper, a method in choosing independent variable of multi-linear regression model is given in order to weaken the multi-collinearity among the independent variables by using method of condition number. Finally, the typhoon intensity prediction experiment based on condition number method is conducted in the northwest Pa- cific Ocean from May to July 2001-2011, and further the forecast method and the stepwise regression method are compared.

关 键 词：复共线性 条件数 预报因子 台风强度 

分 类 号：P457.8[天文地球—大气科学及气象学]