基于主成分分析法的蜀柏毒蛾灾害发生影响因子筛选研究

A Study of Principal Component Analysis in Comprehensive Indicator Screening for Parocneria Orienta Hazard

以中江县蜀柏毒蛾发生区为研究对象,探讨主成分分析方法在筛选影响因子中的作用。选取影响蜀柏毒蛾危害的12个生态因子,通过主成分筛选和优化,划分蜀柏毒蛾危害区域,并采用GIS进行直观展示。研究得到3个主成分,分别代表立地因子、环境关系因子和林分因子,贡献量分别为35.274%、20.544%、14.897%,累计贡献率为70.715%,基本保留了12个生态因子的绝大部分信息。按主成分综合得分值大小,采用自然断点分类法将中江县蜀柏毒蛾危害划分为常发区、偶发区和安全区,与历年发生趋势一致。并用Arc GIS软件以继光乡为例进行了展示。结果表明,主成分分析可以优化蜀柏毒蛾危害综合评价指标,GIS可直观显示危害程度。古店乡、继光乡、回龙乡等以蜀柏毒蛾常发区为主的乡镇,应重点监测。

The aim of this paper is to investigate the application of principal component analysis in impact factors of Parocneria orienta hazard in Zhongjiang County. A principal component analysis was performed by the data of 12 factors associating with the prediction of P. Orienta hazard. The degree of P. Orienta hazard in different regions was obtained and intuitively presented by GIS. The eigenvalues of 3 principal components were 35. 274% 、20. 544% 、14. 897%,respectively. The accumulative contribution rate of the 3 major factors to total variation accounted for 70. 715%,maintaining most of information of 12 characters.According to the comprehensive value of principal component,GIS Natural Breakpoint Method was used to divide the degree of P. Orienta hazard into frequently occurring region,occasionally occurring region and safe area. Results showed that principal component analysis could optimize the comprehensive indicators for the evaluation of P. Orienta hazard,and comprehensive score of Principal Component could quantify and intuitively show the degree of P. Orienta hazard in different regions.