动态聚类法在湖北省农作物遥感估产区划中的应用  

The application of dynamic clustering analysis in regionalization of crop yield estimation by remote sensing in Hubei Province

作　　者：杜桐生[1] 黄进良[1] 薛怀平[1] 李蓉蓉[1] 

机构地区：[1]中国科学院测量与地球物理研究所,武汉430077

出　　处：《华中师范大学学报（自然科学版）》2000年第2期241-244,共4页Journal of Central China Normal University：Natural Sciences

基　　金：中国科学院"九五"重大和特别支持项目资助!KZ95T-0 3 -0 3 -0 4 )

摘　　要：采用动态聚类法 ,进行了湖北省农作物遥感估产区划 .其方法为先对样品数据进行标准化变换 ,用以消除量纲上的差异 ,再用变换取整法进行初始分类 ,然后计算每类的重心坐标作为凝聚点 ,将所有样品按最近的凝聚点归类 ,如此修改 n次使分类结果比较合理为止 ,得到动态聚类结果 .在定量的基础上 ,定性地加以修改 。The regionalization of yield estimation is the precondition of crop yield estimation by remote sensing. This paper discusses dynamic clustering analysis method to regionalize crop estimation of Hubei Province, The following is the process: First, standardizing of the samples to eliminate the difference among dimensions. Next, classifying initially the samples, computing the gravity coordinate as a clustering point. Thirdly, modifying the classification of the recent clustering point. Finally repeating the above steps n times, stopping when the result of the classification is relatively reasonable. Then analyzing qualitatively the quantitative results of clustering and determining the regionalization of crop estimation of Hubei Province.

关 键 词：估产 区划 农作物 动态聚类法 遥感 

分 类 号：S127[农业科学—农业基础科学] F323.1[经济管理—产业经济]