叶面积指数遥感估算的三种回归模型分析  

Comparison of three regression models for remote sensing estimation of leaf area index

作　　者：胡古月 李少达[1] 杨容浩[1,2] HU Guyue;LI Shaoda;YANG Ronghao(College of Earth Sciences,Chengdu University of Technology,Chengdu 610059,China;Sichuan Engineering Research Center for Emergency Mapping Disaster Reduction,Chengdu 610041,China)

机构地区：[1]成都理工大学地球科学学院,成都610059 [2]四川省应急测绘与防灾减灾工程技术研究中心,成都610041

出　　处：《测绘科学》2018年第10期46-50,66,共6页Science of Surveying and Mapping

基　　金：四川省教育厅科研计划重点项目(15ZA0060);四川省应急测绘与防灾减灾工程技术研究中心开放基金资助项目(K2014B001)

摘　　要：针对叶面积指数(LAI)估算方法的适用性和提高估算精度的问题,该文基于Landsat-7ETM+遥感影像,以都江堰市青城山地区为研究对象,利用LAI-2000对样地LAI进行观测采集,建立如归一化植被指数等4种常见植被指数的单植被指数线性、非线性回归模型和多植被指数组合的偏最小二乘回归模型,比较分析了不同植被指数和不同模型估算LAI的精度。实验结果表明,偏最小二乘回归法能有效解决因变量较多时权重系数难确定的问题,是估算LAI的一种有效方法。Aiming at the problem of practicability and estimation accuracy of estimation methods for the leaf area index（LAI）,based on the Landsat-7 ETM＋ remote sensing image,the single vegetation index linear regression and non-linear regression models and the partial least-square regression（PLSR）method of multi-vegetation index of four common vegetation indices,like normalized difference vegetation index,were established in this paper,in which the Mount Qingcheng area of Dujiangyan was taken as the research object and the sample site LAI was collected by LAI-2000.Then LAI estimate accuracy of diverse vegetation indexes and models were compared and analyzed.The experimental results showed that PLSR method could effectively solve the difficulty of determining weight coefficient when there are too many dependentvariable,which is an effective solution of LAI estimate.

关 键 词：叶面积指数 植被指数 回归模型 偏最小二乘回归 

分 类 号：P237[天文地球—摄影测量与遥感] TP79[天文地球—测绘科学与技术]