基于主成分分析法的老秃顶子自然保护区森林蓄积量遥感估测  

Estimation of Laotudingzi nature reserve forest volume based on principal component analysis

作　　者：刘明艳 王秀兰[1,2] 冯仲科 于东海[1,2] 

机构地区：[1]北京林业大学林学院精准林业北京市重点实验室,北京100083 [2]北京林业大学森林培育与保护省部共建教育部重点实验室,北京100083

出　　处：《中南林业科技大学学报》2017年第10期80-83,117,共5页Journal of Central South University of Forestry & Technology

基　　金：北京市自然基金项目"环首都圈森林植被空间环境效应模型与造林决策支持系统研究"(6161001)

摘　　要：以老秃顶子自然保护区为研究区,采用研究区landsat8 OLI遥感影像、DEM数据、实地调查数据作为数据源,提取11个光谱因子、8个纹理因子、3个地形因子,采用主成分分析法对所有因子进行降维处理,以累积方差贡献率大于80%作为指标,选取4个主成分,并以主成分得分为自变量、以每公顷蓄积量为因变量,建立线性回归估测模型,并检验精度。结果表明:回归方程调整后的R^2=0.810,拟合度好。对模型进行精度检验,结果为:蓄积量估测的平均相对误差为12.12%,总相对误差为6.02%,平均预估误差为7.82%,模型预估精度达到92.18%,能够满足林业调查中对于蓄积量遥感估测的要求。By selecting Laotudingzi as the research area, taking the remote sensing images of landsat80LI, DEM data and survey data as the data source, the studied area＇ s 11 spectrum factors, 8 texture factor and 3 terrain factors in the corresponding sample area were acquired. Putting the cumulated variance contribution ratio to more than 80% as index, the principal component affecting forest reserves were extracted by principal component analysis, and then a principal component regression model was obtained using the principal component. The results showed that the regression equation of the adjusted R2 is 0.810, which fits well. To accuracy test, the model precision of the average relative error is 12.12%, the total relative error is 6.02%, the mean prediction error is 7.82% and the model prediction accuracy is 92.18%. It can meet the requirements of forestry investigation for volume of remote sensing to estimate.

关 键 词：蓄积量 主成分分析法 模型检验 

分 类 号：S771.8[农业科学—森林工程]