基于SPOT5影像的杉木胸高断面积估测探讨  

Study on Basal Area Estimation of Chinese Fir Based on SPOT5 Images

作　　者：陈柏海[1] 林辉[1] 孙华[1] 

机构地区：[1]中南林业科技大学林业遥感信息工程研究中心,长沙410004

出　　处：《中南林业调查规划》2012年第1期44-48,56,共6页Central South Forest Inventory and Planning

基　　金：湖南省高等学校科学研究项目"高分辨率遥感影像森林结构参数反演研究"(11C1313)

摘　　要：采用角规实地调查黄丰桥林场90个杉木人工纯林样地胸高断面积,利用样地SPOT5遥感信息与地理信息,建立了杉木胸高断面积多元线性回归估测模型。首先对样地采用GIS软件进行缓冲处理,缓冲后每个样地的面积为1 hm2;然后提取样地遥感光谱信息与纹理信息等21个因子和4个GIS因子,采用逐步回归分析法筛选出6个因子作为模型自变量;最后分别采用普通最小二乘法(OLS)和偏最小二乘法(PLS)建立了杉木胸高断面积多元回归模型。研究结果表明:OLS回归模型的预测精度为82.2%,均方根误差(RMSE)为5.12 m2/hm2;PLS回规模型的预测精度为83.9%,均方根误差(RMSE)为4.21 m2/hm2,PLS和OLS回归模型在杉木胸高断面积估测中均取得了较好的效果,用中高分辨率遥感影像在估测森林结构参数上是可行的。The surveying of Chinese fir basal area by fielding 90 sample plots witn angle gauges liau tied out in Huangfengqiao forest farm,the multiple linear regression estimation model of basal area was set up based on remote sensing and geographic information. First, each sample plot was buffered by GIS software to 1 hectare;Then from which 2l RS index factors such as spectral and texture information and 4 GIS index factors were extracted ,in which 6 index factors were screened out as independent model variables through stepwise re- gression analysis;Last the multiple regression model was built by using OLS and PLS respectively. The results showed that： the model predicted accuracy was 82.2% and RMSE was 5.12 m^2/hm^2 by using OLS; the model predicted accuracy was 83.9% and RMSE was 4.21 m^2/hm^2 by using PLS ;The adoption of OLS and PLS serv- ices well in basal area estimation, to estimate forest structural parameters can achieve good effects by using high resolution remote sensing images.

关 键 词：胸高断面积 多光谱 SPOT5 多元统计分析 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置] S757.2[自动化与计算机技术—控制科学与工程]