基于TM影像的林木参数提取和树高估测  

Trees parameter extraction and tree height estimation based on TM images

作　　者：张巍巍[1] 冯仲科[1] 汪笑安[2] 张雷[3] 李丹丹[1] 张凝[1] 

机构地区：[1]北京林业大学测绘与3S技术中心,北京100083 [2]住房和城乡建设部规划中心,北京100835 [3]西南林业大学,云南昆明650224

出　　处：《中南林业科技大学学报》2013年第9期27-31,共5页Journal of Central South University of Forestry & Technology

基　　金：国家自然科学基金项目(30872038)

摘　　要：基于TM影像和样地数据,利用ERDAS处理软件平台,从影像中提取出可能与树高相关的遥感因子,并结合实测数据,通过多元分析中的因子分析及偏相关分析,获得对树高有显著性影响的自变量因子,运用逐步回归法建立树高估测模型。本研究以旺业甸林场为研究区域,其中以针叶林为主要研究对象,利用3倍标准差法进行数据筛选和偏相关分析,可得到该模型的相关系数较高(R=0.808),再利用剩余的18块未参加建模实地调查数据进行检验。结果表明,估测树高的总精度可达到88.55%。具有较好的估测效果。Based on TM images and sample data, by using the ERDAS processing software platform, the remote sensing factors related to tree height were extracted from the images; and by using the measured data, through the factor analysis and partial correlation analysis, the independent variable factors which have significant influences on tree height were extracted, then by using the multivariate stepwise regression method, the tree height estimation model was establish. Through selecting Wangyedian Forest Farm as the survey region, coniferous forest in the area as the studied objects, by using 3 times standard deviation method, the data screening and partial correlation analysis of the forests were conducted with a higher correlation coefficient （R=0.808）, and then the remaining 18 plots that did not to participate in the survey data modeling test were examined. The results show that the estimated total tree height accuracy can reach 88.55%.

关 键 词：林业遥感 植被指数 多元逐步回归 树高 误差分析 

分 类 号：S771.8[农业科学—森林工程]