文献类型：期刊文献

中文题名：西藏灌木林遥感分类方法对比研究

英文题名：A Comparative Study on Remote Sensing Classification Method of Shrub in Tibet

作者：张超[1,2] 黄清麟[2] 朱雪林[3] 张晓红[2] 普布顿珠[3] 旦增[3]

第一作者：张超

机构：[1]西南林业大学资源学院;[2]中国林业科学研究院资源信息研究所,国家林业局林业遥感与信息技术重点实验室;[3]西藏自治区林业调查规划研究院

年份：2010

期号：5

起止页码：572-578

中文期刊名：山地学报

外文期刊名：Journal of Mountain Science

收录：CSTPCD;;北大核心:【北大核心2008】；CSCD:【CSCD2011_2012】；

基金：中国林科院中央级公益性科研院所基本科研业务费专项资金重点项目"西藏灌木林资源评价与遥感监测技术研究(CAFYBB2007001)"资助~~

语种：中文

中文关键词：西藏;灌木林;遥感分类;空间分布特征;光谱特征

外文关键词：Tibet; shrub; Remote Sensing Classification; spatial distribution characteristics; spectral characteristics

分类号：Q948.15;S771.8

摘要：以西藏主要类型灌木林为对象,开展外业遥感调查,分别应用非监督分类、监督分类、基于空间分布特征的辅助分类和基于光谱特征再分类4种方法,对灌木林类型进行了遥感分类识别,并对各种分类方法所得到的分类结果进行对比评价。研究表明,非监督分类和监督分类的总体精度分别为42.20%和46.79%,除了对沙棘灌木林分类精度较高外,对其他灌木林类型分类精度均较低;基于空间分布特征的辅助分类精度为86.24%,其分类结果未能识别到具体灌木林类型;基于光谱特征的再分类精度为70.64%,对沙棘、红柳能较准确的识别,而对个别灌木林类型(如杜鹃和小檗)的识别精度不高。
As an important part of forest in Tibet,shrub plays an irreplaceable role in regional ecological security maintenance and regional economic and social development.Based on the remote sensing classification of shrub,we can identify the situation and dynamic changes of shrub accurately.Because of the large size of shrub patch and pure stand mostly,it is helpful to the remote sensing classification of shrub in Tibet.Based on the investigation of the main types of shrub in research area,the research was carried out as following by taking the main types of shrub in Tibet as the research object：（ⅰ）unsupervised classification;（ⅱ）supervised classification;（ⅲ）assistant classification based on spatial distribution characteristics;（ⅳ）re-classification based on spectral characteristics.The results showed that the unsupervised classification accuracy was 42.20%,and the supervised classification accuracy was 46.79%,the accuracy of assistant classification based on spatial distribution characteristics was 86.24%,which could meet the requirements of forestry,and the accuracy of re-classification based on spectral characteristics was 70.64%.Hippophae spp.and Tamarix spp.could be identified accurately by using of technical processes and image data,other types of shrub could be classified to types or type groups which depending on the precision request.