应用Sentinel-2A卫星遥感影像估测森林蓄积量  

Estimating Forest Volume Using Sentinel-2A Satellite Remote Sensing Image

作　　者：庞晓燕 刘海松 年学东 张志超 李崇贵[2] 董相军 Pang Xiaoyan;Liu Haisong;Nian Xuedong;Zhang Zhichao;Li Chonggui;Dong Xiangjun(Inner Mongolia Forestry Monitoring and Planning Institute,Hohhot 010031,P.R.China;Xi’an University of Science and Technology;Baiyin Oboo Mechanical Fire Station of Keshketeng Banner)

机构地区：[1]内蒙古自治区林业监测规划院,呼和浩特010031 [2]西安科技大学测绘科学与技术学院 [3]西安科技大学 [4]内蒙古自治区克什克腾旗白音敖包机械防火站

出　　处：《东北林业大学学报》2021年第7期72-77,90,共7页Journal of Northeast Forestry University

摘　　要：为了探讨应用Sentinel-2A遥感影像进行森林蓄积量估测的可行性,以内蒙古自治区某林业局的一类清查样地数据、二类调查小班数据、数字高程模型(DEM)以及林地数据为数据源,以遥感影像的波段灰度信息、比值波段及地形信息为自变量,采用k-近邻法(k-NN)、稳健估计及偏最小二乘估计等方法,分林型构建研究区域的森林蓄积量估测模型,从林业局、林场及小班尺度进行估测精度评价。结果表明:k-NN方法在林业局、林场和小班等3个尺度中的估测精度分别达到了97%、93.2%和83.6%,均表现出良好的估测效果;稳健估计法在3个尺度中的估测精度分别为89.3%、72.4%、69.3%;偏最小二估计法在3个尺度中的估测精度分别为85.7%、75.8%、71.7%。k-NN方法估测效果明显优于稳健估计方和偏最小二估计法,因此,Sentinel-2A遥感影像能够有效应用于森林蓄积量估测。In order to explore the feasibility of applying Sentinel-2 A remote sensing images for forest stock estimation,a forestry bureau in Inner Mongolia Autonomous Region was used as the data source for the first class inventory sample plot data,second class survey small group data,digital elevation model(DEM)and forest land data,and the wavelength grayscale information,ratio band and topographic information of remote sensing images were used as independent variables,and the k-nearest neighbor method(k-NN),robust k-NN method,robust estimation and partial least squares estimation were used to construct the forest stock estimation model in the study area by forest type,and the estimation accuracy was evaluated at the scale of forestry bureau,forestry field and small group.The results showed that the estimation accuracy of k-NN method reached 97%,93.2%and 83.6%in the three scales of forestry bureau,forestry field and small group,respectively,which showed good estimation results;the estimation accuracies of robust estimation method were 89.3%,72.4%and 69.3%in the three scales,respectively;the estimation accuracies of partial least squares estimation method were 85.7%,75.8%and 71.7%in the three scales,respectively.The estimation results of the k-NN method were significantly better than those of the robust estimator and the partial least squares estimator.Therefore,the Sentinel-2 A remote sensing images could be effectively used for forest stock estimation.

关 键 词：Sentinel-2A卫星影像 森林蓄积量 K-近邻法 稳健估计法 偏最小二乘估计法 

分 类 号：S757.2[农业科学—森林经理学]