基于CR-SAM的博斯腾湖西岸湖滨带典型植被分类  

The typical vegetation classification at Bosten lakeshore using CR-SAM

作　　者：张子慧 李新国[1,3] 李勇 ZHANG Zihui;LI Xinguo;LI Yong(College of Geographic Sciences and Tourism,Xinjiang Normal University,Urumqi 830054,China;Xinjiang Institute of Ecology and Geography Chinese Academy of Sciences,Urumqi 830011,China;Xinjiang Key Laboratory of Lake Environment and Resources in Arid Zone,Xinjiang Normal University,Urumqi 830054,China)

机构地区：[1]新疆师范大学地理科学与旅游学院,新疆乌鲁木齐830054 [2]中国科学院新疆生态与地理研究所,新疆乌鲁木齐830011 [3]新疆干旱区湖泊环境与资源重点实验室,新疆乌鲁木齐830054

出　　处：《中山大学学报（自然科学版）（中英文）》2022年第6期36-43,共8页Acta Scientiarum Naturalium Universitatis Sunyatseni

基　　金：国家自然科学基金(41661047,U2003301)。

摘　　要：以博斯腾湖西岸湖滨带为研究区,采取高光谱数据和Sentinel-2A遥感数据,采用包络线去除(CR)法提取并分析箭杆杨、沙枣树和芦苇3种典型植被类别间的可区分特征,利用光谱角匹配(SAM)法对研究区3种典型植被的分布状况进行分类识别。结果表明:①SAM分类方法对3种典型植被的识别百分比为58.47%,基于包络线去除的光谱角匹配(CR-SAM)分类方法对3种典型植被的识别百分比为79.12%,CR-SAM能够突出植被光谱细节特征,减少环境背景对植被光谱的影响。②CR-SAM分类结果的总体分类精度为70.50%,较SAM分类方法提高了17.50%,Kappa系数由SAM分类方法的0.32增加到0.66,CR-SAM分类方法更能满足影像分类过程中的精度需求。③3种典型植被中,箭杆杨呈片状分布于道路两侧,面积3.98 km^(2),占研究区总面积的2.01%;沙枣树主要分布于荒地与耕地之间的过渡带,面积19.76 km^(2),占研究区总面积的9.98%;芦苇主要分布于湖滨带湿地及开都河下游沿岸,面积174.26 km^(2),占研究区总面积的88.01%。Using hyperspectral data and Sentinel-2A remote sensing image data,the continuum removed(CR)method was adopted to extract and analyze the distinguishable characteristics of typical vegetation species including Populus nigra var.thevestina(Dode)Bean,Elaeagnus angustifolia,and Phragmites australis in the west shore of Bosten Lake.The typical vegetation covering in the study area was classified and identified by Spectral Angle Mapper(SAM).The results show that three vegetation species are identified 58.47%by SAM and 79.12%by CR-SAM.The detailed characteristics of vegetation spectrums are highlighted and the influence of environmental background on vegetation spectrums is reduced by CR-SAM.The overall accuracy of the CR-SAM classification result is 70.50%,which is 17.5%higher than that of the SAM classification method;the Kappa coefficient increases from 0.32 in SAM to 0.66 in VR-SAM,indicating that CR-SAM satisfies the accuracy requirements better than SAM in the process of image classification.Among the three vegetation species,Populus nigra var.thevestina(Dode)Bean distributes strictly on both sides of the roads and covers an area of 3.98 km^(2)and 2.01%of the total area of the study area;E.angustifolia distributes mainly in the transitional zones between wetlands and cultivated lands and covers an area of 19.76 km^(2)and 9.98%of the total area;Ph.australis distributes commonly in the lakeside wetlands and along the lower reaches of the Kaidu River,and covers an area of 174.26 km^(2)and 88.01%of the total area.

关 键 词：植被分类 Sentinel-2A遥感数据 高光谱数据 包络线去除 光谱角制图 湖滨带 

分 类 号：P237[天文地球—摄影测量与遥感]