基于Landsat的环洞庭湖区湿地信息提取

Landsat-based Information Extraction of the Wetland Around Dongting Lake Area

大范围湿地信息的精细提取对湿地资源的管理和保护具有关键作用。基于GEE平台和ArcGIS软件,对Google云端提供的Landsat数据进行预处理后,分别采用最小距离、决策树、随机森林、迭代决策树4种不同类型的分类器提取环洞庭湖区域的湿地信息。主要研究结果如下:随机森林分类方法得出的分类结果的精度最高,总体精度(OA)为95.82%,Kappa系数为0.942;最终提取得到2015年环洞庭湖区湿地的各个地类中,水体(河流、湖泊、库塘)、苔草滩地、泥沙滩地、芦苇滩地的面积占比分别为6.385%、4.578%、0.418%、6.060%。研究结果可为大面积湿地的监测和分类工作提供更高效的技术支持和更可靠的数据保障。

The fine extraction of large-scale wetland information plays a key role in the management and protection of wetland resources.In this paper,the Google Earth Engine platform and ArcGIS software were used to preprocess the Landsat data provided by Google Cloud.Four types of classifiers such as minimum distance,decision tree,random forest,and iterative decision tree were used to extract wetland information around Dongting Lake area.The main results were as follows:The accuracy of the random forest classification method is the highest,the overall accuracy(OA)was 95.82%and the score Kappa coefficient was 0.942.Among the various land types of the wetlands around Dongting Lake in 2015,the areas of water(rivers,lakes,ponds),moss beach,mud beach,and reed beach accounted for 6.385%,4.578%,0.418%,and 6.060%,respectively.The results of this studyis to provide more efficient technical support and more reliable data guarantee for the monitoring and classification of large area wetlands.