Landsat-based Information Extraction of the Wetland Around Dongting Lake Area
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.
wetlandGoogle Earth Engineland type information extractionLandsataround Dongting Lake area
NETL
NSTL
万方数据
