基于决策树的天山冰湖提取方法研究OA北大核心CSTPCD

The Tianshan region,located in the middle of the Eurasian continent,is a major distribution area of modern glaciers,and its glacial meltwater has formed a large number of and widespread glacial lakes.Glacial lakes are vital indicators of climate change and an important source of surface and groundwater supply in arid and semi-arid regions of Northwest China.The impact of topographic factors and spectral characteristics of ground objects makes it difficult to distinguish between glacial lakes,mountain shadow,and snow cover when ex-tracting glacial lakes from remote sensing images using a single water index.In this study,based on the Google Earth Engine platform and Landsat 8 remote sensing images as the data source,a decision tree extraction method of the glacial lake was constructed according to the topographic characteristics(slope,elevation,and buffer analy-sis)and spectral characteristics of the glacial lake.This method was compared with the NDWI and MNDWI threshold methods.Experimental results demonstrated that the decision tree method can effectively reduce the im-pact of mountain shadow and snow cover and accurately extract glacial lake information.The overall accuracy of the extraction results was 89.14%,the Kappa coefficient was 0.783,and the F1 score was 87.85%.The decision tree method,which combines spatial topographic features and spectral features,is a relatively efficient extraction method for dynamic monitoring and research analysis of glacial lakes.