基于决策树的干旱区湿地信息自动提取——以疏勒河流域为例

The Automatic Extraction of Wetland Information in Arid Zone based on Decision Tree Algorithm——A Case Study in the Shule River Basin

以疏勒河流域为研究区,探讨了干旱区湿地的遥感影像自动提取方法。以Landsat 8卫星影像数据为主要数据源并辅以数字高程模型(DEM),利用改进的干旱区湿地指数(MAZWI)、归一化植被指数(NDVI)、地表反照率(Albedo)、灰度共生矩阵(GLCM)的非相似性分量等识别指数构建决策树模型,对研究区湿地进行提取,并将结果与最大似然分类结果进行对比。结果表明:该方法在一定程度上提高了湿地提取的精度,与最大似然分类结果相比总体精度和Kappa系数分别提高了6.52%和0.124。证明决策树法是干旱区水域湿地自动提取的一种有效手段。

Selecting the Shule River Basin which locates in the west of Gansu Province as the study area,the automated extraction method for wetlands in arid regions was discussed.Using the Landsat 8satellite images as the data sources,supported by the digital elevation model（DEM）,the modified arid zone wetlands index（MAZWI）,the normalized difference vegetation indices（NDVI）and the surface albedo,and the identification of dissimilarity index of the gray level co-occurrence matrix（GLCM）,were used as the indicators to establish the decision tree model and the wetlands were extracted.Comparing with the results obtaining by the maximum likelihood supervised classification,it showed that the decision tree method based on the indices can improve the overall accuracy by 6.52%and the overall kappa coefficient by 0.1243.The results of this study suggested that the decision tree method based on indices is an effective tool for wetlands classification in arid zone.