基于决策树分类的土地沙化遥感识别--以内蒙古巴彦淖尔市临河区为例

Remote Sensing Identification of Land Desertification Based on Decision Tree Classification--Taking Linhe District,Bayannur City,Inner Mongolia as an Example

土地沙化遥感识别一直以来都被广泛地应用于自然环境监测领域,目前遥感影像的识别分类方法多种多样,精细度和准确度也不尽相同。本研究基于Landsat8卫星的多光谱遥感影像,利用ENVI软件对巴彦淖尔市临河地区的土地沙化进行遥感识别,研究旨在提高沙化土地信息提取的准确性和效率,为防沙治沙工程和生态环境保护提供科学依据。本研究通过提取地表不同属性光谱信息,以支持向量机(SVM)作为分类方法,分别计算NDVI指数、NDBI指数和MSAVI指数,结合ENVI决策树分类,实现对土地沙化的识别。研究结果表明,基于决策树分类法可以较准确地监测土地沙化的空间分布特征,为相关决策提供可靠的数据支持。

Remote sensing iden tification of land desertification has been widely used in the field of natural environment monitoring.At present,there are a variety of recognition and classification methods for remote sensing images,with varying degrees of precision and accuracy.In this paper,based on multispectral remote sensing images from the Landsat8 satellite,the ENVI software was used to remotely sense the desertification of land in Linhe area of Bayannur city.The aim of the research was to improve the accuracy and efficiency of desertified land information extraction,providing a scientific basis for the sand prevention and control project and the ecological environmental protection.In this study,the identification of land desertification was realized by extracting different property spectral information from the earth's surface,using the Support Vector Machine(SVM)as the classification method,calculating NDVI,NDBI and MSAVI indices,respectively,and then combining them with the ENVI decision tree classification.The results showed that the decision tree classification method could accurately monitor the spatial distribution characteristics of land desertification,provide reliable data support for related decision-making.