基于EVI-Albedo特征空间的土地荒漠化遥感分类方法

A Remote Sensing Classification Method of Land Desertification Based on EVI-Albedo Space

针对基于NDVI-Albedo特征空间的土地荒漠化遥感分类方法受大气和土壤影响存在不足的问题,提出基于EVI-Albedo特征空间的辽北亚湿润干旱地区土地荒漠化遥感分类方法。首先,设计基于支持向量机(support vector machine,SVM)的多分类器从总体上提取荒漠化土地覆盖区,并以此作为研究区;其次,对提取的荒漠化研究区构建EVI-Albedo特征空间,计算其荒漠化差值指数(desertification difference index,DDI);随后,依据DDI值,提出土地荒漠化等级分类模型,并与NDVI-Albedo模型进行对比分析;最后,以1989年和2020年两期Landsat图像为数据源,分析辽宁省康平县土地荒漠化动态变化情况。结果表明,EVI-Albedo模型在荒漠化分类的准确度上优于NDVI-Albedo模型;同时,基于EVI-Albedo模型的荒漠化监测结果表明,康平县1989—2020年土地荒漠化逐渐走向好转。

Aiming at the shortage of the remote sensing classification method of land desertification in the NDVI-Albedo space,which is affected by the atmosphere and soil,this paper proposes a remote sensing classification method for land desertification in sub-humid and arid areas of northern Liaoning based on the EVI-Albedo space.Firstly,design a multi-classifier based on Support Vector Machine(SVM) to extract the desertification land cover area in general,and use it as the research area.Secondly,construct EVI-Albedo feature space for the extracted desertification study area,and calculate its desertification difference index(DDI).Next,based on the DDI value,a land desertification classification model is proposed and compared with the NDVI-Albedo model.Finally,Landsat data in 1989 and 2020 is used to analyze the dynamic changes of land desertification in this county by taking Kangping county,Liaoning province as the research area.The results show that the EVI-Albedo model is better than the NDVI-Albedo model in the accuracy of desertification classification;at the same time,based on the classification of the EVI-Albedo model,it can be seen that the land desertification of Kangping county is gradually improved from 1989 to 2020.