摘要
以三垟湿地为研究区,基于2013—2021年间Landsat、Sentinel-2遥感影像,分析影像光谱特征,构建基于光谱、纹理、指数的特征数据集,通过支持向量机、随机森林、极限学习三种不同分类器的比较明确适用于研究区的解译方法,得到研究区内三垟湿地土地分类图,并计算了土地利用动态度,分析了三垟湿地自生态修复开始的土地利用变化,根据解译结果计算研究区土地利用变化率与景观变化。结果表明,遥感影像能够较好的应用于城市湿地动态监测,三垟湿地修复建设已逐步完善,因长期开垦导致的土地流失也逐步恢复。
Taking Sanyang wetland as the study area,based on Landsat and Sentinel-2 remote sensing images from 2013 to 2021,the image spectral features are analyzed,and the feature data set based on spectrum,texture and index is constructed.Through the comparison of three different classifiers of support vector machine,random forest and limit learning,the interpretation methods applicable to the study area are clarified,and the land classification map of Sanyang wetland in the study area is obtained,the dynamic degree of land use is calculated,the land use change of Sanyang wetland since ecological restoration is analyzed,and the land use change rate and landscape change in the study area are calculated according to the interpretation results.The results show that remote sensing images can be better applied to the dynamic monitoring of urban wetlands,the restoration and construction of Sanyang wetland has been gradually improved,and the land loss caused by long-term reclamation has been gradually restored.
作者
沈茗戈
SHEN Mingge(Zhejiang College of Security Technology,Wenzhou 325016,China)
出处
《现代信息科技》
2022年第2期139-142,共4页
Modern Information Technology
基金
浙江安防职业技术学院重点科研项目(AF2021Z03)。
关键词
三垟湿地
遥感解译
随机森林
支持向量机
Sanyang wetland
remote sensing interpretation
random forest
support vector machine