基于GEE的合肥市土地利用/覆盖类型变化研究  

Research on land use and cover change in Hefei City based on GEE

作　　者：王德金 刘春阳[2] WANG Dejin;LIU Chunyang(School of Spatial Information and Surveying and Mapping Engineering,Anhui University of Science and Technology,Huainan 232001,China;Key Laboratory of Aviation-aerospace-ground Cooperative Monitoring and Early Warning of Coal Mining-induced Disasters of Anhui Higher Education Institutes,Anhui University of Science and Technology,Huainan 232001,China)

机构地区：[1]安徽理工大学空间信息与测绘工程学院,安徽淮南232001 [2]安徽理工大学矿山采动灾害空天地协同监测与预警安徽普通高校重点实验室,安徽淮南232001

出　　处：《测绘工程》2025年第2期38-47,共10页Engineering of Surveying and Mapping

基　　金：安徽省自然科学基金面上项目(2108085MD130);矿山采动灾害空天地协同监测与预警安徽普通高校重点实验室(安徽理工大学)开放基金资助项目(KLAHEI202203)。

摘　　要：依赖本地单机状态下的传统处理方式通常难以应对大尺度和长时序的遥感影像处理需求,文中研究以合肥市为例,提出一种基于谷歌地球引擎(GEE)平台,使用已有的土地利用/覆盖分类数据集快速选取可用于逐年影像分类训练样本的方法;然后对合肥市2013—2022年Landsat影像数据进行无云筛选和影像拼接等操作,得到年度影像分类底图;再通过融合光谱波段、光谱指数、地形特征、夜间灯光指数4个维度构建14个分类特征,采用随机森林模型的方法,对合肥市土地利用/覆盖类型变化进行研究分析,结果表明:①使用已有的土地利用/覆盖数据集获取训练样本的方法具有可行性,样本点选取精度高于95.2%。②融合多维度遥感特征的随机森林模型影像分类方法能准确地提取土地利用/覆盖分类信息,4个时间段的平均总体精度OA为92.75%,Kappa系数为0.905。③研究期间内,耕地是主要的转出源,其主要转化为建设用地,建设用地面积1931.01 km^(2)增加到3360.88 km^(2),耕地面积7386.86 km^(2)减少到6029.48 km^(2),水体面积1210.88 km^(2)减少到1105.98 km^(2),林地和未利用地面积总体未发生明显变化。文中研究为合肥市逐年土地利用/覆盖分类提供方法,同时为土地资源管理、生态可持续发展等领域提供数据支撑。Efficient and accurate access to land use/cover information is essential for urban planning and ecological environment protection.Traditional processing methods relying on local clients are often difficult to cope with large-scale and long time series remote sensing image processing requirements.Taking Hefei City as an example,this paper proposed a method based on Google Earth Engine(GEE)platform and using existing land use/cover data sets to quickly select training samples for image classification year by year.Then,the 2013—2022 Landsat image data of Hefei City were screened without cloud and image splicing to obtain the annual image classification base map.Then,14 classification features were constructed by fusing spectral band,spectral index,topographic features and night light index,and the random forest method was adopted to study and analyze the land use/cover change in Hefei City.The results showed that:(1)It is feasible to use the existing land use/cover data set to obtain training samples,and the sample point selection accuracy is higher than 95.2%.(2)The random forest image classification method integrating multi-dimensional remote sensing features could accurately extract land use/cover information,with the average Overall Accuracy of 92.75%and Kappa coefficient of 0.905 in the four time periods.(3)During the study period,cultivated land was the main source of diversion and was mainly converted into construction land,and the area of construction land increased significantly(1931.01 km^(2) to 3360.88 km^(2)),while the area of cultivated land decreased(7386.86 km^(2) to 6029.48 km^(2)).The area of water body decreased slightly(1210.88 km^(2) to 1105.98 km^(2)),and the area of forest land and unused land did not change significantly.This study provides a yearly land use/cover classification method applicable to Hefei area,and provides data support for land resource management and ecological sustainable development.

关 键 词：长时序 土地利用/覆盖变化 时空特征分析 

分 类 号：P237[天文地球—摄影测量与遥感]