基于GEE和Landsat时间序列数据的海南岛土地利用分类研究  

Mapping Land Use in Hainan Island Based on Google Earth Engine and Landsat Time Series Data

作　　者：戴声佩 易小平[1,3] 罗红霞 李海亮[2,5] 李茂芬[2,5] 郑倩 胡盈盈[2,5] DAI Shengpei;YI Xiaoping;LUO Hongxia;LI Hailiang;LI Maofen;ZHENG Qian;HU Yingying(Chinese Land Surveying and Planning Institute/Key Laboratory of Land Use,Ministry of Natural Resources,Beijng 100035,China;Institute of Scientific and Technical Information,Chinese Academy of Tropical Agricultural Sciences/Key Laboratory of Applied Research on Tropical Crop Information Technology of Hainan Province,Haikou,Hainan 571101,China;Institute of Tropical Bioscience and Biotechnology,Chinese Academy of Tropical Agricultural Sciences,Haikou,Hainan 571101,China;Key Laboratory of Tropical Island Land Surface Processes and Environmental Changes of Hainan Province,Haikou,Hainan 571158,China;Key Laboratory of Agricultural Remote Sensing,Ministry of Agriculture and Rural Affairs,Beijing 100081,China)

机构地区：[1]中国国土勘测规划院/自然资源部土地利用重点实验室,北京100035 [2]中国热带农业科学院科技信息研究所/海南省热带作物信息技术应用研究重点实验室,海南海口571101 [3]中国热带农业科学院热带生物技术研究所,海南海口571101 [4]海南省热带海岛地表过程与环境变化重点实验室,海南海口571158 [5]农业农村部农业遥感重点实验室,北京100081

出　　处：《热带作物学报》2021年第11期3351-3357,共7页Chinese Journal of Tropical Crops

基　　金：自然资源部土地利用重点实验室开放研究项目;海南省自然科学基金面上项目(No.619MS100);海南省热带海岛地表过程与环境变化重点实验室开放课题(No.DLZDSYS202101)。

摘　　要：土地利用/覆盖变化(land use/cover change,LUCC)是当前全球变化研究的核心内容之一。土地利用遥感监测是土地利用变化相关研究的重要技术手段,尤其是高分辨率遥感技术和谷歌地球引擎(Google Earth Engine,GEE)云计算平台的出现,为土地利用空间信息的获取提供了新的途径和方法。本研究基于GEE云平台提供的Landsat-8 OLI时间序列卫星影像数据,采用随机森林(random forest,RF)和支持向量机(support vector machines,SVM)分类算法,对海南岛土地利用类型进行了遥感分类研究。结果表明:RF与SVM算法对海南岛土地利用中水体和建筑用地的分类精度均较高,对耕地、园地和林地分类精度较低。与SVM方法相比,RF分类方法能够更准确识别各类地物信息,更适于海南岛土地利用分类的研究。海南岛林地(包括天然林、橡胶林等)所占比例最大,主要分布在海南岛中部;耕地和园地面积接近,相间分布于海南岛大部分区域;水体和建筑用地面积较小,在海南岛均呈零散的分布状态,以沿海地区为主。GEE平台对于开展大区域土地利用分类与遥感动态监测具有重要的意义。Land use/cover change(LUCC) is one of the core contents of global change research. Land use remote sensing monitoring is an important technical for land use change research, especially the emergence of high-resolution remote sensing technology and Google Earth Engine(GEE) cloud computing platform, which provides a new way and method for obtaining land use spatial information. Based on the Landsat-8 OLI time series data provided by GEE cloud platform, the random forest(RF) and support vector machines(SVM) classification algorithm was used to mapping land use in Hainan Island. The results show that both RF and SVM algorithms have higher classification accuracy for water and building land, and have lower accuracy for cultivated land, garden land and forest land. Compared with SVM method, RF classification method could identify all kinds of land features more accurately and is more suitable for the study of land use classification in Hainan Island. The largest proportion of forest land(including natural forest, rubber forest, etc.) in Hainan Island is mainly distributed in the central part of Hainan Island. The area of cultivated land and garden land are distributed alternately in most areas of Hainan Island. The area of water body and construction land is small, which are scattered in Hainan Island, mainly in coastal areas. GEE platform is an useful tool for land use classification and remote sensing dynamic monitoring in large areas.

关 键 词：GEE LANDSAT 土地利用 海南岛 

分 类 号：P237[天文地球—摄影测量与遥感]