建筑区夜间灯光的人口空间化模拟  

Population Spatialization Simulation based on Night Time Light in Built-up Areas

作　　者：李明[1,2] 张婷煜 方敏 肖天兰 王毓乾[1,2,3] 谢相建 LI Ming;ZHANG Tingyu;FANG Min;XIAO Tianlan;WANG Yuqian;XIE Xiangjian(Key Laboratory of Mine Environmental Monitoring and Improving around Poyang Lake of Ministry of Natural Resources,East China University of Technology,330013,Nanchang,PRC;School of Surveying and Geoinformation Engineering,East China University of Technology,330013,Nanchang,PRC;CNNC Engineering Research Center of 3D Geographic Information,East China University of Technology,330013,Nanchang,PRC)

机构地区：[1]东华理工大学自然资源部环鄱阳湖区域矿山环境监测与治理重点实验室,南昌330013 [2]东华理工大学测绘与空间信息工程学院,南昌330013 [3]东华理工大学中核三维地理信息工程技术研究中心,南昌330013

出　　处：《江西科学》2023年第5期887-894,共8页Jiangxi Science

基　　金：江西省自然科学基金项目(20202BABL202045,20202BABL213029);东华理工大学研究生创新基金项目(DHYC-202221);国家级(省级、校级)大学生创新创业训练计划项目(S202210405002)。

摘　　要：基于夜间灯光的人口空间网格化通常使用行政区全境的夜间灯光,忽略了土地覆盖信息,而人口的空间分布与建筑区的夜间灯光具有紧密关系。分别利用建筑区与行政区全境的NPP/VIIRS夜间灯光数据结合DEM和NDVI数据,选取了线性回归、支持向量机与随机森林3种代表性模型,对南昌市2020年人口进行了100 m的空间网格化模拟。结果表明,基于建筑区的网格化结果,不论在不同模型的拟合精度,还是不同区县的人口模拟精度上都要优于基于行政区全境网格化结果;在单个市级区域,选择的3种代表性模型中线性模型能取得最好的效果。基于建筑区的人口空间网格化结果去除了无人区地物的影响,能直观地展现城市内建筑区的人口密度分布,对城市管理有重要的意义。Night time light remote sensing based population distribution spatial gridded simulation usually used the night time light of the whole administrative area and ignored land cover information.But population spatial distribution holds very close relationship with the night time light of the building area.In this paper,the 100 meters spatial gridded population distribution of Nanchang City in 2020 was simulated by three selected typical models,including linear regression,support vector machine and random forest,in the use of NPP/VIIRS night time light data source of the whole administrative area as well as the building area integrated by DEM and NDVI data.The experiment results showed popultion spatial distribution simulation based on night time light of the building area would entirly own higher model fitting acurracy than the whole administrative area either by different models or on different districts and counties;linear models held the best performances among these three chosen typical models in a single municipal area.The population grid simulation based on building area removed the influences of depopulated zone,intuitively showed the population density distribution on the building area of city,which would be of great significance to the urban management.

关 键 词：人口密度 网格化 NPP/VIIRS 建筑区 机器学习 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]