融合建筑物与POI数据的精细人口空间化研究  

Fine-Scale Population Spatialization by Synthesizing Building Data and POI Data

作　　者：刘正廉 桂志鹏[1,2,3] 吴华意 秦昆[1,3] 吴京航 梅宇翱[1] 赵晶 LIU Zhenglian;GUI Zhipeng;WU Huayi;QIN Kun;WU Jinghang;MEI Yuao;ZHAO Jing(School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,China;State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan 430079,China;Collaborative Innovation Center of Geospatial Technology,Wuhan 430079,China;Archives of Surveying and Mapping Achievements in Hubei Province,Wuhan 430070,China)

机构地区：[1]武汉大学遥感信息工程学院,湖北武汉430079 [2]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079 [3]地球空间信息技术协同创新中心,湖北武汉430079 [4]湖北省测绘成果档案馆,湖北武汉430070

出　　处：《测绘地理信息》2021年第5期102-106,共5页Journal of Geomatics

基　　金：国家重点研发计划(2017YFB0503704,2018YFC0809806);国家自然科学基金(41501434,41371372)。

摘　　要：精细的人口分布数据能刻画出行政单元内部细节的人口空间分布信息,为城市规划、灾害评估等相关研究和应用提供有效数据支撑。利用房屋建筑及高德兴趣点(point of interest,POI)数据提取建模因子,结合随机森林模型获取了武汉市2015年常住人口50 m空间化数据集。结果表明,相较于WorldPop数据集结果,所提出方法的结果在武汉市高、中、低3种不同人口密度社区单元均具有更高的拟合精度。Fine-scale population distribution data can depict detailed spatial distribution information of population in administrative unit,which can provide effective data support for urban planning and disaster assessment. The modeling factors extracted from the building data and point of interest(POI)data are used with random forest model to obtain the spatial data set of permanent population with 50 m resolution in Wuhan in 2015. Compared with the result of WorldPop dataset,the result of the proposed method has higher fitting accuracy in community units of Wuhan with high,medium or low population densities.

关 键 词：人口空间化 随机森林模型 建筑物类型 兴趣点 数据融合 武汉市 

分 类 号：P208[天文地球—地图制图学与地理信息工程] C913.31[天文地球—测绘科学与技术]