基于多源地理大数据我国县域人口估算方法研究  

Population mapping in China with multi-sourced geographical open data

作　　者：徐勇 郑从威 XU Yong;ZHENG Cong-wei(School of Geography and Remote Sensing,Guangzhou University,Guangzhou 510006,China)

机构地区：[1]广州大学地理科学与遥感学院,广东广州510006

出　　处：《广州大学学报（自然科学版）》2024年第4期32-45,共14页Journal of Guangzhou University:Natural Science Edition

基　　金：国家自然科学基金面上资助项目(42271477);广东省哲学社会科学“十三五”规划资助项目(GD20CYJ40)。

摘　　要：实时人口数据对于城市规划、资源管理和社会的可持续发展等方面至关重要。为了有效提升现有基于地理大数据的人口估算方法,研究全面对比分析不同开放地理数据的人口模拟性能,并发展综合遥感与新兴社交媒体用户数据,以实现区县级人口的高精度快速估算。文章以中国各区县为试验区,运用多元线性回归及地理加权回归方法,全面评估各类地理遥感数据对我国人口模拟的性能,采用的数据包括腾讯位置服务(LBS)数据、高德兴趣点数据(POI)、夜间灯光遥感数据和基于遥感所得的土地利用/覆盖数据等。研究结果显示,在预估人口分布方面,腾讯定位数据与兴趣点数据比遥感所得的土地利用/覆盖数据和夜间灯光卫星数据都要好,人口模拟精度分别为81.6%、70.8%、68.8%和63.0%。文章进一步综合运用多源地理数据,可实现85.4%总体人口模拟精度,研究结果和发现可为我国人口相关政策提供数据和技术支撑。Real-time population data is crucial for urban planning,resource management,and the sustainable development of society.In order to effectively enhance existing population estimation methods based on geospatial big data,this study comprehensively compares and analyzes the popula-tion simulation performance of different open geospatial datasets,and develops a comprehensive ap-proach integrating remote sensing and emerging social media user data to achieve high-precision rapid estimation of population at the county level.Taking Chinese counties as the experimental area,multi-ple linear regression and geographically weighted regression methods are employed to comprehensively evaluate the population modeling capability of various geospatial remote sensing data.The data uti-lized include Tencent Location-Based Service(LBS)data,Amap Point-of-Interest(POI)data,nighttime light remote sensing data,and land use/cover data derived from remote sensing.The re-search findings indicate that,in estimating population distribution,Tencent location data and POI da-ta outperform remotely sensed land use/cover data and nighttime light satellite data,with population simulation accuracies of 81.6%,70.8%,68.8%,and 63.0%,respectively.Furthermore,the comprehensive use of multi-source geospatial data can achieve an overall population simulation accura-cy of 85.4%.The research results and discoveries can provide data and technical support for popula-tion-related policies in China.

关 键 词：人口 腾讯位置数据 POI数据 土地覆盖数据 夜间灯光 

分 类 号：K909[历史地理—人文地理学]