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 population simulation performance of different open geospatial datasets, and develops a comprehensive approach 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, multiple linear regression and geographically weighted regression methods are employed to comprehensively evaluate the population modeling capability of various geospatial remote sensing data. The data utilized 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 research findings indicate that, in estimating population distribution, Tencent location data and POI data 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 multisource 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.