基于POI数据的人口分布格网化方法研究

A Gridding Method of Redistributing Population Based on POIs

人口数据空间化展现了客观世界的人口分布,以行政区划为单元计算的人口密度难以体现实际人口分布的差异性,精细尺度的人口分布估算对于与人类活动相关的研究有重要意义。该文提出一种基于POI数据的人口分布格网化方法:首先,利用深圳市街道人口统计值和土地利用分类数据进行多元线性回归建模,获得50m格网尺度下的人口分布;然后,借助POI数据与人口数据间的相关性,结合四叉树索引获得相关性较高的POI中心点集;最后,建立一种新的格网人口重分配模型,更接近实际人口空间分布。研究结果表明,所提出的人口分布格网化模型能较好地拟合街道和建筑物人口统计值,为微观和更精确的人口估算奠定了基础。

Many activities,such as demographic studies,public transportation,and urban planning,require data on population at a fine resolution.However,to protect privacy,traditional demographic data are aggregated only based on administrative divisions(provinces,municipalities,districts,streets),thereby not available at finer resolution.Therefore,gridded population distribution data are highly desirable because they are closer to actual population distribution.In this paper,we propose a gridding method of redistributing population based on POIs.Firstly,we establish a modified multi-factor linear regression model based on the land use data.As a result,the population densities of different land use categories in different streets can be different.Secondly,a 50 m gridded population dataset is acquired from the result produced by previous step.Thirdly,sixteen categories of Baidu map POIs are extracted to select POIs with the highest correlation,and the center points from those POIs are identified based on Quadtree.Finally,the population density of each grid is recalculated according to a decaying function of the radial distance from center points.The proposed model is applied to estimate the 50 m gridded population distribution of eight districts in Shenzhen,Guangdong Province,China.Results prove that the population redistribution model is capable of producing high precision population estimates at the level of streets and buildings and depends on the high relevance POIs.In a conclusion,the gridding method of redistributing population based on POIs has great potential for applications.