基于夜间灯光遥感数据与POI数据的城市中心识别——以深圳市为例

City Center Recognition Based on Night Light Remote Sensing Data and POI Data:Taking Shenzhen City as an Example

尝试结合夜间灯光遥感数据(NPP-VIIRS)和POI数据,去识别深圳市的城市中心。首先,通过实验确定了多分辨率分割的最佳参数并确立了统一的空间单元计算POI的密度;其次,通过计算Anselin Local Moran′s I结合人类活动密度把高-高聚类单位定义为城市主中心,并使用地理加权回归模型根据POI密度的平方根与分割单元的几何中心到城市主中心的加权距离来确定子中心,识别出了2个城市主中心和11个城市子中心;最后通过与传统识别城市中心的方法进行对比分析。结果表明:多尺度分割、Anselin Local Moran′s I与地理加权回归结合的方法更好。

This article attempts to combine night light remote sensing data(NPP-VIIRS)and POI data to identify the city center of Shenzhen city.First,the optimal parameters of multi-resolution segmentation are determined through experiments and a unified spatial unit is established to calculate the density of POI.Second,the high-high clustering unit is defined as the main center of the city by calculating Anselin Local Moran′s I combined with human activity density,and it uses the geographically weighted regression model to determine the sub-centers based on the square root of the POI density and the weighted distance from the geometric center of the segmentation unit to the main center of the city.Two main city centers and eleven city sub-centers are identified.The methods of identifying the city center are compared and analyzed.The results show that through multi-scale segmentation,Anselin Local Moran′s I combined with geographically weighted regression is better.