基于NPP/VIIRS和LJ1-01夜间灯光数据的人口空间化对比研究——以北京市为例

A Comparative Study of Population Spatialization Based on NPP/VIIRS and LJ1-01 Night Light Data:Taking Beijing for an Example

为探究珞珈一号01星(LJ1-01)获取的最新一代夜间灯光数据在人口空间化研究的适用性,采用NPP/VIIRS夜间灯光数据、LJ1-01夜间灯光数据、WorldPop人口数据、土地利用数据等,在乡镇(街道)尺度上分别建立线性回归模型、GWR模型和RFR模型,并对建模结果进行评价。结果表明:线性回归模型和GWR模型在人口密度大的区域模拟效果较好,RFR模型在整个区域表现更稳定;GWR模型的精度与空间自相关程度有关,在分区构建GWR模型时还要考虑空间因素的影响;基于LJ1-01夜间灯光数据的人口空间化结果在拟合度以及精度方面均优于NPP/VIIRS。根据实验结果可知,由LJ1-01数据制作的人口空间分布图能够清晰地展示城市内部人口空间分布的差异,在人口空间化研究方向具有巨大潜力。

In order to explore the applicability of the latest generation of night light data obtained by Luojia-101(LJ1-01)in population spatialization studies,NPP/VIIRS night light data,LJ1-01 night light data,WorldPop population data,land use data,etc.are used as the source data,linear regression model,GWR model and RFR model are established on the township(street)scale,and the modeling results are evaluated.The results show that:the linear regression model and the GWR model have better simulation effects in areas with high population density,and the RFR model performs more stable in each area;the accuracy of the GWR model is related to the degree of spatial autocorrelation,and the influence of spatial factors should also be considered when building GWR model by zones;the population spatialization results based on LJ1-01 night light data are better than that based on NPP/VIIRS data in terms of fit and accuracy.According to the experimental results,the population spatial distribution map produced by the LJ1-01 data can clearly show the difference in the spatial distribution of the population within the city,which has great potential in the direction of population spatialization research.