兼顾空间自相关性和异质性的人口空间化模型

The model of population spatial distribution considering spatial autocorrelation and spatial heterogeneity

针对现有人口空间化模型并未兼顾变量间空间自相关性和空间异质性的问题,该文以高分辨率的“珞珈一号”灯光数据融合兴趣点数据,在验证变量间空间自相关性及空间异质性共存的基础上,提出了一种兼顾两类空间关系的地理加权自回归模型对温州市的人口进行空间化实证。结果表明:灯光数据结合POI数据能够有效地刻画出精细尺度的人口空间分布,模型中兼顾两类空间关系的地理加权自回归模型整体要优于空间自回归模型和地理加权回归模型;地理加权自回归模型可有效解决人口空间化中变量间空间自相关性和空间异质性共存的问题,提高人口空间化模拟精度。

In view of the problem that spatial autocorrelation and spatial heterogeneity between variables was not considered at the same time in the existed population spatialization model,this paper used high-resolution LJ1-01 light data to fuse point of interest data,on the basis of verifying the coexistence of spatial correlation and spatial heterogeneity,a Geographically weighted autoregressive model considering both types of spatial relations was proposed to spatialize the population in Wenzhou.The results showed that the light data combined with POI data could effectively describe the spatial distribution of population on a fine scale,the Geographically weighted autoregressive model considering two kinds of spatial relations was better than the spatial autoregressive model and the Geographically weighted regression model as a whole;the Geographically weighted autoregression model could effectively solve the problem of the coexistence of spatial autocorrelation and spatial heterogeneity in population spatialization to improve the precision of population spatialization simulation.