基于LJ1-01夜间灯光影像的苏锡常地区人口空间化研究

基于NPP/VIIRS DNB和"珞珈一号"01星(LJ1-01)两种夜间灯光数据原始影像与苏锡常地区县级人口统计数据进行空间滞后回归建模,得到500 m×500 m和200 m×200 m两种空间尺度的人口密度格网图,并利用乡镇人口数据进行检验证明LJ1-01数据在人口空间化研究中具有更高的精度;在此基础上,基于电子地图兴趣点(POI)数据与人口分布之间的相关性,通过融合POI数据对LJ1-01原始影像的模拟结果进行优化,并在乡镇尺度上对人口空间化结果进行了精度评价。结果表明:(1)LJ1-01夜间灯光影像亮度值与人口数呈显著正相关,相关系数高于NPP/VIIRS夜间灯光数据;(2)LJ1-01新型夜间灯光数据适用于人口空间化研究,且其模型拟合效果整体优于NPP/VIIRS传统夜间灯光数据;(3)将LJ1-01夜间灯光数据与POI数据进行融合可有效改善其人口空间化模拟结果,空间滞后回归模型的复相关系数R~2提高至0.946 3。通过实践,可以发现LJ1-01夜间灯光数据具有实现精细尺度人口数据空间化的巨大潜力。

Modeling the Population Density of Su-Xi-Chang Region Based on Luojia-1A Nighttime Light Image

Abstract：Firstly, we established spatial lag regression model to estimate the population density of Su-Xi-Chang region based on NPP/VIIRS DNB, LJ1-01 nighttime light data original images and county level resident population data. Combining the regression functions of population and nightlight data, we arrived at two sets of gridded population density map with a spatial resolution of 500m×500m and 200m×200m, respectively. The accuracy of the two-gridded population dataset was estimated using demographic data at township level. The results prove that LJ1-01 data has higher precision in population spatialization study. Then, based on the correlation between electronic map point of interest (POI) data and population distribution, we optimized the simulation results of LJ1-01 original image by merging POIs. And the accuracy evaluation of population spatialization results was carried out at the township scale, also. The analysis shows that:(1) LJ1-01 night light image brightness value is significantly positively correlated with population, and the correlation coefficient is higher than NPP/VIIRS night light data.(2) Accuracy assessment results show that the night light of LJ1-01 is superior to NPP/VIIRS in population spatial processing research. We also show that the recently published LJ1-01 night light data is suitable for research on spatial processing of demographic data.(3) When the LJ1-01 night light data were combined with POIs, the complex correlation coefficient(R2) of the spatial lag regression model is increased to 0.946 3, indicating that this method can effectively improve the precision of population spatialization. Through the study of this paper, it can be find that the night light data of LJ1-01 has enormous potential value for the spatial processing of demographic data in the future.