Analysis of urban growth and estimating population density using satellite images of nighttime lights and land-use and population data
Analysis of urban growth and estimating population density using satellite images of nighttime lights and land-use and population data
复制标题
DOI：
10.1080/15481603.2015.1072400
复制
发表时间：
2015-01-01
影响因子：
6.7
通讯作者：
Yamagata, Yoshiki
中科院分区：
文献类型：
--
作者：
Bagan, Hasi;Yamagata, Yoshiki
利用网格化土地利用数据、人口普查数据和夜间灯光卫星图像，研究了1990 - 2006年日本城市扩张的时空动态。首先，我们将国防气象卫星计划（DMSP）的夜间灯光和土地利用数据映射到日本1公里(2)网格单元系统上，以确定每个网格单元内DMSP和城市土地利用的比例面积。然后，研究了人口密度、DMSP与城市面积的关系。城市/建成区面积与人口密度呈显著正相关，特大城市周边城市/建成区面积的快速扩张与人口增长相关。相比之下，农村地区和小城镇的人口密度急剧下降。统计分析表明，种群密度与DMSP的相关系数随着DMSP夜间灯亮度值的增大而增大。接下来，我们使用普通最小二乘（OLS）回归模型估计北海道地区的人口密度。数值评价结果表明，土地利用数据与DMSP相结合可用于人口密度预测。最后，以北海道札幌市为例，比较OLS与GWR模型。与OLS相比，GWR能更好地预测人口密度。
We investigated the spatiotemporal dynamics of urban expansion in Japan from 1990 to 2006 by using gridded land-use data, population census data, and satellite images of nighttime lights. First, we mapped Defense Meteorological Satellite Program (DMSP) nighttime lights and land-use data onto the 1 km(2) grid cell system of Japan to determine the proportional areas of DMSP and urban land use within each grid cell. Then, we investigated the relationships among population density, DMSP, and urban area. The urban/built-up area was strongly positively correlated with population density, and rapid expansion of the urban/built-up area around megacities was associated with population increases. In contrast, population density dropped steeply in rural areas and in small towns. Statistical analysis showed that correlation coefficients between population density and DMSP increased as the DMSP nighttime lights brightness value increased. We next estimated population density in the Hokkaido region using an ordinary least squares (OLS) regression model. Numerical evaluation of the results showed that the combination of land-use data and DMSP could be used to predict the population density. Finally, we compared OLS and geographically weighted regression (GWR) model for Sapporo city, Hokkaido. Compared with the OLS, the GWR can improve predictions of population density.