摘要
夜间灯光的遥感图像可去除大部分的自然干扰,使人类活动获得有效的反映,为社会经济空间化研究提供技术支持。该文用Suomi-NPP卫星获取的夜间灯光数据与GDP数据建立多种空间关系模型,以探索两者的空间分布规律,并对其中的影响因素进行分析。文中先对中国内地各省级行政区进行灯光信息统计,并计算各灯光指标,从而选取最佳灯光指标。实验结果表明,灯光区域的归一化总辐射指数与统计GDP的相关性更强。再以其作为灯光指标与统计GDP建立线性与非线性空间模型,最后选出较好的模型对GDP进行预测,其中线性、幂指数和Logistic曲线模型的拟合优度均达0.8以上。采用幂指数模型对2014年各行政单元的GDP进行预测,其平均相对误差为26.0%。
Nighttime light imagery most natural disturbances, so nighttime images can be used to reflect human activity, especially for research on spatialization of socio-economic development. This study utilized night-light data collected by the Suomi-NPP satellite in a spatial correlation model with GDP data to analyze both the spatial distribution and factors influencing development. The model extracted the lighting information and calculated night-light indexes to select the best night-light index for each Chinese mainland province. Tests show that the normalized total radiance index relates well to the GDP, so this index is related to the GDP using linear and nonlinear spatial models to develop a model to predict the GDP. The fitting results for the linear, power law and logistic models are all greater than 0.8. The power law model predicts the GDP of each mainland provincial-level region in 2014 with an average relative error of only 26.0%.
作者
郭永德
高金环
马洪兵
GUO Yongde GAO Jinhuan MA Hongbing(Department of Electronic Engineering, Tsinghua University, Beijing 100084, China School of Government, Peking University, Beijing 100871, China)
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2016年第10期1122-1130,共9页
Journal of Tsinghua University(Science and Technology)
基金
清华大学自主科研计划资助项目(20131089381)
