| 基于时间序列DMSP/OLS夜间灯光数据的GDP预测模型 |
| |
| 作者姓名： | 顾鹏程 王世新 周艺 刘文亮 尚明 |
| |
| 作者单位： | 1. 中国科学院遥感与数字地球研究所, 北京 100101;
2. 中国科学院大学资源与环境学院, 北京 100049 |
| |
| 基金项目： | 国家重点研发计划（2017YFB0503805）和高分辨率对地观测系统重大专项（00-Y30B14-9001-14/16）资助 |
| |
| 摘 要： | 采用基于不变目标区域法对1992-2013年DMSP夜间灯光数据进行相互校正、饱和校正和影像间的连续性校正，提取出同期中国大陆31个省级行政区夜间灯光强度信息，并与统计GDP数据建立线性、指数、二次项和乘幂回归模型。通过比较预测GDP与统计GDP误差，选出各自的最优拟合模型。结果表明：1）校正解决了DMSP夜间灯光长时间序列影像之间不稳定、不连续的问题；2）校正后的DMSP夜间灯光数据集与GDP强相关；3）中国大陆GDP预测的指数模型最佳，R2达到0.97，平均相对误差仅为11.32%；4）31个省级行政区按时间序列构建GDP预测模型优于每年各省级行政区模型。4个直辖市和经济总量前6的行政区指数模型最优，其余省份二次项模型最优，R2均达到0.95以上，GDP预测相对误差10%左右。
|
| 关 键 词： | DMSP/OLS GDP 长时间序列 空间关系模型 |
| 收稿时间： | 2017-12-26 |
| 修稿时间： | 2018-03-29 |
Estimation of GDP based on long time series of DMSP/OLS nighttime light images |
| |
| Authors: | GU Pengcheng WANG Shixin ZHOU Yi LIU Wenliang SHANG Ming |
| |
| Institution: | 1. Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China;
2. College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, China |
| |
| Abstract: | In this study, the DMSP/OLS nighttime satellite data of China from 1992 to 2013 were used to find the relationships between GDP and nighttime satellite data. The nighttime satellite imageries were corrected by mutual correction, saturation correction, and continuity correction based on the invariant target region method. Then the lighting information of Mainland China and 31 provincial regions were extracted and models between GDP and light information, including linear, quadratic polynomial, power function, and exponential regression models, were tested to find the optimal ones. The results are showed as follows. 1) The corrected DMSP/OLS nighttime satellite data are more stable and continuous than the uncorrected data. 2) There is a strong correlation between the corrected DMSP/OLS night light dataset and GDP. 3) Exponential model was the most suitable one for predicting GDP of Mainland China, with the R2 value of 0.97 and MARE of 11.32%.4)Provincial models of long time series are better than the annual provincial administrative region models. The exponential function models were optimal for the four municipalities and the top six provincial economic entities, and the quadratic polynomial models were optimal for the other administrative regions, whose R2 values are above 0.95 and MARE's are about 10%. |
| |
| Keywords: | DMSP/OLS GDP long time series spatial correlation model |
|
| 点击此处可从《》浏览原始摘要信息 |
| 点击此处可从《》下载免费的PDF全文 |
