基于长时序夜光遥感数据的长三角区域GDP预测研究

Research on GDP Prediction of Yangtze River Delta Based on Long Time-series Night Light Remote Sensing Data

融合了DMSP/OLS(Denfense Meterological Satellite Program/Operational Linescan System)和NPP/VIIRS(Suomi Nation Polar-orbiting Partnership/Visible infrared imaging radiometer suite)两种常见的夜光遥感数据,延长了夜光遥感数据的时间跨度,以长三角区域的三省一市为例,对比了一元ARIMA模型、加入了NPP/VIIRS数据夜光总量ARIMAX模型和加入本文融合两种夜光数据的夜光总量ARIMAX模型,对研究区域的GDP进行了研究,以2019—2021年作为预测年份,与真实值进行对比。实验结果表明,不同经济结构在GDP预测结果的精度不同,加入夜光遥感数据的模型在预测经济发达并且面积较小的区域GDP时精度会显著降低,融合后的夜光数据GDP平均预测精度相比之前有了大幅度提高,可以很好地改进GDP预测结果。

This paper combines DMSP/OLS and NPP/VIIRS two kinds of common night light remote sensing data to extend the time span of night light remote sensing data.Taking three provinces and one city in the Yangtze River Delta region as an example,a comparison was made between the univariate ARIMA model,the ARIMAX model with NPP/VIIRS data added,and the ARIMAX model with the fusion of two types of night light data in this paper to study the GDP of the study area.2019-2021 was used as the predicted year,and compared with the actual value.The experimental results show that different economic structures have different accuracy in GDP prediction results.The addition of night light remote sensing data models will significantly reduce the accuracy of predicting GDP in economically developed and small areas.The average GDP prediction accuracy of the fused night light data has been significantly improved compared to before,which can effectively improve GDP prediction results.