基于GDP/TND建模的贵州省县级GDP估算及时空格局分析

County-level GDP Estimation and Spatio-temporal Pattern Analysis in Guizhou Province Based on GDP/TND Modeling

夜间灯光强度是区域经济发展程度的直接反征,为社会经济参数的估算提供了新的思路和方法。本文将贵州省88个县级行政区作为研究区域,以GDP为其重要的社会经济参数,通过三种建模方法研究NPP-VIIRS灯光影像数据与GDP数据之间的空间关系,利用最优模型对2017年的GDP数据进行估算分析,并在此基础上,探究了贵州省2012~2017年经济发展的时空特征,分析了造成贵州省极核式空间发展格局的原因。研究结果表明:(1)GDP与灯光指标TND的pearson相关系数最高,相关性较好,优于MND和SND两项灯光指标;(2)按GDP/TND大小分级建模的GDP估算精度最高,按GDP大小分级建模的估算精度次之,未分级建模的估算精度最弱;(3)采用GDP/TND分级建模时,统计GDP与灯光指标TND的R~2均高于0.9,平均相对误差为8.29%;(4)贵州省2012~2017年经济发展速度整体呈现西高东低的趋势,花溪区GDP增长最快,与统计年鉴分析结果吻合度高。

The intensity of night light is a direct reflection of regional economic development,which provides a new idea and approach to estimate social and economic parameters.This article 88 county-level administrative regions in guizhou province are taken as the research areas,whose GDP is its important social and economic parameters,through study on three modeling methods to researchthe spatial relations of NPP-VIIRS light image data and GDP data,using the optimal model to estimate the 2017 GDP data analysis,and based on this,exploring the 2012 to 2017 years of economic development in guizhou characteristics of space and time,analyzing the causes of the polar core spatial development pattern in Guizhou province.The results show that:(1)the pearson correlation coefficient between GDP and lighting indicator TND is the highest and the correlation is good,which is better than MND and SND;(2)the accuracy of GDP estimation based on the SCALE of GDP/TND is the highest,followed by that based on the scale of GDP,and the estimation without the scale of GDP is the weakest;(3)when GDP/TND grading modeling was adopted,R~2 of both GDP and lighting indicator TND was higher than 0.9,with an average relative error of 8.29%;(4)the overall economic development rate of Guizhou from 2012 to 2017 performed a trend of higher in the west and lower in the east,with Huaxi District having the fastest GDP growth,which was highly consistent with the analysis results of Guizhou Statistical Yearbook.