基于夜间灯光和土地利用数据的云南沿边地区GDP空间差异性分析

云南沿边地区包括8个地州,共56个县,其中有25个县市与老挝、缅甸和越南直接毗邻,具有重要的地缘位置。本研究利用土地利用数据和夜间灯光数据在实现云南沿边地区GDP空间化的基础上,对GDP的空间分布格局进行深入探讨,这对缩小区域经济差异及促进地区共同发展具有一定的指导意义。采用土地利用数据对国内生产总值（Gross Domestic Product, GDP）数据的第一产业进行空间化拟合,采用DMSP/OLS夜间灯光数据对GDP的第二、三产业进行拟合,将第一产业和第二、三产业空间化拟合的结果相加,实现云南沿边地区1992-2013年的GDP的空间化拟合。在此基础上对云南沿边地区GDP空间分布差异进行分析。结果表明：① 土地利用数据对第一产业建模的效果较好,拟合的多期数据的相对误差均低于1.12%,采用夜间灯光数据,基于“分类回归”方法对第二、三产业拟合相对误差最大仅为6.404%,最终二者之和拟合的GDP拟合精度都较好,相对误差最大仅为4.241%;② 22期GDP数据在空间分布上均呈现正的相关性,且均为显著集聚;③ GDP空间分布局部集聚的高值-高值区域集中在开远、蒙自等县域,低值-低值地区集中在绿春、西蒙等地区;④ 云南沿边地区县域之间的经济差异在1992-1996年逐渐增强,1996年之后,经济差异波动缩小,空间关联效应呈现波动式的增强和减弱;⑤ 云南沿边地区的三维插值结果均呈现出西北至东南一线的“洼地-丘陵-平地-高峰”地势变化格局,沿边地区的东南角地区即红河州的建水、个旧和开远等县市的GDP最高,“丘陵”地势主要集中在腾冲、保山市以及最南部的景洪地区,“洼地-平地”地势主要分布在沿边地区西北角的贡山和福贡等县域、西南角的西蒙和孟连等县及中部区域的绿春和江城县等地区。

Spatial Difference of GDP in Yunnan Border Area based on Nighttime Light and Land Use Data

Yunnan border area is an important geographic location. It is composed of 56 counties in 8 municipalities. Among which, 25 counties are adjacent to Laos, Myanmar, and Vietnam. Land use and nighttime light data were used in this study to explore the spatial pattern of GDP based on the spatialization of GDP in the Yunnan border area. This study was expected to inform policy on reducing economic gaps between regions and promoting regional common development. The land use data was used to spatially fit the Gross Domestic Product (GDP) from the first industry, and the DMSP/OLS nighttime light data was used to fit GDP from the second and third industries. The fitting results were summed up to realize the spatialization of total GDP in the border area of Yunnan province from 1992 to 2013. Based on this, the spatial difference of GDP in the Yunnan border area was analyzed. The results showed that: (1) The land use data could be well used to model the GDP from the first industry, with goodness of fit (R²) being greater than 0.82 in each year and overall relative error being less than 1.12%. The nighttime light data and the classification regression method were used to fit the GDP from the second and third industries. The maximum relative error of fitting was 6.404%, and the fitting accuracy of the sum of the two industries was satisfactory with the maximum relative error being only 4.241%; (2) The 22-phase GDP data of the Yunnan border area was positively correlated in space, presenting an obvious clusters; (3) The distribution of GDP cluster in the county was characterized by High-High values (HH) and Low-Low values (LL). The distribution of Low-High and High-Low values was scattered with no regularity. The clustered high values of GDP were concentrated in Kaiyuan, Mengzi, and other counties, while the clustered low values of GDP were concentrated in Luchun, Ximeng, and other counties; (4) The economic gap between counties in the Yunnan border area gradually increased from 1992 to 1996 followed by a decrease trend afterward. The spatial correlation effect showed a fluctuation of increase and decrease; (5) Results of three-dimensional interpolation in the Yunnan border area presented a topographical pattern of “depression-hill-flat-peak” from the northwest to the southeast. The counties in the southeast corner of the border area such as Jianshui, Gejiu and Kaiyuan and other counties in the Honghe municipality, had the highest GDP. The “hill” terrain was mainly concentrated in Tengchong, Baoshan city, and the southernmost Jinghong area. The terrain of “depression-flat” was mainly distributed in the counties such as Gongshan and Fugong in the northwest corner of the border area, the counties in the southwest corner of Ximen and Menglian, and in the central area, such as Luchun and Jiangcheng counties.