基于多源数据的福建省GDP空间网格化

GDP Spatial Gridding Expression of Fujian Province Based on Multi-source Data

采用DMSP/OLS夜间灯光数据结合土地利用类型数据、DEM数据等派生的多指标数据对福建省2008年GDP进行网格化.结果显示:利用DMSP/OLS数据、土地利用数据、DEM多源数据的多元逐步回归,较单一DMSP/OLS与GDP的一元回归模型增加了农林牧业对GDP的贡献,提高了GDP的预测精度,但从残差显示在预测区域经济发达程度高的区域比区域经济发达程度较低的区域GDP预测精度偏低;因此按多元回归的相对误差(δ=0.5)为界限将多源数据指标重新多元逐步回归,可提高GDP网格化的精度;最后借助统计数据进行调整得到福建省县域尺度零误差的GDP网格化分布图,为DMSP/OLS数据在县域尺度的应用提供了新的思路.生成的结果图能较全面地反映县域尺度下的福建省区域经济的空间分布特征,其结果对区域经济分析、资源环境研究等具有一定的实用价值.

GDP data spatial gridding is the premise of creating space type social economy,how to scientific and accurate spatial gridding GDP is important way,in view of the urban and rural residence in light data and DEM data influence the differences of natural conditions,using the DMSP/OLS night light data combining with land use types of data,DEM data and other multiple index data of GDP data spatial gridding in fujian province 2008 years.The results show that using the DMSP/OLS data,land using data,DEM data of multi-source data of multivariate stepwise regression more than a single DMSP/OLS regression increase with GDP,forestry and animal husbandry＇s contribution to GDP,experimental results improves the prediction precision of GDP,but from the residuals display in the prediction of regional economy with high intensity lower than the economic strength of regional prediction accuracy is low,so according to the relative error of the multiple regression（δ=0.5） to limit the multi-source data indicators to do multiple stepwise regression,improve the accuracy of the GDP spatial gridding,and finally do adjustment with the aid of statistical data,the GDP of county scale zero error for spatial gridding map.