论文部分内容阅读
随着社会经济数据的省域空间化需求的增加和夜间灯光数据应用的日渐成熟,基于北京市的DMSP-OLS夜间灯光数据,根据相关分析和回归分析的方法,建立综合灯光指数与第二、第三产业GDP的回归模型;利用北京市的Landsat8数据,依据CART决策树算法分类出高精度的土地利用类型图,构建耕地、草地、林地和水域面积与第一产业GDP的回归模型;通过对模拟的第一、二、三产业GDP共同求和的方式计算GDP总值,最终建立北京市的1km格网GDP空间分布图。结果表明,模拟的第一产业GDP的平均残差和相对误差分别为0.04亿元和3.55%,第二、三产业GDP的平均残差和相对误差分别为-5.54亿元和3.35%,GDP总值的平均残差和相对误差分别为-6.43亿元和3.36%。生成的GDP密度图能较全面地反映北京市的社会经济分布特征,可为经济决策和GDP产值估算提供依据。
With the increase of spatial demands of provincial and municipal socio-economic data and the maturity of night-time light data application, based on the DMSP-OLS night light data of Beijing and the correlation analysis and regression analysis methods, a comprehensive light index and the second, The regression model of tertiary industry GDP; Landsat8 data in Beijing, based on the CART decision tree algorithm to classify the high-precision land use type map to build a regression model of cultivated land, grassland, forest land and water area and the GDP of the primary industry; Simulation of the first, second and third industry GDP total sum of GDP to calculate the total, the final establishment of Beijing 1km grid GDP spatial distribution map. The results show that the average residual and relative errors of the simulated primary industry GDP are 0.04 billion yuan and 3.55% respectively. The average residuals and relative errors of the secondary and tertiary industries are -5.54 billion yuan and 3.35% respectively. The average residuals and relative errors of the values were -6.43 billion yuan and 3.36% respectively. The generated GDP density map can comprehensively reflect the social and economic distribution of Beijing and provide the basis for economic decision-making and GDP output estimation.