夜间灯光数据与统计数据的相关性研究——以川南经济区为例

Relevant Research Between NPP-VIIRS Data and Statistics-A Case Study of South Sichuan Economic Zone

提取了川南经济区2015年NPP-VIIRS夜间灯光数据灯光计数、灯光总量、灯光均值、灯光标准差、灯光最小值、灯光最大值等6个指标,并收集整理了该地区2015年与城市化密切相关的地区生产总值、第一产业增加值、地方公共财政支出、城镇化率、年末常住人口等5个社会统计数据指标,发现不同尺度下各指标的相关性有较大差异.在此基础上,利用偏最小二乘法对地市级尺度上夜间灯光数据的6个指标和社会统计数据的5个指标进行了回归,比较了县级尺度上偏最小二乘回归与最小二乘回归的效果.结果表明:利用夜间灯光数据回归地区生产总值、第一产业增加值、地方公共财政支出、城镇化率、年末常住人口时均能通过检验,但回归效果上有一定的差异;用夜间灯光数据回归以上社会统计数据时不能只考虑灯光总量,在对年末常住人口等指标进行回归时灯光最大值等也起着及其重要的作用.

Extracted six indicators of NPP-VIIRS nighttime lighting data in southern Sichuan Economic Zone in 2015,including count,sum,mean,stdev,min and max,and sorted out five social statistical data indicators closely related to urbanization in 2015,including GDP,Primary Industry,Expenditure of Local Public Finance,Urbanization Rate and Permanent Population at the Year-end,and it was found that the correlation of each index is quite different under different scales.On this basis,the partial least squares regression(PLSR)model was used to regress six indexes of night light data and five indexes of social statistics on the municipality scale,and the effects of partial least square regression and least square regression on the county level scale are compared.The results showed that the regression of GDP,Primary Industry,Expenditure of Local Public Finance,Urbanization Rate and Permanent Population at the Year-end can pass the test,but there are some differences in the regression effect;When regressing the above social statistical data with night light data,we can not only consider the total amount of light,but also the maximum value of light plays an important role in the regrcssion of Permanent Population at the Year-end and other indicators.