Study On Spatio-temporal Dynamic And Driving Forces Of Energy Consumption In China Based On Nighttime Light Data

Energy is an important material foundation of survival and development of human and society.China is a major power of energy consumption,and its energy consumption lies front rank in the world for many years.In recent years,Energy demand caused by economic development remains strong and the dependence on foreign oil and natural gas energy is rising.This imbalance situation between supply and demand has a profound impact on our country's energy security and economic development.Meanwhile,greenhouse effect and air pollution caused by large amount of energy consumption need to be solved urgently.Therefore,it's essential to understand the change of its total amount and the evolution of its spatial pattern timely and accurately,and evaluate its influencing factors scientifically.Night light image reflects human life and production activities,and its application field is extended from estimating socio-economic factors,extracting urban information to estimating energy consumption.The empirical results show that night light data is an objective and effective way to obtain small-scale energy consumption.However,these studies are based on general regression models such as linear and quadratic polynomials,and the accuracy of the models has not been verified on a small scale.Besides,most researches on the influencing factors of energy consumption do not consider the spatial correlation between variables.Based on the statistical data of energy consumption in prefecture-level cities,this paper compares the precision of different models between provincial night light value and energy consumption statistics.With the better simulation result is selected,the spatiotemporal evolution characteristic is analyzed by exploratory data analysis method.Furthermore,a spatial Durbin panel model is constructed to explore the main influencing factors of energy intensity.The followings are the main conclusions of this paper:(1)Night light data is a reliable mean to acquire the temporal and spatial dynamic changes of energy consumption timely and accurately.The simulation result of panel data model between provincial night lighting value and energy consumption statistics is better than quadratic polynomials.(2)From 1995 to 2016,most parts of the country are low-growth and low-consumption.Three regions have significant differences in energy consumption behavior,and high-growth areas and highconsumption areas are mainly distributed in the eastern region.(3)The result of provincial global Moran Index indicates there isn't an obviously spatial correlation of energy consumption,but its local Moran index depicts different local characteristics.(4)The distribution of municipal energy consumption shows a significant positive spatial correlation pattern,and its spatial agglomeration has been strengthened year by year.In terms of temporal and spatial variation,the change in the distribution of high-high and low-low cluster is the most obvious,and the distribution of the other twocluster types is relatively stable.(5)The spatial spillover effect of provincial energy intensity is evident.The development of economy and the improvement of foreign direct investment(FDI)have promoted the reduction of energy intensity in this region,which is different from the influence of urbanization and industrial structure.And FDI and industrial structure have spatial spillover effects on other regional energy intensity.