Studies On The Spatiotemporal Patterns And Driving Factors Of Energy-induced Carbon Dioxide Emissions In China Based On Multi-source Data

The global climate change caused by carbon dioxide emissions(CO2)has aroused the concern of the international community.CO2 emissions become one of important factors which restrict China's sustainability.Understanding spatial and temporal dynamics of CO2 emissions of China at virious spatia,l scales is crucial for adaption and mitigation of climate change.However,the unavailability of energy statistics at small scales,such as prefectural level,prevents a better understanding of CO2 emissions in previous studies.Though nighttime light(NTL)has been widely used as a proxy for estimating carbon emissions and many other socio-environmental issues,the limited range of sensor radiance of DMSP/OLS(Def'ense Meteorological Satellite Program's Operational Linescan System)prevents its application and estimation accuracy.The aims of this paper are:(1)Improving the performance of DMSP/OLS NTL as a proxy in socio-environmental estimations by fusing multi-year NTL images with population density,Normalized Difference Vegetation Index(NDVI)and water body data;(2)Applying the modified NTL to estimate CO2 emissions in China at 1 km × 1 km scale;(3)Exploring the spatiotemporal dynamics at different spatial scales in China during the last 20 years;(4)Analyzing the the impact factors of CO2 emissions in the context of rapid urbanization.The main findings of this study are as follows:(1)This paper proposed an index named PVANUI,which could be used to calibrate the saturation and blooming problems and to reduce the estimation error in urban core and rural areas.As one of the most important spatiotemporal indicator for human activities,DMSP/OLS NTL are strongly correlated with energy consumption and can be applied for estimating carbon emissions at global,regional,country,and city levels.However,some defects of the data source limit the estimated accuracy at a small scale.We developed an improved index,which combines NTL with time-series NDVI,population density,and water body data to calibrate the saturation and blooming problems of original NTL and to reduce the CO2 estimation error in urban core and rural areas in this paper.The results show that PVANUI can add information about human activities in the unlit pixels and can reduce saturation and spillover effects,which show better performance in capturing spatial details in urban areas than original DMSP/OLS NTL,nighttime light calibrated by population(PNTL)and NDVI-corrected NTL(VANUI).(2)Based on PVANUI,this paper estimated energy-induced carbon emissions in China at 1 km × 1 km scale.The results of the accuracy verification show that the estimations based on PVANUI are better than those based on the original NTL,PNTL and VANUI,as the average of R2 of models using PVANUI is the highest and the estimated error is the smallest.The estimated results from PVANUI can provide basic data to support the study of carbon emissions at the prefecture-city level or at even smaller scale.(3)This paper analyzed the spatiotemporal dynamics of CO2 emissions in China at muti-levels during the past 20 years.The results indicate that the CO2 emissions from energy consumption in China experience a three-stage trajectory.The spatial distribution of CO2 emissions is significant uneven.Carbon emissions in China are highly constrate in eastern and northern area.Besides the developed areas such as Beijing-Tianjin-Hebei,Yangtze River Delta Region and Pearl River Delta Region,carbon emissions are also concentrated in Inner Mongolia,Xinjiang and other northern region,which are rich in coal resources and own heavy industry.We employed two-stage nested Theil decomposition to quantify the regional disparities of emissions.The results show that as the decline of within-province disparity,the total disparities decrease continuously from 1995 to 2013.The between-province disparity becomes the highest contributor to the total disparities.We also used Exploratory Time-Space Data Analysis to explore the spatiotemporal patterns of carbon emissions at a prefectual city level.The results indicate that there exist significant spatial agglomerations of CO2 emissions at a prefectural level.Most cities in eastern region belong to high-high agglomerations,while many cities in the western region belong to low-low agglomerations.The relative small frequency of changes in the local Moran's I transition probability matrix indicate a stable local spatial autocorrelation in China.(4)This paper explored the impact factors of economic growth,urbanization and energy consumption on carbon emissions.Granger causality was applied to discucuss the relationships among economic growth,urbanization,energy consumption and carbon emissions.The results indicate that there exists a long-term equilibrium relationship between per capita GDP,per capita energy consumption and urbanization.There is a two-way causality between energy consumption and GDP in a short-term.There are one-way Granger causalities between urbanization and GDP,energy consumption,and carbon emissions in a short-term.Economic development plays a positive and sustained role in promoting energy consumption and carbon emissions.Based on IPAT,we used STIRPAT regression model to analyze the impact factors.The results show that at the provincial level,the growth of population,per capita GDP,energy intensity,urbanization and construction land are positively correlated with carbon emissions,while the impact from proportion of tertiary industry are negative.There is an inverted U-type relationship between per capita GDP and carbon emissions,which indicate the existence of environmental Kuznets phenomenon in carbon emissions.The results of dynamic panel data model affirm the impact of time dependency on carbon emissions.Furthermore,the results of spatial lag model(SLM)and spatial error model(SEM)at prefectural level show that the impact from spatial autocorrelation is significant.Then,Geographical Weighted Regression was employed to explore the different impact of factors.The results confirm that population,per capita GDP and energy intensity have positive effect on carbon emissions,while the urbanization rate,the tertiary industry and the construction land have both positive and negative effects on carbon emissions.In addition,the impacts of urban forms on carbon emissions were quantified.The results show that urban size,fragmentation,irregularity and the dominance of the largest urban patch are positively correlated with carbon emissions.