Carbon Dioxide Emissions Estimation And Spatio-temporal Distribution Of Guangdong Province Based On Nighttime Light Data

In the process of high-speed development of the Chinese economy,the large amount of fossil energy consumed due to industrial production and population activities which accompanied by huge emissions of carbon dioxide.Today,China has leaped to become the world's largest carbon emitter.In recent years,although the growth trend of China's carbon emissions has slowed down,but because China has a huge energy consumption base,it still has to emit large amounts of carbon dioxide into the atmosphere each year.Faced with the severe situation of environmental degradation,China is undertaking a huge responsibility for energy conservation and emission reduction.The acquisition of accurate carbon emission data which is the first step for China to carry out accurate emission reduction work.At present,most of China's energy consumption carbon emissions research is based on energy statistics yearbook.Due to the inconsistency of statistical data at the national,provincial and prefecture level,and the lack of statistical data on a large number of prefecture-level cities in China and the following administrative scales,most of the current carbon emission research in China can only be carried out for provincial-level regions and some developed cities.DMSP/OLS sensors can detect light from cities,townships,industrial sites,oil and gas combustion,and transient events such as fire and lightning lighting clouds.Therefore,DMSP/OLS nighttime remote sensing data can objectively and truly reflect human activities.Research by a large number of scholars at home and abroad shows that there is a strong correlation between the amount of light at night and the carbon emissions from energy consumption.The purpose of this paper is to use genetic neural networks to construct a model of the relationship between nighttime lighting data and carbon emissions from energy consumption,so as to use lighting data to more objectively and accurately estimate regional energy consumption carbon emissions.This paper focuses on the preprocessing of nighttime light data,the calculation of energy consumption carbon emissions data and the spatial and temporal distribution,as well as the light-carbon emission model construction,and so on.The following studies have been completed.?1?The image of 34 annual nightlight images acquired by F10,F12,F14,F15,F16and F18 sensors from 1992 to 2013 in the same year was multi-sensor image fusion and multi-year image inter-calibration with the method proposed by Elvidge et al.in2009,and calculate the conversion coefficient between the light image obtained by each sensor and the F16 satellite acquisition image in 2006.?2?Using the statistical yearbook data of 21 cities in Guangdong Province from2004 to 2013 to calculate the energy consumption of carbon dioxide emissions in Guangdong Province over a decade.Through time trend analysis,it was found that the rate of carbon emission growth in Guangdong Province was in an overall slowing trend.By calculating the global Moran's I,the spatial autocorrelation of energy consumption in Guangdong Province was analyzed.It was found that the energy consumption of carbon emissions in Guangdong Province as a whole is a high value aggregate.By calculating the Anselin Moran's I,the spatial autocorrelation of carbon emissions in the cities of Guangdong Province was analyzed.It was found that Guangdong's energy-consuming carbon emissions as a whole are high-value aggregates.By calculating the local Moran's I index,the spatial autocorrelation of carbon emissions in the cities of Guangdong Province was analyzed.It was found that Guangzhou,Dongguan,and Shenzhen exhibited high spatial concentration of carbon emissions,and the carbon emission spatial clustering characteristics of Huizhou and Foshan cities changed from high values including low-value anomalies to high-value aggregates in ten years,while Qingyuan City The spatial low-value clustering from carbon emissions has no obvious clustering characteristics.The spatial clustering relationship of carbon emission in Jieyang is still an obvious low-value cluster.In general,the Pearl River Delta Region is a high-value area for energy consumption in Guangdong Province,which is also consistent with the spatial distribution characteristics of the resident population and GDP of Guangdong Province.In addition,the global Moran's I and the Anselin Moran's I were calculated for corrected light images with cities across Guangdong Province.From the perspective of spatial autocorrelation,there is a strong space-related link between the amount of nighttime light data and the CO₂ emissions in Guangdong Province.?3?Proposing to use Genetic Algorithm Neural Networks to construct a light-energy consumption carbon emission estimation model,and estimate the carbon emissions of energy consumption in the cities of Guangdong Province.The experimental results show that using the Genetic Algorithm Neural Networks estimation model constructed to estimate the carbon emissions of energy consumption of 21 cities in Guangdong in 2013,the relative error of estimated carbon emission values of most cities is within 30%,and the average relative error is 18.14%.On the other hand,the average relative error of the extrapolated value of the multi-regression model of fossil energy consumption carbon emissions constructed by the regional light quantity,population,and regional tertiary industry output was 33.43%.It can be seen that the Genetic Algorithm Neural Networks has a very good forecast for Guangdong Province's city-level energy consumption of CO₂ emissions.Not only that,the Genetic Algorithm Neural Networks trained with provincial statistics is used to estimate the energy consumption of carbon emissions from 30 provincial administrative units in China except Hong Kong,Macao,Taiwan,and Tibet.The average relative error of estimated carbon emissions is:20.57%.For the estimation of carbon emissions in urban and county-level cities in Guangzhou from 2007 to 2013,the average relative error of Genetic Algorithm Neural Networks estimations is 2.58%.It can be seen that the use of Genetic Algorithm Neural Networks also has a very good effect on the estimation of carbon emissions from energy consumption in the greater provincial and more residential districts.At the same time,this also fills the gap in the CO₂ statistics data generated by the lack of energy statistics in district and county administrative units in China.