Spatio-Temporal Variations of CO2 Emission from Energy Consumption

Home , Chizhou, Huainan, Huangshan City, Kunming, Lianyungang

sustainability

Article Spatio-Temporal Variations of CO2 Emission from Energy Consumption in the Yangtze River Delta Region of China and Its Relationship with Nighttime Land Surface Temperature

Juchao Zhao 1,2, Shaohua Zhang 2,*, Kun Yang 1,2,*, Yanhui Zhu 1,2 and Yuling Ma 1,2

1 School of Tourism and Geographic Science, Yunnan Normal University, Kunming 650500, China; [email protected] (J.Z.); [email protected] (Y.Z.); [email protected] (Y.M.) 2 The Engineering Research Centre of GIS Technology in Western China, Ministry of Education of China, Yunnan Normal University, Kunming 650500, China * Correspondence: [email protected] (S.Z.); [email protected] (K.Y.)

Received: 27 August 2020; Accepted: 7 October 2020; Published: 12 October 2020

Abstract: The rapid development of industrialization and urbanization has resulted in a large amount of carbon dioxide (CO2) emissions, which are closely related to the long-term stability of urban surface temperature and the sustainable development of cities in the future. However, there is still a lack of research on the temporal and spatial changes of CO2 emissions in long-term series and their relationship with land surface temperature. In this study, Defense Meteorological Satellite Program’s Operational Linescan System (DMSP/OLS) data, Suomi National Polar-orbiting Partnership (NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) composite data, energy consumption statistics data and nighttime land surface temperature are selected to realize the spatial informatization of long-term series CO2 emissions in the Yangtze River Delta region, which reveals the spatial and temporal dynamic characteristics of CO2 emissions, spatial autocorrelation distribution patterns and their impacts on nighttime land surface temperature. According to the results, CO2 emissions in the Yangtze River Delta region show an obvious upward trend from 2000 to 2017, with an average annual growth rate of 6.26%, but the growth rate is gradually slowing down. In terms of spatial distribution, the CO2 emissions in that region have significant regional differences. Shanghai, Suzhou and their neighboring cities are the main distribution areas with high CO2 emissions and obvious patch distribution patterns. From the perspective of spatial trend, the areas whose CO2 emissions are of significant growth, relatively significant growth and extremely significant growth account for 8.78%, 4.84% and 0.58%, respectively, with a spatial pattern of increase in the east and no big change in the west. From the perspective of spatial autocorrelation, the global spatial autocorrelation index of CO2 emissions in the Yangtze River Delta region in the past 18 years has been greater than 0.66 (p < 0.01), which displays significant positive spatial autocorrelation characteristics, and the spatial agglomeration degree of CO2 emissions continues to increase from 2000 to 2010. From 2000 to 2017, the nighttime land surface temperature in that region showed a warming trend, and the areas where CO2 emissions are positively correlated with nighttime land surface temperature account for 88.98%. The increased CO2 emissions lead to, to a large extent, the rise of nighttime land surface temperature. The research results have important theoretical and practical significance for the Yangtze River Delta region to formulate a regional emission reduction strategy.

Keywords: CO2 emission; nighttime land surface temperature; Yangtze River Delta region

Sustainability 2020, 12, 8388; doi:10.3390/su12208388 www.mdpi.com/journal/sustainability Sustainability 2020, 12, 8388 2 of 17

1. Introduction As the largest developing country in the 21st century, China has experienced the progressive transformation from planned economy to market economy and achieved a surprising economic growth over the past 30 years [1]. Since the reform and opening-up in 1978, the gross domestic product has been increasing with an average speed of 9.8% per year and ranked the third of the world in 2009 [2]. Driven by the fast economic development and the increase of primary energy consumption, the emission increase of CO2 in in eastern coastal areas has been accelerated significantly [3]. According to the statistics, the growth rate of CO2 emission in China per year has reached 10% since 2000. The overall CO2 emission in 2008 reached 8.325 billion tons for which China became the country with the largest emission in the world [4]. In one aspect, the huge emission of carbon and the higher acceleration impose a huge pressure on China for emission reduction [5]. A scientific and reasonable analysis on the variation of time and space for CO2 emission has become the fundamental precondition for the strategy of emission reduction. On the other aspect, the continuous increase of CO2 emission has generated a significant warming effect for the land surface temperature which not only impacts the residents’ health, production activities and atmospheric environment of cities, but also makes the soil infertile, impacts the water cycling system and even causes the irreversible destruction to the biological diversity [6,7]. The disclosure of the influence from CO2 emission on land surface temperature has become one of the key challenges for sustainable development. The nighttime light data are of long duration and wide coverage etc. which are ideal for the exploration of city information [8]. Many scholars have successfully applied the DMSP/OLS data into such fields as population estimation [9], urban boundary extraction [10], power consumption estimation [11] etc. Since human activities are the major source of CO2 emission and DMSL/OLS data happens to effectively reflect the strength of human activities [12], many scholars at home and abroad have proved that the application of DMSP/OLS data into the estimation of CO2 emission is feasible. For example, Wang et al. [13] applied spatiotemporal modeling based on DMSP/OLS data to evaluate the CO2 emission on the municipal level in China from 1992 to 2013 and put forward a series of feasible policies for emission reduction. Shi et al. [14] compared and analyzed the temporal and spatial variation and contributory factors for CO2 emissions on the provincial and municipal level from 1997 to 2012 based on DMSP/OLS data. Cheng et al. [15] studied the spatio-temporal dynamics of CO2 strength in energy consumption on the provincial level from 1997 to 2010 and its dominant factors based on the method provided by the International Panel on Climate Change (IPCC). These research studies mainly discuss and explore CO2 emission on the provincial or municipal level, most of which were conducted before 2013. With the launch of a new generation of NPP/VIIRS nighttime light data, the research on the spatio-temporal dynamics of serialized CO2 emissions over a longer duration has become possible. Based on the DMSPOLS and NPP/VIIRS data, this research explores the spatio-temporal characteristics of serialized CO2 emissions on the municipal level with a longer duration, in order to determine the objective of emission reduction for the local governments and provide scientific basis for the preparation of a regional strategy for emission reduction. Besides, the current research on the influence of CO2 emissions on land surface temperature has always been an interesting topic. Research indicates that the concentration enrichment of CO2 emission means more absorption of heat which causes the variation of land surface temperature [16]. For example, Vostok ice-core data suggest that the land surface temperature and CO2 emissions have a close relationship in the past 160 thousand years [17]. Jones et al. [18] discovered that fossil fuel and bio-fuel black carbon has significantly contributed to the climate warming in the last 50 years of the 20th century through the inspection and analysis based on the observation data of global near-surface temperature in the last century. Through selecting the global time series data in 161 years and applying the linear method and non-linear method to inspect the casual relationship between CO2 emissions and air temperature, Bai et al. [19] proved that CO2 emissions are the contributor for the climate warming. Das et al.’s [20] research suggests that the fossil fuel consumption may cause the CO2 level to rise to one and half times by the year 2020 in comparison to that of the 2008 level, thus causing an increase in the surface temperature by 0.0008 percent per Sustainability 2020, 12, 8388 3 of 17 annum. The above stated research studies have obtained comparatively ideal research achievements. However, the current research on the influence of CO2 emissions on land surface temperature is still comparatively insufficient and does not consider that factors such as solar radiation difference in daylight and human activities will influence the result of the correlation analysis on CO2 emissions and land surface temperature. Therefore, this research selects the nighttime land surface temperature as the data source since the data of nighttime land surface temperature have higher precision compared to the data of land surface temperature in daylight, and they are more typical. The Yangtze River Delta is the region with the highest population density and level of urbanization in China [21]. This region’s rapid economic growth inevitably resulted in a high concentration of energy consumption, which greatly increased CO2 emissions. To cope with the huge pressure of emission reduction and enhance the sustainable development of cities, the scientific and precise analysis on the temporal and spatial variation of CO2 emissions in the Yangtze River Delta and its relationship with the nighttime land surface temperature may determine the objective of emission reduction for the local governments and provide the theoretical basis and data support for the improvement of the urban ecological environment. The present study combined Geographic Information System (GIS) technology and remotely sensed data to construct a carbon emissions simulation model for the Yangtze River Delta region. The model was based on the theory of sustainable urban development and on DMSP/OLS data, NPP/VIIRS data, statistics on energy consumption and some basic geographic information data. The temporal and spatial origin, and the distribution pattern of CO2 emissions were determined using trend and spatial autocorrelation analyses. Pearson’s correlation coefficient was used to analyze the relationship between CO2 emissions and night surface temperature.

2. Datasets

2.1. Study Area Located in the lower reaches of the Yangtze River, the Yangtze River Delta is one of the largest urban agglomerations in China and has also been the region with the fastest economic development since China’s reform and opening up [22]. The range which the study selects contains Shanghai, Zhejiang (Hangzhou, Ningbo, Wenzhou, Shaoxing, Huzhou, Jiaxing, Jinhua, Quzhou, Zhoushan, Taizhou, Lishui), Jiangsu (Nanjing, Wuxi, Xuzhou, Changzhou, Suzhou, Nantong, Lianyungang, Huai’an, Yancheng, Yangzhou, Zhenjiang, Taizhou, Suqian), Anhui (Hefei, Bengbu, Wuhu, Huainan) Ma’anshan City, Huaibei City, Tongling City, Anqing City, Huangshan City, Fuyang City, Suzhou City, Chuzhou City, Lu’an City, Xuancheng City, Chizhou City, Bozhou City), as shown in Figure1. This area belongs to subtropical humid monsoon climate, with elevation below 1749 m, and most of the areas are plain, located at Shanghai, Jiangsu, Northern Anhui and Northern Zhejiang. The Yangtze River Delta, as an important intersection of the “Belt and Road” and the Yangtze River Economic Belt, is gradually developing into an inﬂuential world-class metropolis, and playing an important role in China’s social development [23].

2.2. Data Sources and Processing Data sources include DMSP/OLS Data, NPP/VIIRS Data, energy consumption statistics, nighttime land surface temperature data and other relevant data. SustainabilitySustainability 20202020, 12, 12, x, 8388FOR PEER REVIEW 4 of4 of 18 17

Figure 1. The location of research area. Figure 1. The location of research area. DMSP/OLS nighttime light data. These data come from the stable noctilucent remote sensing imageDMSP/OLS which is providednighttime by light National data. These Oceanic data and come Atmospheric from the Administrationstable noctilucent (NOAA) remote and sensing whose imagespatial which resolution is provided is 0.008333 by National degrees andOceanic whose and range Atmospheric of digital numberAdministration (DN) value (NOAA) is 0~63. and The whose larger spatialthe DN resolution value, the is higher 0.008333 the degrees light intensity. and whose Three range satellites, of digital namely, number F15 (2000–2007), (DN) value F16 is (2004–2009)0~63. The largerand F18the (2010–2013),DN value, the were higher used the for light data intensity. acquisition. Three Because satellites, the DMSPnamely, satellite F15 (2000–2007), is influenced F16 by (2004–2009)measurement and error, F18 (2010–2013), attenuation ofwere detection used for ability, data atmosphereacquisition. andBecause so on, the relative DMSP radiometricsatellite is influencedcorrection by must measurement be conducted error, when attenuation studying of and detection using DMSP ability,/OLS atmosphere Data [24 ].and By so referring on, relative to the radiometriccorrection ofcorrection nighttime must light databe conducted by Elvidge when et al. [studying25] and Wu and et al.using [26 ],DMSP/OLS this study uses Data ArcGIS [24]. By 10.2 referringas the data to the processing correction platform of nighttime and conductslight data the by followingElvidge et foural. [25] steps and to Wu correct et al. DMSP [26], this/OLS study data: uses(a) createArcGIS China 10.2 regionalas the data nighttime processing light databaseplatform (theand spatial conducts resolution the following of the data four after steps resampling to correct is 1 km 1 km) through projection change, resampling, clipping and other processing; (b) to correct the DMSP/OLS× data: (a) create China regional nighttime light database (the spatial resolution of the data afterDN valueresampling with multi-sensor is 1 km × data1 km) in thethrough same yearprojection to form change, the unique resampling, nighttime clipping light data and each other year; processing;(c) to correct (b) the to correct continuity the ofDN time value series with to multi ensure‐sensor the latter data DN in the value same of theyear annual to form pixel the is unique greater nighttimethan that light of the data previous each year; year; (c) and to correct (d) create the the continuity nighttime of time light series database to ensure of the the Yangtze latter RiverDN value Delta ofthrough the annual the pixel clipping is greater of nighttime than that light of the data previous based on year; the and administrative (d) create the division nighttime map light of the database Yangtze ofRiver the Yangtze Delta. River Delta through the clipping of nighttime light data based on the administrative divisionNPP map/VIIRS of the nighttime Yangtze River light data.Delta. The data are from the National Geophysical Data Center of theNPP/VIIRS United States. nighttime The selected light data. years The are data 2012–2017. are from The the spatialNational resolution Geophysical of the Data original Center data of the was 0.004167 degrees, and it was converted to 1 km 1 km by resampling. Different from DMSP/OLS data, United States. The selected years are 2012–2017.× The spatial resolution of the original data was 0.004167NPP /VIIRS degrees, data and do not it was remove converted the influence to 1 km of × Aurora, 1 km by firelight resampling. and other Different background from DMSP/OLS noise [27,28 ]. data,These NPP accidental /VIIRS data factors do willnot remove affect the the accuracy influence of of CO Aurora,2 Emission firelight Simulation and other and background must be removed. noise [27,28].Considering These accidental that the NPP factors/VIIRS will data affect are missingthe accuracy in some of CO areas2 Emission from May Simulation to July, and and in must order be to removed.make full Considering use of the information that the NPP/VIIRS of each image, data are the datamissing from in January some areas to April from and May August to July, to and October in ordereach to year make are full corrected, use of andthe information then the annual of each NPP image,/VIIRS the composite data from data January are obtained to April according and August to the tomonthly October NPP each/VIIRS year compositeare corrected, data and [29]. then Then, the in orderannual to NPP/VIIRS overcome the composite influence data of di ffareerent obtained satellite accordingsensors, the to the “two-step” monthly proposed NPP/VIIRS by Zhaocomposite et al. [ data30] is [29]. used Then, to correct in order the continuity to overcome of DMSP the influence/OLS and ofNPP different/VIIRS satellite data. Finally, sensors, we getthe the “two nighttime‐step” proposed light data by of 2000–2017Zhao et al. time [30] series, is used and to Figure correct2 shows the continuitythe time change of DMSP/OLS of total DN and values NPP/VIIRS of nighttime data. Finally, light data we in get the the Yangtze nighttime River light Delta data from of 20002000–2017 to 2017. timeIn Figure series,2, and TDN Figure represents 2 shows the total the DNtime value change of nighttime of total DN light values data in of the nighttime entire Yangtze light data River in Delta. the Yangtze River Delta from 2000 to 2017. In Figure 2, TDN represents the total DN value of nighttime light data in the entire Yangtze River Delta.

Sustainability 2020, 12, 8388 5 of 17 Sustainability 2020, 12, x FOR PEER REVIEW 5 of 18

Figure 2. TotalTotal digital number (TDN) values in the Yangtze River Delta region fromfrom 20002000 toto 2017.2017.

Statistical data of energy consumption in the YangtzeYangtze RiverRiver Delta.Delta. The energy consumption data used to calculate CO2 emissions are from the reference energy terminal consumption in the energy consumption balance table of Zhejiang, Anhui, Jiangsu and Shanghai in ChinaChina EnergyEnergy StatisticalStatistical Yearbook. ChinaChina EnergyEnergy Statistical Statistical Yearbook Yearbook is is from from China China National National Knowledge Knowledge Infrastructure. Infrastructure. Due Due to theto the lack lack of of some some energy energy consumption consumption statistics statistics from from 2000 2000 to to 2002, 2002, the the selected selected timetime rangerange ofof energy statistics is 2003–2017. Nighttime land surface temperature (NLST) data.data. From From the the data of MOD11A2 provided by National Aeronautics andand Space Administration (NASA),(NASA), thethe timetime resolution isis 8 days and the spatial resolution is 1 km 1 km. First, the Modis Reprojection Tool (MRT) software was used to project and resolution is 1km × 1km. First, the Modis Reprojection Tool (MRT) software was used to project and format the MOD11A2 data [[31].31]. The MRT software was provided by NASA. Second, to eliminate the inﬂuenceinfluence of outliers,outliers, the maximummaximum value composition (MVC) [[32]32] was used to synthesize the 8-day data into annual data. Finally, the data were spliced and trimmed to obtain the annual nighttime land surface temperature datadata ofof thethe YangtzeYangtze RiverRiver DeltaDelta urbanurban agglomerationagglomeration forfor 20002000 toto 2017.2017. Relevant auxiliaryauxiliary data. data. It mainlyIt mainly includes includes China’s China's national national boundaries, boundaries, provincial provincial and municipal and administrativemunicipal administrative divisions indivisions the Yangtze in the River Yangtze Delta, River and theDelta, data and is fromthe data the nationalis from the geographic national informationgeographic information resources directory resources service directory system. service system.

3. Methods 3. Methods

3.1. Spatialization of CO2 Emissions 3.1. Spatialization of CO2 Emissions (1) CO2 emission calculation. According to the statistical data of energy consumption in the Yangtze (1) CO2 emission calculation. According to the statistical data of energy consumption in the River Delta region and the 2006 greenhouse gas emission inventory released by the Intergovernmental Yangtze River Delta region and the 2006 greenhouse gas emission inventory released by the Panel on climate change of the United Nations, the CO2 emission generated by energy consumption is Intergovernmental Panel on climate change of the United Nations, the CO2 emission generated by calculated as follows: energy consumption is calculated as follows: 10 44 X SCE = K E (1) PRO 4412 × 10 j j = × j=1 SCEPRO  K j E j (1) 12 j=1 In this formula, SCEPRO represents the CO2 emission of a province (city), the subscript j represents the energy type, Ej represents the consumption of energy j (unit: 10,000 tons of standard coal), and Kj In this formula, SCE PRO represents the CO2 emission of a province (city), the subscript j represents the CO2 emission coefficient of energy j (unit: 10,000 tons of carbon/10,000 tons of standard represents the energy type, E j represents the consumption of energy j (unit: 10000 tons of standard coal). See reference [5] for the CO2 emission coefficient of each energy. coal),(2) and TheK COj represents2 emission the simulation CO2 emission model coefficient was established. of energy Based j (u onnit: lighting 10000 datatons andof carbon CO2 emission / 10000 data,tons of Meng standard et al. coal). [33] See concluded reference that [5] therefor the is CO a linear2 emission correlation coefficient between of each carbon energy. emissions and (2) The CO2 emission simulation model was established. Based on lighting data and CO2 emission data, Meng et al. [33] concluded that there is a linear correlation between carbon emissions Sustainability 2020, 12, 8388 6 of 17 night lighting data. By analyzing the total amount of lighting and carbon emissions from energy consumption in the Henan Province of China, Gu et al [34] also concluded that there is a linear correlation between the two factors. Therefore, using the correlation between the total CO2 emissions and the total DN value of nighttime light in each municipal administrative region in China, five commonly used correlation models (linear model, exponential model, logarithmic model, power function model, quadratic polynomial model) are selected to explore the correlation between CO2 emissions and the total DN value of nighttime lighting. The results show that the fitting degrees of the linear model, exponential model, logarithmic model, power function model and quadratic polynomial model are 0.775, 0.794, 0.570, 0.615 and 0.778, respectively. Among them, the exponential model has the best fitting degree. However, further research shows that the average relative error of Shanghai from 2000 to 2017 is 23.90%, which is obviously underestimated, which may be due to the saturation of lighting. It is difficult to eliminate the problem completely in data correction, and there are still some errors in the city with high light intensity. In order to further improve the accuracy and reliability of CO2 emission simulation value of each province (city), the relationship models between carbon emission and the DN value sum of Zhejiang, Jiangsu, Anhui and Shanghai are established, respectively. The results show that the fitting effect of the partition model is better than that of the overall model, which can make up for the underestimation of carbon emission in Shanghai. The R2 of Zhejiang, Jiangsu, Anhui and Shanghai were 0.953, 0.988, 0.976 and 0.870, respectively, which passed the significance test of 0.01. At the same time, considering the accuracy of downscaling model inversion, a linear fitting model without intercept is adopted:

E = k DN (2) it i × it

In the formula, the independent variable DNit represents the total DN value of nighttime lighting of i province (city) in t year, the dependent variable Eit represents the statistical value of CO2 emission of i province (city) in t year, and ki is the corresponding coeﬃcient of i province (city) (Table1).

Table 1. Coeﬃcient of linear model of CO2 emission.

Province (City) Zhejiang Jiangsu Anhui Shanghai Coeﬃcient (k) 0.0189 0.0215 0.0162 0.0475 Coeﬃcient of 0.9533 0.9878 0.9775 0.8704 determination (R2)

3.2. Trend Analysis Regression analysis can be used to calculate the temporal variation characteristics of a single pixel, and then reflect the spatial variation law of the whole region [34]. Based on the CO2 emission data set from 2000 to 2017, the linear regression analysis of CO2 emission was carried out with pixel as the basic unit. See reference [35] for the specific calculation formula. In addition, in order to better reflect the spatial variation characteristics of CO2 emissions in the Yangtze River Delta region, slope is divided into five grades by the standard deviation classification method: basically unchanged, weakly significant growth, significant growth, relatively significant growth and extremely significant growth.

3.3. Spatial Automocorrelation Analysis Spatial autocorrelation is used to describe the potential interdependence and aggregation characteristics of variable attribute values in the same region. Spatial autocorrelation analysis includes global spatial autocorrelation and local spatial autocorrelation [36]. This paper selects the global Moran’s I index, which is widely used to reﬂect the concentration degree of CO2 emissions in the Yangtze River Delta region. Local spatial autocorrelation is used to determine the speciﬁc location of spatial agglomeration [37,38]. Sustainability 2020, 12, 8388 7 of 17

4. Results Sustainability 2020, 12, x FOR PEER REVIEW 7 of 18 4.1. Accuracy Evaluation of Simulated CO2 Emissions In order to evaluate the accuracy and reliability of the simulated CO2 emission value, this study In order to evaluate the accuracy and reliability of the simulated CO2 emission value, this study adopted the following two methods for accuracy testing: one is to perform regression analysis on the adopted the following two methods for accuracy testing: one is to perform regression analysis on the statistical data and simulated data of CO2 emissions and make a determination based on the statistical data and simulated data of CO emissions and make a determination based on the correlation 2 2 correlation 2coefficient R (Figure 3); the second is to calculate the relative error between the CO2 coefficient R (Figure3); the second is to calculate the relative error between the CO 2 emission statistics emission statistics and simulation data in different regions in different years (Table 2). It can be seen and simulation data in different regions in different years (Table2). It can be seen from Figure3 that from Figure 3 that the CO2 emission statistics data and the simulated data show an obvious linear the CO2 emission statistics data and the simulated data show an obvious linear correlation, and the correlation, and the correlation is high (R2=0.990), which passed the significance test of 0.01. It can be correlation is high (R2 = 0.990), which passed the significance test of 0.01. It can be seen from Table2 seen from Table 2 that the relative errors between the statistical values and the simulated values of that the relative errors between the statistical values and the simulated values of CO2 emissions in CO2 emissions in the Yangtze River Delta from 2003 to 2017 are all within 10%. The annual average the Yangtze River Delta from 2003 to 2017 are all within 10%. The annual average relative errors of relative errors of Zhejiang, Jiangsu, Anhui and Shanghai are -0.21%, 0.25%, 1.81% and -5.17%. It can Zhejiang, Jiangsu, Anhui and Shanghai are 0.21%, 0.25%, 1.81% and 5.17%. It can be seen that the be seen that the accuracy of the CO2 emission− simulation based on nighttime− light data is better, and accuracy of the CO2 emission simulation based on nighttime light data is better, and this accuracy can this accuracy can meet the requirements of CO2 emission change research. meet the requirements of CO2 emission change research.

FigureFigure 3.3. ScatterScatter diagramdiagram betweenbetween COCO22 emissionsemissions calculatedcalculated fromfrom statisticalstatistical datadataand and CO CO22 emissionsemissions simulatedsimulated fromfrom nighttimenighttime lightlight data.data. 4.2. Temporal Characteristics 4.2. Temporal Characteristics From 2000 to 2017, the Yangtze River Delta region’s CO2 emissions experienced a significant From 2000 to 2017, the Yangtze River Delta region's CO2 emissions experienced a significant increase (Figure4), with an average annual growth rate of 6.26%. From 2000 to 2005, from 2005 to increase (Figure 4), with an average annual growth rate of 6.26%. From 2000 to 2005, from 2005 to 2010 and from 2010 to 2017, the average annual growth rates of CO2 emissions in the Yangtze River 2010 and from 2010 to 2017, the average annual growth rates of CO2 emissions in the Yangtze River Delta were 9.42%, 6.79%, and 4.59%, indicating that the CO2 emissions in the Yangtze River Delta are Delta were 9.42%, 6.79%, and 4.59%, indicating that the CO2 emissions in the Yangtze River Delta are increasing, but the growing speed is gradually slowing down, which is mainly due to the impact of increasing, but the growing speed is gradually slowing down, which is mainly due to the impact of economic and social development goals and the continuous improvement of the energy consumption economic and social development goals and the continuous improvement of the energy consumption system. For example, during the “Eleventh Five-Year Plan” period, the Chinese government made system. For example, during the "Eleventh Five-Year Plan" period, the Chinese government made optimizing the industrial structure and improving the efficiency of resource utilization its main goals optimizing the industrial structure and improving the efficiency of resource utilization its main goals for economic and social development. During the “Twelfth Five-Year Plan” period, the Chinese for economic and social development. During the “Twelfth Five-Year Plan” period, the Chinese government set the goals of reducing energy consumption per unit of gross domestic product (GDP) government set the goals of reducing energy consumption per unit of gross domestic product (GDP) by 16% and CO2 emissions per unit of GDP by 17% [14]. The implementation of these policies has by 16% and CO2 emissions per unit of GDP by 17% [14]. The implementation of these policies has played a positive role in mitigating the growth of CO2 emissions. From the perspective of the spatial played a positive role in mitigating the growth of CO2 emissions. From the perspective of the spatial agglomeration degree, the global Moran’s I in the Yangtze River Delta from 2000 to 2017 was between agglomeration degree, the global Moran's I in the Yangtze River Delta from 2000 to 2017 was between 0.663 and 0.731, both reaching a significant level of 0.01 (Figure4). It shows that CO 2 emissions in 0.663 and 0.731, both reaching a significant level of 0.01 (Figure 4). It shows that CO2 emissions in the the Yangtze River Delta are positively correlated with each other, and CO2 emissions have significant Yangtze River Delta are positively correlated with each other, and CO2 emissions have significant spatial agglomeration characteristics. From 2000 to 2010, Moran's I showed a clear upward trend, with a growth rate of 0.6%•a-1, indicating that the spatial accumulation of CO2 emissions continued to increase, the accumulation of CO2 emissions increased, and the spatial heterogeneity among regions is weakened. From 2011 to 2017, the Moran's I index of CO2 emissions stabilized at 0.7301– Sustainability 2020, 12, 8388 8 of 17 Sustainability 2020, 12, x FOR PEER REVIEW 8 of 18

0.7309,spatial agglomerationand the spatial characteristics.distribution pattern From of 2000 em toissions 2010, in Moran’s the Yangtze I showed River a clearDeltaupward was basically trend, with a growth rate of 0.6% a 1, indicating that the spatial accumulation of CO emissions continued to unchanged. The main reason• − was that the CO2 emissions in the Yangtze River2 Delta experienced an obviousincrease, upward the accumulation process from of CO 20002 emissions to 2010, especi increased,ally in and Shanghai, the spatial Suzhou, heterogeneity Wuxi, Changzhou. among regions The is weakened. From 2011 to 2017, the Moran’s I index of CO emissions stabilized at 0.7301–0.7309, spatial distribution characteristics of CO2 emissions tend to2 be from broken distribution to sheet and the spatial distribution pattern of emissions in the Yangtze River Delta was basically unchanged. distribution, and the connectivity between CO2 emission patches tends to be saturated. The main reason was that the CO2 emissions in the Yangtze River Delta experienced an obvious upwardTable process 2. Relative from error 2000 between to 2010, CO especially2 emissions in calculated Shanghai, from Suzhou, statistical Wuxi, data Changzhou. and CO2 emissions The spatial distributionsimulated characteristics from nighttime of light CO 2dataemissions(%). tend to be from broken distribution to sheet distribution, and the connectivity between CO2 emission patches tends to be saturated. Year Zhejiang Jiangsu Anhui Shanghai Yangtze River Delta

Table 2. Relative2003 error8.22 between CO 4.582 emissions 3.43 calculated 13.22 from statistical data 7.36 and CO2 emissions simulated from2004 nighttime6.47 light data 7.75 (%). 10.91 5.18 7.58 2005 -0.12 5.95 10.03 -2.59 3.32 Year Zhejiang Jiangsu Anhui Shanghai Yangtze River Delta 2006 -5.31 1.04 8.16 -5.57 -0.42 20032007 -8.89 8.22 -4.75 4.58 3.79 3.43 -9.81 13.22 -4.92 7.36 20042008 -7.78 6.47 -3.52 7.75 -0.74 10.91 -8.36 5.18 -5.10 7.58 2005 0.12 5.95 10.03 2.59 3.32 − − 20062009 -5.785.31 -3.79 1.04 3.83 8.16 -8.81 5.57 -3.64 0.42 − − − 20072010 -1.118.89 -3.314.75 -1.29 3.79 -10.99 9.81 -4.18 4.92 − − − − 20082011 -1.027.78 -3.863.52 0.32 0.74 -10.20 8.36 -3.69 5.10 − − − − − 2009 5.78 3.79 3.83 8.81 3.64 2012 -1.25− -3.31− -5.20 -10.14− -4.98− 2010 1.11 3.31 1.29 10.99 4.18 2013 -0.65− -0.56− -4.51− -12.16− -4.47− 2011 1.02 3.86 0.32 10.20 3.69 − − − − 20122014 0.961.25 1.633.31 -2.52 5.20 -8.49 10.14 -2.11 4.98 − − − − − 20132015 1.070.65 1.480.56 -2.40 4.51 -5.64 12.16 -1.37 4.47 − − − − − 20142016 3.86 0.96 0.38 1.63 -1.10 2.52 -4.01 8.49 -0.22 2.11 − − − 2015 1.07 1.48 2.40 5.64 1.37 2017 8.18 4.01 4.44− 0.79 − 4.36− 2016 3.86 0.38 1.10 4.01 0.22 − − − 2017 8.18 4.01 4.44 0.79 4.36

Figure 4. Temporal variation of CO2 emissions and global Moran’s I during 2000–2017. Figure 4. Temporal variation of CO2 emissions and global Moran’s I during 2000–2017. 4.3. Spatial Distribution Characteristics 4.3. Spatial Distribution Characteristics Based on the nighttime light data of DMSP/OLS and NPP/VIIRS, the spatial distribution of CO2 emissionsBased onon thethe nighttime pixel scale light in the data Yangtze of DMSP/OLS River Delta and NPP/VIIRS, is simulated. the Figure spatial5 distributionshows the spatial of CO2 emissionsdistribution on of the CO 2pixelemissions scale in thethe YangtzeYangtze River River Delta Delta from is simulated. 2000 to 2017, Figure and it 5 uses shows equal the intervals spatial distributionto divide it into of CO 8 levels,2 emissions 0–0.1, 0.1–0.2,in the Yangtze 0.2–0.3, River 0.3–0.4, Delta 0.4–0.5, from 0.5–0.6, 2000 to 0.6–0.7, 2017, and 0.7–max it uses (unit: equal 10 intervals4 t/km2). Theto divide area occupied it into 8 by levels, 0–0.1 dropped0–0.1, 0.1–0.2, from 73.26%0.2–0.3, in 0.3–0.4, 2000 to 23.94%0.4–0.5, in0.5–0.6, 2017, while 0.6–0.7, the areas0.7–max occupied (unit: 104t/km2). The area occupied by 0–0.1 dropped from 73.26% in 2000 to 23.94% in 2017, while the areas occupied by 0.1–0.2, 0.2–0.3, 0.3–0.4, 0.4–0.5, 0.5–0.6, 0.6–0.7, 0.7–max are on the rise. The proportion Sustainability 2020, 12, 8388 9 of 17

by 0.1–0.2, 0.2–0.3, 0.3–0.4, 0.4–0.5, 0.5–0.6, 0.6–0.7, 0.7–max are on the rise. The proportion of area occupied by 0.7–max rose from 3.08% in 2000 to 15.48% in 2017. From the perspective of spatial distribution, there are obvious regional diﬀerences in CO2 emissions in the Yangtze River Delta. Shanghai, Suzhou and its neighboring cities are the main distribution areas with high CO2 emissions. They are distributed in obvious patches. These regions have a high level of urbanization development, highly dense population, and carried out a large number of economic activities. The low-value areas of CO2 emissions are mainly distributed in Lishui and Quzhou of Zhejiang Province, and Lu’an, Anqing, Chizhou of Anhui Province. These cities have relatively low energy consumption and low Sustainability 2020, 12, x FOR PEER REVIEW 10 of 18 CO2 emissions.

Figure 5. Figure 5. SpatialSpatial distribution distribution of of CO CO2 emissions2 emissions during during 2000–2017. 2000–2017.

Figure6 shows the spatial agglomeration of CO 2 emissions in the Yangtze River Delta from 2000 to 2017. The High-High (HH) aggregation zone indicates that the CO2 emissions of a certain space unit and its surrounding space units are very high, and the Low-Low (LL) aggregation zone indicates that the CO2 emissions of a space unit and its surrounding space units are very low. The High-Low (HL) aggregation zone indicates that the high value is surrounded by the low value, and the Low-High (LH) aggregation zone means that the low value is surrounded by the high value. It can be seen from Figure6 that the spatial agglomeration characteristics of CO 2 emissions in different regions are significantly different. The HH aggregation zone is mainly distributed in Shanghai, southern Jiangsu and northern Zhejiang. The area occupied has increased from 26,949 km2 in 2000 to 50,382 km2 in 2017. The connectivity of these areas is significantly enhanced, and there is a certain link between the CO2 emissions of these regions, and they affect each other. From 2000 to 2010, the degree of spatial agglomeration of CO2 emissions changed significantly. The average annual growth rate of the area occupied by the HH aggregation zone was 5.46%. From 2011 to 2017, the HH aggregation zone showed a clear patchy distribution, and the degree of inter-annual spatial agglomeration was not changed obviously, the average annual growth rate of the occupied area is 0.97%. The HL aggregation zone and the LH aggregation zone occupy a small area and are mainly distributed in areas with large differences in CO2 emissions. The HL aggregation zones are scattered, most of which are located in cities with relatively low economic development in Anhui and Zhejiang, while the LH aggregation zone is attached to the periphery of HH aggregation zone in a band, and its area proportion fluctuates from 2.45% to 3.27%. The LL aggregation zone occupies the largest area. The main distribution areas of the LL aggregation zone have basically remained unchanged in the past 18 years, mainly in Anhui Figure 6. Local spatial autocorrelation analysis map of CO2 emissions from 2000 to 2017.

4.4. Spatial Trend Characteristics The spatial change trend of carbon emissions in the Yangtze River Delta region from 2000 to 2017 is studied based on the unary linear regression model (Figure 7). The research results show that (a) during 18 years from 2000 to 2017, the overall CO2 emissions of that region showed an upward trend, and the basically unchanged area accounted for the largest proportion of area, accounting for 70.14% (Table 3), and it is mainly distributed in Anhui Province and the western and southern regions of Zhejiang Province, and these regions have relatively low CO2 emissions, and the inter‐annual changes are not obvious. (b) The areas whose CO2 emissions were of relatively significant growth accounted for 4.84%, which mainly include Shanghai, Suzhou, southern Nantong, northern Wuxi, northern Zhenjiang, southern Yangzhou and central Nanjing. The areas whose CO2 emissions were

Sustainability 2020, 12, x FOR PEER REVIEW 10 of 18

Sustainability 2020, 12, 8388 10 of 17

Province and the western and southern parts of Zhejiang Province. These regions have relatively low Figure 5. Spatial distribution of CO2 emissions during 2000–2017. CO2 emissions and the mutual inﬂuences between regions are smaller [39,40].

FigureFigure 6. 6. LocalLocal spatial spatial autocorrelation autocorrelation analysis analysis map map of of CO CO2 2emissionsemissions from from 2000 2000 to to 2017. 2017.

4.4.4.4. Spatial Spatial Trend Trend Characteristics Characteristics TheThe spatial spatial change trend ofof carbon carbon emissions emissions in in the the Yangtze Yangtze River River Delta Delta region region from from 2000 2000 to 2017 to 2017is studied is studied based based on the on unarythe unary linear linear regression regression model model (Figure (Figure7). The7). The research research results results show show that that (a) during 18 years from 2000 to 2017, the overall CO emissions of that region showed an upward trend, (a) during 18 years from 2000 to 2017, the overall2 CO2 emissions of that region showed an upward trend,and the and basically the basically unchanged unchanged area accounted area accounted for the for largest the largest proportion proportion of area, of accounting area, accounting for 70.14% for 70.14%(Table3 (Table), and 3), it isand mainly it is mainly distributed distributed in Anhui in Anhui Province Province and and the westernthe western and and southern southern regions regions of Zhejiang Province, and these regions have relatively low CO emissions, and the inter-annual changes of Zhejiang Province, and these regions have relatively low2 CO2 emissions, and the inter‐annual are not obvious. (b) The areas whose CO emissions were of relatively significant growth accounted changes are not obvious. (b) The areas whose2 CO2 emissions were of relatively significant growth accountedfor 4.84%, for which 4.84%, mainly which include mainly Shanghai, include Shanghai, Suzhou, southernSuzhou, southern Nantong, Nantong, northern northern Wuxi, northern Wuxi, northernZhenjiang, Zhenjiang, southern southern Yangzhou Yangzhou and central and Nanjing. central The Nanjing. areas whose The areas CO2 whoseemissions CO2 were emissions of extremely were significant growth accounted for the smallest part, namely, 0.58% or 1989km2, which mainly include Shanghai. (c) The study found that in some large, medium and small urban centers (such as Shanghai, Hefei, Wuxi, etc.), the type of CO2 emission growth is weak and significant. The possible reason is that the population density tends to be saturated in the urban center, which leads to the decrease of the CO2 emission growth rate. The regions with significant growth, relatively significant growth and extremely significant growth are distributed in a circle. This is closely related to the spatial expansion of the city. With the continuous outward expansion of the city, the CO2 emission in the peripheral area of the city center increases rapidly. Sustainability 2020, 12, x FOR PEER REVIEW 11 of 18 of extremely significant growth accounted for the smallest part, namely, 0.58% or 1989km2, which mainly include Shanghai. (c) The study found that in some large, medium and small urban centers (such as Shanghai, Hefei, Wuxi, etc.), the type of CO2 emission growth is weak and significant. The possible reason is that the population density tends to be saturated in the urban center, which leads to the decrease of the CO2 emission growth rate. The regions with significant growth, relatively significant growth and extremely significant growth are distributed in a circle. This is closely related toSustainability the spatial2020 expansion, 12, 8388 of the city. With the continuous outward expansion of the city, the11 CO of 172 emission in the peripheral area of the city center increases rapidly.

Figure 7. Spatial trend characteristics of CO2 emissions during 2000–2017. Figure 7. Spatial trend characteristics of CO2 emissions during 2000–2017. Table 3. Area and proportion of different growth types. Table 3. Area and proportion of different growth types. Change Types Area (km2) Area Ratio (%) Cumulative Percentage (%) Change Types Area (km2) Area Ratio (%) Cumulative Percentage (%) Basically unchanged 240,519 70.14 70.14 weaklyBasically significant unchanged growth 240519 53,700 70.14 15.66 70.14 85.80 weaklySignificant significant growth growth 53700 30,108 15.66 8.78 85.80 94.58 RelativelySignificant significant growth growth 30108 16,597 8.78 4.84 94.58 99.43 RelativelyExtremely significant significant growth 16597 1989 4.84 0.58 99.43 100 Extremely significant growth 1989 0.58 100 5. Relationship between CO2 Emissions and Nighttime Land Surface Temperature 5. Relationship between CO2 Emissions and Nighttime Land Surface Temperature 5.1. Comparative Analysis of Temporal and Spatial Changes 5.1. ComparativeThere is a statisticAnalysis on of theTemporal annual and mean Spatial nighttime Changes land surface temperature in 2000–2017 to make a figureThere of is nighttime a statistic on land the surface annual temperature mean nighttime an anomalyland surface in the temperature Yangtze River in 2000–2017 Delta (Figure to make8). aThe figure result of nighttime shows that land the nighttime surface temperature land surface an temperature anomaly in increase the Yangtze is 0.57 River◦C/10a Delta in the (Figure area, which8). The is resultmore shows than the that rising the ratenighttime of average land surface surface temperature temperature in increase China in is the 0.57 past °C/10a 60 years in the (0.23 area,◦C /which10a) [40 is], moreso it meansthan the that rising the nighttime rate of average land surface surface temperature temperature in in Yangtze China River in the Delta past is60 warming. years (0.23 Compared °C/10a) [40],with so the it emissionmeans that change the nighttime of CO2, bothland surface of them temperature are increasing; in Yangtze but they River are diDeltafferent. is warming. The CO2 Comparedemission is with rising the up emission with straight change line of CO while2, both the of nighttime them are land increasing; surface but temperature they are different. is in upward The COtrend2 emission of fluctuation. is rising Thereup with is correlationstraight line analysis while the on nighttime the carbon land emission surface and temperature nighttime is land in upward surface trendtemperature of fluctuation. in Yangtze There River is correlation Delta in 2000–2017, analysis andon the the carbon result showsemission that and there nighttime is positive land correlation surface temperaturebetween the twoin Yangtze parties withRiver the Delta correlation in 2000–2017, coefficient and of 0.413the result and p valueshows of that 0.088. there is positive correlationTo compare between and the analyze two parties the spatial with the distribution correlation characteristics coefficient of of 0.413 CO2 andemissions p value and of 0.088. nighttime land surface temperature, the average value of the CO2 emission and nighttime land surface temperature in 2000–2017 with a grid computing tool ArcGIS 10.2 is used (Figure9). Through the comparison and analysis of the spatial distribution law of CO2 emissions and nighttime land surface temperature, it could be seen that there is high consistency between CO2 emissions and nighttime land surface temperature in the spatial distribution. Especially, Shanghai, Suzhou, Wuxi, Nanjing and Zhenjiang are the most significant since they are the main distribution area with high CO2 emission, and are flaky in distribution. Accordingly, the nighttime land surface temperature is also in flaky distribution. In Fuyang, Bozhou, Huaibei, Suzhou of Anhui Province, the CO2 emission is scattered, and the nighttime land surface temperature is also in the same feature. What is more, most of the areas with low values of CO2 emission and nighttime land surface temperature are Chizhou, Xuancheng, Huangshan of Anhui Province and Lishui and Wenzhou of Zhejiang Province. SustainabilitySustainability2020 2020, ,12 12,, 8388 x FOR PEER REVIEW 1212 of of 17 18

Sustainability 2020, 12, x FOR PEER REVIEW 13 of 18 Figure 8. Temporal variation of nighttime surface temperature during 2000–2017. Figure 8. Temporal variation of nighttime surface temperature during 2000–2017.

To compare and analyze the spatial distribution characteristics of CO2 emissions and nighttime land surface temperature, the average value of the CO2 emission and nighttime land surface temperature in 2000–2017 with a grid computing tool ArcGIS 10.2 is used (Figure 9). Through the comparison and analysis of the spatial distribution law of CO2 emissions and nighttime land surface temperature, it could be seen that there is high consistency between CO2 emissions and nighttime land surface temperature in the spatial distribution. Especially, Shanghai, Suzhou, Wuxi, Nanjing and Zhenjiang are the most significant since they are the main distribution area with high CO2 emission, and are flaky in distribution. Accordingly, the nighttime land surface temperature is also in flaky distribution. In Fuyang, Bozhou, Huaibei, Suzhou of Anhui Province, the CO2 emission is scattered, and the nighttime land surface temperature is also in the same feature. What is more, most of the areas with low values of CO2 emission and nighttime land surface temperature are Chizhou, Xuancheng, Huangshan of Anhui Province and Lishui and Wenzhou of Zhejiang Province.

FigureFigure 9. 10aSpatial is divided distribution into patternfour grades of CO 2(Tableemissions 4) with (a) and standard nighttime deviation land surface classification: temperature area of low (emissionb) in the Yangtze (0–0.05), River area Delta of medium from 2000 emission to 2017. (0.05–0.41), area of higher emission (0.41–0.77) and Figure 9. Spatial distribution pattern of CO2 emissions (a) and nighttime land surface temperature (b) area of high emission (0.77–2.91). The area proportions are 20.97%, 63.68%, 7.66% and 7.69%, in the Yangtze River Delta from 2000 to 2017. respectively.Figure 10 aHere is divided is the intoresult four of gradesthe study. (Table The4) withnighttime standard land deviation surface temperature classiﬁcation: is area the oflowest, low emission (0–0.05), area of medium emission2 (0.05–0.41), area of higher emission (0.41–0.77) and area of 25.84 Table°C, in 4.the Different area with CO 2low Emission CO emission; Grades and the Theirtemperature Correspondent soars toNighttime 27.32 °C Land with Surfacea year ‐on‐year high emission (0.77–2.91). The area proportions are 20.97%, 63.68%, 7.66% and 7.69%, respectively. increaseTemperature of 5.73% in in Yangtze the area River with Delta. medium CO2 emission; the temperature is 28.09 °C in the area with Here is the result of the study. The nighttime land surface temperature is the lowest, 25.84 C, in higher CO2 emission, so there is a year‐on‐year increase of 8.71% compared with the area with◦ low the area with low CO emission; the temperatureArea soars with to 27.32 AreaC with with a year-on-year increase of CO2 emission. The temperature2 Area is 28.80with °C in the area with high ◦CO2 emission, so thereArea withis a year ‐on‐ 5.73% in the area with medium CO emission; theMedium temperature is 28.09HigherC in the area with higher CO yearCO increase2 Emission of 11.5%Grade comparedLow Emission2 with the area with low CO2 emission.◦ It couldHigh be Emission seen that the2 emission, so there is a year-on-year increase of 8.71%Emission compared withEmission the area with low CO emission. nighttime land surface temperature would rise up along with the increase of the CO2 emission,2 so The temperatureNighttime Land is 28.80 C in the area with high CO emission, so there is a year-on-year increase of there is a certain correlation◦ relationship between2 CO2 emission and nighttime land surface Surface Temperature 11.5% compared with the area with25.84 low CO2 emission.27.32 It could be seen28.09 that the nighttime28.80 land surface temperature.(°C) temperature would rise up along with the increase of the CO2 emission, so there is a certain correlation Area (km2) 71,909 218,367 26,267 26,370 relationship between CO emission and nighttime land surface temperature. Percentage (%) 2 20.97 63.68 7.66 7.69

Table 4. Diﬀerent CO2 Emission Grades and Their Correspondent Nighttime Land Surface Temperature 5.2. inScale Yangtze Analysis River within Delta. the Year

To show the correlation degree between CO2 emission and nighttime land surface temperature Area with Low Area with Area with Higher Area with High CO Emission Grade based 2on the spatial distributionEmission map of COMedium2 emission Emission and nighttimeEmission land surface temperatureEmission in 2000–2017, a grid is built with the data management tool of ArcGIS 10.2, statistics on total CO2 Nighttime Land Surface emissions per grid (SCE) are made25.84 and the maximum 27.32 (NLSTmax), minimum 28.09 (NLSTmin) and 28.80 average Temperature (◦C) 2 (NLSTmeanArea) nighttime (km ) land surface71,909 temperature are 218,367acquired, and finally, 26,267 the correlation coefficient 26,370 of the COPercentage2 emission (%) and the maximum,20.97 minimum and 63.68 average of and 7.66 the nighttime land 7.69 surface temperature are calculated through Pearson correlation analysis method (Table 5). The study result shows that there is a significant positive correlation relationship between the CO2 emission and the maximum, minimum and average of and the nighttime land surface temperature, and all of them have passed the significance test of 0.05. However, the correlation degrees of CO2 emission and nighttime land surface temperature in different years are various. In 2015, there is the most significant correlation between the CO2 emission and the maximum, minimum and average of the nighttime land surface temperature with the correlation coefficients of 0.444, 0.506 and 0.590. In other years, the correlation in 2019 is lower and the correlation coefficients are 0.085, 0.244 and 0.282.

Sustainability 2020, 12, 8388 13 of 17 Sustainability 2020, 12, x FOR PEER REVIEW 15 of 18

FigureFigure 10. Correlation 10. Correlation coefficient coeffi (cienta) and (a significance) and significance level (b level) of CO (b)2 ofemissions CO2 emissions and nighttime and nighttime land surface temperatureland surface from temperature2000 to 2017. from Note: 2000 ESPC: to 2017. extremely Note: significant ESPC: extremely positive significant correlation; positive VSPC: correlation; very significant positiveVSPC: correlation; very significant RSPC: positive relatively correlation; significant RSPC: positive relatively correlation; significant WSPC: positive weak correlation; significant WSPC: positive correlation;weak significant WSNC: weak positive significant correlation; negative WSNC: correlation; weak significant RSNC: negativerelatively correlation; significant RSNC: negative relatively correlation; VSNC:significant very significant negative negative correlation; correlation; VSNC: very ESNC: significant extremely negative significant correlation; negative ESNC: correlation. extremely significant negative correlation.

Table 6. Correlation between CO2 emissions and nighttime land surface temperature. 5.2. Scale Analysis within the Year Correlation p Value Area (km2) Percentage (%) Cumulative Percentage (%) To show the correlation degree between CO emission and nighttime land surface temperature ESPC p < 0.01 11,248 2 3.28 3.28 based on the spatial distribution map of CO emission and nighttime land surface temperature in VSPC p < 0.05 30,176 2 8.80 12.08 2000–2017, a grid is built with the data management tool of ArcGIS 10.2, statistics on total CO RSPC p < 0.10 31,034 9.05 21.13 2 emissions per grid (SCE) are made and the maximum (NLSTmax), minimum (NLST ) and average WSPC p > 0.10 232,693 67.85 88.98 min (NLSTmean) nighttime land surface temperature are acquired, and finally, the correlation coefficient WSNC p > 0.10 37,542 10.95 99.93 of the CO emission and the maximum, minimum and average of and the nighttime land surface 2RSNC p < 0.10 137 0.04 99.97 temperature are calculated through Pearson correlation analysis method (Table5). The study result VSNC p < 0.05 69 0.02 99.99 shows that there is a significant positive correlation relationship between the CO emission and the ESNC p < 0.01 14 0.00 100 2 maximum, minimum and average of and the nighttime land surface temperature, and all of them have passed the significance test of 0.05. However, the correlation degrees of CO emission and nighttime 6. Conclusions 2 land surface temperature in different years are various. In 2015, there is the most significant correlation betweenThis the paper CO 2calculatesemission andthe CO the2 maximum,emissions in minimum the Yangtze and averageRiver Delta of the region nighttime based land on surfaceenergy temperatureconsumption with statistics, the correlation and then coecombinesfficients night of 0.444, lighting 0.506 data and to 0.590. realize In the other spatial years, informatization the correlation inof 2019CO2 emissions is lower and in thea long correlation time series, coeffi whichcients can are solve 0.085, the 0.244 defect and 0.282.of insufficient energy consumption statistics at the municipal level. In terms of data acquisition, energy consumption statistics come from the NationalTable 5. CorrelationBureau of coeStatistics,fficients DMSP/OLS of CO2 emissions data and come the from maximum, the National minimum Oceanic and average and nighttimeAtmospheric Administrationland surface of temperature. the United States, and NPP/VIIRS data come from the National Geophysical Data Center of the UnitedSCE and States.SCE All andthe aboveSCE data and can be obtainedSCE andfree of charge.SCE and In theSCE aspect and of data Year Year processing, theNLST datamax processingNLST minis carriedNLST outmean according to NLSTthe existingmax NLSTcorrectionmin methodNLSTmean of night light data,2000 which 0.376 is simple ** and 0.284 feasible. ** In 0.404 terms ** of 2009CO2 emission 0.085 *simulation, 0.244 it ** is feasible 0.282 to ** use the linear simulation2001 0.327 model ** without 0.284 **intercept 0.389 to retrieve ** 2010 the spatial 0.275 information ** 0.312 of ** CO2 emissions 0.410 ** in the Yangtze2002 River Delta 0.148 **region in 0.200 this ** study, 0.236 and **the simulation 2011 0.438 accuracy ** is relatively 0.416 ** high 0.526 (the ** relative error is2003 within 10%), 0.232 ** which 0.305is a reliable ** new 0.416 type ** of 2012 CO2 emission 0.266 ** spatialization 0.266 ** method. 0.334 However, ** 2004 0.160 ** 0.262 ** 0.339 ** 2013 0.387 ** 0.431 ** 0.574 ** in the process of realizing the spatialization of CO2 emissions in the Yangtze River Delta region, there 2005 0.436 ** 0.255 ** 0.374 ** 2014 0.362 ** 0.277 ** 0.395 ** are still2006 some problems: 0.303 ** the relative 0.252 ** error 0.335 of CO **2 emissions 2015 in 0.444Shanghai ** from 0.506 2005 ** to 2016 0.590 is **negative, and CO20072 emissions 0.369 **are obviously 0.341 ** underestimated, 0.410 ** 2016 which 0.268is due ** to the 0.337 difficulty ** 0.394of completely ** eliminating2008 the 0.407saturation ** problem 0.454 ** of DMSP/OLS 0.577 ** data 2017 during 0.372 correction. ** 0.407 Secondly, ** in 0.480 the process ** of processing NPP/VIIRS data, due toNote: some * means missingp < 0.05; data ** meansfrom pMay< 0.01. to July, annual NPP/VIIRS data can only be synthesized based on monthly NPP/VIIRS data from January to April and August to December. To sum up, although there is a certain deviation in the simulation results of CO2 emissions,

Sustainability 2020, 12, 8388 14 of 17

5.3. Interannual Scale Analysis

To disclose the correlation relationship between CO2 emission and nighttime land surface temperature in a partial area, it is selected with the CO2 emission and nighttime land surface temperature in 2000–2017 in the Yangtze River Delta as analysis indexes for the analysis on the correlation relationship (n = 18). Figure 10a is the spatial distribution of the correlation coefficient between CO2 emission and nighttime land surface temperature at the pixel scale, and the correlation coefficient is between 0.67 and 0.99. There is a significance test for the coefficient and then it would − be divided into an extremely significant positive correlation (coefficient ranging from 0.5897 to 0.9946, p < 0.01), very significant positive correlation (coefficient ranging from 0.4683 to 0.589, p < 0.05), relatively significant positive correlation (coefficient ranging from 0.4000 to 0.44683, p < 0.1), weak significant positive correlation (coefficient ranging from 0 to 0.40000, p > 0.1), weak significant negative correlation (coefficient ranging from 0.40000 to 0, p > 0.1), relatively significant negative correlation − (coefficient ranging from 0.4683 to 0.4000, p < 0.1), very significant negative correlation (coefficient − − ranging from 0.5897 to 0.4683, p < 0.05) and extremely significant negative correlation (coefficient − − ranging from 0.6727 to 0.5897, p < 0.01). All of these are represented with ESPC, VSPC, RSPC, − − WSPC, WSNC, RSNC, VSNC and ESNC and shown in Figure 10b. The result shows that there is regional difference of CO2 emission on nighttime land surface temperature. The correlation relationship between CO2 emission and nighttime land surface temperature is mainly the positive one, and the area of positive correlation is 88.98%. The area of very significant positive correlation and extremely significant positive correlation is 42,030 km2, accounting for 12.08%, and they are mainly in Northern Lianyungang, central Yangzhou, Western Jiaxing, Eastern Huzhou, eastern Hangzhou, northern Shaoxing, central Ningbo, Eastern Taizhou and Eastern Wenzhou. Meanwhile, there is less of an area of negative correlation and they are mainly in Northwest Anhui Province, Jinhua, Quzhou and Lishui in Zhejiang Province. The area of extremely significant negative correlation and the very significant negative correlation only takes up 0.02%. The increase in CO2 emission should not be ignored as a factor in the increase in the nighttime land surface temperature in the Yangtze River Delta. However, CO2 emissions are not the only factor responsible for this increase. There are complex factors affecting the temperature, including climate change, the distribution of the population and vegetation coverage which all have an impact. The present study only explored the relationship between CO2 emission and nighttime land surface temperature (Table6). Consequently, a further study involving a comprehensive analysis of the impacts of all factors on the nighttime land surface temperature is required.

Table 6. Correlation between CO2 emissions and nighttime land surface temperature.

Correlation p Value Area (km2) Percentage (%) Cumulative Percentage (%) ESPC p < 0.01 11,248 3.28 3.28 VSPC p < 0.05 30,176 8.80 12.08 RSPC p < 0.10 31,034 9.05 21.13 WSPC p > 0.10 232,693 67.85 88.98 WSNC p > 0.10 37,542 10.95 99.93 RSNC p < 0.10 137 0.04 99.97 VSNC p < 0.05 69 0.02 99.99 ESNC p < 0.01 14 0.00 100

6. Conclusions

This paper calculates the CO2 emissions in the Yangtze River Delta region based on energy consumption statistics, and then combines night lighting data to realize the spatial informatization of CO2 emissions in a long time series, which can solve the defect of insuﬃcient energy consumption statistics at the municipal level. In terms of data acquisition, energy consumption statistics come from the National Bureau of Statistics, DMSP/OLS data come from the National Oceanic and Atmospheric Administration of the United States, and NPP/VIIRS data come from the National Geophysical Data Sustainability 2020, 12, 8388 15 of 17

Center of the United States. All the above data can be obtained free of charge. In the aspect of data processing, the data processing is carried out according to the existing correction method of night light data, which is simple and feasible. In terms of CO2 emission simulation, it is feasible to use the linear simulation model without intercept to retrieve the spatial information of CO2 emissions in the Yangtze River Delta region in this study, and the simulation accuracy is relatively high (the relative error is within 10%), which is a reliable new type of CO2 emission spatialization method. However, in the process of realizing the spatialization of CO2 emissions in the Yangtze River Delta region, there are still some problems: the relative error of CO2 emissions in Shanghai from 2005 to 2016 is negative, and CO2 emissions are obviously underestimated, which is due to the difficulty of completely eliminating the saturation problem of DMSP/OLS data during correction. Secondly, in the process of processing NPP/VIIRS data, due to some missing data from May to July, annual NPP/VIIRS data can only be synthesized based on monthly NPP/VIIRS data from January to April and August to December. To sum up, although there is a certain deviation in the simulation results of CO2 emissions, the simulation accuracy of CO2 emissions in the Yangtze River Delta region from 2003 to 2017 is controlled within 10%, which can meet the requirements of CO2 emission change research. Under the background of rapid urbanization, CO2 emissions in the Yangtze River Delta region showed an obvious upward trend from 2000 to 2017, but the growth rate of CO2 emissions in the Yangtze River Delta region is gradually slowing down. In terms of spatial distribution, affected by the level of urbanization development, the CO2 emissions in the Yangtze River Delta region are obviously different from those in the east and west. The high-value areas of CO2 emissions are mainly distributed in the eastern coastal regions, with Shanghai being the most obvious. From the perspective of spatial agglomeration degree, CO2 emissions in the Yangtze River Delta region showed significant positive spatial autocorrelation from 2000 to 2017, and the main distribution areas of HH and LL agglomeration areas of CO2 emissions remained basically unchanged. The spatial distribution pattern of CO2 emissions in the Yangtze River Delta region showed a certain path-dependent effect. In the past 18 years, the nighttime land surface temperature in the Yangtze River Delta region has shown a warming trend, and the spatial distribution of nighttime land surface temperature and CO2 emissions has a high consistency. There is a significant positive correlation between CO2 emissions and the maximum, minimum and average nighttime land surface temperature in each year, reaching a significant level of 0.05. On the interannual scale, the correlation degree between CO2 emissions and nighttime land surface temperature is different in different regions, but the relationship between CO2 emissions and nighttime land surface temperature is mainly positive, with the area with positive correlation reaching more than 85%, so there is certain correlation relationship between CO2 emissions and nighttime land surface temperature.

Author Contributions: K.Y. and S.Z. designed and performed this research; J.Z. performed the experiments and drafted the manuscript; Y.Z. and Y.M. edited the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: Funding: This research was funded by the National Natural Science Foundation of China (41461038), Yunnan Provincial Science and Technology Project (2011XX2005) and Specialized Research Fund for the Doctoral Program of Higher Education (20115303110002). Acknowledgments: The authors would like to thank the anonymous reviewers and editor for constructive comments and suggestions. Conflicts of Interest: The authors declare no conflict of interest. Data Availability: The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

1. Li, Y.; Wei, Y.H. The spatial-temporal hierarchy of regional inequality of China. Appl. Geogr. 2010, 30, 303–316. [CrossRef] Sustainability 2020, 12, 8388 16 of 17

2. You, J. China’s energy consumption and sustainable development: Comparative evidence from GDP and genuine savings. Renew. Sustain. Energy Rev. 2011, 15, 2984–2989. [CrossRef]

3. Zhang, L.; Huang, Y.; Li, Y.; Cheng, X. An Investigation on Spatial Changing Pattern of CO2 Emission in China. Resour. Sci. 2010, 32, 211–217. 4. Cheng, L.; Jian, W.; Jia, Y.; Ming, L. Research on Regional Carbon Emissions in China. Acta Geosci. Sin. 2010, 31, 727–732. 5. Su, Y.; Chen, X.; Li, Y. China’s 19-year city-level carbon emissions of energy consumptions, driving forces and regionalized mitigation guidelines. Renew. Sustain. Energy Rev. 2014, 35, 231–243. [CrossRef] 6. Fan, M.; Yue, T.; Chen, C.; Sun, X. Spatial Change Trends of Temperature and Precipitation in China. J. Geo Inf. Sci. 2011, 13, 526–533. [CrossRef] 7. Hafner, J. Urban heat island modeling in conjunction with satellite-derived surface/soil parameters. J. Appl. Meteorol. 1999, 38, 448–465. [CrossRef] 8. Wang, Z.; Yang, S.; Wang, S. Monitoring evolving urban cluster systems using DMSP/OLS nighttime light data: A case study of the Yangtze River Delta region, China. J. Appl. Remote Sens. 2017, 11, 046029. [CrossRef] 9. Chen, L.; Xun, L.; Yao, Y.; Cheng, D. Eﬀects of population agglomeration on urban economic growth in China. Acta Geogr. Sin. 2018, 73, 1107–1120. 10. Liu, X.; Ning, X.; Wang, H.; Wang, C.; Zhang, H.; Meng, J. A Rapid and Automated Urban Boundary Extraction Method Based on Nighttime Light Data in China. Remote Sens. 2019, 11, 1126. [CrossRef] 11. Shi, K.; Yang, Q.; Fang, G.; Yu, B.; Chen, Z.; Yang, C.; Wu, J. Evaluating spatiotemporal patterns of urban electricity consumption within diﬀerent spatial boundaries: A case study of Chongqing, China. Energy 2019, 167, 641–653. [CrossRef] 12. Su, Y.; Chen, X.; Ye, Y.; Wu, Q.; Kuang, Y. The characteristics and mechanisms of carbon emissions from energy consumption in China using DMSP/OLS night light imageries. Acta Geogr. Sin. 2013, 68, 1513–1526.

13. Wang, S.; Liu, X. China’s city-level energy-related CO2 emissions: Spatiotemporal patterns and driving forces. Appl. Energy 2017, 200, 204–214. [CrossRef]

14. Shi, K.; Yu, B.; Zhou, Y.; Chen, Y.; Yang, C.; Chen, Z.; Wu, J. Spatiotemporal variations of CO2 emissions and their impact factors in China: A comparative analysis between the provincial and prefectural levels. Appl. Energy 2019, 233, 170–181. [CrossRef] 15. Cheng, Y.; Wang, Z.; Ye, X.; Wei, Y.H. Spatiotemporal dynamics of carbon intensity from energy consumption in China. J. Geogr. Sci. 2014, 24, 631–650. [CrossRef] 16. Jarvis, A.; Leedal, D.; Taylor, C.; Young, P. Stabilizing global mean surface temperature: A feedback control perspective. Environ. Model. Softw. 2009, 24, 665–674. [CrossRef] 17. Ayeni, B.J.; Pilat, R. Crude Oil Reserve Estimation: An Application of the Autoregressive Integrated Moving Average Model. J. Pet. Sci. Eng. 1992, 8, 13–28. [CrossRef] 18. Jones, G.; Christidis, N.; Stott, P. Detecting the inﬂuence of fossil fuel and bio-fuel black carbon aerosols on near surface temperature changes. Atmos. Chem. Phys. 2010, 10, 799–816. [CrossRef] 19. Bai, W.; Yang, G.; Zhang, X. Research on multiple statistical causal about carbon emission increases with temperature rise. J. Guizhou Coll. Financ. Econ. 2013, 5, 46–51. 20. Das, D.; Srinivasan, R.; Sharfuddin, A. Fossil fuel consumaption, carbon emissions and temperature variation in India. Energy Environ. 2011, 22, 695–709. [CrossRef] 21. Lu, H.; Zhang, M.; Sun, W.; Li, W. Expansion Analysis of Yangtze River Delta Urban Agglomeration Using DMSP/OLS Nighttime Light Imagery for 1993 to 2012. ISPRS Int. J. Geo Inf. 2018, 7, 52. [CrossRef] 22. Du, Y.; Xie, Z.; Zeng, Y.; Shi, Y.; Wu, J. Impact of urban expansion on regional temperature change in the Yangtze River Delta. J. Geogr. Sci. 2007, 4, 387–398. [CrossRef] 23. Gu, C.; Hu, L.; Zhang, X.; Wang, X. Climate change and urbanization in the Yangtze River Delta. Habitat Int. 2011, 35, 544–552. [CrossRef] 24. Wang, T.; Zhang, G.; Li, P.; Li, F. Analysis on the driving factors of urban expansion policy based on DMSP/OLS remote Sensing image. Acta Geod. Cartogr. Sin. 2018, 47, 42–49. 25. Elvidge, C.D.; Ziskin, D.; Baugh, K.E.; Tuttle, B.; Ghosh, T.; Pack, D.; Erwin, E.; Zhizhin, M. A Fifteen Year Record of Global Natural Gas Flaring Derived from Satellite Data. Energies 2009, 2, 595–622. [CrossRef] 26. Wu, J.; He, S.; Peng, J.; Li, W.; Zhong, X. Intercalibration of DMSP-OLS night-time light data by the invariant region method. Int. J. Remote Sens. 2013, 34, 7356–7368. [CrossRef] Sustainability 2020, 12, 8388 17 of 17

27. Elvidge, C.D.; Baugh, K.; Zhizhin, M.; Hsu, F. Why VIIRS data are superior to DMSP for mapping nighttime lights. Proc. Asia Pac. Adv. Netw. 2013, 35, 62–69. [CrossRef] 28. Schroeder, W.; Oliva, P.; Giglio, L.; Csiszar, I.A. The new VIIRS 375 m active ﬁre detection data product: Algorithm description and initial assessment. Remote Sens. Environ. 2014, 143, 85–96. [CrossRef] 29. Shi, F.; Yu, B.; Huang, Y.; Hu, Y.; Yin, B.; Chen, Z.; Chen, L.; Wu, J. Evaluating the ability of NPP-VIIRS nighttime light data to estimate the gross domestic product and the electric power consumption of China at multiple scales: A comparison with DMSP-OLS data. Remote Sens. 2014, 6, 1705–1724. [CrossRef] 30. Zhao, J.; Ji, G.; Yue, Y.; Lai, Z.; Chen, Y.; Yang, D.; Yang, X.; Wang, Z. Spatio-temporal dynamics of urban

residential CO2 emissions and their driving forces in China using the integrated two nighttime light datasets. Appl. Energy 2019, 235, 612–624. [CrossRef] 31. Stow, D.; Petersen, A.; Hope, A.; Engstrom, R. Greenness trends of Arctic tundra Vegetation in the 1990s: Comparison of two NDVI datasets from NOAA AVHRR systems. Int. J. Remote Sens. 2007, 28, 4807–4822. [CrossRef] 32. Holben, N. Characteristics of maximum-value composite images from temporal AVHRR data. Int. J. Remote Sens. 1986, 7, 1417–1434. [CrossRef] 33. Meng, X.; Han, J.; Huang, C. An Improved Vegetation Adjusted Nighttime Light Urban Index and Its Application in Quantifying Spatiotemporal Dynamics of Carbon Emissions in China. Remote Sens. 2017, 9, 829. [CrossRef] 34. Gu, Y.; Feng, X.; Fan, L.; Guan, Z.; Feng, D.; Gao, Y. Spatial analysis of carbon emissions from region energy consumption based on night light data. Sci. Surv. Mapp. 2017, 42, 140–146. 35. Yang, Q.; Wang, T.; Chen, H.; Wang, Y. Characteristics of vegetation cover change in Xilin Gol League based on MODIS EVI data. Trans. Chin. Soc. Agric. Eng. 2015, 31, 191–198. 36. Zhu, W.; Jia, S.; Lu, A.; Yan, T. Analyzing and modeling the coverage of vegetation in the Qaidam Basin of China: The role of spatial autocorrelation. J. Geogr. Sci. 2012, 22, 346–358. [CrossRef] 37. Yan, D.; Lei, Y.; Shi, Y.; Zhu, Q.; Li, L.; Zhang, Z. Evolution of the spatiotemporal pattern of PM2.5 concentrations in China–A case study from the Beijing-Tianjin-Hebei region. Atmos. Environ. 2018, 183, 225–233. [CrossRef] 38. Anselin, L. Local indicators of spatial association—LISA. Geogr. Anal. 1995, 27, 93–115. [CrossRef] 39. Li, F.; Zhou, C.H. Spatial autocorrelation analysis on regional economic disparity of northeast economic region in China. Chin. J. Popul. Resour. Environ. 2009, 7, 27–31. 40. Zhao, J.; Zhu, Y.; Duan, G.; Yang, K. Spatial and temporal dynamics of vegetation in Xishuangbanna from 2001 to 2015 based on MOD13Q1 data. Chin. J. Ecol. 2019, 38, 1083–1092.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).