An Integrated View of Homologous Emissions of Greenhouse Gases and Air Pollutants in China

Background: Air pollution in China has raised great concerns due to its adverse effects on air quality, human health, and climate. Emissions of air pollutants (APs) are inherently linked with CO 2 emissions through fossil-energy consumption. Knowledge of the characteristics of APs and CO 2 emissions and their relationships is fundamentally important in the pursuit of co-benets in addressing air quality and climate issues in China. However, the linkages and interactions between APs and CO 2 in China are not well understood. Results: Here, we conducted an ensemble study of six bottom-up inventories to identify the underlying drivers of APs and CO 2 emissions growth and to explore their linkages in China. The results showed that, during 1980-2015, the power and industry sectors contributed 61–79% to China’s overall emissions of CO 2 , NO x , and SO 2 . In addition, the residential and industrial sectors were large emitters (77–85%) of PM 10 , PM 2.5 , CO, BC, and OC. The emissions of CH 4 , N 2 O and NH 3 were dominated by the agriculture sector (46–82%), while the share of CH 4 emissions in the energy sector increased since 2010. During 1980-2015, APs and greenhouse gases (GHGs) emissions from residential sources generally decreased over time, while the transportation sector increased its impact on recent emissions, particularly for NO x and NMVOC. Since implementation of stringent pollution control measures and accompanying technological improvements in 2013, China has effectively limited pollution emissions (e.g., growth rates of –10% per year for PM and –20% for SO 2 ) and slowed down the increasing trend of carbon emissions from the power and industrial sectors. We also found that areas with high emissions of CO, NO x , NMVOC, and SO 2 also emitted large amounts of CO 2 , which demonstrates the possible common sources of APs and GHGs. Moreover, we found signicant correlations between CO 2 and APs (e.g., NO x , CO, SO 2 , and PM) emissions in the top 5% high-emitting grid cells, with more than 60% common/overlapped grid cells during 2010–2015. Conclusions: We found signicant homology in spatial and temporal aspects for CO 2 , and NO x , CO, SO 2 , and PM emissions in China. We targeted sectorial and spatial APs and GHGs emission hot-spots, which help for management and policy-making of collaborative reductions of them. This comprehensive analysis over 6 datasets improves our understanding of APs and GHGs emissions in China during the period of rapid industrialization from 1980 to 2015. This study helps elucidate the linkages between APs and CO 2 from an integrated perspective, and provides insights for future synergistic emissions reduction.

mitigation. Here, we conducted a comprehensive study, comprehensively considering APs and GHGs emissions to identify the underlying drivers and linkages. Based on the most comprehensive public emissions inventories, we presented a detailed evaluation of the major emission sources, including the agriculture, power, industrial, residential, transportation, and waste sectors. We also aim to characterize trends, drivers and homology of anthropogenic GHGs and APs emissions in China and provide scienti c basis for addressing both air quality and climate problems more effectively from an integrated perspective.

Data And Methods
To explore the spatial patterns and trends of GHGs and APs emissions and their linkages in China, we analyzed six global and regional bottom-up inventories including 5 gridded datasets and 1 tabular dataset. The 5 gridded inventories considered in this study included the Community Emission Data System (CEDS v20210421) [41], Emissions Database for Global Atmospheric Research (EDGAR v5.0) [56], Multi-resolution Emission Inventory for China (MEIC v1.3) [43], Peking University (PKU v2) [57], and Regional Emissions Inventory in Asia (REAS v3.2) [58]. The statistical tabular dataset was retrieved from the published study of Zhao [59]. Speci cally, CEDS v20210421 is a global annual emission inventory based on a mosaic approach that provides country-level emissions by fuel and sector of the GHGs (i.e., CO 2 , CH 4 , and N 2 O) and APs (e.g., CO, NO x , SO 2 , NH 3 , BC, OC, and NMVOC) during 1750-2019 [41]. EDGAR v5.0 was developed by the European Commission's Joint Research Centre (JRC) and the Netherlands Environmental Assessment Agency (PBL), which provides sectoral and country-level emissions of GHGs (i.e. CO 2 , CH 4 , and N 2 O) and APs including ozone precursor gases (e.g., CO, NMVOC, and NO x ), acidifying gases (e.g., NO x , and SO 2 ), and primary particulates (e.g., PM 2.5 , PM 10 , BC, and OC) during 1970-2015 [60]. PKU v2 is a global monthly emission inventory that includes GHGs (i.e., CO 2 and CH 4 ) and APs (e.g., PM 2.5 , PM 10 , CO, NO x , SO 2 , TSP, NH 3 , BC, OC, PAHs) during 1960-2014 based on 64 to 88 (except CH 4 ) individual sources [61,62]. REAS v3.2 produces monthly Asian inventories of sector-speci c emissions of CO 2 , PM 2.5 , PM 10 , CO, NO x , SO 2 , NH 3 , BC, OC and NMVOC during 1950-2015 [58]. In MEIC v1.3, a technology-based approach is implemented by Tsinghua University to produce monthly anthropogenic emissions inventories over mainland China for CO 2 , PM 2.5 , PM 10 , CO, NO x , SO 2 , NH 3 , BC, OC, and NMVOC in 2008 and 2010-2017 [45,63]. In addition, we also used one statistical tabular dataset constructed by Zhao, Zhang (59], which contains a national-scale and sector-speci c emissions for CO 2 and pollutants (e.g. PM 2.5 , PM 10 , CO, NO x , SO 2 , BC, and OC) over the period of 2000-2014 in China [64]. To evaluate the characteristics of emissions under the different inventories on a common scale, speci c anthropogenic sectors were aggregated into six categories (i.e., the agriculture, power, industrial, residential, transportation, and waste sectors). Further details on the inventories adopted in this study are listed in Table 1. The differences and uncertainties of different inventories were showed in Fig. S5 and Table S1.
To quantify the spatial characteristics of CO 2 and APs emissions, we conducted target analysis in seven high-emitting areas, including the Beijing-Tianjin-Hebei region and surrounding provinces (Henan and Shandong) (BTHs), the YRD region, the Pearl River Delta (PRD) region, the Cheng-Yu (CY) region, the Fenwei Plain (FP), Northeast China (NE), and the Triangle of Central China (TC). Moreover, the top 5% high-emitting grid cells derived from speci c emissions were collected to explore the spatial linkages between carbon and pollutant emissions. Speci cally, the spatial locations of the top 5% high-emitting grids for APs were identi ed and then compared to the exact locations of the top 5% CO 2 high-emitting grids to identify common grids.  S1). These results are attributed to the differences in source categories, input activity data, and emission factors between the currently available emissions inventories [40,69]. The N 2 O emissions in the agriculture and energy sectors reported by the Food and Agriculture Organization (FAO), indicated a slightly increasing trend (AAGR: 0.8%) during 2007-2015. As the nationwide reduction in N-fertilizer applied per area has almost been offset by the expansion of the sowing area, the growth rate of N 2 O emissions stemming from croplands declined after 2003 and then plateaued until 2014 [70].
Regarding APs, the ensemble mean of NO x , NMVOC, and SO 2 emissions revealed large variations, and CO, NH 3 , and PM emissions exhibited relatively small variations, but BC and OC indicated negligible variations during 1980-2015 (Fig. 1b). The notable increase in SO 2 emissions was mainly caused by emissions originating from coal combustion in power plants in the 2000s, but considerable reductions subsequently occurred after 2006 due to the introduction of ue gas desulfurization (FGD) systems and the improvement in fuel combustion e ciency [18,71]. The SO 2 emissions exhibited a steep decreasing trend (AAGRs: -5.5%) during 2006-2015, with a peak value of 33.7 Mt in 2006 (Fig. 1b). Furthermore, SO 2 emissions were estimated to continue to decrease (AAGR: -19.6% during 2013-2017, Fig. s1) because of the shutdown of small coal-red industrial boilers and the replacement of residential coal use with electricity and natural gas in recent years [45]. NO x emissions increased rapidly in the 2000s (AAGR: 7.2%), but decreased from 2012 onward (AAGR: -4.6%, Fig. 1b). These results are attributed to the introduction of denitri cation technology (selective catalytic reduction, SCR) in large power plants and regulations targeting road vehicles [58]. NMVOC emissions maintained an almost continuous increase trend, with an AAGR of 2.9% during 1980-2015 (Fig. 1b). This growth was mainly due to the persistent growth in emissions from the industrial sector and solvent use [69], with a lag in effective emissions controls in current policies [45]. PM, such as PM 2.5 and PM 10 , exhibited a consistent trend with that of SO 2 , with peaks reached in approximately 2005. Later, particulate pollution decreased along with SO 2 due to the installation of FGD systems in power plants. Since 2013, the implementation of clean air policies has led to considerable PM 2.5 and PM 10 emissions reductions (AAGRs: -9.5% and -12.1%, respectively) during 2013-2017, which are consistent with the results from previous studies [43,63]. The implementation of stringent pollution control policies in China has effectively reduced growth rates, despite an increase in fossil fuel consumption and vehicle numbers [72,73].
BC and OC contain relatively higher uncertainties among the APs because of the lack of su cient information on the energy consumption, combustion technology, and emissions rate in the rural residential sector [51,57,74]. The differences among the various inventories range from 63% to 71% in 2015 (Fig.  S2). In speci c sectors, there are also considerable discrepancies in NO x , SO 2 and PM emissions originating from industrial sources, and the differences among the current inventories range from 40%-98% in 2015 (Fig. S2). The uncertainties in NO x and PM emissions are attributed to the emissions from cement production and industrial boilers [75]. CO emissions also indicate a large discrepancy in residential sources, and the estimates range from 29.4-67.1 Mt yr -1 in 2015. As evidenced by the substantial emissions of APs and GHGs in China and their consequences for climate change and public health, reliable inventories are of great importance in both understanding emission sources and supporting GHGs reduction and air quality improvement.

Sectoral contributions to the changes in GHGs and APs emissions across China
Quanti cation of the relative contributions of the different sectors and the evolution of high-emitting sources over time allowed us to target sector-speci c emissions reductions. The temporal changes of the ensemble mean GHGs emissions across China in the 1980s, 1990s, 2000s, and after 2010 were comprehensively determined based on the different inventories. Regarding the speci c sources, the power and industry sectors played a dominant role in CO 2 emissions growth, contributing 33.3%-57.1% to the increase during the different periods (Fig. 2a). CH 4 emissions growth was mainly driven by the power sector (31.5%-86.7%), and the impact of the emissions of the waste sector increased after 2010 (Fig. 2b). CH 4 emissions from coal production were curbed by closing a large number of small mines and increasing the e ciency in larger coal mines since 2010 [68]. In contrast, most of the N 2 O emissions originated from agricultural sources, while industrial sources increased their impact on recent emissions, accounting for 62.6% of the growth after 2010 (Fig. 2c).
In terms of speci c sectors, the power and industrial sectors contributed the most (61-79%) to CO 2 , NO x , and SO 2 emissions, while the residential and industrial sectors were large emitters (77-85%) of PM 10 , PM 2.5 , CO, BC, and OC during 1980-2015 (Fig. S2). These results further con rmed that certain APs and CO 2 are homologous, as they are all strongly associated with fossil-energy combustion [15]. CH 4 , N 2 O and NH 3 were dominated by the agriculture sector (46-82%), while the energy sector increased its share of CH 4 emissions after 2010 (Fig. 2b). The shares of both APs and GHGs emissions decreased in the residential sector, especially CO, BC, and NMVOC emissions, with the proportions decreasing by more than 30%. These results are probably attributed to the reduced emissions originating from biofuel combustion in recent years [58]. As the number of vehicles increased, the transportation sector increased its impact on recent emissions. For example, in terms of NO x and NMVOC emissions, the transportation sector contributed 9% and 14%, respectively, to the total emissions in 1980, and the contribution rates later increased to 18% and 27%, respectively. However, APs emissions stemming from transportation increased less than did CO 2 emissions because of vehicle technology improvement and fuel sulfur content reduction [30].
After the implementation of strict control measures, APs emissions began to decline or a negative growth was observed afterward. For example, the industrial sector contributed 84.9% to the decline in CO emissions during 2010-2015, which was bene cial for the improvement in the combustion e ciency and regulation tightening [76]. The power and industrial sectors contributed 45.8% and 51.5%, respectively, to SO 2 emissions reduction since 2010 (Fig. 2l), which resulted from the phasing out of shaft kilns in cement production [77,78] and from the introduction of ultra-low emissions standards for coal-red power plants [73]. These results indicated that mitigation measures were the dominant factor contributing to pollution emissions reduction and reducing carbon emissions originating from the power and industrial sectors.
In regard to BC and OC, residential sources dominated the variation in total emissions, contributing more than 60% to emissions reduction since 2010 (Fig.  2d,i). NH 3 emissions exhibited similar features to those of N 2 O emissions, with agriculture as the largest source, but its share gradually declined during 1980-2015. In terms of PM, PM 10 and PM 2.5 emissions increased steadily due to the growth in emissions in the industrial sector during 1980-2009 (Fig. 2j,k).
Recently, PM emissions have experienced substantial changes since 2010, which can be attributed to the integrated effect of emissions reduction in the power, industrial, and residential sectors. The reduction in emissions of certain APs could also lead to a decrease trend in PM emissions during 2010-2015. For example, BC is a major component of primary PM 2.5 , whereas SO 2 and NO x (precursors of sulfate and nitrate aerosols, respectively) are crucial precursors in the formation of secondary PM 2.5 [58].

Spatial relationship between CO 2 and APs emissions in high-emitting areas
The ve available gridded emissions inventories of CEDS, EDGAR, MEIC, PKU, and REAS were analyzed to explore the spatial characteristics of CO 2 and APs emissions. Regions with high anthropogenic emissions generally host a large population and rapid economic and industrialized development. Therefore, seven high-emitting areas were analyzed to identify the major sources and possible linkages between pollutants and CO 2 emissions via multiple inventories (Fig. 3). During 2010-2015, the BTHs and YRD regions were the main contributors to the national total emissions (Fig. 3d,e) linkages occur because these emissions are largely contributed by power and industrial sources (Fig. 2). Because of the common sources, control measures targeting speci c APs should be considered complementary to CO 2 mitigation strategies [14,24]. The consistent spatial patterns of pollutants and CO 2 emissions reveal the importance of synergy in controlling emissions in high-emitting areas and of setting source-speci c emissions reduction targets.
To further identify the relationship between CO 2 and APs emissions at the grid cell level, the numbers of the top 5% high-emitting grids (representing emissions hotspots) were extracted from CO 2 inventories to detect the consistency and difference between CO 2 and APs emissions. As shown in Fig. 4, the highest correlation was found between CO 2 and NO x , with the coe cient of determination (R 2 ) mostly remaining above 0.9, and the numbers of common grid cells of CO 2 and APs hotspots reached more than 70% of all grid cells during 2010-2015 (Fig. 4). Strong correlations were observed between CO 2 and CO, SO 2 , and PM emissions, with R 2 values generally exceeding 0.6, and the common grid cells accounted for more than 60% of the total number of grid cells. These results revealed that there occurred highly consistent patterns in source locations and high-emission intensity areas. These results were probably due to that the power and industrial sources were the dominant contributors (Fig. 2). There appeared to be a limited correlation between CO 2 and BC or OC based on MEIC, PKU, and REAS, with R 2 values ranging from 0.1 to 0.5. This is because the dominant contributor to BC and OC emissions was the residential sector. However, a strong correlation was observed between CO 2 and BC or OC based on EDGAR and CEDS. This could be attributed to the power and industrial sources largely contributing to BC and OC in EDGAR. Similarly, the relationship between CO 2 and NMVOC emissions was in uenced by the different emission sources. Solvent use and industry dominated the growth in NMVOC emissions (Li 2019), while CO 2 was mainly driven by the power and industrial sectors. Hence, the R 2 values generally ranged from 0.42-0.75. NH 3 emissions hotspots indicated a weak relationship with CO 2 (R 2 <0.3) and the smallest number of common grids (28%-39%). The agriculture sector was the major contributor to NH 3 emissions (Fig. 2a,h). The relatively higher positive correlation (R 2 =0.54) between NH 3 and CO 2 based on PKU was largely because PKU mainly considered emissions from combustion and industrial sources but did not include agriculture [61].

Relationships between the changes in CO 2 and APs emissions over time
China's emissions have dramatically changed, especially after the implementation of control targets for the emissions of speci c APs and carbon emissions in recent years. To quantify the relationships between CO 2 and APs emissions changes over time, 6 inventories, including CEDS, EDGAR, MEIC, PKU, REAS, and Zhao, were evaluated at the sectoral level during 2010-2015. As illustrated in Fig. 5, in the residential sector, effective controls of APs and CO 2 emissions were observed based on MEIC and PKU, which were located in the third quadrant. In the power and industrial sectors, pollution generally exhibited negative changes with CO 2 emissions growth, which were mainly located in the fourth quadrant (Fig. 5c,d,f,h-k). This further con rmed that actions to mitigate air pollution were more effective than was limiting CO 2 emissions. However, for NMVOC and NH 3 , emissions still increased with CO 2 , which was located in the rst quadrant.
This could be because NMVOC and NH 3 emissions were mainly driven by persistent growth due to the industry and solvent use and the lack of relevant emissions controls over SO 2 , NO x , and PM [45,46,69].
Pollutants emissions stemming from the waste sector tended to increase with CO 2 among the inventories. The total amount of municipal solid waste continues to grow with the population, urbanization, and industrialization levels [79]. The amount of solid waste treatment increased by 25% during 2010-2015, and the level consistently increased to 10.11 Mt in 2019 [80]. Although the power and industrial sectors are the dominant emission sources in China, transportation has contributed a growing share to the total emissions due to the increase in motorization (Fig. S2). In the transportation sector, pollutants emissions based on PKU, MEIC, and Zhao re ected negative changes with CO 2 growth (in the fourth quadrant, as shown in Fig. 5a,b,f-h), except for NO x and NH 3 (in the rst quadrant, as shown in Fig. 5d,e). In the agriculture sector, emissions indicated low interannual variability among the inventories, except for PKU. This discrepancy was mainly attributed to PKU only including the enhanced contributions of both fossil fuel and biomass combustion to the agricultural emissions.
After the implementation of stringent air quality control measures for several years after 2011, the majority of pollutants exhibited a decreasing trend. Moreover, compared to the 2000s, the growth in CO 2 emissions has successfully declined in recent years (Fig. S4). During 2000-2009, APs emissions were positively correlated with CO 2 , and the trends were generally located in the rst quadrant (Fig. S4). It is encouraging to nd that China's efforts to mitigate both air pollution and climate change have taken effect. To further reduce pollutants emissions, more effective strategies are needed to strengthen controls on NMVOC and NH 3 emissions and emissions originating from vehicle and waste sources.

Conclusions
Driven by the increase in energy consumption, urbanization, and vehicle number, air pollution and carbon emissions in China have increasingly become a serious problem, especially due to their negative impacts on air quality, human health, and climate change. Knowledge of spatiotemporal characteristics and exploration of possible links between APs and GHGs are imperative to effectively mitigate both air pollution and climate change. Through analysis of six global and regional bottom-up inventories, the results in this study revealed that CO 2 , NO x , and SO 2 emissions were closely linked because they were all mainly driven by the power and industrial sectors during 1980-2015. In regard to PM 10 , PM 2.5 , CO, BC, and OC, the residential and industrial sectors were the largest contributors to the total emissions. Both APs and GHGs exhibited a decreasing emissions share stemming from residential consumption, especially for CO, BC, and NMVOC, with the proportions decreasing more than 30%. However, the transportation sector increased its impact on recent emissions, particularly in NO  Table S1 and Fig. S5 for their differences and uncertainties.