Analysis of the National Air Pollutant Emissions Inventory (CAPSS 2018) Data and Assessment of Emissions Based on Air Quality Modeling in the Republic of Korea

According to the 2018 National Air Pollutant Emissions Inventory (NEI), air pollutant emissions in the Republic of Korea comprised 808,801 tons of CO, 1,153,265 tons of NO X , 300,979 tons of SO X , 617,481 tons of TSP, 232,993 tons of PM 10 , 98,388 tons of PM 2.5 , 15,562 tons of black carbon (BC), 1,035,636 tons of VOCs, and 315,975 tons of NH 3 . As for national emission contributions to primary PM 2.5 and PM precursors (NO X , SO X , VOCs, and NH 3 ), major source categories were the road sector for NO X , the industry sector for SO X and PM 2.5 , and the everyday activities and others sector for VOCs and NH 3 . In the case of emissions by region, the largest amount of NO X was emitted from the Seoul Metropolitan Areas (SMA; Seoul, Incheon, and Gyeonggi-do, hereafter SMA) and the largest amounts of SO X , PM 2.5 , VOCs, and NH 3 were from the Yeongnam region. A 3D chemical transport modeling system was used to examine the uncertainty of the national air pollutant emissions based on the National Emission and Air Quality Assessment System (NEAS). Air quality was simulated using CAPSS 2018, and the simulation data were compared with observed concentrations to examine the uncertainties of the current emissions. These data show that emissions from five si (cities) (Pohang, Yeosu, Gwangyang, Dangjin, and Ulsan) need to be improved. Most of all, it is necessary to examine the emissions from places of business that use anthracite, which is the major PM 2.5 emission source, as fuel in these areas.


INTRODUCTION
The government of the Republic of Korea announced the Comprehensive Measures on Fine Dust Management that aims to reduce particulate matter (PM) emissions by 30% (MOE, 2017), and implemented the strengthened plan on fine dust (MOE, 2019).Evidently, despite these aggressive reduction efforts of the government, atmospheric PM 2.5 concentrations were not significantly reduced and high PM concentrations were not mitigated while public awareness is low.Therefore, the Comprehensive Plans for Fine Dust Management and plan on air Environment Management by Region (SMA, Central area, Southern area, and Southeast area) were established and implemented (Kim et al., 2022;MOE, 2020;MOE, 2019).
Since the characteristics of pollutant emission differ by area, it is necessary to identify the major emission sources and analyze their emission contributions to effectively improve PM emissions (Bae et al., 2021).Air pollutant emissions have different characteristics in different regions depending on the topography and industrial structure.For example, SMA has the largest amounts of car-related pollutants in Korea as it has 50% of the country's total population and cars; whereas, Gangwon-do's annual air pollutant emissions are relatively low because of small population and underdevelopment of industrial complexes, which is the result of its mountainous topography (NAIR, 2021).In recent years, research has been conducted on the PM concentrations and emission status considering such regional characteristics (Gong et al., 2021;Hwang et al., 2021), and mutual impacts among neighboring areas, caused by PM emissions, have been analyzed (Kim et al., 2021a, b, c;You et al., 2020).
NAIR assesses and publishes the emissions of 9 air pollutants (CO, NO X , SO X , TSP, PM 10 , PM 2.5 , black carbon [BC], VOCs, and NH 3 ) for 17 dos (provinces) and metropolitan cities and 250 si (city), gun (county), gu (district) every year (note: Emissions from the sea were managed separately) (Choi et al., 2021).Based on this, the central and local governments need to establish customized PM reduction measures to protect the health and property of local residents and minimize the economic loss of industries.
In this study, the 17 dos (provinces) and metropolitan cities were classified into 5 regions (SMA, Gangwon region, Chungcheong region, Honam region, and Yeongnam region) and changes in air pollutant emissions by region were analyzed using the 2018 national air pollutant emission estimation results.In addition, the uncertainty of domestic emissions was examined by region and pollutant through a comparison between the simulated concentrations using 3D chemical transport model with ground level observed concentrations.

NATIONAL AIR POLLUTANT EMISSION ESTIMATION METHOD AND IMPROVEMENTS
As for national air pollutant emissions, the measurement-based emission data of a tele-monitoring system (TMS) were utilized, as it was in previous studies (Choi et al., 2021;Choi et al., 2020;Yeo et al., 2019), or the emissions of 9 pollutants (e.g., PM 2.5 , NO X , and SO X ) were estimated in 13 first-level categories, 56 second-level categories, and 240 third-level categories by applying  approximately 30,000 emission factors using approximately 300 statistical data from approximately 150 related organizations (e.g., pollutant-emitting places of business, and those related to transportation and meteorology) as activity data (NAIR, 2022).There were improvements in emission estimation method compared to 2017; The way to collect the activity data from road transport, non-road transport, and agriculture were improved.The details are as follows (NAIR, 2021) (Table 1).

1 National Air Pollutant Emissions
In the 2018 NEI, the national emissions of air pollutants comprised 808,801 tons of CO; 1,153,265 tons of NO X ; 300,979 tons of SO X ; 617,481 tons of TSP; 232,993 tons of PM 10 ; 98,388 tons of PM 2.5 ; 15,562 tons of BC; 1,035,636 tons of VOCs; and 315,975 tons of NH3 (Table 2).
The emission contributions of different emission ), industrial process (14.6%) for NH 3 (Fig. 1).For primary PM 2.5 and PM precursors (NO X , SO X , VOCs, and NH 3 ), the 13 first-level source categories were classified into five sectors (energy, industry, road, nonroad, and everyday activities and others), as presented in Table 3.The national air pollutant emissions in 2018 were compared with those in 2017, and major causes of changes in emissions were analyzed.
NO X emissions decreased by 3.1% compared to the previous year due to the replacement old cars with new cars in the road sector and the reinforcement of the emission control for power plants in the energy sector.SO X emissions decreased by 4.6% compared to the previous year due to the reduction in fuel consumption (including B-C oil) of power plants and strengthened emission control.PM 2.5 emissions increased by 7.3% due to the increase in the number of ships and construction machinery registrations in the non-road sector.VOCs emis-sions decreased by 1.1% compared to the previous year due to the decline in paint supply in the everyday activities and others sector.NH 3 emissions increased due to the increase in fertilizer consumption and the number of livestock population in the everyday activities and others sector (Fig. 2).
To examine pollutant emission characteristics by region, the current status of factors related to major pollutant emission, such as population, economy, large-scale places of business, cars, and construction machinery, was analyzed.For the analysis of the current status of the economy by region, Gross Regional Domestic Product (GRDP) data published by the Korean Statistical Information Service (KOSIS) were utilized.GRDP is the sum of the market prices of all final goods and services produced in a fixed economic zone for a certain period of time.It is used to establish local financial and economic policies because it comprehensively represents the status of the local economy.As of 2018, SMA showed the high-est values and proportions for both population (49.8%) and GRDP (52.2%), followed by the Yeongnam, Chungcheong, Honam, and Gangwon regions, and Jeju Island (Table 5).
According to the analysis results, the Yeongnam region  had the largest number of large-scale places of business (annual pollutant emissions>20 tons; 37.5%), followed by SMA (24.4%) and the Chungcheong region (19.1%).
For the analysis of construction machinery, SMA had the largest number of registered vehicles (44.5%) and excavators (33.5%), followed by the Yeongnam region (26.8 and 27.3%, respectively) (Table 6).
The analysis on the major cause of changes in emissions and the comparison of regional and sectoral emissions based on emissions by region and pollutant is presented in the next section.

2. 1 Analysis of Changes in Emissions for SMA
Almost half of the national population of Korea is concentrated in SMA, which consists of Seoul Metropolitan City (the capital), Incheon, and Gyeonggi-do, as it is the center of politics, economy, society, and culture.To improve the air pollution of SMA caused by high population density, traffic congestion, and industrialization, a separate law (Special Act On The Improvement Of Air Quality In Seoul Metropolitan Area, 2003) was enacted.Based on this, the Air Quality Management Plan in Seoul Metropolitan Area (2005) has been established and implemented.The plan includes strengthening of vehicle emission standards, supply of eco-friendly vehicles and expansion of infrastructure, details regarding total air pollutant emissions limitations for places of business, mandatory installation of VRU at gas stations, reinforcing the management of fugitive dust from vacant lands and places of business.
The population and economy indicators showed that SMA had the largest population (approximately 49.8%) and recorded the highest GRDP (approximately 52.2%) in 2018.The electric, electronic, and precision instrument manufacturing sector constituted the highest proportion of GRDP.
Air pollutant emissions from SMA in 2018 were estimated to be 17,162 tons of PM 2.5 , 22,120 tons of SO X , 322,296 tons of NO X , 318,393 tons of VOCs, and 58,023 tons of NH 3 .In addition, the contributions of each pollutant to the national emissions were as follows: PM 2.5 (17.4%),SO X (7.3%), NO X (27.9%),VOCs (30.7%),NH 3 (18.4%).PM 2.5 and VOCs emissions increased by 4.4% and 1.1% compared to the previous year, while SO X , NO X , and NH 3 emissions decreased by 10.8%, 0.2%, and 0.2%, respectively.Meanwhile, the contributions of NO X and SO X to the emissions from the road and industry sector respectively were the largest compared to other pollutants.In addition, PM 2.5 , VOCs, and NH 3 contributed the largest to the emissions from the everyday activities and others sector (Fig. 4).SMA's emissions from the road transport recorded the largest compared to other regions as it recorded the largest number of vehicles registered (44.5%), and VKT (40.5%) (PM 2.5 36.1%,SO X 40.2%, NO X 38.9%, VOCs 46.9%, and NH 3 40.9%).SO X , NO X , VOCs, and NH 3 emissions decreased compared to those of the previous year.This is due to the replacement of old vehicles with new ones, which offset the effects of the increase in the number of vehicle registrations (3.1%, 307,000 units) and VKT (2.7%, 3,428 million km).
SMA's emissions from the non-transport sector were also the largest compared to those of other regions.(NO X : 27.5%, PM 2.5 : 28.0%, VOCs: 27.5%, and NH 3 : 27.6%).The region's PM 2.5 and NO X emissions from the construction machinery increased by 12.3% (366 tons) and 11.3% (6,867 tons) compared to those in the previous year.This was because construction machinery registrations (including excavators) increased by 9.9% (20,671 units) and the swaths of construction sites increased by 7.0% (5,001 m 2 ).VOCs emissions increased by 23.7% (3,568 tons) compared to that of the previous year.This was mainly because of the decrease in the number of registered leisure boat using gasoline in Incheon (by 26.4%, 880 units).SO X emissions decreased by 10.8% (2,686 tons) compared to those of the previous year.In particular, SO X emissions from the industry sector decreased by a large margin (10.3%, 1,077 tons).This was due to the decreased consumption of industrial bituminous coal (66,000 tons, 19.5%) in Gyeonggi-do.

2. 2 Analysis of Changes in Emissions for
the Gangwon Region Most of the Gangwon region, which is located in the northeastern part of Korea, is mountainous.Under the influence of such geographical conditions, industrial complexes are underdeveloped, which led to relatively low proportion of manufacturing-based industries.
The population and economy indicators showed that this region accounted for approximately 3.0% of the national population as of 2018.The GRDP of the region was approximately 2.5% of the national GRDP.More specifically, the public administration, defense and social security-related administration sector showed the highest proportion in GRDP.The manufacturing sector represented approximately 0.9% of the national GRDP.
Air pollutant emissions from the Gangwon region in 2018 were estimated to be 4,109 tons of PM 2.5 , 13,802 tons of SO X , 79,834 tons of NO X , 30,263 tons of VOCs, and 14,848 tons of NH 3 .In addition, the contributions of each pollutant to the national emissions were as follows: PM 2.5 (4.2%), SO X (4.6%), NO X (6.9%), VOCs (2.9%), NH 3 (4.7%).PM 2.5 , SO X , and NO X emissions decreased by 0.1%, 2.3%, and 7.2%, respectively, compared to those of the previous year, whereas VOCs and NH 3 increased by 6.7% and 7.6%, respectively.Meanwhile, in the Gangwon region, the contributions of PM 2.5 , VOCs, and NH 3 to the emissions from everyday activities and others sector respectively were the largest compared to other pollutants.In addition, NO X and SO X contributed the largest to the emissions from the indus- (Unit: metric tons/year) try sector (Fig. 5).
Emissions from the energy sector increased compared to those of the previous year (NO X : 10.9%, SO X : 29.6%, PM 2.5 : 66.8%, VOCs: 50.2%, and NH 3 : 58.0%).This was because the consumption of coal (including bituminous coal) and LNG increased by 43.7% (3,228,000 tons) and 62.2% (269 million m 3 ), respectively, due to the operation of new thermal power plants (coal and LNG).
VOCs emissions from the non-road transport sector increased by 38.0% (1,449 tons) compared to those of the previous year.This was mainly because of the increase in the number of registered leisure boat (by 47.2%, 1,570 units).
On the other hand, NO X and SO X emissions from the industry sector decreased by 10.4% (4,992 tons) and 9.1% (764 tons), respectively, compared to those of the previous year.This was due to the reduction in the fuel (bituminous coal) consumption of cement production facilities.NH 3 emissions increased by 67.1% compared to those in the previous year.This was mainly because emissions from DeNO X facilities in the industry sector increased by 67.7% (826 tons).

2. 3 Analysis of Changes in Emissions for
the Chungcheong region The Chungcheong region, located in the center of Korea, consists of Daejeon Metropolitan City, Sejong Special Self-governing City, Chungcheongnam-do, and Chungcheongbuk-do.In the western part of the region, thermal power plants (coal and LNG), petrochemical complexes, iron and steel mills, and large manufacturing industries are located near trading ports.Meanwhile, in the eastern part of the region, high-value-added manu-facturing industries (e.g., medicine and electronics) and food manufacturing industries are located.
The population and economy indicators showed that the region represented approximately 10.7% of the national population as of 2018.The GRDP of the region was approximately 12.5% of the national GRDP.The electric, electronic, and precision instrument manufacturing sector showed the highest proportion in GRDP, followed by the coal and petrochemical product manufacturing sector.
Air pollutant emissions from the Chungcheong region in 2018 were estimated to be 22,719 tons of PM 2.5 , 77,784 tons of SO X , 189,823 tons of NO X , 143,997 tons of VOCs, and 73,667 tons of NH 3 .In addition, the contributions of each pollutant to the national emissions were as follows: PM 2.5 (23.1%),SO X (25.8%),NO X (16.5%),VOCs (13.9%),NH 3 (23.3%).PM 2.5 and NH 3 emissions increased by 9.5% and 1.5%, respectively, compared to the previous year, whereas SO X , NO X , and VOCs emissions decreased by 1.8%, 5.9%, and 3.8%, respectively.Meanwhile, in the Chungcheong region, the contributions of PM 2.5 , NO X and SO X to the emissions from the industry sector respectively were the largest compared to other pollutants.In addition, VOCs and NH 3 contributed the largest to the emissions from the everyday activities and others sector (Fig. 6).
In the case of the Chungcheong region, pollutant emissions from the energy sector were large compared to other regions (PM 2.5 : 37.4%, SO X : 36.9%,NO X : 25.6%).NO X and SO X emissions from the energy sector decreas ed by 24.1% (7,838 tons) and 16.3% (4,081 tons), respectively, compared to those of the previous year.This was because of the reinforcement of the power plant emission management, which offset the effects of increased consumption (Unit: metric tons/year) of coal (including bituminous coal) in the coal-fired power plants of the region (1.0%, 444,000 tons) compared to the previous year.
PM 2.5 and SO X emissions from the industry sector increased by 20.1% (2,242 tons) and 6.6% (3,295 tons) compared to those of the previous year.This was because of the increase in anthracite consumption in the primary metal industry (23.3%).
VOCs emissions decreased by 3.8% (5,737 tons) compared to those of the previous year.More specifically, VOCs emissions from the everyday activities and others sector decreased by 7.3% (7,246 tons) compared to those of the previous year.This was due to the emissions reductions (6,501 tons, 22.5%) caused by the decrease (21.8%) in the consumption of paint used for architecture and buildings in the region.
Meanwhile, NH 3 emissions represented 23.3% of the national emissions, and increased by 1.5% (1,084 tons) compared to those of the previous year.This was mainly because the emissions from the agriculture-manure management sector increased by 1.3% (720 tons), which was caused by a 3.3% (1,606,000 units) increase in the number of livestock population, including cows, pigs, and chickens.

2. 4 Analysis of Changes in Emissions for
the Honam Region The Honam region, which consists of Gwangju Metropolitan City, Jeollabuk-do, and Jeollanam-do, is located in the southwestern part of Korea.It is Korea's representative breadbasket with wide plains, such as Honam and Naju plains.Thermal power plants (coal and LNG) and the nation's largest petrochemical complex are located in Yeosu, a southern part of the region, in addition to near-by iron and steel mills in Gwangyang.
The population and economy indicators showed that the region accounted for approximately 10.0% of the national population as of 2018.The GRDP of the region is approximately 8.7% of the national GRDP.More specifically, the coal and petrochemical product manufacturing sector showed the highest proportion of GRDP.
Air pollutant emissions from the Honam region in 2018 were estimated to be 17,265 tons of PM 2.5 , 62,5554 tons of SO X , 156,101 tons of NO X , 174,525 tons of VOCs, and 79,892 tons of NH 3 .In addition, the contributions of each pollutant to the national emissions were as follows: PM 2.5 (17.5%),SO X (20.8%),NO X (13.5%),VOCs (16.9%),NH 3 (25.3%).PM 2.5 and NH 3 emissions increased by 11.7% and 6.3%, respectively, compared to those of the previous year, whereas SO X , NO X , and VOCs decreased by 0.1%, 0.8%, and 2.5%, respectively.Meanwhile, the contributions of PM 2.5 , and SO X to the emissions from the industry sector, the contributions of NO X to the emissions from the road sector, the contributions of VOCs and NH 3 emissions from the everyday activities and others sector were the largest in the region (Fig. 7).
In the case of the industry sector, PM 2.5 , SO X , and NO X emissions increased by 30.9% (2,010 tons), 5.8% (2,671 tons), and 6.4% (2,666 tons), respectively, compared to those of the previous year.This was mainly because the increased consumption of coal, including anthracite, in the manufacturing sector (13.3%, 318,000 tons) in Jeollanam-do.
NO X emissions from the road transport and non-road transport sectors decreased by 5.1% (2,611 tons) and 5.9% (2,078 tons), respectively, compared to those of the previous year.For road transport, emissions from the sector decreased because of the decrease in the number (Unit: metric tons/year) of old cars registrations and the replacement of old cars with new ones, which offset the impacts of the increase in the number of car registrations (3.0%, 77,000 units) in the region.In the case of the non-road transport, NO X emissions decreased mainly because of the decrease in emissions from the non-road-construction machinery sector (13.6%, 1,674 tons) caused by the reduction in the number of registered construction machinery (14.6%, 6,162 units) in the region.
VOCs emissions decreased by 2.5% (4,561 tons) compared to those in the previous year.More specifically, this was because emissions by paint that is used for shipbuilding decreased by 16.9% (2,528 tons).For paint consumption, it decreased by 17.1% (4,606 kL) compared to those in the previous year.
NH 3 emissions increased by 6.3% (4,722 tons) compared to those in the previous year.The Honam region exhibited the largest NH 3 emissions in the country from the everyday activities and others sector.This was due to the 21.0% (1,335 tons) increase in emissions caused by a 20.8% (44,000 tons) increase in fertilizer consumption in farmlands, and the 5.0% (2,873 tons) increase in NH 3 emissions from the manure sector caused by a 10.2% (5,902,000 units) increase in the number of livestock population.

2. 5 Analysis of Changes in Emissions for
the Yeongnam Region The Yeongnam region, which consists of Busan Metropolitan City, Daegu Metropolitan City, Ulsan Metropolitan City, Gyeongsangbuk-do, and Gyeongsangnamdo, is located in the southeastern part of Korea.Iron and steel manufacturing, shipbuilding, automobile manufacturing, and petrochemical industries as well as the nation's largest trading port (Busan Port) are located in the region.
The population and economy indicators showed that the region represented approximately 25.3% of the national population as of 2018.The GRDP of the region is approximately 23.1% of the national GRDP.More specifically, the machinery transport equipment, and other product manufacturing sector showed the highest proportion of GRDP, followed by electric, electronic, and precision instrument manufacturing and non-metallic mineral and metal product manufacturing sector.
Air pollutant emissions from the Yeongnam region in 2018 were estimated to be 32,945 tons of PM 2.5 , 113,601 tons of SO X , 302,187 tons of NO X , 344,649 tons of VOCs, and 81,881 tons of NH 3 .In addition, the contributions of each pollutant to the national emissions were as follows: PM 2.5 (33.5%),SO X (37.7%),NO X (26.2%),VOCs (33.3%),NH 3 (25.9%).PM 2.5 and NH 3 emissions increased by 7.0 and 1.2%, respectively, compared to those in the previous year, whereas SO X , NO X , and VOCs emissions decreased by 6.5%, 3.8%, and 1.8%, respectively.Meanwhile, the contributions of PM 2.5 and SO X to the emissions from the industry sector were the largest in the region.In addition, VOCs and NH 3 contributed the largest to the emissions from the everyday activities and others sector (Fig. 8).
In the case of the Yeongnam region, air pollutant emissions from the industry sector were found to be the largest in Korea.In this region, emissions from the industry sector were 17,459 tons of PM 2.5 , 79,097 tons of SO X , 76,780 tons of NO X , 95,344 tons of VOCs, and 17,153 tons of NH 3 .Each pollutant represented 43.1% (PM 2.5 ), 40.1% (SO X ), 31.2%(NO X ), 38.2% (VOCs), and 36.6% (NH 3 ) of national emissions from the industry sector, (Unit: metric tons/year) respectively.PM 2.5 and NO X emissions from the sector increased by 15.8% (2,380 tons) and 7.7% (5,459 tons), respectively, compared to those in the previous year.This was mainly because the consumption of coal, including anthracite, in the manufacturing sector increased by 15.1% (504 tons) compared to that in the previous year.
PM 2.5 emissions increased by 7.0% (2,156 tons) compared to those of the previous year.This was because of the increase in the consumption of anthracite in the industry sector.Meanwhile, NH 3 emissions also increased by 1.2% (982 tons) compared to the previous year.This was mainly because emissions from DeNO X facilities in the industry sector increased by 55.0% (1,282 tons).
NO X emissions decreased by 3.8% (12,100 tons) compared to those in the previous year.These emissions decreased by 18.7% (5,854 tons) and 7.9% (9,581 tons) in the energy production and road transport sectors, respectively.For the energy production sector, this was mainly because of the reduction (4.1%) in the bituminous coal consumption by public power generation facilities and the reduction (19.6%, 5,292 tons) in emissions caused by the reinforcement of environmental facilities for power generation facilities.In the case of the road sector, the main cause of such reduction was the decrease in emissions caused by the reduction in the number of old vehicles, which offset the impacts of the increase in vehicles registrations and VKT increased by 2.0% (124,000 units) and 0.8% (750 million km), respectively, compared to those of the previous year.SO X emissions decreased by 6.5% (7,866 tons) compared to those of the previous year.This was mainly due to the emissions reductions in the energy sector (15.5%, 3,697 tons), the non-road sector (17.1%, 1,966 tons), and the everyday activities and others sector (29.7%, 2,000 tons).More specifically, for the energy sector, the main cause of such reduction was the decrease in emissions from public power generation facilities as it was for NO X .In the case of the non-road sector, the emissions reductions caused by the decrease in the number of cargo ships entering the ports (6.9%, 6,640 units) and the decrease in the sulfur content in fuel (B-C oil).For the everyday activities and others sector, such reductions were due to the emissions reductions (49.3%, 1,424 tons) caused by the reduction in the consumption of fuel oil for cooling and heating (9.4%, 174,000 kL) in commercial and public facilities.
VOCs emissions from the everyday activities and others sector decreased by 1.8% (6,417 tons) compared to those of the previous year.This was mainly due to the decrease in emissions (6.7%, 8,609 tons) caused by the reduction in the consumption of paint at coating facilities (6.5%, 17,719 kL).VOCs emissions from the nonroad sector, on the other hand, increased by 23.6% (2,662 tons).This was due to the emissions increase (47.0%, 2,844 tons) caused by an increase in the number of registered leisure boat (47.2%, 785 units).

1 Methodology
The latest activity data and the best available emission factors were applied to the emissions data estimated above.Nevertheless, there are uncertainties in some emission sources.Old emission factors, activity data (Unit: metric tons/year) with low reliability, and missing emission sources are mentioned as the causes of such uncertainties (Kim et al., 2020a;Lee et al., 2019;Kim and Jang, 2014).Therefore, it is necessary to examine the uncertainty of the estimated emission data.Since air pollutants have different emission characteristics depending on the emission sources, it is difficult to verify emissions in a consistent way and present the results in a quantitative manner.To overcome such difficulties, a method of indirectly examining the accuracy of emissions has been used.This methodology is about comparing the concentrations data from monitoring stations with the results of air quality modeling, a process of converting air pollutants emissions into atmospheric concentrations using 3D chemical transport model (Bae et al., 2020a, b;Kim et al., 2020a).
As such, this study uses a method of utilizing 3D chemical transport model to examine the uncertainty of national air pollutant emissions.This study was conducted based on the National Emission and Air quality assessment System (NEAS).NEAS consists of the Weather Research and Forecasting (WRF) model, the Sparse Matrix Operator Kernel Emissions (SMOKE) model, and the Community Multiscale Air Quality (CMAQ) model.The detailed physico-chemical options used in the WRF and CMAQ models are presented in Supplementary Materials (Table S1).CAPSS 2018 was used for domestic emissions and the KORUSv5 data was used for overseas emissions.The domains and horizontal resolutions were for the simulation were as follows: Northeast Asia (27 km), the Korean Peninsula (9 km), and South Korea (3 km).And 2018 was selected as the target year for simulation (Supplementary Materials Fig. S1).NO 2 and SO 2 were selected as the target pollutants for which the uncertainty of emissions was to be examined by considering the following three aspects: 1) The two pollutants themselves are harmful to people's health.It is important to identify the emissions of their uncertainty as they are major precursors transformed into PM through secondary formation in the atmosphere; 2) It is easy to intuitively interpret the overestimation/underestimation as emissions and concentrations of NO 2 and SO 2 have a relatively linear relationship, nature of primary pollutants; 3) Since are NO 2 and SO 2 less affected by long-range transport, it is easy to assess the emissions of each region.However, this study suggests the results of comparison between simulated and observed concentrations of PM 2.5 as well because of the importance PM 2.5 has.
An error between the simulated and observed concentrations may occur due to various factors.Representative factors are the uncertainties of meteorological input data, emissions input data, and various physical and chemical equations included in atmospheric chemical transport models.This study assumes that the systematic bias found at a similar level in most regions drives from the errors between meteorological input data and atmospheric chemical transport models.This is to examine the model's errors occurring from the perspectives of emissions.And the study presents and analyzes the regions with large errors between observed and simulated concentrations while comparing their annual mean concentrations to examine the uncertainty of emissions from the perspectives of total emissions.

2 Comparison between Simulated and
Observed Concentrations Based on the locations of the urban air pollution monitoring network, errors between the simulated and observed annual mean concentrations across Korea were found to be 0.6 ppb (3%) for NO 2 , 0.1 ppb (4%) for SO 2 , and 5.6 μg/m 3 (24%) for PM 2.5 .The simulated concentrations of gaseous pollutants were similar to the observed concentrations relative to PM 2.5 .And PM 2.5 concentrations were underestimated compared to the observed concentrations.In addition, this study compared the simulated and observed concentrations at a provincial and metropolitan city level.The bias of the simulated NO 2 concentrations was found to range from -7.5 ppb (-35%, Chungcheongbuk-do) to 5.6 ppb (41%, Jeollanam-do).And high reproducibility was observed in Daejeon, Daegu, and Jeju with the error<1 ppb (<5%) (Fig. 9[a]).The bias of the simulated SO 2 concentrations ranged from -2.9 ppb (-66%, Seoul) to 7.4 ppb (144%, Jeollanam-do).Overestimation occurred in 5 out of 17 dos (provinces) and metropolitan cities.And it was particularly notable in Jeollanam-do, Gyeongsangbukdo, and Ulsan (Fig. 9[b]).The bias of the simulated PM 2.5 concentrations ranged from -9.5 μg/m 3 (-35%, Jeollabuk-do) to 6.4 μg/m 3 (26%, Gyeongsangbuk-do), and underestimation occurred in 15 out of the 17 dos (provinces) and metropolitan cities (Fig. 9[c]).
For dos (provinces) and metropolitan cities that exhibited an underestimation, a tendency towards underestimation was generally observed in most of the municipalities as well.On the other hand, for dos (provinces) and metropolitan cities that showed an overestimation, high simulated concentrations intensively occurred in some of the municipalities.Such municipalities include Gyeongsangbuk-do (Pohang), Jeollanam-do (Yeosu), Jeollanamdo (Gwangyang), and Chungcheongnam-do (Dangjin), and the simulated NO 2 , SO 2 , and PM 2.5 concentrations in those municipalities were 2-3 times higher than the observed concentrations of the same pollutants.However, for Ulsan, overestimation of SO 2 concentrations occurred at most of the air quality monitoring stations (11 stations, 73%), which was an exceptional case (Fig. 10).
This study assumes that the uncertainty of emissions would to be high for regions where the errors between the simulated and observed concentrations were large.The PM 2.5 concentrations in the atmosphere, however, are known to be affected in a complex manner by direct emissions from emission sources, secondary formation in the atmosphere by the chemical reactions of precursors, and long-range transport (Kim et al., 2021d;Kim et al., 2017a;Kim et al., 2017b).The possibility that  long-range transport affected the errors between simulated and observed concentrations was determined to be low because it affects the entire country rather than specific regions (Bae et al., 2021).The components generated by the secondary formation caused by precursors accounted for 50-60% of the domestic PM 2.5 concentrations (Kim et al., 2020b).This explains why it is necessary to analyze the uncertainty of precursor emissions in addition to the uncertainty of the air pollutants directly emitted as PM 2.5 .
Comparing observed and simulated concentrations on the basis of the concentrations of PM's detailed components would be the most direct way to distinguish the impacts of secondary formation from those of direct emission in the process of analysis.However, since the number of monitoring stations measuring the concentrations of PM 2.5 components is extremely limited, simulated concentrations were analyzed based on the following two assumptions: 1) If the uncertainty of primary PM emissions is large, the error will be relatively large in regions adjacent to emission sources due to their direct impacts from emission sources, and the proportion of primary PM components will be relatively high in the simulated PM 2.5 concentrations; 2) If the uncertainty of precursor emissions is higher, on the other hand, it takes some time for PM 2.5 to be generated after precursors come from emission sources.Therefore, the errors are likely to be larger in the downwind region relatively far from emission sources, and the proportion of secondary components (such as NO X and SO X ), will be high in the simulated concentrations.
Figure 11 shows the spatial distribution of the relative proportion of primary PM components in the simulated PM 2.5 concentrations.For the Pohang, Yeosu, Gwangyang, and Dangjin regions mentioned above, the proportion of primary PM was>70%, which was relatively high compared to that in other regions.Based on this, the main cause of the error in PM 2.5 simulation for Pohang, Yeosu, Gwangyang, and Dangjin was determined to be the uncertainty of emissions (primary PM).And the major emission sources for the areas were analyzed reflecting this conclusion.
For the five regions where the simulated concentration was distinctively higher than the observed concentrations (Pohang, Yeosu, Gwangyang, Dangjin, and Ulsan), the manufacturing (first-level category)-others (secondlevel category), industrial process (first-level category)iron and steel making (second-level category), industrial process (first-level category)-petroleum industry (second-level category), and non-road transport (first-level category)-ships (second-level category) sectors were major air pollutant emission sources.Among them, four emission sources at the second-level category level accounted for 57% (NO X ), 78% (SO 2 ), and 88% (PM 2.5 ) of the total air pollutant emissions in the five regions.Major emission sources were slightly different by region.In Pohang, Gwangyang, and Dangjin, manufacturing (first-level category)-others (second-level category) and industrial process (first-level category)-iron and steel making (second-level category) were major emission sources.Meanwhile, in Yeosu and Ulsan, industrial process (first-level category)-petroleum industry (secondlevel category) and non-road transport (first-level category)-cars (second-level category) were major emission sources.In particular, manufacturing (first-level category)-others (second-level category) emission sources produces the large amounts of emissions of all the target air pollutants of this study (NO X , SO 2 , and PM 2.5 ).In detail, the emission source of manufacturing (first-level category)-others (second-level category)-primary metal industry (third-level category), in which non-public anthracite is used as fuel, represented>99% of the emissions from manufacturing (first-level category)-others (second-level category).Thus, to improve the accuracy of emissions, it is necessary to first examine the uncertainty that may occur in the process of estimating emissions from corresponding emission sources.The uncertainty ahead, however, does not mean the uncertainty of emissions from point sources.When it comes to point sources of large-scale places of business, errors in emissions are not likely to occur because their emissions are estimated on the basis of TMS data.NAIR estimates national air pollutant emissions and has identified the problems with the activity data and the process of estimating emissions from corresponding emission sources.Accordingly, NAIR is conducting research on the improvement of the emission estimation method and the results of improvement to address these problems.The details will be presented in a follow-up paper.
In summary, in this study, air quality modeling was conducted using CAPSS 2018 emissions, and the uncertainty of the current emissions was examined through comparison between observed and simulated concentrations.It was determined that emissions from five regions (Pohang, Yeosu, Gwangyang, Dangjin, and Ulsan) need to be improved.Most of all, it is necessary to examine the emissions form point sources using non-public anthracite as a fuel in manufacturing (firstlevel category)-others (second-level category)-primary metal industry (third-level category).
Emissions of primary PM 2.5 as well as PM 2.5 , SO X , VOCs, and NH 3 , which contribute to the formation of secondary PM 2.5 were assessed in this study.For PM 2.5 , SO X , VOCs, and NH 3 , Yeongnam region (33.5, 37.7, 33.3, and 25.9%, respectively) produced the largest amounts of emissions compared to other regions.Meanwhile, for NO X , the largest amounts of emissions occurred in SMA (27.9%).In SMA, the large amounts of PM 2.5 , VOCs, and NH 3 emissions were observed in the everyday activities and others sector (49.1, 74.2, and 85.7%, respectively), and the large amounts of SO X emissions were observed in the industry sector (42.6%), and the large amounts of NO X emissions were observed in the road sector (49.1%).In the Gangwon region, the large amounts of PM 2.5 , VOCs, and NH 3 emissions occurred in the everyday activities and others sector (55.2, 74.7, and 84.9%, respectively) and the large amounts of SO X and NO X emissions occurred in the industry sector (55.0 and 54.1%, respectively).In the Chungcheong region, the large amounts of PM 2.5 , SO X , and NO X emissions occurred in the industry sector (59.0, 68.0, and 34.7%, respectively) and the large amounts of VOCs and NH 3 emissions occurred in the everyday activities and others sector (64.2 and 83.5%, respectively).In the Honam region, the large amounts of PM 2.5 and SO X emissions occurred from the industry sector (49.3 and 77.3%, respectively), and the large amounts of NO X emissions occurred from the road sector (31.0%), and the large amounts of VOCs and NH 3 emissions occurred from the everyday activities and others sector (48.3 and 86.8%, respectively).In the Yeongnam region, large amounts of PM 2.5 and SO X emissions occurred from the industry sector (53.0 and 69.6%, respectively), and the large amounts of NO X emissions occurred from the road sector (37.1%), and the large amounts of VOCs and NH 3 emissions occurred from the everyday activities and others sector (64.5 and 77.6%, respectively).
The method of utilizing 3D chemical transport modeling was used to examine the uncertainty of national air pollutant emissions based on the NEAS.In this study, air quality was simulated using CAPSS 2018, and the uncertainty of the current emissions was examined through comparison between the simulated and observed concentrations.The results indicate that the proportion of primary PM in the simulated PM 2.5 concentrations was >70% for Pohang, Yeosu, Gwangyang, and Dangjin, which was relatively high compared to that for other areas.Based on this, the main cause of the errors in PM 2.5 simulation for Pohang, Yeosu, Gwangyang, and Dangjin was determined to be the uncertainty of emissions (primary PM) In addition, it is necessary to examine the emissions from places of business that use anthracite, a major emission source of PM 2.5 , as fuel in these si (cities).
To improve the uncertainty of air pollutant emissions, NAIR of Republic of Korea has been conducting research as follows: development of emission factors for facility using SRF (Solid Refuse Fuel), asphalt concrete manufacturing facility, SRU (Sulfur Recovery Unit), latest car models; improvement of activity data on anthracite consumption, traffic volumes of cars, vacant land, and barbecue grilling; identification of missing emission sources such as CHE (Cargo Handling Equipment), military equipment, GSE (Ground Support Equipment).
Based on these research efforts, NAIR aims to establish and implement air quality improvement policy, including highly effective PM reduction policies whose impacts can be felt by people, so that it can contribute to improve air quality and promote public health.

APPENDIX
in the method of counting the number of registered cars aged 10-15 years (integrated model year counting → individual model year counting) • (Before change) 10 to <15 years • (After change)<15 years, <14 years, <13 years, <12 years, and <11 years <Non-road transport> -(Construction equipment) An increase in the maximum car age subjected to the deterioration rate from 20-30 years -(Marine ships) An improvement in fuel consumption for passenger ships and fishing boats as well as subdivided criteria for the application of the sulfur content by oil type → Detailed classification of the fuels used for each ship <Agriculture> -An improvement in the method of counting livestock population • (Before change) Collect the latest information on the number of livestock population in as of fourth quarter • (After change) Collect the latest information on the number of livestock population as of the first, second, third, and fourth quarters

Fig. 10 .
Fig. 10.Spatial distributions of observed (circle) and simulated (tile) annual mean air pollutant concentrations.(a) NO 2 , (b) SO 2 , and (c) PM 2.5 concentrations in 2018 and the bias between them by measurement point.

Fig. 11 .
Fig. 11.(a) Simulated annual mean concentrations of PM 2.5 in 2018 and (b) the spatial distribution of primary PM 2.5 components' relative proportion.

Table 1 .
Improvements in the emission estimation method.

Table 4 .
Classification of administrative districts by region.

Table 6 .
Current Status of places of business and the number of registered cars and construction machinery by region in 2018

Table 7 .
Air pollutant emissions by administrative divisions in 2018.
*Sea: Air pollutant emissions from maritime transport such as ships and fishing boats