Sangjun Choi1, Ju-Hyun Park2, Seo-Yeon Bae3, So-Yeon Kim3, Hyaejeong Byun4, Hyunseok Kwak5, Sungho Hwang6, Jihoon Park7, Hyunhee Park8, Kyong-Hui Lee9, Won Kim10, Dong-Uk Park 3

Department of Occupational Health, Daegu Catholic University, Gyeongsangbuk-do 38430, Korea
Department of Statistics, Dongguk University, Seoul 04620, Korea
Department of Environmental Health, Korea National Open University, Seoul 03087, Korea
Samsung SDS Co., Ltd., Seoul 05510, Korea
Occupational Lung Diseases Institute, Korea Workers’ Compensation and Welfare Service, Incheon 21417, Korea
National Cancer Control Institute, National Cancer Center, Goyang 10408, Korea
Environmental Safety Group, Korea Institute of Science and Technology Europe Forschungsgesellschaft mbH, 66123 Saarbrücken, Germany
Occupational Safety and Health Research Institute, Ulsan 44429, Korea
Force Health Protection and Preventive Medicine, US Army MEDDAC-Korea, Unit 15281, APO AP 96205-5281, USA
10 Wonjin Institute of Occupational and Environmental Health, Seoul 02221, Korea

Received: May 22, 2019
Revised: August 19, 2019
Accepted: October 9, 2019
Download Citation: ||  

Cite this article:

Choi, S., Park, J.H., Bae, S.Y., Kim, S.Y., Byun, H., Kwak, H., Hwang, S., Park, J., Park, H., Lee, K.H., Kim, W. and Park, D.U. (2019). Characteristics of PM10 Levels Monitored for More than a Decade in Subway Stations in South Korea. Aerosol Air Qual. Res. 19: 2746-2756.


  • The average PM10 levels decreased by year in all stations and city.
  • The PM10 levels were far higher than the yearly average ambient air quality.
  • Some of subway characteristics were found to influence the PM10 level.
  • Platform screen doors, number of transfer lines were factors influencing the PM10.


This study aimed to evaluate the variation in PM10 concentration and identify the factors influencing it in Korean subways during the past decade. The PM10 measured internally by subway companies according to legal requirements was categorized by the subway’s characteristics, which were statistically examined using a mixed effects model to identify the relevant parameters. The average levels monitored near or on the platforms and in the waiting rooms ranged from 53.9 to 92.4 µg m−3, remaining below the Indoor Air Quality Control Act regulatory standard of 150 µg m−3. However, the levels monitored on the platforms far exceeded the average yearly atmospheric environmental standard (50 µg m−3). Based on both univariate and multiple analyses, several subway characteristics, including the presence of a platform screen door (PSD), were found to significantly correlate with the concentration, although slight differences in the significant factors were detected between the cities. Particularly, the absence of transfer lines and the presence of a PSD reduced the platform concentration, except at Busan and during specific years.

Keywords: Subway; PM10; Platform screen door (PSD); Indoor air quality.


Subway systems are the most used public transportation service in South Korea. Subway lines have been expanding continuously since their inception in 1974. Physically, the underground portion of the subway system is a semi-confined environment that may accumulate either internally generated contaminants or those from the outside environment. Proper mechanical ventilation is vital to this situation; otherwise, contaminants may accumulate to a severely harmful level 

(Nieuwenhuijsen et al., 2007). The level of efficiency of the ventilation system varies among subways and according to their year of construction. In general, it has been reported that subway users are likely to be exposed to higher levels of particulate matter (PM) than the outdoor concentration (Kamani et al., 2014; Ramos et al., 2015).

In South Korea, the Indoor Air Quality Control Act (IAQ Act) was first established in 1996 as the Underground Living Space Air Quality Control Act (KMOE, 1996). In 1998, under the IAQ Act, the 24-hour average indoor air quality standard (IAQ standard) for PM10 (defined as particulate matter with an aerodynamic diameter equal to or less than 10 µm) was first set as 250 µg m−3. It was revised to 200 µg m−3 in 2000 and finally to 150 µg m−3 in 2002 (KMOE, 1998). Since 2005, only PM10 in subway stations has been required to be monitored once per year and reported to the Korean Ministry of Environment (KMOE) mandatorily (KMOE, 2004).

No comprehensive studies have been conducted to assess variation in hazardous pollutants, including PM10, that may likely be associated with commuters’ health. The annual variations in PM10 have never been reported. This study aimed to assess the variation in PM10 over the past decade in South Korea and identify subway characteristics influencing the PM10 level.


General Information about Subway Systems in South Korea

A subway transportation system has been fully established in five metropolitan cities in South Korea, including in Seoul. General information on subway systems is shown in Table 1. This information includes the first and last year of construction, the number of lines, and the number of stations covered. A total of 34 lines are currently operating nationwide. About 14 million commuters use the subway every day nationwide. The number of people who use the subway daily are compared among the cities as of the end of 2017 along with the increase by year.

Table 1. General information on the subway system in major metropolitan cities.

Data Collection

According to the IAQ Act, all subway corporations in South Korea are required to monitor five pollutants, including PM10, once per year, and to report the measurements of these to KMOE; a history of recorded measurements is also kept. All PM10 measurements recorded in 13 subway corporations that operate in 7 large cities across Korea (Table 1) were collected and analyzed based on this study strategy. We asked each subway company to report their monitored PM10 measurements from 2005 through 2017 according to both year and location; a total number of 12,174 PM10 measurement data were collected from 2005 to 2017. Among them, 570 measurement data with missing values of concentration were excluded. In addition, we excluded 356 measurement data from subway cabins, driving rooms, and tunnels. In total, 11,248 measurement data taken from platforms, waiting rooms, or transfer passageways were considered to be valid and were consequently selected for our analysis.

The PM10 sampling and analytical methods that subway companies have to use are standard, as designated by the Korean National Institute of Environmental Research (NIER) (NIER, 2017). According to the primary standard method, PM10 samples are collected using nitrocellulose membrane filters with an air sampling pump operated from 1 to 30 L min1 and should be analyzed using a gravimetric method. All samples were monitored at specific locations within the subways, including on station platforms, in concourses, or in transfer passageways for longer than 6 hours. Although PM10 can be monitored by a secondary standard method using a beta attenuation monitor, over 90% of samples were measured by the primary standard method. In addition, outdoor PM10 levels monitored in each city during the same year were also collected and compared with levels in the subways.

Data Analysis

A total number of PM10 measurement data in various subway environments (n = 11,248) were categorized according to the following variables used as an independent analysis unit.

  • The year measured was categorized into one of four groups: 2005–2008, 2009–2011, 2012–2014, or 2015–2017
  • The area measured: platform, waiting rooms, or transfer passageways
  • The presence of a transfer line: yes or no
  • The number of transfer lines: none, 2, or > 2
  • The presence of a platform screen door (PSD): yes or no (screen doors were established to isolate the platform from the railway and ensure the safety of passengers, and the year of establishment varies among the subway stations)
  • The season measured: spring (March–May), summer (June–August), autumn (September–November), or winter (December–February)
  • The age of the station: < 5 years, 5–10 years, or > 10 years

We statistically examined both the environmental and subway characteristics significantly influencing the PM10 level. All PM10 measurements monitored were summarized using the following descriptive statistics: arithmetic mean (AM) and standard deviation (SD) with 95% confidence interval (CI); geometric mean (GM) and geometric standard deviation (GSD) with 95% CI; range; and quartiles. Box plots were used to show the distribution of measurements by subway characteristics, such as region and location monitored. Since PM10 was monitored in each subway station every year and the measurements were thus correlated, both univariate and multiple linear mixed models were implemented to ascertain through a likelihood ratio test whether experimental factors such as region, location, and subway characteristics have a significant influence on the level of mean PM10 while taking into account the correlation. All the statistical analyses were conducted using R software (version 3.5.1; R Foundation for Statistical Computing, Vienna, Austria).


The PM10 levels monitored in subway platforms, waiting rooms, and transfer passageways are compared by year (Fig. 1), city (Table 2), and subway and environmental characteristics (Table 3). Average PM10 values are found to decrease slightly with year, which was consistently detected in all stations and regions. Average PM10 levels monitored near subway platforms or waiting rooms ranged from 53.9 to 92.4 µg m−3, and all levels were below 150 µg m−3, the IAQ standard regulated by the IAQ Act. There have been 14 stations (9 in Seoul in 2007, 2 in Seoul in 2008, and 3 in Daegu in 2011) exceeding 150 µg m−3. These are located above the dashed line in the box plot (Fig. 1). 

Fig. 1. Distribution of PM10 levels by year. A total of 14 stations (9 in Seoul in 2007, 2 in Seoul in 2008, and 3 in Daegu in 2011) exceeded the indoor air quality standard of 150 µg m–3 indicated by the dashed line.Fig. 1. Distribution of PM10 levels by year. A total of 14 stations (9 in Seoul in 2007, 2 in Seoul in 2008, and 3 in Daegu in 2011) exceeded the indoor air quality standard of 150 µg m–3 indicated by the dashed line.

Table 2. PM10 levels (µg m–3) measured over the study period in each city.

Table 3. Univariate analysis examining relationships between PM10 level and subway characteristics.

When comparing the PM10 levels on the platform by city, Seoul (AM = 91.9 µg m−3, max = 170.2 µg m−3) and Daegu (AM = 92.4 µg m−3, max = 153.1 µg m−3) were the highest, and Busan (AM = 56.6 µg m−3, max = 128.6 µg m−3) was the lowest (Table 2). Based on both univariate analysis and mixed effects multiple analysis, several subway characteristics, including the presence of PSDs, were found to be significantly associated with PM10 level (Tables 3 and 4), even though there is a little difference in significant factors among cities. We indicated the univariate and multiple analysis results from Seoul at Table 4 and results from other regions can be found in Tables S1–S5 of the supplement. In terms of Seoul and the surrounding metropolitan area, the PM10 concentration around the subway platform and in the waiting rooms was significantly high for Subway Line 1 when monitored from 2005 to 2007 when there was no PSD. In particular, the presence of PSDs was found to contribute to a significant reduction in PM10 levels on the platform (Fig. 2), with the exception of in Busan (Fig. 2(e)) and in certain cities for specific years (2015–2016 in Gwangju). The PM10 levels in the subway were found to be significantly higher than those monitored outdoors, regardless of city or year (Fig. 3). 

Fig. 2. Comparison of distribution of PM10 levels (µg m–3) on platforms by presence of screen door and year.Fig. 2. Comparison of distribution of PM10 levels (µg m–3) on platforms by presence of screen door and year.

Fig. 3. Comparison of distribution of yearly average PM10 levels (µg m–3) monitored in subway stations and outdoor air.Fig. 3. Comparison of distribution of yearly average PM10 levels (µg m–3) monitored in subway stations and outdoor air.


We analyzed variation in PM10 concentrations over more than a decade in subways in Korea and found several characteristics influencing PM10 level. The yearly average of PM10 has fallen to 65.9 µg m−3 in 2017 from a peak of 102.4 µg m−3 prior to 2010, a 35.6% decrease. This trend is observed in all cities’ subways, although the levels of decrease differ. Our results are far below the results reported by several studies conducted in Seoul. Park and Ha reported that PM10 levels inside Subway Lines 1, 2, and 4 exceeded the Korean IAQ standard of 150 µg m−3. Their average PM10 concentration as monitored inside trains (144.0 µg m−3) was far higher than the 125.8 µg m−3 on platforms (p = 0.026) and the concentration range (35–81 µg m−3) measured in outdoor air in Seoul from January to November in 2004 (SAMC, 2004). The oldest line, Line 1 in Seoul, showed concentration levels in 10 of 12 of its investigated stations that exceeded the IAQ standard for PM10. The highest monitored concentration was 207.5 µg m−3 inside an underground station on Line 1.

Regardless of country, city, or location within the system, it has been characteristic for high concentrations of PM to be measured in subway systems, such as in London (Adams et al., 2001), Stockholm (Johansson and Johansson, 2003), Prague (Braniš, 2006), Rome (Ripanucci et al., 2006), Berlin (Fromme et al., 1998), Seoul (Kim et al., 2008), and Beijing (Li et al., 2007). Levels of PM10, PM2.5, and nanoparticles in the subway environment have all been reported to be far higher than those monitored in the outside environment. When the average annual atmospheric concentration of PM10 (KMOE, 2017) is compared with annual average concentrations measured in the subway, we also found the subway concentration to be higher than the atmospheric concentrations in all cities (Fig. 3). These results demonstrate that the amount of fine and ultrafine dust absorbed into the respiratory system in subway systems can generally be far higher than the amount from outdoors based on both exposure time and exposure level.

Generalizing factors that may influence PM10 levels measured under specific circumstances are very difficult to specify because of the subway characteristics, surrounding environments, and environment measured (i.e., the types of subway, location measured, age of subway, number of subway users). We found that several subway characteristics significantly influence the level of PM10 in subway stations. Year was found to be significantly associated with change in PM10 level. The level of PM10 reduces markedly with year, something statistically detected for all cities and all stations (Fig. 1 and Tables 3 and 4). However, the PM10 levels measured in subway stations are still far above the Korean atmospheric environmental standard for PM10 (yearly average: 50 µg m−3; daily average: 100 µg m−3) intended to protect the general public, including children and elderly people, even though other pollutants such as ozone and nitrogen dioxide (NO2) are not substantially different (KMOE, 2018a). The Korean Ministry of Environment also adopted a PM2.5 standard for the first time in 2015 (yearly average: 25 µg m−3; daily average: 50 µg m−3) and strengthened the standard in March 2018 (yearly average: 15 µg m−3; daily average: 35 µg m−3) (KMOE, 2018b). However, IAQ regulations lacked a standard for PM2.5, even though it accounts for most of the PM10 generated in the subway environment. IAQ standards for PM2.5 in subway stations are scheduled to be set as a daily average of 50 µg m−3 for the first time from 2019 (KMOE, 2018c). Many countries’ national health organizations and influential global organizations such as the World Health Organization (WHO) have stipulated a standard or guideline value for indoor hazardous pollutants (Abdul-Wahab et al., 2015). Air quality standards have been adopted as measures enforceable by a regulatory authority, including in Korea, Taiwan, and Japan. On the other hand, air quality guidelines are designed to offer guidance for reducing adverse health impacts from air pollution, and many countries such as the USA suggest their IAQ values only as guidelines. Generally, IAQ standards except for PM10 are set to be similar to outdoor atmospheric standards (Vahlsing and Smith, 2012). Many people use the subway, including not only adults but also pollutant-sensitive groups such as children, medical patients, elderly people, and pregnant women. Also, people who routinely use a subway system for commuting can be exposed to the air in the subway for much longer periods than to outdoor air. This indicates that IAQ standards for PM10 and PM2.5 should be modified to match atmospheric standards.

In general, subway stations with transfer lines have more PM sources compared with stations without transfer lines, including a greater number of passengers, number of entrances and exits, and frequency of maintenance work. In addition, the passageways of transit stations are often connected to underground shopping areas, and thus, the possibility of the inflow of pollutants from the outside is greater. Therefore, transfer stations should be managed first to reduce fine dust concentrations. This study recommends the installation of boards that display real-time levels of PM10 and PM2.5 on some subway platforms on transfer lines so that citizens can be aware of the quality of subway air.

Several mitigation measures have been developed to reduce PM concentrations in subway systems; Korean researchers conducted most of the evaluations in this field. PSDs were recognized as one of the most efficient measures to improve underground air quality in subways. The average PM10 concentration measured on the platforms after the installation of PSDs significantly reduced by 16% (Kim et al., 2012) and 38% (Han et al., 2014), respectively, compared to the earlier period. In this study, the presence of a PSD was also found to be a significant factor in reducing PM10 levels on platforms (Fig. 2); this is considered true despite the city of Busan (Fig. 2(e)) and specific years not following this trend. However, the mean PM10 concentration measured inside trains after the installation of PSDs increased significantly by 29.9% compared to the concentration before the installation (Son et al., 2014). Son et al. (2014) suggest that air mixing between the platform and the tunnel was extremely restricted after the installation of the PSDs. Kown et al. (2016) investigated the change of PM size distribution in an underground station with PSDs; their results showed that the PM that was suspended in the tunnel flowed into the platform area even in a subway station where the effect of train-induced wind was blocked by the installed PSDs, as this flow occurred when the PSDs were opened. Despite the installation of completely sealed PSDs, the inflow of coarse mode particles from the tunnel seems unavoidable, indicating the need for measures to decrease the generated PM in order to lower subway user exposure.

Mechanical ventilation has been recognized as a key factor affecting indoor air quality in subway systems. Juraeva et al. (2016) experimentally investigated the effects of the train wind, the air curtain, and electric precipitators as well as the proper conditions for electric precipitator operation to decrease the PM concentration. Their results indicated that the average velocity of the airflow in the shaft increased when the velocity of the air curtain increased. The PM concentration after ventilation was reduced significantly in the tunnel when the air curtain and train wind were operated. Station design is also related to the influence of tunnel ventilation and the train piston effect. The effects of ventilation conditions and station design on underground air quality were investigated in the Barcelona subway system. Narrow platforms served by single-track tunnels were dependent on forced tunnel ventilation and could not rely on the train piston effect alone to reduce platform PM concentrations. The PM concentrations of stations with spacious double-track tunnels were not significantly affected when tunnel ventilation was switched off (Moreno et al., 2014). To reduce PM concentration in subway cabins, the subway cabin air purifier (SCAP) was developed and evaluated for its effectiveness (Kim et al., 2014); it was found that the PM10 concentrations inside cabins were reduced by 15.5–26.0% after the SCAP system was installed.

In order to improve indoor air quality in subways, it is important to identify the main sources of air pollution. Several studies in identifying the chemical composition of subway PM demonstrated that Fe was the most abundant metal element of PM2.5 (Aarnio et al., 2005; Loxham et al., 2013; Lee et al., 2018; Minguillón et al., 2018) and PM10 (Jung et al., 2010; Park et al., 2012; Loxham et al., 2013; Moreno et al., 2015; Lee et al., 2018), accounting for 30% and 80% of PM content, respectively. Jung et al. (2010) clearly identified indoor sources of subway PM in comparing four sets of samples collected in tunnels, on platforms, near ticket offices, and outdoors. Fe-containing particles predominated in the samples collected in tunnels, with relative abundances of 75–91% for the four stations. In addition, the amount of Fe-containing particles decreased as the distance of sampling locations from the tunnel increased. These results clearly indicated that Fe-containing subway particles were generated in the tunnel. Park et al. (2012) characterized PM10 sources by positive matrix factorization; railroad-related sources such as the abrasion of the railroad tracks, brakes, and power supply or draft lines during subway operation contributed the most PM10 to subway cabin air. Studies in both New York City (Vilcassim et al., 2014) and Shanghai (Guo et al., 2017) indicated that a potential source of fine particles in subways was the diesel engine cleaning and maintenance vehicles that operated during the night in the underground facilities. Recently, Choi et al. (2019) also identified that the use of diesel engine vehicles in tunnel maintenance was a key contributor to both PM2.5 and black carbon (BC) exposure levels among subway workers. The use of diesel engine vehicles in semi-confined underground environments causes not only exposure to high levels of diesel engine exhaust emissions but also an increase in PM2.5 on subway platforms or in waiting rooms. Therefore, proactive measures, including the installation of diesel particulate filters (DPFs) on diesel engine maintenance vehicles in tunnels, are urgently suggested in order to reduce subway workers’ exposure to both PM2.5 and BC. In addition, after the enforcement of the EURO engine emission standard, NO2 emissions for recent diesel engines are becoming a significant concern in diesel engine exhaust (Grice et al., 2009; Carslaw and Rhys-Tyler, 2013). Electric-battery equipped vehicles, which would be effective in reducing the levels of airborne particles and NO2, should be introduced to improve air quality in subways. In particular, diesel vehicles without diesel exhaust-reducing air treatment systems should be phased out of use in subways.

This study has several limitations. One major limitation is that it is not possible to know how representative these findings obtained from various locations are with regard to subway characteristics that involve various types of physical environments and ventilation levels. Our PM10 measurements were compilations of data measured once in a specific area and on a specific day during the year, which may likely be affected by not only subway characteristics but by outdoor conditions as well. The number of passengers, which likely is associated with the level of PM10, was not examined in this study. It is impossible to obtain the number of passengers who used the stations where PM10 levels were measured. In addition, this study did not examine the impact on PM10 by engineering control measures designed to reduce the infiltration of air pollutants in subways, including fine and ultrafine particles, from the outdoor environment, which is likely one of the factors increasing the level of fine particles in a subway. The facilities for supplying outdoor air were all found to be installed on a street at the same level as a road bearing traffic. The air cleaners are not able to remove fine particles exhausted from vehicles and the outdoor environment. In addition, the non-designed data collection resulted in some potential environmental factors partially crossing, implying that not all the effects of these influential factors were fully estimated and tested within the linear mixed models. For example, all of the PM10 data for Daejeon for the years 2012–2014 were collected in the autumn, and therefore, it was impossible to separate the effect of autumn from that of the years 2012–2014. Although we successfully took into account the correlation structure among PM10 measurements due to their being measured over time at several locations within a station through the use of a linear mixed model, it is assumed that spatial correlations among adjacent stations were not large enough to be considered under the condition that the ventilation system in a tunnel connecting any two stations worked well. This assumption needs to be investigated thoroughly in a future study.

Nevertheless, our results are useful not only for characterizing the level of PM10 in the subway environment but also for identifying specific factors that may significantly influence PM10 concentrations and for recommending mitigation procedures. The general variations in PM10 levels over more than a decade were also characterized. Based on our results, a number of appropriate engineering, administrative, and regulatory measures could be taken to reduce exposure to fine and ultrafine PM in subway stations. Further study is required to monitor PM2.5 and the diesel engine exhaust concentrations in subway air considering several subway characteristics including maintenance at tunnel.

In conclusion, even though the PM10 concentrations in subways have been decreasing over time, they are still higher than outdoor levels and far exceed the yearly atmospheric environmental standard (50 µg m−3). We found that several factors, including the specific year, the location of the station, and the presence of a PSD, significantly influenced the amount of PM10.


This study was supported by the 2018 Extramural Research Fund of the Occupational Safety and Health Research Institute, KOSHA (2018-OSHRI-796).


All authors declare there is no financial/personal interest or belief that could affect their objectivity.

Don't forget to share this article 


Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

Latest coronavirus research from Aerosol and Air Quality Research

2018 Impact Factor: 2.735

5-Year Impact Factor: 2.827

SCImago Journal & Country Rank

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal, promotes submissions of high-quality research, and strives to be one of the leading aerosol and air quality open-access journals in the world.