Introduction

Particulate matter (PM) is an important indicator of air quality, as its presence at high concentrations is detrimental to human health and the environment (Wang et al. 2000; Li et al. 2012; Vellingiri et al. 2015). Particulate matter in air can be operationally classified into several fractions based on aerodynamic diameter: e.g., total suspended particles (TSPs: aerodynamic diameter ≤ 100 μm), PM10 (≤10 μm), PM2.5 (≤2.5 μm), and PM1.0 (≤1.0 μm). Although individual sources of PM are diverse, they can be classified as either anthropogenic (e.g., fuel combustion (mainly coal), vehicle emissions, and tobacco smoke) or natural (volcanoes, surface dust, and sea salt) (Anderson et al. 2012; Kim et al. 2010). A large-scale dust storm, called Asian Dust (AD), which occurs periodically, is a well-known example of the latter (Ahmed et al. 2015). PM released at central Asia can travel long distances in the atmosphere to eastern Asia such as Korea, Japan, and Eastern China. According to the back-trajectory analysis, the influence of long-range sources such as AD is crucial for the considerable build-up of PM in spring at Seoul, South Korea (Nguyen et al. 2010; Vellingiri et al. 2015).

The presence of particulate matter pollution in the urban environment can be affected by a number of variables such as local geography, meteorology, wind, and atmospheric stability, among others (Tecer et al. 2008; Elminir 2005). In recognition of the damaging effects of PM, many countries have regulated PM emissions (along with many other pollutants). A large number of studies have demonstrated that pollution by PM has a detrimental impact on human health from inhibiting respiratory function to increasing human morbidity and mortality (Brook et al. 2010; Wang et al. 2000; Ye et al. 2003). To protect public health, air quality guideline values for key pollutants (PM, ozone, nitrogen dioxide, and sulfur dioxide) have been in place for many years (e.g., WHO 2006). As a long-term exposure guideline, the annual mean limits for PM10 and PM2.5 were set as 20 and 10 μg m−3, respectively (WHO 2006). Likewise, as a short-term exposure guideline, 24-h mean limits for PM10 and PM2.5 were set as 50 and 25 μg m−3, respectively (NIER 2013).

In this study, the concentrations of four PM fractions (TSP, PM10, PM2.5, and PM1.0) were measured at hourly intervals from two mountainous monitoring sites (Buk Han (BH) and Gwan AK (GA)) in Seoul, South Korea from January 2010 to December 2013. These results were compared to measurements at one ground-level reference site (Gwang Jin: GJ). PM concentrations in urban mountainous areas may also have an impact on human health due to the increasing number of people visiting these areas during their leisure time; however, few studies have investigated the importance of such aspect. The aim of our study was to assess the concentrations of particles in various size fractions using the data collected from the two elevated locations and to determine how they interact with various environmental parameters measured simultaneously at all monitoring sites.

Methods

Site characteristics

PM was measured at two mountainous sites: BH (37° 39′ 31.5″ N; 126° 58′ 39.72″ E) and GA (37° 25′ 44″ N; 126° 57′ 49″ E). Additionally, measurements were also made from one reference ground site at GJ (37° 32.40′ N, 127° 05.44′ E) in Seoul, South Korea (Fig. 1). The concentrations of TSP, PM10, PM2.5, and PM1.0, as well as other relevant pollutants (e.g., SO2, CO, O3, and NOX), were measured at these sites and subsequently analyzed. BH, at 837 m above sea level, is a major landmark visible from most districts of the Seoul metropolitan area. In comparison, GA, at 632 m above sea level, is a smaller mountain that lies at the border of Seoul and the neighboring cities of Anyang and Gwacheon. The ground level site, GJ, is located on the north side of the Han River and is close to two universities in Seoul (Konkuk and Sejong). Moreover, it includes a big bus terminal and amusement parks, often leading to heavy local traffic.

Fig. 1
figure 1

Map of the three sampling locations in Seoul, South Korea: (1) Buk Han (BH) Mt., (2) Gwan Ak (GA) Mt., and (3) Gwang Jin (ground site)

Full size image

Data acquisition

All PM data and relevant parameters were measured at hourly intervals at all three monitoring stations over a period of 4 years. Most of the analysis of PM was performed after converting the raw hourly data into daily averages (except for analyses of diurnal variation analysis and the frequency distribution of wind direction). However, in practice, some data were missing due to breakdown of the measurement system. A valid daily average was calculated if at least two thirds of the data were available for a given calendar day. All monitoring was conducted according to the operation guidelines for air quality monitoring stations managed by the Korean Ministry of Environment (KMOE: http://www.airkorea.or.kr/eng/).

The concentration data of PM10 and PM2.5 at each station were acquired initially at every 5 min using a β-ray absorption system (No FH 62 C14, Thermo Electron Co., US) at a flow rate of 16.67 L min−1 (±5 %) and detection limit of <1 μg m−3. For the measurements of SO2, NO2, O3, and CO, the following four systems (SA-633, NA-623, OA-683 (all three by KIMOTO, Japan), and ZRF instruments (FUJI, Japan), respectively) have been employed. A list of basic meteorological parameters (e.g., wind speed/direction, temperature/humidity, and solar radiation) have also been measured by WM05103 (R.M. Young, US), HMP45A (VAISALA Co., Finland), and LI200SZ (LI-COR Co., US), respectively. Detailed descriptions of the data acquisition system for the four PM fractions and other relevant parameters, sampling methods, and QA/QC parameters are detailed in our previous study (Vellingiri et al. 2015).

Results and discussion

Spatial distribution of PM

A statistical summary of the PM data for all four fractions (TSP, PM10, PM2.5, and PM1.0) measured at the three study sites (BH, GA, and GJ) from 2010 to 2013 is provided in Table 1. The daily mean concentrations of PM10 at BH, GA, and GJ during the study period were 37.0 ± 24.6, 34.5 ± 24.4, and 45.6 ± 28.0 μg m−3, respectively. The PM2.5 concentrations at these three sites were 20.6 ± 12.9, 21.5 ± 14.1, and 24.8 ± 15.1 μg m−3, respectively. The mean concentrations of PM10 and PM2.5 at the three sites exceeded the standard guidelines for ambient air set by the WHO (20 and 10 μg m−3, respectively). In the case of PM10, the values exceeded the limit by 72.5 to 128 %, while they exceeded the limits of PM2.5 by 106 to 148 %. However, the measured values did not exceed the less stringent annual guideline values set by the Korean Ministry of Environment (KMOE) for PM10 and PM2.5 of 50 and 25 μg m−3, respectively. The mean concentrations of PM1.0 during the study period were 15.3 ± 9.53, 14.8 ± 8.89, and 17.5 ± 10.6 μg m−3, respectively.

Table 1 Statistical summary of the concentrations of the four PM fractions (TSP, PM10, PM2.5, and PM1.0) and other relevant parameters using the daily mean values measured at Buk Han Mountain (BH), Gwan AK Mountain (GA), and Gwang Jin (GJ) in Seoul, South Korea between 2010 and 2013
Full size table

The average concentrations of all four PM fractions measured at GJ were greater than those found at the mountainous sites, with average concentrations decreasing in the order of GJ > BH > GA (Fig. 2). The average concentrations of PM1.0 were similar among the three sites, while the differences between the concentrations of the other fractions increased with increasing diameter of the PM component. For instance, the mean concentration of PM2.5 at GA was slightly higher (by 4.4 %) than that at BH.

Fig. 2
figure 2

Comparison of the mean concentrations of a TSP, PM10, PM2.5, and PM1.0 (μg·m−3); b SO2, and O3 (ppb); c NO, NO2, and NOX (ppb); and d CO (ppb): results are compared between BH, GA, and GJ (using daily data) from 2010 to 2013 (error bars denote SD)

Full size image

The frequency distribution of the PM data is shown in Fig. 1S. The PM10 data had the highest frequency at between 20∼30 μg m−3 at the BH and GA sites, while it was 30∼40 μg m−3 at the ground level site. In contrast, the patterns for PM2.5 were almost identical, with maximum frequencies of 10∼20 μg m−3 at all three sites. TSP (between 40 and 50 μg m−3) and PM1.0 (between 10 and 20 μg m−3) were also more concentrated at the GJ site than at the BH site by 10 μg m−3. The maximum frequency of measured mass concentrations of the PM fractions was in the rank order of PM1.0< TSP < PM2.5 < PM10 at all three study sites. The maximum frequency patterns of the two fine PM fractions, PM1.0 and PM2.5, were highly comparable with each other, regardless of site.

The seasonal mean values of wind speed ranged from 0.89 ± 0.16 to 1.16 ± 0.16 m s−1 at BH and from 1.90 ± 0.17 to 2.38 ± 0.27 m s−1 at GJ. In contrast, the mean values were 3.27 ± 0.91 to 4.24 ± 1.25 m s−1 at GA; these values are three times larger than those measured at BH and approximately two times larger than those measured at GJ. All three sites showed the maximum and minimum wind speeds during spring and fall, respectively. Figure 2S shows the frequency distribution of wind direction for BH, GA, and GJ, computed using hourly datasets. The relative frequency of wind direction at BH was dominated by N (21.1 %) and NNE (19.1 %), while that at GA was predominantly from NE (15.2 %). In contrast, winds blowing at GJ occurred most frequently from W (27.1 %), ENE (22.1 %), and E (16.0 %). High concentrations of the four PM fractions were found when the wind was blowing from the west in spring (35.5 %) and winter (33.8 %) at GJ, which may reflect the influence of the AD event (Hong et al. 2010; Oh et al. 2015).

Temporal variation in PM

A statistical summary of the four PM fractions and other relevant parameters from the three sites are presented in Table 1S. The yearly mean concentrations of the four PM fractions were similar across the 4 years at each monitoring site. Nonetheless, there were significant seasonal variations at all three sites (Fig. 3S). The spring and winter values for all four PM fractions were significantly higher than the summer and fall values at all three sites (e.g., TSP levels recorded during spring and winter were nearly 88 % higher than those observed during fall and summer) (Table 2S). Moreover, the mean concentrations of PM10 and PM2.5 measured at GJ in spring and winter exceeded the KMOE guideline values by up to 27.6 %. This threshold was not exceeded at the two mountainous sites. These results indicate the influence of intense precipitation events in summer washing PM out of the atmosphere, while the effect of AD events was reflected in increased concentrations of PM in the spring and winter (Kim and Kim 2003; Kim et al. 2007).

Figure 3 shows the temporal variation in monthly mean of the four PM fractions at the three sites. The monthly concentrations of the four target pollutants at all three study sites were higher in 2010 and 2011 than 2012 and 2013 (no values for TSP at GA were available). The concentrations of the four PM fractions measured between late 2010 and early 2011 were noticeably higher, probably due to strong AD events than those measured over the rest of the period. The influence of AD event on the PM fractionation was observed previously (e.g., Ahmed et al. 2015; Kim et al. 2016).

Fig. 3
figure 3

Comparison of monthly mean concentrations of TSP, PM10, PM2.5, and PM1.0 (μg m−3) measured at a BH, b GA, and c GJ (using daily data) between 2010 and 2014

Full size image

Figure 4a shows the diurnal variation for all four PM size fractions at the three study sites. The mean hourly variations of TSP and PM10 were very similar at all three sites. However, the mean hourly values of PM2.5 and PM1.0 decreased in the early morning (12 a.m. to 8 a.m.) and subsequently increased suddenly at the two mountainous sites. These concentrations were relatively stable at the GJ site. Noticeably, the mean hourly values of PM2.5 and PM1.0 at BH and GA exceeded those at GJ between 2 and 10 PM. Figure 4b presents the average daily patterns of the four PM fractions at all three sites (TSP data not available at GA). In most cases, an abrupt increase was found after 8 a.m. The lack of difference in diurnal patterns between mountainous sites and the reference site suggests that city-scale meteorological parameters and widespread diffuse emission sources rather than local PM emission sources affected PM concentrations at all these sites.

Fig. 4
figure 4

Diurnal variation of the four PM fractions between BH, GA, and GJ (hourly data). a Comparison between the different PM fractions (no data available for TSP at GA from 2010 to 2013). b Comparison between the three different sites

Full size image

Correlation of PM and other relevant parameters among the three sites

Table 2a presents the results of a correlation analysis between the PM concentrations of the four different fractions and other pollutants at each of the three study sites. The observed correlations were generally high at all three sites (e.g., r values were 0.40 to 0.66 for SO2 and 0.48 to 0.68 for CO; p < 0.01). This indicates that most of the pollutants measured from all three sites should share some communality (e.g., fossil fuel burning, forest fires). The findings of generally enhanced ozone levels at the two mountain sites as well as the presence of stronger correlations between NO2 and PM2.5 at BH (r = 0.523; p < 0.01) and GA (r = 0.484; p < 0.01) are indicative of the more direct influence of manmade sources (e.g., vehicle emissions) at elevated sites (Chen et al. 2014). The effects of such manmade sources are apparent at our reference site of GJ, which is also comparable to our previous work made at Yongsan, Seoul (Vellingiri et al. 2015).

Table 2 Results of the correlation analysis of TSP, PM10, PM2.5, and PM1.0 with chemical pollutants and meteorological parameters
Full size table

Table 2b summarizes the calculated correlation coefficients between the four PM fractions and meteorological parameters. The results were generally similar across all the three study sites. PM concentrations are expected to be higher at low temperatures, because lower temperatures are generally accompanied by poorer atmospheric dispersion conditions (Kim et al. 2016; Oh et al. 2015). This effect was observed consistently for all four PM fractions in both spring and winter. Comparatively weak correlations between humidity and PM concentrations were noticed at BH (r values up to −0.205; p < 0.01), GA (r value up to −0.313; p < 0.01), and GJ (r values up to −0.270; p < 0.01). This indicates that the higher humidity and associated precipitation lowered the observed concentration of PM at the two mountain sites because of atmospheric washout. The correlation between PM and wind speed was relatively weak. If there was one local dominant point source, one would expect a strong inverse correlation between PM and wind speed; this is explained by the greater dispersion of PM as wind increases (Jones et al. 2010; Lianou et al. 2013). The poor correlation observed suggests the role of a larger-scale, area source emission dominating the observed concentrations at the monitoring stations, such as would be found from a large city like Seoul. This hypothesis is supported by the data shown in Table 3S, which shows the results of a correlation analysis for each of the four PM fractions at each study site. The very high correlation (r ≥ 0.966; p < 0.01 for PM10 between BH and GJ) for all four PM fractions indicates that the three study sites were affected by the same large source.

Comparison with other studies

Table 3 compares the results of this study with those obtained in other relevant studies. Due to a lack of data from mountainous sites, we compared our results with the concentrations reported from urban roadside and urban background or suburban areas in Europe and Asia. The concentrations at GJ (TSP (61.1 μg m−3), PM10 (45.6 μg m−3), PM2.5 (24.8 μg m−3), and PM1.0 (17.5 μg m−3)) were higher than those observed in most European studies (with the exceptions of Zonguldak, Turkey and Madrid, Spain) but half the levels reported in Asian counties. Ahmed et al. (2015) reported similar values (TSP (75.3 μg m−3), PM10 (54.0 μg m−3), and PM2.5 (26.6 μg m−3)) in Yongsan, Seoul, South Korea between 2004 and 2013. This indicates that the PM concentrations in Seoul are much lower than in other Asian megacities (e.g., Beijing, Hong Kong); this is in part because of the strategic abatement programs setup by KMOE. However, concentrations at the two mountain sites were higher than those reported in European countries and approximately the same as the concentrations recorded in Taipei, Taiwan, and Yokohama, Japan (Gugamsetty et al. 2012; Khan et al. 2010). In contrast, they were much lower than those reported in other urban background studies in Seoul (Shin Chon and Dae Hung) and Hong Kong. Notably, Kim and Kim (2003) reported 50.7 μg m−3 (PM2.5), 107 μg m−3 (PM10), and 145 μg m−3 (TSP) at a mountainous site located in the center of Seoul, which is approximately two times higher than the corresponding concentrations measured at BH and GA in this study. This was mainly due to local human activities and heavy local traffic in the mountain site evaluated in the previous study compared to the mountain sites investigated in our study.

Table 3 Comparison of the concentrations (μg m−3) of PM measured in this study with those reported in previous studies
Full size table

Conclusions

Concentrations of four PM fractions (TSP, PM10, PM2.5, and PM1.0) were measured at two mountainous sites and one reference ground site in Seoul, South Korea from 2010 to 2013. The concentrations of the four PM fractions at the two mountainous sites were much lower than those measured at the reference site. Seasonal variations were large at all three study sites, and PM concentrations were much higher in spring and winter than in summer and autumn, regardless of site. In addition, PM concentrations were higher in both spring (35.5 %) and winter (33.8 %) at GJ when the wind was from the west. This may reflect the influence of the monsoon in summer washing PM out of the atmosphere, while the effects of AD events dominated in late winter to spring. A similar pattern of diurnal variation was observed at all three sites, suggesting a single large area source of PM, such as Seoul, at all three sites. Diurnal variations in PM2.5 and PM1.0 at the two mountainous sites were less pronounced than at the reference site in the early morning. The lack of a significant difference in diurnal patterns when comparing the mountain sites and the reference site suggests that city-scale meteorological parameters and emissions, rather than local emission sources, are the dominant determinants of PM concentrations in these areas. The enhanced correlations between NO2 and PM2.5 at the reference site indicate the influence of traffic at that site. The PM concentrations at the two mountainous sites in this study were similar to those measured at an urban background site in Seoul but still much higher than those in most European countries.