Journal Archive

Particle-phase concentrations were measured at the roadside of National Highway 45, which is located about 35 km southeast of downtown Seoul and runs in front of the Global Campus of HUFS (see Fig. 1). This site was selected because: (1) there are no other thoroughfares except the access road to HUFS to the south; (2) there are no major emission sources except vehicle emissions from the roads; and (3) the surrounding areas have relatively flat terrain, where small-sized buildings, farmlands, and open spaces are scattered. In fact, fugitive dust from nearby farmlands or vacant lots also could become an important emission source. However, it was thought that its effect should be examined separately if fugitive dust is another key factor in determining particulate matter concentrations in Korea (Korean Society for Atmospheric Environment, 2009).

Fig. 1. 
Locations of the Global Campus Hankuk University of Foreign Studies (HUFS) in Korea, National Highway 45, and the measurement site. Low-volume air samplers were deployed at distances of 10, 80, and 200 m from the outer lane of two lanes each way.

National Highway 45 around the measurement site has two lanes in each direction. Its average daily traffic volume of 1,070 vehicles is not as high as that for highways or thoroughfares of the city center, and it remains constant, except for a decrease in the middle of the night or weekend (Korea Ministry of Construction and Transportation, 2006). Passenger cars and light trucks on National Highway 45 account for the majority of the total traffic volume (about 95%), while buses and heavy trucks account for about 3.1% and 0.2%, respectively.

Measurements were carried out during two different periods - January 4 through 24, 2008, and May 11 through 29, 2008 (Table 1). The former period was chosen to represent the winter with high combustion-related emissions, while the latter was chosen for comparison with winter emissions. Three measurement locations were established at 10, 80, and 200 m from the northbound outer lane of the highway. The maximum distance of 200 m was chosen because it has been reported that direct influence of on-road emissions becomes insignificant beyond a distance of about 200 to 300 m (Zhu et al., 2002a, b; Roorda-Knape et al., 1998). Wind direction and speed were measured at a height of about 1.8 m using the automatic weather station (AWS; Campbell 05103-L R.M. Young Wind Monitor) at a distance of 80 m from the roadside.

Samples of PM₁₀ and PM_2.5 were collected using a low-volume air sampler with a flow rate of 16.7 L/min, which consists of one-stage Teflon filter packs and cyclones (URG-2000-30ENB, URG-2000-30EH) of cut-size 10 μm and 2.5 μm each. The sampling height was roughly 1.5 m above the surface. Samples for measuring the concentrations of total mass and ionic components in both PM₁₀ and PM_2.5 were collected using a Teflon membrane filter (Zefluor filter) with a diameter of 47 mm. Only PM₁₀ samples were collected for measuring OC and BC concentrations using a quartz filter (Whatman 1851-047, QMA grade), because the number of filter packs was not sufficient to collect PM_2.5 as well at three locations. Considering that OC and BC are present mostly in PM_2.5, only PM_2.5 could have been collected. However, PM₁₀ rather than PM_2.5 was collected to quantify all carbonaceous particles originating from the roads, because OC and BC can be included in coarse particles such as those from tire wear (Watson et al., 1998).

With regard to sampling design, installing upstream denuders and a backup filter/denuder to the filter pack can minimize positive or negative sampling artifacts (Kim et al., 2015). However, such installation is complex and labor-intensive, and was not feasible for this study. Samples were collected for 24 hours starting at 10 a.m., except in special circumstances. Teflon filters were conditioned at 40% relative humidity for 24 hours before and after the sampling and were weighed using an electronic balance (DVG215CD, Ohaus). For ion analysis, a Teflon filter, which is hydrophobic, was first soaked with ethanol in a vial to facilitate the contact with distilled water. After filling the vial with distilled water, ions were eluted using ultrasonic extractor (Power Sonic #420, Hwashin Tech). Detailed procedures for handling filters are described in Won et al. (2010) and Kim et al. (2015).

Three anions (Cl^-, NO₃^-, and SO₄^2-) and five cations (Na⁺, NH₄⁺, K⁺, Mg²⁺, and Ca²⁺) were analyzed using a Metrohm’s module ion chromatography (IC). Anions were analyzed using a Metrosep A Supp 5 column with a solution of 3.5 mM Na₂CO₃ and 1 mM NaHCO₃ as the eluent. Cations were analyzed using Metrosep C2 150 and C4 150 columns with solutions of 4 mM HNO₃ and 1.7 mM HNO₃/0.7 mM dipicolinic acid as eluents, respectively. The minimum detection limits for the IC analysis were determined by analyzing standard solutions of minimum concentrations used for the IC calibration seven times repetitively. The values in μg/L (with uncertainties in parentheses as %) were 4.4 (2.0) for Cl^-, 9.8 (2.7) for NO₃^-, 8.1 (2.2) for SO₄^2-, 6.2 (3.1) for Na⁺, 11.9 (6.1) for NH₄⁺, 14.1 (5.1) for K⁺, 10.3 (3.6) for Mg²⁺, and 5.3 (6.5) for Ca²⁺.

A quartz filter for OC and BC analysis was used after baking overnight at 450°C. Samples collected by filters were sent to the Center for Air Resources Engineering and Science (CARES) at Clarkson University, Potsdam, New York, United States, where OC and BC were analyzed by the National Institute of Occupational Safety and Health (NIOSH) Method 5040 using thermal-optical transmittance (TOT).

One of the most important factors when investigating the effects of road vehicles on ambient air quality is wind direction. Air pollutant concentrations vary considerably depending on whether the measurement location is upwind or downwind (Zhu et al., 2002a). However, calm winds (less than 0.5 m/s) predominated throughout the study periods, accounting for 61 and 87% of the time in January and May, respectively. Northerly wind, which blew along the north-south running road, accounted for 30% in January and 7.2% in May. Although average wind speed from the north was high compared with that from other directions, it was merely 1.19±0.50 m/s and 1.58±0.90 m/s in January and May, respectively. In other words, wind direction and speed were not significant factors in this study.

Because only wind direction and speed data were collected at the measurement site, other meteorological data were taken from the nearest weather station, which is located about 22 km to the east-southeast. During the study periods, average temperature was - 2.0°C (ranging from - 8.4 to 2.0°C) in January and 17°C (ranging from 11 to 23°C) in May. Fig. 2 shows variations of 24-hour average PM₁₀ and PM_2.5 concentrations over the study periods. In January, three occasions of precipitation, including a heavy snow on the 11th, occurred, and there was a relatively large amount of rainfall in May.

Fig. 2. 
Variations in 24-hour average concentrations of PM₁₀ and PM_2.5 with distance from the roadside over the (a) January and (b) May measurement periods. Measurements were made for 24 hours, starting from 11 a.m. or noon.

Table 1 summarizes the measurement results during the study periods. Samples were collected for 20 days from January 4 to 24. In January, the total number of samples was 16 for PM₁₀ and 15 for PM_2.5, which fell 4 and 5 short of a possible maximum of 20, respectively. For one set of measurements, nine filters, which consisted of three Teflon filters and three quartz filters for PM₁₀ and three Teflon filters for PM_2.5, were collected at three measurement locations. If any one of the nine samples was missing during the sampling and analysis process, all samples for those measurements were excluded. Another reason for the reduced number of data was due to collecting samples for 48 hours; a heavy snow on January 11 made it difficult to access the measurement site (Fig. 2(a)).

Initially, we considered not collecting samples in the presence of precipitation. However, we continued taking measurements because it was believed that the effects of precipitation on fugitive dust generation could be large and thus worth checking (Choi et al., 2014). However, variations in concentrations with distance did not change significantly due to precipitation, even on January 11 through 13 when significant amounts of snow piled up.

Fig. 3 shows variations in concentrations of PM, BC, and OC with distance from the roadside. In January, relative concentrations of PM₁₀ and PM_2.5 at 200 m were 0.83 and 0.89, respectively, compared with their concentrations at 10 m. Furthermore, the relative concentration of PM₁₀ at 80 m was 0.79, which is slightly less than that at 200 m. As pointed out previously, a decrease in PM₁₀ concentrations with distance is relatively small compared with BC or OC, for which vehicle exhaust accounts for the majority of total emissions (Zhu et al., 2002a, b; Roorda-Knape et al., 1998). As shown in Fig. 3, PM₁₀ concentrations at 200 m increased compared with those at 80 m, albeit there was a relatively small sample size at the three measurement locations. PM₁₀ concentrations at 80 m were lower than those at 200 m in May as well, as in January. However, relative concentrations were generally higher, and PM_2.5 concentrations at both 80 m and 200 m were higher than those at 10 m. If concentrations at 10 m accurately represent the effects of vehicle emissions, then effects other than vehicle emissions were greater for PM_2.5 than for PM₁₀, and greater in May than in January.

Fig. 3. 
Variations in relative concentrations of PM₁₀ and PM_2.5, and BC and OC in PM₁₀ with distance from the roadside, relative to those at 10 m. Error bars denote standard deviation. Relative concentrations and error bars are displayed at the left and right of the measurement distance for better readability.

As in Zhu et al. (2002a), relative concentrations of BC, a greater proportion of which is emitted from vehicle exhausts, were lower than those of PM₁₀ and PM_2.5 at 200 m in Fig. 3. However, relative concentrations of BC at both 80 and 200 m in January were higher than those in May, which shows that the effects of emissions other than vehicle emissions were relatively large in January. Relative concentrations of OC with a variety of emission sources were larger than those of BC; the small differences in relative concentrations between January and May denote a small seasonal influence.

Fig. 4 shows variations in major ion concentrations of PM₁₀ and PM_2.5 in January. SO₄^2-, NO₃^-, and NH₄⁺ were used to represent secondary products, and Ca²⁺, K⁺, and Cl^- were used as markers of crustal component, biomass burning, and open burning, respectively (Watson et al., 1998). Although Cl^- is generally known to be a tracer of sea salt, a correlation coefficient of - 0.01 between Cl^- and Na⁺ and a high equivalent ratio of Cl^-/Na⁺ at 5.95 (from both PM₁₀ and PM_2.5) suggests that the origin of Cl^- is other than sea salt and presumed to be open burning (Table 1).

Fig. 4. 
Variations in concentrations of primary ions in PM₁₀ and PM_2.5 with distance from the roadside in January. Meanings of symbols and error bar are the same as in Fig. 3

In January, PM₁₀ concentrations were higher at 200 m than at 80 m, and OC and BC concentrations decreased gradually with distance from the roadside (Fig. 3). However, all ion concentrations were higher at 200 m than at 80 m, as in Fig. 4. Compared with concentrations at 10 m, Ca²⁺ (a tracer of crustal component) decreased more rapidly than other ions, but a slight decrease in K⁺ (a tracer of biomass burning) occurred. In other words, the effects of biomass burning are considered to be greater than fugitive dust from surrounding areas beyond 200 m. However, the reason for the highest K⁺ concentration in PM₁₀ at 10 m is unclear.

In Table 2, correlations between total mass, OC, and BC and ions were examined for PM₁₀ and PM_2.5, and for January and May. Correlations of OC and BC in PM_2.5, which were not measured, were calculated using OC and BC in PM₁₀. This was based on the assumption that most OC and BC exists in PM_2.5, although some OC and BC from tire and brake wear can exist in coarse particles (Watson et al., 1998). As in Table 2(a), correlation of total mass is the highest with K⁺, high with NO₃^-, NH₄⁺, and Cl^-, and not as high with Ca²⁺ in the case of PM₁₀ in January. This corroborates that the primary causes of variations in PM₁₀ are biomass burning, open burning, and secondary formation rather than fugitive dust. Considering that K⁺ and Cl^- were highly correlated with OC and BC, gradual declines of OC and BC are related to biomass and open burning in Fig. 3.

Fig. 3 shows a decrease of PM_2.5 in January, but there is no discernible trend except a high Ca²⁺ concentration at 80 m in Fig. 4. Because of differences in variations of Ca²⁺ and PM_2.5 mass, Ca²⁺ has the weakest correlation with total mass. Despite a higher correlation between secondary ions and total mass in Table 2(a), there are no distinct variations with distance in Fig. 4, which can be explained by the fact that the secondary formation on the road and in surrounding areas is similar.

It is highly likely that a decrease of PM_2.5 is primarily due to OC and BC, if variations in ions fail to support a decrease of PM_2.5 in January (Fig. 3). As shown in Table 1, PM_2.5 concentration at 200 m is 5 μg/m³ lower than that at 10 m, but OC and BC concentration in PM₁₀ at 200 m is 2.62 μg/m³ lower than that at 10 m. In contrast, the ion sum is smaller by only 0.64 μg/m³. Despite the use of OC and BC in PM₁₀, it is sufficient to verify the effects of OC and BC in a decrease of PM_2.5 considering that most of OC and BC are present in PM_2.5.

Even in May, secondary ions exhibited small variations, as shown in Fig. 5. However, secondary ions in PM_2.5 increased with distance, similar to variations of PM_2.5 in Fig. 3, showing that variations of PM_2.5 are associated with secondary ions. However, a correlation between total mass and secondary ions is lower than that in January. It is interesting to note that mass concentrations exhibited a significant correlation with NH₄⁺ and NO₃^- in January but with NH₄⁺ and SO₄^2- in May. It is generally known that NO₃^- is primarily influenced by local effects while SO₄^2- is more influenced by regional effects (Edgerton et al., 2006; Brook et al., 2002). Different correlations of NO₃^- and SO₄^2- in January and May should also have resulted from the influence of surrounding areas rather than the influence of roads, but further study is warranted to confirm the current assertion.

Fig. 5. 
Variations in concentrations of primary ions in PM₁₀ and PM_2.5 with distance from the roadside in May. Meanings of symbols and error bar are the same as in Fig. 3

Compared with that in January, a higher correlation with Ca²⁺ in PM₁₀ is noteworthy in May. Substantial fugitive dust generation can be seen to occur at 200 m more affected by surrounding areas in May with a higher temperature than in January with a lower temperature below freezing. However, absolute concentrations of Ca²⁺ were higher at 10 m as well as at 200 m in January than in May (Table 1).

Fig. 6 shows the correlation between OC and BC. It is presumed that a larger slope in January than in May is due to the influence of biomass burning and open burning, which emit more OC than vehicle exhaust. It can also be interpreted that the largest slope at 200 m PM₁₀ May PM₁₀ PM_2.5 PM_2.5 in January is associated with emissions from biomass and open burning. As pointed out previously, the influence of secondary formation is significant in May, thereby causing a small variation of slope with distance (standard deviation of 0.34 compared with 0.59 in January) and less steep slopes than in January.

Fig. 6. 
Correlation between OC and BC with distance from the roadside. At 10, 80, and 200 m, the slopes are 3.62, 3.34, and 4.48, respectively, in January, and 2.59, 3.27, and 2.84 in May.

Fig. 7 shows average composition variations of PM₁₀ and PM_2.5 with distance. In January, an abundance of primary ions, OC, and BC confirms again the importance of biomass or open burning in variations of PM. In May, considerable amounts of secondary ions and others exist in both PM₁₀ and PM_2.5. As photochemical reactions become active due to temperature rise, the influence of secondary formation increased in PM₁₀ as well as PM_2.5, leading to decreased variations with distance.

Fig. 7. 
Chemical compositions of PM₁₀ and PM_2.5 with distance from the roadside.