On-Road Measurements of Ultrafine Particles and Associated Air Pollutants in a Densely Populated Area of Seoul, Korea

Spatial distributions of ultrafine particles (UFPs; 6 < Dp < 560 nm) and related gaseous and particulate pollutants were estimated from on-road measurements undertaken on busy roadways of Seoul, Korea, using a mobile laboratory (ML). The objectives of the study were to determine the spatial variations in UFP size distributions and concentrations of associated gaseous and particulate pollutants and to observe the relationships of UFP number concentrations with other pollutants on roadways in an urban area in Korea. The pollutants associated with diesel vehicles such as black carbon (BC) and particlebound polycyclic aromatic hydrocarbons (PM-PAHs) exhibited a high determination coefficient (r = 0.65), indicating the influence of diesel vehicles on emissions in the study area. Further supporting evidence for the influence of diesel vehicles on emissions was given by the higher determination coefficients of PM-PAHs and BC concentrations with larger sizeclassified particles, ranging from 60 < Dp < 220 nm, than with total UFP number concentrations or smaller particles in the 6 < Dp < 60 nm size range. Peak concentrations of measured pollutants were observed mostly at intersections, reflecting the relationships of transient driving modes (i.e., deceleration and acceleration) with emissions of UFPs, associated pollutants, and concentrated traffic volumes at such locations.

Atmospheric ultrafine particles (UFPs; D p < 0.1 µm) originating from natural sources and processes as well as from anthropogenic activities have attracted increasing interest in the last decade (Morawska et al., 2008).UFPs are important from the perspective of adverse health effects on humans as well as their subsequent growth into fine particles (FPs; D p < 2.5 µm), which have the longest atmospheric residence time of all particulates and can be transported for long distances from the source area, contributing to regionalscale air pollution.Among the anthropogenic particulate matter (PM) sources, emissions of industrial combustion processes and traffic-related emissions account for the highest contributions of both UFPs and FPs (Schauer et al., 1996;Shi et al., 1999;Cass et al., 2000;EPA, 2000;Harrison et al., 2000;Hitchins et al., 2000).For traffic-related emissions, the greater part of the particle number concentration from vehicle exhaust is in the 20-130 nm size range for diesel engines (Kittelson, 1998;Morawska et al., 1998) and in the 20-60 nm for gasoline engines (Kittelson, 1998;Ristovski et al., 1998).PM emissions from internal combustion engines have traditionally been regulated solely on the basis of total PM mass emissions (Kittelson et al., 2004a).In response to such regulations, engines that emit much lower PM mass concentrations have been developed (Kittelson et al., 2004a).However, there is a concern that modern vehicles may emit smaller sized particles because of their advanced exhaust-gas management systems and factors such as driving conditions, which influence the emission of smaller particles.
Many studies have investigated the physical and chemical characteristics of UFPs or FPs near roadsides (Hitchins et al., 2000, Zhu et al., 2002a, b;Polidori et al., 2008;Buonanno et al., 2009;Hagler et al., 2009;Kim et al., 2011a, b).Studies have reported UFP concentration gradients near sources, including the rapid decrease in particle number concentrations with distance downwind of roadways (Hitchins et al., 2000, Zhu et al., 2002a, b;Polidori et al., 2008;Buonanno et al., 2009;Hagler et al., 2009;Kim et al., 2013).These stationary measurement studies provide useful data at roadsides but are less suitable for investigating the spatial distribution of UFPs and sharp gradients that characterize UFP concentrations on real roadways.Roadside and vehicle tail pipe exhaust measurements must be linked to understand the spatial nature of UFPs, as well as differences in actual pollutant concentrations along roadways.Recent studies in various countries (China, Finland, Germany, Netherland, Switzerland, and United States) have developed mobile monitoring platforms using real-time instruments in attempts to estimate on-road air pollution concentrations together with spatial and temporal gradients of UFPs and related air pollutants (Bukowiecki et al., 2002a, b;Canagaratna et al., 2004;Kittelson et al., 2004a, b;Pirjola et al., 2004;Weijers et al., 2004;Schneider et al., 2008;Wang et al., 2009).Highways and arterial roads have been targeted for these measurements in both rural and urban areas.These studies provide information regarding UFP number concentrations with size distributions and levels of associated gaseous pollutants such as CO, CO 2 , and NO x under real-world conditions.However, most of these studies have been conducted in European countries or the United States.Few studies have been carried out using mobile monitoring platforms in Asian countries (Wang et al., 2009).
We carried out on-road measurements on busy roadways of Seoul, Korea, using a mobile laboratory (ML) to estimate air pollutants concentrations together with the spatial distributions of UFPs (6 < D p < 560 nm in mobility diameter) and related gaseous and particulate pollutants.Our objectives were a) to provide UFP spatial variations and size distributions with associated gaseous and particulate pollutant concentrations and b) to observe the relationships of UFPs number concentrations with other pollutants on roadways in an urban area in Korea.

Mobile Laboratory
We developed the ML for real-time on-road measurement in 2009 (Ministry of Environment, Republic of Korea).The ML enabled us to measure vehicle-related gaseous and particulate pollutants while driving.Sampling inlets were located at the top of the ML and consisted of a stainless steel pipe and Teflon tubing for particulate and gaseous pollutants, respectively.The sampling height was approximately 2 m above ground level as shown in Fig. 1.
The inlet for particulate monitoring was funnel-shaped with an inner entrance diameter of 10 mm.A stainless steel sampling line (48 mm inner diameter [I.D.] and 1.6 m length) was connected to the end of the funnel-shaped inlet and was also connected to a chamber (150 mm I.D.) inside the ML, labeled as A in Fig. 1.Air was pulled into the chamber at approximately 65 L/min by an inline pump located downstream of all sampling ports.The chamber contained five small inlet ports, which were connected to several instruments measuring aspects of particulate pollution as shown in Fig 1 .All sample lines for particles were connected with conductive tubing.Teflon tubing was used for sampling gaseous pollutant and all sampling lines were kept very short (0.5-1 m) to minimize residence time and particle losses.

Measurement Equipment and Data Acquisition
Fig. 1 illustrates the ML instrument setup and Table 1 gives details of each instrument with the parameter measured and its time resolution.A fast mobility particle sizer (FMPS Model 3091, TSI Inc., Shoreview, MN, USA) was used to measure the particle number size distribution with the particle mobility diameter D p at a flow rate of 10 L/min.The particles, which flow through a region of differing electrical fields, are repelled by the voltage from the central column.When the particles hit the outer cylinder, consisting of multiple cylindrical rings, the particle impact creates a current measured by electrometers (Weimer et al., 2009).The electrometer backgrounds were monitored and set to zero using particle-free air before each measurement.In the present study, UFPs size range was defined from 6 to 560 nm in mobility diameter.
The surface area of nanoparticle aerosols that deposit in the lung, as corresponding to the International Commission of Radiological Protection (ICRP) lung deposition curves for the alveolar (A) regions of the human respiratory tract, was monitored by a nanoparticle aerosol monitor (NAM Aerotrak 9000, TSI Inc.) equipped with a PM 1.0 cyclone for the removal of particles larger than 1.0 µm in aerodynamic diameter.Other measured pollutants included black carbon (BC), particle-bound PAHs, NO x , and CO 2 .The BC sample line was equipped with a PM 2.5 cyclone for the removal of particles larger than 2.5 µm in aerodynamic diameter, before utilizing a portable aethalometer (AE42 dual-wavelength, Magee Scientific, Berkeley, CA, USA), operating with quartz filter tape, illuminated by 370 and 880 nm wavelength light, with the allowable optical attenuation depth of the filter set to 1.25 (adjusted for 370 nm) at a flow rate of 5 L/min.Nitrogen oxides were measured with a NO x analyzer, utilizing chemiluminescence (AC32M, Environmental S.A., Poissy, France).Carbon monoxide and carbon dioxide were measured with a non-dispersive infra-red (NDIR) analyzer (CO12M, Environmental S.A.).A photoelectric aerosol sensor (PM-PAHs; PAS 2000, EcoChem, League City, TX, USA) was used to monitor representative particulate PAHs.PM-PAHs with three or more rings were detected through the measurement of electrons emitted by organic molecules on particles irradiated by UV light.Further information regarding the principle of measurement by this technique is described in detail elsewhere (Burtscher, 1992;Dunbar et al., 2001).The air flow rate for the PAS 2000 sensor was set at 2.0 L/min and the lamp wavelength was fixed at 220 nm.As shown in Table 1, 6-s intervals, during which the lamp was off for 2 s and then on for 4 s, were used for measurements.

Driving Sections and Measurement
According to Statistics Korea (2011), about 18.4 million motor vehicles are registered in the country, with vehicles powered by gasoline, diesel, and liquefied petroleum gas (LPG) accounting for 49.7, 36.4,and 13.2% of the total, respectively.Of the total motor vehicles in Korea, 16.2% are registered in Seoul.Additional information about the traffic composition and volume is provided in Table S1 (Supplement).The Gangnam district is one of the most affluent areas of Seoul, with an area of 40 km 2 and a population of over 500,000.Several popular shopping and entertainment areas are located in the Gangnam district and an important business area runs east-west from Gangnam Station to Samseong Station.The Gangnam district is also surrounded by several expressways which could influence the air quality of the area.We chose this area for our study and drove the ML on fixed routes in the Gangnam district, as shown in Fig. 2.
The driving route covered the business, residential, shopping, and entertainment areas of the Gangnam district.We divided the driving route into five sections according to their characteristics, as shown in Fig. 2 and described in Table 2.Sections [1], [2], [3], [4], and [5] included business, residential, shopping, entertainment, and business with entertainment areas, respectively.The total driving distance was 16.31 km, with the distances covered in sections and [5] were 4 .12, 3.74, 3.05, 2.19, and 3.21 km, respectively.To avoid the direct influence of vehicle exhaust plumes, we maintained a distance greater than 2-3 m from the nearest vehicle in front of the ML while driving.The measurements were conducted in the winter of 2011 and this study presents intensive and consecutive measurements from 11:19 to 16:40 hrs on 22 December 2011 (five measurements for each study area).Urban background concentrations at the roadside were measured when driving in section [2] (the location is indicated by the red square in Fig. 2).The blue circles, from A to F, indicate points where peaks of high pollutant concentrations were observed.These peaks are discussed later.Morning and evening rush hours were intentionally avoided to eliminate the influence of commuting (i.e., heavy traffic congestion) in the study area.

Concentration Levels in the Different Driving Sections
Average vehicle speeds with corresponding meteorological information are shown in Table 3.The average vehicle speed was 16.5 ± 3.1 km/h and the wind predominantly blew  3.21 ± 0.04 10 Business, residential, shopping and entertainment mixed area "±" indicates standard deviations of measured distances (n = 9).from the west-northwest or northwest to southeast with an average speed of 3.2 ± 0.3 m/s during the five measurements.Table 4 presents the average concentrations of observed pollutants.The sample period was on a weekday and the measurements were repeated five times consecutively, to estimate the average concentrations.
The highest concentrations of pollutants, except for NO x (NO and NO 2 ), were observed in section [1], and the lowest concentrations were found in section [2], as shown in Table 3.The winter monsoon typically blows from the northwest to southeast in Korea.The prevailing wind directions were northwest and north on the measurement day.These winds were assumed to have influenced the lower concentrations recorded in section [2] by reducing the direct influences from vehicles.In addition, dilution effects may have arisen from a stream that runs through section [2] in a northerly direction.Furthermore, a relatively smaller traffic volume due to fewer road lanes and the characteristics of section [2] (residential and school areas) could also have been factors in the lower pollutant concentrations in comparison to sections.Relatively low pollutant concentrations were expected in section [5] through which the Han River flows in a northward direction.However, pollutant concentrations in section [5] were similar to those in section [4].Relatively wider road lanes (10 lanes) with large traffic volumes and high rise buildings, which may act as obstacles against horizontal mixing, could have influenced the observed pollutant concentrations in section [5].
Section [2] exhibited the smallest concentration variations, whereas section [4] showed the largest concentration variations for the five measurements (except for CO 2 concentrations).These temporal variations in pollutant concentrations reflect the mobility of vehicles travelling through the measurement areas.Despite the temporal variations for each driving section, generally high concentrations in section [1] and low concentrations in section [2] were observed throughout the measurement periods.This reflects the traffic volumes and the characteristics of these sections, i.e., a business area in section [1] and a residential area in section [2]."a" indicates the number of data based on 1 s data recording for each section (n = 5).

Correlations among Various Pollutants
Correlations were calculated to identify any relationships among pollutants (Fig. 3 and Table 5).Relatively high correlations were observed among PAHs, BC, and particle surface area, while UFP number concentrations showed moderate correlations with BC (r 2 = 0.32), CO 2 (r 2 = 0.33), and NO (r 2 = 0.42) and weak correlations with PAHs (r 2 = 0.20).The pollutants associated with diesel vehicles such as BC and PAHs showed high determination coefficients (r 2 = 0.65) indicating the influence of diesel vehicles in our study area.CO 2 was weakly or moderately correlated with other pollutants.Large quantities of CO 2 are emitted from all vehicles powered by fossil fuel and therefore a moderate correlation is not a surprising result.Similar moderate correlations (r 2 = 0.4-0.5)for CO 2 with other pollutants were reported by Westerdahl et al. (2005).Westerdahl et al. (2005) also reported determination coefficients (r 2 ) above 0.7 for UFPs (0.007-1 µm) with BC, NO, and PM-PAHs.Although the monitoring instruments used in the current study were not identical to those used by Westerdahl et al. (2005), only moderate or weak correlations of UFPs with BC, PM-PAHs, and NO were obtained in our measurements.Therefore, we divided particle sizes into five groups  and 6, size-classified particle groups (c) 60 < D p < 107 nm and (d) 107 < D p < 220 nm exhibited much higher determination coefficients with PM-PAHs and BC concentrations, whereas poor determination coefficients remained for the sizeclassified particle groups (a) 6 < D p < 20 nm and (b) 20 < D p < 60 nm.These results may reflect the relationship of relatively larger particles with PM-PAHs and BC.Venkataraman and Friedlander (1994) investigated the size distributions of PAHs (PAHs with more than four rings) and elemental carbon (EC) in Los Angeles, CA, USA, and observed ambient PAH and EC size distributions with bimodal peaks in the 0.05-0.12µm (mode I) and 0.5-1.0µm (mode II) size ranges.They concluded that mode I particles were attributable to primary emissions from combustion sources whereas mode II particles were attributable to the accumulation of secondary reaction products on primary particles.Hence the lack of any relationship between UFP number concentrations in the 220 < D p < 560 nm particle size range with measured pollutants may have been a consequence of this bimodal characteristic of PAHs and BC.The poor correlations of the UFP number concentrations in the 6 < D p < 60 nm particle size range with other pollutants, can also be explained by the findings of Venkataraman and Friedlander (1994).The higher correlations of UFP number concentrations in the 60 < D p < 220 nm particle size range with PM-PAHs, BC, and particle surface area strongly reflect Fig. 3. Scatter plots from 60-s average concentrations on 22 December, 2011 (n = 324).Units are 1000-s of particles/cm 3 for UFPs, µm 2 /cm 3 for surface area (NAM), µg/m 3 for BC, ng/m 3 for PAHs, ppb for NO, and ppm for CO 2 (range 540-740 ppm).Note x and y are identical for each pollutant.the dominant contribution of diesel vehicles to emissions of relatively larger particles (60 < D p < 220 nm) over the measurement period in our sampling area because PM-PAHs and BC are known to be emitted from diesel engines.Furthermore, as noted in the Introduction, the greater part of the particle number concentration from vehicle exhausts is in the 20-130 nm size range for diesel engines (Kittelson, 1998;Morawska et al., 1998) and in the 20-60 nm range for gasoline engines (Kittelson, 1998;Ristovski et al., 1998).Therefore, the linear relationships of the 60 < D p < 220 nm particle size range with PAHs and BC indicate the dominance of diesel exhaust emissions for relatively larger particles in our study area, although it was difficult to determine the relative contributions of vehicle emissions for smaller particles (6 < D p < 60 nm).
As Giechaskiel et al. (2005) observed, the nucleation mode is a particle size classification emitted by vehicles and consists fully of volatile compounds (Kittelson et al., 2006) whereas the non-volatile fraction accounts for the largest part of the soot particle mode (D p > 50 nm).Baltensperger et al. (2002) and Wehner et al. (2004) observed a high non-volatile fraction for particles in the 80-150 nm size range, in ambient measurements with a high traffic influence.Kittelson et al. (2002Kittelson et al. ( , 2004a, b) , b) conducted onroad measurements with two scanning mobility particle sizers (SMPS) equipped both with and without a thermal denuder (TD) operated at 300°C, to determine the nonvolatile portion of the particle size distribution.They found that the TD removed between 87 and 95% of the particle number (nucleation mode) suggesting that most of the particles consisted of volatile material and that smaller particles were composed of more volatile materials.Similar observations were also made by Weimer et al. (2009).In the present study, the 6 < D p < 60 nm particle size accounted for 93% of the total particle number concentrations, which is consistent with the findings of Kittelson et al. (2002Kittelson et al. ( , 2004a, b) , b) and Weimer et al. (2009).Therefore, the nanoparticle (D p < 60 nm) number concentrations in our study area may be related to volatile compounds, which can be emitted from either gasoline or diesel vehicles.

Time-Series Plots of Pollutants for Each Location
Figs. 5(a)-5(d) demonstrate very similar trends between UFP number concentrations and other associated pollutants.The 60-s average concentration data represent and summarize the trends of air pollutants better than the 1-s concentration data.Average concentrations of measured pollutants tended to be high in section [1] and low in section [2] for all measurements.The business district is located mainly in section [1] with dense residential areas in section [2].Moreover, the sections differed in terms of number of road lanes, with up to 12 (8-12) lanes in section [1] and 4 lanes in section [2].The large traffic volumes and the characteristics of section [1] (business district) may have contributed to the high concentrations of air pollutants.The region labeled as 1A in section [2], as shown in Figs.5(a) through 5(d), indicates the measurement at the roadside background site.UFPs, particle surface area, and NO concentrations tend to show sharp concentration peaks over the first half of the distance driven in section [3], where a large interchange (Seocho I.C. of the Gyeongbu Expressway, the largest expressway in Korea) is located.The Seocho I.C. is located in the northwest of the study area and is about 500 m away from the locations sampled in first half of section [3].The main wind directions during the measurement periods were northwest and west-northwest with an average wind speed of 3.2 ± 0.3 m/s (Korea Meteorological Administration).Hence, the sharp peaks in UFPs, particle surface area, and NO concentrations, labeled as 2A in Figs. 5 (a)-5(d), may have been generated by traffic at the Seocho I.C.Local wind speed plays an important role in determining the particle number size distributions in an urban area.Wu et al. (2008) observed that the nucleation (3 < D p < 20 nm) and coarse mode (D p > 1 µm) particle number concentrations  Several peaks of measured pollutants were observed across all the sections (A-F in Fig. 5).Interestingly, these peaks were commonly found at intersections reflecting the relationships of transient driving modes (i.e., deceleration and acceleration) with UFPs and associated pollutants (see Fig. 2, A-F indicated by blue circles).This is also related to the concentrated traffic volumes at intersections.Tong et al. (2000) previously reported that transient driving modes are more polluting than steady-speed driving modes (i.e., constant velocity) in terms of the pollutant amounts per distance (g/km) and time (g/s), although they measured only CO, HC, NO x , and soot.They also observed that idling emissions of CO, HC, and NO x were higher than those for acceleration and deceleration modes for both petrol and diesel vans in terms of fuel consumed (g/kg).

Comparisons with Other Studies
The UFP number concentrations and the concentrations of other pollutants measured in this study are compared with other studies in Table 7.Because only arterial roadways were studied in the present study, we compared the concentration levels observed on arterial roadways and background sites here.Note that the instruments may differ between studies, and the traffic volumes or other factors such as the condition of background sites must also be considered.For the roadway (not freeway) average, our study measured 97,000 particles/cm 3 , whereas UFP concentrations of 40,000, 160,000, and 133,000 particles/cm 3 were reported by Westerdahl et al. (2005), Kittelson et al. (2004b), andSchneider et al. (2008), respectively.Our results are comparable to these previous findings, except those of Westerdahl et al. (2005).Our BC concentration was almost five times higher than that observed by Westerdahl et al. (2005), whereas we measured almost half the BC concentration observed by Schneider et al. (2008).Although PM-PAH concentrations were not reported by the studies carried out with a mobile platform shown in Table 7, roadside and onroad measurements taken with the same PM-PAH monitor (PAS 2000) in Mexico City showed up to 100-200 ng/m 3 (during morning rush hour) and 210-790 ng/m 3 of PM-PAHs, respectively (Marr et al., 2004(Marr et al., , 2006)).The concentrations of nitrogen oxides in our measurements were higher than those observed in the USA (Kittelson et al., 2004;Westerdahl et al., 2005).The background pollutant concentrations in our study also tended to be higher than those observed in other studies.However, it should be noted that our background area was located immediately beside the roadside in the center of an urban area in Seoul, whereas background sites for other studies were located outside urban areas.

UFP Numbers and Size Distributions
Overall, high particle number concentrations in the 6 < D p < 20 nm size range were measured in sections [1] and [4], while high particle number concentrations in the 60 <  D p < 220 nm size range were recorded in sections [1] and [5] (Fig. 6).The variations in BC and PM-PAH concentrations in the different sections were closely related to the particle numbers in the 60 < D p < 220 nm size range and the number of road lanes and hence traffic volumes (Tables 2 and 3).Average vehicle speeds for sections [1], [2], [3], [4], and [5] were 17.4 ± 6.3, 12.5 ± 6.8, 16.7 ± 6.8, 11.1 ± 5.5, and 22.8 ± 8.0 km/h, respectively, indicating traffic congestion in section [4] when considering the number of road lanes available (see Table 2).Traffic congestion was frequently observed in section [4] due to the presence of shopping and entertainment areas and many crosswalks with traffic signals.The higher standard deviations recorded in section [4], particularly for the smaller particle size, strongly support this tendency.Large particles (60 < D p < 220 nm) accounted for 7.4 and 8.0% of the total UFP number concentrations observed in sections [1] and [5], whereas in sections [2] to [4] the corresponding figures were 5.1 to 5.7%, as shown in Fig. 6.In summary, the relative contribution of smaller particles (6 < D p < 60 nm) to the total measured was high in sections [2] to [4].

CONCLUSIONS
In the present study, we successfully measured UFPs and associated air pollutants using a mobile laboratory (ML).This study provides a link between roadside and vehicle emission measurements by estimating the spatial distributions of UFPs and associated air pollutants, on real-world roadways, to enhance understanding of traffic emissions and characterize pollution gradients along roadways.
High determination coefficients (r 2 = 0.65) among the pollutants associated with diesel vehicles (BC and PM-PAHs) indicate the influence of diesel vehicles in our study area.Although poor correlations between total UFP number concentrations and BC or PM-PAHs were recorded, sizeclassified larger particles ranging from 60 < D p < 220 nm showed much higher determination coefficients with PM-PAHs and BC concentrations whereas poor correlations were recorded for the size-classified smaller particles ranging from 6 < D p < 60 nm with PM-PAHs and BC concentrations.The higher correlations of UFP number concentrations for the 60 < D p < 220 nm size range with PM-PAHs, BC, and particle surface area strongly reflect the dominant contribution of diesel vehicle exhaust emissions for relatively larger particles (60 < D p < 220 nm) over the measurement period in our monitoring area.Several peaks of measured pollutants were observed, mostly at intersections, reflecting the relationships of transient driving modes (i.e., deceleration and acceleration) with UFPs, associated pollutants, and concentrated traffic volumes.
quality-Climate change information, Seoul" provided from a meteorological station located in Gangnam.Table4.Average pollutant concentrations on December 22, 2011.indicates standard deviations of measured pollutants (n = 5).
[(a) 6 < D p < 20 nm, (b) 20 < D p < 60 nm, (c) 60 < D p < 107 nm, (d) 107 < D p < 220 nm, (e) 220 < D p < 560 nm)] and plotted these size categories against the other pollutants measured to clarify the relationship between fractional particle concentrations and air pollutants.No relationships were found for UFP number concentrations in the 220 < D p < 560 nm particle size range with measured pollutants.Scatter plots of size-classified particles and PM-PAHs are shown in Figs.4(a) and 4(b).The third of the five measurements, which included the roadside background measurement, was excluded from these plots owing to the different conditions at the background site.Unlike the poor correlations of total UFP number concentrations in Tables 5

Fig. 5 .
Fig. 5. Time-series plot of measured pollutants during 22 December arterial road ways.(a) UFPs total number vs. NO concentration, (b) UFPs total number vs. particle surface area (NAM), (c) UFPs total number vs. CO 2 concentration, (d) black carbon vs. PM-PAHs concentration ("A"-"F" indicates abundant peaks of pollutants; 1A indicates background measurement at roadside; 2A indicates sharp peaks of UFPs, particle surface area, and NO concentrations in the first half driving section [3].

Fig. 6 .
Fig. 6.Comparison of particle number concentration for five driving sections measured on December 22, 2011.

Table 2 .
Divided driving sections with distances and characteristics.

Table 3 .
Measurement periods and the average vehicle speed with meteorological information.

Table 7 .
Comparison of reported average concentrations for UFPs and associated pollutants for roadways and background areas.