Aerosol and Air Quality Research, 13: 957–976, 2013 Copyright © Taiwan Association for Aerosol Research ISSN: 1680-8584 print / 2071-1409 online doi: 10.4209/aaqr.2012.07.0184

Characteristics of PM2.5 Haze Episodes Revealed by Highly Time-Resolved Measurements at an Air Pollution Monitoring Supersite in Korea Seung-Shik Park1*, Sun-A Jung2, Bu-Joo Gong2, Seog-Yeon Cho3, Suk-Jo Lee2 1 Department of Environmental Engineering, Chonnam National University, 300 Yongbong-dong, Gwangju 500-757, Korea 2 Division of Climate & Air Quality Research, National Institute of Environmental Research, Kyungseo-dong, Seo-Gu, Incheon 404-170, Korea 3 Department of Environmental Engineering, Inha University, 100 Inha-Ro, Nam-gu, Incheon 402-751, Korea ABSTRACT

Hourly measurements of PM2.5, organic and elemental carbon (OC and EC), inorganic ionic species, and elemental constituents were made between February 1 and March 31, 2011, at a South Area Supersite at Gwangju, Korea. Over the two-month study period, daily PM2.5 mass concentration exceeded the 24-hr average Korean NAAQS of 50.0 μg/m3 on 20 days, of which two pollution episodes (episodes I and II) are investigated. Episode I (February 01–08) is associated with regional pollution along with wild fire smoke emissions over southern China, and also characterized by high CO/NOx ratios and high K+ concentrations. While episode II (March 11–12) is characterized by locally produced pollution with low CO/NOx ratios, and broad variations in OC, EC, and NO3

– concentrations. For episode I, the 1-hr PM2.5 mass concentration ranged from 27 to 159 μg/m3 with a mean of 88 μg/m3. Although a low

OC/EC ratio (3.4) was observed for February 04–05 when high K+ concentrations occurred, significant linear relationships between EC – OC (R2 = 0.82), OC – K+ (R2 = 0.77), and K+ – K+/OC (R2 = 0.77) suggest the contribution of carbonaceous aerosols from long-range transport of biomass burning plumes occurring in the southern regions of China during episode I, as well as local traffic emissions.

For episode II, the hourly PM2.5 mass concentration ranged from 23 to 116 μg/m3 with a mean of 78 μg/m3. Hourly maximum contributions of SO4

2– and NO3– to the PM2.5 mass were 54 and 66%, respectively. An elevated NO3

– concentration was observed along with high OC and EC concentrations, suggesting the influence of local emissions. The pattern of SO4

2– variations in relation to wind direction and the strong correlation between SO2 and SO4

2– suggest local SO2 emissions were likely an important source of SO4

2– at the site. Keywords: Korean PM Supersites; Semi-continuous measurements; Wildfires burning emissions; OC/EC ratio; OC – K+ relationship. INTRODUCTION

Particulate matter (PM) smaller than 2.5 μm in aerodynamic diameter (PM2.5) is of great concern due to the transportability in the human body. PM2.5 in the atmosphere has been linked to the formation of haze and changes in the atmosphere’s radiative balance (Charlson et al., 1987), as well as adverse effects on human health, crops, and materials (Chameides et al., 1999). Haze pollution has been studied intensively on its impact on air quality, visibility, climate, and public health (Chameides et al., 1999; Kim et al., 2001; Okada et al., 2001; Schichtel et al., 2001; Menon et al., 2002; * Corresponding author. Tel.: 82-62-530-1863;

Fax: 82-62-530-1859 E-mail address: park8162@chonnam.ac.kr

Watson, 2002; Watson and Chow, 2002; Chen et al., 2003; Weber et al., 2003; Yadav et al., 2003; Kang et al., 2004; Park et al., 2006a; Sun et al., 2006; Kim et al., 2008; Jung and Kim, 2011; Oanh and Leelasakultum, 2011). Results from these studies have revealed that haze pollution is strongly correlated with meteorological conditions, emissions of pollutants from anthropogenic sources, and gas-to-particle conversion. Emissions of air pollutants and air stagnation conditions are favorable to haze formation which alters the chemical composition of atmospheric aerosol particles. It has been suggested that haze is not restricted to urban areas and is often regional in nature. Urban PM2.5 is typically composed of local and regional sources such as carbon and nitrate are contributed significantly from local sources, while sulfate particles are mostly from regional sources. Therefore, understanding the association between local and regional emissions and the formation of haze is critical to effective air quality regulation.

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 958

Although long integrated filter-based measurements in the above-mentioned studies have mainly been made to determine aerosol chemical constituents during haze events, their temporal resolution is typically insufficient to observe short-term variations. The daily averaging times used in long integrated filter-based measurements tend to smooth out much of the variability, thus limiting the understanding of the factors influencing of haze formation dynamics and the short-term exposures that result in adverse health impacts. For these reasons, time-resolved measurements of PM2.5 and its chemical components were conducted to better understand PM2.5 sources and the dynamics of haze formation because chemical composition of atmospheric aerosol particles is highly variable in space and time, and is dependent on meteorological parameters (Weber et al., 2003; Park et al., 2005b, 2005c, 2006a; Solomon and Sioutas, 2008). It has been indicated that during haze events, excess PM2.5 is often largely due to elevations in the concentration of one or two of the major chemical species. Such elevations are often induced by near-by sources and/or regional sources. These studies indicated that short-term transients in carbonaceous species, nitrate, CO, and NOx levels were often observed in the morning hours or sometimes in evening hours. In contrast, sulfate concentrations peak during summertime afternoons when ozone levels are elevated, or can be strongly associated with the regional haze events.

As a part of the development of effective control strategies to reduce the impacts of atmospheric PM including yellow sand and long-range transported aerosols, the Korean EPA in 2007 launched an intensive air pollution monitoring research program known as the Supersite program, which is similar to the U.S. EPA Supersite program (Solomon and Sioutas, 2008). The primary objectives of the Korean Supersite program are to monitor the physicochemical characteristics of the yellow sand and long-range transported aerosols, estimate the domestic background concentrations of the aerosol particles at remote background sites, provide unprecedented physicochemical characterization of ambient PM in urban areas, and evaluate semi-continuous measurement methods of PM and its chemical speciation data. This study was carried out at a South Area Supersite in Gwangju, Korea, which began in 2009. The South Area air pollution supersite study was designed to develop an extended set of time-resolved and compositionally resolved PM2.5 data for use in apportioning sources and resolving their local and regional contributions. The objectives of this study was to compare results between filter-based integrated and semi-continuous measurements of PM2.5 mass and its major chemical constituents from data gathered for approximately two months in 2011 at a South Area Supersite in Gwangju, Korea. In addition to evaluating the near continuous methods, an attempt was made to elucidate the key chemical characteristics of two PM2.5 haze pollution episodes revealed by time-resolved measurement methods.

EXPERIMENTAL Description of the South Area Supersite at Gwangju

The South Area Supersite (latitude 35.23°N, longitude

126.85°E) is located in the southern part of Korea, as depicted in Fig. 1. The site is located in a semi-urban commercial and residential area which is surrounded by agricultural lands and traffic roads, and is ~20 km northwest of downtown Gwangju and northeast of two industrial complexes. Emissions from these areas to the site could be resolved during typical western and southern flow regimes. The closest major traffic road lies approximately 0.3 km southeast of the site, and the Honam express highway is located 1.5 km west of the site. Meteorological data were monitored at the Gwangju regional meteorological station, located 7 km south of the site. Gwangju has a population of about of about 1.4 million in a 501.4 km2 area, with 350,000 motor vehicles, of which gasoline vehicles account for ~70%. In addition, ~85% of total air pollution emissions in the city are attributed to vehicular sources. Air quality in Gwangju has been reported to be influenced by local pollution and long-range transported pollution from polluted regions of China (Kim et al., 2001; Park et al., 2005a, 2006b, 2010, 2012; Jung et al., 2010; Jung and Kim, 2011; Park and Cho, 2011). High levels of PM pollution are also observed due to the burning of agricultural wastes in farming areas nearby the Supersite (Ryu et al., 2004, 2007; Park et al., 2006b; Jung and Kim, 2011). Thus, the South Area Supersite located in Gwangju is an excellent choice for studying the properties of local, regional, and biomass burning aerosol emissions and their effect on urban air quality, in order to investigate our hypothesis regarding time-resolved measurements and aerosol age.

Highly time-resolved measurements of PM2.5, organic and elemental carbon (OC and EC), ionic species, and elemental species have been made at the Gwangju Supersite at 1-hr intervals since January 2010 using commercial semi-continuous instruments. However, in the present study, only the measurement data recorded between February 1 and March 31, 2011 were utilized due to frequent malfunction of some instruments. Hourly ambient temperature ranged from –7.4 to 16.7°C (a mean of 3.1°C) during February and from –2.9 to 19.7°C (a mean of 5.8°C) during March. Average relative humidity (RH) during February and March was 64.8% (18.2–99.9%) and 55.5% (16.9–99.5%). 24-hr PM2.5 Mass and Speciation Measurements

For comparison of the semi-continuous measurements, 24-hr integrated filter-based PM2.5 samples were collected between midnight and midnight with three sets of sequential PM2.5 samplers (PMS103, APM Korea) and then used for determining the total mass concentration, the amounts of carbonaceous species (OC and EC), eight ionic species (Cl–, NO3

–, SO42–, Na+, NH4

+, K+, Ca2+, and Mg2+), and 29 elemental species (Si, S, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Br, Rb, Sr, Mo, Cd, Sn, Sb, Te, Cs, Ba, Hg, Pb, and Bi). During the filter-based sampling of aerosol particles, the artifacts that can arise due to particle-to-particle interactions, gas-particle interactions and the dissociation loss of semi-volatile species can change the collected aerosol composition. Positive and negative artifacts may have been created in this study because the sequential sampler used did not have a denuder to remove gases prior to sampling

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 959

Fig. 1. Area map of Gwangju PM Supersite.

particles or a backup filter to collect vapor from particles collected on the filter. Issues arising from the sampling artifacts are discussed below.

The Teflon filters (Teflo membrane, 2.0-μm pore size, Gelman Science, USA) were weighed before and after sample collections with a microbalance of 1 μg sensitivity and analyzed for elemental components using X-ray fluorescence (XRF). The filters were conditioned for about 48hr in a clean chamber maintained at RH of 40% and temperature of 20°C. Aerosol samples collected on 47-mm diameter quartz fiber filters were analyzed for OC and EC using the NIOSH thermal-optical transmittance (TOT) method. In our protocol, OC and EC were determined as follows. OC was evolved under a stream of high purity He (99.999% higher). The sample was heated in five steps to a final temperature of 800°C in order to volatilize OC and EC in a helium atmosphere. To evolve EC and pyrolyzed OC, which were removed in the second part of the analysis,

the sample was first cooled to 550°C, and then heated under a mixture of 2% O2/98% He, in three temperature steps, to a final temperature of 800°C. The TOT analyzer utilized laser transmission to correct for OC charring. EC was determined as the carbon evolved after the filter transmittance returned to its initial value. The detection limits of OC and EC, which are defined as twice the average of the field blanks, were 0.15 and 0.03 μg C/m3, respectively. The precision of the OC and EC measurements was 4.0 and 7.5%, respectively.

Zefluor filters (Zefluor, 2 μm pore size, Gelman Science, USA) from a sequential sampler were analyzed for eight ionic species (Cl–, NO3

–, SO42–, Na+, NH4

+, K+, Mg2+, and Ca2+). In order to extract the ionic species from the sampled filters, each filter was first put into a 125-mL HDPE vial, and wetted in 0.2 mL of HPLC-grade C2H5OH and then 30 mL of distilled deionized water. The solution was extracted using ultra-sonication for 30 min and mechanically extracted again for 60 min. All the extracts were filtered using 0.4-μm

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 960

Teflon filters and analyzed by ion chromatography (Metrohm Modula IC, model 818 IC pump/819 IC detector). Filter blank values were also applied for background subtraction for each of the filters. The IC system consisted of an anion IC (Metrohm Modula with a suppressor, equipped with a Metrohm Metrosep A Supp-5, 4 × 150 mm, anion column) and a cation IC (Metrohm Modula without a suppressor, equipped with a Metrohm Metrosep C4, 4 × 150 mm, cation column). The eluants were 3.5 mM sodium carbonate (Na2CO3)/1.0 mM sodium bicarbonate (NaHCO3) for the anion IC and 4.0 mM nitric acid (HNO3)/0.7 mM dipicolinic acid for the cation IC. The eluant flow rate for the anion and cation was maintained at 0.7 and 0.9 mL/min, respectively. The measured detection limits for Cl–, NO3

–, SO42–, Na+,

NH4+, K+, Ca2+, and Mg2+ were 12.3, 12.1, 14.9, 11.3, 4.7,

8.5, 12.4, and 4.7 ng/m3, respectively. Semi-Continuous PM2.5 Measurements PM2.5 Mass Concentration

PM2.5 mass concentration was determined continuously every one hour with a beta-attenuation monitor (BAM1020, MetOne Instrument Inc., USA). Integrated 24-h averages of 1-hr PM2.5 data were compared with those from 24-h integrated PM2.5 speciation sampler. The corresponding regression relationships: PM2.5BAM (μg/m3) = (1.02 ± 0.03) PM2.5FRM (μg/m3) + (1.50 ± 1.27), R2 = 0.97, indicated that the continuous BAM1020 measurements provided excellent agreement with 24-h integrated filter-based PM2.5 mass.

Organic and Elemental Carbon Measurements

Concentrations of OC and EC in PM2.5 were measured using a model-4 semi-continuous OC-EC field analyzer (Sunset Laboratory Inc., USA) with 1-hr time resolution. This analyzer is based on the NIOSH method 5040 (NIOSH, 1996). In the carbon analyzer, OC and EC were determined as follows: OC was evolved under a stream of ultrahigh purity He (99.999% higher) while the sample was heated in two temperature steps (600°C for 80 s and 840°C for 90 s). The evolved carbon was oxidized to CO2 in an oxidizing oven. The CO2 is then swept out of the oxidizing oven by a helium stream and measured directly by a self-contained non-dispersive infrared (NDIR) detector system. To evolve EC and pyrolyzed OC, the sample was cooled to 650°C, and then heated under a mixture of He/O2 in two temperature steps (650°C for 35 s and 880°C for 105 s). The analyzer accounted for this pyrolysis by defining the split between the OC and EC as the point when the analysis transmittance is equal to the transmittance before analysis. The EC is determined as the carbon evolved after the filter transmittance returned to its initial value.

Ambient air was drawn at 8.0 L/min through a PM2.5 sharp-cut cyclone. A carbon impregnated multi-channel parallel-plate diffusion denuder was used to remove any semi-volatile organic vapors absorbed on the quartz filter media (Turpin et al., 2000). However, the use of a denuder also disturbs the gas/particle equilibrium at the quartz fiber filter, which can cause OC to volatilize from the filter (Turpin et al., 2000). During the 2-month study period, a total of 1,338 1-hr measurements, out of a possible 1,415, were recorded,

yielding a data capture efficiency of 95.0%. However, 89.0% of these data were flagged as good data and the remaining 11% as invalid because of filter replacement, maintenance, regular check-up, filter blowouts during sampling, power failure, etc. To determine the degree of equivalence, 24-hr integrated OC and EC concentrations (52 measurements available) were regressed against 24-hr averages of the 1-hr OC and EC measurements from the semi-continuous carbon analyzer. In calculating the 24-hr averages, regression analyses were made using only valid OC and EC data. Data pairs were used in the regressions only if < 30% of the 24 1-hr measurements were missing. Regressions were made for 43 data pairs during the study period.

Fig. 2 represents the relationship between 24-hr averages of the 1-hr data and 24-hr integrated data for OC and EC measurements. Sunset OC concentrations were ~27% lower than 24-hr integrated OC concentrations [OCsunset (μg C/m3) = (0.73 ± 0.04)OCfilter (μg C/m3) + (0.05 ± 0.20), R2 = 0.94]. Sunset EC data were also ~27% lower than 24-hr integrated EC concentrations [ECSunset (μg C/m3) = (0.73 ± 0.03)ECfilter (μg C/m3) + (0.12 ± 0.05), R2 = 0.90]. The difference in the OC data could have been due to positive and negative artifacts between the semi-continuous carbon analyzer (operated with the denuder) and the PM2.5 speciation sampler without the denuder (Park et al., 2005b; Polidori et al., 2006), to the different face velocities of air, or to the different TOT temperature programs (Schauer et al., 2003). The face velocity of air passing through the filters was 20.1 and 111.1 cm/s for the speciation sampler and semi-continuous carbon analyzer, respectively. Therefore, the OC data from the PM2.5 speciation sampler were expected to be clearly higher than the semi-continuous measurements. The large discrepancy for the EC data was attributed to the high detection limit of EC in the semi-continuous carbon analyzer at the 1–2 hr time resolution (Lim and Turpin, 2002) and differences in the temperature programs. A previous study has indicated that EC measurements are sensitive to operating conditions despite the use of the same analytical method, i.e., TOT (Schauer et al., 2003). In this work, the original OC and EC data were used to test the mass balance closure of PM2.5 and to estimate the relative contribution of carbonaceous particles to the PM2.5. Ionic Species Measurements

Hourly concentrations of eight ionic species in PM2.5 were measured using an ambient ion monitor (AIM, URG9000D, URG Corporation). The AIM monitor drew air in through a PM2.5 sharp-cut cyclone at a volumetric-flow controlled rate of 3 L/min to remove the large particles from the air stream. In order to minimize the loss of particles during sampling, a Teflon-coated aluminum pipe was set up in the vertical direction. The sample was then drawn through a liquid diffusion denuder where the interfacing acid and alkaline gases were removed. In order to achieve high collection efficiencies, the particle-laden air stream then entered the aerosol super-saturation steam chamber to enhance the particle growth. Grown particles, i.e., droplets, were collected every hour by an inertial particle separator and were then injected into the ion chromatograph. Both particles and

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 961

Fig. 2. Comparison between filter-based and semi-continuous measurements for PM2.5 major chemical components.

gases were collected and injected using syringe pumps, PEEK valves, Teflon® tubing and PEEK tubing. A detailed description of the monitor can be found elsewhere (http://www.urgcorp.com).

The 24-hr averages of hourly Cl–, SO42–, NO3

–, NH4+,

and K+ were regressed against 24-hr integrated ionic

species data measured with a PMS103 speciation sampler. Comparison results between the two methods for SO4

2–, NO3

–, NH4+, and K+ concentrations are shown also in Fig. 2.

Very strong correlations were found between the two methods with R2 of 0.90–0.97. AIM SO4

2– concentrations were about 10% lower than the filter-based SO4

2– measurements with

24-hr average filter-based K+ conc (g/m3)

0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0

24-h

r av

erag

e A

IM K

+ c

onc

( g/

m3 )

0.0

0.4

0.8

1.2

1.6

2.0

2.4

2.8

3.2

3.6

4.0

AIM vs. filter-based K+

Linear fit: y = 090x - 0.02, R2=0.97

24-hr average filter-based OC conc (g C/m3)

0 2 4 6 8 10 12 14 16 18 20

24-h

r av

erag

e S

unse

t OC

con

c (

g C

/m3 )

0

2

4

6

8

10

12

14

16

18

20

Sunset carbon monitor vs filter-based OC

Linear fit: y = 0.73x + 0.05, R2=0.94

24-hr average filter-based SO42- conc (g/m3)

0 3 6 9 12 15 18 21 24 27 30

24-h

r av

erag

e A

IM S

O42-

con

c (

g/m

3 )

0

3

6

9

12

15

18

21

24

27

30

AIM vs filter-based SO42-

Linear fit: y = 0.89x + 0.49, R2=0.90

24-hr average filter-based NO3- conc (g/m3)

0 4 8 12 16 20 24 28 32 36 40

24-h

r av

erag

e A

IM N

O3- c

onc

( g/

m3 )

0

4

8

12

16

20

24

28

32

36

40

AIM vs. filter-based NO3-

Linear fit: y = 0.76x + 0.28, R2=0.95

24-hr average filter-based EC conc (g C/m3)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

24-h

r av

erag

e S

unse

t EC

con

c (

g C

/m3 )

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Sunset carbon monitor vs. filter-based ECLinear fit: y = 0.73x + 0.12, R2=0.90

24-hr average filter-based NH4+ conc (g/m3)

0 2 4 6 8 10 12 14 16 18 20

24-h

r av

erag

e A

IM N

H4+

con

c (

g/m

3 )

0

2

4

6

8

10

12

14

16

18

20

AIM vs. filter-based NH4+

Linear fit: y = 1.00x + 0.00, R2=0.91

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 962

an R2 of 0.90. The SO42– from AIM has been reported to

exhibit a positive interference at SO2 concentration exceeding 30 ppbv and negative readings at SO4

2– levels exceeding 20 µg/m3 (Wu and Wang, 2007). Although not shown here, however, no SO2 interference was observed in the present study. The observed AIM SO4

2– data were corrected based on the regression slope between the two methods to check the mass balance closure of PM2.5. AIM NO3

– concentrations were strongly correlated with the filter-based NO3

– measurements with a slope of 0.76 and an R2 of 0.95, indicating that the AIM NO3

– concentrations were about 24% lower than those from the filter measurements. The higher NO3

– concentrations on the filter samples may have been due to absorption of HNO3 on the filter, causing positive bias. The sampling artifacts of nitrate and ammonium also occurred due to evaporation of semi-volatile NH4NO3 from particles collected on the filter (Koutrakis et al., 1992; Zhang and McMurry, 1992), leading to a negative bias which can be substantial (45–75%) at high temperature and at low aerosol loading (Nie et al., 2010; Saarnio et al., 2010). However, the negative artifact of NO3

– particles in this study may not have been significant because of low ambient temperatures and very high aerosol loading. Low temperature in winter favors the partition of ammonium nitrate in the particulate phase. Hence, the sampling artifacts probably were not a major problem in winter. AIM hourly NO3

– data were corrected using a regression relationship (slope and intercept) between the 24-hr averages of AIM hourly data and 24-hr integrated speciation data and used to investigate the characteristics of ionic species observed during two haze pollution episodes, which are discussed below.

Measurements of Elemental Constituents

Hourly concentrations of elemental constituents (K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Ag, Cd, Sn, Sb, Ba, Ag, Tl, and Pb) in PM2.5 were measured with an ambient metals monitor (model: XactTM 620, Cooper Environmental Services LLC, USA). The Xact 620 monitoring system uses reel-to-reel filter tape sampling and nondestructive XRF (Ed- this acronym has already been defined above) analysis to monitor ambient air. The air was sampled at a flow rate of 16.7 L/min through a PM2.5 inlet and drawn through a filter tape. The resulting PM2.5 deposit was automatically advanced and analyzed by XRF for selected metals as the next sample was being collected. Sampling and analysis were performed continuously and simultaneously, except during advancement of the tape (~20 sec) and during daily automated quality assurance checks. The measurement method of the metal species in ambient air particles was based on EPA Method IO 3.3 (“Determination of metals in ambient PM using XRF). The minimum detection limits of K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Cd, Sn, Sb, Ba, Tl, and Pb for time resolution of 1-hr were 0.81, 0.32, 0.18, 0.14, 0.11, 0.07, 0.08, 0.05, 0.04, 0.06, 0.04, 0.03, 0.03, 1.35, 2.54, 0.67, 0.40, 0.05, and 0.05 ng/m3, respectively. The 24-hr averages of hourly elemental species measurements (K, Ca, Fe, Mn, Cu, Zn, As, and Pb) were compared with those derived from 24-hr integrated filter sample (results from other elemental

species are not provided here because of insufficient paired data). Agreement between elemental species concentrations measured by the two methods showed correlation coefficients (R2) greater than 0.80, with the exception of Fe (R2 = 0.70). Slopes between the measured concentrations from the two methods ranged from 0.74 (Zn) to 1.80 (Ca), with the majority in the range of 0.83–1.15, suggesting high accuracy of continuous XRF measurements for these species. RESULTS AND DISCUSSION Reconstructed PM2.5 Mass Balance Closure

Hourly-based chemical mass balance closure was assessed by comparing BAM1020 PM2.5 mass concentration. The reconstructed PM2.5 mass was computed by summing the concentrations of organic matter (OM), EC, SO4

2–, NO3–,

NH4+, Cl–, crustal material, and trace metals. OM was

obtained by applying a multiplicative factor of 1.6 (i.e., urban aerosols) to OC to account for the oxygen and hydrogen associated with OC (Turpin and Lim, 2001). We estimated the concentration of crustal material by using the following equation reported by Malm et al. (1996): Crustal material (μg/m3) = 2.20[Al] + 2.49[Si] + 1.63[Ca] + 2.42[Fe] + 1.94[Ti]. The sum of the concentrations of all the other elements is reported here as the trace metals value. In this study, because Al and Si data were not available, the contribution of these elements was not included in the estimated concentrations of the crustal material, thus resulting in an underestimation of the crustal material in the reconstructed fine mass. Fig. 3 compares the BAM and reconstructed PM2.5 concentrations. As shown, the reconstructed PM2.5 was highly correlated with the measured BAM PM2.5 concentrations with an R2 of 0.92. However, the reconstructed PM2.5 was about 19% lower than the measured PM2.5. As discussed before, the reconstructed PM2.5 mass concentrations may have been underestimated due to low contributions of crustal materials, OM, and EC to PM2.5 mass and particle water content. Descriptions of Haze Pollution Episodes

As shown in Fig. 3, it is apparent that PM2.5 mass was highly variable on a time scale of 1 hr, whereas the use of 24-hr average largely reduces this variability. For example, the hourly averaged PM2.5 mass concentration on February 4 reached 159.0 μg/m3, i.e., approximately 1.3 times greater than the 24-hr average mass of 122.8 μg/m3. Over the two-month study period, the average 1-hr PM2.5 mass concentration was 44.4 ± 30.0 µg/m3 with a maximum of 159.0 μg/m3, and the 24-hr averaged BAM PM2.5 mass ranged from 14.6–122.8 μg/m3. The 24-hr averaged Korean PM2.5 NAAQS of 50.0 μg/m3 was exceeded on 20 days during the study period. These daily PM2.5 mass concentration ranged from 50.4 μg/m3 on March 20th to 122.8 μg/m3 on February 4th. These pollution episodes were associated with regional haze sulfate and nitrate events, biomass burning emissions, and local traffic emissions under air stagnation and low-level inversions conditions. Out of 20 days exceeding the 24-hr averaged Korean PM2.5 NAAQS, two distinct PM2.5 haze episodes are discussed herein: February 1–8

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 963

Fig. 3. Comparison of hourly measured and reconstructed PM2.5 mass concentrations.

(episode I) and March 11–12 (episode II), 2011. Two haze episodes were classified based on CO/NOx ratios and K+ concentrations. Haze episode I which developed between February 01 and 08, was strongly associated with regional haze pollution along with wildfire smoke emissions over southern China. This episode was further characterized by elevated afternoon SO4

2– and O3 concentrations (O3 = 46–60 ppb on February 01–05). Episode II was from 14:00 March 11–17:00 March 12 during which hourly PM2.5 increased from a background level of 23 μg/m3 to a maximum of 116 μg/m3. This episode was characterized by locally produced pollution with low CO/NOx ratios and broad peaks for OC, EC, and NO3

– concentrations. During episode I, local wind speeds were generally weak to moderate averaging 1.1 m/s (range: 0.1 to 3.7 m/s). Ambient temperatures ranged from –7.4 to 9.9°C (mean: 2.7°C). Relative humidity (RH) had a range sufficient to cause hygroscopic growth (26.1 to 96.6%; mean = 67.1 ± 19.6%), and ranged from > 75–80% in the evening/morning hours. During episode II, temperatures were in the range of –0.1 to 16.2°C (mean = 7.6°C). Winds ranged from 0.1 to 6.3 m/s (mean = 2.2 m/s), and were weak during the night (0.1–1.6 m/s). Relative humidity ranged from 19.8–92.9% with a mean of 56.3%. Daytime RH was low (< 40%), but was > 75% during the night, especially on the evening/morning.

To examine the differences in the chemical composition of PM2.5 observed during the two episodes, we analyzed MODIS images of two types, i.e., Aqua & Terra MODIS image (http://earthdata.nasa.gov/firms) and MODIS true color images (http://lance-modis.eosdis.nasa.gov/cgi-bin/imagery/ realtime.cgi), and transport pathways of the air mass, as illustrated in Fig. 4. The MODIS images show that during episode I, numerous fire counts were observed and a thick haze layer lingered over northern and southern China, and the Korean peninsula. While during episode II, a clear sky was observed over northern China and the Korean peninsula. The transport pathways of the air mass prior to arriving at the site were determined using the Hybrid Single-Particle Lagrangian Integrated Trajectories (HYSPLIT 4.5) model (Draxler and Ralph, 2012). Isentropic five-day backward trajectories were computed for every 12-hr (0000 UTC and 1200 UTC time) interval for each day of the two episodes, at three altitudes (500, 1000, and 1500 m) above ground level (AGL). Two of the trajectory results are also presented in Fig. 4, one for episode I (1200 UTC February 04) and the other for episode II (1200 UTC March 11). For episode I, air masses originating from the polluted regions of northern China (at 500 and 1000 m AGL) or regions of Southern China (at 1500 m AGL), passed over the Yellow Sea and reached the site. Air mass trajectories for episode I

Measurement period (2011)

02/01 02/08 02/15 02/22 03/01 03/08 03/15 03/22 03/29

PM

2.5

mas

s co

ncen

trat

ion

( g/

m3 )

0

20

40

60

80

100

120

140

160

180

200

MetOne BAM PM2.5

Reconstructed PM2.5

MetOne BAM PM2.5 conc (g/m3)

0 20 40 60 80 100 120 140 160 180 200

Rec

onst

ruct

ed P

M2.

5 co

nc (g

/m3 )

0

20

40

60

80

100

120

140

160

180

200

Reconstructed versus BAM PM2.5

Linear fit: y = 0.81x - 2.3, R2=0.91

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 964

Fig. 4. MODIS images and back trajectories of air mass arriving at the site for the two haze episodes. Symbols in trajectories; : 500 m AGL, : 1000 m AGL, : 1500 m AGL.

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 965

indicated a combination of transport directions of northern and southern China and the air mass was classified as a mixture of wild-fire plumes, regionally transported and local pollution. Similar to the transport pathway for episode I, the air mass for episode II originated in the regions of northeastern China and arrived at the site, but without any observable influence of wild-fire plumes.

In this study, ratio of CO to NOx (= NO + NO2) was also used to examine the evolution of the aerosol chemical composition with respect to the age of the air masses as a proxy for proximity to major pollution sources and atmospheric processing (Mogan et al., 2010; Freney et al., 2011). Morgan et al. (2010) classified CO/NOx ratios of 5–15, 10–50, and > 50 as urban, near-source and aged regional conditions, respectively. Temporal variation in the CO/NOx ratio is shown in Fig. 5. The mixing ratios of CO and NOx for episodes I and II were 896 (475–1319 ppb) and 32 (6–139 ppb), and 482 (247–688 ppb) and 40 (7–97 ppb), respectively, with greater impact of CO during episode I. The CO/NOx ratios ranged from 9 to 159 with an average value of 41 ± 26 in episode I and from 6 to 44 with an average of 18 ± 11 in episode II, respectively, confirming that the air masses arriving at the site during episode I were associated with more aged regional components than those during episode II. Fig. 6 shows the mass factions of EC, OM, SO4

2– and NO3– to the PM2.5 mass against the CO/NOx

ratio for the two episodes. As shown in Fig. 6, the significant amounts of EC, OM, SO4

2– and NO3– observed during

episode I were associated with high CO/NOx ratio (> 50). Higher EC and OM contributions to the PM2.5 favor low CO/NOx ratios arising from presumably local emission sources. However, their absolute concentrations were not greatly affected by the CO/NOx ratio (not shown here). That is, it is found that high concentrations of EC, OC, SO4

2– and NO3

– were also related to either the boundary layer air or regional aged air masses.

Table 1 summarizes the PM2.5 concentrations and the chemical components for episodes I and II. As shown in Table 1, as much as 50% of the PM2.5 mass is attributed to secondary SO4

2–, NO3–, and NH4

+ particles, with NO3– being

most significant fraction of the PM2.5 mass for the two episodes. The large difference in chemical compositions between the two haze episodes is due to the NO3

– and K+ concentrations. Fig. 7 shows temporal profiles of PM2.5, OC, EC, K+, and SO4

2– concentrations for the two pollution episodes. Fig. 8 shows correlation relationships between OC and K+, and between K+ and K+/OC ratio for the period of February 04–05. Temporal profiles of NO3

–, NO2 and O3 concentrations, ambient temperature, and RH are shown in Fig. 9. Characteristics of Carbonaceous Particles

OC and EC concentrations were 6.9 (2.6–14.0 μg C/m3) and 2.2 (0.5–4.5 μg C/m3) for episode I and 6.3 (2.0–10.3 μg C/m3) and 2.0 (0.7–5.1 μg C/m3) for episode II, respectively. OM and EC contributed 13.1 ± 2.7% and 3.3 ± 0.7% to the measured PM2.5 for episode I, and 14.2 ± 3.8% and 3.1 ± 0.8% for episode II, respectively. The EC and OC concentrations were strongly correlated with the following regression relationships; for episode I: OC = 2.76 EC + 0.95, R2 = 0.71, for episode II: OC = 3.01 EC + 0.46, R2 = 0.83, suggesting common source emissions (i.e., vehicles) during episodes I and II. The OC/EC ratio was 3.2 ± 0.6 (1.8–5.4) and 3.3 ± 0.6 (2.1–4.8) for episodes I and II, respectively. Diurnal OC and EC peak concentrations over the two episodes occurred mainly during morning and evening rush-hour periods, indicating the significant impact of traffic. As the boundary layer rose throughout the morning and early afternoon, the OC and EC concentrations tended to decrease. However, close inspection of Fig. 7 revealed high background OC and EC levels during episode I with values of 3.0–4.0 and ~2.0 μg C/m3, respectively, probably due to long-range transport of polluted air masses (shown in Fig. 4) and air stagnation conditions. The OC and EC background levels during episode II were approximately 1.0 and 0.5 μg C/m3, respectively.

According to MODIS satellite images and air mass back trajectory analysis (Fig. 4), huge forest wildfires occurred in the southern regions of China on February 01–02, and the air mass passing over these wildfire regions reached the

Fig. 5. Variations of CO/NOx ratio for the two haze episodes.

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

CO

/NO

x ra

tio

(-)

0

20

40

60

80

100

120

140

160

180

200

CO/NOx ratioFebruary 1~8

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

CO

/NO

x ra

tio (

-)

0

20

40

60

80

100

120

140

160

180

200

CO/NOx ratio

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 966

Fig. 6. Fractions of EC, OM, SO4

2–, and NO3– to the PM2.5 against CO/NOx ratio for the two episodes.

sampling site on February 04. Therefore, relationships among OC, K+, and the K+/OC ratio from February 04–05, when the K+ concentration significantly increased, were examined to evaluate the effect of the forest fire plumes on carbonaceous particles observed at the site (Fig. 8). It has been known

that biomass burning emissions contribute significantly to atmospheric OC, EC, and K+ concentrations (Andreae, 1983; Echalar et al., 1995; Andreae and Merlet, 2001; Reid et al., 2005; Park et al., 2006b; Ram and Sarin, 2010, 2011). Vehicles and fossil fuel emissions are the dominant sources

February 01~08

CO/NOx ratio

0 20 40 60 80 100 120 140 160 180 200

EC

/PM

2.5 (

%)

0

1

2

3

4

5

6

7

8

9

10

EC/PM2.5 vs. CO/NOx ratio

March 11~12

CO/NOx ratio

0 10 20 30 40 50 60 70 80 90 100

EC

/PM

2.5 (

%)

0

1

2

3

4

5

6

7

8

9

10

EC/PM2.5 vs. CO/NOx ratio

February 01~08

CO/NOx ratio

0 20 40 60 80 100 120 140 160 180 200

Org

anic

mas

s (O

M)/

PM2.

5 (%

)

0

3

6

9

12

15

18

21

24

27

30

OM/PM2.5 vs. CO/NOx ratio

March 11~12

CO/NOx ratio

0 10 20 30 40 50 60 70 80 90 100

Org

anic

mas

s (O

M)/

PM2.

5 (%

)

0

4

8

12

16

20

24

28

32

36

40

OM/PM2.5 vs. CO/NOx ratio

February 01~08

CO/NOx ratio

0 20 40 60 80 100 120 140 160 180 200

SO42-

/PM

2.5 (

%)

0

7

14

21

28

35

42

49

56

63

70

SO42-/PM2.5 vs. CO/NOx ratio

March 11~12

CO/NOx ratio

0 10 20 30 40 50 60 70 80 90 100

SO

42-/P

M2.

5 (%

)

0

7

14

21

28

35

42

49

56

63

70

SO42-/PM2.5 vs. CO/NOx ratio

March 11~12

CO/NOx ratio

0 10 20 30 40 50 60 70 80 90 100

NO

3- /PM

2.5 (

%)

0

7

14

21

28

35

42

49

56

63

70

NO3-/PM2.5 vs. CO/NOx ratio

February 01~08

CO/NOx ratio

0 20 40 60 80 100 120 140 160 180 200

NO

3- /PM

2.5 (

%)

0

7

14

21

28

35

42

49

56

63

70

NO3-/PM2.5 vs. CO/NOx ratio

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 967

Table 1. Concentration summary of chemical species in PM2.5 for the two episodes.

Parameter Unit Episode I1) Episode II2)

Mean Range Mean Range PM2.5 μg/m3 88 27–159 78 23–116 EC μg C/m3 2.2 0.5–4.5 2.0 0.7–5.1 OC μg C/m3 6.9 2.6–14.0 6.3 2.0–10.3

SO42– μg/m3 18.5 3.1–48.4 16.5 8.4–27.7

NO3– μg/m3 22.3 6.2–53.1 32.3 5.7–49.5

Cl– μg/m3 1.7 0.3–4.8 1.5 0.3–3.0 Na+ μg/m3 0.9 0.6–1.1 0.8 0.7–0.9

NH4+ μg/m3 8.9 4.0–15.1 6.3 2.8–8.2

K+ μg/m3 1.8 0.3–5.6 0.5 0.2–0.7 OC/EC - 3.2 1.8–5.4 3.3 2.1–4.8 K+/OC - 0.26 0.07–0.63 0.10 0.06–0.20 SOR3) - 0.55 0.25–0.76 0.50 0.37–0.71

K μg/m3 2.0 0.3–5.9 0.9 0.4–1.3 Ca μg/m3 0.27 0.03–0.74 0.26 0.14–0.41 Fe μg/m3 0.46 0.09–0.92 0.54 0.24–1.31 Ti ng/m3 17 2–40 15 8–28 V ng/m3 8 3–29 8 2–13 Cr ng/m3 9 2–16 13 4–28 Mn ng/m3 39 10–87 56 21–130 Ni ng/m3 6 3–12 8 2–14 Cu ng/m3 23 6–54 28 11–64 Zn μg/m3 0.13 0.05–0.32 0.30 0.10–0.60 As ng/m3 17 3–34 16 6–28 Se ng/m3 8 2–13 6 2–8 Cd ng/m3 6 1–46 23 2–57 Sb ng/m3 73 23–111 86 46–121 Ba ng/m3 97 11–325 36 17–61 Pb ng/m3 130 28–308 107 41–151

Note) 1)Episode I covers the period of February 01–08, 2011; 2)Episode II covers the period of March 11–12, 2011; 3)SOR indicates SO4

2–/(SO2 + SO42–).

of OC and EC, but have little impact on K+ concentrations. In this study, OC was strongly correlated with EC (R2 = 0.82) during the 2-days period, but the OC/EC ratio (range: 2.8–5.3, mean: 3.4) did not increase, which was significantly lower than those reported for Savannah burning, forest fires, and agricultural waste burning emissions (Andreae, 1983; Park et al., 2005b, 2006b; Reid et al., 2005; Ram and Sarin, 2011; Ram et al., 2012a, 2012b; Zhang et al., 2012b), or slightly lower than that (5.7) from emissions of forest fires in the Indochina (Chuang et al., 2012). Other studies have shown even when biomass materials are burned, a low OC/EC ratio is possible because it depends greatly on the type of materials burned and combustion conditions (Turn et al., 1997; Saarnio et al., 2010; Bae et al., 2012; Zhang et al., 2012a). For example, Turn et al. (1997) indicated that for wood fuels such as Walnut prunings, Almond prunings, and Ponderosa pine slash, the OC/EC ratio in PM2.5 was in the range of 1.7–2.1. It was also found that when pine needles, cherry trees, and acacia trees are combusted, their OC/EC ratios in PM2.5 were 0.83, 0.20, and 0.21, respectively (Bae et al., 2012). As shown in Fig. 7, the concentration of K+, an indicator of biomass burning emissions (Simoneit et al., 2004; Park et al., 2006b; Zhou et al., 2009; Saarnio et al., 2010; Chuang et al., 2012), increased from 1.5 µg/m3 at

00:00 to 5.6 µg/m3 at 12:00 on February 04, and remained high for 2 days. K+ concentration accounted on average 2.9% of the PM2.5 (1.2–4.8%), which was comparable to that (2.5%) reported by a recent study of Chuang et al. (2012). In addition to the biomass burning, the abundance of K in urban air is attributed to soil dust. To assess the origin of K in this study, the relationship between total K and K+ was examined. Previous studies indicated quite low K+/K ratio of 0.1–0.2 in geological material (Gertler et al., 1995) and 0.01–0.02 in Mongolia soil dust (Kim et al., 2010). Water-soluble K+ and total K were strongly correlated (slope = 0.82; R2 = 0.96) for episode I (not shown here), suggesting these two constituents were clearly derived from a similar emission source, most likely forest fire emissions. This was evidenced by the MODIS satellite image, in which numerous fire counts were observed during episode I, and transport pathway of air masses. Scatter plots between OC – K+ and K+ – K+/OC for high K+ episode (February 04–05) were strongly correlated (R2 of 0.77 and 0.77, respectively) (Fig. 8), suggesting an emission source from biomass burning (Zhang et al., 2012b). The K+/OC ratio for the 2-days period showed a range of 0.15–0.63 with a mean of 0.38, which was significantly higher than those reported for biomass burning aerosols. Previous studies have reported K+/OC

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 968

Fig. 7. Temporal trends of PM2.5, OC, EC, K+ and SO4

2– concentrations for the two haze episodes.

ratios from 0.08–0.10 for savanna burning (Echalar et al., 1995), 0.04–0.13 for agricultural waste burning (Andreae and Merlet, 2001), 0.19–0.21 for wheat straw burning/fertilizer (Duan et al., 2004), 0.02-0.14 for biomass burning (Ram and Sarin, 2010, 2011; Ram et al., 2012a), and 0.04–0.24 for wheat and/or stalk burnings (Zhang et al., 2012b). The high K+/OC ratio found in this study may be attributed to traffic and fossil fuels emissions, as well as biomass burning emissions. Accordingly, even though low OC/EC and high K+/OC ratios were observed for the February 04–05 episode,

relatively high K+ concentrations and the significant linear relationships between EC – OC (R2 = 0.82), OC – K+ (R2 = 0.77), and K+ – K+/OC (R2 = 0.77) suggest contribution of carbonaceous aerosols from long-range transport of biomass burning plumes occurring in the southern regions of China during episode I, as well as the traffic emissions.

Sulfate Concentration

The concentrations of SO42– observed during episodes I

and II were 18.5 ± 8.7 μg/m3 (3.1–48.4 μg/m3) and 16.5 ±

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

OC

con

cent

rati

on (g

C/m

3 )

0

2

4

6

8

10

12

14

16

18

20

EC

con

cent

rati

on (g

C/m

3 )

0

1

2

3

4

5

6

7

8

9

10

OCEC

February 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

OC

con

cent

ratio

n (

g C

/m3 )

0

2

4

6

8

10

12

14

16

18

20

EC

con

cent

ratio

n (

g C

/m3 )

0

1

2

3

4

5

6

7

8

9

10

OCEC

Februry 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

PM2.

5 co

ncen

trat

ion

( g/

m3 )

0

20

40

60

80

100

120

140

160

180

200

PM2.5 time-series plot

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

PM

2.5

conc

entr

atio

n (

g/m

3 )

0

20

40

60

80

100

120

140

160

180

200

PM2.5 time-series plot

February 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

SO

42- c

once

ntra

tion

( g/

m3 )

0

6

12

18

24

30

36

42

48

54

60

K+ c

once

ntra

tion

( g/

m3 )

0

1

2

3

4

5

6

7

8

9

10

SO42-

K+ (right "y" axis)March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

SO

42- c

once

ntra

tion

(g

/m3 )

0

6

12

18

24

30

36

42

48

54

60

K+ c

once

ntra

tion

(g

/m3 )

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

SO42-

K+ (right "y" axis)

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 969

Fig. 8. Regression relationships between K+ and OC (left) and between K+/OC and K+ (right) for February 04–05.

5.7 μg/m3 (8.4–27.7 μg/m3), respectively, contributing 20.7 % (6.4–53.4%) and 25.2 % (9.4–53.5%) to the PM2.5. The SO4

2– aerosols in fine particles are formed through gas phase reactions, in-cloud processes and aerosol droplet processes (Guo et al., 2010). The sulfur oxidation ratio [SOR = SO4

2–

/(SO2 + SO42–)] has been used as a good indicator to assess

the atmospheric transformation of SO2 to SO42– (Kaneyasu

et al., 1995; Sahu et al., 2009). SOR in winter is expected to be low due to the low temperature and low photochemical flux, lowering the possibility of gas-phase oxidation of SO2 to SO4

2–. Non-sea salt SO42– was calculated by subtracting the

sea-salt SO42– from the measured SO4

2–. SO42– in seawater

was estimated from the SO42–/Na+ ratio in seawater.

Previously reported 24-hr measurements made at a Gwangju urban site of Korea, which is 10 km away from our site, indicated an SOR of 0.51 in summer and 0.14 in winter (Park et al., 2010, 2012). The calculated SOR was 0.55 ± 0.11 (0.25–0.76) and 0.50 ± 0.09 (0.37–0.71) for episodes I and II, respectively, which were substantially high even in winter. Urban SO4

2– in PM2.5 is composed of urban and regional components. The regional SO4

2– contribution can be evaluated by its diurnal behavior and background levels. For episode I, the temporal profile of SO4

2– concentration (Fig. 7) exhibited a continuous increase in the SO4

2– background level from 3.0 μg/m3 on February 1 to 20.0 μg/m3 on February 5 suggest a greater regional contribution, and are probably due to long range transport and regional sulfate formation, or air stagnation. An obvious diurnal pattern was observed with maximum SO4

2– in the afternoon and the SO42–

transient increased gradually from 26.6 μg/m3 on February 1 to 48.4 μg/m3 on February 4. The maximum hourly O3 and SO2 concentrations on February 1, 2, 3, 4, and 5 were 46 and 13 ppb, 50 and 16 ppb, 58 and 14 ppb, 52 and 19 ppb, and 48 and 12 ppb, respectively, which were relatively high even in winter. The SO4

2– concentration was highly correlated with SO2 with an R2 of 0.64 (data not shown here). This implies that local SO2 emissions were also an important source of SO4

2–formation, and that slow gas-phase oxidation of SO2 probably played a critical role in

the SO42– production. During episode I, the temporal variation

of sulfate concentrations and air mass trajectories suggested that some of the sulfate excess over background could be attributed to sources in the local region. As a result, the high SOR and highly elevated SO4

2– concentration observed during episode I suggest that in addition to local production, SO4

2– was probably caused by long-range secondary atmospheric processing.

For episode II, SO42– concentrations continued to increase

from 14:00 on March 11 until 19:00 to a maximum of 23.6 μg/m3, accounting for 33.7% of the PM2.5, after which the SO4

2– dropped to 9.3 μg/m3 at 07:00 on March 12. At this time it increased again and peaked at 13:00 hr with a maximum of 27.7 μg/m3, constituting 40.2% of the PM2.5, then decreased gradually. Hourly O3 and SO2 concentrations were in the ranges of 47–50 and 5–11 ppb between 14:00 and 19:00 on March 11, respectively. They subsequently decreased to 17 and 2 ppb at 07:00 hr of March 12 and again increased to 66 and 13 ppb at 13:00 hr, respectively. Hourly temperature and RH were 8.2–13.0°C and 20–43% between 14:00 and 19:00 on March 11, and 4.6–15.0°C and 28–75% between 07:00 and 13:00 on March 12, respectively. The SO4

2– peaks occurring in the afternoon hours on both March 11 and 12 coincided with maximum ozone concentrations, thus, both SO4

2– peaks were driven by photochemistry. Throughout episode II, the SO4

2– concentration was strongly correlated with SO2 (R

2 = 0.85) (not shown here). Local wind directions in the afternoon hours were between 230 and 250° (not shown here) where industrial sources are located. This suggests a number of industrial sources located in southwest direction from the site may have contributed to the observed SO4

2– concentration. Consequently, the pattern of these SO4

2– excursions in relation to wind direction and the strong correlation between SO2 and SO4

2– suggests local SO2 emissions were likely an important source of SO4

2– at the site during episode II.

Nitrate Concentration Concentrations of NO3

– particles during episodes I and II

February 04~05

OC concentration (g C/m3)

0 2 4 6 8 10 12 14 16 18 20

Wat

er-s

olub

le K

+ c

once

ntra

tion

(g

/m3 )

0

1

2

3

4

5

6

7

8

9

10

K+ vs. OCLinear fit: y = 0.46x- 0.14, R2=0.77

February 04~05

Water-soluble K+ concentration (g/m3)

0.0 0.7 1.4 2.1 2.8 3.5 4.2 4.9 5.6 6.3 7.0

K+/O

C r

atio

(-)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

K+/OC vs. K+

Linear fit: y = 0.09x + 0.07, R2=0.77

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 970

were 22.3 ± 9.7 μg/m3 (6.2–53.1) and 32.3 ± 13.3 μg/m3 (5.7–49.5), respectively, accounting for 24.2% (9.5–40.4) and 40.9% (19.0–66.3%) of the PM2.5 mass. Atmospheric nitrate particles are mainly formed through gas-phase oxidation process of NO2 and OH during daytime and heterogeneous reactions involving nitrate radical and N2O5 during nighttime (Smith et al., 1995; Wittig et al., 2004; Park et al., 2005c). The nitrate ambient concentration is also driven by gas-to-particle partitioning of ammonium nitrate and boundary layer dynamics (Seinfeld and Pandis, 2006). For episode I, the NO3

– concentration of 7.3 μg/m3 at 10:00 on February 01 was gradually increased to reach a maximum of 53.1 μg/m3 at 12:00 on February 04, accounting for 34.1% of the PM2.5 mass. As shown in Fig. 9, during this period, the NO3

– pollution event exceeding 20 μg/m3 lasted about 5 days and relatively high NO3

– background levels were also maintained, which could be attributed to the thick haze layers lingering over southern and northern China and the Korean peninsula, and to transport pathways of air masses arriving at the site (Fig. 4). Some studies have indicated that a strong diurnal pattern in PM2.5 nitrate concentration was exhibited with a maximum in the morning and minimum in the afternoon (Wittig et al., 2004; Park et al., 2005c). Low nitrate concentrations during the daytime are typically because the nitrate in PM2.5 is mostly in the gas phase as nitric acid due to the higher temperature and lower RH compared to the morning and evening. However, as shown in Fig. 9, morning peaks and afternoon minimums in the nitrate concentrations were not readily observed at the site. Instead, the daytime nitrate transients with very high levels over background were often found around 12:00–13:00. Temperatures, RH, and wind speeds in the afternoon were 4.0–10.0°C, 26–59%, 1.5–2.5 m/s, respectively. The nitrate peaks were more enhanced in the daytime than in the nighttime. The nitrate daytime peak concentrations on February 2, 3, 4, 5 and 6 were 42.0, 40.8, 53.1, 38.0, and 21.4 μg/m3, accounting for 40.4, 40.0, 34.1, 27.0, and 24.3% of the PM2.5, respectively. Periods showing these peaks were reasonably consistent with those showing high O3 concentrations. Concentrations of NO2 and O3 at these periods were 34 and 41, 15 and 47, 21 and 48, 11 and 48, and 11 and 60 ppb, respectively. These O3 levels were comparable to the highest O3 concentrations on these days, which were observed around 15:00–17:00, with levels of 50, 58, 54, 48, and 60 ppb, respectively. During the daytime excursion periods, the concentrations of NOx and carbonaceous species (OC and EC) were reduced to their background levels on the corresponding day. Typically the afternoon peak concentrations of fine nitrate particles have been attributed to the photochemical formation of nitric acid due to the increased ozone concentration, especially in summertime (Seinfeld and Pandis, 2006). However, the present situation, in which ozone levels of 40-60 ppb were observed in winter, facilitated the formation of nitric acid through gas-phase oxidation of NO2. These daytime nitrate transients during episode I were probably associated with transports of severe haze lingering over the upwind locations of China (Fig. 4) to the sampling site, as well as photochemical secondary aerosol production of nitrate and

low temperatures of 4.0–10°C. This hypothesis is further supported by the relatively high CO/NOx ratio, as illustrated in Fig. 5. A closer look at the CO/NOx ratio for episode I shows that the ratios at the times of the elevated NO3

– levels were 46–80, 44–78, 42–91, 81–93, and 72–91, suggesting that the air masses arriving at the site were in regional aged conditions. Evening NO3

– peaks were observed four times, at 20:00 on each of 02, 04, 05, and 07 February, with levels of 26.9, 38.3, 22.1, and 24.9 μg/m3, accounting for 26.1, 24.1, 25.1, and 26.3% of the PM2.5, respectively. NO2 concentrations at the times of these nighttime peaks were 26, 25, 35, and 43.3 ppb, respectively. The increases in NO3

– in the evening were probably related to the lower mass concentrations of sulfate at night, which allowed excess NH3 to react with HNO3 (Seinfeld and Pandis, 2006). These evening peaks coincided with the enhanced NO2 concentrations, but the ambient concentrations of the nitrate were not necessarily proportional to the concentrations of precursor emissions because the rates at which they form and their gas-to-particle partitioning are controlled by meteorological parameters other than the concentration of the precursor gas. A similar result was found elsewhere (Park et al., 2005c). After its evening peaks, nitrate decreased for some time, but was maintained at high levels due to continuous haze phenomena and poor atmospheric dispersion conditions at night. Because the RH during the nighttime (70–93%) was high enough to form aqueous nitric acid, the nitrate observed at night may have accumulated in wet aerosol particles. Thus, the significantly elevated concentrations of NO3

– particles observed during the nighttime may have been related to low ambient temperatures, high RH, and the regional aged air masses. In summary, the high NO3

– values during episode I probably reflect long-range secondary atmospheric processing of nitrate particles, in addition to local emission sources.

Episode II was characterized by a rapid increase of nitrate starting at 14:00 on March 11 and persisting until 12:00 on March 12, with a small drop in concentration between 01:00 and 07:00 on March 12. As shown in Figs. 7 and 9, OC, EC, and NO3

– concentrations peaked in the morning of March 11 prior to the onset of episode II, and were probably associated with morning rush hour traffic patterns. The enhancement of morning NO3

– concentration, to a maximum of 18.4 μg/m3 at 09:00, was attributed to high RH and low temperatures. RH and temperatures between 02:00 and 09:00 on March 11 were 84–93% and 0.0–1.0°C, respectively. After 09:00, the NO3

– concentration decreased to 4.6 μg/m3 at 13:00, started to increase again until 20:00–21:00, with a maximum of 49.5 μg/m3, accounting for 46.3% of the PM2.5, and then decreased to about 36.0 μg/m3, which was still high. From 14:00 to 21:00 on March 11, as shown in Fig. 7, the temporal behaviors of OC, EC, and SO4

2– concentrations were similar to that of NO3

–, probably due to local emissions. Moreover, NO2 increased from 12 to 57 ppb, and O3 was maintained at 40–50 ppb until 20:00 hr, but then was sharply reduced to 10–20 ppb (Fig. 9). Hourly ambient temperature and RH were 7.8–13.0°C and 24–48%, which is below the deliquescence RH (DRH) of NH4NO3 (61% @25°C), respectively. Hourly wind speed was 2.2–4.6 m/s. Based on

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 971

Fig. 9. Temporal profiles of NO3

– concentrations along with NO2, O3, temperature, and relative humidity for the two haze episodes.

the concentrations of precursor gases and meteorological conditions, the continuous increase of NO3

– concentration between 14:00 and 20:00 on March 11 was probably due to gas-phase production of HNO3 via the reaction of NO2 and OH at low RH (< 50%), with the majority of the HNO3 residing in the particle phase. In addition to the photochemical production of HNO3, air masses transporting from regions of northern China to the site could have partially contributed to the enhanced NO3

– concentrations in the afternoon (see Fig. 4). As shown in Fig. 9, the nitrate concentrations

remained high between 22:00 March 11 and 07:00 March 12, ranging from 36.2 to 46.9 μg/m3. This NO3

– excursion was associated with low NO/high NO2, and high RH, following high afternoon O3 levels of about 50 ppb, presumably due to nighttime radical chemistry. During these periods, the wind speed was 0.1–2.2 m/s with a mean of 1.0 m/s, and RH and the temperature were 72–79% and 3.4–8.4°C, respectively. The mixing ratios of NO and NO2 at 22:00 were 29 and 73 ppb, respectively. Heterogeneous reactions involving nitrate radical and N2O5, which are

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

Am

bien

t tem

pera

ture

(de

g C

)

-9

-6

-3

0

3

6

9

12

15

18

21

Rel

ativ

e hu

mid

ity (

%)

20

30

40

50

60

70

80

90

100

110

120

Ambient temperatureRelative humidity

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

NO

2 an

d O

3 co

ncen

trat

ions

(pp

b)

0

10

20

30

40

50

60

70

80

90

100

NO2

O3

February 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

Am

bien

t tem

pera

ture

(de

g C

)

-9

-6

-3

0

3

6

9

12

15

18

21

Rel

ativ

e hu

mid

ity

(%)

20

30

40

50

60

70

80

90

100

110

120

Ambient temperatureRelative humidity

February 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

NO

2 an

d O

3 co

ncen

trat

ions

(pp

b)

0

10

20

30

40

50

60

70

80

90

100

NO2

O3

February 01~08

Measurement period (2011)

02/01 02/02 02/03 02/04 02/05 02/06 02/07 02/08 02/09

SO

42- a

nd N

O3- c

onc

( g/

m3 )

0

6

12

18

24

30

36

42

48

54

60

NO3-

March 11~12

Measurement period (2011)

03/11 00 03/11 08 03/11 16 03/12 00 03/12 08 03/12 16 03/13 00

SO

42-an

d N

O3- c

onc

(g/

m3 )

0

6

12

18

24

30

36

42

48

54

60

NO3-

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 972

formed during nighttime (Smith et al., 1995), are a potential source of the nighttime nitrate. Nitric acid formed through the heterogeneous hydrolysis of N2O5 at night can retain its molecular form in the gas phase, or be transferred to the particulate phase. The NO3– observed between 22:00 and 07:00 may have existed in the aqueous phase because of the RH of 72–79%. The consequently elevated NO3

–, which induced the high PM2.5 levels during episode II, was attributed to gas-phase photochemical reactions during the daytime and heterogeneous hydrolysis during the nighttime. Possible Sources of Crustal and Trace Metals in Ambient Particles

One of the major anthropogenic sources in urban air is vehicle emissions. Local sources have a greater effect on EC concentrations. Variations in local diesel engines or combustion sources could affect EC at a site. Like EC concentrations, ambient concentrations of metals were dominated by local sources. Road dust is generally re-suspended by the turbulent passage of motor vehicles over local roads. The road dust profile contains crustal elements (Al, Si, Ca, Fe), along with metal constituents such as Mo, Sb, Cu, Zn, Ba, Mn, Cr, and Pb, reflecting enrichment from other sources and urban pollutants (Schauer et al., 2006). These metals are known to be emitted from road traffic abrasion processes or are important constituents of road dust (Harrison et al., 2003; Schauer et al., 2006; Bukowiecki et al., 2009; Amato et al., 2010). Brake wear emissions have been reported to contain significant amounts of Ba, Zn, Cu, and Sb, as well as Fe and other crustal elements (Garg et al., 2000; Sternbeck et al., 2002; Thorpe and Harrison, 2008). Previous studies indicated also that Sb in urban environments is associated with traffic emissions, and more specifically with emissions emerging from brake wear (Weckwerth, 2001; Furuta et al., 2005; Gomez et al., 2005; Johansson et al., 2009).

As shown in Table 1, the concentration differences of Sb and the other brake wear-related elements (Fe, Cr, Mn, Cu, and Pb) between episodes I and II were not significant. The concentration of Sb was 73 ± 16 ng/m3 (22.5–111.0) and 86 ± 19 ng/m3 (46–121) for episodes I and II, respectively. The mean concentration of Fe, one of the most abundant measured elements, was similar during the two episodes, ranging from 0.09 to 0.92 µg/m3 with a mean of 0.46 µg/m3 and from 0.24 to 1.31 µg/m3 with a mean of 0.54 µg/m3, respectively. However, Ba, another marker of brake wear, was about three times higher in episode I (97 ng/m3) than in episode II (36 ng/m3), while Zn was two times higher in episode II (0.30 µg/m3) than in episode I (0.13 µg/m3). This suggests that in addition to the road dust, there were relevant local emissions of these elements at the site. Taking EC, Fe, and Sb as measures of road traffic influence, correlation analyses among EC, crustal elements and other road traffic-related metals were performed and their results are presented in Table 2. During episode I, Sb was correlated with Fe (R2 = 0.71), Ca (R2 = 0.67), Cr (R2 = 0.64), and Mn (R2 = 0.53) and to a lesser extent with Zn (R2 = 0.42), Pb (R2 = 0.41), Cu (R2 = 0.34), and Ba (R2 = 0.26), which is suggestive of other sources such as road dust for these elements. During

episode II, Sb had strong correlations with Cr, Ca, Ba, Zn, Pb, Mn, Fe, and Cu (R2 = 0.88, 0.86, 0.83, 0.77, 0.76, 0.76, 0.64, and 0.46, respectively). Ca, Fe, Cr, Mn and Pb are typically present at significant levels in crustal material and soils, but the relatively high correlations between Sb and the metals indicated that the concentrations of these metals must be affected by local sources other than soil. Also EC had moderate correlations with the metals during episodes I and II. Because re-suspended road dust is generally too large to dominate PM2.5, the good relations of Sb and EC with the other metals indicate that tailpipe and brake wear emissions were important sources for these elements at the site, with a greater effect in episode II. Sb and Cd were very weakly correlated (R2 = 0.04) during episode II, indicating that other sources, rather than vehicle-related emissions, dominated the concentration of Cd at the site. SUMMARY AND CONCLUSION

Hourly measurements of PM2.5 mass, elemental and organic carbon (EC and OC), ionic species, and elemental species were made at the South Area Supersite at Gwangju, Korea, using commercial semi-continuous instruments. During the 2 months between February 01 and March 31, 2011, a total of 20 days wherein daily PM2.5 mass concentrations exceeding 24-hr average Korean NAAQS of 50.0 μg/m3 were identified. Two haze episodes out of them were discussed; the episode I (February 01–08) was associated with regional pollution along with severe haze layers and wildfires emissions, and characterized further by high CO/NOx ratios and high K+ concentrations. The episode II (March 11–12) was characterized by locally produced pollution with low CO/NOx ratios, short-term SO4

2– excursions, and broad peaks in OC, EC, and NO3

– concentrations. The EC and OC concentrations were strongly correlated

with R2 of 0.71 and 0.83 for episodes I and II, respectively, suggesting common source emissions (i.e., vehicles). Their respective OC/EC ratio was 3.2 and 3.3. However, the concentration of K+, an indicator of biomass burning emissions, was increased for February 04–05 during the episode I and reached an hourly maximum of 5.6 μg/m3, accounting for ~93% of the total K concentration. This enhancement was clearly evidenced by the MODIS satellite image, in which numerous fire counts were observed over the southern regions of China, and transport pathways of air masses arriving at the site. Abnormally, however, the OC/EC ratio (mean: 3.4) did not increase for the 2-days period. Previous studies suggested that a low OC/EC ratio is possible because it depends greatly on the type of materials burned and combustion conditions. Accordingly, although low OC/EC ratio was observed when relatively high K+ concentrations occurred, the significant linear relationships between EC – OC (R2 = 0.82), OC – K+ (R2 = 0.77), and K+ – K+/OC (R2 = 0.77) between February 04 and 05 suggest contribution of carbonaceous aerosols from long-range transport of biomass burning plumes occurring in the southern regions of China during episode I, as well as the traffic emissions.

The good correlations between SO2 and SO42– (R2 = 0.64

and 0.85 for episodes I and II) and local wind directions

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 973

Table 2. Correlation coefficients among EC, crustal elements, and metals.

(a) Episode I Component EC Ca Fe Ti Cr Mn Cu Zn Sb Ba Pb

EC - 0.44 0.64 0.56 0.58 0.36 0.51 0.62 0.51 0.33 0.74 Ca 0.44 - 0.76 0.91 0.80 0.70 0.63 0.54 0.82 0.67 0.71 Fe 0.64 0.76 - 0.81 0.87 0.84 0.61 0.82 0.84 0.54 0.74 Sb 0.51 0.82 0.84 0.79 0.80 0.73 0.58 0.65 - 0.51 0.64 Ba 0.33 0.67 0.54 0.46 0.82 0.64 0.90 0.45 0.51 - 0.66

(b) Episode II Component EC Ca Fe Ti Cr Mn Cu Zn Cd Sb Ba Pb

EC - 0.52 0.63 0.62 0.65 0.64 0.55 0.72 0.35 0.66 0.76 0.69 Ca 0.52 - 0.74 0.92 0.88 0.81 0.46 0.74 0.15 0.93 0.81 0.68 Fe 0.63 0.74 - 0.88 0.91 0.97 0.50 0.85 0.64 0.80 0.89 0.68 Sb 0.66 0.93 0.80 0.89 0.94 0.87 0.68 0.88 0.20 - 0.91 0.87 Ba 0.76 0.81 0.89 0.90 0.94 0.91 0.60 0.88 0.35 0.91** - 0.78

All figures are significant at p < 0.01.

suggest that local SO2 emissions for the two episodes were an important source of SO4

2– formation, and that the gas-phase oxidation of SO2 due to high O3 probably played a critical role in the SO4

2– production. The time-series plot of SO42–,

the SO2-to-SO42– relationship, local wind directions, and

the air mass trajectories indicated that regional contributions during episode I probably dominated increased SO4

2– concentrations, while the local urban contribution dominated the SO4

2– concentration during episode II. For NO3– particles,

the daytime NO3– concentration was often significantly

elevated over background levels during episode I, which suggests that these particles can be attributed to the transport of severe haze lingering over the Chinese locations upwind of the sampling site, as well as the photochemical production of nitric acid and low temperatures of 4.0–10°C. Similar to the formation processes for episode I, the elevated NO3

– leading to high PM2.5 during episode II was attributed to gas-phase photochemical reactions during the daytime and heterogeneous hydrolysis during the nighttime.

To investigate the influence of road traffic on the concentrations of elemental constituents observed in PM2.5 during the two episodes, correlation analyses among crustal elements (Ca, Fe, and Mn) and road traffic-related metals (Sb, Cu, Zn, Ba, Mn, Cr, and Pb) were performed. Good correlations of Sb, a specific marker for brake wear, with crustal elements and other metals revealed brake wear emission to be an important source for these elements during episodes I and II, with greater road traffic influence in episode II than in episode I. These metal correlation analyses support the conclusion that the enrichment of crustal and trace elements at sites close to the roadside can be mainly attributed to vehicle wear products rather than tailpipe exhaust emissions. ACKNOWLEDGEMENTS

This work was supported by project “Investigation on chemical characteristics of PM2.5 and its formation processes by areas” funded from the National Institute of Environmental Research. This work was also partially supported by the

General Researcher Program through a NRF grant funded by the Korea government (MEST) (No. 2011-0007222). The authors also gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www. arl.noaa.gov/ready.php) used in this publication. REFERENCES Amato, F., Nava, S., Lucarelli, F., Querol, X., Alastuey, A.,

Baldasano, J.M. and Pandolfi, M. (2010). Comprehensive Assessment of PM Emissions from Paved Roads: Real-world Emission Factors and Intense Street Cleaning Trials. Sci. Total Environ. 408: 4309–4318.

Andreae, M.O. (1983). Soot Carbon and Excess Fine Potassium: Long Range Transport of Combustion-Derived Aerosols. Science 220: 1148–1151.

Andreae, M.O. and Merlet, P. (2001). Emission of Trace Gases and Aerosols from Biomass Burning. Global Biogeochem. Cycles 15: 955–966.

Bae, M.S., Shin, J.S., Lee, K.Y. and Kim, Y.J. (2012) Characteristics of Carbonaceous Aerosols Analyzed by the Analysis of Organic Molecular Markers at Gosan Supersite, Proce. 54th Meeting of KOSAE, p. 30.

Bukowiecki, N., Lienemann, P., Hill, M., Figi, R., Richard, A., Furger, M., Rickers, K., Falkenberg, G., Zhao, Y., Cliff, S.S., Prevot, A.S.H., Baltensperger, U., Buchmann, B. and Gehrig, R. (2009). Real-world Emission Factors for Antimony and Other Brake Wear Related Trace Elements: Size-Segregated Values for Light and Heavy Duty Vehicles. Environ. Sci. Technol. 43: 8072–8078.

Chameides, W.L., Yu, H., Liu, S.C., Bergin, M., Zhou, X., Mearns, L., Wang, G., Kiang, C.S., Saylor, R.D., Luo, C., Haung, Y., Steiner, A. and Giorgi, F. (1999). Case Study of the Effects of Atmospheric Aerosols and Regional Haze on Agriculture: An Opportunity to Enhance Crop Yields in China through Emission Controls? Proc. Nat. Acad. Sci. U.S.A. 96: 13626–13633.

Charlson, R.J., Lovelock, J.E., Andreae, M.O. and Warren, S.G. (1987). Oceanic Phytoplankton, Atmospheric Sulfur,

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 974

Cloud Albedo and Climate. Nature 326: 655–661. Chen, L.W.A., Chow, J.C., Doddridge, B.G., Dickerson,

R.R., Ryan, W.F. and Mueller, P.K. (2003). Analysis of a Summertime PM2.5 and Haze Episode in the Mid-Atlantic Region. J. Air Waste Manage. Assoc. 53: 946–956.

Chuang, M.T., Chou, C.C.K., Sopajareepom, K., Lin, NH., Wang, J.L., Sheu, G.R., Chang, Y.J. and Lee, C.T. (2012) Characterization of Aerosol Chemical Properties from Near-Source Biomass Burning in the Northern Indochina during 7-SEAS/Dongsha Experiment. Atmos. Environ., In Press, http://dx.doi.org/10.1016/j.atmosenv.2012. 06.056.

Draxler, R.R. and Rolph, G.D. (2012). HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model Access via NOAA ARL READY Website (http://ready. arl.noaa.gov/HYSPLIT.php), NOAA Air Resources Laboratory, Silver Spring, MD.

Duan, F., Liu, X., Yu, T. and Cachier, H. (2004). Identification and Estimate of Biomass Burning Contribution to the Urban Aerosol Organic Carbon Concentrations in Beijing. Atmos. Environ. 38: 1275–1282.

Echalar, F., Gaudichet, A., Cachier, H. and Artaxo, P. (1995). Aerosol Emissions by Tropical Forest and Savanna Biomass Burning: Characteristic Trace Elements and Fluxes. Geophys. Res. Lett. 22: 3034–3042.

Freney, E.J., Sellegri1, K., Canonaco, F., Boulon1, J. Hervo1, M., Weigel, R., Pichon, J.M., Colomb, A., Prevot, A.S.H. and Laj, P. (2011). Seasonal Variations in Aerosol Particle Composition at the Puy-de-Dome Research Station. Atmos. Chem. Phys. Discuss. 11: 27139–27170.

Furuta, N., Iijima, A., Kambe, A., Sakai, K. and Sato, K. (2005). Concentrations, Enrichment and Predominant Sources of Sb and Other Trace Elements in Size Classified Airborne Particulate Matter Collected in Tokyo from 1995 to 2004. J. Environ. Monit. 7: 1155–1161.

Garg, B.D., Cadle, S.H., Mulawa, P.A., Groblicki, P.J., Laroo, C. and Parr, GA. (2000). Brake Wear Particulate Matter Emissions. Environ. Sci. Technol. 34: 4463–4469.

Gertler, A.W., Lowenthal, D.A. and Coulombe, W.G. (1995). PM10 Source Apportionment Study in Bullhead City, Arizona. J. Air Waste Manage. Assoc. 45: 75–82.

Gomez, D.R., Gine, M.F., Bellato, A.C.S. and Smichowski, P. (2005). Antimony: A Traffic-Related Element in the Atmosphere of Buenos Aires, Argentina. J. Environ. Monit. 7: 1162–1168.

Guo, S., Hu, M., Wang, Z.B., Slanina, J. and Zhao, Y.L. (2010). Size-Resolved Water-Soluble Ionic Compositions in the Summer of Beijing: Implication of Regional Secondary Formation. Atmos. Chem. Phys. 10: 947–959.

Harrison, R.M., Tilling, R., Romero, M.S.C., Harrad, S. and Jarvis, K. (2003). A Study of Trace Metals and Polycyclic Aromatic Hydrocarbons in the Roadside Environment. Atmos. Environ. 37: 2391–2402.

Johansson, G., Norman, M. and Burman, L. (2009). Road Traffic Emission Factors for Heavy Metals. Atmos. Environ. 43: 4681–4688.

Jung, J. and Kim, Y.J. (2011). Tracking Sources of Severe Haze Episodes and Their Physicochemical and Hygroscopic Properties under Asian Continental Outflow: Long-Range Transport Pollution, Post-harvest Biomass

Burning, and Asian Dust. J. Geophys. Res. 116: D02206. Jung, J., Kim, Y.J., Lee, K.Y., Cayetano, M.G., Batmunkh,

T., Koo, J.H. and Kim, J. (2010). Spectral Optical Properties of Long-Range Transport Asian Dust and Pollution Aerosols over Northeast Asia in 2007 and 2008. Atmos. Chem. Phys. 10: 5391–5408.

Kaneyasu, N., Ohta, S. and Murao, N. (1995). Seasonal Variation in the Chemical Composition of Atmospheric Aerosols and Gaseous Species in Sapporo, Japan. Atmos. Environ. 29: 1559–1568.

Kang, C.M., Lee, H.S., Kang, B.W., Lee, S.K. and Sunwoo, Y. (2004). Chemical Characteristics of Acidic Gas Pollutants and PM2.5 Species during Hazy Episodes in Seoul, South Korea. Atmos. Environ. 38: 4749–4760.

Kim, D.R., Kim, J.S. and Ban, S.J. (2010). A Study on the Characteristics of Soil in the Asian Dust Source Regions of Mongolia. J. Korean Soc. Atmos. Environ. 26: 606–615 (in Korean).

Kim, K.W., Kim, Y.J. and Oh, S.J. (2001). Visibility Impairment during Yellow Sand Periods in the Urban Atmosphere of Kwangju, Korea. Atmos. Environ. 35: 5157–5167.

Kim, K.W., Kim, Y.J. and Bang, S.Y. (2008). Summer Time Haze Characteristics of the Urban Atmosphere of Gwangju and the Rural Atmosphere of Anmyon, Korea. Environ. Monit. Assess. 141: 189–199.

Koutrakis, P., Thompson, K.M., Wolfson, J.M., Spengler, J.D., Keeler, G.J. and Slater, J.L. (1992). Determination of Aerosol Strong Acidity Losses due to Interactions of Collected Particles: Results from Laboratory and Field Studies. Atmos. Environ. 26A: 987–995.

Lee, Y.L. and Sequeira, R. (2001). Visibility Degradation across Hong Kong: Its Components and Their Relative Contributions. Atmos. Environ. 34: 5861–5872.

Lim, H.J. and Turpin, B.J. (2002). Origins of Primary and Secondary Organic Aerosol in Atlanta: Results of Time-Resolved Measurements during the Atlanta Supersite Experiment. Environ. Sci. Technol. 36: 4489–4496.

Malm, W.C., Molenar, J.V., Eldred, R.A. and Sisler, J.F. (1996). Examining the Relationship among Atmospheric Aerosols and Light Scattering and Extinction in the Grand Canyon Area. J. Geophys. Res. 101: 19251–19265.

Menon, S., Hansen, J., Nazarenko, L. and Luo, Y. (2002). Climate Effects of Black Carbon Aerosols in China and India. Science 297: 2250–2253.

Morgan, W.T., Allan, J.D., Bower, K.N., Highwood, E.J., Liu, D., McMeeking, G.R., Northway, M.J., Williams, P.I., Krejci, R. and Coe, H. (2010). Airborne Measurements of the Spatial Distribution of Aerosol Chemical Composition across Europe and Evolution of the Organic Fraction. Atmos. Chem. Phys. 10: 4065–4083.

National Institute of Occupational Safety and Health (NIOSH) (1996). Method 5040 Issue 1: Elemental Carbon (Diesel Exhaust), NIOSH Manual of Analytical Methods, 4th ed., Cincinnati, OH.

Nie, W., Wang, T., Gao, X., Pathak, R.K., Wang, X., Gao, R., Zhang, Q., Yang, L. and Wang, W. (2010). Comparison among Filter-Based, Impactor-Based and Continuous Techniques for Measuring Atmospheric Fine Sulfate and

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 975

Nitrate. Atmos. Environ. 44: 4396–4403. Oanh, N.T.K. and Leelasakultum, K. (2011). Analysis of

Meteorology and Emission in Haze Episode Prevalence over Mountain-Bounded Region for Early Warning, Sci. Total Environ. 409: 2261–2271.

Okada, K., Ikegami, M., Zaizen, Y., Makino, Y., Jensen, J.B. and Gras, J.L. (2001). The Mixture State of Individual Aerosol Particles in the 1997 Indonesian Haze Episode. J. Aerosol Sci. 32: 1269–1279.

Park, S.S., Bae, M.S., Schauer, J.J., Ryu S.Y., Kim, Y.J., Cho, S.Y. and Kim, S.J. (2005a). Evaluation of the TMO and TOT Methods for OC and EC Measurements and Their Characteristics in PM2.5 at an Urban Site of Korea during ACE-Asia. Atmos. Environ. 39: 5101–5112

Park, S.S., Harrison, D., Pancras, P. and Ondov, J.M. (2005b). Time Resolved Elemental and Organic Carbon Measurements at the Baltimore Supersite in 2002. J. Geophys. Res. 110: D07S06, doi: 10.1029/2004JD004610.

Park, S.S., Ondov. J.M., Harrison, D. and Nair, N.P. (2005c). Seasonal and Shorter-term Variations in Particulate Atmospheric Nitrate in Baltimore. Atmos. Environ. 39: 2011–2020.

Park, S.S., Kleissl, J., Harrison, D., Kumar, V., Nair, N.P., Adam, M., Ondov, J. and Parlange, M. (2006a) Characteristics of PM2.5 Episodes Revealed by Semi-Continuous Measurements at the Baltimore Supersite at Ponca St. Aerosol Sci. Technol. 40:845–860.

Park, S.S., Bae, M.S., Schauer, J.J., Kim, Y.J., Cho, S.Y. and Kim, S.J. (2006b). Molecular Composition of PM2.5 Organic Aerosol Measured at an Urban Site of Korea during the ACE-Asia Campaign. Atmos. Environ. 40: 4182–4198.

Park, S.S., Cho, S.Y. and Kim, S.J. (2010). Chemical Characteristics of Water-Soluble Components in Fine Particulate Matter at a Gwangju Area. Korean J. Chem. Eng. 48: 20–26 (in Korean).

Park, S.S. and Cho, S.Y. (2011). Tracking Sources and Behaviors of Water-Soluble Organic Carbon in Fine Particulate Matter Measured at an Urban Site in Korea. Atmos. Environ. 45: 60–72.

Park, S.S., Kim, J.H. and Jeong, J.U. (2012). Abundance and Sources of Hydrophilic and Hydrophobic Water-Soluble Organic Carbon at an Urban Site in Korea in Summer. J. Environ. Monit. 14: 224–232.

Polidori, A., Turpin, B.J., Lim, H.J., Cabada, J.C., Subramanian, R., Pandis, S.N. and Robinson, A.L. (2006). Local and Regional Secondary Organic Aerosol: Insights from a Year of Semi-Continuous Carbon Measurements at Pittsburgh. Aerosol Sci. Technol. 40: 861–872.

Ram, K. and Sarin, M.M. (2010). Spatio-Temporal Variability in Atmospheric Abundances of EC, OC and WSOC over Northern India. J. Aerosol Sci. 41: 88–98.

Ram, K. and Sarin, M.M. (2011). Day-Night Variability of EC, OC, WSOC and Inorganic Ions in Urban Environment of Indo-Gangetic Plain: Implications to Secondary Aerosol Formation. Atmos. Environ. 45: 460–468.

Ram, K., Sarin, M.M. and Tripathi, S.N. (2012a). Temporal Trends in Atmospheric PM2.5, PM10, Elemental Carbon, Organic Carbon, Water-Soluble Organic Carbon, and

Optical Properties: Impact of Biomass Burning Emissions in the Indo-Gangetic Plain. Environ. Sci. Technol. 46: 686–695.

Ram, K., Sarin, M.M., Sudheer, A.K. and Rengarajan, R. (2012b). Carbonaceous and Secondary Inorganic Aerosols during Wintertime Fog and Haze over Urban Sites in the Indo-Gangetic Plain. Aerosol Air Qual. Res. 12: 359–370.

Reid, J.S., Koppmann, R., Eck, T.F. and Eleuterio, D.P. (2005). A Review of Biomass Burning Emissions Part II: Intensive Physical Properties of Biomass Burning Particles. Atmos. Chem. Phys. 5: 799–825.

Ryu, S.Y., Kim, J.E., Kim, Y.J. and Gang, G.U. (2004). Chemical Composition of Post-Harvest Biomass Burning Aerosols in Gwangju, Korea. J. Air Waste Manage. Assoc. 54: 1124–1137.

Ryu, S.Y., Kwon, B.G., Kim, Y.J., Kim, H.H. and Chun, K.J. (2007). Characteristics of Biomass Burning Aerosol and Its Impact on Regional Air Quality in the Summer of 2003 at Gwangju, Korea. Atmos. Res. 84: 362–373.

Saarnio, K., Aurela, M., Timonen, H., Saarikoski, S., Teinila, K., Makela, T., Sofiev, M., Koskinen, J., Aalto, P.P., Kulmala, M., Kukkonen, J. and Hillamo, R. (2010). Chemical Composition of Fine Particles in Fresh Smoke Plumes from Boreal Wild-land Fires in Europe. Sci. Total Environ. 408: 2527–2542.

Sahu, L.K., Kondo, Y., Miyazaki, Y., Kuwata, M., Koike, M., Takegawa, N., Tanimoto, H., Matsueda, H., Yoon, S.C. and Kim, Y.J. (2009). Anthropogenic Aerosols Observed in Asian Continental Outflow at Jeju Island, Korea, in Spring 2005. J. Geophys. Res. 114: D03301, doi: 10.1029/2008JD010306.

Schauer, J.J., Mader, B.T., Deminter, J.T., Heidemann, G., Bae, M.S., Seinfeld, J.H., Flagan, R.C., Cary, R.A., Smith, D., Huebert, B.J., Bertram, T., Howell, S., Kline, J.T., Quinn, P., Bates, T., Turpin, B., Lim, H.J., Yu, J.Z., Yang, H. and Keywood, M.D. (2003). ACE-Asia Intercomparison of a Thermal-Optical Method for the Determination of Particle-Phase Organic and Elemental Carbon. Environ. Sci. Technol. 37: 993–1001.

Schauer, J.J., Lough, G.C., Shafer, M.M., Christensen, W.F., Arndt, M.F., DeMinter, J.T. and Park, J.S. (2006). Characterization of Metals Emitted from Motor Vehicles, Health Effect Institute.

Schichtel, B.A., Husar, R.B., Falke, S.R. and Wilson,W.E. (2001). Haze Trends over the United States, 1980–1995. Atmos. Environ. 35: 5205–5210.

Seinfeld, J.H. and Pandis, S.N. (2006). Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, 2nd Ed., John Wiley & Sons, Inc.

Simoneit, B.R.T., Kobayashi, M., Mochida, M., Kawamura, K., Lee, M., Lim, H.J., Turpin, B.J. and Komazaki, Y. (2004). Composition and Major Sources of Organic Compounds of Aerosol Particulate Matter Sampled during the ACE-Asia Campaign. J. Geophys. Res. 109: D19S10.

Smith, N., Plane, J.M.C., Nien, C. and Solomon, P.A. (1995). Nighttime Radical Chemistry in the San Joanquin Valley. Atmos. Environ.29: 2887–2897.

Solomon, P.A. and Sioutas, C. (2008). Continuous and Semicontinuous Monitoring Techniques for Particulate

Park et al., Aerosol and Air Quality Research, 13: 957–976, 2013 976

Matter Mass and Chemical Components: A Synthesis of Findings from EPA’s Particulate Matter Supersites Program and Related Studies. J. Air Waste Manage. Assoc. 58: 164–195.

Sternbeck, J., Sjodin, A. and Andreasson, K. (2002). Metal Emissions from Road Traffic and the Influence of Resuspension – Results from Two Tunnel Studies. Atmos. Environ. 36: 4735–4744.

Sun, Y.L., Zhuang, G.S., Tang, A.H., Wang, Y. and An, Z.S. (2006). Chemical Characteristics of PM2.5 and PM10 in Haze–Fog Episodes in Beijing. Environ. Sci. Technol. 40: 3148–3155.

Thorpe, A. and Harrison, R. M. (2008). Sources and Properties of Non-Exhaust Particulate Matter from Road Traffic: A Review. Sci. Total Environ. 400: 270–283.

Turn, S.Q., Jenkins, B.M., Chow, J.C. and Pritchett, L.C. (1997). Elemental Characterization of Particulate Matter Emitted from Biomass Burning: Wind Tunnel Derived Source Profiles for Herbaceous and Wood Fuels. J. Geophys. Res. 102(D3): 3683–3699.

Turpin, B.J., Saxena, P. and Andrews, E. (2000). Measuring and Simulating Particulate Organics in the Atmosphere: Problems and Prospects. Atmos. Environ. 34: 2983–3013.

Turpin, B.J. and Lim, H.J. (2001). Species Contributions to PM2.5 Mass Concentrations: Revisiting Common Assumptions for Estimating Organic Mass. Aerosol Sci. Technol. 35: 602–610.

Watson, J.G. (2002). Visibility: Science and Regulation. J. Air Waste Manage. Assoc. 52: 628–713.

Watson, J.G. and Chow, J.C. (2002). A Wintertime PM2.5 Episode at the Fresno, CA, Supersite. Atmos. Environ. 36: 465–475.

Weber, R., Bergin, M., Kiang, C.S., Chameides, W., Orsini, D., St, J.J., Chang, M., Bergin, M., Carrico, C., Lee, Y.N., Dasgupta, P., Slanina, J., Turpin, B., Edgerton, E., Hering, S., Allen, G. and Solomon, P. (2003). Short-Term Temporal Variation in PM2.5 Mass and Chemical Composition during the Atlanta Supersite Experiment. J.

Air Waste Manage. Assoc. 53: 84–91. Weckwerth, G. (2001). Verification of Traffic Emitted

Aerosol Components in the Ambient Air of Cologne (Germany). Atmos. Environ. 35: 5525–5536.

Wittig, A.E., Takahama, S., Khlystov, A.Y., Pandis, S.N., Hering, S., Kirby, B. and Davidson, C. (2004). Semi-continuous PM2.5 Inorganic Composition Measurements during the Pittsburgh Air Quality Study. Atmos. Environ. 38: 3201–3213.

Wu, W.S. and Wang, T. (2007). On the Performance of a Semi-continuous PM2.5 Sulphate and Nitrate Instrument under High Loadings of Particulate and Sulphur Dioxide. Atmos. Environ.41: 5442–5451.

Yadav, A.K., Kumar, K., Kasim, A., Singh, M.P., Parida, S.K. and Sharan, M. (2003). Visibility and Incidence of Respiratory Diseases during the 1998 Haze Episode in Brunei Darussalam. Pure Appl. Geophys. 160: 265–277.

Zhang, X.Q. and McMurry, P.H. (1992). Evaporative Loss of Fine Particulate Nitrates during Sampling. Atmos. Environ. 26A: 3305–3312.

Zhang, Y., Shao, M., Lin, Y., Luan, S., Mao, N., Chen, W. and Wang, M. (2012a). Emission Inventory of Carbonaceous Pollutants from Biomass Burning in the Pearl River Delta Region, China. Atmos. Environ., In Press, doi: 10.1016/j.atmosenv.2012.05.055.

Zhang, R., Tao, J., Ho, K.F., Shen, Z., Wang, G., Cao, J., Liu, S., Zhang, L. and Lee, S.C. (2012b). Characterization of Atmospheric Organic and Elemental Carbon of PM2.5 in a Typical Semi-Arid Area of Northeastern China. Aerosol Air Qual. Res. 12: 792–802.

Zhou, Y., Wang, T., Gao, X., Xue, L., Wang, X., Wang, Z., Gao, J., Zhang, Q. and Wang, W. (2009). Continuous Observations of Water-Soluble Ions in PM2.5 at Mount Tai (1534 m a.s.l.) in Central-Eastern China. J. Atmos. Chem. 64: 107–127.

Received for review, July 12, 2012 Accepted, December 16, 2012