Apportioning and Locating Nonmethane Hydrocarbon Sources to a

Jul 7, 2010 - Nonmethane hydrocarbons (NMHCs) measured between April 2004 and March 2005 at a background monitoring site on Sukmo Island, Korea ...
0 downloads 0 Views 4MB Size
Download PDF
Environ. Sci. Technol. 2010, 44, 5849–5854

Apportioning and Locating Nonmethane Hydrocarbon Sources to a Background Site in Korea EUNHWA CHOI, JONG-BAE HEO, AND SEUNG-MUK YI* Department of Environmental Health, Graduate School of Public Health, Seoul National University, Gwanak-gu, Seoul, 151-742, South Korea

Received December 2, 2009. Revised manuscript received June 22, 2010. Accepted June 23, 2010.

(PSCF) has been used to indicate the possible source areas of specific NMHC species (11). Long-range NMHC transport has been studied using backward trajectories at remote sites in Japan (12, 13) and China (14). However, backward trajectory calculations include uncertainty as to whether the air mass passed through the source region. Therefore, PSCF, which utilizes both the air back trajectory and the concentration of the contaminant is a more effective tool for locating possible source regions. In this study, we identified and apportioned the sources of NMHCs measured at a background monitoring site in Korea using PMF. We also examined the locations and boundaries of regional NMHC sources using the PSCF model coupled with the PMF results.

2. Experimental Section Nonmethane hydrocarbons (NMHCs) measured between April 2004 and March 2005 at a background monitoring site on Sukmo Island, Korea were analyzed to identify and apportion NMHC sources. A total of 7694 samples and 35 NMHC species were analyzed. Positive matrix factorization (PMF), applied to identify and apportion the sources of NMHCs, resolved six sources: two fuel evaporative sources (36.3%), solvent sources (25.4%), mixed sources of vehicle exhaust and combustion (22.8%), petrochemical sources (9.6%), and biogenic sources (5.4%). During the summer, the largest contributors to ozone formation were biogenic sources (48.9% and 79.7% by maximum incremental reactivity and propene-equivalent concentration, respectively), which were situated locally, and secondary sources included solvent sources (22.2% and 7.4%) and fuel evaporative sources (15.6% and 8.2%). For evaporative-1 sources composed of long-lived alkanes, the potential source contribution function (PSCF) technique using 48 h back trajectories revealed oil and gas fields in China as potential source areas of fresh “regional” air masses. In addition, the PSCF results for evaporative-2 sources and a long-lived marker species of vehicle exhaust/combustion sources showed that the NMHC mixing ratio in Sukmo, South Korea was enhanced by longrange transport from the Shandong area in China.

1. Introduction Anthropogenic nonmethane volatile organic compound (NMVOC) emissions in Asia are greater than those in Europe or North America, and Asian total NMVOC emissions are predicted to increase by 99% over 2000 levels by 2020 (1). To establish effective measures for reducing ambient nonmethane hydrocarbon (NMHC) concentrations, it is very important to identify the main sources of NMHCs and quantify the contribution of each source. Several receptor models, such as principal components analysis (PCA), chemical mass balance (CMB), UNMIX, and positive matrix factorization (PMF) have been used to identify and apportion the NMHC sources (2-6). PMF is a very effective tool that produces non-negative factors, utilizes data matrix error estimates, and apportions NMHC sources using only measured data without source-specific chemical profiles (7, 8). Conditional probability function (CPF) plots applied to PMF results have been used to identify the direction of local NMHC sources (2, 9, 10). The potential source contribution function * Corresponding author tel: +82 2 880 2809; fax: +82 2 743 8240; e-mail: [email protected]. 10.1021/es903634e

 2010 American Chemical Society

Published on Web 07/07/2010

2.1. Sampling Sites. Nonmethane hydrocarbon samples were collected at Sukmo Island (Sukmo) situated at the northwestern tip of the Republic of Korea (Figure 1). This location has small local anthropogenic sources and low population density. It is approximately 64 km from Seoul, 35-50 km from the Incheon industrial complex, 40 km from the port of Incheon, and 35 km from Incheon International Airport. In addition, the North Korean Gaesong industrial complex and the armistice line are approximately 40 km northeast and 8 km northwest of the sampling site, respectively. The shortest distance to China is approximately 300 km northwest and west from the site. Therefore, Sukmo is a good site for examining the impact of transported NMHCs. 2.2. Samples of NMHCs. Sampling and analysis of NMHCs was conducted in accordance with the Photochemical Assessment Monitoring Stations (PAMS) Technical Assistance Document (TAD) produced by the U.S. Environmental Protection Agency (15). In total, 55 species of NMHCs were measured every hour using a combined gas chromatograph with a flame ionization detector. Also following the EPA TAD, quality assurance and quality control procedures such as calibration, assessment of sampling background and carry-over, detection limit determination, precision, and accuracy were performed (16-18). Hourly meteorological data, including wind speed and wind direction, were obtained from the Ganghwa weather station (126.3° E longitude and 37.4° N latitude), which is approximately 15.4 km east of the sampling site. 2.3. PMF. The following PMF model for source identification of NMHCs was used: p

Xij )

∑g

ik

· fkj + eij

(1)

k)1

A data set can be expressed as a matrix X of i by j dimensions, where i is the number of samples and j is the species measured; gik is NMHC concentration from the kth source contributing to the ith sample, fkj is the mass fraction of the jth species from the kth source, eij is the residual of the jth species concentration measured in the ith sample, and p is the total number of independent sources. The objective of PMF analysis is to determine the number of sources, p, the chemical composition profile, fk, and the source contributions, gk. Concentrations and associated uncertainties of the PMF model were prepared according to the procedure suggested by Polissar et al. (19). Of the 55 species measured at Sukmo, those with at least 1% of the measurements above the method detection level and n-undecane, which is a marker of diesel VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5849

FIGURE 1. Study site location. vehicle emissions, were included in the PMF analysis. As a result, 7694 samples and 35 species were analyzed (Table S1 of the Supporting Information). 2.4. CPF. To estimate likely local source impacts, the CPF analysis was performed using the source contributions estimated from PMF coupled with the surface wind directions and speeds. The CPF is the probability that the source contribution from a given wind direction will exceed a prespecified value and defined as CPF )

m∆θ n∆θ

(2)

where m∆θ is the number of occurrences from wind sector ∆θ that exceed the thresholds, and n∆θ is the total number of occurrences from ∆θ. In this study, 16 sectors (∆θ ) 22.5°) and a threshold criterion of the upper 25th percentile were used to show the directionality of the sources. Data for periods when the wind speed was below 1 m/s were excluded, thereby eliminating 30.9% of the total observations. 2.5. PSCF. Potential source contribution function analysis has been applied to identify possible source areas and their preferred pathways that give rise to high concentrations of air pollutants (20, 21). Backward trajectories were calculated using the hybrid single particle Lagrangian integrated trajectory (HYSPLIT 4) model and gridded meteorological data fromtheU.S.NationalOceanicandAtmosphericAdministration. PSCF is the conditional probability of a high concentration upon arrival at the monitoring site that passes through each cell and is defined as PSCFij )

P[Bij] mij ) P[Aij] nij

(3)

where P [Aij] represents the probability of a randomly selected air parcel on the ijth grid, and P[Bij] reflects the probability in the same cell for contaminated air parcels. mij end points among nij correspond to the trajectories that pass over the 5850

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010

receptor site with pollutant concentrations exceeding the threshold criterion. Since backward trajectories starting at different heights traverse different distances and pathways (21, 22), multiple height PSCF analysis was performed with starting elevations of 100, 500, and 1000 m above ground level for 48 h backward time. To minimize the effect of small nij values resulting in high PSCF values with high uncertainties, the PSCF values were down-weighted when the total number of end points per cell was less than three times the average value of the end points (16, 23).

Wij )

{

3nave〈nij 1.0 0.7 1.5nave〈nij e 3nave 0.4 nave〈nij e 1.5nave nij e nave 0.2

}

(4)

The PSCF model relies on quantitative information of NMHC sources; thus only contributions (G-factors) of each source identified by the PMF model were used as input data. An upper 25% concentration criterion was used. There were 369,312 total end points and the geographical region was divided into 16,950 grid cells of 0.3° latitude by 0.3° longitude.

3. Results and Discussion 3.1. General Characteristics of Measured NMHCs. As summarized in Table S1 of the Supporting Information, the most abundant NMHCs by category were alkanes (55.6%), followed by aromatics (27.5%), alkenes (11.7%), and acetylene (5.3%). The dominating NMHCs were relatively long-lived species in the atmosphere: ethane (arithmetic mean, 4.67 ppbc), toluene (4.49 ppbc), propane (3.00 ppbc), ethylene (1.92 ppbc), and acetylene (1.71 ppbc). The “photochemical age” of an air mass can be assessed by examining the ratios of two species with similar emission sources, but with different chemical lifetimes in the atmosphere (24). The major decay of hydrocarbons in the atmosphere is by mixing and by reaction with the OH radical

TABLE 1. OH Rate Constant, k, and the Corresponding Lifetimes in the Troposphere for Selected Hydrocarbons species

1012 × ka cm3 molecule-1 s-1

lifetimec

ethane propane n-butane iso-pentane benzene toluene ethylbenzene m-xylene p-xylene ethylene propylene isoprene acetylene

0.268 1.15 2.54 3.95b 1.23 5.96 7.1 23.6 14.3 8.52 26.3 101 0.90

58 days 13 days 6.1 days 3.9 days 13 days 2.6 days 2.2 days 7.8 h 1.1 days 1.8 days 7.0 h 1.8 h 17 days

a Values (at 298 K) from Atkinson (33) unless otherwise noted. b Values from Harley and Case (34). c 12 h daytime OH concentration of 1.5 × 106 molecules cm-3.

(the OH rate constants and corresponding lifetimes of selected NMHCs are shown in Table 1). A higher ratio of the two species (the more reactive hydrocarbon to the less reactive hydrocarbon) indicates relatively little photochemical processing of the air mass and a major impact from local emissions. By contrast, a lower ratio reflects older NMHC air masses; i.e., emissions from more distant sources (3, 24). The ratios of propane-to-ethane, butane-to-propane, propylene-to-ethylene, m,p-xylenes-to-ethylbenzene (X/E), and toluene-to-benzene (T/B) have been widely used to assess the age of air masses and to understand source signatures (3, 24-27). Ethane and propane are emitted mainly from natural gas (9, 26). Butane and propane are emitted from vehicle exhaust, fuel, liquid petroleum gas (LPG), and industrial processes (10, 24, 28). The main sources of propylene and ethylene are vehicle exhaust and petrochemical plants (9, 10). Toluene, ethylbenzene, and m- and p-xylenes are produced mainly from traffic emissions and paint and industrial solvents, whereas benzene is produced mainly from traffic sources; therefore, a high T/B ratio indicates local solvent sources (29). The NMHC concentrations and characteristic ratios (ppbc/ppbc) calculated at Sukmo were compared with those at 22 areas used in previous studies (Table S2 and Figure S1 of the Supporting Information). All sampling locations were classified into remote (Re), rural (Ru), suburban (SU), urban (U), urban background (UB), industrial (I), and road side (Ro) to understand the levels and the characteristics of the NMHCs by sampling location. At Sukmo, butane-to-propane, propylene-to-ethylene, and X/E ratios were very low, similar to a remote or rural site, suggesting the transport of an aging air mass and similar photochemical ages with remote and rural sites. While the T/B ratio at Sukmo (3.15) was relatively

higher than the other ratios, the correlation between toluene and benzene was very low (R ) 0.44), suggesting that the two species originated from different sources and that the sampling site is influenced more strongly by solvent sources than other areas. According to Ministry of Environment and National Institute of Environmental Research of Korea, solvent sources were the largest contributor to total NMVOC emissions in the Republic of Korea (South Korea) accounting for 58.2% of the total in 2004 (30). 3.2. PMF Results. Evaporative-2 and biogenic sources were not separated in the five-source model, and the evaporative-2 source was split in the seven-source model. Therefore, a six-source model with FPEAK ) 0 provided the most physically meaningful solution (Figure S2 of the Supporting Information). The largest NMHC sources at Sukmo were two fuel evaporative sources, which accounted for 36.3% to total NMHC concentrations during all sampling periods (Table 2). The evaporative-1 source was identified by the abundance of C2-C5 alkanes, such as ethane, propane, iso-butane, n-butane, and n-pentanes. The abundance of ethane and propane was due mainly to natural gas (9, 26), while the enrichment of propane, butanes, and pentanes was probably the result of emissions from evaporative sources such as LPG and gasoline vapor (10, 28). Therefore, the source was probably natural gas, LPG, gasoline vapor, or fugitive emissions from industrial plants, such as natural gas drilling, crude oil storage tanks, or petroleum facilities (9, 10, 31). The evaporative-2 source was identified as C4-C6 alkanes, such as iso-pentane, n-pentane, 2-methylpentane, isobutane, n-butane, and n-hexane which were most likely from gasoline vapor and liquid gasoline (10, 26, 32). The solvent source, which had high loadings of toluene, ethylbenzene, and m-, p-, and o-xylenes was the second highest contributor to the total NMHC concentration (25.6%). The next source contains an abundance of ethane, ethylene, acetylene, benzene, toluene, n-decane, and nundecane (9, 26, 32). These compounds are typical markers of vehicle or diesel vehicle exhaust. However, ethylene, acetylene, and heavy alkanes such as n-decane and nundecane are also produced by combustion process (10). No noticeable increase in mixing ratios of these compounds was seen during typical rush hour timeframes (Figure S3 of the Supporting Information). Thus, we labeled this source as a mix of vehicle exhaust and combustion source. The petrochemical source was characterized by ethylene and propylene (9, 10). A biogenic source was identified by the high proportion of isoprene. As shown in Table 2, the annual average percentage of the biogenic source was the lowest among the six sources (5.4%); however, its summermean contribution was the second highest (23.4%). 3.3. Contribution of NMHC Sources to Ozone Formation. To estimate the contribution of NMHC sources to ozone formation at the sampling sites, the ozone formation potential (OFP) and propene-equivalent concentration (prop-equiv) (5, 24, 35) were estimated for the NMHC sources resolved by

TABLE 2. Photochemical Characteristics of Sources Extracted by Positive Matrix Factorization (Absolute, O3-Formation Potential, and Propene-Equivalent Concentration) total sampling periods

summer

source

absolute (%)

absolute (%)

O3-formation potential (%)

propene-equivalent concentration (%)

evaporative-1 solvent vehicle exhaust/combustion evaporative-2 petrochemical biogenic

25.8 25.6 22.9 10.6 9.7 5.4

10.8 30.5 13.5 18.1 3.6 23.4

3.8 22.2 8.2 11.8 5.1 48.9

1.9 7.4 3.6 6.3 1.2 79.7

VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5851

FIGURE 2. Diurnal variation of the contributions to ozone formation during the summer: (a) evaluated by ozone formation potential, (b) evaluated by propene-equivalent concentration.

TABLE 3. Comparison of Mean Values by Groups Segregated Using the Concentration of Evaporative-1 Sources and the Ratio of Propane to Ethane (Values in Parentheses Are Standard Deviations)

total number of event propane-to-ethane (ppbc/ppbc) total NMHCs (ppbc) evaporative-1 (ppbc)

whole

common

fresh local air

fresh regional air

7694 0.70 (0.93) 39.28 (30.71) 10.09 (8.06)

4191 0.67 (1.21) 31.63 (20.63) 4.34 (3.09)

881 1.20 (0.52) 82.96 (60.95) 20.64 (9.85)

2622 0.57 (0.07) 36.82 (9.92) 15.75 (4.51)

PMF. The OFP indicates the degree to which NMHCs may contribute to ozone formation and can be calculated by multiplying the mean concentrations of hydrocarbons (µg/ m3) of each source category and the maximum incremental reactivity (MIR) coefficient (ozone produced per hydrocarbon, dimensionless). Prop-equiv (i) (ppbc) is the photochemical ozone creation potential defined relative to the ozone increment formed by propene and defined as prop-equiv(i) ) concn.(i) × kOH(i)/kOH(propene)

(5)

where concn.(i) is the concentration of hydrocarbon i and kOH(i) and kOH(propene) are rate constants for the reaction of hydrocarbon i and propene with OH. The OFP and the prop-equiv were determined during the summer using the average concentration of 35 NMHC species from each source from June to August 2004. Table 2 presents the absolute contributions during all sampling periods and the summer for sources extracted by PMF, and the summermean source contributions to ozone formation. The biogenic source was the greatest contributor to ozone formation from June to August 2004 (48.9% and 79.7% by MIR and propequiv, respectively), followed by solvent (22.2% and 7.4%), evaporative-2 (11.8% and 6.3%), vehicle exhaust/combustion (8.2% and 3.6%), petrochemical (5.1% and 1.2%), and evaporative-1 (3.8% and 1.9%) sources. The contributions from the biogenic source increased from 4 a.m. to a maximum at 2 p.m. at 79.5% and 93.9% for the OFP and prop-equiv concentration, respectively (Figure 2). The six emission sources deduced by PMF were based on data measured at receptor sites without considering reaction. Therefore, their contribution was likely underestimated due to photochemical reactivity (4) and the mass contribution of the biogenic source to ozone production may be much higher due to highly reactive isoprene (36). 3.4. CPF and PSCF Results. CPF was used to identify the direction of local sources, and PSCF, which considers material transfer from multiple vertical heights, was used to locate potential source regions. The PSCF model does not require 5852

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010

an emissions inventory for NMHCs. Its output is a map of likely source areas whose emissions can be transported to the sampling site (37) and thus PSCF plots are very useful in identifying probable source areas without the need for an NMHC emissions inventory. Each NMHC source resolved by PMF consisted of various NMHC species with different lifetimes. Biogenic sources identified by isoprene, which is highly reactive, were excluded from the PSCF analysis, and the two evaporative sources composed of light alkanes, which are more stable in the atmosphere (Table 1), were analyzed using starting elevations of 100, 500, and 1000 m and 48 h backward time. As discussed in Section 3.1, the propane to ethane ratio can be used to investigate the impact of local sources, the relative ages of air parcels, and transport histories (10). Both ethane and propane are abundant in evaporative-1 sources and have long but different lifetimes (Table 1). To determine the probable locations of evaporative-1 sources using PSCF, we classified data into three groups: (1) Sukmo background air (termed “common” in Table 3), which has a concentration below the mean of evaporative-1 sources, (2) fresh “local” air (higher than the mean concentration of evaporative-1 sources and a higher ratio of propane to ethane than the mean), and (3) fresh “regional” air (higher than the mean concentration of evaporative-1 sources and a lower ratio of propane to ethane than the mean). For Groups 2 and 3, the propane to ethane ratios were 1.20 and 0.57 and the mean concentrations of total NMHCs and evaporative-1 sources were 82.96 and 20.64 ppbc and 36.82 and 15.75 ppbc, respectively. This result suggests that impacts by relatively local sources of NMHC were largely responsible for elevated concentrations at the sampling site. However, based on the total number of events, the sampling site was more frequently impacted by regional air masses than local sources (881 events had higher propane to ethane ratios and 2622 had lower ratios). The 48 h PSCF results for the evaporative-1 sources with lower propane to ethane ratio (0.57) indicated a region near Shandong, Liaoning, and Inner Mongolia, where oil and gas fields are located (Figure 3a). The Shung-ri oil and gas field,

FIGURE 3. Likely source areas identified using conditional probability function (top left box) and potential source contribution function plots, 48 h backward time: (a) evaporative-1 sources with low propane to ethane ratio, (b) evaporative-1 sources with high propane to ethane ratio, (c) evaporative-2 sources, and (d) acetylene sources. which is the second largest in China, is situated northeast of Shandong and extends over eight cities. The oil and gas fields near the reaches of the Liao River constitute the third largest oil field in China and stretch over 13 cities in Liaoning and Inner Mongolia. Emissions of ethane and propane from Shandong and Liaoning, China were higher than the total emissions of those compounds from South Korea in 2000 (38). The CPF results for the evaporative-1 sources points to the southeast and is consistent with the PSCF result of higher propane to ethane ratio (1.20) which represents the impact of more fresh “local” sources (Figure 3b). Whereas, PSCF plots applied to unsegregated data indicated only Shandong, Liaoning, and Inner Mongolia as potential source regions similar to lower ratio modeling results (Figure S4 of the Supporting Information and Figure 3a) and thus, segregated application of PSCF appears to be more useful for inferring both regional and relatively local source areas than unsegregated application. The CPF and PSCF results for the evaporative-2 sources (composed mainly of pentanes and butanes) are in good agreement and indicate the source to the southeast in the southwestern part of South Korea and Shandong province in China (Figure 3c). In 2000, South Korea reported that the emissions of pentanes and butanes were 63.30 and 138.97 Gg, respectively, and in Shandong, which ranked as the second highest emitting province in China, they were 54.93 and 71.45 Gg, respectively (38). Solvent and petrochemical sources whose typical indicators include very reactive hydrocarbons such as xylenes and

propylene, respectively (Table 1), were analyzed using starting elevations of 50, 100, and 500 m and 6 h backward time PSCF (Figure S5 of the Supporting Information). Both CPF and PSCF results suggest the solvent source to the southeast where the industrial complex consisting of many factories using solvents and paints is located (Figure 1). The CPF and PSCF results for the petrochemical sources were also in good agreement indicating the southeast, north, and northwest. Vehicle exhaust/combustion sources, which were composed of less reactive NMHCs than solvent and petrochemical sources, were analyzed using 6, 12, and 24 h backward time PSCF. For the longer-lived and marker species of this source, 48 h backward time PSCF was used to complement the shorter duration PSCF results (Figure S6 of the Supporting Information and Figure 3d). The 48 h PSCF plot for acetylene, which has a long lifetime of 17 days (Table 1) and is the dominant compound in the vehicle exhaust/combustion sources (Figure S2 of the Supporting Information) indicates Shandong in China, northeast of Pyongyang in North Korea, and the part of South Korea as potential source areas. In summary, the transport of the air mass from distant sources was predicted by low NMHC concentrations, a high proportion of long-lived species, and very low NMHC ratios of propane-to-ethane (0.64), butane-to-propane (0.53), propylene-to-ethylene (0.21), and m,p-xylenes-to-ethylbenzene (1.37). Long range transport was confirmed by 48 h PSCF maps for fresh “regional” air mass of evaporative-1 sources and those for evaporative-2 and acetylene sources indicating Shandong area in China as a common probable source region. VOL. 44, NO. 15, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5853

Therefore, we identified the possibility that the NMHC mixing ratio in South Korea could be enhanced by long-range transport from China.

Supporting Information Available Additional tables and figures. This information is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Ohara, T.; Akimoto, H.; Kurokawa, J.; Horii, N.; Yamaji, K.; Yan, X.; Hayasaka, T. An Asian emission inventory of anthropogenic emission sources for the period 1980-2020. Atmos. Chem. Phys. 2007, 7, 4419–4444. (2) Zhao, W.; Hopke, P. K.; Karl, T. Source identification of volatile organic compounds in Houston, Texas. Environ. Sci. Technol. 2004, 38, 1338–1347. (3) Guo, H.; So, K. L.; Simpson, I. J.; Barletta, B.; Meinardi, S.; Blake, D. R. C1-C8 VOC in the atmosphere of Hong Kong. Atmos. Environ. 2007, 41, 1456–1457. (4) Na, K.; Kim, Y. P. Chemical mass balance receptor model applied to ambient C2-C9 VOC concentration in Seoul, Korea: Effect of chemical reaction losses. Atmos. Environ. 2007, 41 (32), 6715– 6728. (5) Duan, J.; Tan, J.; Yang, L.; Wu, S.; Hao, J. Concentration, sources and ozone formation potential of volatile organic compounds (VOCs) during ozone episode in Beijing. Atmos. Res. 2008, 88, 25–35. (6) Song, Y.; Dai, W.; Shao, M.; Liu, Y.; Lu, S.; Kuster, W.; Goldan, P.; Xie, S. Comparison of receptor models for source apportionment of volatile organic compounds in Beijing, China. Environ. Pollut. 2008, 156, 174–183. (7) Paatero, P.; Tapper, U. Positive matrix factorization: a nonnegative factor model with optimal utilization of error estimates of data values. Environmetrics 1994, 5, 111–126. (8) Paatero, P. Least squares formulation of robust, non-negative factor analysis. Chemom. Intell. Lab. Syst. 1997, 37, 23–35. (9) Kim, E.; Brown, S. G.; Hafner, H. R.; Hopke, P. K. Characterization of non-methane volatile organic compounds sources in Houston during 2001 using positive matrix factorization. Atmos. Environ. 2005, 39, 5934–5946. (10) Xie, Y.; Berkowitz, C. M. The use of positive matrix factorization with conditional probability functions in air quality studies: An application to hydrocarbon emissions in Houston, Texas. Atmos. Environ. 2006, 40, 3070–3091. (11) Xie, Y.; Berkowitz, C. M. The use of conditional probability functions and potential source contribution functions to identify source regions and advection pathways of hydrocarbon emissions in Houston, Texas. Atmos. Environ. 2007, doi:10.1016/ j.atmosenv.2007.03.049. (12) Sharma, U. K.; Kajii, Y.; Akimoto, H. Seasonal variation of C2-C6 NMHCs at Happo, a remote site in Japan. Atmos. Environ. 2000, 34, 4447–4458. (13) Kato, S.; Pochanart, P.; Kaji, Y. Measurements of ozone of nonmethane hydrocarbons at Chichi-jima island, a remote island in the western Pacific: long range transport of polluted air from the Pacific rim region. Atmos. Environ. 2001, 35, 6021– 6029. (14) Tang, J. H.; Chan, L. Y.; Chan, C. Y.; Li, Y. S.; Chang, C. C.; Liu, S. C.; Li, Y. D. Nonmethane hydrocarbons in the transported and local air masses at a clean remote site on Hainan Island, south China. J. Geophys. Res. 2007, 112, D14316. (15) U.S. Environmental Protection Agency. Technical assistance document for sampling and analysis of ozone precursors; EPA/ 600-R-98/161; 1998. (16) Choi, E.; Heo, J.-B.; Hopke, P. K.; Jin, B.-B.; Yi, S.-M. Identification, apportionment, and photochemical reactivity of non-methane hydrocarbon sources in Busan, Korea. Water, Air Soil Pollut. 2010, DOI: 10.1007/s11270-010-0459-0.

5854

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 15, 2010

(17) Hwang, S. M. A study on the measurement of VOCs using Online and Off-line GC system. In Proceedings of the 40th Meeting of Korean Society for Atmospheric Environment, 2005; pp 356358. (18) Hwang, S. M. QC/QA for Photochemical assessment monitoring station. In Proceedings of the 45th Meeting of Korean Society for Atmospheric Environment, 2007; pp 351-353. (19) Polissar, A. V.; Hopke, P. K.; Paatero, P.; Malm, W. C.; Sisler, J. F. Atmospheric aerosol over Alaska 1. Elemental composition and sources. J. Geophys. Res. 1998, 103 (D15), 19045–19057. (20) Ashbaugh, L. L.; Malm, W. C.; Sadeh, W. D. A residence time probability analysis of sulfur concentrations at Grand Canyon National Park. Atmos. Environ. 1985, 19, 1263–1979. (21) Hsu, Y.-K.; Holsen, T. M.; Hopke, P. K. Comparison of hybrid receptor models to locate PCB sources in Chicago. Atmos. Environ. 2003, 37, 545–562. (22) Cheng, M. D.; Hopke, P. K.; Barrie, L. A.; Rippe, A.; Olson, M.; Landsberger, S. Qualitative determination of source regions of aerosol in Canadian high Arctic. Environ. Sci. Technol. 1993, 27, 2063–2071. (23) Hoe, J.-B.; Hopke, P. K.; Yi, S.-M. Source apportionment of PM2.5 in Seoul, Korea. Atmos. Chem. Phys. 2009, 9, 4957–4971. (24) So, K. L.; Wang, T. C3-C12 non-methane hydrocarbons in subtropical Hong Kong: spatial-temporal variations, sourcereceptor relationships and photochemical reactivity. Sci. Total Environ. 2004, 328, 161–174. (25) Nelson, P. F.; Quigley, S. M. The m,p-xylenes:ethylbenzene ratio. A technique for estimating hydrocarbon age in ambient atmospheres. Atmos. Environ. 1983, 17, 659–662. (26) Brown, S. G.; Fankel, A.; Hafner, H. R. Source apportionment of VOCs in the Los Angeles area using positive matrix factorization. Atmos. Environ. 2007, 41, 227–237. (27) Tang, J. H.; Chan, L. Y.; Chan, C. Y.; Li, Y. S.; Chang, C. C.; Liu, S. C.; Wu, D.; Li, Y. D. Characteristics and diurnal variations of NMHCs at urban, suburban, and rural sites in the PRD and a remote site in South China. Atmos. Environ. 2007, 41, 8620– 8632. (28) Na, K.; Kim, Y. P.; Moon, I.; Moon, K. C. Chemical composition of major VOC emissions sources in the Seoul atmosphere. Chemosphere 2004, 55, 585–594. (29) Wang, T.; Guo, H.; Blake, D. R.; Kwok, Y. H.; Simpson, I. J.; Li, Y. S. Measurements of trace gases in the inflow of south China sea background air and outflow of regional pollution at Tai O, southern China. J. Atmos. Chem. 2005, 52, 295–317. (30) Korean MoE and NIER. Annual Report of Air Quality in Korea; 2006. (31) Sexton, K.; Westberg, H. Photochemical ozone formation from petroleum refinery emissions. Atmos. Environ. 1983, 17, 467– 475. (32) Watson, J. G.; Chow, J. C.; Fujita, E. M. Review of volatile organic compound source apportionment by chemical mass balance. Atmos. Environ. 2001, 35, 1567–1584. (33) Atkinson, R. Gas phase trophospheric chemistry of organic compounds: a review. Atmos. Environ. 2007, 41, S200–S240. (34) Harley, R. A.; Cass, G. R. Modeling the atmospheric concentrations of individual volatile organic compounds. Atmos. Environ. 1995, 29, 905–922. (35) Carter, W. P. L. Development of ozone reactivity scales for volatile organic compounds. J. Air Waste Manage. Assoc. 1994, 44, 881– 899. (36) Choi, Y.-J.; Ehrman, S. H. Investigation of source of volatile organic carbon in the Baltimore area using highly time-resolved measurements. Atmos. Environ. 2004, 38, 775–791. (37) Polissar, A.; Hopke, P. K. Atmospheric Aerosol over Vermont: Chemical Composition and Sources. Environ. Sci. Technol. 2001, 35, 4604–4621. (38) NMVOC emissions, 2000. Available at http://www.cgrer.uiowa. edu/EMISSION_DATA/anthro/table/nmvoc_2000_speciation_ final.htm.

ES903634E