[ Research Article ]

Asian Journal of Atmospheric Environment - Vol. 12, No. 4, pp.311-325

ISSN: 1976-6912 (Print) 2287-1160 (Online)

Print publication date 31 Dec 2018

Received 15 Jun 2018 Revised 26 Aug 2018 Accepted 18 Sep 2018

DOI: https://doi.org/10.5572/ajae.2018.12.4.311

Estimation of the Source Contributions for Carbonaceous Aerosols at a Background Site in Korea

Sanghee Han ; Ji Yi Lee ; Jongsik Lee¹⁾ ; Jongbae Heo²⁾

; Chang Hoon Jung³⁾ ; Eun-Sill Kim⁴⁾ ; Yong Pyo Kim⁵⁾^{, *}

Department of Environmental Science and Engineering, Ewha Womans University, 52 Ewhayeodae-gil, Seodaemun-gu, Republic of Korea
1)Department of Renewable Energy Convergence, Chosun University, Seoseok-dong, Dong-gu, Gwangju, Republic of Korea
2)Department of Environmental Health, Seoul National University, Daehak-dong, Gwanak-gu, Seoul, Republic of Korea
3)Department of Health Management, Kyungin Women’s College, Gyesan 2-dong, Gyeyang-gu, Incheon, Republic of Korea
4)Korea Global Atmosphere Watch Center, Korea Meteorological Administration, Anmyeon, Taean-gun, Chungcheongnam-do, Republic of Korea
5)Department of Chemical Engineering & Materials Science, Ewha Womans University, 52 Ewhayeodae-gil, Seodaemun-gu, Republic of Korea

Correspondence to: ^* Tel: +82-2-3277-2832, E-mail: yong@ewha.ac.kr

Copyright © 2018 by Asian Journal of Atmospheric Environment
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

To identify and quantify the contribution of the major sources for the ambient carbonaceous aerosols at a Global Atmosphere Watch (GAW) station in Korea, a receptor model, the Positive Matrix Factorization (PMF) model was applied for the one-year long measurement data. Particulate matter less than or equal to 2.5 μm in aerodynamic diameter (PM_2.5) aerosols were sampled at Anmyeon Island GAW station from June 2015 to May 2016 and carbonaceous species including ~80 organic compounds were analyzed. According to the performance parameter evaluation, 5 or 7 factors were considered as optimal number of factors. It was found out that the results of 7 factors gave less contribution from the factor designated as mixed sources which we could not separate clearly. The major sources with 7 factors were identified with various analyses including chemical characteristics and air parcel movement analysis. The 7 factors with their relative contributions are; anthropogenic Secondary Organic Aerosols (SOA) (14%), biogenic SOA (15%), primary biogenic source (8%), local biomass burning (13%), transported biomass burning (16%), combustion related source (15%), and mixed sources (19%). The air parcel movement analysis results also support the identification of these factors. Thus, the Anmyeon Island GAW station has been affected by both regional and local sources for the carbonaceous aerosols.

Keywords:

Positive Matrix Factorization, Source apportionment, Carbonaceous aerosol, Anmyeon Island, Global Atmosphere Watch station

1. INTRODUCTION

Organic Aerosols (OA) contain thousands of organic compounds and contribute to 20-50% to the total fine aerosol mass at mid-latitudes (Putaud et al., 2004; Saxena and Hildemann, 1996). OA are originated from both anthropogenic such as fossil fuel combustion, and biogenic sources. Also, OA are either emitted directly in particulate form (Primary Organic Aerosol, POA) or generated in the air (Secondary Organic Aerosol, SOA) (Kanakidou et al., 2005).

Some compounds of organic aerosols have been used as a marker for source identification. For instance, levoglucosan and Polycyclic Aromatic Hydrocarbons (PAHs) are known as markers of biomass burning and incomplete combustion of organic matters, respectively (Schauer et al., 1996; Simoneit et al., 1991).

Receptor oriented models infer source contributions by determining the best-fit linear combination of emission source chemical composition profiles needed to reconstruct the measured chemical composition of ambient samples (Watson, 1984). Receptor models have been used to identify major source of OA in the ambient air (Heo et al., 2009; Jaeckels et al., 2007; Watson et al., 2001).

Two receptor models, Chemical Mass Balance (CMB) and Positive Matrix Factorization (PMF) model, have been widely used as a tool in the area of air pollution source apportionment studies. Since PMF is a multivariate factor analysis tool that decomposes a matrix of speciated sample data into two matrices (factor contributions and factor profiles), source profiles are not necessary to run the model (Paatero and Tapper, 1994). Thus, PMF model has been commonly used at certain places in which (1) composition is changed by chemical reaction during the transport like backgrounds areas (Yuan et al., 2012) or (2) source profiles are not developed (Lee et al., 2008).

Anmyeon Island is located in the mid-western coast of South Korea where a Global Atmospheric Watch (GAW, 36°32'N; 126° 19'E, 45.7 m above sea level) regional station (AMY) has been established by the Korean Meteorological Administration (KMA). One of the goals of the AMY is to monitor the variations of the components that might affect climate change at a background area. There have been few reports on the levels of the atmospheric trace species, ionic species (Park et al., 2010) and n-alkanes and PAHs in the ambient aerosols (Lee et al., 2013, 2011).

The AMY is considered as a background site, but the concentrations of some organic compounds measured at the site were comparable to those in Seoul (Lee et al., 2011). In Lee et al. (2011), the reported concentration of water soluble organic carbon (WSOC) in 2010 at the AMY was 4.65±2.18 μgC m^-3, comparable to the level at Seoul (4.35±2.76 μgC m^-3). On the other hand, the average concentration of n-alkane at the AMY was 23.7±4.02 ng m^-3, which was about 1/3 of at Seoul (66.87±27.01 ng m^-3). It suggests that the ambient air at the AMY have characteristics of both background and polluted area. Thus, it is essential to characterize the AMY by identifying the levels of the atmospheric trace species and estimating the relative contribution of major contributors of the observed atmospheric trace species. By characterizing the AMY, we can understand the background level of carbonaceous aerosols and, thus, to estimate the effect of regional transport in Northeast Asia.

In Korea, several studies have been carried out on source apportionment of carbonaceous aerosols by using receptor models ( Jung et al., 2015; Kim et al., 2013; Heo et al., 2009; Lee and Kim, 2007). However, for background areas, only a few studies have been carried out for Gosan, a background area at Jeju Island (Moon et al., 2008; Han et al., 2006).

In this study, based on the one-year long measurement, the characteristics of carbonaceous species in PM_2.5 at the AMY is evaluated by applying the PMF model. At first, optimal number of factors for the PMF receptor modeling result on the carbonaceous species in PM_2.5 at the AMY was determined, then the characteristics of these factors were identified and quantified by various analyses including characteristic identification of organic compounds and air movement analyses. Finally, the relative importance of local emissions to the carbonaceous species in PM_2.5 is quantified.

2. METHODOLOGY

2. 1 Measurement Data

Details on the sampling and analysis are given elsewhere (Lee et al., 2018). In brief, samples were collected every 6th day 2016 on the rooftop of the AMY station building from Jun. 3, 2015 to May. 27. A high-volume air sampler with a PM_2.5 inlet was operated with 1000 L min^-1 of flow rate. PM_2.5 samples were collected on quartz fiber filters (203 mm×254 mm, Whatman Inc., Maidstone, UK). The number of samples was 58.

Organic Carbons (OC), Elemental Carbon (EC), WSOC, HUmic-LIke Substances of Carbon (HULIS-C) were classified. Seventy-eight organic compounds were identified and quantified; Polycyclic Aromatic Hydrocarbons (PAHs) (14), n-alkanes (17), Dicarboxylic Acids (DCAs) (19), n-alkenoic acids (18), and sugars (10). In addition, meteorological parameters were observed simultaneously at the same place by the KMA. Information of the data set is summarized in Table 1.

Table 1.

Summarized information of the input measurement data of carbonaceous species (Unit: ng m-3).

2. 2 Positive Matrix Factorization (PMF)

The US EPA PMF ver 5.0 (Norris et al., 2014) was used in this study. PMF is a one of the tools for multivariate factor analysis that decomposes a matrix of speciated sample data into two matrices, i.e., source profile and source contribution. According to Paatero (1997), fundamental equations about PMF can be presented with the definitions of matrices. Define the ‘residual matrix’ E, the difference between measurement X_ij and model Y_ij, as a function of the factors G and F as described in Eq. (1). Where X is a matrix of observed data and n×m dimensions where G is the unknown left hand factor matrix (scores) of dimensions n×p, and F is the unknown right hand factor matrix (loadings) of dimensions p×m (Paatero and Tapper, 1994).

E i j = X i j - Y i j = X i j - ∑ h = 1 p G i h F h j i = 1, …, m, j = 1, …, n

(1)

Define the ‘object function’ Q, to be minimized, as a function of the factors G and F, as in Eq. (2):

Q E = ∑ i = 1 m ∑ i = 1 n E i j / σ i j 2

(2)

The values σij are the standard deviation of the observed value X_ij. The task of the non-negatively constrained weighted factor analysis is: Minimize Q(E) with respect to G and F under the constraint that all or some of the elements of G and F are constrained to non-negative values (Paatero, 1997). More details on PMF were described in Paatero and Tapper (1994) and Norris et al. (2014).

In this study, the analyzed concentration values were used for the measured data, and the error fractions and 1/2 of the methods detection limit (MDL) values were used as the overall uncertainty assigned to each observation, as (Norris et al., 2014):

u n c e r t a i n t y = e r r o f r a c t i o n × c o n c e n t r a t i o n 2 + 12 × M D L 2

(3)

The error fraction for the ambient data was set as 15% (the variation of the measurement) of the concentration for individual carbonaceous species since the uncertainty for the MDL values of individual organic compounds were small (Choi et al., 2016; Lee and Kim, 2007). Values below the MDL were replaced by half of the MDL and their overall uncertainties were set at 5/6 of the MDL. Missing values were replaced by the median of the measured values and associated uncertainties were set at four times the median.

Optimal number of factors was determined by checking the maximum individual column mean (IM), the maximum individual column standard deviation (IS) values following the approach from Lee et al. (1999) and displacement (DISP), bootstrap (BS) parameter values from Paatero et al. (2014) and Brown et al. (2015). IM and IS were indicators to identify the species having the least fit and the most imprecise fit (Lee et al., 1999). They can also be used to identify the number of factors in the PMF. When the number of factors increases to a critical value, IM and IS will be stable after experiencing drastic drop.

DISP intervals include effects of random errors and partially include effects of rotational ambiguity. For DISP, focus is on the number of swaps at the lowest dQ^max level and the percent change in Q (%dQ) (Brown et al., 2015). We focused on the %dQ of the DISP and acceptable range of %dQ was set as 0.1, following Paatero et al. (2014). The US EPA PMF performs BS by randomly selecting non-overlapping blocks of consecutive samples (block size is suggested by the software or by the user) and creating a new input data file of the selected samples, with the same dimensions (i.e., number of samples and number of species) as the original data set. The PMF is then run on the new resampled data set, and each BS factor is mapped to a base run factor by comparing factors’ contributions (G matrix) for those samples included in the resampled data set (Brown et al., 2015).

Since the number of samples (58) was smaller than the number of compounds used in the PJMF modeling (82 as shown in Table 1), there might be limitations of the reliability of the modeling result. Still, the result in this study is a first step to identify major contributors of the observed carbonaceous aerosols at the AMY.

2. 3 Air Parcel Movement Analysis

To identify sources more accurately, the results of the Conditional Probability Function (CPF) and backward trajectories are used. In the CPF, the wind speed and wind direction at the sampling site are used to calculate the possible contribution of local sources. The CPF alone could not identify the regional transport. Thu, backward trajectory analysis is applied to estimate the air parcels’ movement of regional scale, outside of Korea. Thus, the combination of the results of both CPF and backward trajectory analysis along with the PMF modeling result could be used to determine whether the source is impacted by regional transport or local emission.

In the CPF, the probability that a source is located within a particular wind direction sector, ΔΘ is calculated:

C P F = m ∆ Θ n ∆ Θ

(4)

where n_ΔΘ is the number of times that the wind passed through direction sector ΔΘ, and m_ΔΘ is the number of times that the source contribution peaked while the wind passed through sector ΔΘ (Pekney et al., 2006; Ashbaugh et al., 1985). m_ΔΘ is calculated as the number of times that contribution shows higher than mean value with the wind passed through direction sector ΔΘ. The hourly wind direction and wind speed data were from the KMA at the same site.

Hybrid Single-Particle Lagrangian Intergrated Trajectory (HYSPLIT4) model (http://www.ready.noaa.gov/HYSPLIT.php) developed in National Oceanic and Atmospheric Administration (NOAA) is used to estimate backward trajectories. Three day backward trajectories are estimated for every 3 hours at the starting height of 1500 m to clarify the impact of the regional sources. For each factor, backward trajectories are estimated for 6 samples, about 10% of the total samples to identify major trajectories for each factor.

3. RESULTS AND DISCUSSIONS

3. 1 Determination of Factors

Table 2 shows the IM and IS values drastically dropped when the number of factor became 5 and 7. The DISP value that indicates random errors was zero and BS value that shows the sensitivity of the result was 85.40% and 84.29% concord when the number of factor became 5 and 7, respectively. For both cases, the performance parameters show comparable results. Furthermore, the 7 factor case shows how the two factors for the 5 factor case are divided into 2 related factors, respectively. Thus, we decided to show both cases.

Table 2.

The value of parameters (IM, IS, DISP, BS) to decide the optimal number of factors.

The characteristics of these factors could be identified with the criteria related to the known characteristic of emissions and reactions of carbonaceous species. For example, PAHs are mainly formed during incomplete combustion processes of organic matters and, thus, the factor with high contribution of PAHs were considered as primary emissions from combustion. EC is usually emitted from primary sources, while OC is formed by both primary and secondary aerosol formations. WSOC is mainly produced from secondary formation and biogenic sources.

Carbon Preference Index (CPI_odd) is defined as the ratio of odd to even carbon numbered n-alkanes and has been used to identify the origin of n-alkanes. Ambient n-alkanes can be originated from anthropogenic and biogenic sources (Simoneit, 1991). A CPI_odd value close to 1 indicates anthropogenic sources contributing more than biogenic sources. When CPI_odd value is larger than 1 and the maximum concentration of n-alkanes is observed in C₂₇, C₂₉, C₃₁, it suggests a predominant biogenic input form plants (Kotianova et al., 2008). In addition, rural environments with more biogenic influence generally have CPI_odd value above 2.0 (Gogou et al., 1996; Simoneit, 1989). Thus, a factor with high CPI_odd with dominant concentration in C₂₇, C₂₉, C₃₁ is determined as biogenic sources, since these three compounds were marker for the biogenic source.

DCAs are important indicator of SOAs because it can be formed by chemical reactions in the atmosphere (Kawamura et al., 1996; Rogge et al., 1993). Usually DCAs are further divided into two groups; regional ones and locally generated ones since compounds of DCAs can be grouped by seasonal variation (Aggarwal and Kawamura, 2009). Thus, seasonal contribution should be considered together to clarify the source of factors with high DCAs.

Monocarboxylic acids are emitted from both biogenic sources and fossil fuel combustions. cis-pinonic acid is the major product of the photochemical oxidations of monoterpenes derived from biogenic origins (Zhang et al., 2010) and, thus, considered as secondary biogenic marker. The maximum carbon number in monocarboxylic acids from fossil fuel combustion was C₁₆ (Simoneit, 1986, 1985) and the ratio of unsaturated mono-carboxylic acid (C_18:1) to saturated analog (C₁₈) in the ambient air represents the residence time since C_18:1 is less stable than C₁₈. When the concentration ratio of C₁₈/C_18:1 shows high value, ambient organic compounds should experience more degradation (Cheng et al., 2004).

Among sugars, levoglucosan is commonly used as a marker of biomass burning (Fraser and Lakshmanan, 2000; Simoneit et al., 1999). Other sugar compounds are mainly derived from primary biogenic sources, such as airborne pollen and fungal spores (Fu et al., 2012).

3. 2 PMF Results with 5 Factors

The profiles and relative contributions of the 5 factors are presented in Fig. 1 and summarized in Table 3(a). To identify a source, both factor profile and temporal variation of the relative contribution of the factor were considered.

Fig. 1.

Source profiles and relative contributions for the five factors of the PMF modeling result at the Anmyeon Island GAW station (AMY).

Table 3.

summarized characterisitcs of factor profiles.

The first factor shows high level of DCAs. The concentration of WSOC is the highest for the first factor which is considered as an indicator of aerosols aging. Also the C₁₈/C_18:1 ratio, another indicator of the aged aerosol, shows the highest value in the first factor as 16.2. In addition, the relative contributions of this factor are generally higher in summer and spring. According to the characteristics of profile and contribution, the first factor could be identified as SOA with 11% annual contribution.

For the second factor, the CPI_odd value is the highest as 2.24 with the high concentrations in the range of C₂₇, C₂₉, and C₃₁ among n-alkanes. Comparing with other factors, the C₁₈/C_18:1 ratio shows the lowest value among the factors as 6.3. Furthermore, relative contribution of second factor was the lowest in winter and highest in summer. Thus, it is determined as the biogenic source due to the characteristic of the profile and seasonal trend of the relative contribution and the annual contribution is 23%.

The third and fourth factors show high contribution of PAHs and levoglucosan in the profiles. Especially, the third factor shows high concentrations of short chain n-alkanes and both factors show high contributions of levoglucosan in the factor profiles, especially, the fourth factor shows the highest ratio of levoglucosan. Moreover, the third factor shows higher contribution in winter than other seasons while the fourth factor shows higher contribution in fall and spring than winter. During winter and fall, prominent wind direction to Korea is from northwest. Local biomass burning has shown high contribution in fall due to open burnings after harvest and in spring due to cleaning of fields (Kim et al., 2013; Lee and Kim, 2007). Thus, the characteristics of seasonal variation suggest that the third factor is a regional one, transported biomass burning from outside of Korea and fourth factor is local biomass burning. The annual average values of the relative contribution are both 17% for third and fourth factor.

The fifth factor shows high contributions of levoglucosan and C₁₈/C_18:1 ratio in the profile. In addition, the concentration of C₂₇, C₂₉, and C₃₁ among n-alkanes and CPI_odd is also high. The seasonal variation of the relative contribution shows high level in spring and fall. According to the criteria, the fifth factor shows mixed characteristics. Thus, it could be considered as a mixed one from various sources including vehicular sources and other combustion sources including biomass burning. Moreover, it might be also affected by regional source due to the profile that C_18:1 is almost 0 in the fifth factor. Previous studies using measurement data in background area have suggested mixed sources as an important contributor (Minguillion et al., 2012; Choi et al., 2010). Thus, the fifth factor is designated as mixed sources. It is possible to say that this fifth factor might be the transported carbonaceous aerosols, the source characteristics of which might be lost during transport and mixed together. However, since it takes the largest contribution of 32%, further study is warranted for this fifth factor.

3. 3 PMF Result with 7 Factors

In the results of the PMF model with 7 factors, biogenic sources and mixed sources of the result with 5 factors are divided into two, respectively. Factor profiles and relative contributions are presented in Fig. 2 and summarized in Table 3(b).

Fig. 2.

Source profiles and relative contributions for the seven factors of the PMF modeling result at the Anmyeon Island GAW station (AMY).

First, biogenic source (23%) is divided into biogenic SOA (15%) and primary biogenic source (8%). The biogenic SOA factor shows high relative contribution of the n-alkenoic acids and cis-pinonic acid. Since cis-pinonic acid is the marker of the biogenic SOA (Choi et al., 2016; Rogge et al., 1993) and shows the highest concentration in this factor, this factor is determined as the biogenic SOA. This factor shows high contribution in summer (from June to September in 2015), as the typical trend for the SOA source. The other factor is designated as primary biogenic source. The profile of this factor presents the highest concentration of sugar which are mainly emitted from primary biogenic source (Rogge et al., 1993), such as fluctose and glucose, except levoglucosan. Moreover, value of the CPI_odd is the highest as 10.49 among the factors. Seasonal variation of the relative contribution of this factor shows high contribution in spring. Thus, this factor might show the naturally emitted primary biogenic sources.

Second, the mixed source (32%) is also divided into combustion source (15%) and mixed sources (19%). The first one is the combustion related source. In the profile, high concentration of the n-alkanes and EC suggest that this factor is a primary source. In addition, PAHs and levoglucosan are presented in this factor. According to these characteristics of the profile, it was determined as the combustion related factor. The last factor is still left as mixed source since it presents characteristics of several sources. Comparison of the identified sources and relative contributions of 5 and 7 factors is summarized in Table 4.

Table 4.

Identified sources and relative contributions of 5 and 7 factors.

3. 4 Conditional Probability Function (CPF) and Backward Trajectory Analysis for the Factors

To further validate the PMF modeling result, analyses of air parcel movement were carried out by applying the CPFs and backward trajectory estimation for each factor and the results are presented in Fig. 3, Fig. 4, and Fig. 5 for 5 and 7 factors, respectively.

Fig. 3.

The result of the Conditional Potential Function (CPF) and backward trajectories for three factors (Anthropogenic SOA, local biomass burning, transported biomass burning).

Fig. 4.

The result of the Conditional Potential Function (CPF) and backward trajectories for biogenic sources.

Fig. 5.

The result of the Conditional Potential Function (CPF) and backward trajectories for mixed sources.

In Fig. 3, analysis result for the three factors (anthropogenic SOA, local biomass burning, and transported biomass burning) that are the same in both 5 and 7 factors are presented. Backward trajectories for these three factors are the same for 5 factor and 7 factor cases. In case of the anthropogenic SOA factor, the major local wind directions were from northeast and southwest for both 5 and 7 factor cases and the CPF was high with the wind from southwest. The major backward trajectories were from northwest and southwest. Thus, the results of both backward trajectory and CPF of anthropogenic SOA agree and show impact of western side of AMY, probably China.

The backward trajectories for the local biomass burning factor show that the air parcels were moved from various directions while those for the transported biomass burning were mainly from north. The CPF results for the local biomass burning factor show that local winds were mainly from north while those for the transported biomass burning were mainly from northwest. The dominant backward trajectories of transported biomass burning were from north and northwest. Impact of biomass burning from China and North Korea to Seoul has been verified previously (Kim et al., 2013; Lee and Kim, 2007). Thus, these air parcels movement also support the classification of the local and transported biomass burning factors.

The biogenic source and mixed sources for the 5 factor case are divided into two factors each, respectively for the 7 factor case. The divided factors show clearly distinctive air parcels movements as shown in Fig. 4 and Fig. 5. The biogenic source factor for the 5 factor case, the backward trajectories were from various directions. However, in the biogenic SOA factor for the 7 factor case, backward trajectories were mainly from west while those of the primary biogenic source were mainly from northwest. The CPF results for the biogenic SOA were mainly from southeast and southwest while those for the primary biogenic source were mainly from northeast. Marine origin aerosols were commonly reported as one of the organic sources in coastal areas (Meskhidze et al., 2011; Faccini et al., 2008) and southwest area from the AMY site is the Yellow Sea. Thus, the high possibility of southwest local wind to the sampling site might suggest impact from marine origin organic aerosols. The CPF results also support that the primary biogenic source is mainly affected from inland of Korea.

Mixed source was also divided into two factors, combustion related source and mixed sources in which local winds are mainly near to the site. The CPF results for these two factors were quite different with the result of mixed sources with 5 factors. While combustion related source shows high possibility at the northern side of sampling site, mixed sources factor shows high possibility at the southern side of it. Moreover, backward trajectories were clearly divided into two. Combustion source factor has backward trajectories mainly from northwest while mixed sources has trajectories from various directions. Thus, the mixed source in the 7 factor case is thought to be the combination of both local and regional one.

4. SUMMARY

The PMF model was applied to the one-year measurement data at the Anmyeon GAW station (AMY) to identify the major sources of carbonaceous aerosols. Based on the evaluation of the performance parameters, the optimal number of factors was verified as 5 or 7.

The major sources with 5 factors with relative contributions are; Secondary Organic Aerosols (SOA) (11%), biogenic sources (23%), local biomass burning (17%), transported biomass burning (17%), and mixed sources (32%). The major sources with 7 factors with relative contributions are; anthropogenic SOA (15%), biogenic SOA (15%), primary biogenic source (8%), local biomass burning (17%), transported biomass burning (16%), combustion related source (15%), and mixed sources (19%). With the case of 7 factors, the contribution from the mixed sources has decreased and the biogenic sources are further classified into primary and secondary biogenic sources, respectively.

Temporal variation of the relative contributions of each factor and the air parcel movement analysis results also support the characterization of the factors be a reasonable one. Based on the receptor modeling result, it is suggested that the AMY site is affected by both the regional transport and local emissions which were mainly biogenic sources and biomass burning. It turned out that at AMY, carbonaceous aerosols from anthropogenic origin were not as significant as urban areas.

Acknowledgments

This work was supported by Korea Meteorological Administration (CATER 2014-3190) and National Research Foundation of Korea (NRF-2017R1A2B4006760).

References

Aggarwal, S.G., Kawamura, K., (2009), Carbonaceous and inorganic composition in long-range transported aerosols over northern japan: Implication for aging of water-soluble organic fraction, Atmospheric Environment, 43, p2532-2540. [https://doi.org/10.1016/j.atmosenv.2009.02.032]
Ashbaugh, L.L., Malm, W.C., Sadeh, W.Z., (1985), A residence time probability analysis of sulfur concentrations at grand canyon national park, Atmospheric Environment (1967), 19, p1263-1270. [https://doi.org/10.1016/0004-6981(85)90256-2]
Brown, S.G., Eberly, S., Paatero, P., Norris, G.A., (2015), Methods for estimating uncertainty in pmf solutions: Examples with ambient air and water quality data and guidance on reporting pmf results, The Science of the Total Environment, 518-519, p626-635.
Cheng, Y., Li, S.-M., Leithead, A., Brickell, P.C., Leaitch, W.R., (2004), Characterizations of cis-pinonic acid and n-fatty acids on fine aerosols in the lower fraser valley during pacific 2001 air quality study, Atmospheric Environment, 38, p5789-5800. [https://doi.org/10.1016/j.atmosenv.2004.01.051]
Choi, E., Heo, J.-B., Yi, S.-M., (2010), Apportioning and locating nonmethane hydrocarbon sources to a background site in korea, Environmental Science & Technology, 44, p5849-5854. [https://doi.org/10.1021/es903634e]
Choi, N.R., Lee, S.P., Lee, J.Y., Jung, C.H., Kim, Y.P., (2016), Speciation and source identification of organic compounds in PM₁₀ over Seoul, South Korea, Chemosphere, 144, p1589-1596. [https://doi.org/10.1016/j.chemosphere.2015.10.041]
Facchini, M.C., Decesari, S., Rinaldi, M., Carbone, C., Finessi, E., Mircea, M., Fuzzi, S., Moretti, F., Tagliavini, E., Ceburnis, D., O’Dowd, C.D., (2008), Important source of marine secondary organic aerosol from biogenic amines, Environmental Science & Technology, 42(24), p9116-9121. [https://doi.org/10.1021/es8018385]
Fraser, M.P., Lakshmanan, K., (2000), Using levoglucosan as a molecular marker for the long-range transport of biomass combustion aerosols, Environmental Science & Technology, 34, p4560-4564. [https://doi.org/10.1021/es991229l]
Fu, P., Kawamura, K., Kobayashi, M., Simoneit, B.R., (2012), Seasonal variations of sugars in atmospheric particulate matter from gosan, jeju island: Significant contributions of airborne pollen and asian dust in spring, Atmospheric Environment, 55, p234-239. [https://doi.org/10.1016/j.atmosenv.2012.02.061]
Gogou, A., Stratigakis, N., Kanakidou, M., Stephanou, E.G., (1996), Organic aerosols in eastern mediterranean: Components source reconciliation by using molecular markers and atmospheric back trajectories, Organic Geochemistry, 25, p79-96. [https://doi.org/10.1016/s0146-6380(96)00105-2]
Han, J., Moon, K., Lee, S., Kim, Y., Ryu, S., Cliff, S., Yi, S., (2006), Size-resolved source apportionment of ambient particles by positive matrix factorization at gosan background site in east asia, Atmospheric Chemistry and Physics, 6, p211-223. [https://doi.org/10.5194/acp-6-211-2006]
Heo, J.-B., Hopke, P., Yi, S.-M., (2009), Source apportionment of PM_2.5 in seoul, korea, Atmospheric Chemistry and Physics, 9, p4957-4971.
Jaeckels, J.M., Bae, M.-S., Schauer, J.J., (2007), Positive matrix factorization (pmf) analysis of molecular marker measurements to quantify the sources of organic aerosols, Environmental Science & Technology, 41, p5763-5769. [https://doi.org/10.1021/es062536b]
Jung, C.H., Lee, J.Y., Kim, Y.P., (2016), Estimation of optical properties for HULIS aerosols at Anmyeon island, Korea, Aerosl Air Qual. Res. is submitted.
Jung, D.B., Cho, Y.S., Kim, I.S., Lee, J.Y., Kim, Y.P., (2015), Impact of energy consumption in northeast asia to the particulate pahs levels and composition at seoul, Aerosol and Air Quality Research, 15, p2190-2199. [https://doi.org/10.4209/aaqr.2015.01.0038]
Kanakidou, M., Seinfeld, J., Pandis, S., Barnes, I., Dentener, F., Facchini, M., Dingenen, R.V., Ervens, B., Nenes, A., Nielsen, C., (2005), Organic aerosol and global climate modelling: A review, Atmospheric Chemistry and Physics, 5, p1053-1123.
Kawamura, K., Kasukabe, H., Barrie, L.A., (1996), Source and reaction pathways of dicarboxylic acids, ketoacids and dicarbonyls in arctic aerosols: One year of observations, Atmospheric Environment, 30, p1709-1722. [https://doi.org/10.1016/1352-2310(95)00395-9]
Kim, B.M., Seo, J., Kim, J.Y., Lee, J.Y., Kim, Y., (2016), Transported vs. Local contributions from secondary and biomass burning sources to PM_2.5, Atmospheric Environment, 144, p24-36. [https://doi.org/10.1016/j.atmosenv.2016.08.072]
Kim, I.S., Lee, J.Y., Kim, Y.P., (2013), Impact of polycyclic aromatic hydrocarbon (pah) emissions from north korea to the air quality in the seoul metropolitan area, south korea, Atmospheric Environment, 70, p159-165. [https://doi.org/10.1016/j.atmosenv.2012.12.040]
Kotianová, P., Puxbaum, H., Bauer, H., Caseiro, A., Marr, I., Cik, G., (2008), Temporal patterns of n-alkanes at traffic exposed and suburban sites in vienna, Atmospheric Environment, 42, p2993-3005.
Lee, E., Chan, C.K., Paatero, P., (1999), Application of positive matrix factorization in source apportionment of particulate pollutants in hong kong, Atmospheric Environment, 33, p3201-3212. [https://doi.org/10.1016/s1352-2310(99)00113-2]
Lee, J.Y., Kim, Y., (2007), Source apportionment of the particulate pahs at seoul, korea: Impact of long range transport to a megacity, Atmospheric Chemistry and Physics, 7, p3587-3596. [https://doi.org/10.5194/acpd-7-1479-2007]
Lee, J.Y., Hwang, E.J., Lim, H.B., Kim, Y.W., Kim, E.S., Kim, Y.P., (2013), Composition of organic compounds in the ambient PM₁₀ of the Anmyeon Island, Particle and Aerosol Research, 9, p187-197, (in Korean). [https://doi.org/10.11629/jpaar.2013.9.3.187]
Lee, J.Y., Kim, Y.W., Kim, E.S., Kim, S.Y., Lee, H., Yi, S., Kwon, S.H., Kim, Y.P., (2011), Analysis of Water Soluble Organic Carbon (WSOC) and n-alkanes for the ambient PM₁₀ in the Anmyon Island, Particle and Aerosol Research, 7, p131-138, (in Korean).
Lee, J., Jang, Y.W., Kim, E.S., Kim, Y.P., Jung, C.H., Lee, J.Y., (2018), Temporal variation and characteristics of carbonaceous fractions in PM_2.5 measured at Anmyeon Island, a background site in Korea, In preparation.
Lee, S., Liu, W., Wang, Y., Russell, A.G., Edgerton, E.S., (2008), Source apportionment of PM_2.5: Comparing pmf and cmb results for four ambient monitoring sites in the southeastern united states, Atmospheric Environment, 42, p4126-4137. [https://doi.org/10.1016/j.atmosenv.2008.01.025]
Meng, J., Wang, G., Li, J., Cheng, C., Ren, Y., Huang, Y., Cheng, Y., Cao, J., Zhang, T., (2014), Seasonal characteristics of oxalic acid and related soa in the free troposphere of mt. Hua, central china: Implications for sources and formation mechanisms, Science of the Total Environment, 493, p1088-1097. [https://doi.org/10.1016/j.scitotenv.2014.04.086]
Meskhidze, N., Xu, J., Gantt, B., Zhang, Y., Nenes, A., Ghan, S.J., Liu, X., Easter, R., Zaveri, R., (2011), Global distribution and climate forcing of marine organic aerosol: 1. Model improvements and evaluation, Atmospheric Chemistry and Physics, 11(22), p11689-11705. [https://doi.org/10.5194/acp-11-11689-2011]
Minguillón, M., Querol, X., Baltensperger, U., Prévôt, A., (2012), Fine and coarse pm composition and sources in rural and urban sites in switzerland: Local or regional pollution?, Science of the Total Environment, 427, p191-202.
Moon, K., Han, J., Ghim, Y., Kim, Y., (2008), Source apportionment of fine carbonaceous particles by positive matrix factorization at gosan background site in east asia, Environment International, 34, p654-664. [https://doi.org/10.1016/j.envint.2007.12.021]
Norris, G., Duvall, R., Brown, S., Bai, S., (2014), Epa positive matrix factorization (pmf) 5.0 fundamentals and user guide prepared for the us environmental protection agency office of research and development, washington, dc, DC EPA/600/R-14/108.
Paatero, P., Eberly, S., Brown, S.G., Norris, G.A., (2014), Methods for estimating uncertainty in factor analytic solutions, Atmospheric Measurement Techniques, 7, p781-797. [https://doi.org/10.5194/amt-7-781-2014]
Paatero, P., Tapper, U., (1994), Positive matrix factorization: A non‐negative factor model with optimal utilization of error estimates of data values, Environmetrics, 5, p111-126. [https://doi.org/10.1002/env.3170050203]
Park, S.-H., Panicker, A., Lee, D.-I., Jung, W.-S., Jang, S.-M., Jang, M., Kim, D., Kim, Y.-W., Jeong, H., (2010), Characterization of chemical properties of atmospheric aerosols over Anmyeon (South Korea), a super site under global atmosphere watch, Journal of Atmospheric Chemistry, 67, p71-86. [https://doi.org/10.1007/s10874-011-9205-2]
Pekney, N.J., Davidson, C.I., Zhou, L., Hopke, P.K., (2006), Application of pscf and cpf to pmf-modeled sources of pm2.5in pittsburgh, Aerosol Science and Technology, 40, p952-961. [https://doi.org/10.1080/02786820500543324]
Polissar, A.V., Hopke, P.K., Paatero, P., Malm, W.C., Sisler, J.F., (1998), Atmospheric aerosol over alaska: 2. Elemental composition and sources, Journal of Geophysical Research: Atmospheres, 103, p19045-19057. [https://doi.org/10.1029/98jd01212]
Putaud, J.-P., Raes, F., Van Dingenen, R., Brüggemann, E., Facchini, M.-C., Decesari, S., Fuzzi, S., Gehrig, R., Hüglin, C., Laj, P., (2004), A european aerosol phenomenology—2: Chemical characteristics of particulate matter at kerbside, urban, rural and background sites in europe, Atmospheric Environment, 38, p2579-2595. [https://doi.org/10.1016/j.atmosenv.2004.01.041]
Rogge, W.F., Mazurek, M.A., Hildemann, L.M., Cass, G.R., Simoneit, B.R., (1993), Quantification of urban organic aerosols at a molecular level: Identification, abundance and seasonal variation, Atmospheric Environment. Part A. General Topics, 27, p1309-1330. [https://doi.org/10.1016/0960-1686(93)90257-y]
Saxena, P., Hildemann, L.M., (1996), Water-soluble organics in atmospheric particles: A critical review of the literature and application of thermodynamics to identify candidate compounds, Journal of Atmospheric Chemistry, 24, p57-109. [https://doi.org/10.1007/bf00053823]
Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R., (1996), Source apportionment of airborne particulate matter using organic compounds as tracers, Atmospheric Environment, 30, p3837-3855. [https://doi.org/10.1016/1352-2310(96)00085-4]
Simoneit, B.R., (1985), Application of molecular marker analysis to vehicular exhaust for source reconciliations, International Journal of Environmental Analytical Chemistry, 22, p203-232. [https://doi.org/10.1080/03067318508076422]
Simoneit, B.R., (1986), Characterization of organic constituents in aerosols in relation to their rigin and transport: A review, International Journal of Environmental Analytical Chemistry, 23, p207-237. [https://doi.org/10.1080/03067318608076446]
Simoneit, B.R., (1989), Organic matter of the troposphere—v: Application of molecular marker analysis to biogenic emissions into the troposphere for source reconciliations, Journal of Atmospheric Chemistry, 8, p251-275. [https://doi.org/10.1007/bf00051497]
Simoneit, B.R., Schauer, J.J., Nolte, C., Oros, D.R., Elias, V.O., Fraser, M., Rogge, W., Cass, G.R., (1999), Levoglucosan, a tracer for cellulose in biomass burning and atmospheric particles, Atmospheric Environment, 33, p173-182. [https://doi.org/10.1016/s1352-2310(98)00145-9]
Simoneit, B.R., Sheng, G., Chen, X., Fu, J., Zhang, J., Xu, Y., (1991), Molecular marker study of extractable organic matter in aerosols from urban areas of china, Atmospheric Environment. Part A. General Topics, 25, p2111-2129. [https://doi.org/10.1016/0960-1686(91)90088-o]
Watson, J.G., (1984), Overview of receptor model principles, Journal of the Air Pollution Control Association, 34, p619-623. [https://doi.org/10.1080/00022470.1984.10465780]
Watson, J.G., Chow, J.C., Fujita, E.M., (2001), Review of volatile organic compound source apportionment by chemical mass balance, Atmospheric Environment, 35, p1567-1584. [https://doi.org/10.1016/s1352-2310(00)00461-1]
Yuan, B., Shao, M., Gouw, J., Parrish, D.D., Lu, S., Wang, M., Zeng, L., Zhang, Q., Song, Y., Zhang, J., (2012), Volatile organic compounds (VOCs) in urban air: How chemistry affects the interpretation of positive matrix factorization (PMF) analysis, Journal of Geophysical Research: Atmospheres, 117. [https://doi.org/10.1029/2012jd018236]
Zhang, Y., Müller, L., Winterhalter, R., Moortgat, G., Hoffmann, T., Pöschl, U., (2010), Seasonal cycle and temperature dependence of pinene oxidation products, dicarboxylic acids and nitrophenols in fine and coarse air particulate matter, Atmospheric Chemistry and Physics, 10, p7859-7873. [https://doi.org/10.5194/acpd-10-13253-2010]
Zhang, Y., Sheesley, R.J., Schauer, J.J., Lewandowski, M., Jaoui, M., Offenberg, J.H., Kleindienst, T.E., Edney, E.O., (2009), Source apportionment of primary and secondary organic aerosols using positive matrix factorization (PMF) of molecular markers, Environment, 43(34), p5567-5574. [https://doi.org/10.1016/j.atmosenv.2009.02.047]

Species		Average	Max	Min	MDL	Missing+BDL (%)
Organic aerosols (4)	OC	4243	17491	701	234	0.0
	EC	453	1827	88	29	0.0
	WSOC	2416	11029	154	51	0.0
	HULIS-C	1769	7291	263	88	1.7
PAHs (14)	Phen	0.301	1.395	0.008	0.007	0.0
	Anth	0.050	0.100	0.000	0.014	28.3
	Flt	0.438	1.954	0.008	0.008	0.0
	Pyr	0.324	1.292	0.010	0.010	0.0
	BaA	0.112	0.483	0.005	0.012	3.4
	Chr	0.270	1.200	0.010	0.009	0.0
	BbF	0.609	11.062	0.000	0.007	0.0
	BeP	0.310	4.837	0.004	0.003	0.0
	BaP	0.139	0.574	0.003	0.012	3.4
	Per	0.040	0.070	0.010	0.009	74.1
	IdP	0.222	1.070	0.015	0.008	19.0
	DahA	0.028	0.027	0.085	0.004	42.0
	BghiP	0.119	0.439	0.000	0.013	15.5
	Cor	0.020	0.090	0.000	NA	72.4
n-Alkane (17)	C20	0.204	1.255	0.018	0.012	1.7
	C21	0.474	3.286	0.004	0.016	5.2
	C22	0.725	5.614	0.017	0.014	0.0
	C23	1.042	6.940	0.043	0.017	0.0
	C24	1.095	6.279	0.046	0.010	0.0
	C25	1.516	6.440	0.077	0.012	0.0
	C26	1.049	3.796	0.070	0.009	0.0
	C27	1.801	8.924	0.218	0.007	0.0
	C28	0.779	2.345	0.092	0.009	0.0
	C29	1.928	10.357	0.171	0.010	0.0
	C30	0.512	1.690	0.067	0.006	0.0
	C31	1.594	7.714	0.094	0.011	0.0
	C32	0.360	1.349	0.022	0.010	1.7
	C33	0.501	2.300	0.025	0.008	1.7
	C34	0.209	0.861	0.007	0.038	6.9
	C35	0.380	1.484	0.048	0.056	10.3
	C36	0.280	0.960	0.028	0.073	37.9
DCAs (19)	Malonic acid	70.4	425	0.283	0.094	1.7
	Methylmalonic acid	0.588	5.648	0.005	0.125	8.6
	Maleic acid	8.04	29.0	1.06	0.337	0.0
	Succinic acid	20.3	83.6	3.78	0.118	0.0
	Methylsuccinic acid	1.94	5.81	0.49	0.115	0.0
	Methylmaleic acid	0.848	3.28	0.139	0.209	0.0
	Fumaric acid	1.29	5.32	0.18	0.011	0.0
	Glutaric acid	5.01	16.9	0.571	0.108	0.0
	2-Methylglutaric acid	0.545	1.427	0.095	0.189	0.0
	D-Malic acid	16.39	52.1	0.388	0.120	0.0
	Adipic acid	2.19	7.80	0.454	0.408	0.0
	Pimelic acid	0.801	5.95	0.107	0.297	0.0
	Phthalic acid	8.38	36.2	0.470	0.181	0.0
	Suberic acid	0.990	3.92	0.209	0.270	0.0
	iso-Phthalic acid	0.972	3.31	0.094	0.243	0.0
	tere-Phthalic acid	6.91	23.0	0.416	0.102	0.0
	Azelaic acid	3.03	9.00	0.575	0.192	0.0
	Sebacic acid	0.506	1.932	0.030	0.293	0.0
	Undecanedionic acid	0.381	1.458	0.061	0.363	0.0
n-Alkenoic acids (18)	C8	0.690	1.87	0.070	NA	0.0
	C9	0.789	1.64	0.186	0.024	0.0
	C10	1.053	15.3	0.047	0.040	1.7
	C11	0.220	0.581	0.026	0.076	0.0
	C12	0.322	0.872	0.091	0.087	0.0
	C13	0.206	0.588	0.055	0.106	0.0
	C14	0.955	3.62	0.299	0.089	0.0
	C15	0.645	1.81	0.121	0.125	0.0
	C16	10.604	54.1	1.64	0.179	0.0
	C17	0.863	3.09	0.024	0.107	61.0
	trans-C18	0.415	1.848	0.026	0.207	0.0
	C18	5.29	33.4	0.659	0.695	0.0
	C20	1.26	7.81	0.100	0.085	0.0
	C21	1.60	10.1	0.032	0.217	0.0
	C22	2.66	23.5	0.122	0.041	0.0
	C23	2.60	11.7	0.000	0.062	0.0
	C24	3.50	27.3	0.159	0.089	0.0
	cis-Pinonic acid	3.21	16.7	0.357	0.103	0.0
Sugar (10)	Arabinose	1.11	14.7	0.087	0.123	10.3
	Ribose	5.94	21.1	0.056	0.132	22.4
	Levoglucosan	54.7	374	0.257	0.122	0.0
	Xylose	3.34	43.2	0.059	0.077	29.3
	Fructose	1.39	22.	0.031	0.186	3.4
	Mannose	1.32	3.79	0.063	0.041	72.4
	Galactose	2.71	15.0	0.138	0.058	74.1
	Glucose	2.32	23.3	0.020	0.073	0.0
	Sucrose	6.70	64.5	0.090	0.101	62.1
	Maltose	0.027	0.383	0.014	0.128	87.9

	4	5	6	7
IM	2.69	2.21	1.55	1.54
IS	5.38	3.31	3.10	3.00
%dQ	0.00021	0	0.01181	0
BS	94.75	85.40	89.83	84.29

	EC	WSOC, HULIS	PAHs	CPI_odd	C₂₇, C₂₉, C₃₁	DCAs	cis-Pinonic acid	Levo
(a) Summarized characteristics of the 5 factors.
Facto 1: SOA (11%)	-	+	-	1.62	-	++	-	-
Facto 2: Biogenic sources (23%)	++	+	-	2.24	+	+	-	-
Facto 3: Transported BB (17%)	-	-	++	1.5	-	-	-	++
Facto 4: Local BB (17%)	-	+	++	1.47	-	-	+	++
Facto 5: Mixed sources (32%)	-	-	+	2	+	+	-	+

(b) Summarized characteristics of the 7 factors.
	EC	WSOC, HULIS	PAHs	CPI_odd	C₂₇, C₂₉, C₃₁	DCAs	cis-Pinonic acid	Levo
Facto 1: Anthropogenic SOA (14%)	+	+	-	1.77	-	++	-	-
Facto 2: Biogenic SOA (15%)	-	+	-	1.89	-	-	+	-
Facto 3: Primary biogenic sources (8%)	-	-	-	10.49	++	-	-	+
Facto 4: Transported BB (16%)	-	-	++	1.51	-	-	-	+
Facto 5: Local BB (13%)	-	-	++	1.39	-	-	-	++
Facto 6: Combustion related sources (15%)	+	-	+	2.1	-	-	-	+
Facto 7: Mixed sources (19%)	-	-	+	1.64	+	-	+	+

Number of factors	5	7
Identified sources	SOA (11%)	Anthropogenic SOA (14%)
	Biogenic sources (23%)	Biogenic SOA (15%) Primary biogenic source (8%)
	Transported BB (17%)	Transported BB (16%)
	Local BB (17%)	Local BB (13%)
	Mixed sources (32%)	Combustion related sources (15%) Mixed sources (19%)