Water-Soluble Ionic Characteristics of Aerosols in the Marine Boundary Layer over the Yellow Sea during the KORUS-AQ Campaign

Major compositions of water-soluble ionic species in particulate matter less than 10 and 2.5 μm in diameter (PM10 and PM2.5, respectively) over the Yellow Sea were collected during the Korea–United States Air Quality (KORUS-AQ) campaign in 2016 onboard the research vessel Gisang 1. The secondary ionic species (NH4+, nss-SO42−, and NO3−) in PM10 and PM2.5 accounted for 84% and 89% of the total analyzed species. NH4+ was strongly correlated with non-sea salt (nss) SO42− (nss-SO42−) in PM10 and PM2.5; NO3− was closely correlated with Na+, Mg2+, and nss-Ca2+ in PM10 and NH4+ in PM2.5. High mass concentrations of methane sulfonic acid (MSA, CH3SO3−), the main source of natural sulfates over the Yellow Sea, were observed. The concentrations of MSA were found to show an increasing trend over the Yellow Sea in recent years. Biogenic sulfur contributions to the total nss-SO42− (MSA/nss-SO42− ratio) over the Yellow Sea ranged from 1.4% to 9.2% in PM10 and from 0.68% to 9.5% in PM2.5 during the cruise. Thus, biogenic nss-SO42− must be included, especially in the spring and early summer seasons, when biological activities are elevated in Northeast Asia. We classified the high aerosol mass concentration cases such as Asian dust and haze cases. In Asian dust cases, the ratio of NO3− to nss-SO42− in the aerosols showed that mobile (stationary) sources mainly affected PM10 (PM2.5). The major chemical species for Asian dust cases over the Yellow sea were CaCO3, Ca(NO3)2, Mg(NO3)2, Na(NO3)2, and sea salt. In haze cases over the Yellow sea, the contributions from stationary sources are high and the major species were (NH4)2SO4 and NH4NO3 in PM10 and PM2.5, respectively.


Introduction
Recent economic development in Northeast Asia, especially in China, has resulted in frequent occurrences of high aerosol mass concentration events in the region (Ding and Liu 2014;Wang and Chen 2016). Aerosols originating over land, from natural and/or anthropogenic sources, are deposited into surrounding ocean surfaces, which become the main sources of continental aerosols Zhang et al. 2004). The Gobi Desert and the Loess Plateau areas are key sources of mineral aerosols in Northeast Asia and the North Pacific Ocean (Zhang et al. 1993;Gao et al. 1997). Previous observations of aerosols over coastal seas and at a number of inland sites in Northeast Asia have focused on the spring and early summer seasons (Kim et al. 1998(Kim et al. , 2009Lee et al. 2002;Zhang et al. 2002). Northeast Asia has emerged as the world's largest source of SO 2 in recent years (Su et al. 2011). Furthermore, it has been reported that by 2020, NO x emissions in the region may increase five-fold compared to the levels in 1990 (Akimoto 2003). Recent some studies (Gu et al. 2013;Liu et al. 2016) reported the reduction in NO x emission trends over China. Although the reduction of NO x emission is found over China, the Northeastern Asia is the regions of highest NO x concentration in the world (Huang et al. 2017). Emissions from land sources in Northeastern Asia affect the aerosol field over the Yellow Sea, depending on atmospheric circulation patterns. Hence, the recent increase in anthropogenic emissions can alter aerosol compositions and their characteristics over the Yellow Sea.
The Korea-United States Air Quality (KORUS-AQ) campaign was an international, multi-organization mission to observe air quality across the Korean peninsula and surrounding waters. KORUS-AQ was conducted by the National Aeronautics and Space Administration (NASA) and their international partners from April to June 2016. National Institute of Meteorological Science (NIMS) of the Korea Meteorological Administration (KMA) measured atmospheric aerosols over the Yellow Sea, which was affected by aerosols originating in various regions of China and Korea during the campaign periods. KORUS-AQ integrated observations from ships, aircrafts, ground sites, and satellites, in conjunction with air quality models, to understand the factors governing air quality across urban, rural, and coastal interfaces in Northeast Asian (https://espo.nasa.gov/home/korus-aq).
Especially, MSA in the marine boundary layer is one of the major end products of the oxidation of dimethyl sulfide (DMS) produced by marine biota. MSA and DMS studies were conducted over the Yellows Sea by Chinese scientists to understand sulfur chemistry in the past (Chen et al. 2012). This study focuses on the NO 3 radicals which play an important role as an oxidant in DMS decomposition in the regions of highest NO x concentration such as Northeastern Asia (Gao et al. 1996). China emits large amounts of SO 2 and NO x (Han et al. 2005), which are transported to the Yellow Sea by prevailing westerlies and deposited there to enhance MSA production. For example, Gao et al. (1996) reported an MSA mass concentration of 0.029 μg m −3 in TSP (Total Suspended Particle) over the Yellow Sea and East China Sea. Zhang et al. (2015) reported an MSA mass concentration of 0.061 μg m −3 in TSP over the Yellow Sea in May 2012. Cha et al. (2016) reported an MSA mass concentration of 0.13 μg m −3 in PM 10 over the Yellow Sea in April 2015 (Table 1).
This study aims to understand the chemical characteristics of water-soluble aerosols over the Yellow Sea from shipborne sampling data of aerosols acquired during the KORUS-AQ campaign because the Yellow Sea is under the strongest air pollutant influence in the world (Wang and Chen 2016). Water-soluble ions are important constituents of atmospheric aerosols over ocean surfaces. In particular, heavy aerosol events, such as Asian dust and haze over the Yellow Sea, are mainly due to long-range aerosol transport from inland regions, such as China, Korea, and Japan. Therefore, this study focuses on analyzing the concentrations of the major watersoluble ionic compounds and methanesulfonic acid (MSA) in the high aerosol concentration cases observed during the KORUS-AQ campaign. The internal correlations between the chemical ionic species are examined in terms of the chemical species formation in these cases.

Data and Methodology
The KORUS-AQ campaign involved three cruises conducted using the research vessel Gisang 1, operated by the National Institute of Meteorological Science (NIMS) of the Korea Meteorological Administration (KMA). The first cruise started on May 3, 2016, and ended on May 12, 2016; the second cruise started on May 18, 2016, and ended on May 29, 2016; and the third cruise started on June 5, 2016, and ended on June 13, 2016. The samples obtained during these cruises over the region 35.3-37. 3°N and 123.2-125.2°E were selected for analysis in this study (Fig. 1).
Five-minute average PM 10 (particulate matter less than 10 μm in diameter) mass concentrations were measured using a PM 10 suspended particulate analyzer (β-ray PM 10 analyzer, Thermo Scientific Inc., FH62-C14; hereafter, β-ray PM 10 ) on the basis of a β-ray absorption method. A total of 17 samples were collected using 47-mm Teflon filters and a particle measuring system (APM Inc., PMS-104; hereafter, PMS) equipped with a PM 10 and PM 2.5 (particulate matter less than 2.5 μm in diameter) separator. The study mainly focused on the characteristics of the samples collected through the PM 10 and PM 2.5 inlets. An aerodynamic particle sizer (TSI Inc., APS-3321; hereafter, APS) was used to observe the aerosol particle size distribution. The measurable concentration range of APS is 0-10,000 cm −3 and the observation range is 0.5-20 μm with 52-bin channels. The β-ray PM 10 measures the PM 10 mass concentration every 5 min, while the APS measures the size-segregated number concentration every 3 min. The PMS was placed on the deck of the ship at 8 m above sea level. Samples of PM 10 and PM 2.5 aerosols were collected for around 10 h during daytime. That is, the water-soluble ionic mass concentration obtains from PMS for PM 10 and PM 2.5 and the total PM 10 mass concentration does from β-ray PM 10 . The size-segregated total PM mass concentration such as PM 10 and PM 2.5 is from APS.
Meteorological and oceanic parameters, such as temperature, wind direction, wind speed, and sea surface pressure, were observed every 5 min using automatic meteorological instruments installed on the ship. An aerosol observation container was installed on the bow of the ship to protect the instruments from marine hazards as well as to protect the ship from the effects of pollution sources, such as the smokestack. Figure 1 shows the main route of the research vessel over the Yellow Sea during the KORUS-AQ campaign. At the end of each cruise, aerosol samples collected on the filter were analyzed using an ion analyzer (ion chromatograph; hereafter, IC) to measure five types of cations (NH 4 + , Na + , K + , Ca 2+ , and Mg 2+ ) and five types of anions (SO 4 2− , NO 3 − , HCOO − , CH 3 COO − , and CH 3 SO 3 − ). The detection limit and the coefficient of variation of IC vary from 0.29 to 7.48 μg L −1 and 0.19% to 7.33%, respectively, depending on the ion type.

Characteristics of Aerosol Ionic Species in Total Samples over the Yellow Sea
The major secondary aerosol mean mass concentrations for nss-SO 4 2− , NO 3 − , and NH 4 + were 10.195, 3.025, and 3.436 μg m −3 , respectively, in PM 10 , and 8.823, 1.021 and 2.966 μg m −3 , respectively, in PM 2.5 . Further, the diameters of nss-SO 4 2− and NH 4 + were mostly less than 2.5 μm; hereafter, size denotes diameter. , respectively. In this study, these secondary ionic species accounted for 84% and 89% of the total analyzed species in PM 10 and PM 2.5 , respectively. This implies that the aerosols collected over the Yellow Sea during the campaign were strongly affected by the anthropogenic emissions over land.
Non-sea salt potassium (nss-K + ) is a good indicator of biomass burning (Simoneit 2002), and its diameter is less than 2.5 μm (Gaudichet et al. 1995). The nss-K + mass concentration over the Yellow Sea in this study is decreased than that in previous studies (e.g. Zhang et al. 2013;Zhang et al. 2015). The nss-Ca 2+ from soil (Kang et al. 2003) clearly shows a decreasing trend from land to ocean (Fig. 2). The nss-Ca 2+ mean mass concentration in this study was 0.4 μg m −3 in PM 10 and 0.1 μg m −3 in PM 2.5 . The difference between the nss-Ca 2+ mean mass concentrations of PM 10 and PM 2.5 was 0.3 μg m −3 on average, and the nss-Ca 2+ mean mass concentration of PM 2.5 was 0.1 μg m −3 on average. Thus, most of the nss-Ca 2+ was between 10 and 2.5 μm in diameter over the Yellow sea.
The correlation coefficient matrix was employed to analyze the internal relationships among different species. NH 4 + is strongly correlated with nss-SO 4 2− and the correlation coefficients between NH 4 + and nss-SO 4 2− are 0.84 and 0.97 for PM 10 and PM 2.5 , respectively (Tables 5 and 6). This implies that most of the NH 4 + was combined with nss-SO 4 2− . The correlation coefficients between NO 3 − and Na + , Mg 2+ , and nss-Ca 2+ for PM 10 were 0.73, 0.74, and 0.72, respectively. Thus, NO 3 − in the PM 10 samples existed as compounds of Na + , Mg 2+ , and nss-Ca 2+ . These results indicate that the marine and continental sources simultaneously affected the samples during the cruise. The mean mass concentration of MSA in PM 10 and PM 2.5 was 0.318 and 0.239 μg m −3 , respectively. The difference between the MSA mean mass concentrations of PM 10 and PM 2.5 was 0.079 μg m −3 on average and the MSA mean mass concentration of PM 2.5 was 0.239 μg m −3 on average. Thus, most of the MSA was in PM 2.5 ; the MSA mass concentration in PM 10 and PM 2.5 ranged from 0.096 to 0.61 μg m −3 and from 0.082 to 0.48 μg m −3 , respectively. In conjunction with the previous measurements (Zhang et al. 2013Cha et al. 2016), this study shows an increasing trend of the MSA mass concentration over the Yellow Sea in recent years and the MSA concentration in the central Yellow Sea is larger than that in other regions (see Fig. 2 and Table 1). The increase in MSA within aerosols in the Yellow Sea may be related to the changes in the inputs of the materials, which are related to the formation of MSA from dimethyl sulfide (DMS). High nutrient inputs from the Yellow River and the Yangtze River in China, which  (Wei et al. 2015), may have increased the formation of DMS and thus MSA. The deposition of mineral matter transported by air flow from inland deserts to the Yellow Sea (Hsu et al. 2009) can also affect the formation of DMS. The relationship among DMS, MSA, and marine productivity has been observed in many other regions (Calhoun 1992;Ayers et al. 1986;Park et al. 2017). Therefore, this study analyzed the detailed MSA mass concentration even though the MSA was not major watersoluble ions such as for nss-SO 4 2− , NO 3 − , and NH 4 + .

Classification of High PM Cases
High aerosol mass concentration cases during the experiment were classified using the β-ray PM 10 and APS. The differential mass concentration of a given aerodynamic diameter (dM Dae ) for each channel is calculated as where D ae is the aerodynamic diameter, dN Dae is the differential number concentration for a given aerodynamic diameter, ρ p is the density of the particle (1 g cm −3 , Katrib et al. 2005), and D ve is the volumetric equivalent diameter, all of which were obtained from the APS. PM 10 and PM 2.5 mass concentrations were calculated from the summation of channels 1-42 and 1-23, respectively. The unit for the differential mass concentration is μg m −3 . Seventeen-day data were analyzed after eliminating anchoring periods due to severe weather, fuel supply, and buoy release, as well as obvious contamination from the research vessel. We performed a mass-frequency distribution analysis for the PM 10 hourly series measured on the ship and in Seoul during the study period (Fig. 3). The average PM 10 mass concentrations over the Yellow Sea and in Seoul during the study period were around 40 and 48 μg m −3 , respectively. The measurements at the two sites were similar despite differences in the station type: the ship (Gisang 1) is a background station located on the sea, whereas Seoul is a megacity affected by many types of aerosol sources. The peak aerosol mass concentration was 30-40 μg m −3 on the ship and 40-50 μg m −3 at the Seoul station, and the frequency above 50 μg m −3 at Seoul was higher than that on the ship. Further, the frequency above 100 μg m −3 was higher than the frequency for the range 70-100 μg m −3 both on the ship and at Seoul station. It shows that both sites have similar frequency of mass concentration distribution. The cases above 100 μg m −3 during the campaign was from same high mass concentration episodes (the Asian dust and haze cases) at both site (not shown). That is, longrange aerosol transport from Chinese sources. Classification of the dust and haze events over the Yellow Sea in this study follows the work of Cho et al. (2013): for "Asian dust" ("haze"), the hourly-mean PM 2.5 /PM 10 ratio was below 40% (above 80%). Figure 4 shows the method for the classification of Asian dust and haze event by flowchart. Figure 5a shows the PM 10 mass concentration from the βray PM 10 , and Fig. 5b shows the ratios of PM 2.5 /PM 10 over the Yellow Sea. In Fig. 5b, the blue (red) regions indicate ratios below (above) 80% (40%). The dust and haze events were selected when the PM 10 concentration exceeded 50 μg m −3 , i.e., the value corresponding to the sum of the mean and one standard deviation of the PM 10 mass concentration over the Yellow Sea. On the basis of this criterion, one Asian dust event (Case I: May 7, 2016) and three haze events (Case II: May 12, 2016; Case III: May 21, 2016; and Case IV: May 29, 2016)) were identified ( Table 2). The aerosol chemical compositions were sampled in Cases I, II, and III for PM 10 , and in Cases I, II, and IV for PM 2.5 . The cases were selected if more than 50 μg m −3 was measured at least once using the β-ray PM 10 . The cumulative PM 10 mass concentrations collected using 47-mm Teflon filters using PMS during each case day were 73, 47, and 52 μg m −3 in Cases I, II, and III, respectively, and the PM 2.5 mass concentrations were 22, 30, and 52 μg m −3 in Cases I, II, and IV, respectively. The aerosol chemical compositions were not sampled in Case IV for PM 10 and Case III for PM 2.5 during the cruise.

Origins of High PM Cases
The HYSPLIT 4 model developed at the National Oceanic and Atmospheric Administration/Air Resources Laboratory was used to estimate the upstream path of air flow over the Yellow Sea during the KORUS-AQ campaign. The HYSPLIT simulations were run using the Unified Model-Global Data Assimilation and Prediction System (UM-GDAPS) weather data from KMA for 72 h prior to each case, at 500 m above the center of the vessel observation route (36.16°N, 124.29°E). Figure 6 shows the 72-h backward trajectories for the selected high PM 10 mass concentration cases from the HYSPLIT simulations. In Case I (Asian dust case; hereafter, AD), air from the Inner Mongolia region passed through the main source regions of Asian dust to affect the sampling Fig. 2 Comparison between the observed ranges for water-soluble ionic mass concentrations using PMS in this study and those in previous studies over the sea near Northeast Asia Korean Meteorological Society region. In Case II, i.e., the haze and mist from Northeast China (HMNC) case, air from Liaoning province in Western Manchuria stayed over the Yellow Sea for extended periods after leaving the land. The classification criterion of mist and haze is 75% RH: Mist ≥75% RH and Haze <75% RH (http:// www.kma.go.kr). In Case III, i.e., the haze from the Korean Peninsula (HKP) case, air pollutants from South Korea affected the sampling area. In Case IV, i.e., the haze from the Shandong Peninsula in China (HSPC) case, air flows originating in inland China passed through the Shandong Peninsula, an area of recent rapid industrialization, and stayed over the Yellow Sea for extended periods before arriving at the sampling site. Table 2 provides detailed information on the observations for the four cases.

Characteristics of Aerosol Ionic Species in High PM Cases
The mass concentration of nss-SO 4 2− is larger than that of the other ions, even though PM 10 and PM 2.5 were not sampled in HSPC and HKP, respectively (Fig. 7) Fig. 7 and Table 3). These results show that during the campaign, the Shandong Peninsula (HSPC) was the largest anthropogenic source of the ions over the Yellow Sea.
To identify the main sources of the sampled water-soluble ions, we calculated the Enrichment factor (EF) in aerosols using their contents in seawater and the soil crust. In this method, Na + and nss-Ca 2+ are regarded as conservative elements of the marine and crustal sources (Nishikawa et al. 1991;Millero 2006). The EFs of an ion X are defined as follows: and where the subscript "aerosol" denotes the mass concentration ratio between X and Na + or nss-Ca 2+ in the aerosol samples. The subscripts "seawater" and "crust" denote the ratio of X to Na + in seawater and X to nss-Ca 2+ in the crust, respectively. The Na + and nss-Ca 2+ mass concentrations for seawater and crust obtain from Nishikawa et al. (1991). . The EFs of nss-SO 4 2− for seawater source ranged from 13 in PM 10 in AD to 83 in PM 2.5 in HSPC. This implies that nss-SO 4 2− is from nonsea salt sources. The nss-SO 4 2− in AD is smaller than those in the haze cases (HMNC, HKP, and HSPC), suggesting that aerosols over the Yellow Sea in AD are affected more by the sulfates from seawater than in the other three cases. The EF of K + and nss-Ca 2+ in the four cases varies as follows: 5.1-10.0 (4.2-9.5) for K + in PM 10 (PM 2.5 ) and 10.0-33.8 (2.8-4.5) for nss-Ca 2+ in PM 10 (PM 2.5 ). The EF of K + for crustal source lies in the range of 0.3-0.5 (0.2-0.5) in PM 10 (PM 2.5 ), indicating that for the four cases, the K + and nss-Ca 2+ in the samples are mostly from the crustal source.
In recent years, Northeast Asia has emitted large amounts of SO 2 and NO x (Han et al. 2005). The oxidation of SO 2 and NO x can contribute to high levels of secondary ions, such as SO 4 2− and NO 3 − . NH 3 from agricultural activity and livestock farming (Galloway et al. 1996) and mineral dust transported from desert areas are rich in Ca 2+ ions, and they play an important role in neutralizing acid aerosols. To understand the relationship between the cations (NH 4 + and nss-Ca 2+ ) and the anions (nss-SO 4 2− and NO 3 − ), these two groups of ionic species were converted to equivalent concentrations for comparison. Figure 8 shows the correlation between the sum of the acidic ionic species (nss-SO 4 2− and NO 3 − ) and the sum of the alkaline ionic species (NH 4 + and Ca 2+ ) in the form of equivalent concentration, in aerosols sampled over the Yellow Sea. These two groups are strongly correlated, with a slope of 1.  from anthropogenic SO 2 and NO x are adsorbed onto wet sand aerosol particles, and are then combined with Ca 2+ by replacing CO 3 2− in CaCO 3 from crustal sources during the longrange transport (Zhang et al. 2013). In this study, nss-Ca 2+ showed a good correlation with NO 3 − (Tables 5 and 6).
Thus, the reaction processes of Ca 2+ play significant roles in reducing acidic aerosols over the Yellow Sea. NH 4 + has a strong correlation with nss-SO 4 2− and NO 3 − (Tables 5 and 6). This study applied the method of Rogula-Kozłowska et al. (2014) for estimating (NH 4 ) 2 SO 4 and NH 4 NO 3 . Figure 9 shows the estimated (NH 4 ) 2 SO 4 and NH 4 NO 3 . Most of the (NH 4 ) 2 SO 4 is in PM 2.5 , and the mass concentration in HSPC is extremely high. The mass concentration of (NH 4 ) 2 SO 4 varies in the order HSPC > HMNC > HKP > AD (Fig. 9), i.e., in the high aerosol mass concentration cases during the campaign, most of the (NH 4 ) 2 SO 4 was from China. NH 4 NO 3 does not appear in HMNC, and HKP shows large amounts of NH 4 NO 3 , indicating that the amount of NH 4 NO 3 is substantially reduced, while air flows travel over the sea surface for long periods after leaving the land, and vice versa. The NO 3 − to nss-SO 4 2− ratio in aerosols can be used to track the relative contribution of stationary and mobile sources to the secondary aerosols (Park and Lim 2006). Ko et al. (2017) estimated a ratio of 0.01-13.72 for PM 10 and 0.00-0.92 for PM 2.5 over the Yellow Sea. The ratios were 2.27, 0.23, and 0.73 for PM 10 in AD, HMNC, and HKP, respectively, and 0.36, 0.009, and 0.39 for PM 2.5 in AD, HNMC, and HSPC, respectively. The contributions from stationary sources are high for HMNC. Mobile (stationary) sources had a greater

MSA and Contribution of Biogenic SO 4 2− in High PM Cases
MSA in the marine boundary layer is one of the major end products of the oxidation of dimethyl sulfide (DMS) produced by marine biota. The formation of MSA from DMS oxidation is affected by the concentrations of OH and NO 3 radicals and temperature (Gao et al. 1996). The MSA mass concentrations are 0.21, 0.61, and 0.61 μg m −3 in PM 10 in AD, HMNC, and HKP, respectively, and 0.20, 0.48, and 0.17 μg m −3 in PM 2.5 in AD, HMC, and HSPC, respectively. The MSA mass concentration varies in the order HMNC > HKP > AD > HSPC, while the NO 3 − mass concentration varies in the order HSPC > AD > HKP > HMNC. MSA can be used to track the contribution of biogenic SO 4 2− to the total nss-SO 4 2− ; the ratio MSA/ nss-SO 4 2 represents the biogenic contribution. The ratio ranges from 1.4% to 9.2% for PM 10 and from 0.68% to 9.5% for PM 2.5 collected during the campaign. The ratio is 5.4%, 3.3%, and 5.8% for PM 10 in AD, HMNC, and HKP, respectively, and 6.1%, 3.6%, and 0.83% for PM 2.5 in AD, HMC, and HSPC, respectively. The ratios reported in previous studies vary widely according to the seasons and geographical locations. Chen et al. (2012) reported a ratio of 0.2%-6% in tropical regions, 6%-12% in unpolluted mid-latitudes, and 15%-93% near coastal Antarctica. Gao et al. (1996) observed a biogenic contribution of 10%-19% over the East China Sea for March-Jun. Arimoto et al. (1996) reported that marine biogenic sources accounted for 3.6% and 10.9% at the eastern and western sides of Jeju Island around the Yellow Sea, respectively. Zhang et al. (2013) also estimated the contribution of biogenic nss-SO 4 2− to be 12% over the North Yellow Sea in the spring season (April 23-May 5, 2007). The differences between this study and previous ones may be explained by the differences in the sampling periods and locations as well as the seasonal variation in MSA concentration, at least partially. For example, Mukai et al. (1995) found that the maximum (minimum) biogenic SO 4 2− occurred in the spring and early summer (winter) seasons. Even though the observation regions over the Yellow Sea were substantially affected by anthropogenic pollutants from Asia, the local biogenic nss-SO 4 2− cannot be ignored, especially in the spring and early summer seasons, when biological activities are elevated. In addition, estimations of biogenic contributions include considerable uncertainties, as the MSA formation in aerosols from DMS must consider a number of factors, such as the formation time of DMS by nutrient-fed phytoplanktons in the Yellow Sea, and the reaction processes of NO 3

Aerosol Size Distribution (ASD) and Chemical Species in High PM Cases
The APS observed the aerosol volume number concentration (Nvc) during the cruise (Fig. 10). Figure 10a shows that dust particles in AD increase the Nvc in the diameter range above 2.5 μm. Asian dust mostly affects the particle above 2.5 μm and Haze does it below 2.5 μm in Korea (Cho et al. 2013). In Fig. 10a, the increased the Nvc above 2.5 μm is from Asian dust. Figure 10b shows the ASD in HMNC. The Nvc rapidly increased in the diameter range below 1 μm owing to the haze effects, while the Nvc increased slightly in the diameter range of 1-4 μm owing to the sea salt effects. The sea salt concentration depends on the wind speed (Jaeglé et al. 2011). The wind speed in HNMC is the highest among the four cases, with a maximum (minimum) of 12 m s −1 (10.8 m s −1 ). Thus, the strong winds during the sampling period contributed to the formation of atmospheric sea salt over the Yellow Sea in HMNC. Figure 10c shows the ASD in HKP. The Nvc below the diameter range of 1 μm increased because of the haze. The haze had the strongest influence on the observation in HSPC among the four cases (Fig. 10d).
To identify the main chemical forms in the four cases, we compared the correlation of the ionic species with the ASD.  Basically, Asian dust mostly affects the particle above 2.5 um and Haze affects it below 2.5 um in Korea (Cho et al. 2013) and then the chemical forms of these ionic species were estimated on the basis of their correlation coefficients and concentrations. In Beijing in 2007, Wang et al. (2005) reported that the main chemical forms are CaCO 3 , CaSO 4 , and CaNO 3 in dust events, and (NH 4 ) 2 SO 4 and NH 4 NO 3 in haze events. We also calculated the CO 3 − , which was the main , which are from continental sources. The nss-CO 3 − is estimated using Eq. (3) with nss-Ca 2+ and nss-Mg 2+ . In Tables 5 and 6, the nss-SO 4 2− is very weakly correlated with the nss-Ca 2+ , indicating that during the KORUS-AQ campaign, CaSO 4 was not the main chemical over the Yellow Sea and the most of the NO 3 − is contained in PM 10 (See Fig. 9), and that nss-CO 3 − is closely correlated with Na + , Mg 2+ , and nss-Ca 2+ in PM 10 . Thus, during the KORUS-AQ campaign, NO 3 − was present in PM 10 over the Yellow Sea. Table 7 summarizes the relationships between the major ions and the concentrations of the main chemical species in the PM 10 and PM 2.5 samples over the Yellow Sea. CaCO 3 and MgCO 3 are the clearly dominant species for AD. In addition, coarse nitrate particles in PM 10 can be produced via the reaction of gaseous nitric acid with mineral aerosols, because NO 3 − is well correlated with Na + , Mg 2+ , and nss-Ca 2+ , and weakly correlated with NH 4 + in PM 10 . Thus, the Nvc increases in the diameter range above 2.5 μm in AD.
In HMNC, RH ranges from 83% to 90%. The increased Nvc in the diameter range below 1 μm was due to haze and mist.
When haze occurred over land, the air polluted with haze induced mist formations over the Yellow Sea. In such an environmental condition, the secondary inorganic aerosols (NH 4 + , nss-SO 4 2− , and NO 3 − ) exist mainly in the form of (NH 4 ) 2 SO 4 and NH 4 NO 3 in the diameter range below 1 μm (Rogula- Kozłowska et al. 2014). The (NH 4 ) 2 SO 4 mass concentration was 24 μg m −3 , and NH 4 NO 3 was not present in HMNC (Fig. 9). Thus, when RH and the mass concentration of nss-SO 4 2− are large, (NH 4 ) 2 SO 4 is the main component of the particles having a diameter of less than 1 μm (Song et al. 2008).
We used 3.27 × Na + for estimation of the sea salt mass concentration for the four cases, as shown in Fig. 11 (Ohta and Okita 1984). The mass concentration of sea salt is the highest in HNMC among the four cases. Sea salt is generally formed as NaCl. The sea salt size distribution ranges from 0.005 to 10 μm. Thus, (NH 4 ) 2 SO 4 and sea salt were the main chemical species in HNMC. The sea salts enhanced the mist in the observation site over the Yellow Sea, because they are hydrophilic particle in high RH (83% to 90%) environments.
In HKP and HSPC, the Nvc rapidly increased in the diameter range below 1 μm owing to the haze. The main species in both the cases were (NH 4 ) 2 SO 4 and NH 4 NO 3 . The mass concentration of (NH 4 ) 2 SO 4 is higher than that of NH 4 NO 3 . In particular, NH 4 + in PM 2.5 is better correlated with NO 3 − than that in PM 10 (Tables 5 and 6). The reactions in which HNO 3 replaces water-soluble particulates, such as formate, acetate, and oxalate, are also important formation pathways for fine   (Tabazadeh et al. 1998). Organic anions may be abundant in fine particles, such as those from biomass burning Andreae et al. 1988). In the future, to understand the generation of fine nitrate particles over the Yellow Sea in the presence of the reaction of gaseous nitric acid with gaseous ammonia, we will require additional details regarding NH 4 NO 3 chemical reactions of the sampled gaseous nitric acid with gaseous ammonia.

Summary and Conclusions
This study investigated the water-soluble ionic characteristics of atmospheric aerosols collected over the Yellow Sea during the KORUS-AQ campaign. High aerosol mass concentration cases during the campaign were classified in terms of the aerosol size distribution and mass concentration: "Asian dust" and "haze" for the hourly mean PM 2.5 /PM 10 ratio below 40% and over 80%, respectively. Backward trajectories corresponding to each case were analyzed using the HYSPLIT model. On the basis of these criteria and trajectory analyses, the high aerosol concentration cases during the campaign were classified into four groups: AD, HMNC, HKP, and HSPC. After the classification, we analyzed the watersoluble ions in the samples over the Yellow sea.
In four high PM cases, the mass concentration of nss-SO 4 2− varies in the order HSPC > HMNC > HKP > AD. Thus, most of the nss-SO 4 2− over the Yellow Sea in May 2016 came from the Shandong Peninsula and Northeast China. The NH 4 + mass concentration varies nearly identically to that of nss-SO 4 2− , except for the differences in the mass concentrations. The mass concentration of NO 3 − varies in the order HSPC > AD > HKP > HMNC. For the four cases, NO 3 − , Na + , Mg 2+ , and nss-Ca 2+ were mostly contained in PM 10 . These results suggest that NO 3 − is mostly combined with Na + , Mg 2+ , and nss-Ca 2+ in PM 10 during high aerosol mass concentration events over the Yellow Sea. This implies that ions such as Mg 2+ and nss-Ca 2+ in PM 10 among dust particles during AD reacted more with nitrates than with sulfates from anthropogenic sources in China. In AD case, the ratio of NO 3 − to nss-SO 4 2− in the aerosols showed that mobile (stationary) sources mainly affected PM 10 (PM 2.5 ). The major chemical species for Asian dust cases over the Yellow sea were CaCO 3 , Ca(NO 3 ) 2 , Mg(NO 3 ) 2 , Na(NO 3 ) 2 , and sea salt. In haze cases (HSPC, HKP, and HMNC) over the Yellow sea, the contributions from stationary sources are high and the major species were (NH 4 ) 2 SO 4 and NH 4 NO 3 in PM 10 and PM 2.5 , respectively.
The MSA mass concentrations are 0.21, 0.61, and 0.61 μg m −3 in PM 10 in AD, HMNC, and HKP, respectively, and 0.20, 0.48, and 0.17 μg m −3 in PM 2.5 in AD, HMC, and HSPC, respectively. The MSA mass concentration varies in the order HMNC > HKP > AD > HSPC, while the NO 3 − mass concentration varies in the order HSPC > AD > HKP > HMNC. High mass concentrations of methane sulfonic acid (MSA, CH 3 SO 3 − ), the main source of natural sulfates over the Yellow Sea, were observed. The concentrations of MSA were found to show an increasing trend over the Yellow Sea in recent years. Biogenic sulfur contributions to the total nss-SO 4 2− (MSA/nss-SO 4 2− ratio) over the Yellow Sea ranged from 1.4% to 9.2% in PM 10 and from 0.68% to 9.5% in PM 2.5 during the cruise. Thus, biogenic nss-SO 4 2− must be included, especially in the spring and early summer seasons, when biological activities are elevated in Northeast Asia.
In the future, we need to analyze element species such as Al and Fe in the aerosols because the element species such as Fe involves in production of DMS as the source of MSA and understand the more detail changes in various chemical species over the Yellow Sea.