Identification and Chemical Characteristics of Distinctive Chinese Outflow Plumes Associated with Enhanced Submicron Aerosols at the Gosan Climate Observatory

From October till November in 2010 and during March of 2011, when Chinese outflow events were frequently encountered, the chemical composition of submicron particles (PM1.0) was determined hourly using a particle-into-liquid sampler at the Gosan Climate Observatory. Three distinctive pollution plume types were identified: haze aerosols impacted by biomass combustion, nanoparticle bursts associated with outflow from Beijing, and saline soil particles from salt deposits. The highest PM1.0 concentration was observed in a fall haze event, under near-stagnant high-pressure synoptic conditions that were characterized by the lowest visibility (< 5 km) and the highest K and OC concentrations, indicating the influence of biomass combustion. When strong high-pressure systems develop in China, they efficiently export fresh urban emissions from Beijing to the study region, as distinguished by nanoparticle bursts of > 10 cm with highly elevated SO2 levels, even during the night. When air masses move rapidly from northeastern China to Gosan under strong wind conditions, the Ca concentration, along with that of Cl and Na, is enhanced in PM1.0, which is attributed to the influence of saline transport from dry lakes. The results of this study reveal compositional details and information on both number and mass concentration for different PM1.0 plumes from anthropogenic and natural sources, all of which are associated with different kinds of Chinese outflow events.


INTRODUCTION
Atmospheric aerosols are of great public concern owing to their effect on health (Lelieveld et al., 2015) and their association with radiative forcing, which occurs directly by light scattering and absorption and indirectly by affecting the cloud properties (IPCC, 2013).In Asian countries, the emissions of short-lived climate forcers, including lightabsorbing aerosols, are the largest warming sources (Sand et al., 2016).However, there is a large uncertainty in the by the public (Liu et al., 2013a).Numerous studies have focused on haze, particularly in eastern China where emissions from coal burning for energy requirements in the cold season, and biomass burning during harvests, result in the occurrence of haze (Kong et al., 2015).It is known that the mixture resulting from biomass burning and coal combustion, defined as "biosmoke" and used to be find in the late September in Northern China, deteriorates air quality far more than the individual components (Liu et al., 2013c) and impairs visibility (Huang et al., 2012).
In addition to high mass concentrations of submicron particles, elevated number concentrations of nanoparticles are also of concern.A sharp increase in number concentrations of nanoparticles to over 10 4 cm -3 has generally been observed in association with new particle formation (NPF; Kulmala et al., 2014).NPF involves a burst of particles in the nucleation mode (Dp = 3-20 nm), followed by continuous growth in size (Dall'Osto et al., 2013).Recently, Wang et al. (2013) reported that NPF events occur during haze episodes in Beijing and are concurrent with SO 2 concentrations high enough to dominate the condensation sink (Wu et al., 2007).Despite a reduction of around 50% in SO 2 emissions since 2005, the Beijing-Tianjin-Hebei region is still the highest SO 2 emitter globally (Fioletov et al., 2016).The high levels of SO 2 , mainly due to fossil fuel combustion, drives particle bursts with or without subsequent growth.While SO 2 plays a primary role in NPF events, the growth of particles is dependent on the concentration of organic species under low SO 2 concentration (Yue et al., 2010).In Beijing, approximately 12c of nanoparticle bursts occurred without subsequent growth during 9 months of continuous measurements by Wang et al. (2013).
In China soil particles, as distinct from urban aerosols, also occasionally cause high levels of aerosol mass loading that exceed air quality standards.During 2000-2013 the average PM 10 mass concentration ranged from 70 to 239 µg m -3 in different regions of Xinjiang with a highest annual concentration of 582 µg m -3 (Zhang et al., 2016).It is well known that PM 10 levels in Korea are heavily dependent on dust outbreaks transported from China (Ahmed et al., 2015;Crawford et al., 2015).Soil particles that originate from desert regions in northwestern China are efficiently transported by frontal systems to the Korean peninsula, Japan, and eventually to the west coast of the US (Hanaoka et al., 2011;Kim et al., 2016).Recently, alkaline soil was recognized as a source of fine dust particles in northeastern China, where dry lake deposits are spread over semi-desert regions (Li and Zhang, 2013).In northeastern China, concentrations of salt species including Cl -and Na + have increased due to the increase of saline land area and wind speed (Chen et al., 2009;Wang et al., 2014).Under prefrontal conditions, these saline soils are remobilized by strong winds and transported to the Korean peninsula, elevating salt contents in PM 1.0 as well as PM 10 (Park et al., 2011).
The Gosan Climate Observatory (GCO) is an ideal place to observe various types of Chinese outflows under different weather conditions.In Korea, PM 10 aerosols were the primary concern originally, and they were usually most prominent in spring (Arimoto et al., 1996).Nowadays the relevance of PM 2.5 has been recognized in East Asia with frequent haze occurrence, particularly in the cold season (September to February).The national exceedence threshold for PM 2.5 mass (1-day average) was established as 25 µg m -3 in Korea and 50 µg m -3 in China.At the GCO, measurement of PM 2.5 began with the Aerosol Characterization Experiments-Asia (ACE-Asia) program in 2001 (Schauer et al., 2003).Its chemical composition was also continuously determined using a particle-into-liquid sampler (PILS) during the ABC East Asian Regional Experiment 2005 (EAREX, 2005;Lee et al., 2007).In particular, PILSs are very useful for examining the evolution of aerosol events due to their high temporal resolution (approximately 1 hour; Orsini et al., 2003).However, few studies have been conducted at GCO that investigated PM 1.0 (Lim et al., 2012(Lim et al., , 2014)).In this study we present the first hourly PILS measurements of PM 1.0 at the GCO in spring and fall.Three contrasting types of Chinese aerosol outflow events were encountered during these observation periods and their chemical structure is comprehensively examined.

EXPERIMENTS
Two intensive PM 1.0 measurement periods were conducted at GCO (126°10′E and 33°17′N) on Jeju Island from October 21, 2010-November 05, 2010 and from March 09, 2011-March 21, 2011.A PILS system (ADI 2081, Applikon Analytical, Netherlands) was employed to continuously collect ambient PM 1.0 aerosols for subsequent measurement of water-soluble inorganic ions.The PILS was coupled with a PM 1.0 sharp-cut cyclone (SCC 2.229, BGI, USA) and a PM 10 impactor that were installed 10 m above the ground.Before entering the PILS body ambient air was passed through acid and base denuders (2000, URG, USA) to remove ambient gases.Inside the PILS aerosols grow continuously in an environment that is supersaturated by generating water vapor.A 0.05-ppbv lithium bromide solution (Sigma Aldrich, USA) was pumped into the head of the PILS at 0.37 mL min -1 as the internal standard for calculating the dilution factor (Timonen et al., 2010).The dilution factor ranged from 1.0 to 1.3 for the entire experiment (Appendix 1).The liquid effluent from the PILS was collected at 1-h intervals using a fraction collector (WFC III, Waters, USA) and frozen for subsequent analysis.
The water-soluble ions Cl -, SO 4 2-, NO 3 -, Na + , NH 4 + , K + , Ca 2+ , and Mg 2+ were analyzed via ion chromatography (IC; IC25, DIONEX, USA).For the IC analysis, the IonPac AS18 and IonPac CS12A columns were used with 39-mM potassium hydroxide and 20-mM methanesulfonic acid as eluents for the anions and cations, respectively.The detection limits for the anions and cations were no higher than 0.03 and 0.05 µg m -3 , respectively.A scanning mobility particle sizer (SMPS 3034, TSI, USA) was used to continuously measure the number concentrations of particles in 54 channels from 10 to 487 nm every three minutes.The SMPS analyzed the ambient aerosols at 2-and 1-h intervals along with aerosols in the potential aerosol mass (PAM) reactor (Kang et al., 2011(Kang et al., ) in 2010(Kang et al., and 2011, respectively. , respectively.Masses of the gaseous and aerosol species O 3 , CO, NO 2 , SO 2 , PM 10 , and PM 2.5 in the samples were measured by the National Institute of Environmental Research.In addition, organic carbon (OC) and elemental carbon (EC) for PM 2.5 were determined at 1-h intervals using an OC/EC analyzer (NIOSH Method 5040, Sunset Laboratory).Details about OC and EC measurements can be found in Batmunkh et al. (2011).Meteorological data were provided by the Korea Meteorological Administration (KMA).

RESULTS
Throughout the two intensive observation periods Chinese aerosol outflow events accounted for approximately one third of the total samples (Table 1).Sulfate was the most dominant component, followed by NH 4 + and NO 3 -.On average these three components comprised 75% of the water-soluble ions detected in PM 1.0 during entire experiment period.The episodic elevations of secondary ion levels were concurrent with increases in PM 10 and PM 2.5 mass concentrations.The hourly maximum concentrations of PM 10 (170 µg m -3 ), PM 2.5 (156 µg m -3 ), and SO 4 2-(21.7 µg m -3 ) were recorded during between 4-6 November 2010 when the air mass originated from the Jiangsu province of East China (Fig. 1).In this period, relative humidities were slightly higher, and considerable enhancement in concentrations of PM 1.0 particulate species and gaseous precursors were noted due to the relatively low wind speeds (Fig. 2).In particular, OC in PM 2.5 increased up to 27.3 µgC m -3 , resulting in a high OC/EC ratio (5.5).This period was identified as a haze event by the KMA.In comparison, NO 3 -, NH 4 + , and gaseous species, including NO 2 , CO, and SO 2 (up to 18.8 ppbv), were more abundant in air masses passing through Beijing areas during October 31, 2010-November 02, 2010, March 11, 2011-March 12, 2011, and March 16, 2011.These Beijing plumes were characterized by a high particle number concentration in Fig. 1.The three-day backward trajectories of air masses arriving at the Gosan Climate Observatory (GCO) for the six periods.Trajectories were calculated using TrajStat_1.2.2.6 and GDAS data (ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1).Hourly trajectories were averaged for each plume period specified in Table 1: H1 and H2 for the haze plume, B1, B2, and B3 for the Beijing plume, and D1 and D2 for the dust plume., ppbv, and km, respectively.Event periods are marked in the plot and specified at the top (see Table 1).
the nucleation mode exceeding 10 4 cm -3 (Table 2).In addition, the maximum PM 10 level (116 µg m -3 ) was recorded during March 16, 2011-March 18, 2011 when air masses were transported from the Mongolian desert region.Note that the soluble salts of Ca 2+ and Mg 2+ were also enhanced in PM 1.0 .On November 02, 2010-November 03, 2010, the Na + and Cl -levels in PM 1.0 were elevated to 1.7 and 1.3 µg m -3 , respectively.In this case, the air masses were transported from semi-arid regions in northeastern China, where dry lakes spread (Liu et al., 2013b).
In this study, different aerosol episodes are distinguished by the different characteristics of the submicron particles.All plumes, as summarized in Table 2, were associated with Chinese outflow events.It is noteworthy that the influence from local emissions was insignificant, based on the decadal measurements conducted at Gosan by Chambers et al. (2016).Haze plumes were identified by highly elevated PM 1.0 mass concentration, Beijing plumes were identified by nanoparticle bursts over 10 4 cm -3 , and dust plumes were identified by an enhanced soluble salt content in PM 1.0 that was not accompanied by an increase in mass.The haze episode was further divided into two phases based on visibility: thick (H1) and thin (H2) periods.The highest concentration and chemical content of PM 1.0 were concurrent with the lowest visibility in the H1 period.The three Beijing plumes observed in fall (B1) and spring (B2 and B3) were slightly different in chemical content, particularly the level of SO 2 and nanoparticle number concentration.Additionally, the dust plumes were separated into D1 and D2 based on the origin of their air masses.For these periods the air masses were traced backward for three days at a starting altitude of 850 m and the average trajectories are shown in Fig. 1.Some other types of air masses observed at the GCO came through the Korean Peninsula, East Sea, and East China Sea, which comprise two thirds of the total number of samples (Table 1).In these periods, the relative humidity was higher than that of Chinese outflow plumes but with lower concentrations of SO 2 , CO, major PM 1.0 constituents, and smaller nanoparticle  number than average (Table 2).However, the relatively high concentration of NO 2 was observed in Korean outflows.Therefore, the following discussion focuses on the three distinct kinds of polluted plumes of submicron aerosols observed in the Chinese outflows.

Haze Particles in Biomass Combustion-Affected Outflow
The haze event including H1 and H2 period was observed on November 05, 2010 when a high-pressure system (Figs.3(a) and 3(b)) resulted in near-stagnant conditions with relatively low wind speeds and slightly elevated relative humidity.Under the near stagnant synoptic conditions, concentrations of the aerosol species in PM 1.0 increased on average by a factor of four for SO 4 2-(15.7 µg m -3 ) and K + (1.3 µg m -3 ) and factor of six for PM 2.5 OC (19.2 µgC m -3 ) in comparison to the rest of the fall measurements.
Atmospheric visibility is often found to be reduced significantly below 10 km when the PM 1.0 and PM 2.5 levels are elevated during haze periods in China (Huang et al., 2014a;Tian et al., 2015;Zhang et al., 2015b).These studies concluded that the secondary species in PM 1.0 or PM 2.5 , including SO 4 2-, NO 3 -, NH 4 + , and OC, are responsible for visibility reduction.In this research, multi-regression analysis was performed to determine whether there is a clear relation between the chemical composition of PM 1.0 and visibility reduction during the haze events.Among the observed species, OC, EC, SO 4 2-, and NH 4 + were found to be strongly correlated with the variability in visibility, where the OC and EC concentrations of PM 2.5 were continuously measured every 1 h and used for analysis based on a previous study by Lim et al. (2012).In their study at GCO, the average OC and EC concentrations for PM 1.0 were 80% and 82% of those in PM 2.5 , respectively.The strong correlations between visibility and chemical species resulted in the following equation:

ln[OC] -0.4 ln[NH 4
+ ]+ 0.5 ln[NO 3 -] (µg m -3 ) (r 2 = 0.94, P = 0.03) (1) Although Eq. ( 1) suggests that visibility is most sensitive to EC concentration, the overall contributions were the highest from SO 4 2-(56%) and OC (38%), owing to their high concentrations.In comparison, the contributions of SO 4 2and OC to light scattering were 24% and 54%, respectively, during the winter haze event in Beijing, where the percentages were also used to evaluate the contribution to visibility (Wang et al., 2015).Regarding radiative forcing, the optical properties are found to be sensitive to chemical composition (Lim et al., 2014).Chemical composition and their relative abundance and aging processes are intimately coupled, thereby modifying absorption and scattering properties of aerosols while air masses are transported over the Yellow Sea.In Beijing winter, the amount of OC emitted from vehicles and near-cooking sources were greater than that of SO 4 2-by an average factor of 3, leading to high OC contribution to visibility reduction.In haze events observed at Gosan, the concentration of SO 4 2-and OC was comparable but with lager contribution of SO 4 2-to visibility degradation than OC.It implies the importance of sulfate in secondary formation due to its high hygroscopicity when air mass is transported over the Yellow Sea.
The visibility was calculated using Eq. ( 1) for the whole haze period as defined by the KMA and was found to be in good agreement with the experimental data (Fig. 4(a)).Particularly for the H1 event, when visibility was reduced to less than 5 km and the PM 2.5 level was over 75 µg m -3 , the agreement was excellent.However, the measured visibility was lower than that calculated for the H2 event.The China Meteorological Administration considers a haze event to have occurred when the visibility is below 5 km for over 6 h (Vautard et al., 2009;Zhang et al., 2015a).In Korea, haze is treated as a meteorological phenomenon for which the relative humidity (RH) and visibility are less than 75% and 10 km, respectively (KMA, http://www.kma.gov).According to the CMA definition only the H1 event constituted a haze event, which was when the visibility is reduced in proportion with the major constituents of fine aerosols (Fig. 4(b)).The high concentrations of SO 4 2-and OC suggest hygroscopic growth of particles under high RH, promoting visibility reduction (Tan et al., 2013).
The sulfur oxidation molar ratio SO 4 2-/(SO 4 2-+ SO 2 ) reflects the extent of the transformation from gaseous to aerosol sulfur, which was found to increase gradually up to 0.7 during a haze episode in the North China Plain (Yuan et al., 2015).In this study, the ratio was even higher (i.e., ≤ 0.8), in the thick haze period (H1), but decreased to 0.2 in the thin haze period (H2).In addition, the secondary OC (SOC) value, which is calculated by OC -EC × (OC/EC) min (Gu et al., 2010) comprised more than 60% and 40% of OC during the H1 and H2 periods, respectively.These results imply that the air mass was more aged for the H1 event (that meets the Chinese criteria for haze) than for the H2 event.In addition, based on the findings of Han et al. (2017), it is likely that the thick haze period (H1) was impacted by biomass combustion because K + and OC were significantly enhanced.Early November is the season for frequent agricultural burning in Central China (Cong et al., 2015).In China, the clearing of crops by burning has been reported to be a major source of OC during May-June and October-November (Liu et al., 2013c).In the H2 period, visibility was increased with a rapid increase in the SO 2 and NO 2 concentrations but with a slight decrease in the OC concentration (Fig. 2).

Nanoparticle Burst in Megacity Outflow
The B1, B2 and B3 air masses were transported from the Beijing region during October 31, 2010-November 02, 2010 (B1), March 11, 2011-March 12, 2011 (B2), and March 16, 2011 (B3) (Fig. 1).In these episodes, the GCO was located at the edge of a strong high-pressure system centered in Southeast China (Figs. 3(c) and 3(d)).This meteorological setting expedites the export of air pollutants from Beijing to the study region.The air masses were characterized by higher concentrations of the gaseous species SO 2 (B1 and B2, but lack of data for B3), NO 2 (B2, but lack of data for  B3), and CO (especially B2, and sharp increase in B3 event) in addition to particulate NH 4 + compared to those for the air masses of other episodes (Table 2).In particular, the ratio of NO 3 -/SO 4 2-was significantly enhanced on March 11.While the concentrations of submicron particles were highly elevated in the haze event, the large number of nanoparticles was conspicuous for these Beijing outflows.As the number concentration in the nucleation mode (i.e., particle size < 30 nm) was highly elevated (> 10 4 cm -3 ), these events were recognized as particle bursts.In a previous study at the GCO, particle bursts with subsequent growth events were frequently observed (Kim et al., 2013) and were considered as NPF events, even though the bursts began with particles larger than 10 nm.In the B1 plume, the particle burst around 20-30 nm at night was accompanied by a growth in size up to 30-70 nm (Fig. 5(a)).For the B2 plume, the particle number concentration reached 10 6 cm -3 with the highest concentrations of SO 2 , CO, and pre-existing particles occurring in the condensation mode (Fig. 5(b)).In particular, a large portion of the particles were smaller than 10 nm.In addition, a particle burst was observed during a dust event on March 16 (B3) (Fig. 5(c)) that was distinguished by an enhanced Ca 2+ concentration (Table 2).However, the SO 2 and nanoparticle concentrations were the lowest among the three events at 3.1 ppbv and 10 4 cm -3 , respectively, which was probably due to the influence of dust particles.
An NPF event was identified as a burst of particles smaller than 10 nm with a gradual growth in size up to 70 nm during the day (Wu et al., 2013).In addition, the dilution of pre-existing particles was recognized as a key factor for NPF (Song et al., 2010).In this research, high concentrations of nanoparticles (> 10 4 cm -3 ) were encountered when the air mass was transported from Beijing areas with high concentrations of SO 2 and pre-existing particles at night as well as in the day.Therefore, these three burst events were not necessarily the cases of NPF, even though the particles grew slightly in size for a short time period.This study was conducted as a part of the Gosan Pollution Experiment (GoPoEx) 2011, and the 14 C levels in PM 2.5 were measured (Krillova et al., 2013).Krillova et al. (2013) reported that the contribution of fossil fuels was dominant during the two burst events of B2 and B3 in March, 2011.Therefore, the burst of nanoparticles is a distinct characteristic of Beijing outflow plumes laden with urban emissions that are rapidly transported by a high-pressure system in the cold season.

Saline Dust Particles in Strong Northerly Outflows
During 2-3 November 2010 (D1) and 16-18 March 2011 (D2), the concentrations of both aerosol and gaseous species tended to decrease to their minimum levels (Table 2).However, the Ca 2+ concentration was enhanced in PM 1.0 without an increase in PM 10 mass.The concentration of Ca 2+ increased in PM 10 during dust events reported by Stone et al. (2011).However, there are no reports on dust occurrence during these two periods, for which the GCO was under strong storm conditions with heavy winds (Figs. 3(e) and  3(f) and Table 2).These are not typical weather conditions for dust occurrence that is usually accompanied with a frontal system (Kim et al., 2010).Instead, the weather conditions represent the continental background conditions of the study region for the cold season.Particularly for D1, the air mass directly came from the north and passed through northeastern China (Fig. 1), where alkaline soil spreads.
In a previous study, alkaline soil particles were reported to be observed in Korea (Park et al., 2013).The soluble salts mixed with typical soil particles were defined as saline dust in China (Abuduwailli et al., 2008(Abuduwailli et al., , 2010)).During the D1 period, the Ca 2+ concentration was enhanced in PM 1.0 , which is concurrent with the elevation of the Na + and Cl - levels, resulting in a high Cl -/SO 4 2-ratio (Fig. 6) that is comparable to those of saline dust particles observed in northeastern China and Duolun (Jiang and Zhang, 2001;Zhang et al., 2009).In contrast, the Cl -/SO 4 2-ratio of the D2 period was in the range of typical soil dust in Beijing (Zhang et al., 2009) and of pure dust events over the East Sea (Maxwell-Meier et al., 2004).While there were clear peaks for Na + , Mg 2+ , and Ca 2+ in November (D1), these species were enhanced for approximately two days in March (D2).In the weather map shown in Fig. 3(f), there are several places with dust outbreaks in dust source regions, where the air mass originated from (Fig. 1).Additionally, dust events were found occur on the previous days.Therefore, the soluble salts observed during March 16-March 18 (D2) were associated with typical soil particles being transported from the desert regions in Mongolia and Inner Mongolia to China.In contrast, the relatively high concentrations of Na + and Cl -and high Na + / Cl -ratio (1.14) during the D1 period possibly indicate the influence of saline dust (Vengosh, 2003), which has not been previously recognized as a typical dust particle in Chinese outflow plumes.These cases reveal that Northeast Asia is under the constant influence of soil dust particles that are transported from China and Mongolia, even though this is not recognized as episodes due to relatively low aerosol mass and submicron size.

SUMMARY AND CONCLUSIONS
This study provides the semi-continuous measurement results of water-soluble ions in PM 1.0 using a PILS at the GCO from October till November in 2010 and during March of 2011.For 36% of the total samples, air masses with highly elevated concentrations of aerosol and gaseous species were transported from China.In these Chinese outflows, the haze, urban, and dust plumes were distinguished by noticeable increases in mass, number, and soluble salts in submicron aerosols.All these plumes were observed under high-pressure systems, depending on their evolution phase during movement toward the study region (Fig. 3).
In late fall (November 5, 2010), a haze event was recognized by a significant increase in PM 1.0 mass concentration under near stagnant weather conditions, during which all measured species reached their maximum concentrations.In conjunction with the highest SO 4 2concentration (15.7 µg m -3 ), the K + and PM 2.5 OC concentrations increased to 1.3 µg m -3 and 19.2 µgC m -3 , respectively, indicating that the air mass was affected by biomass combustion.In this thick haze period (H1), the visibility was reduced to less than 5 km, which is one of the criterion for the occurrence of haze in China, and the air mass was slowly transported over the Yellow Sea from Jiangsu Province.The reduction in visibility was well expressed by a logarithmic increase in the major constituents of the aerosols, particularly during the H1 period.Overall, the visibility reduction was attributed mainly to SO 4 2-(56%) and OC (38%) for the whole haze episode.The contribution of SO 4 2-to visibility reduction was greater at Gosan compared to Beijing.These results indicate the importance of sulfate aerosol to visibility degradation in Chinese outflow plumes Fig. 6.Relation between the Cl -and SO 4 2-ions in PM 1.0 for saline soil dust (D1) and typical soil dust (D2).The Cl -/SO 4 2ratios are quoted from previous studies: 0.6-0.8 in northeastern China (Jiang and Zhang, 2001;Zhang et al., 2009) and 0.2 in Beijing (Zhang et al., 2009).
through chemical transformation and hygroscopic growth when air masses are transported over the Yellow Sea.A sharp increase in reactive gases, including SO 2 , NO 2 , and CO, was observed in air masses that were rapidly transported from the Beijing region.In one of these air masses, the number of nanoparticles was increased to approximately 10 6 cm -3 .In particular the highest concentration of SO 2 suggests fossil fuel combustion as the main source of the Beijing plumes.Under transportdominated condition, the nanoparticle burst occurred during the night as well as day with or without subsequent growth in size.Thus, nanoparticle bursts, regardless of the occurrence of NPF, are likely to be a good indicator of Beijing urban emissions pumped by strong high pressure in the cold season.
In addition to the submicron aerosol plumes of anthropogenic origin, the influence of soil particles was observed in PM 1.0 as enhanced Cl -, Na + , Ca 2+ , and Mg 2+ concentrations while all other species remained at their minimum levels for the entire experiment.For these periods, there were no reports on dust occurrence in Korea.Thus, the increase in Cl -without concurrent increase in SO 4 2-was attributed to the influence of alkaline soils, also referred to as saline dust, transported from dry lake deposits in the semi-arid regions of northeastern China.When the Ca 2+ concentration was increased in PM 1.0 for the D1 period, the enhancements in the Na + and Cl -concentrations were more visible compared to those for the typical soil plume of D2.This resulted in a high Cl -/SO 4 2-ratio of 0.6 for saline soil dusts, which distinguishes them from typical soil dust.This result reveals that the influence of soil particles is more prevalent than currently thought in the study region and suggests that saline particles need to be considered as soil dust particles.

Fig. 4 .
Fig. 4. (a) Comparison of measured and calculated visibility and (b) logarithmic decrease of visibility according to SO 4 2-

Fig. 5 .
Fig. 5. Size segregated ambient aerosol number distribution for Beijing plumes on (a) October 31-November 1 (B1), (b) March 11 (B2), and (c) March 16 (B3).Mode diameters are given in black circle for burst periods.The deep violet color in between indicates no measurement because ambient measurements were performed every hour.

Table 1 .
Number of hourly particle-into-liquid sampler (PILS) measurements classified into three types of Chinese outflow plumes and others during the experiments conducted at the Gosan Climate Observatory (GCO) in fall 2010 and spring 2011.

Table 2 .
Measurement statistics (average ± σ) for comparison of aerosol event (Chinese outflow plumes) and non-event days observed at the GCO, where NA means not available due to lacking of data (only two samples).