Observed Chemical Characteristics of Long-Range Transported Particles at a Marine Background Site in Korea

ABSTRACT

Deokjeok Island is located off the west coast of the Korean Peninsula and is a suitable place to monitor the long-range transport of air pollutants from the Asian continent. In addition to pollutants, Asian dust particles are also transported to the island during long-range transport events. Episodic transport of dust and secondary particles was observed during intensive measurements in the spring (March 31–April 11) and fall (October 13–26) of 2009. In this study, the chemical characteristics of long-range-transported particles were investigated based on highly time-resolved ionic measurements with a particle-into-liquid system coupled with an online ion chromatograph (PILS-IC) that simultaneously measures concentrations of cations (Li⁺, Na⁺, NH₄ ⁺, K⁺, Ca²⁺, Mg²⁺) and anions (F⁻, Cl⁻, NO₃ ⁻, SO₄ ²⁻). The aerosol optical thickness (AOT) distribution retrieved by the modified Bremen Aerosol Retrieval (M-BAER) algorithm from moderate resolution imaging spectroradiometer (MODIS) satellite data confirmed the presence of a thick aerosol plume coming from the Asian continent towards the Korean peninsula. Seven distinctive events involving the long-range transport (LRT) of aerosols were identified and studied, the chemical components of which were strongly related to sector sources. Enrichment of acidic secondary aerosols on mineral dust particles, and even of sea-salt components, during transport was observed in this study. Backward trajectory, chemical analyses, and satellite aerosol retrievals identified two distinct events: a distinctively high [Ca²⁺+Mg²⁺]/[Na⁺] ratio (>2.0), which was indicative of a preprocessed mineral dust transport event, and a low [Ca²⁺+Mg²⁺]/[Na⁺] ratio (<2.0), which was indicative of severe aging of sea-salt components on the processed dust particles. Particulate Cl⁻ was depleted by up to 85% in spring and 50% in the fall. A consistent fraction of carbonate replacement (FCR) averaged 0.53 in spring and 0.55 in the fall. Supporting evidences of Cl⁻ enrichment on the marine boundary layer prior to a dust front were also found.

Supplemental materials are available for this article. Go to the publisher's online edition of the Journal of the Air & Waste Management Association for sector and air mass classifications of clean and LRT cases.

IMPLICATIONS

The chemical characteristics of aerosol particles evolve as they undergo long-range transport (LRT) in the atmosphere from the source region. Aside from meteorological conditions, the signature of the source region influences the loading and chemical characteristics of LRT aerosols.

INTRODUCTION

Asian continental outflow transports large quantities of primary and secondary aerosols over the Korean peninsula during the spring and fall seasons, mostly influenced by the passage of cold fronts.1 Observations during the spring have been extensively studied during the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia)2–5 and other field campaigns (Joint Research on Long-Range Transboundary Air Pollutions in Northeast Asia [LTP],6 Transport and Chemical Evolution over the Pacific (TRACE-P)7,8) These aerosols are either emitted directly from multiple sources or produced through various atmospheric processes. Furthermore, atmospheric aerosols engage in various processes during transport that modify their chemical composition, mixing state, and reactivity.9–12

Particularly in the northeast region (China, Mongolia, Siberia, Korea, Japan) of the Asian continent, source regions of mineral dusts include loess soils from the arid Gobi desert of northern China and Mongolia, and the Taklimakan desert of western China.13 Mineral dust particles are considered the largest single contributor to transported particles in the atmosphere14 and are susceptible to processing and chemical transformation during transport. Mineral dust particles can also be coated with sulfate and nitrate,12,15,16 as well as a carbonaceous10 or sea-salt17 component via heterogeneous processes during transport. These in turn may be removed from their surface, making them more available for processing and the production of secondary aerosols.9 Such processes would determine the chemical characteristics of the resulting aerosol. The long-range transport of mineral dust associated with urban pollution has therefore become one of the major challenges in Northeast Asia.

Long-range-transported aerosol particles can impact an environment a considerable distance from its source. However, limitations concerning the availability of long-term monitoring data have made it difficult to achieve scientific and political consensus regarding the relative impacts of long-range-transported aerosol particles. The Northeast Asian countries have undertaken a wide range of collaborative activities in order to build a scientific consensus on the basis of coordinated multilateral research activities. The Joint Research on Long-Range Transboundary Air Pollutions in Northeast Asia (LTP),6,18 the so-called “LTP Project”, was established to characterize the transport and transformation of atmospheric aerosols over Northeast Asia, and to identify their sources based on continuous, intensive, and remote sensing measurements and receptor models. Long-term (continuous) and intensive (seasonal) measurement campaigns were implemented at multiple surface sites in Northeast Asia. The 2009 LTP Project initialized the use of highly time-resolved aerosol chemistry measurements to study the transport of pollutants to a background site in Korea.

This study reports the results from the 2009 LTP intensive measurement campaigns at a background marine site in Korea. The aim of the intensive measurement was to investigate the chemical characteristics of local and long-range-transported Asian aerosols under different synoptic conditions from the unique Northeast Asian outflow. In addition, satellite-retrieved aerosol optical thickness and air mass back trajectory analyses were used to identify the spatial distribution and transport of aerosols. Furthermore, detailed analysis was performed to characterize the degree of processing of long-range-transported mineral dust and secondary species.

EXPERIMENTAL METHODS

The LTP ground measurement site (Deokjeok Island, 37°13′33″N, 126° 8′51″E, 50 m above sea level [ASL]) is located off the west coast of the Korean peninsula in the effective downwind area of the Asian continent (Figure 1). Incheon Metropolitan City is located just 40 km east, whereas the capital city, Seoul, is located 75 km east of the measurement site.

Figure 1. Sampling locations (clockwise from left): (a) Northeast Asian region (including the Asian Continent, mainly China and Mongolia, Korea, and Japan), divided by dashed lines into sector sources (Sectors I–IV, KP). (b) Deokjeok Island off the west coast of the Korean peninsula. The island is a background marine site. (c) GIST mobile laboratory, situated on Deokjeok Island, equipped with ambient air quality measurement instruments.

Measurements were taken during the spring (March 31 to April 11) and fall (October 14 to October 26) of 2009. A particle-into-liquid system (PILS; BMI, Hayward, CA, USA)¹⁹ coupled with an online ion chromatograph (IC) that simultaneously measures cations (Li⁺, Na⁺, NH₄ ⁺, K⁺, Ca²⁺, Mg²⁺) and anions (F⁻, Cl⁻, NO₃ ⁻, SO₄ ²⁻) was installed inside a mobile laboratory at the Deokjeok Island monitoring site. Ambient aerosols pass through a PM_2.5 (particulate matter with an aerodynamic diameter ≤2.5 μm) size cut impactor, and then through an annular and carbon denuder for removal of reactive gaseous components. The overall PILS sampling flow rate was maintained at 12.8 L/min by a critical orifice. The remaining particles were mixed with high-purity steam (18 MΩ,) injected coaxially, and maintained at a flow rate of 1.5 mL/min and a steam temperature of 98 ± 5 °C. The grown particles, now droplets, were collected by an impactor. An internal standard (LiF) of known flow rate and concentration was used to collect and dissolve the droplets. These droplets are then subsequently injected into the IC (Metrohm, model 850, Herisau) for ionic analysis in a 30-min interval. Recovery of LiF was equal to 75% ± 17% for Li⁺ and 85% ± 10% for F⁻. Particulate matter (PM_2.5) mass concentration was also measured in a 30-min interval using an optical particle counter (OPC; Grimm, model 1.107, Ainring, Germany) 120-Hr back trajectories ending at 200 m ASL at the Deokjeok Island site were analyzed using Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model 4.20 Input data for HYSPLIT utilized the meteorological data from the FNL (Final Run) framework that are archived from the GDAS (Global Data Assimilation System) of NCEP (National Centers for Environmental Protection).20 The computation period from March 31 to April 11, 2009, included all possible in situ monitoring every 6 hrs, ending at 12:00 a.m. Coordinated Universal Time (UTC) (09:00 a.m. Korean Standard Time [KST]), 06:00 a.m. UTC (3:00 p.m. KST), 12:00 p.m. UTC (9:00 p.m. KST), and 6:00 p.m. UTC (03:00 a.m. KST). The aerosol optical thickness (AOT) over the measurement region was retrieved using the MODIS Terra and Aqua hierarchical data format (hdf) satellite data utilizing the modified Bremen Aerosol Retrieval (M-BAER) algorithm.21–23 The study domain covered approximately the Northeast Asian region from 100°E to 145°E and 20°N to 45°N, which includes all of Korea, Japan, most of eastern China, and parts of Mongolia and Russia. Aerosol optical thickness images by M-BAER are integrated over the entire atmospheric column and at a fixed time frame. Meteorological parameters, including wind speed, wind direction, precipitation, temperature, and humidity, were also measured by an automated weather station (AWS). Highly time-resolved ionic chemical composition, PM mass concentration from OPC, and meteorological parameters were averaged to hourly data.

RESULTS AND DISCUSSION

Air Mass History and Temporal Evolution of Secondary Species

The synoptic events and source regions are summarized in Table 1. Descriptive statistics and the contribution of each chemical component to the PM_2.5 mass are presented in Table 2. Aerosol optical thickness (AOT) retrieved from satellite data using the M-BAER19 algorithm combined with HYSPLIT18 back trajectory analyses were used to examine how the aerosol was transported spatially. The air mass pathways arriving in Deokjeok Island are categorized into five sectors. Those originating from Siberia, followed by Northeast China and the Yellow Sea, are categorized as Sector I. Those coming from the arid regions of Mongolia and Central China, through the Shandong peninsula, and passing over the Yellow Sea are categorized as Sector II. Those coming from the east coast of China (through the Yellow Sea) in a southern direction are categorized as Sector III. Those originating from Russia, northeastern China, and entering the Korean peninsula through the East Sea towards Deokjeok Island are categorized as Sector IV. Lastly, stagnant air mass pathways circulating over the Korean peninsula are categorized as KP. The sector regions are illustrated together in Figure 1a.

Table 1. Summary of event classifications and source regions

Synoptic Event	Source Regions	Timeline (KST, 2009)
Clean	Sectors II and III	April 2 11:00 p.m. to April 4 09:00 a.m.
LRT-1	Sector I	March 31 09:00 a.m. to April 2 10:00 p.m.
LRT-2	Sector II	April 4 10:00 a.m. to April 8 09:00 a.m.
LRT-3	Sector III-KP	April 8 10:00 a.m. to April 11 6:00 p.m.
LRT-4	Sectors I and II	October 13 8:00 p.m. to October 15 11:00 p.m.
LRT-5	Sector II	October 16 12:00 a.m. to October 22 12:00 a.m.
LRT-6	Sector II, extending from West Russia	October 22 01:00 a.m. to October 24 08:00 a.m.
LRT-7	Sector II-KP	October 24 09:00 a.m. to October 26 12:00 p.m.

Table 2. Descriptive statistics of PM_2.5 and its chemical components. ^a

	PM2.5	Na^+	NH4 ^+	K^+	Ca^2+	Mg^2+	Cl^−	NO3 ^−	SO4 ^2−
DL ^b	0.1 ^c	0.001	0.0003	0.00002	0.001	0.0004	0.001	0.001	0.001
Synoptic events
Clean	14.9 ± 2.7 (100)	0.6 ± 0.0 (4.0)	1.1 ± 0.2 (7.2)	0.08 ± 0.01 (0.6)	0.2 ± 0.2 (1.2)	0.10 ± 0.03 (0.7)	BDL ^d (0.0)	0.5 ± 0.2 (3.5)	2.2 ± 0.5 (15.3)
LRT-1	43.5 ± 22.9 (100)	0.7 ± 0.1 (2.0)	3.1 ± 1.6 (6.4)	0.3 ± 0.2 (0.7)	0.5 ± 0.3 (1.3)	0.3 ± 0.1 (0.7)	0.2 ± 0.2 (0.5)	3.3 ± 2.1 (6.8)	5.6 ± 3.4 (11.1)
LRT-2	69.8 ± 20.3 (100)	0.7 ± 0.1 (1.3)	6.3 ± 3.2 (9.5)	0.5 ± 0.5 (0.7)	0.6 ± 0.2 (1.2)	0.2 ± 0.1 (0.4)	0.2 ± 0.1 (0.2)	6.0 ± 3.8 (8.6)	14.6 ± 8.1 (22.4)
LRT-3	74.1 ± 15.1 (100)	0.3 ± 0.5 (0.5)	10.3 ± 3.0 (14.7)	0.8 ± 0.4 (1.1)	0.5 ± 0.4 (0.8)	0.1 ± 0.1 (0.1)	0.1 ± 0.2 (0.2)	5.6 ± 3.0 (7.8)	25.4 ± 9.9 (36.4)
LRT-4	13.2 ± 7.0 (100)	0.8 ± 0.1 (8.1)	0.9 ± 0.8 (7.7)	0.2 ± 0.1 (1.4)	0.4 ± 0.3 (3.3)	0.1 ± 0.1 (1.2)	0.7 ± 0.1 (8.0)	0.9 ± 0.4 (8.3)	3.3 ± 2.2 (28.8)
LRT-5	23.0 ± 15.0 (100)	1.2 ± 0.4 (7.0)	0.1 ± 1.4 (0.4)	0.1 ± 0.2 (0.4)	0.6 ± 0.8 (2.8)	0.2 ± 0.2 (1.0)	1.3 ± 1.0 (8.0)	1.0 ± 2.5 (4.9)	1.1 ± 3.9 (5.5)
LRT-6	35.2 ± 14.7 (100)	1.0 ± 0.2 (4.3)	1.2 ± 1.3 (2.6)	0.4 ± 0.3 (1.0)	1.1 ± 0.8 (3.2)	0.3 ± 0.2 (1.0)	0.9 ± 0.2 (3.8)	3.2 ± 2.5 (7.8)	4.1 ± 3.5 (11.5)
LRT-7	37.1 ± 9.6 (100)	0.79 ± 0.05 (2.2)	2.9 ± 1.6 (6.5)	0.4 ± 0.2 (0.9)	0.9 ± 0.4 (2.4)	0.3 ± 0.1 (0.7)	0.9 ± 0.1 (2.5)	3.2 ± 1.3 (7.6)	8.4 ± 4.3 (19.8)
Notes: ^aAverage concentration ± standard deviation, μg/m^3. Contribution to PM2.5, in percent, %.^bDL = detection limit, μg/m^3; calculated using the data from periodic injection of method blank sample during sampling flow rate check and transport liquid flow rate check.^cManufacturer specifications.^dBDL = below detection limit.

Long-range transport events of secondary aerosol species and mineral dust can thus be verified over the Northeast Asian grid from satellite-based AOT distribution and pathway directions (see Supplementary Materials; available in the publisher's online edition). In addition, the differences in chemical composition can be associated with air mass sources (Figure 2a and b). Generally, changes in ionic composition observed on the ground are strongly associated with changes in air mass origins. The chemical characteristics of each event are summarized in Table 1.

Figure 2. The temporal evolution of ground-based chemical composition during clean and long-range transport events. The top arrow range describes the time periods corresponding to different air mass source regions. The x-axes are scaled to 1 day, with grid lines corresponding to 0000 KST. (a) Spring and (b) fall intensive periods.

On the basis of local wind patterns (Figure 3), most LTP cases influenced by westerly wind experienced an increase in the distribution of ionic components (Figure 3b, c, e) and dust components (Figure 3f, g). When Kim et al. conducted a 24-hr sampling and measurements of aerosol components in Deokjeok and Gosan from about 2005 to 2007, they observed a 40% increase in secondary aerosols, and most wind cases appeared to be westerly in origin, indicating the influence of long-range-transported (LRT) pollutants from eastern China.6

Figure 3. Wind rose plots of episodic events during spring and fall intensive periods. Top row, from left to right: (a) Clean, northwest wind; (b) LRT-1, west and northwest; (c) LRT-2, west; and (d) LRT-3, southeast. Bottom row, from left to right: (e) LRT-4, northwest; (f) LRT-5, west; (g) LRT-6, north and southwest; and (h) LRT-7, southeast.

It is apparent from Figure 2b that ground-based concentrations of chemical components of atmospheric aerosols were minimal for the period April 2 11:00 p.m. to April 4 09:00 a.m. KST. Although the local wind direction indicated it was predominantly northwest of the island (Figure 3a), the measurement site was mostly impacted by clean marine air from the Yellow Sea. Moreover, the average PM_2.5 mass concentration was also found to be minimal (14.9 ± 2.7 µg/m³). Thus, this period has been classified as “Clean”.

The spring intensive measurement period is characterized by air mass coming from the northeast and eastern regions of China. The fall intensive measurement period is characterized by a 120-hr air mass originating from the arid regions of Mongolia and China, as well as LRT dust transport events. The dust front that passed over the Korean peninsula (KP; Table 2) also brought in elevated levels of secondary aerosol particles. The LRT-3 event during spring and the LRT-7 event during fall resulted in the highest average PM_2.5 masses of 74.1 ± 15.1 and 37.1 ± 9.6 µg/m³, respectively. A common factor between the two events is the possible accumulation of aerosol particles while the air mass lingers slowly around the KP (Figure 3d and h). In addition, the relatively warm (17 to 22 °C) and humid (relative humidity [RH] >80%) air favored the formation of secondary aerosol such as (NH₄)₂SO₄. Both events also accumulated the highest concentration of NH₄ ⁺ (10.3 ± 3.0 and 2.9 ± 1.6 µg/m³ during LRT-3 and -7, respectively) and SO₄ ²⁻ (25.4 ± 9.9 and 8.4 ± 4.3 µg/m³ during LRT-3 and -7, respectively).

The air mass originating from Sector I during the LRT-1 event in spring exhibited the lowest concentration of secondary acids: 3.3 ± 2.1 µg/m³ for NO₃ ⁻ and 5.6 ± 3.4 µg/m³ for SO₄ ²⁻. Almost the same sector source during the LRT-4 event in the fall marked the lowest NO₃ ⁻ concentration of 0.9 ± 0.4 µg/m³, but not of SO₄ ²⁻. Sector II air mass can be seen to contribute significantly to the concentration of secondary species in PM_2.5, the highest being 22.4% for SO₄ ²⁻ during the LRT-2 event in spring and 19.8% for SO₄ ²⁻ during the LRT-7 event in fall. Figure 2b also suggests the arrival of a marine and dust front between the periods LRT-5 and -6. The increase in sea-salt components was first observed in LRT-5, with contributions reaching as high as 7% for Na⁺ and 8% for Cl⁻. The event was succeeded by elevated levels of dust components during LRT-6, having a Ca²⁺ concentration of 1.1 ± 0.8 µg/m³ and Mg²⁺ concentration of 0.3 ± 0.2 µg/m³. Mineral dust sampled in its preprocessing form was commonly detected during the dust transport event from October 16 12:00 a.m. to October 22 12:00 a.m. Dust in combination with secondary aerosols coming from the KP was also sampled between October 24 09:00 a.m. and October 26 12:00 p.m. In general, spring LRT events contained aerosols that were already processed during transport, and thus represented more aged aerosols. LRT events during fall measurements contain mineral dust and sea-salt particles in their preprocessing form. The chemical composition can be influenced by the enrichment of acidic secondary aerosols on mineral dust particles, and even of sea-salt components. These observations and their processes will be discussed in the succeeding sections.

Chemical Characteristics of Sea-Salt Components in Long-Range-Transported Particles

On a single-particle basis, Sullivan et al. established the least reported occurrence of the presence of chloride on mineral dust.17 This could happen when the NaCl in the marine boundary layer has been converted into HCl(g) via the heterogeneous reaction with SO₄ ²⁻, CH₃SO₃ ⁻, NO₃ ⁻, and their precursors,24 prior to the dust front. The released HCl(g) would then enrich the mineral dust by exchanging with more volatile CO₃ ²⁻ (ibid.). Sea-salt particles are important tracers for understanding the heterogeneous reaction with reactive gases during transport. If the sea-salt particles were freshly injected into the measured aerosols, it would be expected that the measured particles would follow the seawater ratio (SWR) proposed by Wilson25 based on the equivalent concentrations of Na and Cl ([Na⁺]:[Cl⁻] = 1:1.18). However, it can be seen from Figure 4 that there were distinct characteristics of transported sea-salt components for each seasonal measurement period, aside from the noticeable deviation from the SWR line. In spring, there is almost no correlation (r ² = 0.06) between the major sea-salt components [Na⁺] and [Cl⁻]. On the other hand, sea-salt aerosol components during the fall intensive period are moderately correlated (r ² = 0.74). The deviations from the SWR line indicate that the sea-salt aerosols have already been processed and have undergone depletion. Furthermore, the correlations show that the spring sea-salt aerosol components have undergone more extensive processing compared with the fall intensive sea-salt aerosol components. Thus, a formula for the calculation of Cl⁻ depletion (% Cl⁻ depleted) in the marine background site is proposed as follows:

Figure 4. Scatter plot of the major sea-salt components Na⁺ and Cl⁻, expressed in µEq/m³. Dotted line corresponds to the sea water ratio. (a) Spring and (b) fall.

(1)

based on the assumption that freshly released sea-salt aerosols have an SWR of 1.18. More processed sea-salt aerosol component in spring was depleted (% Cl depleted = 85% ± 12%) compared to the fall sea-salt aerosol component (% Cl depleted = 50% ± 9%).

Specifically, we observed remarkable concentrations of Na⁺ (LRT-5: 1.2 ± 0.4 µg/m³, LRT-6: 1.0 ± 0.2 µg/m³) and Cl⁻ (LRT-5: 1.3 ± 1.0 µg/m³, LRT-6: 0.9 ± 0.2 µg/m³) during the arrival of marine and dust fronts in the fall, as seen in Table 2. This could have occurred after Cl⁻ liberation²⁴ prior to the transport of mineral dust, followed by Cl⁻ enrichment in the mineral dust. Thus, HCl(g) was most likely elevated in the boundary layer before the arrival of the dust at Deokjeok Island. The HCl(g) produced by an acidic gas displacement reaction with NaCl was then mixed with incoming dusts, and the liberated Cl⁻ could then enrich the mineral dust making the particles susceptible to a CO₃ ²⁻ and Cl⁻ heterogeneous reaction. On the other hand, prior to reaching the measurement site, the LRT-7 air mass passed over dust-laden Sector II of Northeast Asia, followed by the reactive-species-rich (NO₃ ⁻ and SO₄ ²⁻) Korean peninsula (see Supplementary Figure 7; Figure 2b). Note that this event resulted in NO₃ ⁻ and SO₄ ²⁻ enrichment, which could have displaced existing Cl⁻ in the transported mineral dust, thus producing NO₃ ⁻- and SO₄ ²⁻-enriched aerosols in the LRT-7 case. The sea-salt component has been actively modified in the marine boundary layer, prior to its association with dust components and later associations with secondary acids.

Chemical Characteristics of Long-Range-Transported Mineral Dust and Secondary Aerosols

The atmospheric chemistry, processing, uptake and formation of NO₃ ⁻ and SO₄ ²⁻ on mineral dust particles during long-range transport has been studied at field sites in Northeast Asia,2,15,16,26,27 as well as in laboratory studies¹⁵ and with aerosol interface modeling.¹ Previously, it was shown that the sea-salt particles sampled with the transported mineral dust had already undergone depletion and processing. The mineral dust components associated with these sea salts had undergone aging to a similar extent. It is known that the reactive constituents of dust particles in East Asia are CaCO₃ (calcite), MgCO₃ (magnetite), and CaMg(CO₃)₂ (dolomite), among which CaCO₃ is dominant and accounts for ∼90% of the total dust component.28 Other major components of mineral dust such as Al₂O₃, SiO₂, and Fe₂O₃ can be considered unreactive,29 whereas the volatile CO₃ ²⁻ is expected to be reactive. Our method is unable to measure CO₃ ²⁻ directly due to its interference in the measured anionic conductivity. Assuming the particle is neutrally charged, it was assumed that CO₃ ²⁻ is the only anionic species not measured by the PILS-IC instrument, and thus the difference between cations and anions measured is designated as [CO₃ ²⁻]_estimated ⁴:

(2)

where n and m are the total number of cations and anions measured by the IC, respectively.

Considering the major cationic and anionic species, and the assumption that the only unmeasured anionic species is CO₃ ²⁻, we maintain that the association of Ca²⁺ and Mg²⁺ with carbonate represents preprocessed dust. The fraction of carbonate replacement (FCR; eq 3) method proposed by Song et al.29 was used to investigate the degree of aging of mineral dust aerosols.

(3)

where [CO₃ ²⁻]_replaced is the amount of CO₃ ²⁻ that remained in the aerosol after deducting the estimated CO₃ ²⁻ (in equivalent concentrations), that is, [CO₃ ²⁻]_original − [CO₃ ²⁻]_estimated. [CO₃ ²⁻ _original], on the other hand, is estimated from the sum of Ca²⁺ and Mg²⁺, assuming that all CO₃ ²⁻ is the product of the dissociation of CaCO₃ (calcite), MgCO₃ (magnetite), and CaMg(CO₃)₂ (dolomite) along its transport into the atmosphere.

Results from FCR estimations are presented in Table 3, as well as comparable results from previous studies in the Northeast Asia region. FCR data were collected from aircraft measurements of aerosols in the marine boundary layer, whereas this study dealt with ground-based measurements. As such, the differences in the levels of reactive acidic gases and the interaction time with gas-phase precursors could have caused the differences between the independent FCR measurements.²⁹

Table 3. Degree of dust processing in Northeast Asia using the FCR method

Location	Periods	Results	Method^a	References
West Sea	11 April 2001	Average FCR 0.13	ACE Asia Flight 6	Song et al.^4
	12 April 2001	Average FCR 0.32	ACE Asia Flight 7
Kyushu Island	13 April 2001	Average FCR 0.61	ACE Asia Flight 8^b
Deokjeok Island (Spring)	31 March∼8 April 2009	Ave. of all qualified FCR 0.53	Background marine	This study
Deokjeok Island (Fall)	13∼26 October 2009	Ave. of all qualified FCR 0.55	Background marine	This study
Notes: ^aMeasurements were made using PILS-IC instrument.^bThe ACE-Asia Flight 8 passes over the Marine Boundary Layer (MBL) over the South of Kyushu Island.

Despite the direct observation that Ca²⁺ and Mg²⁺ do not always show a good correlation with CO₃ ²⁻, which can be interpreted as processed or aged mineral dust, correlation analysis (Figure 5) between the same charged ions still suggests that (SO₄ ²⁻ + NO₃ ⁻) and (NH₄ ⁺ + K⁺) show a slope, s, of >1 (1.1 < s < 1.3), which means that the acidic components are in excess. Moreover, a strong correlation coefficient of r ² = 0.95 suggests that these components have been emitted from the same source or have resulted from the same aerosol processes. The range in concentration of SO₄ ²⁻ (spring: 1.4∼48.6 µg/m³; fall: 0.25∼15.9 µg/m³) and NO₃ ⁻ (spring: 0.1∼14.3 µg/m³; fall: 0.3∼23.3 µg/m³) during the spring intensive campaign is higher than that of the fall campaign. These major components are indicators of dust aerosol processing because they are secondary in origin.¹² Results of atmospheric chemical processing show dust processing on pre-emitted mineral dust transported to Deokjeok Island. Secondary acids that have accumulated in mineral dust can be a sink for NH₃, through heterogeneous nucleation30 or direct uptake from gas-phase coagulation with nitrate or sulfate rich ammonium particles.31 Looking at the magnitude of the components, in which the spring transported aerosol event is higher than the fall event, it can be said that the spring event produced more secondary aerosols during the processing. High RH during the spring intensive period may have contributed more to the fine-particle formation induced by a damp atmosphere and abundant acidic sources, as explained previously.

Figure 5. Equivalent concentrations of major anion components (SO₄ ²⁻ and NO₃ ⁻) and major cation components (NH₄ ⁺ + K⁺) for the intensive measurements during the spring and fall seasons. The magnitude of the spring intensive measurement data is higher than that of the fall intensive measurement period. Their slopes lie between 1.1 < s < 1.3, and for both periods r ² = 0.95.

In this study, the presence of Ca²⁺ and Mg²⁺ in varying amounts is thought to characterize the long-range-transported mineral dust. When the mineral dust is preprocessed, we would expect it to be in a carbonate form, and with a strong association with depleted sea-salt particles. However, when mineral dusts are chemically modified during transport, the volatile CO₃ ²⁻ is replaced by less volatile and non-volatile NO₃ ⁻ and SO₄ ²⁻, respectively.²⁹ Meanwhile, after subtracting the contribution of the equivalent [NH₄ ⁺] from the combined equivalent [SO₄ ²⁻] and [NO₃ ⁻], the difference is believed to represent excess acids, which would process the transported dust. Thus, strong associations of Ca²⁺ and Mg²⁺ with excess acids are regarded as indicators of processed mineral dust.

To complement the FCR of Song et al.,29 we propose a simple indicator of mineral dust aging and suggest a range for the degree of associations. The triplot in Figure 6 shows the associations in normalized equivalent concentrations of excess acids with mineral dust components and [CO₃ ²⁻]_estimated. In the triplot, a sample containing primarily Ca²⁺ and Mg²⁺ would appear at the top vortex, primarily [CO₃ ²⁻]_estimated at the right vortex, and primarily excess acids at the left vortex. Using this tool, we can see the extent to which the transported mineral dust has been processed. From both measurements, it can be seen that Ca²⁺ + Mg²⁺ are present in a smaller percentage compared with the excess acids. However, the range of the fraction of mineral dust components and excess acids measured during the spring intensive period are less varied, mostly between about 0.25 and 0.4 for Ca²⁺ + Mg²⁺, and about 0.6 and 0.75 for the excess acids. In contrast, data for the fall intensive measurement period show that the range of the fraction of mineral dust components and excess acids is wider, between about 0.1 and 0.5 for Ca²⁺ + Mg²⁺ and about 0.5 and 0.8 for the excess acids. In addition, there are more fractions of mineral dust components that are associated with excess acids.

Figure 6. Ternary plots of the associated chemical components. Plotted data are normalized such that equivalent concentrations of excess acids, Ca²⁺ + Mg²⁺ and CO₃ ²⁻ are summed to unity. Most of the mineral dust has already undergone complete aging, although not all because levels of [CO₃ ²⁻] were estimated in 10% of the samples in spring and in 80% of the aerosol measured in the fall. Symbols correspond to the episodic events described in each intensive measurement period. (a) Spring and (b) fall intensive periods.

The spring intensive data (Figure 6a) show the equivalent ratio of Ca²⁺ + Mg²⁺ in the range of 25∼50%. When the proportion of CO₃ ²⁻ is 40∼70%, the mineral dust component is present in the preprocessed form, as in the case of LRT-1. Mineral dust during local events does not show evidence of associations with secondary acids. However, mineral dust components during the LRT-2 period are associated at a constant amount with CO₃ ²⁻ and excess acids, which are in inverse proportions. This indicates the exchange in the anionic component of the transported mineral dust processed by reactive acids such as NO₃ ⁻ and SO₄ ²⁻. The fall intensive data show unique results. Around 65∼90% of excess acids are in strong association with 10∼25% of mineral dust components in the LRT-4 event. The poor association of mineral dust with the secondary acid component in the case of LRT-7 only confirmed that most of the aerosols sampled are in the processed form, and mostly comprise secondary acids.

One remarkable feature of the fall intensive data is that the degree of acidity of associated secondary acids in mineral dust can be seen by just looking at the percentage of associations with excess acids (Figure 6b). For example, by looking at the left-most panel of Figure 6b (excess acids), the LRT events are shown to be grouped in increasing order of acidity: LRT-4 aerosols have an association of 75∼87%, LRT-6 aerosols possess an association of 62∼75%, and LRT-7 aerosols display an association of 50∼60%. Most of the sampled aerosols during LRT-5 were associated with CO₃ ²⁻ (ranging from about 15% to 70%), owing to the arrival of a marine and dust front during this period. Thus, the order of acidity of the LRT events is represented by the relationship LRT-4 > LRT-6 > LRT-7.

It is therefore apparent that transported mineral dust particles can process and accumulate reactive acids. A laboratory-scale experiment has shown that nitrate products after calcite mineral processing are still intact even after significant water uptake,32 whereas reaction with SO₂ is relatively slow.33 Competitive reactions between SO₂ and HNO₃ show no apparent effect on the reactivity of both gases (ibid.). On the basis of single-particle measurements, Zhang et al.34 observed that the degree of SO₄ ²⁻ formation may have been dependent on the oxidation of SO₂, but would later be neutralized by surface-adsorbed or gas-phase NH₃ present in the troposphere at the time of the measurements. Observations of Fairlie et al.35 have shown that particulate nitrate is primarily associated with dust, sulfate is primarily associated with ammonium, and mineral dust remained alkaline across their observational grid over the Pacific. Their results were even reproduced using models.

There are several studies in the Northeast Asian region that also utilized the ratio method for indicators of dust episodes, some of which are tabulated in Table 4. Most of the values were obtained by analyzing filter-based samples for chemical composition, followed by the ratio method of mineral dust components using Ca as the common chemical indicator.36–40 However, most of the studies were conducted at continental sites (Beijing, Mongolia, China), such that the partner species in their ratio method was either Al or Si. Because our study site was conducted at a background marine site, we employed the ratio method using Ca and Mg in partnership with the sea-salt component Na. As shown in Figure 7, the degree of associations of mineral dust with reactive acids was explored. The extent of processing was based on the magnitude of reactive acids, utilizing the scatter plot of NO₃ ⁻ and SO₄ ²⁻ in relation to Ca²⁺. The sea-salt and dust components were reconciled using the ratio method between mineral dust and sea-salt indicators, namely, [Ca²⁺ + Mg²⁺] and [Na⁺], respectively. To reconcile the associations of reactive acids with transported mineral dust, the ratio of NO₃ ⁻ and SO₄ ²⁻ individually to Ca²⁺ was used. The color bar indicates the degree of association of mineral dust to sea-salt components. The range of the [Ca²⁺ + Mg²⁺]/[Na⁺] ratio itself reflects the extent of the difference in aerosol events. A higher [Ca²⁺ + Mg²⁺]/[Na⁺] (ratio >2) indicates that major dust events dominate during that period, which decreases the possibility of dust processing. A lower [Ca²⁺ + Mg²⁺]/[Na⁺] (ratio <2) ratio suggests that sea-salt aging has already occurred on the dust particles. The equivalent ratio of SO₄ ²⁻/Ca²⁺ and NO₃ ⁻/Ca²⁺ shows the degree of association of the acidic component on dust particles. Observed higher magnitudes for the SO₄ ²⁻/Ca²⁺ or NO₃ ⁻/Ca²⁺ ratio (>10) suggest that the ground aerosols have already been processed by the reactive gaseous acids, whereas a lower ratio (<10) suggests that the dust has been processed to a lesser extent with estimated amounts of CO₃ ²⁻.

Table 4. Ratio method for mineral dust and soil indicators at different sites in Northeast Asia

Ratio Method	Indicative	Site	Period	Method	Reference
Ca/Si<0.2, Al/Ca>1.7	Dust storm	Urban Beijing	Jan, Apr, Jul, Oct 2000	24 hrs filter based, XRF	Song et al^38
Ca/Si 0.37 Al/Ca 1.00	Mineral dust	Chinese Loess Area	Not applicable	Single particle-EDX	Krueger et al^39
Al/Ca 1.6∼1.7	Surface Soils	Mongolia and Inner Mongolia	Not available	Emission inventory	Xuan^37
Al/Ca 2.05	Mineral Dust	Hunshan Dake Sandland	Spring, 2001	PM2.5	Cheng et al^40
Al/Ca^2+ ∼2.17	Dust	Urban	2001–2004	PM2.5	Wang et al^36
Al/Ca^2+ ∼0.74	Haze	Beijing
Al/Ca^2+ ∼1.04	Clear
[Ca^+2 +Mg^+2]/[Na^+] ratio > 2.0 * [Ca^+2 +Mg^+2]/[Na^+] (ratio < 2) **	Transported mineral dust seasalt ageing on the processed dust particles	Deokjeok island	Spring and Fall, 2009	PILS-IC	This study
*[Ca^+2 +Mg^+2]/[Na^+] ratio ranged from2.0 to 23.4 **[Ca^+2 +Mg^+2]/[Na^+] ratio ranged from 0.02 to 2.0

Figure 7. Ratio method of major reactive acids to Ca²⁺ to describe the extent of LRT aerosol processing. For both intensive periods, aerosol particles with aged sea salt (i.e., [Ca²⁺+Mg²⁺]/[Na⁺] ratio <2) are shown to have higher magnitudes of acidic components. (a) The Ca²⁺-normalized equivalent ratio of NO₃ ⁻ and SO₄ ²⁻ during the spring intensive measurement period occurs mostly between the 1:2∼1:6 line, and (b) between the 1:1∼1:6 line for the fall intensive measurement data. (Inset) Larger x-y range to show the acid/[Ca²⁺] ratio >10.

SUMMARY AND CONCLUSION

The chemical characteristics of long-range-transported aerosols downwind of the Asian continent were investigated to identify the levels of mineral dust transport and processing using ground-based measurements. Two mineral dust transport events and details of the degrees of associations of their components have been identified. Evidence was found for the preprocessing of mineral dust by Cl⁻ and reprocessing by reactive acids (NO₃ ⁻ and SO₄ ²⁻). Sea-salt components were aged to the same extent as the mineral dust. The prior reaction of HCl(g) with mineral dust caused the displacements of various amounts of CO₃ ²⁻, and activated the mineral dust for further reaction with nitrates and sulfates. The spring observations indicate that CO₃ ²⁻ was almost fully depleted by excess NO₃ ⁻ and SO₄ ²⁻, whereas the fall observations show that fine-particle mode dust particles were not completely depleted by these excess acids. Additionally, the estimated [CO₃ ²⁻] was strongly associated with [Ca²⁺] and [Mg²⁺] in the fall. The analysis of measurement data suggests that the fraction of carbonate replacement (FCR), which is a measure of the degree of chemical interaction between mineral dust and reactive gases, was comparable with that obtained from independent ground-based measurements. In addition, our triplot analysis and ratio approach supported the degree of association between CO₃ ²⁻, excess acids, and mineral dust components. This study is important for an understanding of the chemical evolution of dust particles in Northeast Asia, and clarifies the chemical processes of aerosols from anthropogenic and natural sources. It was possible to estimate CO₃ ²⁻ from the difference of the summed anions and excess cations, and the degree of replacement as shown by the FCR method, utilizing time-resolved ion data. Our observations on the processing of long-range-transported mineral dust and associated sea-salt components suggest important evidence on the temporal evolution of naturally occurring atmospheric particles and the subsequent anthropogenic reactive components. Generally, source regions influence the loading and chemical characteristics of LRT aerosols.

ACKNOWLEDGMENTS

This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (MEST) (Project No. 2008-0060618) and by the 2009 LTP Project supported by the National Institute of Environmental Research (NIER). It was also funded by the Korea Meteorological Administration Research and Development Program under grant RACS_2010-1002 through the Advanced Environmental Monitoring Research Center at Gwangju Institute of Science and Technology. The authors also gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY Web site (http://www.arl.noaa.gov/ready.php) used in this publication. Lastly, the authors would like to acknowledge the assistance of local residents from Deokjeok Island.