Chemical composition of atmospheric aerosols from Zhenbeitai, China, and Gosan, South Korea, during ACE-Asia



[1] Studies were conducted as part of Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) to characterize the major ion and elemental composition of aerosol particle samples collected at Gosan, an ACE-Asia supersite (GOS, Korea, total suspended particle or TSP samples) and at Zhenbeitai (ZBT, China, TSP and particles < 2.5 μm diameter or PM2.5 samples), a site closer to the sources for Asia dust. The concentrations of 24 elements in the ZBT PM2.5 samples were correlated with Al (an indicator of mineral dust), and the ratios of these elements to Al were similar to those in a loess certified reference material, but a second group of elements was enriched over crustal proportions most likely as a result of pollution emissions. The concentrations of various water-soluble (WS) cations (Na+, K+, Ca2+, Mg2+) also were generally well correlated with Al in both the ZBT and GOS samples, with the exception being WS K+ at ZBT, where biomass burning may have had an effect. The percentage of calcium that was soluble approached 100% at ZBT versus ∼60% at GOS, and the ratio WS Ca2+/Al also was higher at ZBT. The molar ratio of sulfate to WS Ca2+ was ∼0.1 at ZBT but increased to near unity at GOS, where the aerosol nitrate/WS Ca2+ ratio was tenfold to hundredfold higher compared with ZBT, presumably because of anthropogenic influences. The observed differences in aerosol characteristics between sites can only be explained as the end product of different source contributions combined with complex processes involving gas-particle conversion, size-dependent fractionation, and aerosol mixing.

1. Introduction

[2] The effects of tropospheric aerosol particles (henceforth simply aerosols) on climate [e.g., Charlson et al., 1987, 1992; Hobbs, 1993] are generally understood in qualitative terms but still not quantitatively [National Research Council Panel on Aerosol Radiative Forcing and Climate Change, 1996]; this is mainly due to limited information on the aerosols' chemical composition, optical and physical properties (including shape and size), and geographical distributions. The Aerosol Characterization Experiments (ACE-1, ACE-2 and ACE-Asia) have combined in situ measurements, remotely sensed data, and models to investigate interactions between aerosols and the Earth's climate system. The overall goals of the ACE studies have been to advance the understanding of aerosols' effects on climate and in so doing improve the ability of modelers to predict the effects of changing atmospheric aerosol loads [Bates et al., 1998; Raes et al., 2000]. The studies presented here were conducted as part of ACE-Asia, the Asian Pacific Regional Aerosol Characterization Experiment. ACE-Asia was conducted in conjunction with the East Asian/North Pacific Regional Experiment (APARE), an activity of the International Global Atmospheric Chemistry Project [Uematsu and Arimoto, 2000].

[3] The attention of atmospheric chemists, and aerosol scientists in particular, increasingly has turned to the Asia/Pacific region, largely because of the strength and diversity of Asian sources for both natural and pollution-derived aerosols. In late winter and early spring, mineral dust from the Asian deserts is a ubiquitous, oftentimes dominant, component of the atmospheric aerosol over large parts of Asia [Parungo et al., 1995]. Every spring, large quantities of Asian dust [Duce et al., 1980; Shaw, 1980; Prospero et al., 1989, 2002] and pollutant aerosols [Prospero et al., 2003] are transported from Asia to the North Pacific and beyond.

[4] The atmosphere over much of Asia already has been heavily contaminated with anthropogenic trace gases and aerosols, and further large-scale perturbations are all but certain to occur. High concentrations of potentially toxic trace elements, such as Pb and As, have been found in aerosol samples collected from large cities such as Beijing, Chengdu, and, Lanzhou [Hashimoto et al., 1994]. Aerosol sulfate concentrations at a network of coastal/continental sites in eastern Asia were found to be 10 or more times those at remote sites in the North Pacific [Arimoto et al., 1996]. Several missions involving aircraft, most notably PEM-West A and B [Hoell et al., 1997] and especially TRACE-P (see special section in Journal of Geophysical Research, 108(D20), 2003) have begun providing information on the vertical distributions of aerosols over Asia and the western North Pacific. In addition to the influences of aerosols on climate, the distributions and transport of Asian aerosols have important implications for human health [Harrison and Yin, 2000], oceanic biogeochemical cycles [Young et al., 1991] and chemical fluxes between continents [Jaffe et al., 1999].

[5] ACE-Asia was an international program that involved sampling from aircraft, onboard ships and at a network of ground stations [Huebert et al., 2003]. Here we present the results of aerosol studies conducted at two ACE-Asia ground stations: Zhenbeitai, in China, and Gosan, in South Korea. The studies were designed to characterize the elemental and major ion composition of the aerosols. Particular attention was paid to (1) the composition of dust and the relationships between soluble cations and Al, a commonly used indicator of mineral dust; (2) non-sea-salt sulfate (nss SO42−) and nitrate, which both have significant natural and pollution sources; (3) several pollution-derived trace elements, and (4) methanesulfonate (MSA, a biogenic sulfur compound produced via the oxidation of dimethylsulfide, which is produced by marine photoplankton). Elemental data for a Chinese loess certified reference material (CRM) [Nishikawa et al., 2000] were used for comparisons with the composition of mineral aerosol.

2. Methods

2.1. Aerosol Sampling

[6] One set of aerosol samples was collected at the ACE-Asia supersite at Gosan (34.28°N, 126.17°E, abbreviated as GOS, formerly known as Kosan), which is on the western end of Jeju Island (previously Cheju, Figure 1). Daily samples were collected during an intensive observation period (IOP, 25 March 2001 to 1 May 2001) and over intervals of 1–7 days from 1 May 2001 to 30 September 2001. The high-volume bulk aerosol sampler at GOS was previously used at the same site for the NASA Pacific Exploratory Mission in the Western Pacific Ocean Experiments (PEM-West) [see Hoell et al., 1996, 1997; Arimoto et al., 1996, 1997].

Figure 1.

Locations of ACE-Asia sampling stations at Gosan (GOS) and Zhenbeitai (ZBT). Map courtesy of M. Uematsu, University of Tokyo.

[7] The GOS sampler operated at ∼60 m3 h−1, and single 20 × 25 cm unwashed Whatman 41® filters (Whatman International Ltd., Maidstone, England) were used as the collection substrates. To minimize contamination from local sources, the sampler at Gosan was controlled with respect to wind speed (< 1 m s−1) and wind direction (sampling between 180° and 353° magnetic); however, local swirling winds at the site sometimes caused the sampler to operate when the airflow was not from the clean oceanic sector.

[8] A second set of samples was collected from 8 to 30 April 2001 at Zhenbeitai, China (ZBT, 38.3°N, 109.7°E, Figure 1). This site is downwind of the Mu Us Desert, which is a significant regional source for desert dust. Twenty-five pairs of PM2.5 aerosol samples (particles ≤ 2.5 μm aerodynamic equivalent diameter) were collected at ZBT using an IMPROVE-type sampler located on the first level (∼10 m) of a 20-m-tall air-sampling tower. The airflow for this sampler was split between two filters. A 25-mm-diameter Telfo® filter (3.0 μm pore size, Pall Gelman Laboratory, Ann Arbor, Michigan) was used to determine the concentrations of major ions and water-soluble trace elements and a Metricel® filter (mixed cellulose ester, 25-mm diameter, 0.8-μm pore size, Pall Gelman) was used for trace element determinations following acid digestion. The flow rate for this sampler was ∼11.5 L min−1. Filters were changed daily except for 10 April 2001, when they were changed twice because of high filter loadings caused by a severe dust storm. Blanks were collected by placing filters in the samplers, but not pumping any air through them.

[9] Seventy-three, daytime (12-hour), total suspended particle (TSP) aerosol samples were collected from the second level of the ZBT tower (20-m height) under northerly surface wind flow from 19 February to 21 May 2001. These samples were collected using a Partisol®-Plus Model 2025 Sequential Air Sampler, Rupprecht & Patashnick Co., Inc., Albany, New York) using a flow rate of ∼16.7 L min−1. A control system (Global Water Instrumentation Inc., Gold River, California) was used to ensure that only air masses arriving from the west (through north) to the east (270°–90°) were sampled. The low-wind-speed cutoff for the control system was set at 7.2 km h−1 until 19 April 2001 and 3.6 km h−1 thereafter. The change in the wind speed cutoff was made because relatively few samples were collected in the first part of experiment, and the intention was to increase the number of samples collected during the low-dust period later that spring. Teflo® membrane filters (Pall Gelman Laboratory) were used as collection substrates. An additional set of 21 TSP samples was collected at the ZBT tower under southerly surface winds for comparison with the northerly flow. The southerly wind samples were more likely to be influenced by anthropogenic materials, especially from Yulin, a small city ∼10 km south of the sampling site.

2.2. Preparation of Samples

2.2.1. High-Volume Filters From GOS

[10] The Whatman 41® sample filters from the high-volume sampler at GOS were shipped to the Carlsbad Environmental Monitoring and Research Center, New Mexico State University (CEMRC) where they were cut into quarters using plastic scissors in a clean bench with HEPA-filtered air. One quarter of each filter was used for the ion chromatographic and soluble elements analysis; these aliquots were placed in plastic bags and extracted in an ultrasonic water bath in two steps using 20 mL of deionized (specific resistance = 18 MΩ) water in each step (10 min each) in sealed plastic bags (Precision Clean II®, Fisher Container Corp., Buffalo Grove, Illinois).

[11] Elemental analyses were performed on separate quarter-filter aliquots of the GOS filters. For this, the filter quarters were placed in Teflon® microwave digestion vessels, and to each vessel was added 5 mL ultrapure H2O, 7 mL HNO3, 3 mL HCl, 5 mL HF, and 2 mL H2O2. The vessels were sealed and heated to 180°C in a microwave digestion unit (Model MDS 2100, CEM Corporation, Matthews, North Carolina) for 30 min.

2.2.2. Low-Volume PM2.5 Filters From ZBT

[12] Separate filters were collected at ZBT for analyses of (1) major ions/water soluble trace elements and (2) total recoverable trace elements. For the aqueous extractions, the Teflo® filters were extracted at CEMRC with ultrapure water in sealed polyethylene bags using isopropanol as a wetting agent. These extractions were done in an ultrasonic water bath at room temperature in three steps. The Metricel® filters from the PM2.5 sampler were prepared for total recoverable elemental analyses using a procedure similar to that described above for the GOS TSP filters.

2.2.3. TSP Filters From ZBT

[13] The University of Hawaii (UH) group received one quarter of each Teflo® TSP sample filter. Each of the filter portions was placed into a microclean polyethylene bag, to which 0.2 mL ethyl alcohol was added as a wetting agent. Then 9.8 mL of 10−5 M trifluoroacetic acid were added (the acidic pH prevents the loss of NH4+), and the bag was placed in an ultrasonic bath for 15 minutes. Portions of these filters also were analyzed for elemental composition without pretreatment by proton-induced X-ray emission at Beijing Normal University (see description below).

2.2.4. Loess Certified Reference Material

[14] The loess CRM (CJ-1) [Nishikawa et al., 2000] was digested and analyzed for comparisons with the aerosol elemental data. Analyses of some initial loess digests indicated that the recoveries of several elements were less than optimal when a modification of U.S. Environmental Protection Agency (U.S. EPA) [1996] Method 3052 was used to prepare the samples. Therefore a second aliquot of the loess CRM was prepared using a different procedure in which boric acid was added to the mixture of acids to improve the digestion of the mineral matrices. For these digestions, 0.1 g of the loess CRM was placed in a microwave vessel, and to this 5 mL HNO3 and 1 mL HF were added. The samples were heated in microwave (MDS 2100, Milestone s.r.l., Monroe, Connecticut) for 15 min and then rapidly cooled in a freezer. When cool, 5 mL of a 6% boric acid solution were added; the samples were resealed and reheated in the microwave for 5 min. Finally, the samples were cooled and diluted in preparation for instrumental analysis. The recoveries of the certified elements for the CJ-1 loess averaged 102%, with an average relative percent difference (RPD) between replicates of 8.0% (minimum RPD of 0.4% for K and a maximum RPD of 16.8% for Fe).

2.3. Instrumental Analyses

2.3.1. Ion Chromatography

[15] The concentrations of the ions of interest in the aqueous extracts of the GOS high-volume and ZBT low-volume aerosol samples were determined at CEMRC using a Dionex 500 ion chromatography system (Dionex Corp, Sunnyvale, California). Nitrate, sulfate, and other anions in the aqueous extracts of both high- and low-volume filters were separated using an AS-14 column, and the MSA concentrations were determined in separate aliquots using an AS-11 HC microbore column. Non-sea-salt sulfate was calculated using Na as an indicator of sea salt and assuming the seawater SO42− /Na+ mass ratio to be 0.2515 [Millero, 1996]. The major cations (Na+, Mg2+ and K+) were separated using a CS-12A column.

[16] For the aqueous extracts of the TSP samples from ZBT, Cl, NO3, SO42− and oxalate were analyzed by ion chromatography at UH (Dionex 300 series) with an OmniPac 500-column, 30-mM NaOH/5% CH3OH as eluent and an AMMSIII MicroMembrane Suppressor. Na+, NH4+, K+, Mg2+ and Ca2+ were analyzed with an IonPac CS12 column, 20-mM MSA eluent and a CSRS-ULTRA Self-Regenerating Suppressor.

2.3.2. Inductively Coupled Plasma Mass Spectrometry

[17] The elemental analyses of the ZBT PM2.5 and GOS TSP samples were done by inductively coupled plasma mass spectroscopy (ICP-MS) using a Perkin-Elmer Elan 6000 (Perkin-Elmer Corp., Norwalk, Connecticut) following U.S. EPA [1994] Method 200.8. The ICP-MS analyses can provide data for up to ∼35 elements in aerosol samples, and the lower limits of quantitation are in the range of 1–100 parts per billion (1 part in 109) of the sample digest. Typically, the instrument precision for the ICP-MS is ±1–5%, method precision is ±10–20%, and accuracy is ±10–20%. Background corrections were made for all analytes whose average concentrations in blanks were above their method detection limits.

2.3.3. Proton-Induced X-Ray Emission Spectrometry

[18] The elemental analyses of the ZBT TSP filters were done by proton-induced X-ray emission (PIXE). The tandem accelerator at Beijing Normal University was used for these analyses (for details of the methods, see Zhang et al. [1993]). The PIXE technique provides information on up to nineteen elements [Zhang et al., 2003a], but here we consider only the data for Al and Ca in detail. All concentrations were corrected for backgrounds from blank filters.

3. Results and Discussion

3.1. Loess Certified Reference Material

[19] Loess is a “terrestrial windblown silt deposit consisting chiefly of quartz, feldspar, mica, clay minerals and carbonate grains in varying proportions” [Pye, 1987, p. 199]. As such, one would expect that in the absence of extensive weathering or reworking, loess would share many of the physical and chemical characteristics of atmospheric mineral dust even though variations in loess composition as a function of particle size are a potentially confounding influence. The elemental composition of atmospheric aerosol particles has commonly been evaluated through comparisons with average crustal material, using various compilations such as Taylor and McLennan [1995] as compositional references. However, one might expect Chinese loess to be more representative of the composition of Asian dust than average crustal material, and for that reason we used the data for the CJ-1 CRM for comparisons with the aerosol data.

[20] In general, the normalized concentrations of most elements determined (X/Al mass ratios where X is any element and Al is used as crustal indicator) in the CJ-1 loess were similar to those in the Earth's crust as presented by Taylor and McLennan [1995]. However, the concentrations of several elements Ag, As, Cd, Co, Sb, and Sc, which were measured in our study but were not certified in the loess, were enriched over average crustal values by factors of 9, 20, 8, 4, 11, and 4, respectively. Of these elements, Sc is typically crustal in origin, but Ag, As, Cd, and Sb and, to a lesser extent, Co often are enriched in aerosols because of pollution emissions [see Rahn, 1976]. These results thus suggest that the loess CRM may have contained some trace element contaminants, but one cannot entirely discount the possibility that the enrichments reflect natural variations in mineral composition. Such enrichments, if real, would have obvious implications for evaluating natural versus anthropogenic sources in aerosol samples collected downwind.

3.2. Zhenbeitai

3.2.1. Mineral Aerosol and Trace Elements at Zhenbeitai

[21] A major objective for our studies was to characterize the chemical composition of Asian mineral dust, and for this we calculated arithmetic mean trace element concentrations and the ratios for the trace elements to Al, comparing the latter with the ratios for the loess CRM. In addition to the thirteen elements whose concentrations were certified in the loess, the ICP-MS analysis provided data for 25 elements not certified.

[22] The ratios of 24 elements (Ba, Ca, Ce, Co Dy, Er, Eu, Fe, Gd, K, La, Li, Mg, Mn, Na, Nd, Pr, Sc, Si, Sm, Sr, Th, Ti, and U) to Al in the ZBT PM2.5 samples were similar to those in the loess CRM thus demonstrating a strong chemical connection between the dust and the eolian sedimentary material. The correlation coefficients (r) for the crustal elements versus Al ranged from 0.73 to 1.0 and exceeded 0.95 in most cases (Tables 1a and 1b). In the discussion of the crustal elements that follows, special attention is given to Al and Ca, the second and third most abundant elements certified in the CJ-1 loess CRM (Si is the most abundant element certified: O was not determined).

Table 1a. Certified Trace Elements in PM2.5 Aerosol Samples From Zhenbeitai and a Loess Certified Reference Material: Concentrations and Relationships to Aluminuma
ElementnConcentration, ng m−3Correlation With Al (r)Ratio to Al, unitlessZBT/CJ-1
MeanStandard DeviationZBT AerosolsCJ-1 CRMb
  • a

    Here, n, number of samples; ZBT, Zhenbaitai; NA, not applicable.

  • b

    CRM, loess certified reference material [Nishikawa et al., 2000].

Al224.3 × 1036.1 × 103NA11NA
Ba214.3 × 1015.5 × 1010.899.0 × 10−38.4 × 10−31.1
Ca185.2 × 1036.6 × 1030.981.0 × 1009.7 × 10−11.1
Fe212.5 × 1033.7 × 1030.995.9 × 10−14.9 × 10−11.2
K221.5 × 1031.9 × 1030.973.2 × 10−13.2 × 10−11.0
Mg221.5 × 1032.1 × 1030.953.5 × 10−12.6 × 10−11.4
Mn226.2 × 1019.0 × 1010.991.5 × 10−21.1 × 10−21.4
Na189.4 × 1029.9 × 1020.931.5 × 10−12.2 × 10−10.7
Si218.8 × 1031.2 × 1040.871.9 × 1004.6 × 1000.4
Sr222.3 × 1013.0 × 1011.005.0 × 10−34.6 × 10−31.1
Table 1b. Noncertified Trace Elements and Water-Soluble Cations in PM2.5 Aerosol Samples From Zhenbeitai and a Loess Certified Reference Material: Concentrations and Relationships to Aluminuma
ElementnConcentration, ng m−3Correlation With Al (r)Ratio to Al, unitlessZBT/CJ-1
MeanStandard DeviationZBT AerosolsCJ-1 CRMbCJ-1 Analysis
  • a

    Here, n, number of samples; ZBT, Zhenbaitai; WS, water-soluble; NA, not available.

  • b

    CRM, loess certified reference material [Nishikawa et al., 2000].

  • c

    Noncertified values in parentheses.

Sc84.6 × 1004.8 × 1000.815.5 × 10−4(1.8 × 10−4)c5.8 × 10−40.9
Ti202.6 × 1023.2 × 1020.985.1 × 10−2(6.0 × 10−2)6.0 × 10−20.9
Ce225.0 × 1007.6 × 1000.981.3 × 10−3NA1.0 × 10−31.2
Co204.5 × 1005.2 × 1000.738.3 × 10−4NA5.1 × 10−41.6
Dy223.5 × 10−15.1 × 10−10.998.4 × 10−5NA6.2 × 10−51.3
Er222.2 × 10−13.0 × 10−10.985.0 × 10−5NA4.1 × 10−51.2
Eu171.3 × 10−11.6 × 10−10.972.5 × 10−5NA1.5 × 10−51.6
Gd146.8 × 10−18.1 × 10−10.981.2 × 10−4NA9.7 × 10−51.2
La163.3 × 1004.1 × 1000.996.1 × 10−4NA5.1 × 10−41.2
Li223.4 × 1004.7 × 1000.967.8 × 10−4NA5.9 × 10−41.3
Nd222.1 × 1003.2 × 1000.985.3 × 10−4NA4.2 × 10−41.3
Pr225.7 × 10−18.6 × 10−10.981.4 × 10−4NA1.1 × 10−41.3
Sm224.3 × 10−16.5 × 10−10.981.1 × 10−4NA8.7 × 10−51.2
Th221.0 × 1001.5 × 1000.992.5 × 10−4NA1.8 × 10−41.4
U223.3 × 10−14.7 × 10−10.957.8 × 10−5NA5.2 × 10−51.5
WS Ca2+96.0× 1034.2× 1030.875.4 × 10−1NANANA
WS Mg2+93.8 × 1021.2 × 1020.813.5 × 10−2NANANA
WS Na+99.9 × 1021.1× 1030.891.4 × 10−1NANANA

[23] Al commonly is used as an indicator of mineral matter, and water-soluble calcium (WS Ca2+) also has recently used for this purpose for aerosol samples [e.g., Seinfeld et al., 2004; Jordan et al., 2003] and rainwater [Mace et al., 2003]. Data for cations in aqueous extracts of ZBT PM2.5 samples are limited, and no data for water-soluble extracts of the loess CRM are available for comparison, but WS Ca2+, WS Mg2+, and WS Na+ all were significantly correlated with aerosol Al (determined by ICP-MS) in the PM2.5 samples (Table 1b). The TSP samples from ZBT showed weaker, though still significant relationships between these cations and Al. For 66 daily samples collected under northerly flow, the correlation coefficients (r-values) for WS Na+, WS Mg2+, and WS Ca2+ versus Al were 0.70, 0.78, and 0.77, respectively.

[24] The ratios of Ca/Al and especially water soluble (WS) Ca2+/Al were higher at ZBT than at GOS (Figures 2a and 2b, the latter is discussed below), and the percentage of Ca that was water soluble at ZBT also was higher than at GOS: In fact, in several of the PM2.5 samples from ZBT, the concentrations of WS Ca2+ were the same (within 15 to 20% error) as total recoverable Ca. The ratio of total Ca to Al was approximately 1:1, and this is consistent with the similar concentrations of Ca and Al in the loess CRM.

Figure 2.

Relationships between water-soluble (WS) calcium and total recoverable calcium versus aluminum at (a) Zhenbeitai and (b) Gosan.

[25] In contrast to the crustal group, the concentrations of a second group of elements (Ag, Bi, Cd, Cr, Cu, Hg, Mo, Ni, Pb, Sb, Tl, V, and Zn) in the ZBT samples either were uncorrelated or weakly correlated with Al (Table 2). These elements also displayed ratios to Al (calculated as the arithmetic means of the ratios for the samples rather than the slopes from regression models) that ranged from 6 to more than 100 times those in the loess CRM. This second group of elements was therefore moderately to highly enriched over the levels expected from mineral dust, and their main sources likely include pollution emissions [e.g., Rahn, 1976; Nriagu and Pacyna, 1988]. Among these elements, the detection of Tl was noteworthy because of potential health concerns and because Tl is emitted by coal-fired power plants, from smelting operations and cement factories; all of which are important sources of pollutants in Asia.

Table 2. Concentrations of Enriched Elements and Water-Soluble Potassium in PM2.5 Aerosol Particle Samples From Zhenbeitai and a Loess Certified Reference Materiala
ElementnConcentration, ng m−3Correlation With Al (r)Ratio to Al, unitlessZBT/CJ-1
Arithmetic MeanStandard DeviationZBT AerosolsCJ-1 CRMb
  • a

    Here, n, number of samples; ZBT, Zhenbaitai; WS K+, water-soluble potassium; ND, no data.

  • b

    CRM, loess certified reference material [Nishikawa et al., 2000].

Certified Elements (Certified Element to Aluminum Ratio for CJ-1 CRM)
Cu209.98 × 1006.95 × 1000.766.2 × 10−33.9 × 10−416
Ni131.37 × 1011.52 × 1010.633.0 × 10−35.2 × 10−46
Zn224.78 × 1013.39 × 1010.373.5 × 10−21.2 × 10−330
Noncertified Elements (Observed Element to Aluminum Ratio for CJ-1 CRM)
Ag131.27 × 10−18.38 × 10−20.347.0 × 10−55.4 × 10−613
Bi117.43 × 10−13.05 × 10−10.287.9 × 10−45.9 × 10−6135
Cd214.13 × 10−12.27 × 10−10.033.7 × 10−41.0 × 10−536
Cr82.50 × 1018.77 × 1000.711.9 × 10−28.7 × 10−422
Hg75.27 × 10−11.07 × 10−10.304.1 × 10−46.2 × 10−666
Mo81.32 × 1001.33 × 1000.084.7 × 10−42.2 × 10−521
Pb222.34 × 1011.73 × 1010.062.5 × 10−24.2 × 10−460
Sb191.17 × 1001.42 × 1000.266.5 × 10−42.6 × 10−524
Tl225.54 × 10−15.14 × 10−10.186.8 × 10−41.1 × 10−560
V122.39 × 1011.76 × 1010.501.1 × 10−21.1 × 10−310
WS K+85.1 × 1022.9 × 1020.571.2 × 10−1ND

[26] Unlike the other cations, WS K+ was not correlated with Al or with any of alkali/alkaline earths or with the water-soluble cations at ZBT in the PM2.5 samples (Table 3). For the TSP samples, WS K+ was significantly correlated with Al, but the r-value for WS K+ versus Al (r = 0.54) was substantially (∼40%) lower than the r-values for other cations. This suggests a unique source for small-particle water-soluble K+; this most likely involves fine particles produced by biomass burning [Andreae, 1983; Cachier et al., 1996], but we have no way of rigorously evaluating the contributions from this source with the data available.

Table 3. Correlations Between Alkali/Alkaline Earths and Major Cations in PM2.5 Samples From Zhenbeitaia
 CaWS Ca2+KWS K+MgWS Mg2+NaWS Na+
  • a

    WS, water-soluble. The upper right portion of this table shows the number of sample pairs used in the calculations. The lower left portion of the table shows the correlation coefficients. Correlations (r-values) in bold type are significant p < 0.01, those in plain type are significant at p < 0.05, and dashes are not significant.

Ca 8197198138
WS Ca2+0.90 888979
K0.79  7248188
WS K+  7868
Mg0.790.98 8188
WS Mg2+0.790.820.760.74 79
Na0.960.760.940.940.84 7
WS Na+0.840.700.960.950.780.97 

[27] The results from ZBT suggest that even during strong dust events, significant quantities of pollutants can be mixed with Asian dust even near source regions, at least over the 24-hour sampling intervals used for ACE-Asia. Even so, the dust concentrations at ZBT at times are so high that the pollution signals are partially masked. This can be demonstrated by plotting Al concentrations along with enrichment factors (X/Al in the sample divided by X/Al in the CJ-1 loess reference material [Rahn, 1976], where X represents any element of interest; as noted above, the concentrations of some elements in the CJ-1 loess were up to 20 times higher than in average crustal rock). The enrichment factors for Pb and Sb, for example (Figure 3), were lowest when dust concentrations were high, but even during the 9 April dust storm when the Al concentrations increased 150-fold over typical nondust conditions, the concentrations of these elements still were elevated 2–3 times over the levels expected from the CJ-1 loess CRM.

Figure 3.

Time series plots of enrichment factors for lead and antimony in PM2.5 samples from Zhenbeitai. Enrichment factors are the concentration of the element of interest (X) divided by the Al concentration in the sample divided by X/Al ratio in the CJ-1 loess reference material.

[28] Whether the mixing of dust and pollutants occurs during transport to ZBT or the enrichments result from the resuspension of previously deposited chemical contaminants remains an open question. Other observations and models for ACE-Asia show that dust, pollution aerosols, and smoke from biomass burning often exist as distinct layers initially, but mixing and reactions between trace gases and aerosol particles can cause the cycles of various chemical constituents to become linked in complex ways [Seinfeld et al., 2004].

3.2.2. Aerosol Ions and Their Relation to Dust at Zhenbeitai

[29] It is evident from time series plots of TSP nitrate, sulfate, and WS Ca2+ (Figures 4a–4c) that the cleanest air at ZBT with respect to NO3 and SO42− comes from the north. Paradoxically, during dust storms, northerly winds also bring the highest calcium and sulfate concentrations to the site (Table 4 and Figures 4b and 4c). Southerly winds periodically transport pollutants from the nearby Yulin urban and industrial region to ZBT, elevating each of the ions' concentrations above their values in nondusty northerly airflow.

Figure 4.

Time series plots of (a) nitrate, (b) sulfate, and (c) water-soluble calcium in total suspended particle samples from Zhenbeitai.

Table 4. Summary Statistics for Major Ions in Total Suspended Particle Samples From Zhenbeitai Under Different Wind and Dust Storm Conditions
AnalyteSouth Winds (n = 21)aNorth Winds (n = 73)
MedianbMeanStandard DeviationcMaximumMedianMeanStandard DeviationMaximum
  • a

    Here, n, number of samples.

  • b

    Concentrations in micrograms per cubic meter.

AnalyteNorth Winds (Non-Dust-Storm, n = 62)North Winds (Dust Storm, n = 11)
MedianMeanStandard DeviationMaximumMedianMeanStandard DeviationMaximum

[30] The ZBT samples offer an opportunity to examine the dust composition prior to its passage over the heavily populated regions of eastern China. However, since much of the dust originates to the west of ZBT [Zhang et al., 2003a], there is still the potential that it has mixed with pollutants prior to our sampling or that pollutants previously deposited in the source region have become resuspended. Indeed, we have shown above that several elements commonly associated with pollution are enriched at ZBT, so there is little doubt that some anthropogenic materials were mixed with the dust.

[31] Figure 5 offers one approach to assessing the amount of SO42− (upper panels) and NO3 (lower panels) in relatively clean dust. In these figures, the ratios of SO42−/WS Ca2+ and NO3/WS Ca2+ are plotted against the WS Ca2+ concentration, the latter being used to distinguish high-dust from low-dust samples. We note that Jordan et al. [2003] have used non-sea-salt calcium for the same purpose. The intent here is to find the ratios of analytes that exist when the dust concentrations are at their highest because under those conditions the relative influence of pollution is reduced. In samples with high soluble Ca concentrations, the molar ratios of sulfate to WS Ca2+ and nitrate to WS Ca2+ did indeed stabilize (Figure 5). Using an arbitrary cut point of 40 μg m−3 for soluble Ca as an indicator of high-dust conditions, the SO42−/WS Ca2+ molar ratio at ZBT in dusty air averaged 0.090 ± 0.035, while NO3/WS Ca2+ molar ratio averaged 0.015 ± 0.005. In contrast to these results, Jordan et al. [2003] found that in a set of high-dust samples that did not have enhanced pollution, non-sea-salt calcium was independent of nitrate and sulfate. There were some noteworthy differences between the two studies, however. First, the samples in the Jordan et al. study were not from near ground level but instead were collected using the DC-8 aircraft, and second, the dust concentrations in their samples were in general much lower than those at ZBT, most likely because of a greater distance from the dust sources. More important, Jordan et al. concluded that the colocation of dust and pollution sources can lead to the uptake of nitrate and sulfate on dust.

Figure 5.

(a) Sulfate versus soluble calcium and (b) molar ratio of sulfate to soluble calcium versus soluble calcium in total suspended particle (TSP) and PM2.5 samples from Zhenbeitai. The corresponding plots for (c) nitrate versus soluble calcium and (d) molar ratio of nitrate to soluble calcium. TSP samples are stratified by surface wind direction.

[32] The molar sulfate-to-soluble Ca slope for all northerly wind samples from ZBT was 0.096, while that for nitrate was 0.012: Both of these values are close to the ratios calculated for the samples collected under high-dust conditions. The values thus derived can be considered proxies for the amounts of sulfate and nitrate in dust prior to its mixing with the large quantities of air pollutants. It is not surprising that little NO3 appears to be in the soils from which this dust lofted. Not only were there less than 2% as many moles of NO3 as of WS Ca2+, but the correlation between the two analytes was weak, except at very high dust loads, and the intercept was about a third of the mean nitrate value. This implies that the source of the NO3 under nondust storm conditions was different from that of WS Ca2+. Small amounts of urban and industrial NOx emissions may have been responsible.

[33] In comparison to NO3/WS Ca2+, the SO42−/WS Ca2+ relationship was far more robust. The r-value for SO42− versus WS Ca2+ was 0.91, and the intercept for the regression for all samples collected during northerly winds was indistinguishable from zero. These results show that the during the aqueous extraction of the samples, dust releases ∼10% as many moles of sulfate as of calcium; therefore anhydrite or gypsum, the presumed sources for crustal SO42−, evidently make up ∼10% of the soluble calcium salts in the “unpolluted” dust reaching ZBT. The main form of calcium in Chinese desert dust, in fact, has been shown to be CaCO3 (J. Cao, Institute of Earth Environment, Chinese Academy of Sciences, personal communication, 2004).

3.3. Gosan

3.3.1. Mineral Aerosol at Gosan

[34] The dust concentrations at GOS during ACE-Asia tended to be higher than during the PEM-West experiments and in the intervening years (Figure 6), but the data are discontinuous, and therefore no conclusions should be drawn from these data regarding long-term trends. The average Al concentration for 71 samples collected for ACE-Asia was 10.1 μg m−3 with a standard deviation (sd) of 21.7; this is equivalent to a dust concentration of ∼170 μg m−3 assuming the dust has the same Al to mass ratio as the CJ-1 loess CRM. Closer inspection shows that the dust concentrations were high at GOS during the massive dust storm of 5–15 April 2001, but not extraordinarily so compared with other samples collected during ACE-Asia. The fact that the dust concentrations during this event were not especially high at Gosan but were at other sites serves to illustrate the heterogeneity of large dust storms.

Figure 6.

Aluminum (mineral aerosol) concentrations at Gosan during PEM-West and ACE-Asia. IOP marks the intensive observation period, which was in March and April 2001.

[35] With reference to trends in dust concentrations in the past few decades, an analyses of meteorological records by Natsagdorj et al. [2003] led to the conclusion that the number of dusty days in Mongolia increased from the 1960s to the 1980s but then started to decrease in 1990. Furthermore, these authors found that the annual precipitation increased in the 1990s, particularly from 1990 to 1994, and they offered this as one explanation for the observed decrease in dusty days. Trends in loess accumulation in China over geological scales similarly have been linked to variations in the monsoonal circulation and associated changes in the hydrological cycle [An et al., 1990]. Meteorological explanations highlighting the effects of drought and large-scale circulation patterns also have been invoked to explain decadal patterns in African dust storm activity (Prospero and Nees [1986] and Moulin et al. [1997], respectively).

[36] Human activities also can affect atmospheric dust loads, and even though modeling studies by Tegen and Fung [1995] and others have brought considerable attention to this possible effect, the relative influence of human activities versus natural variability is poorly constrained. Complicating the assessment of the relative influences of human activities versus climate on dust fluxes still further are large-scale engineering projects and land management practices designed to suppress dust production [Parungo et al., 1994]. Nevertheless, the existing evidence suggests that dust production is dominated by natural sources [Prospero et al., 2002]. In this regard, analyses of Asian dust emissions over the past 43 years [Zhang et al., 2003b] similarly led to the conclusion that meteorology and climate were more important influences on modern Asia dust fluxes than were human influences and desertification.

3.3.2. Aerosol Ions and Their Relation to Dust at Gosan

[37] To compliment the studies of dust and major ions at ZBT, we plotted the Ca and WS Ca2+ concentrations versus Al at GOS. As Figure 2b shows, the concentrations of both WS Ca2+ and total recoverable Ca at GOS are strongly correlated with total recoverable Al. For total Ca versus Al, r = 0.93 (Table 5) and for WS Ca2+ versus Al, r = 0.70; both of these are significant at a probability of p < 0.01. A plot of these data (Figure 2b) shows (1) the WS Ca2+ concentrations are substantially less than (∼52% of) the total recoverable Ca, (2) the majority of samples have WS Ca2+/Al mass ratios between 0.1 and 0.33 (for 54 samples, geometric mean equation imageg = 0.24, geometric standard deviation sdg = 2.15), and (3) the total recoverable Ca/Al ratios are typically between 0.33 and 1 (n = 58, equation imageg = 0.62, sdg = 1.54). Differences in the relationships between Ca and Al at GOS and ZBT are discussed below.

Table 5. Correlations Among Alkali/Alkaline Earths, Major Cations, and Aluminum at Gosana
 AlCaWS Ca2+KWS K+MgWS Mg2+NaWS Na+
  • a

    The upper right portion of this table shows the number of sample pairs used in the calculations. The lower left portion of the table shows the correlation coefficients. Correlations (r-values) in bold type are significant p < 0.01; dashes are not significant.

Al 6363645363546353
Ca0.93 64635363556451
WS Ca2+0.700.76 635363556451
K0.870.930.67 5363546352
WS K+0.710.730.760.64 52535350
Mg0.820.820.460.890.58 546351
WS Mg+0.580.610.680.570.820.65 5551
Na0.370.590.43 51
WS Na+0.640.630.470.630.770.720.820.55 

[38] Further analyses showed that with one exception, strong correlations existed among Al and the total recoverable and the water-soluble concentrations of K, Mg, Ca, and Na at GOS (Table 5). The one exception was total Na, which was not significantly correlated with Al, Ca, WS Ca2+, or WS K+ (WS Na+ was positively correlated with all other analytes). The Na/Al ratio for the GOS samples (arithmetic mean = 7.6, sd = 16.2, n = 63) was more than 30 times that in the loess CRM, and therefore noncrustal sources for Na predominated, with sea-salt aerosol the most obvious candidate. In comparison, the arithmetic mean K/Al ratio at GOS was 0.61 (sd = 0.64, n = 64) and for Mg/Al the mean ratio was 1.0 (sd = 1.9, n = 63); these ratios are only ∼2 and 3 times those in the loess CRM, respectively, indicating much weaker influences of sea salt on these elements compared with Na. Strong seasonality also is evident in the constituent ratios, that is, the Mg/Al (and K/Al) ratios are especially close to crustal values during the springtime high-dust events. For example, the 34 samples with data for both Mg and Al collected in March and April, had a Mg/Al ratio of 0.43 (sd = 0.41); in contrast, the ratio was 1.76 (sd = 2.55) for the 29 samples collected during other months.

[39] Having established a relationship between WS Ca2+ and Al, we next examined the GOS data for relationships between SO42−, NO3 and WS Ca2+ for comparison with the ZBT results. Using the same approach as described above for ZBT, plots were constructed of the nss SO42− and NO3 concentrations and molar SO42−/WS Ca2+ and NO3/WS Ca2+ ratios versus WS Ca2+ (Figures 7a–7d). A linear regression of nss SO42− versus WS Ca2+ for the daily samples demonstrated a highly significant correlation, with r = 0.81, a slope of 1.7 and an intercept of 5.8. At high WS Ca2+ concentrations, the molar SO42−/WS Ca2+ ratio approached unity, and this is approximately tenfold higher than what was observed at ZBT. The molar ratio of NO3/WS Ca2+ was between 10 and 100 times higher at GOS compared with ZBT, even when matched to the lower dust loadings (Figure 7d). These differences between the GOS and ZBT results for Ca, SO42−, NO3, and dust are discussed in detail below.

Figure 7.

(a) Sulfate versus soluble calcium and (b) molar ratio of sulfate to soluble calcium versus soluble calcium in total suspended particle samples from Gosan. The corresponding plots for (c) nitrate versus soluble calcium and (d) molar ratio of nitrate to soluble calcium.

3.3.3. Aerosol Ions at Gosan

[40] Samples from GOS that were collected for ACE-Asia and the various other programs were not matched by season, and they also were collected over different time intervals, ranging from less than a day to roughly a week. Therefore we caution that the summary statistics for the different studies summarized in Table 6 are not directly comparable. The differences in nitrate and sulfate concentrations between experiments are small in any case: Nitrate concentrations during ACE-Asia are at most slightly higher than those previously measured while the differences in nss SO42− concentrations are even smaller than those for nitrate (Table 6).

Table 6. Summary Statistics for Non-sea-salt Sulfate (nss SO42−) and Nitrate (NO3) for TSP Samples From Gosan From PEM-West to ACE-Asia
AnalyteProgramaNumber of SamplesSum of WeightsWeighted MeanbStandard ErrorLower 95%Upper 95%
  • a

    PEM-West: 10 September 1991 to 18 March 1994. Interim: 31 October 1995 to 27 May 1996. ACE-Asia: 25 March 2001 to 30 September 2001.

  • b

    Concentrations weighted by sample volumes in cubic meters.

nss SO42−, μg m−3PEM160146,0757.120.366.407.84
nss SO42−, μg m−3Interim31103,0157.020.436.177.88
nss SO42−, μg m−3ACE5322,6917.780.935.959.60
NO3, μg m−3PEM160146,0754.110.183.754.46
NO3, μg m−3Interim31103,0154.330.213.914.75
NO3, μg m−3ACE6525,9015.310.434.476.16
nss SO42−/NO3PEM160146,0751.800.081.641.95
nss SO42−/NO3Interim31103,0151.840.101.652.02
nss SO42−/NO3ACE5322,6911.560.201.161.96

[41] The ratio of nss SO42− relative to NO3 in aerosols showed no strong trends from PEM-West to ACE-Asia, with the ratios most commonly between 1 and 2. While the data coverage at GOS is not continuous, these values are similar to those observed during long-term studies at Midway [Prospero et al., 2003]. At that central North Pacific site (28°13′N, 177°22′W), the ratio of the annual mean nss SO42− to NO3 was 1.9, and there has been no evidence for a systematic change over time in the relative concentrations of anthropogenic SO42− and NO3. The ratios of these two species do vary by about a factor of 2 from year-to-year at Midway, and because of this variability it is unlikely that small changes in the Asia emission signature would be detectable. Moreover, direct comparisons of the GOS and Midway results are not warranted because the data from GOS are not necessarily representative of Asian outflow. However, the overall consistency in the ion concentrations at GOS and the general absence of large swings in the nss SO42−/NO3 ratios suggests that the aerosols there are probably not greatly impacted by local sources.

[42] The MSA concentrations at GOS during ACE-Asia were similar to those observed for PEM-West during the same season in the early to mid-1990s [Arimoto et al., 1996], and as in the prior experiments, the nss SO42−/MSA mass ratio (Figure 8) was much higher than the canonical ratio of ∼14 that has been used to evaluate the contribution of marine biogenic sources to nss SO42−[Saltzman et al., 1983; Arimoto et al., 1996; Savoie et al., 2002; Prospero et al., 2003]. Using MSA as an indicator, Arimoto et al. [1996] concluded that biogenic sources could account for only ∼10 to 20% of the nss SO42− at a network of coastal/continental sites in eastern Asia; the ACE-Asia results for GOS are simply further evidence for the dominance of nonbiogenic sources of nss SO42−. MSA was not detected in the PM2.5 samples from ZBT; this is not unexpected given the distance from ZBT to any significant sources for marine biogenic substances.

Figure 8.

Methanesulfonate and non-sea-salt sulfate concentrations at Gosan. “Biogenic” line marks a ratio of ∼14, which has been used an empirical indicator of biogenic sulfate.

3.4. Dust and Aerosol Chemistry: Some Implications for Climate

[43] The studies summarized here were conducted to characterize the elemental and major ion composition of aerosols from two sites in Asia. A thorough treatment of aerosol transport and evolution is beyond the scope of this paper, but the data from our study do provide some insights into the relationship between water-soluble ions and mineral dust, a topic that has received relatively little attention beyond a few studies in the 1980s [Savoie and Prospero, 1980; Talbot et al., 1986] and the recent study by Jordan et al. [2003]. It also is clear from our studies and others that it is common for dust to be mixed with pollution even close to the dust source region. This is important in the context of climate because the mixing of aerosol populations, in particular whether the aerosol is internally or externally mixed, can influence their radiative effects. For example, calculations by Seinfeld et al. [2004] indicate that under cloud-free conditions the internal mixing of organic and elemental carbon with sulfate and Asian dust can result in a net change in climate forcing of >1 W m−2 relative to the externally mixed case.

[44] At ZBT, the molar ratio of sulfate to WS Ca2+ in TSP samples collected under high-dust conditions is approximately 0.1 Therefore, when evaluating SO42− in coarse particles downwind, one might assume as a first approximation that anything above this sulfate/soluble calcium molar ratio is the result of gas-to-particle conversion, with alkaline calcium either picking up SO2 gas, adsorbing H2SO4 vapor, or coagulating with submicron sulfate aerosols. In this regard, the observation that the SO42−/WS Ca2+ ratio at GOS during high-dust conditions is roughly 10 times that at ZBT implies that significant quantities of SO2 or sulfate were taken up by the dust during transit. Parallel comparisons of the ZBT and GOS data for NO3 suggest that uptake of nitrogen oxides by dust is of the same order as the uptake of SO2 and SO42− if not higher. Removal of the sulfur and nitrogen oxides onto aerosol surfaces will not only affect chemical budgets but also alter the number and size distributions of aerosols. Such changes in aerosol size, number and composition not only can affect the aerosol light scattering and absorption but also can indirectly affect climate through clouds by altering cloud albedo and lifetime.

[45] Another factor that may have contributed to the differences in the relative concentrations of the ions at ZBT and GOS is the preferential removal of WS Ca2+ relative to SO42− or NO3 by dry deposition. The mass of Asian dust is predominantly in the 2–20 μm diameter size class [Zhang et al., 2003a], that is, on particles larger than would be collected with a PM2.5 sampler. Plots of the PM2.5 and TSP data from ZBT show that a major fraction of the SO42− can occur on giant particles during some dust events, such as the one in late April/early May (Figures 4b and 4c), but in general one would expect nss SO42− to be on particles smaller than dust. Thus the larger WS Ca2+-bearing dust particles may be removed more rapidly by dry deposition than the SO42− containing particles, leading to higher SO42−/WS Ca2+ ratios as the aerosols are transported downwind. Studies at other sites such as Barbados similarly have shown that the nss SO42− and dust concentrations are correlated and that the largest percentage of coarse particle (>1 μm diameter) nss SO42− occurs during dust storms [Li-Jones and Prospero, 1998].

[46] A recent study by Mori et al. [2003] provides an estimate of the amount of NO3 that attaches to Asian dust surfaces during transport which is comparable to but toward to the lower end of the tenfold to hundredfold increase of the molar NO3/WS Ca2+ ratio our data suggest. These authors analyzed data for size-separated aerosol samples collected along a pathway for dust moving from inland China to Japan, and they concluded that the amount of NO3 attached to aerosol surfaces in the kosa (dust) samples from Japan was nearly 10 times higher than in samples from near the dust source in China. They also presented evidence showing that SO42− accumulates on the surface of Asian dust during transport, but to a lesser degree. Studies such as these are important because they provide a means for constraining numerical models of the dust's influence on tropospheric chemistry [Phandis and Carmichael, 2000] and climate.

[47] How can the higher Ca solubility in dust samples close to the source region be explained? The most straightforward explanation is that the Ca was associated with different types of aerosols at the two sites. Some of this effect may be due to different combinations of minerals from the dust sources, but it is likely that during transport the dust at GOS was mixed with aerosols containing insoluble Ca, possibly anthropogenic particles such as coal fly ash or oil ash or construction dusts. Mattigod et al. [1990] showed that the concentrations of Al and Ca in fossil fuel combustion wastes were within their ranges in soils, but S (which mainly existed as sulfites and sulfates) was more likely to be enriched over soil values in the combustion residues. Furthermore, these authors showed that the Ca-containing solid phases in the fossil fuel wastes included various mixed silicates, which presumably are less soluble than the Ca-sulfate minerals. The presence of aerosols from fossil fuel combustion would thus explain both the lower solubility of Ca in the GOS samples (more insoluble anthropogenic Ca-silicates at GOS) and the higher SO42−/WS Ca2+ ratio (more anthropogenic particles enriched in sulfates relative to Ca at GOS).

[48] The issue of whether the higher SO42− to WS Ca2+ ratio in the samples at GOS compared with ZBT was mainly due to gas-particle conversion, size selective fractionation, or mixing of aerosol types is fundamental to understanding what controls the composition of Asian outflow and how the aerosols affect climate. In this regard, analysis of individual particles collected from a balloon near Beijing showed evidence that sulfate and sulfuric acid formed on mineral particles heterogeneously [Xu et al., 2001]. Along these same lines, Kim and Park [2001] concluded that significant amounts of both SO42− and NO3 formed on Ca-rich coarse aerosol particles collected in Korea, and similar conclusions reached by Jordan et al. [2003] have been mentioned above.

[49] The effects of mixing of anthropogenic particles, especially the occurrence of glasses, would likely be evident through electron microscopic examinations, and one study of this type [Ro et al., 2001] showed that the aluminosilicates and other particles in “Asian Dust” deposited in rainwater in Seoul, Korea were similar to those in a China Loess sample. These authors thus concluded that there were minimal influences from local dust sources. Along these same lines, elemental analyses of size-separated samples by Mori et al. [2003] showed that the concentrations of crustal elements, e.g., Al and Fe, in coarse particles from China and Japan varied in the same way as mass loadings, implying no fractionation of dust-associated elements. Other elements such as Pb and Zn, however, were higher in the downwind aerosol particle samples, demonstrating pollutant mixing with the dust.

[50] Limited information on the importance of heterogeneous processes, particle mixing and size-selective removal exists because few studies have sampled dust at multiple sites during the same dust event, and none have followed the chemical evolution of a dust plume, measuring trace elements and major ions in the process. In addition, second-order effects, such as the chemical consequences of particle mixing have not been studied in detail. One possible effect involves the mixing of sea salt and dust. Along these lines, Fan et al. [1996] collected samples from the same dust storm in China and Japan, and they found a large proportion of internal dust/sea-salt mixtures in samples from Nagasaki, thus demonstrating the mixing of dust with marine aerosols. Single particle analyses [Zhou et al., 1990] have similarly shown evidence of dust mixed with sea salt, and it stands to reason that these mixing processes may modify aerosol composition and properties in coastal regions. Therefore, at GOS the oxidation of SO2 to SO42− in internally mixed dust/sea-salt aerosols via reactions with ozone [Sievering et al., 1992] could play a role in determining aerosol ion composition and the relationships between the ions and dust (or Al).

[51] The most definitive approach to addressing the uptake, mixing and fractionation issues and how they relate to climate, would involve Lagrangian-type experiments in which the evolution of dust plumes was followed in space and time. Studies of this nature have been the hallmark of some ocean-based studies, but none have specifically targeted dust over the continents. The Lagrangian approach would be especially appealing for studies of dust mixing with pollution because heterogeneity within dust plumes would likely make it difficult to interpret data from multiple ground sites even if the samples were collected in a coordinated manner.

[52] Our investigations into the relationship betweens Ca, WS Ca2+ (and other cations) and Al in Asian mineral dust may be reasonably representative of mineral dust collected near the surface troposphere, but it is important to point out that the dust sampled at ZBT or GOS likely differs in composition from that transported long distances. This is the case because the extent of mixing and the reactions with pollutants are sure to be different for dust transported aloft compared with that transported near the Earth's surface. With reference to regional variability in Asian dust, Sun [2002] concluded that dust from different source regions, and presumably with different compositions, differs not only in its susceptibility to long-range transport but also in climatic effects, due in large measure to interactions with pollutants. Further studies of heterogeneous reactions, aerosol mixing, and size-selective fractionation of what Sun termed local, medium-distance, and long-distance dust will better constrain budgets of key atmospheric constituents and improve our understanding of the interactions between dust and climate.


[53] The loess certified reference material (CJ-1) was generously provided by M. Uematsu, University of Tokyo. We thank J. Conca for insightful discussions of mineralogy. This paper is based on work supported by the National Science Foundation under grants NSF ATM 0002054 and ATM 0002599. The University of Hawaii work was supported by NSF grant ATM00-02604. This research also was supported by grants G2000048703 and KZCX2-305 from CAS and MOST of the People's Republic of China. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsors. The National Research Laboratory of the Meteorological Research Institute (Korea Meteorology Administration) contributed to the measurements of aerosol at Gosan during the ACE-Asia IOP. This research is a contribution to the International Global Atmospheric Chemistry (IGAC) Core Project of the International Geosphere Biosphere Program (IGBP) and is part of the IGAC Aerosol Characterization Experiments (ACE).