Journal of Geophysical Research: Atmospheres

Inorganic composition of fine particles in mixed mineral dust–pollution plumes observed from airborne measurements during ACE-Asia



[1] Chemical characteristics of inorganic water-soluble aerosol particles measured in large Asian springtime dust events during the Asian Pacific Regional Characterization Experiment (ACE-Asia) were investigated. Three specific flights (flights 6, 7, and 10) in the Yellow Sea boundary layer with high mineral dust concentrations mixed with pollutants from Asian urban centers are presented. Measurements during a similar campaign, Transport and Chemical Evolution over the Pacific (TRACE-P), in the same region suggested that fine-particle ammonium sulfate and nitrate salts, and potassium, apparently from biomass burning, are common particle ionic constituents in polluted air. Observations from the ACE campaign show similar characteristics and found that the main component of water-soluble mineral dust was Mg2+ and Ca2+. Ion charge balances of measured fine and total aerosol suggest that a significant fraction of the Mg2+ and Ca2+ observed were in the form of carbonates. In polluted air mixed with dust that advected directly from large urban regions in roughly half a day to 1 day (flights 6 and 7), much of the fine-particle nitrate and sulfate (approximately 80%) was apparently associated with ammonium or potassium, the rest likely associated with mineral dust. Only air masses that spent 2–5 days over the Yellow Sea (flight 10) had clear evidence of Cl depletion. Initial mass accommodation coefficients much less than 0.1 for uptake of SO2 or HNO3 by mineral dust in urban plumes containing fossil fuel and biomass-burning emissions could explain the observations. The data suggest an accommodation coefficient dependence on relative humidity.

1. Introduction

[2] The combination of Asian springtime frontal activity, large arid regions, such as the Gobi desert, and dry open fields prepared for planting, can produce massive Asian dust storms capable of transporting mineral aerosol particles across oceans to distant continents [Uematsu et al., 1983]. Owing to the extent and mass of these dust outbreaks, the airborne particles can influence the global radiation balance and alter climate [Seinfeld et al., 2004]. Downwind and in close proximity to the Asian dust source are some of the planet's largest emissions of industrial pollution and biomass burning which when blended with mineral dust produces a complex chemical mix of aerosol particles and gases. Possible interactions between dust and pollution are important because they can change the physical properties of the mineral dust, such as transforming a solid fairly insoluble dust particle into a soluble salt that can form a droplet, altering its light scattering properties, ability to serve as a cloud condensation nuclei (CCN), and propensity for precipitation scavenging [Krueger et al., 2003].

[3] The Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) was conducted in the spring of 2001 to investigate the radiative properties and physical and chemical characteristics of these Asian dust events. The experiment was a large multi-investigator study involving ground, ship, and airborne measurements, during the time of year when dust storms are most frequent and intense. This paper uses the ACE-Asia C-130 aircraft data to investigate the interaction of dust and pollution and focuses on the particulate water-soluble inorganic compounds. For comparison, a flight conducted one month prior to the ACE-Asia mission from a similar experimental campaign, Transport and Chemical Evolution over the Pacific (TRACE-P), is used to provide insight into the region's pollution before the dust storm season.

[4] Laboratory experiments show that mineral dust calcium carbonate (CaCO3) reacts with nitric oxide (NO2), nitric acid (HNO3), and sulfur dioxide (SO2) to produce calcium nitrate, Ca(NO3)2, and calcium sulfate, CaSO4 [Underwood et al., 2001; Mamane and Gottlieb, 1989; Dentener et al., 1996; Zhang et al., 1994; Krueger et al., 2003]. Calcium nitrate can be formed via two major daytime pathways, either through the direct condensation of gaseous nitric acid onto a CaCO3 particle which then reacts to produce calcium nitrate, equation (1), or via adsorption of nitric oxide (NO2(g)) onto a layer of condensed water coating the dust particle surface, which subsequently reacts with the water to produce absorbed nitric acid (HNO3(a)) and release gaseous nitrous acid (HONO(g)), equation (2) [Pakkanen, 1996]. This nitric acid is then converted to calcium nitrate via reaction (1).

equation image
equation image

Daytime HONO reactions can go on to produce OH and NO radicals that continue the cycle for NO2 and HNO3 formation. Also, as the reaction with CaCO3 proceeds by reaction (1), the production of water allows for continuing NO2 oxidation as well as a solution for soluble ion dissociation.

[5] Underwood et al. [2001] made a comparison of the mass accommodation coefficient (γ) for NO2 and HNO3 uptake on mineral dust particles ranging from aluminum oxide to calcium oxide and China loess dust particles. (The accommodation coefficient is the fraction of molecules that stick after collision with the particle.) The initial mass accommodation coefficients for the dust particle uptake of these gases indicate that HNO3 (γ ≈ 10−3 to 10−5, depending on mineral type; China loess γ = 5.2 × 10−5) is adsorbed more effectively than NO2 (γ ≈ 10−5 to 10−8, China loess γ = 2.1 × 10−6) on mineral particles. These numbers are also dependent on mass loading of the mineral aerosol. The initial accommodation coefficient of NO2 for the dry particle is very small, but increases with relative humidity (RH). As this happens, NO2 is able to adsorb onto the available H2O of a wetted particle surface and readily becomes aqueous HNO3 [Underwood et al., 2001; Mamane and Gottlieb, 1989; Dentener et al., 1996], although likely a very minor process compared to the direct adsorption of HNO3. As the surface becomes more wetted, the accommodation coefficients for HNO3 and SO2 approach 0.1, the value for pure water [Finlayson-Pitts and Pitts, 1999; Dentener et al., 1996; Van Doren et al., 1990; Worsnop et al., 1989]. Gaseous nitric acid-calcium carbonate reactions are thought to dominate over the NO2 pathway for production of calcium nitrate.

[6] The formation of CaSO4 can occur by direct deposition of the acid (gaseous or particulate H2SO4) shown in equation (3), or by adsorption of SO2 onto a wet film on the particle surface followed by heterogeneous oxidation. In the latter case, the oxidation of SO2 to SO42− requires a pH-dependent oxidation step and can be much more complex than that of NO2. The initial dissolution of SO2 in water (equation (4)) is followed by the subsequent oxidation of SO32− in the presence of O3, H2O2 and other peroxides, OH radicals, or O2 catalyzed by transition metals, to form SO42− [Warneck, 1999; Finlayson-Pitts and Pitts, 1999].

equation image
equation image

[7] The rate for the continued oxidation of the species in reaction (4) to form SO42− has been shown to increase with pH and available oxidizing species [Warneck, 1999; Finlayson-Pitts and Pitts, 1999]. At pH > 8, and in the presence of water (RH > 50%), the reaction of SO2 on mineral aerosol can be sufficiently fast so that the rate limiting step is gas diffusion [Judeikes et al., 1978; Dentener et al., 1996]. At sufficient RH, the dust alkalinity may influence the oxidation of SO2 and therefore the reaction rate with the mineral aerosol. Alkaline calcium carbonate should enhance the oxidation of SO2 relative to less alkaline mineral dust compounds such as Al2O3.

[8] In this paper, measurements of bulk, fine-, and, to a lesser extent, total (fine + coarse), particle ionic composition are used to investigate evidence for heterogeneous reactions in Asian dust plumes that have been recently mixed with urban pollution. Our approach is to use the more time-resolved airborne measurements of fine-particle composition instead of the highly time-integrated filter measurements of total aerosol to investigate the ion balances for each type of emission source: mineral dust, urban pollutants, and sea salt. Although most of the mineral dust mass is associated with coarse particles, the fine-particle chemistry does provide insights into dust-pollution interactions since an appreciable fraction (∼20% by mass, based on our data) of the dust appears to reside in the fine mode, and the mass transfer rate of vapor to fine particles is more effective than to coarse particles. This is addressed in more detail in the Discussions.

2. Experiment

[9] The National Center for Atmospheric Research (NCAR) C-130 aircraft was one component of the ACE-Asia research mission deployed in the western Pacific for measurements of particle physical and chemical properties. The aircraft was stationed at the Iwakuni Marine Corps Air Station (MCAS), Japan, from 30 March 2001 to 6 May 2001 and flew within the range of 20°–45°N latitude and 120°–145°E longitude. Most of the instruments deployed to measure the aerosol chemical composition sampled from a low-turbulence inlet (LTI), which tends to enhance concentrations of supermicron particles in the diffuser section of the inlet (i.e., inlet penetrations are greater than 1 for particles larger than ∼1 μm diameter). Comparison of particle sampling efficiencies between a solid diffuser inlet and LTI were conducted on the C-130 prior to ACE-Asia [Huebert et al., 2004].

[10] Care was also taken in the plumbing running from inlet to detectors by minimizing turbulence at flow splitters, using shallow bends, maintaining laminar flows, and locating instruments as close as possible to the inlet. Particle sampling and transport issues are mainly of concern for measurements of coarse particles (diameters greater than roughly 1 μm) and lead to uncertainties in reported coarse particle concentrations. Calculated fine-particle transmission efficiencies are greater than 90% (see Willeke and Baron [1993] for a summary of transmission efficiency formulas). Further discussion on sampling efficiencies and results from intercomparison studies of the various instruments deployed are given by Ma et al. [2004] and Moore et al. [2004]. The methods used to measure the fine and total aerosol ionic composition are briefly described.

2.1. Measurements of Water-Soluble Fine-Particle Composition With the PILS-IC

[11] The Georgia Tech particle-into-liquid sampler coupled to a dual-channel ion chromatograph (PILS-IC) operated continuously to quantify the soluble ionic species: Na+, NH4+, Ca2+, Mg2+, K+, Cl, NO3, and SO42−. The separation for the selected anions and cations takes 4 min for our IC configuration. During that period only particles sampled over a 3 min 24 s period are analyzed. Thus 3 min 24 s integral measurements are repeated continuously every 4 min. A more detailed instrument description, and a discussion of IC calibrations and operation can be found in the literature [Weber et al., 2001; Orsini et al., 2003].

[12] On the basis of baseline noise from measurements of ambient air filtered to remove particles, detection limits were calculated to be in the range of 40–60 ng/m3 for cations and approximately 10 ng/m3 for anions. Measurement precision determined through side-by-side operation of identical instruments is roughly 5% [Orsini et al., 2003]. However, the overall uncertainty associated with each reported ion is estimated at 20% mainly because of uncertainties in liquid and sample-air flow rates. Ratios of measured ions for a single simultaneous cation-anion injection are more precise since they mainly depend on IC calibrations and the accuracy of the peak integration, which is estimated to be better than 5% when peaks are significantly above the limits of detection.

[13] Recent comparisons with impactor measurements of fine-particle ionic composition, also located on the C-130 during ACE Asia [Ma et al., 2004], and laboratory experiments, suggest that the PILS may systematically under measure NH4+ by approximately 15%. Comparisons from other studies, however, do not show this same bias [Orsini et al., 2003]. Although this is a systematic error, it is within the range of our measurement uncertainty and does not significantly influence our conclusions. (Note that a systematic NH4+ undermeasurement of 15% results in an approximately 13% lower slope when (NO3 + SO42−) is regressed against (K+ + NH4+); see Figures 6, 10, 13, and 16. In the following analysis, the ion charge balance is calculated using the measured NH4+ and this additional uncertainty is applied to the regression slopes involving (NO3 + SO42−) versus (K+ + NH4+)).

[14] The size-dependent sampling efficiency of the PILS employed for this mission was determined from laboratory experiments with calibration aerosols produced via a Vibrating Orifice Aerosol Generator (VOAG, TSI Inc, St. Paul, Minnesota) and nebulizer/differential mobility analyzer (DMA) combination. The efficiency curve is shown in Figure 1 and indicates that the upper size limit at 50% efficiency is ∼1.3 μm. However, the curve is fairly shallow; 2 μm diameter particles are sampled at approximately 25% efficiency and 3 μm particles with ∼5% efficiency. This version of the PILS was also not 100% efficient. At the smaller sizes the instrument efficiency is ∼87%. Experimental results show that this is due to losses in the PILS turbulent steam-ambient air mixing region and is accounted for by multiplying the measured concentrations by 1.15. The more recent PILS design has 100% collection efficiency for particles up to 10-μm diameter (see Orsini et al. [2003] for a more complete discussion). The PILS has been extensively intercompared with other measurement techniques. For example, on the basis of a number of ground-based studies of fine particles (PM2.5), PILS sulfate typically agrees with other measurement methods to within 15% or better [Drewnick et al., 2003; Orsini et al., 2003; Weber et al., 2003].

Figure 1.

PILS sampling efficiency based on laboratory calibrations.

2.2. Bulk Particle Composition, SO2, and Other Measurements

[15] In an effort to obtain airborne measurements of aerosol mass that includes coarse particles without bias due to sampling losses, B. Huebert and group at the University of Hawaii deployed the total aerosol sampler (TAS) [Huebert et al., 1998; Kline et al., 2004]. The TAS is located in a small enclosure outside the aircraft cabin. The inlet to the sampler is a standard solid diffuser, but by locating the filter immediately behind the diffuser cone, and by extracting the cone itself for deposited particles, sampling biases due to inlet and transport losses are removed. Typical sampling integral times for the TAS were 20 min to 1 hour, and typically 2 to 5 measurements were made per flight. The extracts from the TAS cone/filter assemblies were analyzed by suppressed ion chromatography on two Dionex ICs (one for anions and one for cations), using procedures identical to those described by Huebert et al. [1998] for the ionic components; Na+, NH4+, Ca2+, Mg2+, K+, Cl, NO3, C2O42− and SO42−. For both the PILS and TAS measurements, all volumetric concentrations are reported at the standard conditions of 20°C and 1 atmosphere.

[16] Sulfur dioxide (SO2) was measured with an atmospheric pressure ionization mass spectrometer (APIMS) [Thornton et al., 2002]. A. Clarke and group of the University of Hawaii measured aerosol number size distributions via a combination of differential mobility-condensation particle counter and optical particle counters. Further details on these measurement techniques can be found in the literature [Clarke et al., 2004]. The NCAR Research Aviation Facility measured various meteorological parameters. Air mass back trajectories are calculated at 3-min intervals along the flight paths (G. Carmichael, University of Iowa) using Regional Atmospheric Modeling System (RAMS) 80 km resolution prediction driven by European Center for Medium-Range Weather Forecasts (ECMWF) 1° by 1° reanalysis data. Emission inventories [Streets et al., 2003] are used with the back trajectories to give some sense of possible sources for the various observed plumes.

2.3. Measurement Limitation

[17] Our measurements of aerosol chemical composition can be adversely influenced by solubility limitations, issues relating to particle volatility, and sampling artifacts associated with adsorption of gases or volatilization of particulate species during sample collection and manual handling of samples. For all the water-based analysis systems, calculations suggest that aqueous dilutions are sufficient such that solubility of various mineral compounds (e.g., CaCO3, CaSO4) measured during this experiment are not limited by reaching saturation levels.

[18] Aerosol heating and loss of semivolatile components can also be problematic when sampling from an airborne platform. For ionic components this could lead to losses of particulate ammonium and nitrate. In the case of the PILS located in the aircraft cabin, the temperature, pressure, and relative humidity of sample air just upstream of the instrument was measured. Even though the sample lines were relatively short (2.2 m), sample heating occurred mainly because of differences between ambient and cabin temperatures. In the boundary layer where most plumes were detected, the sample air temperature just upstream of the detector was typically increased above ambient by ∼10°C.

[19] Adsorption of gaseous compounds onto collected particles or filter substrates can lead to positive artifacts [Chow, 1995]. The PILS-IC is operated with gas denuders to minimize this effect. The TAS integrated filter measurements were not denuded and thus more prone to overmeasurement. This artifact can be especially problematic if the chemical composition of the sampled air varies widely during the sample integration period.

[20] Because the TAS measurement involves off line analysis, the storage of samples for subsequent manual extraction can lead to evaporation losses and sample contamination. To minimize these effects the sample substrates were stored in a clean cold environment and the analysis performed shortly after each flight.

3. Results

[21] Asian mineral dust, which contains significant amounts of calcium in the form of CaCO3 [Nishikawa et al., 2000], is likely the main source for Ca2+ observed during ACE-Asia. There are other possible sources of Ca2+. In Shanghai, for example, dust from roads and building construction typically comprises ∼22% of primary total PM (particulate matter) emissions and 7% of primary PM2.5 emissions; Ca2+ from cement manufacture comprises 21% of total PM emissions and 13% of PM2.5 [Li et al., 2004; Carmichael et al., 2003; Streets et al., 2003]. Sea-salt particles also contribute Ca2+ in coastal regions. However, in the spring during the ACE-Asia campaign, the measured Ca2+ is clearly associated mainly with large dust storms readily identified in satellite images (e.g., NASA TOMS products available at

[22] Although most of the mineral dust is coarse particles and would not be measured by the PILS, Figure 2 shows that there is a high correlation (r2 = 0.95) between the PILS measurements of fine Ca2+ and the TAS measurements of total Ca2+concentrations. (PILS Ca2+ is calculated by averaging over each TAS sampling time interval). This could be due to a consistent fine-mode tail of the dust size distribution, or due to the apparent nonzero PILS efficiency for particle diameters larger than 2 μm (see Figure 1). (Note that the correlation is not as good when only considering the lower concentration Ca2+ events. When TAS Ca2+ is less than 5 μg/m3, r2 = 0.65 and the PILS versus TAS slope is 15%). In any case, this correlation suggests that the higher-resolution PILS Ca2+ data can be used as an indicator for the presence of fine and coarse Ca2+, especially in the cases where the dust concentrations are high, which is the focus of this paper.

Figure 2.

PILS fine versus TAS fine plus coarse measurements of water-soluble calcium for all ACE-Asia data. Data points are flight numbers. The linear regression with 95% confidence intervals does not include the one outlier of flight 5.

[23] Using the PILS Ca2+ measurements, Figure 3 shows the C-130 flight paths during ACE-Asia and identifies the regions of highest fine water-soluble Ca2+ concentrations. For the intensive missions performed in ACE-Asia, there are essentially two geographical locations where highest Ca2+ concentrations were detected; during flight 5 over the Sea of Japan, and four separate flights, 6, 7, 10 and 13, all over the Yellow Sea (see also Figure 2). In this paper we focus on the mineral dust measured in flights 5, 6, 7, and 10. Flights 5, 6 and 7 were all part of a large dust storm that blanketed the region from 5 to 15 April. The meteorological conditions associated with these events are discussed extensively elsewhere [e.g., Tang et al., 2004]. One flight from the Transport and Chemical Evolution over the Pacific (TRACE-P) experiment is included in this analysis to provide a pre-dust-storm contrast to the dust-pollution mixed air masses found in the ACE-Asia Yellow Sea flights. This flight was conducted one month before the ACE mission, prior to the large springtime Asian dust storms.

Figure 3.

ACE-Asia flight tracks and location of highest dust (based on PILS fine Ca2+) encountered by the C-130 during ACE-Asia (see also Figure 2).

3.1. Correlations Between Fine-Particle Soluble Ions

[24] The large amount of data collected with the PILS (a total of 1,923 measurements) makes correlations between fine species a useful method for investigating possible associations between the various ions. A more detailed discussion of correlations for both ACE Asia and TRACE-P PILS data is provided elsewhere [Lee et al., 2003]. Here we summarize the ACE-Asia results focusing on those relevant to the dust chemistry. Highest correlations are found between Cl and Na+ with an r2 of 0.85 (excluding 6 outlying points) and the linear regression slope of 1.04 is somewhat near to that of seawater at 1.16. A high correlation is found between Ca2+ and Mg2+ (r2 = 0.83) and a low correlation between Mg2+ and Na+ (r2 = 0.11) and Ca2+ and Na+ (r2 = 0.04). The mass ratio of fine soluble Mg2+ to Ca2+ ranged from 7 to 13%, as seen in Table 1. For ACE data, the association between SO42− and NH4+ is r2 = 0.88 and NO3 and NH4+r2 = 0.73. A higher correlation exists between NH4+ and the sum of NO3 and SO42− (r2 = 0.93). SO42− and NO3 are also somewhat correlated (r2 = 0.53) suggesting colocated sources.

Table 1. Median Ion Mass and Molar Ratios for Fine Particles (PILS) and All Particles (TAS) Measured at Specified Locations and Altitudesa
 SO42−/Ca2+, g/gMg2+/Ca2+, g/gNO3/SO42−, g/gNH4+/SO42−, Eq/EqCl/Na+, Eq/EqK+/SO42−, Eq/Eq
  • a

    The TAS measurement sampling intervals are identified in Figures 7, 9, 12, and 15.

  • b

    Yellow Sea, 124.43°–124.97°N longitude, 29.94°–36.08°N latitude, altitudes below 1.15 km asl.

  • c

    Sea of Japan, 133.47°–137.51°N longitude, 33.07°–37.24°N latitude, altitudes above 2.5 km asl.

  • d

    Yellow Sea, 124.25°–124.59°N longitude, 33.07°–37.24°N latitude, altitudes below 1.2 km asl.

  • e

    Yellow Sea, 124.24°–124.41°N longitude, 33.19°–36.86°N latitude, altitudes below 0.50 km asl.

  • f

    Yellow Sea, 124.26°–124.71°N longitude, 33.08°–37.05°N latitude, altitudes below 1.07 km asl.

TRACE-P flight 14b      
ACE flight 5c      
ACE flight 6d      
ACE flight 7e      
ACE flight 10f      

[25] Fine water-soluble K+ is found to be a marker for biomass/biofuel-burning emissions and high correlations between K+ and NH4+ and NO3 were observed in plumes recorded in TRACE-P [Ma et al., 2003]. For ACE data, the correlation (r2) between K+ and NH4+ is 0.47, and K+ and NO3 is 0.64, pointing to biosmoke as a significant source for these species in the ACE regions investigated. For both fine (PILS) and total aerosol particles (TAS) the ratio of K+ to Ca2+ is lowest in the pure dust plume of flight 5, compared to the mixed dust/pollution plumes of flights 6, 7, and 10 (e.g., see average concentrations given in Table 2). This finding is consistent with other dust studies in the region [Zhang et al., 1993]. No correlation is found between K+ and Na+ (r2 = 0.03). Thus most K+ is clearly not of mineral or sea-salt origin.

Table 2. Median and Standard Deviation of Fine-Particle and Total Aerosol Particle Measurements for the Level Legs in the Specified Regions and Altitudesa
 Fine-Particle (PILS) Measurements
  • a

    Concentrations are in μg/m3 (1 atm, 20°C).

  • b

    Yellow Sea, 124.3°–124.5°N longitude, 30.0°–36.8°N latitude, level legs below 1400 m. PILS total measurement time in region: 1 hour 7 min.

  • c

    Sea of Japan, 133.8°–137.1°N latitude, level legs above 2600 m (dust aloft). PILS total measurement time in region: 2 hours 13 min.

  • d

    Yellow Sea, 124.25°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m. PILS total measurement time in region: 1 hour 56 min.

  • e

    Yellow Sea, 124.25°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m. PILS total measurement time in region: 1 hour 57 min.

  • f

    Yellow Sea, 124.21°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m. PILS total measurement time in region: 2 hours 47 min.

  • g

    Sea of Japan, 133.8°–137.1°N latitude, level legs above 2600 m (dust aloft). Two TAS measurements, totaling 1 hour 9 min.

  • h

    Yellow Sea, 124.25°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m. Two TAS measurements, totaling 1 hour 50 min.

  • i

    Yellow Sea, 124.25°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m. One TAS measurement in Yellow Sea, with total time of 1 hour 3 min.

  • j

    Yellow Sea, 124.21°–124.42°N longitude, 32.5°–37.2°N latitude, level legs below 500 m.

TRACE-P flight 14b0.81 (0.39)0.49 (0.63)2.03 (1.26)0.08 (0.06)0.23 (0.10)7.84 (4.31)10.64 (12.14)12.90 (5.14) 
ACE-Asia flight 5c1.47 (1.31)0.14 (0.07)0.05 (0.05)0.16 (0.09)0.08 (0.07)0.04 (0.07)0.25 (0.10)1.06 (0.56) 
ACE-Asia flight 6d3.59 (1.14)0.62 (0.30)0.71 (.32)0.39 (0.09)0.30 (0.11)1.37 (0.45)2.42 (0.98)3.53 (0.90) 
ACE-Asia flight 7e2.34 (0.85)0.66 (0.34)1.01 (0.65)0.17 (0.06)0.26 (0.11)2.35 (0.44)6.32 (2.36)5.77 (0.68) 
ACE-Asia flight 10f1.44 (0.84)0.10 (0.05)0.25 (0.16)0.20 (0.09)0.14 (0.06)0.89 (0.57)0.68 (0.44)3.95 (2.15) 
 Total Aerosol Particle Measurements (TAS, Fine Plus Coarse)
ACE-Asia flight 5g24.88 (3.39)2.65 (1.13)0.35 (0.40)1.79 (0.89)4.80 (1.53)1.43 (0.75)2.30 (0.98)16.49 (2.76)17.69 (2.87)
ACE-Asia flight 6h12.59 (6.25)2.93 (0.67)0.91 (0.48)0.95 (0.31)2.66 (0.82)1.51 (0.70)5.63 (2.53)8.10 (3.76)8.77 (3.97)
ACE-Asia flight 7i8.355.612.070.924.233.0010.0213.7614.82
ACE-Asia flight 10j6.300.350.260.440.821.051.735.836.04

[26] These results are consistent with there being three distinctly different fine-particle sources influencing the measurements of the water-soluble inorganic ions during this experiment: (1) mineral dust composed of calcium (Ca2+), and magnesium (Mg2+), (2) urban pollutants from fossil fuel and biomass combustion that includes, nitrate (NO3), sulfate (SO42−), potassium (K+), and ammonium (NH4+), and (3) sea-salt particles composed mainly of chloride (Cl) and sodium (Na+).

[27] In the following analysis we use ion charge balances within each of the three sources to investigate the extent of interaction of compounds between these various sources. For example, the adsorption of acidic compounds associated with urban pollutants with the alkaline mineral dust would lead to the loss of CO32− and formation of CaSO4, or Ca(NO3)2 (see equations (1) and (3)). Similar reactions may also occur with MgCO3, forming MgSO4, or Mg(NO3)2. If these reactions had occurred to a significant extent, and all particle ionic constituents measured, it would be observed as a slope less than one for CO32− versus (Mg2+ + Ca2+), and a slope significantly above one for (NO3 + SO42−) versus (K+ + NH4+). Similarly, a charge balance between Cl and Na+ compared to that of seawater can provide insight into the extent of Cl depletion due to reactions between Na+ and acidic compounds. Prior to investigating individual plumes, the charge balance including all the data is presented.

3.2. Ion Charge Balance for All ACE-Asia Data

[28] Calculating the charge balance of cations and anions can give clues regarding chemical reactions between acidic gases and the mineral aerosol and, with certain assumptions, provide an estimate for the completeness of the measured aerosol ionic constituents. When all ions in the ambient aerosol particle are measured, anions and cations, in units of equivalents, are equal (i.e., balance). For acidic aerosol particles, high sulfate and nitrate values that are not balanced by ammonium, or another cation, is expected since H+ is not measured. Conversely, an excess of cations due to a missing anion is also possible and was commonly observed during ACE Asia dust events. Figure 4a shows the imbalance between cations and anions for all fine-particle data. Clearly, excess cations are associated with highest fine water soluble Ca2+. The missing anion is likely carbonate (CO32−), such that a large fraction of the fine water-soluble Ca2+ measured is calcium carbonate (CaCO3), a common component of Asian dust [Nishikawa et al., 2000]. Figure 4a also shows that the balance is acidic (less than 1:1) when Ca2+ concentrations are low. This occurs in regions of high sulfate concentrations (not indicated on the graph).

Figure 4.

Ion charge balance for the measured fine (PILS) inorganic compounds. Cations, Ca2+ + Mg2+ + K+ + Na+ + NH4+; anions, Cl + NO3−+ SO42−. Figures 4a and 4b show that the excess cation charge is associated with Ca2+concentrations and apparently due to the lack of a CO32− measurement. In Figure 4a, higher Ca2+concentration is indicated by lighter color and larger data points. In Figure 4b the cations minus anions essentially equals the total CO32− concentration if there is no bisulfate, and all major ion pairs are measured. The linear regression shown in Figure 4b is based on N points, has an intercept of I, and a slope of S.

[29] To further test the association between the measured ion charge balance and mineral dust, the measured ion charge balance (sum of all measured cation equivalents minus the sum of all measured anion equivalents), is plotted in Figure 4b against the sum of Mg2+ and Ca2+. If all the major aerosol particle ions are measured, the ion charges from the cations should balance the anion charge. For example, if there is no bisulfate (a reasonable assumption under alkaline conditions [Kim and Seinfeld, 1993a, 1993b]), then Ca2+ + Mg2+ + Na+ + K+ + NH4+ = CO32− + Cl + SO42− + NO3, and the measured ion balance from cations minus anions equals the CO32− that was not measured. (Note that CO32− is the total carbonate, the sum of carbonate and bicarbonate.)

[30] We also are assuming that the contributions of other ions, like organic ions, which were not measured, are at low concentrations. This appears to be a reasonable assumption on the basis of other airborne measurements made during ACE-Asia. Mader et al. [2004] found oxalate concentrations the highest of all organic acids measured, and the combined mass of NO3 and SO42− was factors of 9–17 that of the identified organic ions. Oxalate was measured with a micro-orifice impactor over extended integration times on the C-130 during many of the level flight legs discussed in the following sections [see Kline et al., 2004]. For these data, the equivalence ratio of oxalate to NO3 + SO42− is only 0.5% to 1.5%.

[31] On the basis of these assumptions we infer the total CO32− equivalence concentration from

equation image

Estimating that for a single IC sample liquid injection each measured ion has a relative uncertainty of ∼5% (based on the precision experiment of Orsini et al. [2003]) and that the errors are random and independent, the overall quadrature sum of squares CO32− uncertainty is ∼15%. (Note that the larger, possibly systematic, NH4+ uncertainty does not significantly influence our calculations of the mineral dust charge balance since in this case NH4+ is a minor cation.) If the mineral dust, Mg2+ and Ca2+, are only associated with CO32− (e.g., MgCO3 and CaCO3), then a plot of the measured ion balance versus Mg2+ + Ca2+ should be highly correlated and have a slope of 1 and zero intercept. Figure 4b shows the result of a linear regression for all data in which Mg2+ + Ca2+ is greater than 50 nEq/m3. The lower dust concentrations in the fit are ignored since the aerosol is apparently acidic at these times and our assumption of ignoring the H+ ion is not valid. The regression result relating inferred total CO32− to Mg2+ + Ca2+ has a slope of 1.00 and intercept of −27 nEq/m3. The mean of the equivalence ratio of the measured ion balance to the sum of Ca2+ and Mg2+ is 0.77. These results suggest that much of the mineral dust encountered by the C-130 during ACE-Asia (especially for highest dust concentrations) was in the form of MgCO3 and CaCO3. However, given the unusually high dust concentrations in these large storms, this analysis does not indicate that the dust had little effect on the physical-chemical properties of the urban pollutant, just that the fraction of reacted dust was likely small.

[32] Results from four ACE-Asia individual flights, identified in Figure 3, are now described in detail to investigate interactions of mineral dust with the pollution from large Asian urban regions. Of these four flights, three involved measurements from an unusually large Asian dust storm. To place these events in perspective, measurements in the same location over the Yellow Sea during TRACE-P, prior to the dust outbreaks, are first discussed.

3.3. Case Studies

3.3.1. TRACE-P Flight 14: Pollution Over the Yellow Sea With Little Mineral Dust

[33] One month prior to the large dust outbreak of 5–15 April recorded during ACE-Asia and discussed in the next section, measurements of a large urban/industrial air mass were recorded over large regions of the Yellow Sea during the NASA TRACE-P mission on 18 March 2001 (flight 14). The TRACE-P P3-B aircraft deployed a duplicate PILS instrument as well as gas phase measurements. This particular air mass was unique in that it contained the highest CO, and fine-particle K+, NH4+, NO3, and SO42− of both ACE-Asia and TRACE-P missions.

[34] Backward air mass trajectories [see Ma et al., 2003] indicate the polluted air mass likely originated from the Beijing and Tianjin urban regions. Two passes were made in the boundary layer at different locations and are shown in Figure 5. Combined, they indicate that the Yellow Sea boundary layer between 30° and 36°N latitude was highly polluted. SO2 values peak at ∼10 ppbv for altitudes below 1 km asl, during passage through the region (Figure 5a). In Figure 5b, the concentrations of the cations are stacked (summed), and plotted opposite the negative of the stacked anion concentrations. The sum of cations minus anions is shown as the net charge. Average concentrations, including both passes, are summarized in Table 2.

Figure 5.

(a and b) TRACE-P flight 14 measurements of fine water-soluble ions over the Yellow Sea on 18 March 2001. The inorganic composition of the plume is mainly nitrate, sulfate, and ammonium.

[35] The region contains significant nitrate and the median equivalence ratio of NO3/SO42− is 2.4. The air mass was ammonium rich; the median NH4+/SO42− equivalence ratio is 1.5 for both the surface level legs (Table 1), and 2.0 for the first leg with the higher NO3 concentrations. The equivalence ratios indicate that there is more than sufficient NH4+ to completely neutralize the SO42− (i.e., form (NH4)2SO4). The Deming linear regression slope between the major acidic (NO3 + SO42−) and basic compounds (NH4+) is 1.14 ± 0.17 (uncertainty is due mainly to that of NH4+). If K+ is included in the regression between pollution species (NO3 + SO42−) and (NH4+ + K+), the Deming linear regression slope is 1.01 ± 0.13 (see Figure 6). Figures 5 and 6 demonstrate that both NH4NO3 and (NH4)2SO4 salts are likely the major ionic compounds in this plume. The aerosol composition when large dust events blanketed the region is now investigated.

Figure 6.

Balance between the measured main anthropogenic anions and cations. Deming regression intercept I and slope S are given in the plot. Data point error bars are based on an estimate of the relative uncertainty of 5% for each measured ion for a single ion chromatograph injection.

3.3.2. Meteorological Aspects of the Large Dust Outbreak of 5–15 April

[36] During the ACE Asia study, an unusually large dust storm developed over central Asia because of a midlatitude cyclone that produced strong surface winds over arid regions near approximately 45°N latitude. In the regions investigated by the aircraft during this period, dust storms persisted from 5 to 15 April, and the first encounters with dust from the storm were on flight 5, on 8 April 2001, over the Sea of Japan (see Figure 3). The dust had been transported eastward and lofted by frontal activity and had not mixed extensively with surface-level pollution below. However, as the dust storm evolved, postfrontal activity mixed the dust down to the surface where it could be blended with anthropogenic emissions. Examples of this were observed in the two subsequent C-130 flights over the Yellow Sea on 11 April 2001, flight 6, and the following day, 12 April 2001, flight 7.

[37] At a later date, 18 April 2001, during flight 10, another dust encounter was recorded over the Yellow Sea. In this case backward air trajectories indicated that we intercepted a plume that had come from a more southerly region, and had spent some time over the ocean. These four flights are investigated.

3.3.3. Flight 5, Dust Over a Pollution Layer: Beginning of the Dust Event

[38] Flight 5 was conducted on 6 April 2001, over the Sea of Japan and was the first interception of the large dust event. Figure 7a shows fine-particle ion concentrations, and Figure 7b gives the measured charge balance. The bars at the bottom of the graph indicate the total aerosol sampler (TAS) integration periods. The aircraft altitude alternated between several level legs from the surface up to 5 km asl (Figure 7c). The several soundings between legs identified a temperature inversion near 2.5 km asl (Figure 7d) that separates the dust layer aloft from a sulfate-rich surface layer. The median concentrations of the various measured fine and total aerosol inorganic compounds are given in Table 2. The median total water-soluble particulate Ca2+ was 24.88 ± 3.39 μg/m3, based on the two TAS measurements (one of the TAS measurements is the high-concentration outlier in Figure 2).

Figure 7.

(a–d) ACE-Asia flight 5 measurements over the Sea of Japan on 6 April 2001, the first interception of the large dust event. Comparisons of fine water-soluble calcium and sulfate concentrations to altitude show a dust layer above and pollution layer below a temperature inversion at 2.5 km above sea level. The ion balance (Figure 7b) (cations minus anions) indicates an alkaline dust aerosol and slightly acidic pollution aerosol. The median total mineral dust Ca2+concentration measured by the TAS in this region is 25 μg/m3.

[39] The fine-particle measured ion balance, Figure 7b, shows excess cations associated with the dust, suggestive of an alkaline CaCO3 aerosol and a more neutral, or possibly slightly acid aerosol, associated with the sulfate separated from the dust below the inversion. Inspection of the periods when sampling below 1 km asl, in what is apparently polluted air, shows that fine NO3 concentrations are largest when the ion balance is near zero, suggesting a neutral aerosol. For example, for the periods labeled “I” and “II” in Figure 7b, the median nitrate concentration is 1.14 and 1.15 μg/m3, respectively, and the NH4+ equivalent concentrations are similar to that of SO42−. In contrast, for the period labeled “III” in Figure 7b the median nitrate concentration is only 0.04 μg/m3 and the NH4+ equivalent concentrations are much less than SO42− suggesting the aerosol particles may be more acidic. These observations are consistent with thermodynamic predictions that nitrate is lost as the aerosol becomes more acidic. We now focus on flights 6, 7, and 10 where mineral dust had moved into the boundary layer and mixed with anthropogenic pollutants from the urbanized regions of northern China.

3.3.4. Mixed Dust and Pollution Over the Yellow Sea, Flight 6, 11 April 2001

[40] Flight 6 was carried out over the Yellow Sea on 11 April 2001. The flight path with selected backward air mass trajectories and relative emission sources of NOx and SO2 are plotted in Figure 8. These trajectories indicate that the air masses had likely passed through the urban regions of Qingdao, Beijing, and Tianjin.

Figure 8.

Flight 6 backward air mass trajectories every 6 min along the C-130 flight path for the northbound Yellow Sea leg A identified in Figure 9. Sources of NOx (squares) and SO2 (triangles), based on emission inventories, are shown, where symbol size is relative to emission strength. Gray diamonds are spaced every 6 hours apart along one air mass backward trajectory that was sampled at 0208 UTC.

[41] During the flight 6 Yellow Sea runs, the aircraft flew on a north-south flight path at 124.3 to 124.4°E longitude from 33°N to 37°N latitude and repeated runs along this path at three different altitudes. The backward trajectories are for the first leg labeled A in Figure 9. This figure shows the time series of aircraft altitude, SO2, and many of the fine-particle concentrations measured during repeated Yellow Sea transits between 33° and 37°N latitude.

Figure 9.

The 11 April 2001, flight 6, PILS water-soluble measurements of fine-particle ionic constituents over the Yellow Sea between the latitudes of 33.3° and 37°N along 124°E longitude (see Figure 8).

[42] From Figure 9, shortly after 0130 UTC, the aircraft descended to the ocean surface at the southern most end of the Yellow Sea track (33°N latitude). SO2, sulfate, nitrate, potassium, ammonium, and calcium concentrations increased then remained elevated throughout this region (leg A in Figure 9). Upon reaching 37°N (0308 UTC) the aircraft climbed to an altitude of ∼5 km while traveling southward, and then descended to 500 m asl at 34.8°N latitude, again traveling northbound repeating a portion of the first leg (this second pass is leg B in Figure 9). A final southbound leg followed, at 300 m asl, retracing the path from 37°N back to 33.3°N (leg C in Figure 9). Although there is considerable variation in fine-particle concentrations when measurements are repeated in similar locations, suggesting concentrations in these plumes are variable, the whole region is a mix of dust and pollution.

[43] TAS measurements of fine plus coarse particles during the intervals indicated in Figure 9 are given in Table 1. Figure 9 shows that fine Ca2+ concentrations in equivalents are more than a factor of 2 larger than SO42− and NO3 and that nitrate and sulfate track during the three level legs (r2 = 0.78) suggesting colocated anthropogenic sources.

[44] Studying the SO2 trends gives an indication of the extent of the mixing of pollution with the dust over this region. The SO2 concentration rose to ∼4 ppbv during descent into the polluted Yellow Sea boundary layer and the beginning of the 300 m asl run of leg A. We then immediately passed through a region of high SO2, where peak concentrations of 6 ppb were recorded. The backward trajectories in this high SO2 plume measured at ∼0208 UTC are plotted in Figure 8 and suggest that it may have originated from Beijing and Tianjin (in Figures 9 and 10 we label it as the Beijing plume). The SO2 data plotted are of high resolution (1 Hz), and in the Beijing plume SO2 is fairly uniform suggesting that the plume was uniform and well mixed where we intercepted it. Fine-particle SO42−, NO3, and NH4+ are also higher in this plume (see Figures 9 and 10), and Ca2+ is only slightly lower indicating that although the urban plume is well defined, it is well mixed with the mineral dust. For the rest of the level run of leg A, the SO2 was fairly uniform and lower at ∼4 ppb suggesting a more regionally polluted air mass and apparently not as clearly associated with as strong a pollution source.

Figure 10.

Ion balances based on Deming least squares fits for the three major types of aerosol sources for data collected during the three Yellow Sea boundary layer legs (legs A, B, and C) identified in Figure 9. In each plot, I is the intercept, and S is the slope. All data points are included in the fit, except for Cl versus Na+, where the Beijing plume data have been excluded.

[45] To investigate interactions between the various aerosol sources, ion balances based on linear regressions are plotted in Figure 10 for the three types of sources: sea salt, mineral dust, and urban pollutants. The regression fit in the plot of Cl versus Na+ does not include the points identified as the Beijing plume since the measured Cl at these times is significantly higher than typical. This could be due to anthropogenic sources of Cl and/or interferences from organic acids that could add to the Cl peak in the anion chromatogram. In any case, Cl and Na+ are correlated (r2 = 0.82), and because the slope is not significantly less than that of sea salt, the data do not indicate chloride depletion through reactions of sea salt with acidic compounds. The mineral dust also appears to be mainly unreacted. The inferred fine CO32− (equation (5)) versus Mg2+ + Ca2+ are correlated (r2 = 0.94); however, the slope is ∼5% below unity and the intercept is not zero, possibly suggesting some interaction with other ions.

[46] For the fine-particle ionic constituents from the urban regions the slope of (NO3 + SO42−) versus (K+ + NH4+) is 1.22 ± 0.16. (Note that the slope uncertainty is due mainly to the 15% uncertainty in NH4+ and is estimated by performing the linear regressions with and without this possible bias.) The greater than unity slope indicates that more NO3 and SO42− are present than what is needed to neutralize the cations K+ and NH4+. These excess anions could be associated with the mineral dust, consistent with the minor depletion of CO32−, or associated with sea salt. (Note that the reported SO42− is essentially non-sea-salt since we are measuring fine particles. For example, in the leg A boundary layer measurement over the Yellow Sea, the sulfate associated with sea salt, based on measured Na+, is less than 1% of the measured SO42−.)

[47] Throughout the boundary layer legs identified in Figure 9, SO42− and NH4+ are often equal and track, suggesting (NH4)2SO4 may be a common salt. For the period of the first TAS measurement (this does not include the Beijing plume), fine-particle NH4+-SO42−r2 = 0.72 and the NH4+/SO42− equivalence ratio, based on linear regression, is 1.2. Size-resolved composition measurements with a micro-orifice impactor for this same period, indicate that the NH4+ and SO42− have similar size distributions, consistent with the presence of fine-particle ammonium sulfate salts [Kline et al., 2004].

[48] The bulk fine-particle equivalents ratio suggests that NH4+ is greater than SO42− indicating that NH4+ is also possibly associated with other anions, although to a much lesser extent. This is even more obvious in the region identified as the Beijing plume in Figure 9, where NH4+ equivalence concentrations are clearly greater than SO42−. The NH4+ in excess of SO42− could be associated with NO3, similar to the earlier observations of possible fine-particle ammonium nitrate salts in the urban plumes prior to the dust outbreaks (Figure 5). Overall, these data suggest, but do not prove, that for the dusty air mass that mixed with urban pollutants 24 to 30 hours prior to the measurement (based on the backward air mass trajectories, see Figure 8), the mineral dust was mainly CaCO3 and MgCO3 and the urban pollutants were likely mainly some combination of the following salts; (NH4)2SO4, NH4NO3, K2SO4 and KNO3. There is evidence that a smaller fraction of the SO42− and NO3 could also be associated with the mineral dust.

3.3.5. Mixed Dust and Pollution Over the Yellow Sea, Flight 7, 12 April 2001

[49] On the day following flight 6, a large dust region was again encountered over the Yellow Sea. Of the mixed aerosol plumes, this flight has the highest influence of pollution relative to dust, demonstrated by the higher fine- and total particle SO42−/Ca2+ ratios (see Table 1, and also see Figure 12 to compare SO42− and Ca2+). The flight path, backward trajectories, and magnitude of NOx and SO2 point sources based on emission inventories are shown in Figure 11. In this case the back trajectories are from a more westerly direction than the previous day and appear to pick up pollutants from more southerly located urban regions. Also, on the basis of back trajectories, the age of the urban emissions when mixed with dust appear to be only 12 to 18 hours old versus 24 to 30 hours for flight 6.

Figure 11.

Flight 7 backward air mass trajectories, similar to Figure 8. The various trajectories are at 6 min measurement time intervals and span the three legs shown in Figure 12: Leg A, 0316–0416 UTC, which covers the complete north-south range of the Yellow Sea measurement, is shown in black, and legs B and C, 0543 to 0631 UTC, are shown in blue. The red trajectory with diamonds, indicating 6-hour intervals along the trajectory, is for the measurement at 0607 UTC, which is apparently from a fresh plume.

[50] A plot similar to that of flight 6 for the various species is given in Figure 12 and only includes the period of sampling along the nearly constant longitude Yellow Sea runs (124.22°–124.43°E longitude). As indicated in Figure 12, this flight involved a northbound leg near the sea surface (leg A). The aircraft then returned to the southern most region of the Yellow Sea at ∼4 km asl. This was followed by a shorter northbound run (leg B), this time only up to a latitude of 35°N, and then turning around and heading south for leg C, ending up near where leg A started.

Figure 12.

(a–c) Flight 7 measurements over the Yellow Sea on 12 April 2001.

[51] In the surface level legs of A, B, and C, SO2 concentrations are consistently above 4 ppbv, but peak above 18 ppbv at ∼0600 UTC in what is apparently a plume from Zaozhuang, Jinan, and/or Qingdao (it is labeled as the Zaozhuang plume in Figure 12), based on the back trajectories of Figure 11. Fine water-soluble Ca2+ is nearly half that of the day before, whereas the SO42− and NO3 concentrations are similar or higher (see also Table 2). Potassium concentrations are also somewhat higher, possibly suggesting a greater influence from biomass-burning smoke.

[52] The ion charge balances for the three types of sources, shown in Figure 13, are similar to those of flight 6, except that the correlation between anions and cations is better and the linear regression is closer to 1 for the urban pollutants, suggesting less interaction between pollution and dust for these two sources compared to what was observed on flight 6 the day before. This is consistent with the shorter interaction time between the dust and pollutants (12–18 hours in this case versus 24–30 hours in flight 6), based on the time the air mass advected from urban regions. For the total aerosol (TAS measurement) the cation to anion equivalence ratio is 1.43, larger than one and consistent with significant amounts of CO32−. In Figure 13, the particle ClNa+ molar ratio is significantly higher than sea salt, and again may be due to the light organic acids being included with the Cl measurement or anthropogenic Cl sources.

Figure 13.

Ion balances based on Deming least squares fits for the three major types of aerosol sources for data collected in the three Yellow Sea boundary layer (legs A, B, and C) identified in Figure 12. In each plot, I is the intercept, and S is the slope. All data points are included in the fit, except for Cl versus Na+, where the Zaozhuang plume data have been excluded for the regression.

[53] For the level surface leg A identified in Figure 12, NH4+ and SO42− concentrations are somewhat correlated and of similar equivalence concentrations (r2 = 0.56, NH4+ versus SO42− slope is 0.92). However, in the stronger urban plume, identified as the Zaozhuang Plume in Figure 12, NH4+ greatly exceeds SO42− (equivalence concentration units). For these data the Deming regression r2 = 0.81, and NH4+ versus SO42− slope equals 2.7. NO3 and SO42− concentrations are high in this plume. The data again suggest that both fine-particle ammonium nitrate and sulfate salts are likely present in the plume along with the mineral dust that apparently is mainly CaCO3 and MgCO3.

3.3.6. Mixed Dust and Pollution Over the Yellow Sea, Flight 10, 18 April 2001

[54] Six days following the large dust outbreak of 5–15 April, flight 10 returned the C-130 to the Yellow Sea. The flight path, backward air mass trajectories for selected times, and NOx and SO2 point sources are shown in Figure 14. Unlike flights 6 and 7 where the intercepted air masses had advected directly off the eastern China coast spending roughly 12 hours over the sea, most of these air masses have spent approximately 2 to 5 days over the ocean.

Figure 14.

Flight 10 air mass back trajectories, similar to Figure 8. Consecutive trajectories are shown at 6-min intervals for legs A (0201–0258 UTC) and B (0334–0440 UTC), shown in Figure 12. Gray diamonds show the 6-hour intervals back in time along two trajectories; the most northerly one is the apparently fresh plume recorded at 0258 UTC, and other is during the middle of leg B at 0410 UTC. In the latter case, leg B, 0410 UTC, the air mass spent approximately 2 days over the ocean prior to the measurement. For leg A, 0201–0231 UTC, which looped around south of the measurements, the air mass spent approximately 5 days over the ocean.

[55] The data for only the north-south runs over the Yellow Sea in the longitude range 124.21° to 124.42°N are plotted in Figure 15. The aircraft made only one north-southbound Yellow Sea transit between 32.5° and 37.2°N latitude during this flight. During the higher-altitude sampling time between approximately 0300 and 0330 UTC the aircraft circled at the most northerly end of the Yellow Sea, then headed south on leg B and then departed the region.

Figure 15.

Flight 10 measurements over the Yellow Sea on 18 April 2001.

[56] A unique feature of this flight is that now fine-particle NO3 is low throughout the flight at 0.68 ± 0.44 μg/m3 compared to 2.42 ± 0.98 μg/m3 for flight 6 and 6.32 ± 2.36 μg/m3 for flight 7 (mean ± the standard deviation). Total NO3 (TAS) is also lower at 1.73 μg/m3 compared to 5.63 μg/m3 for flight 6 and 10.02 μg/m3 for flight 7. The concentrations of fine-particle K+ and NH4+ are also much lower; however, fine SO42− tends to be of similar magnitude. The lower concentrations of fine-particle K+, NO3, and NH4+ in this air mass compared to those of flights 6 and 7 are consistent with weaker biomass-burning emissions.

[57] On the basis of the back trajectories for this flight, the SO2 peak observed between 0.6 and 2.2 km asl at ∼0250 to 0258 UTC is possibly from Jinan and/or Qingdao (it is labeled as Qingdao in Figure 15, and the back trajectory is shown in Figure 14). This plume has high SO2, and fine-particle SO42− but only a relatively small enhancement in fine-particle NO3.

[58] For these Yellow Sea measurements, the ion balances are plotted in Figure 16. The Cl-Na+ correlation and molar ratios are much different than before. Now there is little correlation between these ions, and Cl concentrations are significantly less than the Na+, both observations are consistent with chloride depletion by reaction of the salts with acidic compounds [Andreae et al., 1986; ten Brink, 1998; Song and Carmichael, 2001; Zhuang et al., 1999]. A similar result is found for total aerosol concentrations (TAS): Cl/Na+ equivalents ratios are much lower in this case than in flights 6 and 7 (see Table 1).

Figure 16.

Ion balances for the three sources, sea salt, mineral dust, and urban pollutants, for the legs identified in Figure 15. The Deming regressions are included on the data plotted.

[59] Both fine-particle ion balances for mineral dust (CO32− versus Mg2+ + Ca2+) and pollution components (NO3 + SO42− versus K+ + NH4+) show evidence for interaction between the dust and pollution since the slopes generally differ from one and the intercepts are nonzero. For example, the ion balance on the urban pollutants has a slope significantly higher than 1, because of SO42− concentrations exceeding NH4+ (NO3 and K+ concentrations are low, see Figure 15). Given that the Na+ concentrations are approximately an order of magnitude lower than both SO42− and Ca2+, a small fraction of this excess SO42− may be associated with the Na+, but most could be associated with the mineral dust. Note that the concentrations of the cation mineral dust components exceed inferred CO32− because of a combination of slope less than one and large negative intercept. This is also consistent with interaction between dust and pollution resulting in the loss of CO32−.

4. Discussion

[60] The range of measurements from both TRACE-P and ACE-Asia provide snapshots of various mixtures of mineral dust and emissions from large Asian cities and surrounding regions during the spring of 2001. The apparent prevalence of nitrate and sulfate salts of ammonium and potassium, and lack of extensive interaction between acidic compounds with the alkaline mineral dust to form calcium (and lesser extent magnesium) nitrate and sulfate salts is of interest. This suggests that coagulation of urban particulates containing, for example, ammonium nitrate and sulfate salts, with fine-particle mineral dust is minimal, and that the uptake of acidic gases (SO2, HNO3, etc.) by the mineral dust was relatively slow.

[61] To investigate the uptake of acidic gases, we focus on the large dust event of flight 6 during the time period identified as the first TAS sampling interval in Figure 11. For a mass accommodation coefficient of 0.1 (i.e., a wet dust particle) and HNO3 diffusivity of 0.1 cm2 s−1, the characteristic HNO3 scavenging time by the fine-particle mineral dust (Dp < 2 μm) is ∼4 min, based on the measured dust size distribution (A. D. Clarke, personal communication, 2003) and a mass transfer rate calculated from the Fuchs and Sutugin [1970] expression. (Note that the characteristic time for uptake by the course mode is 7 min, and for the complete measured distribution ∼3 min.) The characteristic lifetime for an accommodation coefficient of 10−2, 10−3, and 10−4 is 44 min, 7 hours, and ∼3 days, respectively. For this data, back trajectories indicate that the dust had passed Beijing and surrounding urban regions 24 to 30 hours prior to the measurement. However, the observations, based on the linear regression slope (Figure 10), suggest only a minor fraction (on the order of 20%) of the fine-particle nitrate and/or sulfate was not associated with potassium or ammonium. During flight 7, where the dust plume was only 12 to 18 hours out of the urban center, there is evidence that an even smaller fraction of the nitrate and sulfate were associated with the mineral dust (Figure 13).

[62] These results may be due to the details of the mixing of the dust and urban emissions. For example, if the sources for the ions K+, NH4+, NO3, and SO42− were in close proximity, some combination of these salts (e.g., (NH4)2SO4, NH4NO3, K2SO4, and KNO3) could have formed rapidly, before significant interaction occurred with the mineral dust. However, any remaining reactive gases (e.g., SO2) or subsequent acidic gas production (e.g., HNO3) during transport out of the urban region should have readily reacted with the dust, if the mass accommodation coefficient is near 0.1, the value assumed if the dust is wet. Figures 9, 12, and 15 show the high SO2 concentrations in these plumes. No HNO3 measurements were made on the C-130 as part of ACE-Asia; however, at this rate of uptake, gas phase HNO3 concentrations should be near zero. Nitric acid was measured during TRACE P on the NASA DC-8, along with integrated filter measurements of aerosol ionic compounds associated with the total aerosol (fine plus coarse) [Dibb et al., 2003; Talbot et al., 2003]. Measurements over the Yellow Sea (flight 13) on 21 March 2001 in a mixed polluted/mineral dust plume recorded peak HNO3 concentrations of 3.7 ppb and total Ca2+ concentrations of 571 μEq/m3 (11,450 μg/m3). On the basis of an ion balance estimate of CO32−, the equivalence ratio of CO32−/(Mg2+ + Ca2+) is 0.83 suggesting that ∼80% of the mineral dust remained as carbonate. This plume was apparently transported from the Shanghai metropolitan region to the Yellow Sea within a day [Dibb et al., 2003; Fuelberg et al., 2003; Talbot et al., 2003]. These results are consistent with a slow initial reaction between the reactive acidic pollutants with Asian mineral dust.

[63] The observations suggest that mass accommodation coefficients for HNO3 uptake by the mineral dust are a few orders of magnitude below 1. This may be a result of the complex dependence of the mass accommodation coefficient on the amount of condensed water associated with the dust. To explore this, we estimate the fraction of carbonate replaced (FCR),

equation image

where CO32−(lost) is the difference in the carbonate associated with Ca2+ and Mg2+ (equals Ca2+ + Mg2+) minus the inferred CO32− from equation (5). Focusing on the Yellow Sea legs A and C of flight 6, since they have the greatest variability in RH (see Figure 9), Figure 17 compares the FCR to RH. The results suggest that there may be some evidence for an increasing loss of carbonate (e.g., reaction of acidic compounds with mineral dust), with increasing relative humidity. Note that in flight 7 the RH in the boundary layer runs is actually higher than in flight 6, however, there is little evidence for fine-particle loss of CO32− from the mineral dust. Compared to flight 6, in this case the time for advection of the dust/pollution plume from the urban regions to the point of measurement was less, and the plume had apparently also spent less time over the Yellow Sea where the RH is likely to be high. The accommodation coefficient's specific RH-dependence will likely differ for the various acidic compounds condensing. This, however, cannot be explored in more detail with this data set. Box modeling studies testing the sensitivity of mass uptake of the urban pollutants to parameters, such as the amount of water associated with the mineral dust, and constrained by these observations, may provide further insights into these processes.

Figure 17.

Fraction of carbonate replaced (FCR; see equation (6)), estimated from the measured fine-particle ion balance, as a function of ambient relative humidity for two legs of flight 6 (see Figure 9). Three data points are not included in the linear regression for leg C; two points had FCR less than zero, and one had a FCR of 0.35 that did not follow the temporal trend during the leg. Including all the points for leg C, r2 = 0.26. Legs A and C regressed together produce r2 = 0.31.

5. Summary

[64] The characteristics and possible interactions between anthropogenic pollutants from large Chinese urban regions with mineral dust are investigated. The airborne data are based mainly on ion chromatography analysis of the water-soluble ionic constituents of fine ambient aerosol particles collect and measured online with the particle-into-liquid sampler (PILS). The following ions were measured; Ca2+, Mg2+, K+, Na+, NH4+, SO42−, NO3, and Cl. In the initial stages of a large dust outbreak originating over the Gobi Desert, dust-laden air masses are observed over the Sea of Japan residing over a pollution layer, separated by a temperature inversion. The water-soluble component of the fine-particle mineral dust is composed mainly of Ca2+ and to a lesser extent, Mg2+, with Mg2+ typically 7 to 13% of Ca2+, in terms of mass. For air masses with large dust concentrations, ion charge balance calculations indicate that the water-soluble Mg2+ and Ca2+ is apparently associated with an unmeasured anion, likely CO32− (e.g., carbonate plus bicarbonate).

[65] In subsequent measurements following the passage of a frontal system, the mineral dust moved into the surface layers and passed through urban regions. This mix of dust and pollutants was subsequently measured on two consecutive days over the Yellow Sea marine boundary layer. High correlations and near-unity equivalence concentrations between the inferred mineral dust CO32− and Mg2+ + Ca2+ concentrations suggest that a large fraction of the fine mineral dust remained unreacted and alkaline. Ion balances amongst the fine-particle inorganic constituents emitted by the urban regions also suggested that much of the anthropogenic acid compounds were mostly associated with NH4+ and K+. In mixed pollution-dust air masses, 12 to 18 hours advection time from the urban centers, the equivalence ratio between (NO3 + SO42−) and (K+ + NH4+) based on a Deming linear regression slope was 1.03 ± 0.11. For a mixed air mass 24 to 30 hours out of an urban region the slope was 1.22 ± 0.16, suggesting some fraction of the nitrate and sulfate were associated with other cations, possibly Mg2+ and Ca2+.

[66] On the basis of measured particle size distributions in one of the mixed dust-pollution plumes, with an assumed mass accommodation coefficient of 0.1, HNO3 is predicted to have a characteristic lifetime of ∼5 min due to scavenging by the fine-particle alkaline mineral dust. This contrast to the observations may suggest that the initial mass accommodation coefficients for acidic gas (e.g., SO2, HNO3) reactions with mineral dust are orders of magnitude below 1. For one flight over the Yellow Sea there is some evidence suggesting that the fraction of mineral dust carbonate replaced by reaction with acidic compounds depends on the ambient relative humidity.

[67] In these mineral dust–urban mixed plumes, fine-particle NH4+ and SO42− were ubiquitous and spatially correlated and often had equivalence ratios near 1, suggesting that fine-particle ammonium sulfate salts were also found mixed with the dust. In the strongest urban plumes, identified from backward air mass trajectories intersecting urban centers, and peaks in SO2 concentrations, NH4+ concentrations often exceeded that needed to completely neutralize the SO42−. Concurrent high concentrations of fine-particle NO3 points to (but does not prove), the presence of ammonium nitrate salts also mixed with the alkaline mineral dust. Our measurements over the same region, as part of TRACE-P, one month prior to the large dust outbreaks, strongly indicated that the fine-particle ionic composition was mainly ammonium nitrate and sulfate salts. Enhanced levels of fine particulate K+ in most of the plumes emanating from the region between Beijing and Zaozhuang recorded during ACE Asia suggest a significant and widespread influence from biomass smoke (biomass or biofuel burning) during the measurement period. This may have been a major source for the observed fine-particle NO3 [Ma et al., 2003].

[68] For mixed pollution and dust plumes that had spent 2 to 5 days over the Yellow Sea prior to the measurement, there was evidence for extensive Cl depletion based on a ClNa+ slope of 0.58 and poor correlation (r2 = 0.33). The other plumes that had spent roughly 12 hours over the Sea showed no clear evidence for chloride loss. Because of low NO3 concentrations over this region, and SO42− in excess of NH4+, most of the Cl may have been displaced by SO42− to form Na2SO4. There was also some evidence that the SO42− may have reacted with the mineral dust based on a CO32− to (Mg2+ + Ca2+) slope approximately 5% below 1 and intercept indicating nonzero Mg2+ + Ca2+ at zero CO32+ concentrations.

[69] Evidence for limited interaction between the fine-particle mineral dust and acidic compounds could be due in part to the fact that the plumes had only recently been in contact. It is also noted that interactions between acid compounds and coarse mineral dust, which was not investigated in detail in this paper, may be substantially different than our observations which are based on fine dust particles. Other studies, possibly of more aged plumes, that include coarse particles have found evidence for NO3 and SO42− associated with dust particles [e.g., Wu and Okada, 1994; Choi et al., 2001; Gao and Anderson, 2001; Kim and Park, 2001; Jordan et al., 2003]. The data presented here could be useful to constrain model simulations aimed at investigating the processes by which urban pollutants react with alkaline mineral dust particles.


[70] The authors gratefully acknowledge the financial support from the National Science Foundation for support of ACE-Asia under grant ATM 0080471, the National Aeronautics and Space Administration GTE program for support of TRACE-P through grant NCC-1-411L, and NASA Stennis Space Center for Graduate Student Research Fellowship grant NGT13-52742, which supported graduate work. B. Huebert is acknowledged for the use of the TAS data. A. Bandy is also acknowledged as the source for the SO2 data on both the ACE-Asia and TRACE-P missions.