SEARCH

SEARCH BY CITATION

Keywords:

  • Elemental carbon;
  • black carbon;
  • ACE-Asia;
  • Carbonate carbon;
  • mineral dust

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[1] Airborne levels of carbonaceous aerosols were measured using the Twin Otter aircraft during the Aerosol Characterization Experiment (ACE)-Asia. Particles were collected using a newly developed honeycomb denuder sampler and organic carbon (OC), elemental carbon (EC), and carbonate (CC) carbon levels determined using a thermal–optical carbon analyzer. During some flights, atmospheric layers could be identified as marine boundary, pollution dominated, or mineral dust dominated. Ångstrom exponent (å) values, calculated based on data from an onboard three-wavelength nephelometer, were used to discern the nature of some individual layers. Values of å for individual layers ranged from 0.2 to 2, corresponding to dust- and pollution-dominated layers, respectively. OC and EC concentrations below 3 km ranged from 0.58 to 29 μg C m−3 and from 0.20 to 1.8 μg C m−3, respectively. In general, for a given type of atmospheric layer, higher levels of total carbon (TC) were observed during ACE-Asia than those observed during ACE-2, Tropospheric Aerosol Radiative Forcing Observational Experiment (TARFOX) and Indian Ocean Experiment (INDOEX). Mixed layers of dust and pollution were found on some occasions. CC was detected in samples taken from layers in which å = 1.6, indicating that significant amounts of dust can be present even though å > 0.2. A linear regression of light absorption coefficient σap (Mm−1) versus EC concentration had an r2 of only 0.50, indicating that parameters other than the mass of EC significantly affected the value of σap. The mass absorption coefficient Eabs (m2 g−1) varied by as much as a factor of 8 between sampling events, and the average value of 11 m2 g−1 (±5.0) agrees well with previous published values.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[2] Atmospheric aerosol particles affect the Earth's radiative balance directly by scattering or absorbing light, and indirectly by acting as cloud condensation nuclei (CCN), thereby influencing the albedo and lifetime of clouds. A series of Aerosol Characterization Experiments (ACE) have been conducted which utilize ground, sea and aircraft-based measurements, satellite observations and atmospheric modeling with the goal of improving the prediction of climate forcing due to aerosol particles [Bates et al., 1998; Raes et al., 2000]. The third in this series of experiments is ACE-Asia, the intensive field measurement component of which was conducted from 31 March through 1 May 2001.

[3] One of the principal goals of ACE-Asia was to determine the chemical and physical properties of atmospheric aerosol particles originating from the Asian continent, including the levels of carbonaceous compounds. Carbonaceous material influences the optical properties, atmospheric lifetime, and ability of aerosol particles to act as CCN [Seinfeld and Pandis, 1998]. Particulate carbon is often classified into three broad categories: organic carbon (OC), so-called elemental carbon (EC), and carbonate carbon (CC). OC is carbon associated with organic compounds, either emitted directly to the atmosphere (primary emissions) or formed by the condensation of products formed via the atmospheric oxidation of volatile organic chemicals (secondary OC). Typically OC comprises 10–50% of the mass concentration (μg m−3) of atmospheric aerosols [Seinfeld and Pandis, 1998]. EC is formed during the combustion of hydrocarbons and is a major light absorbing component of the atmospheric aerosols. EC is operationally defined by the analysis method and no EC standard exists. In surveys of sites in the United States it has been observed that EC comprises from 1 to 13%, and from 2 to 9% of the mass concentration of atmospheric aerosol from rural and urban locations, respectively [Shah et al., 1986; Malm et al., 1994]. CC is present in mineral dusts; in most urban situations CC is a minor component of particles of diameters less than 2.5 μm (PM2.5); CC comprising less than 1.4% of the mass of Pasadena PM2.5 [Mueller et al., 1972].

[4] As a component of ACE-Asia, this study had the following goals (1) determine the levels of OC, EC, and CC in individual layers of the atmosphere, (2) characterize the type of layer that was sampled (i.e., marine boundary, pollution-dominated or dust-dominated), and (3) evaluate the relationship between measured aerosol absorption coefficients and measured EC levels.

2. Sampling and Analytical Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[5] Two major challenges exist in the sampling and measurement of aerosol OC and EC. First, many compounds comprising OC are semivolatile, having vapor pressures between 10−11 and 10−4 atm and, as a result, partition between the gas (G) and particle (P) phases in the atmosphere [Pankow, 1994]. It is important to collect atmospheric aerosol in a way that separates gaseous OC from the particle-phase OC without biasing the measurement of the G/P distribution of the OC. The second challenge in measurement of atmospheric OC and EC is determination of the relative amount of total evolved carbon allocated as OC and EC (i.e., the OC/EC split) [Cadle and Mulawa, 1990; Chow et al., 2001] (J. J. Schauer et al., “ACE-Asia intercomparison of a thermal–optical method for the determination of particle-phase organic and elemental carbon,” submitted to Environmental Science and Technology, 2002, hereinafter referred to as Schauer et al., submitted manuscript, 2002).

2.1. Measurement of Semivolatile Organic Compounds

[6] Gas- and particle-phase OC are typically separated using filter samplers. It is well documented that adsorption of gaseous OC to filter surfaces can cause a positive bias in measured OC concentrations (μg C m−3) [McDow and Huntzicker, 1990; Turpin et al., 1994; Mader and Pankow, 2001a, 2001b] (and references therein). When using filter samplers, a common method to correct for positive biases involves the use of a second filter (the so-called backup filter) placed downstream of the front filter. Since the backup filter is exposed only to particle-free air, OC measured on this filter originates only from the G-phase. To correct for positive artifacts, the mass of OC measured on the backup filter is subtracted from the mass observed on the particle-loaded front filter. This correction assumes: (1) the mass/filter area amounts of gaseous-OC adsorbed on the front and backup filters are equal; (2) the gas adsorption capacities of the front and backup filter are equal; and (3) OC observed on the backup filter did not result from evaporation of particles collected on the front filter. Mader and Pankow [2001a, 2001b] have suggested that, since the front filter will tend to reach equilibrium with the incoming gaseous OC first, overall G/filter equilibrium might be achieved only after both filters have reached equilibrium with the gaseous OC in the sample air. If sampling ends before equilibrium is reached on both filters, such backup filter corrections may underestimate the extent of gas adsorption on the front filter, and the mass subtracted from the front filter accumulation will be too small.

[7] Aircraft measurements of OC are often conducted at land-speeds ≈180–540 km h−1; sampling times are often kept short in order to achieve as much spatial resolution between OC measurements as possible. For this reason, measurements of OC made using filter samplers present in aircraft are especially prone to gas adsorption artifacts. To minimize biases resulting from the adsorption of gaseous OC to filters, denuders have been used to first remove gaseous OC from the sample air stream before collecting the particles on a filter [Eatough et al., 1989; Fitz, 1990; Eatough et al., 1993; Cui et al., 1998]. If gaseous OC is removed in a denuder, equilibration of the front filter with the gas phase is not an issue, thus allowing for shorter sampling times while minimizing gas-adsorption artifacts. Previous denuder systems are generally too large to fit into most research aircraft. A high-volume coated honeycomb denuder sampler has recently been reported by Mader et al. [2001] for use in aircraft measurements of OC and EC; versions of this sampler were used to acquire the data in the present work.

2.2. Measurement of OC and EC

[8] The OC/EC split is operationally defined by the analysis method itself [Johnson et al., 1981; Cadle and Mulawa, 1990; Fung, 1990; Novakov et al., 2000a, 2000b; Chow et al., 2001] (Schauer et al., submitted manuscript, 2002). The levels of EC measured for the same sample may vary by as much as an order of magnitude between different analysis methods [Cadle and Mulawa, 1990; Hering et al., 1990; Turpin et al., 1990, 2000]. Furthermore, for a given thermal evolved gas analysis method, the OC/EC split is sensitive to the temperature program used for analysis, and the magnitude of the sensitivity dependent on the types of aerosol particles collected [Chow et al., 2001] (Schauer et al., submitted manuscript, 2002).

[9] Most measurements of OC and EC involve thermal-evolved-gas analysis, where a filter sample is placed into a chamber and heated in a series of temperature steps. During the heating period, gases, such as He or N2, or gas mixtures such as 90%He + 2%O2, are passed through the heating chamber. Compounds evaporated from the sample are carried from the chamber to a catalyst that converts the evolved gases to either CO2 or CH4, and these gases are then quantified using infrared or flame ionization detection, respectively. A major problem with such methods is that; while heating the sample, some components of OC pyrolize or “char” [Yu et al., 2002]. The char might then incorrectly be considered as EC in the original sample. Thermo-optical methods are a common, commercially available method, in which a laser absorbance procedure is used to correct for charring [Johnson et al., 1981; Chow et al., 1993; Birch and Cary, 1996; Chow et al., 2001]. In the thermo-optical method, a sample is heated under a stream of pure He similar to the thermal evolved gas technique, but a laser beam is either reflected or transmitted through the filter with the amount of light absorption measured. As the sample is heated, charring causes the light absorption to rise above the initial value. After reaching the maximum operating temperature (selected by the operator, usually between 550 and 870°C), the temperature is set to 550°C, and the analysis gas switched is to a He/O2 mixture. The sample is heated to near 800°C and all remaining carbon is oxidized to CO or CO2 and evaporates with a concomitant decrease in the light absorption. Carbon evolving from the sample after the light absorption has returned to the initial value is considered EC. The accuracy of the charring correction has been evaluated by analyzing samples containing only OC. For example, sucrose is often used since this compound is known to char. Furthermore, organic aerosol generated in chamber studies can also be used. Since no EC standard exists, it is not currently possible to absolutely determine the accuracy of the charring correction for ambient aerosol samples, nor the accuracy of the EC determination. The precision of an OC/EC method can be tested by analyzing replicate samples using one or more thermal-optical analyzers.

3. Aircraft Sampling Platform and Sample Collection

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[10] Samples were collected onboard a modified De Havilland DHC-6 Twin Otter aircraft operated by the Center for Interdisciplinary Remotely Piloted Aircraft Studies (CIRPAS). A total of 19 Research Flights (RFs) were conducted between 31 March and 1 May 2001. The center of aircraft operations was located at the Marine Corps Air Station (MCAS) Iwakuni, Japan and the sampling area included portions of the Sea of Japan south and east of the Korean Peninsula, the East China Sea between China, Japan and Korea, and the Philippine Sea south of Japan. Depending on the scientific goals of a mission, one or more flight profiles were used: Sampling was conducted over (1) level legs at a particular altitude, (2) spiral ascents and descents between altitudes of 50 m and 3000 m, and (3) level descents from 3000 m to 50 m.

[11] Two inlets were mounted on the Twin Otter; one off the roof of the aircraft and a second mounted to the nose. The nose-mounted inlet was dedicated to sampling carbonaceous aerosols and designed to sample air isokinetically at an airspeed of 50 m s−1. This inlet was constructed entirely of metal and could be removed from the aircraft and cleaned between flights if necessary. A manifold was used to separate the sample air among the three denuder samplers, while maintaining isokinetic flow conditions. The manifold included two impactors to remove cloud droplets from the sample air leading to the low-flow denuder samplers (Samplers A and B in Figure 1). A third impactor was a component of the high-volume particle trap impactor-denuder sampler (Sampler C in Figure 1). In addition to removing cloud droplets, these impactors could possibly remove dust particles. For the particle trap impactor present in Sampler C (Figure 1), the aerodynamic diameter of particles collected with 50% efficiency (d50) was 7 μm [Mader et al., 2001]. The d50 of the impactors present upstream of Samplers A and B was not measured, but as described in an upcoming section of the manuscript, it was determined that particles of 2.3 μm diameter were transmitted through these impactors with 97.5 % efficiency. The compartment housing the denuder samplers was neither heated nor pressurized during flights. The temperature was monitored during flights using a temperature probe located in the samplers flow controller. The inlet and manifold were made entirely of aluminum, the impactors for samplers A and B were made of stainless steel, and the transfer lines connecting the manifold to the denuders samplers were aluminum (sample C) or copper (sampler A and B); no organic materials were in contact with the sample air throughout the entire system.

image

Figure 1. Schematic of the Twin Otter sampling inlet and denuder samplers used for the collection of carbonaceous aerosols during ACE-Asia.

Download figure to PowerPoint

[12] Carbonaceous aerosol particles were collected using the denuder samplers described by Mader et al. [2001]. As shown in Figure 1, the sampling system consisted of three samplers: a pair of low-flow denuder samplers (Samplers A and B) to collect filter samples to be analyzed for OC, EC, and CC using a thermal-optical carbon analyzer, and a single, high-volume particle trap impactor-denuder sampler [Mader et al., 2001] (Sampler C) for collection of samples to be used for the determination of water-soluble organic carbon (WSOC) and individual organic compounds comprising the OC. The low-flow samplers operated at a flow rate of 16 Lpm and consisted of a XAD-4 coated honeycomb denuder placed upstream of a pair of either front and backup quartz fiber filters (QFFs) (4.7 cm diameter Tissuequartz QUO-UP 2500, Pall Gelman, Ann Arbor, MI) or a front QFF and backup carbon impregnated glass fiber filter (CIG). The high-volume particle trap impactor-denuder sampler was operated with a pair of front and backup QFFs (19.4 cm diameter), at a flow rate of 300 Lpm. Flow through each denuder sampler was controlled using individual volumetric flow controllers interfaced to an onboard computer; samplers could be turned on/off at any point during the flight. During air sampling, the volumetric flow, pressure, and temperature of the sample air downstream of the filters were monitored at one-minute intervals. The flow controllers had an accuracy of 0.5% and a precision of 1%.

[13] While in transit to the sampling site, there was no flow through the denuder samplers so as to prevent contamination of the samples. When the samplers were off, solenoid valves closed, isolating the low flow denuders (Samplers A and B in Figure 1) from the pumps, and preventing airflow from transporting gases and particles through the sampler. Samplers A and B could be operated in the two following configurations: (1) Samplers A and B1 configured with a denuder to remove gaseous OC, a filter (QFFf) to collect P-phase OC, EC and CC, and a backup filter (QFFb) to collect OC evaporated from particles collected on QFFf (Note that if the denuder does not remove 100% of G-phase OC, it is possible that these compounds can adsorb to QFFf and QFFb). In this configuration two samples could be collected per flight; it was therefore possible to evaluate carbonaceous aerosol mass concentrations in different layers of the atmosphere, such as the marine boundary layer (MBL) or a mineral dust layer. (2) Sampler A was configured with a denuder to remove gaseous OC, a filter (QFFf,a) to collect P-phase OC, EC, and CC and a backup filter (QFFb,a or CIGb,a) to collect OC evaporated from particles collected on QFFf,a (Note that if the denuder does not remove 100% of G-phase OC, it is possible that these compounds can adsorb to QFFf,a, QFFb,a or CIGb,a). Sampler B2 was configured with a 2.0 μm Teflon membrane filter (TMF) (Zefluor, Pall Gelman Sciences, Ann Arbor MI) to remove P-phase OC, EC and CC, a denuder to remove gaseous OC, and a filter (QFFb,b or CIGb,b) to adsorb gaseous semivolatile OC not collected by the denuder. With this configuration, samplers A and B2 were run in parallel; it was possible to measure the OC, EC, and CC content of a sample (QFFf,a), the amount of OC evaporated from collected particles (QFFb,a or CIGb,a), and the ability of the denuder to remove gaseous OC (QFFb,b or CIGb,b).

[14] At the MCAS-Iwakuni, filters were prepared in a room where access was limited to investigators involved with sampling carbonaceous aerosols. An air purifier containing a HEPA filter and activated carbon filter was used to reduce the levels of volatile organic compounds (VOC) and ambient particles in this room. Prior to sampling, QFF were precleaned by baking at 550°C in a muffle furnace for 12–16 hr. QFF were removed hot from the muffle furnace and immediately loaded into the denuder samplers. After loading a denuder sampler with filters, the inlet and outlet of the sampler were capped with Swagelock™ fittings to prevent airflow through the denuder samplers prior to sampling. The average OC and EC loading on field blank filters was 0.30 and 0.12 μg C cm−2, respectively. Since CIGs adsorb gaseous OC during storage in a freezer, CIGs were precleaned immediately prior to sampling. CIGs were baked in a muffle furnace at 490°C for 16 h under an inert gas of either N2 or He. On a μg C cm−2 basis, the field blank value for CIG on flight 15 was ≈7 μg C cm−2. This value is similar to that of 6 μg C cm−2 calculated for Lewtas et al. [2001] using data given in that reference and equal to the value for CIG prepared in the laboratory at Caltech by Mader et al. [2001]. The field blank value for CIG on flight 19 was considerably higher, ≈28 μg C cm−2, due possibly to the presence of high levels of VOC in the lab, or possibly due to a problem with the precleaning process. If, while baking CIG in a muffle furnace, N2 leaks through an improperly sealed filter holder rather than flowing through the CIG filters, it has been observed in the laboratory that the blank levels of CIGs are high.

[15] TMFs were precleaned by rinsing with 3–100 mL aliquots of acetone and 3–100 mL aliquots of dichloromethane (EM Science, Gibbstown, NJ) and then dried using ultra high purity N2. Immediately prior to sampling, filters were loaded into the respective samplers using stainless steel forceps. Sampling times depended on the particle mass concentrations measured at the sampling site and the overall goal of the particular RF, and ranged from 32 to 223 min. When the aircraft returned from a flight, the samplers were immediately returned to the filter handling room and filters unloaded from the samplers and placed into individual aluminum-lined petri dishes. The aluminum liners had been prebaked under the same conditions as the QFF. For each sampling event, blank QFFs and CIGs (only if CIG were used in the sampling event) were loaded into filter holders, removed, and stored with the other samples. The forceps and aluminum liners had been precleaned by baking at 500°C in a muffle furnace for 12 h. Samples were stored in a freezer until analysis. All samples were analyzed for OC and EC within 48 h of sampling; CIG were analyzed within 8 h of sampling

4. Carbon Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[16] OC, EC, and CC concentrations of the material deposited on the filters were determined using a thermo-optical OC/EC analyzer (Sunset Laboratories, Forest Grove, OR) [Birch and Cary, 1996]. Briefly, a 1.45 cm2 punch of a QFF or CIG filter was loaded into the thermo-optical OC/EC analyzer. For QFF samples, OC and EC were determined as follows: OC was evolved under a stream of ultra-high purity He while heating the sample in four temperature steps of 1 min at 310°C, 1 min at 450°C, 1 min at 575°C, and 1.5 min at 870°C. To evolve EC and pyrolized OC, the sample was heated under a mixture of 10% O2, 90% He in six temperature steps of 0.75 min at 550°C, 0.75 min at 625°C, 0.75 min at 700°C, 0.75 min at 775°C, and 0.75 min at 850°C, and 2.0 min at 900°C.

[17] For QFF samples, CC was determined as follows. After a 1.45 cm2 punch of a QFF was analyzed for OC and EC, a second punch was taken from the same QFF. Using a disposable pipette, 2 drops of 6 M HCl was added to the QFF punch and the sample placed into the OC/EC analyzer. The sample was then heated using the same temperature/gas program used for the OC and EC analysis described previously. Based on experiments in which pure CaCO3 was spiked onto a QFF punch and the sample analyzed using the OC/EC analysis procedure previously described, it was observed that carbonate evolved during the fourth temperature step (870°C) of the thermal/optical analysis. By comparing the thermograms of the acid-treated and untreated QFF punch, the presence of CC could be determined; if CC was present, a peak present in the fourth temperature step of the analysis of an untreated sample was not present in the acid treated sample. Using the OC/EC analyzer software, this peak could be integrated and the mass of CC determined.

[18] CIGs were analyzed as follows: OC was evolved under a stream of ultra-high purity He while heating the sample in five temperature steps of 1 min at 250°C, 1 min at 300°C, 1 min at 350°C, 1 min at 400°C and 0.5 min at 450°C.

5. Instrument Calibration and Performance

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

5.1. Inlet Calibration

[19] It is necessary to determine the particle transmission efficiency of the inlet, manifold, transfer line and samplers used to collect carbonaceous aerosols. Calibrations are typically conducted by preparing aerosols of known diameter using a Vibrating Orifice Aerosol Generator (VOAG, TSI, St. Paul, MN) [Berglund and Liu, 1973; Liu et al., 1974]. Mader et al. [2001] describe a method by which a VOAG is used to generate monodisperse aerosols of dioctyl phthalate (DOP) or dioctyl sebacate (DOS). These aerosols are passed through an impactor and the components of the impactor extracted with solvent; gas chromatography/mass spectrometry (GC/MS) is used to quantify the amount of DOS or DOP in the individual extracts. This method was used in the current study in which particles of diameter 2.3 μm were prepared with a VOAG using a solution of DOS (Fluka, Switzerland) in isopropyl alcohol (EM Science, Gibbstown, NJ). An Aerodyne Aerosol Mass Spectrometer (AMS) [Jayne et al., 2000] was used to confirm the size of particles produced by the VOAG. Briefly the AMS sizes particles by measuring their time of flight between two points and calculating the particle velocity, from which the particle aerodynamic diameter can be calculated. The aerodynamic diameter of these particles determined by the AMS was 2.0 ± 0.5 μm.

[20] DOS aerosol was pulled from the VOAG through the inlet, manifold, and transfer lines and into the three denuder samplers using vacuum pumps operating at the same flow rate as used during airborne sampling. (The XAD-4 coated honeycomb denuders were removed from the sampler so as avoid the possibility of contaminating them with large amounts of DOS.) Particles transmitted through the sampler were collected on a QFF, precleaned by baking at 500°C in a muffle furnace. Aerosol was sampled for 9.3 hours, after which the various components of the system were disassembled and their internal surfaces extracted with three 40 mL aliquots of dichloromethane (EM Science, Gibbstown, NJ). Some components were extracted with an additional three 40 mL aliquots of dichloromethane to determine the relative amount of DOS extracted in the first rinsing. Filters were folded, placed into a glass funnel, and rinsed with 100 mL of dichloromethane. Each extract was spiked with 500 μL of a ≈10 μg μL−1 solution of the internal standard tetracosane, and all extracts stored in a freezer (−20°C) until analysis. Extracts were analyzed using Hewlett Packard G1806 GCD GC/MS (Agilent Technologies, Palo Alto, CA) as described by Mader et al. [2001].

[21] The following components were individually extracted: inlet probe + manifold, impactors A and B, transfer lines A, B, and C, particle trap impactor nozzle + wall (Sampler C), the particle trap impactor (impactor C), wall + filter of samplers A, B and C. The masses of DOS extracted in the first rinsing of the manifold and filters were 92% and 97% of the total extracted mass, respectively. These values were deemed sufficiently close to 100% so that the mass of DOS measured on the various components was not corrected for the extraction efficiency.

[22] It was observed that 85% of 2.3 μm DOS particles that entered the sampling system were collected on the QFFs. The mass of DOS collected on individual QFFs present in the three samplers normalized by the flow rate through each filter was 280, 310 and 290 (ng Lpm−1) for filters present in samplers filters A, B, and C, respectively. These values were within 5% of each other; the overall transmission efficiency of 2.3 μm DOS particles was considered equivalent for the three sample air streams. The transmission efficiencies of the individual components of the inlet and samplers are presented in Table 1. As deduced from these data, 3% of 2.3 μm DOS particles entering the particle trap impactor jet nozzle were lost to the walls of the nozzle. Because DOS particles are a viscous oil, it is likely that they stick to metal surfaces with high efficiency; the transmission efficiencies determined in this experiment are judged to represent a lower bound of the particle transmission of ambient atmospheric aerosol articles.

Table 1. Transmission Efficiency of 2.3-μm DOS Particles Through Individual Components of the Inlet and Denuder Samplers
ComponentTransmission Efficiency, %
Inlet probe and manifold93
Transfer line A98
Transfer line B91
Transfer line C93
Impactor A97
Impactor B98
Particle trap impactor (Sampler C)96
Particle trap impactor nozzle and wall97

5.2. Quality Assurance of OC/EC Analysis Data

[23] The accuracy of the determination of OC was evaluated by analyzing filter samples spiked with a known mass of an organic compound. Sucrose was used as the standard compound since it is known to pyrolize during analysis and can thus be used to test the ability of the analyzer to correct for charring. The OC/EC analyzer was operated in the field during ACE-Asia. Standards were prepared as follows: a prebaked 1.45 cm2 QFF punch was spiked with a precisely measured volume (5 to 25 μL) of a sucrose solution of known concentration (0.5512 or 5.586 μg C μL−1). The punch was then analyzed for OC and EC as described previously. A series of standards were analyzed; the range of filter loadings encompassed the range observed in samples collected during ACE-Asia. The measured carbon loading is shown against the true values in Figure 2. The mean percent difference between measured and true values is −2%. No EC was measured for the sucrose standards, indicating that the instrument properly corrected for the charring of this compound. Since no EC standard exists, it was not possible to determine the accuracy of the EC concentrations measured using this method.

image

Figure 2. Comparison of measured carbon filter loading (μg C cm−2) with true carbon loading.

Download figure to PowerPoint

[24] The precision of the OC and EC analysis method was determined by replicate analysis of samples having the same OC/EC filter loading. These samples were comprised of standards made using sucrose or filter samples obtained during ACE-Asia. At a given filter loading, from three to five replicate analyses were completed and the relative standard deviation calculated. As illustrated in Figure 3, the precision of the OC and EC measurements was a function of the filter's carbon loading. The precision of OC and EC reported were determined using the equation provided in Figure 3.

image

Figure 3. Percent relative standard deviation (RSD) of OC and EC measurement as a function of the OC and EC filter loading on quartz fiber filters.

Download figure to PowerPoint

[25] An inter-laboratory comparison of OC and EC measurements was conducted among eight laboratories operating OC/EC analyzers manufactured by Sunset Laboratories. Blind subsamples from four filters were sent to each participant; the samples were analyzed while the OC/EC analyzers were operating in the United States, China and in the field during ACE-Asia. The results of the intercomparison are described by Schauer et al. (submitted manuscript, 2002). Briefly, the precision of the measurements was a function of the filter's carbon loading; OC and EC filter loadings were within 4 and 13% of the consensus values, respectively.

[26] The method detection limit (MDL) (μg C m−3) was defined as two times the ambient mass concentration of OC or EC measured using a field blank filter. For a given sampling event,

  • equation image

where Mfield blank is the mass of carbon (μg C) measured on the field blank filter and Vs the volume of air that was sampled (m3). Thus the method detection limit decreases as the air sample volume increases. If the measured OC or EC concentration did not exceed the MDL the sample was labeled a “nondetect” and the concentration reported was the MDL. The field blank was determined as follows: A 4.7 cm QFF was precleaned and stored in a petri dish as discussed previously. Prior to the sampling mission or on days without a sampling mission, the precleaned filter was loaded into denuder sampler A or B present in the nose of the aircraft. No air was drawn through the sampler (i.e., the inlet was capped and the solenoid valves downstream of the filter closed). After 15 min to 4 hours, the filter was removed from the sampler and handled identically to a field sample. The average OC and EC loading on field blank filters was 0.30 and 0.12 μg C cm−2, respectively. That EC comprised approximately 30% of total carbon in field blanks was likely due to the fact that the OC/EC split is rather ambiguous at such low carbon loadings, since the initial light transmission through the filter is high and the change in filter transmission virtually nondetectible.

5.3. Denuder Performance

[27] The purpose of the denuder section of the samplers is to remove the gaseous OC that adsorbs to QFF causing a positive bias in the measured P-phase OC concentration. During transport through a denuder, gaseous OC is ad/absorbed to/into the stationary phase coating the denuder. It is possible that while sampling under some conditions, such as high gaseous OC levels, long sampling times and/or high temperatures, the sorption capacity of the denuder could be exceeded. Gaseous OC levels and chemical compositions may vary among different locations and meteorological conditions (i.e., Los Angeles in the summer versus East China Sea in the spring). For this reason, the ability of the denuder to collect gaseous OC should be monitored during actual field sampling. During 6 of the 19 airborne sampling missions, samplers A and B were run in parallel with one sampler in the B2 configuration (Figure 1). As shown in Table 2, in all such experiments the concentration of OC determined using filter QFFb,b mass of OC was less than the MDL, that is the concentration determined on QFFb,b was less than twice the value of the field blank. All samples had been field blank-corrected, for the six samples the mass of OC present on QFF field blanks was 11–36% of the mass of OC observed on QFFf,a. These results indicate that under these field situations the levels of gaseous OC adsorbed to QFFf,a were minimal. The same denuder sampler when operated side by side with a filter pack sampler in Pasadena, CA was shown by Mader et al. [2001] to remove gaseous OC that otherwise would have adsorbed to QFFf,a and caused a positive gas adsorption artifact. Moreover, Fitz [1990] observed that quartz denuders removed gaseous OC that would have adsorbed to QFF.

Table 2. Concentrationsa of OC and EC as Measured Using Different System Configurations for the Evaluation of Sampler Performance
Flight NumberSampler A (Denuder/Filter/Filter)bSampler B2 (Filter/Denuder/Filter)bField Blanks
FilterOCErrorECErrorFilterOCErrorECErrorFilterOCErrorECError
  • a

    Concentrations are expressed in μg C m−3.

  • b

    Values for Samplers A and B are field blank corrected.

  • c

    ND = nondetect.

  • d

    NA = not applicable.

1QFFf,a5.041.52NDcNAd     QFFblnk0.640.180.260.07
 QFFb,aNDNANDNAQFFb,bNDNANDNA     
2QFFf,a2.390.59NDNA     QFFblnk1.600.410.650.15
 QFFb,aNDNANDNAQFFb,bNDNANDNA     
3QFFf,a0.790.22NDNA     QFFblnk0.420.120.170.04
 QFFb,aNDNANDNAQFFb,bNDNANDNA     
4QFFf,a1.300.370.200.06     QFFblnk0.410.120.160.04
 QFFb,aNDNANDNAQFFb,bNDNANDNA     
13QFFf,a2.330.601.000.25     QFFblnk1.010.260.410.10
 QFFb,aNDNANDNA          
15QFFf,a1.210.31NDNA     QFFblnk0.430.110.170.04
 CIGb,a10.072.86NDNACIGb,b8.682.44NANACIGblnk11.223.12NANA
18QFFf,a6.601.65NDNA     QFFblnk1.650.410.670.15
 QFFb,aNDNANDNAQFFb,bNDNANDNA     
19QFFf,aNDNANDNA     QFFblnk1.260.320.510.12
 CIGb,a32.59.44NDNACIGb,b49.414.4NANACIGblnk11633.6NANA

5.4. Backup Filters in Denuder Samplers

[28] As noted earlier, some compounds that comprise carbonaceous aerosols are semivolatile and partition between the gas (G) and particle (P) phases in the atmosphere. At equilibrium, a given compound will have a P/G distribution governed by an equilibrium coefficient, Kp,

  • equation image

where cp (ng μg−1) is the particle phase concentration and cg (ng m−3) the gas phase concentration. When gas-phase semivolatile organic compounds (SOCs) are removed during flow through a denuder, cg [RIGHTWARDS ARROW] 0, and SOCs will evaporate from the P-phase in an attempt to reestablish P/G equilibrium. This process can occur either during transport of particles though denuders [Kamens and Coe, 1997] or when SOC-free gas is passed through a particle-loaded filter [Liang and Pankow, 1996]. Mader et al. [2001] derive equations that identify parameters that influence the magnitude of the negative sampling artifact. Briefly, the mass of OC evaporated from filter-bound particles present downstream of a denuder depends on: (1) the volume of SOC-free gas passed through the filter; (2) the P-phase concentration and Kp values of the compounds comprising the P-phase OC; (3) the temperature (values of Kp are inversely proportional to T); and (4) the mass fraction of carbon in the compounds comprising P-phase OC. For these reasons, the magnitude of evaporative losses of OC in denuder samplers may vary among different sampling events and types of particles (i.e., aged versus freshly emitted particles).

[29] The Kp value of a given compound is a function of temperature [Yamasaki et al., 1982];

  • equation image

each compound having an individual set of m and b values. For the partitioning of SOCs such as polycyclic aromatic hydrocarbons (PAHs) at temperatures between 273 to 303 K, a change of 10°C typically results in a factor of three change in Kp [Yamasaki et al., 1982; Bidleman et al., 1986]. As discussed by Mader et al. [2001], such a change in Kp could cause at most a factor three increase in the mass of a given compound evaporated over a given sampling event. Recall that P-phase OC is the sum of all the individual organic compounds comprising the particle phase. To minimize the mass of OC evaporated from collected particles, one should avoid an increase in temperature during sampling. This is particularly important during aircraft sampling since sampling equipment is typically placed inside the heated cabin, and the difference in air temperature inside and outside the aircraft can be quite large, especially at higher altitudes where this difference can be ≈60°C. In addition, if the aircraft descends while sampling, the temperature will increase by approximately 1°C/100 m, such an increase in temperature would also cause an increase in evaporation of particle bound OC.

[30] The denuder samplers were mounted in the unheated nose of the Twin Otter aircraft, and temperature measurements were made using probes located both outside of the aircraft and inside the flow controller of the denuder sampler (and immediately downstream of the filters). These measurements indicate that the temperature inside the samplers was typically within 4°C of the outside temperature. When sampling at a fixed altitude, the temperature varied by less than 1°C.

[31] Several authors have observed that QFF can adsorb gaseous OC [McDow and Huntzicker, 1990; Cotham and Bidleman, 1992; Hart and Pankow, 1994; Turpin et al., 1994; Storey et al., 1995; Mader and Pankow, 2001a, 2001b]. Moreover, Eatough et al. [1993] observed that QFF can adsorb gaseous OC evaporated from collected particles. In the current study QFFb,a was placed behind QFFf,a to collect gaseous OC evaporated from collected particles (Figure 1). As discussed earlier, the denuders removed gas phase OC that would adsorb to QFF, therefore should OC be found on QFFb,a, it must have evaporated from particles collected on OFFf,a. As shown in Table 2, no significant OC was observed on QFFb,a; the levels of OC on QFFb,a less than twice the levels field blank QFF. Since QFFs may not collect gaseous OC with 100% efficiency, CIGs were used during two RFs as a backup filter in Trains A and B. (It has been reported that carbon impregnated cellulose filters (CIFs) collect gaseous OC with 80–100% efficiency [Eatough et al., 1993; Tang et al., 1994; Lewtas et al., 2001]). As shown in Table 2, for RF 15 a significant amount of OC was found on CIGb,a and CIGb,b; the values were approximately half of the blank value (OC levels determined on CIGb,a and CIGb,b in Table 2 are field blank corrected) and about 6 times the levels measured on QFFf,a. The relatively high level of OC found on CIGb,b suggests that gaseous OC escapes the XAD coated denuders; however, this OC is not adsorbed by QFF (Since the levels of OC on QFFb,b are insignificant (Table 2)). This phenomenon has also been observed with XAD-coated glass annular denuders used by Lewtas et al. [2001], who suggested this OC was gaseous volatile organic compounds (VOCs). Although the XAD-coated denuders do not remove gaseous OC with 100% efficiency, as shown previously such denuders did remove gaseous OC that would adsorb to QFF, and thus these denuders did minimize positive gas adsorption artifacts.

[32] Since VOC escapes the XAD-coated denuder, the amount of OC evaporated from collected particles must be calculated from the difference between the levels of OC on CIGb,a and CIGb,b. The relatively large amount of VOC collected on CIGs (i.e., 8.68 to 49.4 μg C m−3) as compared to particle-phase OC collected on front QFFs (i.e., nondetectible to 1.21 μg C m−3) makes it difficult to evaluate possible evaporative losses of OC using CIG. For example, the typical ≈10% uncertainty in the OC levels measured for the CIGs is large relative to the OC levels found on QFFf,a. Moreover, the blank value for CIG in RF 15 was nine times higher than the OC level found on QFFf,a. That the amount of OC on CIGb,b was greater than CIGb,a during RF 19 is not consistent with the idea that OC is evaporated from collected particles. In RF 15 the amount of OC on CIGb,a and CIGb,b were equivalent considering the uncertainty in the measured values. This indicates that the majority of OC on backup CIG was due to breakthrough of VOC through the denuder rather than evaporation of OC from collected particles. Mader et al. [2001] sampled under nearly constant temperatures in Pasadena, CA and determined that less than 30% of particle phase OC could have evaporated during sampling.

[33] These data indicate that when a CIG is placed downstream of a QFF, breakthrough of VOC from the denuder could be misinterpreted as evaporation of OC from collected particles. For example, the concentration of OC determined using CIGb,b is a lower bound estimate of the total gas phase OC concentration; the average value for RF 15 was 9.4 μg C m−3 and the particle phase OC concentration was 1.21 μg C m−3. A denuder operating with 98% collection efficiency for gaseous OC would still allow enough gaseous OC to exit the denuder and be collected on a backup CIG to suggest that at least 15% of particle phase OC had evaporated from collected particles. Therefore it is particularly important to measure the denuder gaseous OC collection efficiency when using CIGs, especially if the ambient carbonaceous aerosol concentration is low.

[34] QFF were used as backup filters in 17 of the 19 sampling events during ACE-Asia since (1) QFF can adsorb gaseous OC, (2) QFF have lower levels of OC in field blanks than CIGs and, (3) the XAD-coated denuder remove gaseous OC that would adsorb to QFFs. The level of OC on backup QFF was insignificant, on average the level was a factor 0.7 the level measured for field blank QFFs. Had a significant amount of OC been observed on QFFb,a, this amount would have been added to that determined on QFFf,a. The collection efficiency of QFF for OC evaporated from collected particles is not known and is a major question. However, in this study, the error in assuming that (1) the amount of OC evaporated from collected particles under conditions of nearly constant temperature is small and (2) QFFs can adsorb a significant amount of OC evaporated from collected particles is likely less than the error introduced when using CIGs as backup filters downstream of an XAD-coated denuder since the levels of VOC collected on CIGs are so high relative to the levels of particle-phase OC collected on front QFFs.

6. Ambient OC/EC Concentrations in ACE-Asia

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

6.1. OC/EC in Individual Layers of the Atmosphere

[35] Current instrumental methods for the quantitative determination of the chemical composition of aerosol particles typically require a longer sampling time than measurements of the particle physical properties (i.e., particle size distribution, light scattering, light absorption). Therefore when both particle chemical and physical properties are needed for extinction calculations, some approximations must be made regarding the homogeneity of the atmosphere. In many ACE-Asia sampling events the atmosphere was observed to be vertically layered (Figures 4, 5, 6, and 8), and sampling was conducted at a constant altitude within a layer so as to collect sufficient aerosol for the determination of particle chemical composition in that layer. During such flights the temperature and relative humidity were observed generally to vary by less than 1°C and 8%, respectively. (As the sampling time for quantitative chemical analysis of aerosol particles decreases, it may be shown that spatial heterogeneity in the chemical composition of the particles may have been important in these sampling events; for the time being approximations regarding the homogeneity of the individual vertical layers is necessary.) For atmospheric concentrations prevalent during ACE-Asia, the minimum amount of time for OC and EC analysis was ≈30 min, and sampling was conducted at land speed ≈146 km h−1. Legs were often flown in a grid pattern rather than a straight leg; the standard deviation of the latitude and longitude during a given sampling event is provided in Table 4. During ACE-Asia certain layers could be characterized as: mineral dust, pollution-dominated and marine boundary layer (MBL). The average values of OC and EC in these layers are summarized in Table 5.

image

Figure 4. Temperature (T), relative humidity (RH), particle number concentration (N), light scattering coefficient (σsp) at wavelengths of 450, 550, and 700 nm, and Ångstrom exponent (equation image) as measured during a descent during Twin Otter Research Flight 5. Sampling was conducted for 66 min between altitudes of 2700 to 3700 m with the intention of collecting carbonaceous aerosol in a dust layer. Also shown on the y axis is the altitude at which sampling was conducted.

Download figure to PowerPoint

image

Figure 5. Temperature (T), relative humidity (RH), particle number concentration (N), light scattering coefficient (σsp) at wavelengths of 450, 550, and 700 nm, and Ångstrom exponent (equation image) as measured during a descent during Twin Otter Research Flight 11 and immediately prior to sampling for carbonaceous aerosol. Also shown on the y axis is the altitude at which sampling was conducted.

Download figure to PowerPoint

image

Figure 6. Temperature (T), relative humidity (RH), particle number concentration (N), light scattering coefficient (σsp) at wavelengths of 450, 550, and 700 nm, and Ångstrom exponent (å) as measured during a descent during Twin Otter Research Flight 17 and immediately prior to sampling for carbonaceous aerosol. (Note: large variation in the value of å above 1500 m is likely a consequence of the measured scattering coefficient being close to the detection limit of the three wavelength nephelometer). Also shown on the y axis is the altitude at which sampling was conducted.

Download figure to PowerPoint

6.1.1. OC/EC in Mineral Dust Layers

[36] The presence of dust was most easily identified by examining the values of the scattering coefficient as a function of incident light wavelength. Scattering of light by atmospheric particles was measured at 450, 550 and 700 nm using an integrating three-wavelength nephelometer (Model 3536, TSI, St. Paul MN) on the Twin Otter. An impactor was not present upstream of the nephelometer, therefore the size distribution of particles reaching the scattering chamber of the nephelometer will depend on the magnitude of the size dependent losses of particles both in the roof-mounted aircraft inlet and in the airflow pathway through the instrument. The measured particle transmission efficiency of the roof-mounted aircraft inlet and nephelometer are not available at the time of this publication, but it has been estimated that the overall d50 ≈ 8 μm. The RH inside the nephelometer was not controlled but the incoming air was heated. The RH inside the nephelometer was monitored and was typically less than 30%.

[37] For particles having diameters less than the wavelength of incident light, the magnitude of the measured scattering coefficient has a λ−4 dependence on the wavelength of the incident light [Seinfeld and Pandis, 1998], the scattering coefficient being greater for blue (450 nm) than for red (700 nm) light. For particles having diameters near the wavelength of incident light, the value of the scattering coefficient is not strongly dependent on incident light wavelength. The dependence of the scattering coefficient σsp (Mm−1) on incident light wavelength λ (nm) is often parameterized in terms of the Ångstrom exponent,

  • equation image

values of å were calculated from the slope of a linear regression with log σsp and log λ as the independent and dependent variables, respectively. (Note: the nephelometer data were not corrected for truncation errors but such errors have a minimal effect on the value of å calculated from these data.). As shown in Figure 4, during RF 5 the magnitude of the light scattering coefficient exhibited a dependence on wavelength at altitudes below 3000 m having a value of å between 1 and 2; it is likely that a significant fraction of aerosol particles in this layer had diameters <1 μm. Above 3000 m, however, the magnitude of the light scattering coefficient was not a strong function of incident light wavelength. The derived value of å ≈ 0.2 suggests that light scattering was dominated by particles having diameters >1 μm, and most likely mineral dust [Sabbah et al., 2001; Vaughan et al., 2001]. Marine aerosol that are dominated by sea-salt can also have low values of å, Delene and Ogren [2002] observed values of å ≈ 1 for such aerosols having diameters below 10 μm, this value is significantly larger than the value measured for mineral dusts.

[38] During RFs 5 and 6, dust layers were encountered during sampling, OC levels were approximately 2.5 μg C m−3 with EC concentration below the method detection limit, but likely less than ≈1.2 μg C m−3 (Tables 3 and 5). Interestingly, CC was not detected in these samples. It is not immediately clear why CC was not observed, perhaps the particle transmission efficiency and concentration of dust particles was too low in these samples.

Table 3. OC and EC Levels,a CC Filter Concentrations,b Absorption Coefficients (σap),c Mass Absorption Coefficients (Eabs),d and Ångstrom Exponents (å)e
DateFlightSample Time, UTOC(±)EC(±)CCσap(±)Eabs(±)åLayer
  • a

    OC and EC levels are expressed in μg C m−3.

  • b

    CC filter concentrations are expressed in μg C cm−2.

  • c

    Absorption coefficients are expressed in Mm−1.

  • d

    Mass absorption coefficients are expressed in m2 g−1.

  • e

    Italics indicates that the ambient concentration was less than the MDL; the italicized value reported is the MDL.

2 April 2001201:21:41–03:10:002.390.591.300.300.00NANANANANAmultiple
4 April 2001301:07:45–04:56:260.790.220.340.090.0044NA12NAmultiple
6 April 2001400:22:31–04:13:121.300.370.200.060.00874037NAmultiple
8April 2001504:24:36–07:41:461.230.300.410.090.00NANANANANAmultiple
8 April 2001506:14:56–07:21:262.600.631.210.270.00NANANANA0.5dust
9 April 2001602:38:40–05:32:100.580.140.460.110.00NANANANANApollution/dust
9 April 2001603:54:20–04:54:103.270.791.320.300.00NANANANA1.1dust
9 April 2001602:38:39–05:32:262.940.721.270.290.00NANANANANAmultiple
12 April 2001702:40:07–03:48:38; 04:31:48–04:59:492.150.540.560.130.081532780.8multiple
12 April 2001703:48:48–04:31:385.501.282.230.480.05163NA20.9mbl
13 April 2001801:06:46–02:04:36; 03:58:27–04:13:281.410.351.240.300.00111NA21.2mbl
13 April 2001802:32:37–03:50:061.190.290.920.210.001211331.2mbl
17 April 20011103:41:58–06:24:482.610.650.580.140.3413112220NAmultiple
17 April 20011104:55:08–05:29:086.941.592.810.630.05256NA31.1pollution/dust
19 April 20011201:37:49–04:58:291.530.380.380.090.11NANANANANAmultiple
19 April 20011203:14:59–04:41:191.390.331.560.340.06NANANANA1.6pollution/dust
19 April 20011201:37:59–05:57:572.520.630.800.190.13NANANANANAmultiple
20 April 20011300:29:16–01:58:362.330.601.000.250.00710710NAmultiple
23 April 20011401:02:34–04:54:051.680.430.410.100.0095NA13NAmultiple
23 April 20011401:36:34–02:55:440.950.230.530.120.0083NA71.8mbl
25 April 20011502:50:03–06:50:441.400.340.850.200.004354NAmultiple
26 April 20011600:45:41–04:38:223.690.980.400.100.00149NA24NAmultiple
26 April 20011601:34:51–02:41:416.731.740.630.140.00152NA61.7pollution/mbl
27 April 20011700:50:04–01:36:14; 02:12:24–03:12:246.631.690.890.200.00271NA71.7mbl
27 April 20011701:38:24–02:10:1428.947.441.620.370.0028121781.6pollution
27 April 20011700:50:04–04:20:2912.974.311.800.570.00246135NApollution/mbl
28 April 20011802:39:16–03:39:066.601.651.580.360.00212NA32pollution
28 April 20011802:18:16–04:48:533.211.011.020.310.009999NAmultiple
1 May 20011901:14:58–02:44:492.970.701.200.260.0066NA51.9pollution

[39] On RFs 11 and 12 indicators of both dust and pollution were observed in the same layer. Profiles of temperature, relative humidity, particle number concentration, scattering coefficient at 450, 550 and 700 nm, and Ångstrom exponent, are shown for RF 11 in Figure 5. As illustrated in Figure 5, a dust layer was present above 2200 m, å for this layer was 0.24 (±0.25). Sampling for carbonaceous aerosol was conducted at 1388 (±30) m, and on the basis of the profile, apparently below the dust layer and near the bottom of a pollution plume. The level of OC and EC for this sample was below the MDL likely because sampling was conducted near the bottom of the plume and the sampling time was only 34 min. However, CC was detected (Table 3), suggesting that mineral dust was in fact present although å = 1.1, a value too high for a primarily dust layer, and too low for a primarily urban pollution layer. CC was also present in the sample obtained in RF 12; however, the value of å between 1.5 and 1.9 suggests the presence of a pollution layer rather than a major dust layer [Sabbah et al., 2001; Vaughan et al., 2001]. It is likely that during RFs 11 and 12 sampling was conducted in a layer consisting of a mixture of dust and pollution. These observations indicate significant levels of mineral dust can be present even though å is much greater than 0.25.

6.1.2. OC/EC in Pollution Layers

[40] Profiles of temperature, relative humidity, particle number concentration, scattering coefficient at 450, 550 and 700 nm and å, are shown for RF 17 in Figure 6. We note that three reasonably well-defined layers were present below 3000 m. The layer present between 500 m and 1500 km had a significantly higher concentration of OC and EC than was observed in the marine boundary layer (MBL) present below 500 m. For this layer, the value of å was 1.61 (± 0.13), a value typical for an urban pollution layer [Sabbah et al., 2001; Vaughan et al., 2001], moreover modeled back trajectories [ACE-Asia Field Catalog, 2001, available at www.joss.ucar.edu/ace-asia/planning.html] suggest that this layer originated over Mainland China. We will designate this layer as a pollution layer. OC and EC concentrations in this layer, 28.9 and 1.62 μg C m−3, respectively, were among the highest observed aboard the Twin Otter during ACE-Asia. These measurements were not affected by a sampling error such as lab contamination, since blank levels were not significantly higher different than those on other days. Moreover, on this day similar levels of OC were observed in the vicinity by those sampling aboard the NCAR C-130 (B. J. Huebert et al., personal communication, 2002).

[41] Some authors have used ratios of total carbon (TC) to black carbon (BC) (TC/BC) to identify the source of ambient carbonaceous aerosols. For example Novakov et al. [2000a, 2000b] report values of the TC/BC ratio of 9 and 2 for aerosol formed from biomass burning and fossil fuel combustion, respectively. Comparing these ratios to those measured during Indian Ocean Experiment (INDOEX), these authors concluded that fossil fuel combustion is the major source of carbonaceous aerosols collected during INDOEX. A comparison of TC/BC or TC/EC ratios must be made with some care, since the OC/BC or OC/EC split depends on the analysis method and for evolved gas techniques, the temperature program used [Turpin et al., 2000; Chow et al., 2001] (Schauer et al., submitted manuscript, 2002). The accuracy of a comparison of such ratios will be optimized if the same OC/BC or OC/EC sampling and analysis methods are used for the source and ambient samples. Using the same OC/EC analysis technique and a similar temperature program as was used during ACE-Asia, Schauer et al. (submitted manuscript, 2002) observed a TC/EC ratio of 7 for wood smoke particulate material formed from the burning of pinewood logs in a fireplace, Fine et al. [2002] observed an average TC/EC ratios of 7.2, 13, 26, 30 and 63 and for wood smoke formed from burning of pine, ash, poplar, sweetgum, and hickory, respectively, and Birch and Cary [1996] observed a TC/EC ratio ≈ 50 for environmental tobacco smoke. Using the same OC/EC analysis technique and a slightly different temperature program as was used during ACE-Asia, Turpin and Huntzicker [1995] observed TC/EC ratios of 5.4 to as high as ≈14 for Los Angeles ambient samples dominated by secondary organic aerosol (SOA). An average TC/EC ratio of 2.5 has been measured for primary emissions in Los Angeles; these primary emissions are most likely from fossil fuel combustion rather than biomass burning [Gray et al., 1986; Turpin and Huntzicker, 1995]. Schauer et al. [1999] report a TC/EC ratio of 1.6 for exhaust from midsize diesel trucks. The ratio of TC/EC for the pollution layer observed in RF 17 was 18 and was the highest TC/EC ratio observed in this study. On the basis of a comparison of the TC/EC value for RF 17 to those from the literature, it is likely that the pollution layer observed during RF 17 originated from biomass burning rather than primary fossil fuel emissions. It is also possible that a significant portion of ambient carbonaceous aerosol mass may have also been SOA.

[42] For RF 17 the aircraft altitude, temperature, relative humidity, light scattering coefficient, particle absorption coefficient, and particle number concentration, are shown as a function of the mission time in Figure 7. These data define characteristics of the MBL and pollution layer, and also illustrate the degree of horizontal homogeneity in temperature, relative humidity, particle number concentration, light scattering coefficient, and particle absorption coefficient at the given altitude. (Note: since the Particle Soot Absorption Photometer (PSAP) integrating time was short relative to the levels of light absorbing particles the particle light absorption data were averaged over 4 min; therefore, some fine structure as seen in other measurements was lost). Comparing the two layers sampled, differences in particle number concentration and particle light scattering suggest a difference in the particle size distribution and/or chemical composition between the two layers. From similar considerations, i.e., on the basis of particle number concentration, particle light scattering, and å values it was determined that carbonaceous aerosols were collected in pollution layers during RFs 16, 17, 18, and 19, and indeed values of å in these layers were ≈1.8. For these flights and for samples having OC and EC levels greater than the MDL, the OC and EC levels were on average ≈14 and 1.3 μg C m−3, respectively (Tables 3 and 5). The high TC/EC ratio observed for the pollution layers observed in RF 16, 17, and 18 suggest that biomass burning and/or SOA may have been important sources of the collected carbonaceous aerosols. We do not necessarily conclude that such sources dominate all pollution layers observed in this region since RF 16, 17 and 18 were conducted on successive days and back trajectories indicate that the air masses all originated from same region of Mainland China [ACE-Asia Field Catalog, 2001] (the air mass encountered during RF 18 also spent time over South Korea). Excluding data from RF 16, 17, and 18 and samples where the levels of OC and EC were less than the MDL, but including samples collected in multiple atmospheric layers the average TC/EC ratio was 3.9. This value is lower than observed for RF 16, 17 and 18 and indicates a generally greater contribution of fossil fuel emissions to TC than was observed for RFs 16,17, and 18. Note: On the basis of estimated back trajectories observed during [ACE-Asia Field Catalog, 2001], an air mass containing carbonaceous aerosols typically took on the order of 1–2 days to travel from the sites of where carbonaceous aerosols were emitted, to the location where aerosol samples were collected. An important question is whether it is possible that the formation of SOA could have caused an increase in the TC/EC ratio during the time required to transport carbonaceous aerosol from the point of emission to the location where samples were collected. This amount of SOA cannot be accurately estimated at this time, and for this reason an estimate the relative amounts of TC from fossil fuel and biomass burning was not attempted.

image

Figure 7. Flight profile for Twin Otter Research Flight 17 indicating the times when carbonaceous aerosol samples were taken. Also shown are the temperature (T), relative humidity (RH), scattering coefficient (σsp) at 550 nm, particle absorption coefficient (σap), and particle number concentration (N). (Note: PSAP response was less reliable while changing altitudes consequently measurements of particle absorption coefficient have greater noise during such times. Since the PSAP integrating time was short relative to the levels of light absorbing particles the particle light absorption data were averaged over four minutes.)

Download figure to PowerPoint

6.1.3. OC/EC in the Marine Boundary Layer (MBL)

[43] Profiles of temperature, relative humidity, particle number concentration, and light scattering at 450, 550 and 700 nm, and å are shown for RF 14 in Figure 8. Primarily from the analysis of temperature and relative humidity profiles it was determined that sampling events conducted during RFs 7, 8, 14, 16, and 17 occurred in the marine boundary layer (MBL). In Figure 9 it is shown that, while sampling the MBL during RF 14, scattering and absorption coefficients were relatively constant and the temperature fluctuated by less than 2°C. The relative humidity and particle number concentration varied significantly. In RFs 7, 8, and 14 the MBL was observed to be relatively clean, having an OC concentration ≈ 1.1 μg C m−3 and EC levels below the method detection limit and likely less than 1 μg C m−3 (Tables 3 and 5). For RF 7 and 8, å ≈ 1 a value typically observed for relatively pristine marine aerosol [Delene and Ogren, 2002]. For RF 14, å ≈ 1.8 and suggests the presence of anthropogenic aerosols [Sabbah et al., 2001; Vaughan et al., 2001; Delene and Ogren, 2002]. A significantly higher level of OC was observed during RF 17 than was observed for other samples taken in the MBL and å ≈ 1.7. As shown in Figures 6 and 7 and discussed previously, it is likely that a highly polluted layer was present above the MBL, the two layers being separated by a weak inversion. It is possible that some mixing of the polluted and MBLs occurred in some places, especially since sampling was conducted near the top of the MBL. It is also possible that ship exhaust may have been present in the MBL, as ships were occasionally present in the area. It was likely that on RF 16 anthropogenic emissions were present in the MBL; the OC and EC concentrations were 6.73 and <0.63 μg C m−3, respectively, and å = 1.7. When the MBL was affected by incursions of pollution, the levels of OC and EC were ≈9 and 1 μg C m−3, respectively (Tables 3 and 5).

image

Figure 8. Temperature (T), relative humidity (RH), particle number concentration (N), light scattering coefficient (σsp) at wavelengths of 450, 550, and 700 nm, and Ångstrom exponent (equation image) as measured during a descent during Twin Otter Research Flight 14. Sampling was conducted for 80 min at 606 m (±11 m) with the intention of collecting carbonaceous aerosol in the marine boundary layer (MBL). Also shown on the y axis is the altitude at which sampling was conducted.

Download figure to PowerPoint

image

Figure 9. Flight profile for Twin Otter Research Flight 14 indicating the time when a carbonaceous aerosol sample was taken. Also shown are the temperature (T), relative humidity (RH), scattering coefficient (σsp) at 550 nm, particle absorption coefficient (σap), and particle number concentration (N). (Note: PSAP response was less reliable while changing altitudes consequently measurements of particle absorption coefficient have greater noise during such times. Since the PSAP integrating time was short relative to the levels of light absorbing particles the particle light absorption data were averaged over four minutes).

Download figure to PowerPoint

6.2. Presence of Carbonate

[44] It is possible that carbon in the form of carbonate (CC) could be present in samples collected in the presence of dust. A significant fraction of the total mass of atmospheric CC could be present on particles larger than 2.3 μm diameter, since the transmission efficiency of such particles through the aircraft inlet was not determined; the ambient CC concentration was not calculated. Significant amounts of CC were observed during RFs 7, 11, and 12. The amount of CC detected on a filter punch was subtracted from the OC concentration, and this value is reported in Table 3. In light of the possible losses of CC during transmission through the aircraft inlet, it is only possible to get a lower bound estimate of the amount of CC present in these samples. On average for the samples containing CC, CC comprised from 6 to 52% of total carbon and averaged 18%.

6.3. Comparison of Ambient OC, EC, and TC Concentrations Measured During ACE-Asia to Those Measured in Previous Field Campaigns

[45] Concentrations (μg C m−3) of OC and EC as determined from samples collected on RFs conducted during ACE-Asia are summarized in Table 3. The average latitude, longitude, and altitude where the aerosol sample was collected and the standard deviation of these values are presented in Table 4. Concentrations of OC and EC ranged from 0.58 to 28.94 μg C m−3 and 0.20 to 1.80 μg C m−3, respectively. Samples were collected at altitudes from ≈50 m to 3000 m. Pooling the data for samples collected at all latitudes and longitudes but during sampling legs conducted at a nearly constant altitude, concentrations of OC and EC are shown versus altitude in Figures 10a and 10b. In these figures the horizontal bar represents the standard deviation of the of OC or EC measurement and the vertical arrow the range in aircraft altitude during sampling.

image

Figure 10. a. Ambient EC concentrations measured in ACE-Asia, Rubidoux, California [Kim et al., 2001], Pasadena, California [Mader et al., 2001], and on Lake Michigan downwind of Chicago, IL [Offenberg and Baker, 2000]. Data from literature plotted as the average (symbol) and the range (bars) of values observed in that study. b. Ambient OC concentrations measured during ACE-Asia, and in Pasadena, California [Mader et al., 2001]. Data from Pasadena plotted as the average (symbol) and the range (bars) of values observed in that study. Note: For ACE-Asia data where the ambient concentration was above the method detection limit (MDL), error bars in the horizontal direction correspond to the standard deviation for the measurement, error bars in the vertical direction correspond to the range of the aircraft altitude during the sampling period. For ACE-Asia data where the ambient concentration was below the MDL, the concentration corresponding to the MDL is presented; the true ambient concentration is likely below this value.

Download figure to PowerPoint

Table 4. Navigational and Meteorological Dataa
DateFlightSample Time, UTAverage Latitude(±)Average Longitude(±)Average Altitude, m(±)LayerRH, %RSD, %T, °CRSD, %
  • a

    RSD = relative standard deviation.

2 April 2001201:21:41–03:10:0032.3380.2672132.6220.27031300.4999multiple38588.261
4 April 2001301:07:45–04:56:2636.5170.3622133.2790.07451665.71144multiple34552.29196
6 April 2001400:22:31–04:13:1233.0630.2434127.8280.29631357.11255multiple43676.41112
8 April 2001504:24:36–07:41:4637.6840.4169133.5860.13712515968multiple44441.9558
8 April 2001506:14:56–07:21:2637.9020.0601133.7350.09132885.8202dust49274.15106
9 April 2001602:38:40–05:32:1037.8740.3676133.5860.15572271.11111pollution/dust47375.39108
9 April 2001603:54:20–04:54:1038.0560.1398133.6360.15512914.4285dust37331.9972
9 April 2001602:38:39–05:32:2637.8730.3692133.5860.15562273.31108multiple51393.64164
12 April 2001702:40:07–03:48:38; 04:31:48–04:59:4933.1010.1343127.6870.20861024.31008multiple38315.67141
12 April 2001703:48:48–04:31:3833.0690.0512127.5570.1059919.910.5mbl47115.613
13 April 2001801:06:46–02:04:36; 03:58:27–04:13:2832.3580.1364132.5830.141247.85.7mbl35816.52
13 April 2001802:32:37–03:50:0632.3830.1247132.6180.13611231.55.9mbl45165.855
17 April 20011103:41:58–06:24:4832.9640.0734128.1360.13611774.11093multiple54318.279
17 April 20011104:55:08–05:29:0832.9460.0497128.1410.10521388.829.5pollution/dust53510.74
19 April 20011201:37:49–04:58:2937.8380.6239133.7110.31041242.71030multiple446611.834
19 April 20011203:14:59–04:41:1937.6280.3986133.6190.21811757.3466pollution/dust27409.9724
19 April 20011201:37:59–05:57:5737.3171.158133.5430.44181723.21286multiple563610.644
20 April 20011300:29:16–01:58:3632.4520.1346132.6830.13771129.7683multiple248615.614
23 April 20011401:02:34–04:54:0533.0940.068134.2480.3101852.4719multiple495212.223
23 April 20011401:36:34–02:55:4433.070.0563134.1770.274760610.8mbl681012.16
25 April 20011502:50:03–06:50:4432.4540.1477132.6710.15461824.81168multiple69159.0943
26 April 20011600:45:41–04:38:2232.4040.1555132.6160.16161283.41109multiple28489.9652
26 April 20011601:34:51–02:41:4132.4560.1466132.6740.1511084.55.5pollution/mbl2199.364
27 April 20011700:50:04–01:36:14; 02:12:24–03:12:2434.0180.0044129.1790.34104.159.6mbl61716.33
27 April 20011701:38:24–02:10:1434.0160.0031129.4520.2537456.16.1pollution363214.13
27 April 20011700:50:04–04:20:2934.0190.007129.1410.3454426.2412pollution/mbl493813.118
28 April 20011802:39:16–03:39:0636.5630.1185133.0070.098810993.4pollution49912.12
28 April 20011802:18:16–04:48:5336.5520.1255133.0310.11871510977multiple44418.561
1 May 20011901:14:58–02:44:4935.8310.0035134.4410.2751123823.6pollution542513.54

[46] Since it is not likely that adsorption of gaseous OC to filter surfaces would significantly affect measurements of the EC concentration, the concentrations of EC measured during ACE-Asia can be compared to values measured at other locations where filter pack samplers were used to collect aerosol samples. In addition, since the relative amounts of OC and EC measured for a given sample often depend on the instrument used for analysis [Hering et al., 1990; Turpin et al., 2000; Schmid et al., 2001], data are compared only among samples analyzed using a thermal-optical carbon analyzer. In Figure 10a, EC concentrations measured during ACE-Asia are compared to values measured during sampling campaigns conducted in other locations: on the ground in Rubidoux, California [Kim et al., 2001], on the roof of the Keck Engineering Laboratory at the California Institute of Technology in Pasadena, California [Mader et al., 2001] and aboard a ship on Lake Michigan, downwind of Chicago, IL [Offenberg and Baker, 2000]. Data from each study are presented as the average value (symbol) and the range (bars) of values observed. Moreover, for sampling events during ACE-Asia where the ambient concentration exceeded the method detection limit, the error bars in the horizontal direction correspond to the standard deviation for the measurement, and the bars in the vertical direction correspond to the range of the aircraft altitude during the sampling period. For sampling events during ACE-Asia in which the OC or EC concentrations were below the MDL, the concentration corresponding to the MDL is presented. This value can be considered as an upper estimate of the true ambient EC or OC concentration.

[47] As shown in Figure 10a, the EC concentrations measured during ACE-Asia were, on average, higher than those measured downwind of Chicago, nearly equal to those observed in Pasadena, CA, and generally lower than those observed in Rubidoux, CA. Since differences among OC/EC analysis methods and aerosol sampling techniques may complicate comparisons of measured OC concentrations made during different sampling campaigns, OC concentrations measured during ACE-Asia are compared only to values measured in Pasadena, CA in which the same denuder sampler sampling system was used. As shown in Figure 10b, the concentration of OC measured in Pasadena was, on average, higher than that observed during ACE-Asia. However, in one ACE-Asia sampling event (RF 17), the OC concentration was significantly higher than any OC concentration measured at the Pasadena location.

[48] As discussed previously the relative amounts of OC and EC measured for a sample is a function of both the sample collection method and OC/EC analysis method. If these methods differ between a set of experiments meaningful comparisons of OC and EC levels measured in the experiments may not be possible. Differences in the OC/EC analysis method between experiment generally will cause the biggest difference in the relative amounts of OC and EC measured for a given sample, however on the basis of total carbon (TC) different OC/EC analysis methods generally agree to within ≈10% [Hering et al., 1990; Turpin et al., 2000; Schmid et al., 2001]. Therefore with proper consideration of differences in the methods used to collect carbonaceous aerosols, it is sometimes possible to make meaningful comparisons of the TC levels measured during different experiments. The levels of TC measured using aircraft during the field projects: ACE-2, Tropospheric Aerosol Radiative Forcing Observational Experiment (TARFOX), and INDOEX are currently available and these levels summarized in Table 5.

Table 5. Summary of Average or Approximate Levels of OC, EC, and TCa in Different Atmospheric Layersb
ExperimentLayerOCECTCMinimum TCMaximum TCReference
  • a

    OC, EC, and TC levels are all expressed in μg C m−3.

  • b

    All measurements were made from aircraft platforms.

  • c

    ≈ indicates approximate value; value was determined considering data above and below the MDL.

  • d

    ≤ indicates ambient level less than the MDL; the value reported is the MDL and is an upper estimate of the true concentration.

  • e

    Average value.

  • f

    Not applicable, the OC/EC analysis method used to analyze sample collected in this experiment was not the same as that used to analyze samples collected during ACE-Asia.

ACE-AsiaDust≈2.5c≤1.2d≤3.7dThis manuscript
Clean Marine Boundary Layer1.1e≈0.9c≈2c≈1.9c≈2.3cThis manuscript
Polluted Marine Boundary Layer9e≈1c≈10c≈7.6c≈14cThis manuscript
Pollution Layer14e≈1.3c≈15c≈7.9c≈30cThis manuscript
 
INDOEXResidual Continental Boundary LayerNAfNAf7.4e3.415.7[Mayol-Bracero et al., 2002]
Marine Boundary LayerNAfNAf5.7e≤0.413.3[Mayol-Bracero et al., 2002]
 
TARFOXAltitudes below 1 kmNAfNAf5.6e1.99.1[Novakov et al., 1997]
Between altitudes of 1 to 3.8 kmNAfNAf4.3e0.69.9[Novakov et al., 1997]
 
ACE-2Dust≤0.6d≤0.7d≤1.3d[Schmeling et al., 2000]
Clean Marine Boundary Layer1.4e≤0.6d≤2d≤1.7d≤2.2d[Schmeling et al., 2000]
Anthropogenically Influenced Marine Boundary Layer3.9≤0.6d≤4.5d[Schmeling et al., 2000]

[49] Briefly ACE-2 was conducted over the subtropical North-East Atlantic 16 June–24 July 1997 [Raes et al., 2000]. During ACE-2 Schmeling et al. [2000] measured OC and EC levels in dust layers as well as in the MBL during relatively clean conditions and when the MBL was influenced by anthropogenically generated aerosols. TARFOX was conducted in July of 1996 in a polluted region off the coast of Virginia [Novakov et al., 1997; Hobbs, 1999]. During TARFOX sampling was conducted at altitudes between 0.1 and 3.8 km, but the levels of TC in specific atmospheric layers were not reported. INDOEX was conducted over the tropical Indian Ocean during the Northern Hemisphere dry monsoon season. During February–March of 1999 aircraft measurements of carbonaceous aerosols were made in two polluted layers: the residual continental boundary layer (rCBL) and the MBL.

[50] During ACE-Asia, flights were purposely conducted in dust, pollution or the MBL, that is the sampling area, altitude and sampling times were not randomly chosen, and the sampling strategy employed was not intended to provide average OC and EC values for the entire study region; rather, the strategy was to identify the chemical composition of different layers of the atmosphere. Therefore it is best to make comparisons of TC levels between different field experiments (i.e., ACE-Asia versus INDOEX) on the basis of the type of atmospheric layer sampled (i.e., the MBL or pollution layer) rather than basing comparisons on the average TC levels measured during an entire individual experiment. The average or approximate levels of TC observed in specific atmospheric layers are summarized in Table 5 for ACE-2, INDOEX, and ACE-Asia. Also presented in Table 5 are the average levels of TC observed during TARFOX for samples collected between altitudes of 0.1 and 1 km and 1 and 3.8 km. It must be noted that the number of TC data for each experiment are different, and in some cases only a few data are available. It was not possible to use statistical techniques such as a paired T-test to determine whether, for a given type of layer, statistically significant differences in TC levels between field experiments were observed; nonetheless, some general comparisons can be made.

[51] Dust layers were sampled during both ACE-2 and ACE-Asia. The levels of TC in dust layers observed in ACE-2 were below the MDL of 1.3 μg C m−3. Considering the MDL, these levels were likely lower than the TC levels observed in dust layers sampled during ACE-Asia, where TC was greater than 2.5 and less than 3.7 μg C m−3. It is possible that the levels of organic material in the soils of the regions that are the source of the dust collected in ACE-Asia are greater than the levels in the source regions for ACE-2. It is also possible that the dust layers observed in ACE-Asia also contained some anthropogenically generated carbonaceous aerosol.

[52] The MBL was sampled during ACE-2, TARFOX, INDOEX, and ACE-Asia. When sampling the MBL under relatively clean conditions, the average and range of TC levels were nearly equal during ACE-2 and ACE-Asia (Table 5), suggesting that the levels of TC in remote background marine air may be similar. It is likely that the layers termed anthropogenically influenced MBL, altitude less than 1 km, MBL, and polluted-MBL, in ACE-2, TARFOX, INDOEX, and ACE-Asia, respectively; all describe a similar type of atmospheric layer. The level of TC for the one ACE-2 sample was below the average for INDOEX and TARFOX but within the range of values observed during those studies. The TC value for ACE-2 was significantly below the average level and range of TC levels observed in a similar layer in ACE-Asia. The levels of TC observed in this layer were greater during ACE-Asia than in similar layers sampled during INDOEX and TARFOX, however the ACE-Asia average value is influenced by an exceptionally high level during RF 17. On average it appears that the levels of TC in polluted MBLs were in the order ACE-2 < TARFOX ≈ INDOEX < ACE-Asia.

[53] As discussed previously, during ACE-Asia, layers present above the MBL were occasionally observed to be affected by anthropogenic aerosols. On the basis of back trajectories these layers were mostly observed to originate from the Mainland China and in this regard are probably similar in character to the layers termed rCBL sampled during INDOEX. For ACE-Asia an approximate average TC level in such a layer was 15 μg C m−3 and ranged from 7.6 to 30 μg C m−3. This approximate average is greater than the average value of 7.4 μg C m−3 observed during INDOEX. Moreover, compared to INDOEX, the range of TC values in this layer were greater and the maximum TC level more than twice the value observed during INDOEX.

[54] In general, for a given type of atmospheric layer, higher levels of TC were observed during ACE-Asia than were observed during ACE-2, TARFOX and INDOEX. Important parameters influencing the concentration of TC observed in a given layer, and which are not considered here, are the mixing height and amount of dispersion that occurred between point at which the TC was emitted and the location where sampling occurred. These factors, rather than only differences in the magnitude of the sources of TC in these regions, might also explain why, for a given type of atmospheric layer, some differences in the TC levels were observed between different field experiments, i.e., INDOEX versus ACE-Asia.

7. Relationship Between Levels of EC and Aerosol Particle Light Absorption

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[55] Ambient levels of light-absorbing carbon (often assumed to be exclusively EC, GC [Rosen and Novakov, 1977], or BC) are either measured directly using filter-based thermal-evolved gas analysis techniques [Chow et al., 1993; Birch and Cary, 1996] or are inferred from measurements of aerosol particle light absorption [Hansen et al., 1984; Reid et al., 1998; Heintzenberg and Bussemer, 2000]. The mass absorption coefficient, or specific absorption cross-section, Eabs (m2 g−1), expresses the light absorbing property of a substance and provides a means of relating measurements of EC mass concentration to those of EC light absorption,

  • equation image

where σap (Mm−1) is the light absorption coefficient and [EC] the mass concentration of EC (defined here as the primary light absorbing species). The value of Eabs for BC has been measured or calculated in previous studies to be in the range of 2–25 m2 g−1 [Liousse et al., 1993]. Variation in the value of Eabs between studies arises because Eabs depends on the chemical composition, mixing state (internal versus external), and size distribution of the light absorbing aerosol particles, as well as the methods used to (1) determine the mass concentration of light absorbing carbon (i.e., thermal-optical method versus MnO2 oxidation), and (2) measure σap (i.e., Particle Soot Absorption Photometer versus Aethalometer) [Ouimette and Flagan, 1982; Liousse et al., 1993; Horvath, 1995; Petzold et al., 1996].

[56] Simultaneous measurements of the ambient EC concentration and σap were made aboard the Twin Otter during ACE-Asia. The value of σap was measured using a Particle Soot Absorption Photometer (PSAP) (Radiance Research, Seattle WA); a description of this instrument is provided by Bond et al. [1999]. Briefly, ambient air is drawn through a filter at a flow rate of 1 Lpm, and the transmission of light (λ = 567 nm) through the filter is continuously monitored. With knowledge of the change in filter transmission over some period of time, the flow rate of air through the filter, and the cross-sectional area of the filter, it is possible to calculate σap. Light scattering by filter bound particles will also cause a reduction in filter transmission, resulting in a positive artifact in the value of σap. Bond et al. [1999] provide a method to correct for such an artifact, as well as correcting errors in the manufacturer's calibration. This method was utilized in the current study to correct σap data acquired by the PSAP. Moreover, values of σap were corrected from a reference pressure of 1 atm to that at which the OC/EC sample was taken; volumetric flow rate measurements in the PSAP are made by using a mass flow controller and assuming an ambient pressure of 1 atm. Another special concern when operating a PSAP onboard an aircraft is that the PSAP response appears to be affected by changes in altitude, ostensibly due to physical or chemical changes in the filter unrelated to the levels of light absorbing species. The PSAP signal, specifically light transmission through the filter, was erratic during ascents and descents, but became reliable after operating a few minutes at a constant altitude. When determining the average value of σap for a particular sampling period, data were averaged over time periods during which the PSAP response was reliable. Typically for a 30 min sampling leg a few minutes of data were not used. An impactor was not present upstream of the PSAP, therefore the size distribution of particles reaching the filter present inside the PSAP will depend on the magnitude of the size dependent losses of particles in the roof-mounted aircraft inlet and the airflow pathway through the instrument. The measured particle transmission efficiency of the roof-mounted aircraft inlet and PSAP nephelometer are not available at the time of this publication, but it has been estimated that the overall d50 ≈ 8 μm.

[57] As shown in Tables 3 and 4, sampling times for EC ranged from ≈30 min to 4 h. The PSAP measured light absorption nearly continuously; the averaging time of the PSAP was 5 s. Since the time required to measure σap was much shorter than the time required to collect ambient aerosol particles for determination of the EC concentration, values of σap were averaged over the time required to collect the carbonaceous aerosols. Among the sampling events during which carbonaceous aerosol particles were collected, values of σap ranged from 4 to 28 Mm−1 (see Table 3) which can be compared to the 1.2 to 38 Mm−1 range encountered during INDOEX [Mayol-Bracero et al., 2002]. (Note: in both studies measurements of σap were made using a PSAP and the same correction scheme used to correct for light scattering and errors in the manufactures calibration). During INDOEX, a frequently observed pollution layer was termed the residual continental boundary layer (rCBL). It was believed that this layer was formed when polluted continental air masses originating from the Indian subcontinent were advected over the Indian Ocean. The average value of σap in such layers was 20 (±10) Mm−1[Mayol-Bracero et al., 2002]. As previously discussed, during ACE-Asia it is likely that the pollution layer observed during RF 17 had originated from Mainland China. The average of σap in this layer was 26 (±6) Mm−1; the average σap value for all pollution layers observed in ACE-Asia was 21 (±8) Mm−1.

[58] In Figure 11 values of σap are shown versus the ambient EC concentration for sampling events in which the ambient EC concentration exceeded the method detection limit. Also shown are error bars corresponding to the standard deviation in the EC concentration and the standard error in the average value of σap. In some cases the standard deviation in σap is rather large, not because of problems with the PSAP, rather aircraft sampling was conducted over different altitudes, each altitude having different levels of light absorbing species. The more complex both the flight profile and the levels of light absorbing species with altitude, the higher the variability in the average measured value of σap.

image

Figure 11. Average measured particle light absorption coefficient σap (Mm−1) as a function of the measured EC concentration (μg C m−3). Indicated are the regression parameters, regression line, and type of atmospheric layer from which the carbonaceous aerosol particle sample was taken. The average mass absorption coefficient (Eabs) for the study was determined from the slope of the regression line and is equal to 11 (±5) m2 g−1.

Download figure to PowerPoint

[59] Values of σap were plotted against the EC concentration. There is significant scatter in the relationship; the r2 of a linear regression was only .50, indicating parameters other than the mass of EC significantly affected the value of σap. The average value of Eabs for EC collected during these flights, determined from the slope of the linear regression line of the data presented in Figure 11, was 11 m2 g−1 (±5.0). While this value lies within the range of values calculated by other authors [Liousse et al., 1993; Martins et al., 1998], significant event-to-event variability existed. Individual values of Eabs ranged from 5 to 40 m2 g−1 (Table 3). Since the aerosol collection technique and OC/EC analysis method were identical for all samples, the source of this variability is likely due to a difference (or differences) in the chemical composition, mixing state, and/or size distribution of the EC. Based on theoretical modeling of light absorption by black carbon (BC) [Martins et al., 1998] observed that for light of 550 nm, Eabs may vary by a factor of eight depending on the radius of a pure spherical BC particle. Moreover if this particle is coated with nonabsorbing organic material, for light of 550 nm Eabs may vary by a factor 15 depending on the particle size and coating thickness.

[60] Another possible source of event-to-event variability in the measured value of Eabs could be the presence of light absorbing species other than EC. Some minerals can absorb light. If significant amounts of light-absorbing dust were collected by the PSAP, the value of σap would be influenced by both the levels of EC and dust. The value of Eabs was calculated (equation 5) by normalizing σap only by the EC concentration; the level of dust not considered. Hematite is the primary light-absorbing mineral commonly found in ambient dust aerosol [Sokolik and Toon, 1999]. No hematite was observed in dust samples collected aboard the Twin Otter. The transmission efficiency of dust through the PSAP was not measured; however, no impactors were present upstream of the PSAP, but they were present upstream of the samplers used to collect carbonaceous aerosols. CC is a proxy for dust and was detected in several carbonaceous aerosol samples; therefore, it is likely that dust could also have reached the PSAP. Two of the nine data points plotted in Figure 11 correspond to samples having significant levels of CC but significant scatter exists among the other seven data points. Therefore the presence of dust cannot explain all the event-to-event variability in the measured value of Eabs.

[61] It might be possible that values of Eabs may be more similar in samples from a given type of atmospheric layer than among samples from different types of layers. That is, the average Eabs value for EC in a pollution layer may be different than the value measured for EC present in a dust layer. Also shown in Figure 11 and Table 3 is the type of layer from which the OC/EC sample was taken. The value of Eabs for the one sample collected in the MBL was nearly equal to the average value determined for samples taken in more than one atmospheric layer. However, an insufficient number of Eabs data were available for each type of atmospheric layer to rigorously determine whether each individual atmospheric layer had a characteristic value of Eabs.

8. Conclusions/Implications

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References

[62] At altitudes below 3 km, values of OC and EC measured during ACE-Asia were mostly lower than the levels observed in an urban area such as Pasadena, CA but were often significantly greater than values determined in remote areas. On the basis of TC values, the levels of carbonaceous aerosol in pollution layers observed during ACE-Asia were larger than the levels observed in INDOEX. Overall the levels observed in ACE-Asia were greater than those observed in TARFOX and greater than those observed during aircraft measurements conducted during in ACE-2. During ACE-Asia most samples were taken at locations hundreds of kilometers from pollution sources suggesting that these sources can affect OC and EC levels over a wide area. Pollution incursions were shown to significantly influence the levels of OC and EC in the MBL. In the MBL the levels of OC and EC increased from 0.95 to 12.97 and from <0.53 to 0.92 μg C m−3, respectively between clean and polluted conditions. Key questions are (1) Do the levels of OC and EC observed in the MBL influence the levels and physical-chemical properties of cloud condensation nuclei (CCN) in the MBL? (2) Would the influence of OC and EC on the levels and physical-chemical properties of CCN affect the formation and lifetime of clouds formed in the MBL? (3) How much does the chemical composition of OC affect the levels and physical-chemical properties of CCN, that is would carbonaceous aerosols from biomass burning affect CCN more than carbonaceous aerosols from the burning of fossil fuels?

[63] Radiative forcing caused by atmospheric aerosols depends on both aerosol physical properties, such as size distributions, and aerosol chemical properties such as OC, EC, and CC content. To properly test the accuracy of model predictions, concurrent measurements of these properties are necessary, however even when using state of the art sampling techniques, the sampling times required for the quantitative determination of OC, EC, and CC concentrations were on the order of 20 times greater than that for the measurements of particle size distributions during ACE-Asia. Therefore evaluation of the spatial and temporal resolution of model predictions of radiative forcing is currently limited by the time necessary for measurements of aerosol chemical composition. Until faster quantitative chemical sampling techniques are developed, differences in the sampling rates of instrumentation used for the determination of the physical and chemical properties of aerosols must be considered when planning aircraft sampling campaigns for the study the radiative effects of aerosols. Future work is necessary to decrease the sampling time required to quantitatively determine the ambient levels of carbonaceous aerosols. In the short term this might be accomplished using existing denuder samplers and thermal-optical techniques by concentrating collected particles over an even smaller area of filter than is currently used and decreasing the levels of carbon on field blank QFF. In the longer term, new, more sensitive analysis techniques must be developed for aircraft that combine carbonaceous aerosol collection and quantitative analysis instrumentation into a single unit.

[64] CC was occasionally observed in pollution layers present below a dust layer indicating the existence of mixtures of dust and pollution. The Ångstrom coefficient (å) measured using a three-wavelength nephelometer is helpful in identifying dust or pollution layers. The value of å is related to the size of the particles that dominate the scattering of incident light. In a pollution layer, small numbers of relatively large dust particles may not affect the value of å, but would result in a detectible mass of CC in filter samples taken in that layer. This might explain why CC was detected in a filter sample collected in a layer in which å = 1.6, a value more typical of a pollution layer.

[65] The value of the mass absorption coefficient (Eabs) was observed to vary by as much as a factor of eight from sample to sample during ACE-Asia. The practice of using measured EC concentrations and a single average Eabs value to estimate σap, or using measured σap values and a single average Eabs value to estimate the EC concentration could result in a predicted value with an error of a factor of three.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Sampling and Analytical Methodology
  5. 3. Aircraft Sampling Platform and Sample Collection
  6. 4. Carbon Analysis
  7. 5. Instrument Calibration and Performance
  8. 6. Ambient OC/EC Concentrations in ACE-Asia
  9. 7. Relationship Between Levels of EC and Aerosol Particle Light Absorption
  10. 8. Conclusions/Implications
  11. Acknowledgments
  12. References
  • ACE-Asia Field Catalog, ACE-Asia field catalog products: Model back-trajectories, 15 April 2001.
  • Bates, T. S., B. J. Huebert, J. L. Gras, B. Griffiths, and P. A. Durkee, The international global atmospheric chemistry (IGAC) project's first aerosol characterization experiment (ACE 1)—Overview, J. Geophys. Res., 103, 16,29716,318, 1998.
  • Berglund, R. N., and B. Y. H. Liu, Generation of monodisperse aerosol standards, Environ. Sci. Technol., 7, 147153, 1973.
  • Bidleman, T. F., W. N. Billings, and W. T. Foreman, Vapor–particle partitioning of semivolatile organic compounds: Estimates from field collections, Environ. Sci. Technol., 20, 10381043, 1986.
  • Birch, M. E., and R. A. Cary, Elemental carbon-based method for monitoring occupational exposures to particulate diesel exhaust, Aerosol Sci. Technol., 25, 221241, 1996.
  • Bond, T. C., T. L. Anderson, and D. Campbell, Calibration and intercomparison of filter-based measurements of visible light absorption by aerosols, Aerosol Sci. Technol., 30, 582600, 1999.
  • Cadle, S. H., and P. A. Mulawa, Atmospheric carbonaceous species measurement methods comparison study: General Motors results, Aerosol Sci. Technol., 12, 128141, 1990.
  • Chow, J. C., J. G. Watson, L. C. Pritchett, W. R. Pierson, C. A. Frazier, and R. G. Purcell, The DRI thermal optical reflectance carbon analysis system: Description, evaluation, and applications in U.S. air quality studies, Atmos. Environ., 27, 11851202, 1993.
  • Chow, J. C., J. G. Watson, D. Crow, D. H. Lowenthal, and T. Merrifield, Comparison of IMPROVE and NIOSH carbon measurements, Aerosol Sci. Technol., 34, 2334, 2001.
  • Cotham, W. E., and T. F. Bidleman, Laboratory investigations of the partitioning of organochlorine compounds between the gas phase and atmospheric aerosols on glass fiber filters, Environ. Sci. Technol., 26, 468478, 1992.
  • Cui, W., D. J. Eatough, and N. L. Eatough, Fine particulate organic material in the Los Angeles basin, 1, Assessment of the high-volume Brigham Young University organic sampling system, BIG BOSS, J. Air Waste Manage. Assoc., 48, 10241037, 1998.
  • Delene, D., and J. A. Ogren, Variability of aerosol optical properties at four North American surface monitoring sites, J. Atmos. Sci., 59, 11351150, 2002.
  • Eatough, D. J., B. Sedar, L. Lewis, L. D. Hansen, E. A. Lewis, and R. J. Farber, Determination of semivolatile organic compounds in particles in the Grand Canyon area, Aerosol Sci. Technol., 10, 438449, 1989.
  • Eatough, D. J., A. Wadsworth, D. A. Eatough, J. W. Crawford, L. D. Hansen, and E. A. Lewis, A multiple-system, multi-channel diffusion denuder sampler for the determination of fine-particulate organic material in the atmosphere, Atmos. Environ., 27, 12131219, 1993.
  • Fine, P. M., G. R. Cass, and B. R. Simoneit, Chemical characterization of fine particle emissions from the fireplace combustion of woods grown in the southern United States, Environ. Sci. Technol., 36, 14421451, 2002.
  • Fitz, D. R., Reduction of the positive organic artifact on quartz filters, Aerosol Sci. Technol., 12, 142148, 1990.
  • Fung, K., Particulate carbon speciation by MnO2 oxidation, Aerosol Sci. Technol., 12, 122127, 1990.
  • Gray, H. A., G. R. Cass, J. J. Huntzicker, E. K. Heyerdahl, and J. A. Rau, Characteristics of atmospheric organic and elemental carbon particle concentration in Los Angeles, Environ. Sci. Technol., 20, 580589, 1986.
  • Hansen, A. D., H. Rosen, and T. Novakov, The Aethalometer—An instrument for the real-time measurement of optical absorption by aerosol particles, Sci. Total. Environ., 36, 191196, 1984.
  • Hart, K. M., and J. F. Pankow, High volume air sampler for particle and gas sampling, 2, Use of backup filters to correct the adsorption of gas phase polycylic aromatic hydrocarbons to the front filter, Environ. Sci. Technol., 28, 655661, 1994.
  • Heintzenberg, J., and M. Bussemer, Development and application of a spectral light absorption photometer for aerosol and hydrosol samples, J. Aerosol Sci., 31, 801812, 2000.
  • Hering, S. V., et al., Comparison of sampling methods for carbonaceous aerosols in ambient air, Aerosol Sci. Technol., 12, 200213, 1990.
  • Hobbs, P. V., An overview of the University of Washington airborne measurements and results from the Tropospheric Aerosol Radiative Forcing Observational Experiment (TARFOX), J. Geophys. Res., 104, 22332238, 1999.
  • Horvath, H., Size segregated light absorption coefficients of the atmospheric aerosol, Atmos. Environ., 29, 875883, 1995.
  • Jayne, J. T., D. C. Leard, X. Zhang, P. Davidovits, K. A. Smith, C. E. Kolb, and D. R. Worsnop, Development of an aerosol mass spectrometer for size and composition analysis of submicron particles, Aerosol Sci. Technol., 33, 4970, 2000.
  • Johnson, R. L., J. J. Shah, R. A. Cary, and J. J. Huntzicker, An automated thermal–optical method for the analysis of carbonaceous aerosol, in Atmospheric Aerosol: Source/Air Quality Relationships, edited by E. S. Macias, and P. K. Hopke, pp. 223233, American Chemical Society, Washington, D. C., 1981.
  • Kamens, R. M., and D. L. Coe, A large gas-phase stripping device to investigate rates of PAH evaporation from airborne diesel soot particles, Environ. Sci. Technol., 31, 18301833, 1997.
  • Kim, B. M., J. Cassmassi, H. Hugo, and M. D. Zeldin, Positive organic carbon artifacts on filter medium during PM2.5 sampling in the south coast air basin, Aerosol Sci. Technol., 34, 3541, 2001.
  • Lewtas, J., Y. Pang, D. Booth, S. Reimer, D. J. Eatough, and L. A. Gundel, Comparison of sampling methods for semivolatile organic carbon associated with PM2.5, Aerosol Sci. Technol., 34, 922, 2001.
  • Liang, C., and J. F. Pankow, Gas/particle partitioning of organic compounds to environmental tobacco smoke: Partition coefficient measurements by desorption and comparison to urban particulate material, Environ. Sci. Technol., 30, 28002805, 1996.
  • Liousse, C., H. Cachier, and S. G. Jennings, Optical and thermal measurement of black carbon aerosol content in different environments: Variation of the specific attenuation cross-section, sigma, Atmos. Environ., 27, 12031211, 1993.
  • Liu, B. H., R. N. Berglund, and J. K. Agarwal, Experimental studies of optical particle counters, Atmos. Environ., 8, 717732, 1974.
  • Mader, B. T., and J. F. Pankow, Gas/solid partitioning of semivolatile organic compounds (SOCs) to air filters, 2, Partitioning of polychlorinated dibenzodioxins, polychlorinated dibenzofurans and polycyclic aromatic hydrocarbons to quartz fiber filters. Atmos. Environ., 35, 12171223, 2001a.
  • Mader, B. T., and J. F. Pankow, Gas/solid partitioning of semivolatile organic compounds (SOCs) to air filters, 3, An analysis of gas adsorption artifacts in measurements of atmospheric SOCs when using Teflon membrane filters and quartz fiber filters, Environ. Sci. Technol., 35, 34223432, 2001b.
  • Mader, B. T., R. C. Flagan, and J. H. Seinfeld, Sampling atmospheric carbonaceous aerosols using a particle trap impactor/denuder sampler, Environ. Sci. Technol., 35, 48574867, 2001.
  • Malm, W. C., J. F. Sisler, D. Huffman, R. A. Eldred, and T. A. Cahill, Spatial and seasonal trends in particle concentration and optical extinction in the United States, J. Geophys. Res., 99, 13471370, 1994.
  • Martins, J. V., P. Artaxo, C. Liousse, J. S. Reid, P. V. Hobbs, and Y. J. Kaufman, Effects of black carbon content, particle size, and mixing on light absorption by aerosol from biomass burning in Brazil, J. Geophys. Res., 103, 32,04132,050, 1998.
  • Mayol-Bracero, O. L., R. Gabriel, M. O. Andreae, T. W. Kirchstetter, T. Novakov, J. Ogren, P. Sheridan, and D. G. Streets, Carbonaceous aerosols over the Indian Ocean during INDOEX: Chemical characterization, optical properties and probable sources, J. Geophys. Res., 107, 10.1029/2000JD000039, in press, 2002.
  • McDow, S. R., and J. J. Huntzicker, Vapor adsorption artifact in the sampling of organic aerosol: Face velocity effects, Atmos. Environ., 24, 25632571, 1990.
  • Mueller, P. K., R. W. Mosley, and L. B. Pierce, Chemical composition of Pasadena aerosol by particle size and time of day, J. Coll. Interface Sci., 39, 235239, 1972.
  • Novakov, T., D. A. Hegg, and P. V. Hobbs, Airborne measurments of carbonaceous aerosols on the East Coast of the United States, J. Geophys. Res., 102, 30,02330,030, 1997.
  • Novakov, T., M. O. Andreae, R. Gabriel, T. W. Kirchstetter, O. L. Mayol-Bracero, and V. Ramanathan, Origin of carbonaceous aerosol over the tropical Indian Ocean: Biomass burning or fossil fuels? Geophys. Res. Lett., 27, 40614064, 2000a.
  • Novakov, T., T. S. Bates, and P. K. Quinn, Shipboard measurements of concentrations and properties of carbonaceous aerosols during ACE-2, Tellus, 52B, 228238, 2000b.
  • Offenberg, J. H., and J. E. Baker, Aerosol size distributions of elemental and organic carbon in urban and over-water atmospheres, Atmos. Environ., 34, 15091517, 2000.
  • Ouimette, J. R., and R. C. Flagan, The extinction coefficient of multicomponent aerosols, Atmos. Environ., 16, 24052419, 1982.
  • Pankow, J. F., An absorption model of gas/particle partitioning of organic compounds in the atmosphere, Atmos. Environ., 28, 185188, 1994.
  • Petzold, A., C. Kopp, and R. Niessner, The dependence of the specific attenuation cross-section on black carbon mass fraction and particle size, Atmos. Environ., 31, 661672, 1996.
  • Raes, F., T. Bates, F. McGovern, and M. V. Liedekerke, The 2nd Aerosol Characterization Experiment (ACE-2): General overview and main results, Tellus, 52B, 111125, 2000.
  • Reid, J. S., P. V. Hobbs, C. Liousse, J. V. Martins, R. E. Weiss, and T. F. Eck, Comparisons of techniques for measuring shortwave absorption and black carbon content of aerosols from biomass burning in Brazil, J. Geophys. Res., 103, 32,03132,040, 1998.
  • Rosen, H., and T. Novakov, Raman scattering and the characterization of atmospheric aerosol particles, Nature, 266, 708, 1977.
  • Sabbah, I., C. Ichoku, Y. J. Kaufman, and L. Remer, Full year cycle of desert dust spectral optical thickness and precipitable water vapor over Alexandria, Egypt, J. Geophys. Res., 106, 18,30518,316, 2001.
  • Schauer, J. J., M. J. Kleeman, G. R. Cass, and B. R. T. Simoneit, Measurement of emissions from air pollution sources, 2, C1 through C30 organic compounds from medium duty diesel trucks, Environ. Sci. Technol., 33, 15781587, 1999.
  • Schmeling, M., L. Russell, C. Erlick, D. Collins, H. Jonsson, Q. Wang, P. Kregsamer, and C. Streli, Aerosol particle chemical characteristics measured from aircraft in the lower troposhere during ACE-2, Tellus, 52B, 185200, 2000.
  • Schmid, H., et al., Results of the “carbon conference” international aerosol carbon round robin test stage I, Atmos. Environ., 35, 21112121, 2001.
  • Seinfeld, J. H., and S. N. Pandis, Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, pp. 99, 970, 1113, John Wiley, New York, 1998.
  • Shah, J. J., R. L. Johnson, E. K. Heyerdahl, and J. J. Huntzicker, Carbonaceous aerosol at urban and rural sites in the United States, J. Air Pollut. Control Assoc., 36, 254257, 1986.
  • Sokolik, I. N., and O. B. Toon, Incorporation of mineralogical composition into models of the radiative properties of mineral aerosol from UV to IR wavelengths, J. Geophys. Res., 104, 94239444, 1999.
  • Storey, J. M., W. Luo, L. M. Isabelle, and J. F. Pankow, Gas/solid partitioning of semivolatile organic compounds to model atmospheric solid surfaces as a function of relative humidity, 1, Clean quartz, Environ. Sci. Technol., 29, 24202428, 1995.
  • Tang, H., E. A. Lewis, D. J. Eatough, R. M. Burton, and R. J. Farber, Determination of the particle size distribution and chemical composition of semivolatile organic compounds in atmospheric fine particles with a diffusion denuder sampling system, Atmos. Environ., 28, 939947, 1994.
  • Turpin, B. J., and J. J. Huntzicker, Identification of secondary organic aerosol episodes and quantitation of primary and secondary organic aerosol concentrations during SCAQS, Atmos. Environ., 29, 35273544, 1995.
  • Turpin, B. J., J. J. Huntzicker, and K. M. Adams, Intercomparison of photoacoustic and thermal–optical methods for the measurement of atmospheric elemental carbon, Atmos. Environ., 24, 18311835, 1990.
  • Turpin, B. J., S. V. Hering, and J. J. Huntzicker, Investigations of organic aerosol sampling artifacts in the Los Angeles Basin, Atmos. Environ., 28, 30613071, 1994.
  • Turpin, B. J., P. Saxena, and E. Andrews, Measuring and simulating particulate organics in the atmosphere: Problems and prospects, Atmos. Environ., 34, 29833013, 2000.
  • Vaughan, J. K., C. Claiborn, and D. Finn, April 1998 Asian Dust event over the Columbia Plateau, J. Geophys. Res., 106, 18,38118,402, 2001.
  • Yamasaki, H., K. Kuwata, and H. Miyamoto, Effects of ambient temperature on aspects of airborne polycylic aromatic hydrocarbons, Environ. Sci. Technol., 16, 180194, 1982.
  • Yu, J. Z., J. Xu, and H. Yang, Charring characteristics of atmospheric organic particulate matter in thermal analysis, Environ. Sci. Technol., 36, 754761, 2002.