Optical properties of atmospheric extinction, scattering, and absorption coefficients were measured continuously with a transmissometer, an integrating nephelometer, and an aethalometer, respectively. Three Asian dust storm events had been observed at Kwangju on 22 March, 11–13 April, and 25–26 April 2001. The physicochemical and optical properties of Asian dust aerosols were analyzed for those three cases and compared with those observed under clean, marine, and hazy urban atmospheric conditions. Their chemical composition varied depending on the source region and the transport path of the air mass. The first Asian dust storm particles, which originated from the northwestern Chinese desert regions, showed typical dust aerosol characteristics of high loading of mineral dust. The second one, which originated initially from the northwestern Chinese desert regions, had been impacted by long-range-transported air pollutants, resulting in increased concentrations of sulfate and organic carbon particles. The third one, which originated from the northeastern Chinese sandy areas, had traveled south to Kwangju, resulting in increased elemental carbon and organic carbon concentrations. Aerosol chemical and optical properties under clean continental, southeastern marine, and stagnant local pollution conditions were also analyzed. The mass scattering coefficient and single-scattering albedo in the fine and coarse modes were determined for three Asian dust event days. The concentration of black carbon (BC) aerosol in the fine and coarse modes was measured with an aethalometer by alternately switching between a particulate-matter-smaller-than-2.5-μm (PM2.5) and a PM10 inlet to it. It was found that BC mass concentration in the coarse mode measured by an aethalometer (BCac) increased because of agglomerated black carbon particles and high loading of dust particles. Single-scattering albedo ω increased to 0.93, 0.90, and 0.84 for the three Asian dust events, respectively, while it was 0.85 for mean ω during other times.
 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 [Intergovernmental Panel on Climate Change (IPCC), 1996]. At present, tropospheric aerosols pose one of the largest uncertainties in model calculations of the climate forcing because of man-made changes in the composition of the atmosphere [IPCC, 2001]. The International Global Atmospheric Chemistry Program (IGAC) has organized a series of Aerosol Characterization Experiments (ACE) that integrate in situ measurements, satellite observations, and models to reduce the uncertainty in calculation of the climate forcing due to aerosol particles. ACE-Asia is the fourth in this series of experiments. It includes an intensive field study during the spring of 2001 (B. J. Huebert et al., Project prospectus: Radiative forcing due to anthropogenic aerosols over the Asian Pacific region, 2001, available at http://saga.pmel.noaa.gov/aceasia/). ACE-Asia focuses on aerosol outflow from Asia to the Pacific basin since both anthropogenic aerosols and mineral dust from the Asian continent greatly affect the atmospheric environment and radiation balance in the downwind regions.
 Continuous in situ aerosol measurement at a network of ground stations is an essential component of the ACE-Asia activities to quantify the chemical, physical, and radiative properties of aerosols and assess their spatial and temporal variability (B. J. Huebert et al., Project prospectus: Radiative forcing due to anthropogenic aerosols over the Asian Pacific region, 2001). Asian dust particles that originated from the deserts and loess area of the Asian continent are often transported over Korea, Japan, and the North Pacific Ocean during the spring season [Shaw, 1980; Parrington et al., 1983; Iwasaka et al., 1983; Arao and Ishizaka, 1986; Nagajima et al., 1991; Parungo et al., 1994; Zhou et al., 1994; K. W. Kim et al., 2001]. Since much more coal and biomass are burned, Asian aerosol sources unlike those in Europe and North America add more absorbing soot and organic aerosol to parts of the Asian atmosphere [Chameides et al., 1999]. Additionally, widespread fossil fuel combustion from Chinese industries results in the strong emission of carbonaceous particles. Wang and Shi  reported that total annual carbonaceous fuel consumption in China was approximately 1 × 109 t including 60% from coal. China's high rates of usage of coal and biofuels were responsible for high BC emissions, and roughly one fourth of global anthropogenic emissions were believed to originate from China [Streets et al., 2001]. The presence of east Asian desert dust adds complexity, since it can lead to both a cooling effect and a warming effect [Sokolik and Toon, 1999]. According to Meteorological Research Institute (METRI) of Korea (the trend of frequency of Asian dust storms is available from METRI at http://yellow.metri.re.kr/new_kor/datadb05_1.php) the frequency and strength of Asian dust storms have gradually increased for the past 3 years; thus it is important to investigate the chemical and optical properties of Asian atmospheric aerosol thoroughly [Huebert et al., 2003]. The Kwangju monitoring site (35°10′N, 126°54′E, 70.5 m above mean sea level) is an appropriate site for investigation of the continental impact of aerosol outflows before they affect the northern Pacific basin [Y. J. Kim et al., 2001]. This paper reports the temporal variations in chemical characteristics of aerosol and optical properties including visibility, single-scattering albedo, and mass extinction efficiency at Kwangju surface site during the ACE-Asia intensive observation period (IOP) to further understand the effect of Asian dust particle on radiative forcing. An accompanying paper will discuss more detailed chemical characteristics of ACE-Asia aerosols by incorporating current data set with time-resolved 3-hour average aerosol chemistry data obtained from a collocated DRUM sampler (S. Y. Ryu et al., manuscript in preparation, 2004).
 During the ACE-Asia IOP, aerosol measurements were performed from approximately 22 March to 4 May 2001 in the urban atmosphere of the city of Kwangju, Korea. Routine aerosol monitoring was carried out to collect aerosol samples using two samplers: a versatile air pollutant sampler, URG-VAPS (model 2000J), with three filter packs and a Wide Inlet Sequential air sampler (WINS, R&P Partisol-Plus Model 2025) with one filter pack. Diurnal 12-hour sampling beginning at 0600 LT and 1800 LT was carried out every third day during the ACE-Asia IOP, 22 March to 4 May 2001 in Kwangju, Korea. However, sampling was carried out every day during the Asian dust storm event days. Because an Asian dust storm event has generally lasted for 1 or 2 days over the Korean peninsula because of the extremely strong wind speed associated with it, 12-hour rather than 24-hour sampling was introduced to get better time-resolved representative samples. Table 1 shows the sampling methods and analytic parameters for aerosol collection.
Table 1. Sampling Methods and Filter Treatments for Aerosol Analysisa
Particle Size, μm
Analytical Method and Organization
Sampling frequency for all sampling methods: for routine sampling, 12-hour time base (daytime and nighttime) per 3 days; for Asian dust intensive sampling, 12-hour time base (daytime and nighttime) per day.
 Two arms of the URG-VAPS collect both fine particles with aerodynamic diameter ≤2.5 μm on nylon (Nylasorb) and a quartz filter and coarse particles with aerodynamic diameter up to 10 μm on a polycarbonate (Nuclepore) filter. The R&P WINS Sequential Air sampler employed a Teflon (Teflo) filter to collect aerosol samples with a 2.5-μm cut size inlet. A Sartorius model Micro microbalance was used to measure gravimetric mass of particles collected on Teflon and polycarbonate Nuclepore filters. Each filter was preweighed after conditioning for 24 hours under relative humidity (RH) < 50% in desiccators. Once exposed, the filter was weighted after conditioning again for 24 hours in the desiccators. They were then analyzed for trace elements using Perkin Elmer model Optima 4300DU Inductively Coupled Plasma/Atomic Emission Spectrometry (ICP/AES) and/or VG Elemental model PQ3 Mass Spectrometry (MS) and UNICAM model SOLAAR989 Atomic Absorption Spectrometry (AAS). To analyze the trace elements, each sampled Teflon filter was put into a 60-mL low-pressure digestion vessel. Five milliliters of aqua regia solution (HNO3/HCl) were added to the vessel, followed by addition of 1 mL of HF, and it was capped to prevent the solution from being volatilized. The vessel was heated overnight. The filter was removed from the vessel after cleaning with 1% nitric acid, and then the solution was volatilized by a heater. The concentration solution was diluted by adding 20 mL of a 1% nitric acid into the vessel. Twenty-one species (Na, Mg, Al, P, S, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Se, Cd, Ba, Pb, and Sr) were analyzed by the isotope research team at Korea Basic Science Institute (KBSI). When the metal species were analyzed by ICP/MS, internal standard materials of 115In and 205TI were added to the samples and standard solution to correct for the instrument drift and matrix effect [Vanhaecke et al., 1992]. Nylon filters followed by two denuders in series coated with sodium carbonate and citric acid were used to collect and analyze nitrate and ammonium aerosols, respectively. They were prepared for ion chromatography (IC) analysis of ionic species (Cl−, NO3−, SO2−, Na+, NH4+, K+, Mg2+, and Ca2+) after being extracted by 10 mL of distilled water using Dionex model DX-120 IC. Each sampled filter was first put into a 20-mL vial, wetted in 1 mL of HPLC-grade methanol, and then mechanically extracted for 30 min with 9 mL of distilled deionized water. Two denuders were installed upstream of the Nylasorb filter to eliminate overestimation of acidic gas (H2SO4 and HNO3, etc.). The 242-mm and 150-mm annular denuders were prepared with zero gas generated by a drying train consisting of silica gels, potassium permanganates, activated carbons, and citric acids. Once sampled, denuders were extracted with 10 mL of distilled deionized water by carefully shaking and rolling them for IC analysis. Blank tests were performed for all filter measurements. The noise levels of gravimetric mass, ion, element, and carbon analysis were calculated to be ±0.07, ±0.04, ±0.02, and ±0.02 μg m−3, respectively.
 In addition, in order to characterize the physical and chemical properties of aerosol during Asian dust periods, the particles collected on fine and coarse filters were examined with a Hitachi model S-4700 scanning electron microscope (SEM) interfaced with Kevex model sigma energy dispersive X-ray analysis (EDX) for morphology, size, and elemental composition of individual particles.
 Optical monitoring provides a quantitative measure of each component of ambient light extinction coefficient representing visibility conditions. Optical monitoring at the Kwangju site included continuous measurements of light extinction coefficient, bext, light scattering coefficient, bscat, and light absorption coefficient, babs, using a LPV-2 long-path transmissometer, a Belfort model 1597 integrating nephelometer, and a Magee Scientific model AE-14U aethalometer, respectively. A transmissometer system, which was operated by Advanced Environmental Monitoring Research Center (ADEMRC), Kwangju Institute of Science and Technology (K-JIST), consists of a transmitter and a receiver that were installed at a distance of 1.91 km across the downtown of Kwangju [K. W. Kim et al., 2001]. The transmissometer measures extinction of light centered on 550 ± 50 nm, and calibration was carried out using the differential path method. Calibration was done by measuring the uncertainties in lamp intensity, in path length, and in atmospheric transmittance over the calibration path. The atmospheric transmittance was calculated to be 0.9423 using MODTRAN 3.0 [Berk et al., 1989] over the calibration path (the distance between transmitter and receiver during calibration) of 366 m. Uncertainty in transmissometer measurement was estimated to be 8 Mm−1. The nephelometer measures scattering of green light centered on 530 nm determined by Kodak Wratten 58 filter. The intensity of scattered light in the nephelometer is proportional to the scattering coefficient and related to mass concentration of aerosol particles [Richards et al., 1999]. Absolute calibration was done using Rayleigh scattering of pure gases, and CFC-12 gas was used to calibrate the span gain value to 1.880V of the instrument. Uncertainty in the nephelometer measurement was reported to be ±6 Mm−1 [Belfort Instrument Company, 1992].
 Concentrations of black carbon (BC) aerosols in the fine and coarse modes were measured with an aethalometer by alternately switching between a particulate-matter-smaller-than-2.5-μm (PM2.5) and a PM10 inlet to it. The aethalometer calculated the BC concentration at a single wavelength of 880 nm. The optical method for the aethalometer is a measurement of the attenuation of a beam of light transmitted through the sample when collected on a fibrous filter. This quantity is assumed to be linearly proportional to the amount of BC in the filter deposit. I0 can be defined as the intensity of light transmitted through the original filter, or through a blank portion of the filter, and I can be defined as the intensity of light transmitted through the portion of the filter on which the aerosol deposit is collected. The optical attenuation (ATN) is defined as ATN = 100 × ln (I0/I). The factor of 100 is for numerical convenience. The absorption of light by a broadband absorber such as graphitic carbon is inversely proportional to the wavelength of the light used. Thus, for a given mass of black carbon [BC], the optical attenuation at a fixed wavelength λ may be written as ATN(λ) = σ (1/λ) × [BC], where [BC] is the mass of black carbon and σ (1/λ) is the optical absorption cross section that is wavelength dependent. A specific attenuation cross section of 19 m2 g−1 was used to convert the attenuation per unit length to BC mass concentration in the analytic algorithm of the aethalometer [Liousse et al., 1993]. In this study, an absorption efficiency of 10 m2 g−1 was used to calculate the atmospheric light absorption coefficient from BC measured by the aethalometer [Hansen et al., 1993; Parungo et al., 1994; Malm et al., 1996; K. W. Kim et al., 2001; Watson, 2002]. Uncertainty in the aethalometer measurement was reported to be ±50 ng m−3 [Magee Scientific Company, 1996]. The appropriate calibration procedure, “optical test strip procedure,” was performed systematically at routine intervals in the aethalometer [Hansen et al., 1982]. Reported comparisons between the aethalometer BC data and the EC component of the EC/OC data using a thermo/optical reflectance (TOR) method have shown good agreement since EC is expected to be the principal (visible) light-absorbing component in ambient air [Hansen and Rosen, 1990; Wolff et al., 1981]. According to the results of Allen et al. , BC and EC concentrations were highly correlated (R2 = 0.925).
 For additional continuous carbon measurement, an ambient particulate carbon monitor (R&P Model 5400), which performs a thermal CO2 analysis, was used to measure 3-hour-averaged organic carbon (OC) and elemental carbon (EC) concentration. It employs a direct measurement approach to determine the concentration of carbon in particulate matter suspended in ambient air. Using a nondisperse infrared (NDIR) CO2 detector, the instrument measures the amount of CO2 released when a sample collected in a collector is oxidized at elevated temperatures under nitrogen. It draws the sampling stream onto an impactor and combusts it at 340°C to determine organic carbon (OC) and at 750°C to determine total carbon (TC). Elemental carbon (EC) is then calculated as the difference between TC and OC [Rupprecht et al., 1995]. The monitoring methods and data treatments of the optical monitoring instruments are summarized in Table 2.
Table 2. Measurement Parameters and Conditions of Optical Monitoring at Kwangju, Korea
Output Data Interval
total light extinction coefficient (550 nm)
continuous (16-min integration and 20-min cycle)
1 min of integration
light scattering coefficient (530 nm)
routine monitoring: continuously measure with PM10 inlet; Asian dust intensive monitoring: alternately measure for 1-hour time interval with PM10 and PM2.5 inlet
every 1 min
BCa and babs
black carbon conc. and light absorption coefficient (880 nm)
routine monitoring: continuously measure with PM2.5 inlet; Asian dust intensive monitoring: alternately measure for 1-hour time interval with PM10 and PM2.5 inlet
1 min of integration
Ambient carbon particulate monitor
continuously measure with PM2.5 inlet
3 hours of integration
 Air mass transport pathways were used to assess the possible aerosol sources for different episodes. Serious visibility impairment in the urban atmosphere of Kwangju resulted from high loading of dust aerosol under northwesterly strong wind conditions. Air mass backward trajectories that ended at Kwangju were computed at 500, 1000, and 2000 m above ground level (agl) with Hybrid Single-Particle Lagrangian Trajectory (HYSPLIT, NOAA/ARL) [Draxler, 1996]. The Final Run (FNL) meteorological data, which were 6-hourly archive data from National Centers for Environmental Prediction's Global Data Assimilation System (GDAS), were used for the trajectory calculation. Twenty percent errors of the traveled distance are typical for those trajectories computed from the analyzed wind field [Stohl, 1998]. Thus calculated air mass pathways indicate the general airflow rather than the exact pathway of an air mass [Parungo et al., 1994]. All back-trajectories at 2100 and 0900 UTC (0600 and 1800 local time (LT)) were calculated extending to backward 72 hours with a 1-hour time interval.
3. Results and Discussion
3.1. Temporal Variation of Aerosol Properties
Figure 1 shows the temporal variation of ambient extinction coefficient and relative humidity during the intensive monitoring period, 22 March to 4 May 2001. Table 3 gives average particulate concentrations in the fine (<2.5 μm) and coarse (2.5–10 μm) modes and average atmospheric optical coefficients (extinction, scattering, and absorption) for each aerosol-sampling period. As shown in Table 3, 12-hour mean ambient light extinction, scattering, and absorption coefficients were observed to be 406 ± 242, 319 ± 222, and 42 ± 12 Mm−1, respectively. Because of frequent dust inflow from the Chinese continent in the spring season, average coarse mode concentration was measured to be 45.4 ± 63.6 μg m−3 compared to 26.5 ± 9.2 μg m−3 in the fine mode. Three Asian dust storm (AD) events were observed on 22 March (first AD), 11–13 April (second AD), and 25 April (third AD) at Kwangju during the intensive observation periods as shown in Figure 1 and Table 3. The 12-hour average ambient light extinction coefficient during those Asian dust events varied from 919 Mm−1 on 25 April (D) up to 1318 Mm−1 on 22 March (D) (where N refers to nighttime sampling (1800–0600 LT) and D refers to daytime sampling (0600–1800 LT)). A general description of the air mass pathway is summarized in Table 3. The first Asian dust event on 22 March was the strongest one of the season, with the highest PM10 loading of 393.9 μg m−3 resulting in a fine-to-PM10 ratio of 0.11. In contrast, during 8 April (N) when the air mass pathway passed over the Korean peninsula, the ratio significantly increased to a higher value of 0.84. In general, urban aerosols are dominated by fine particles, which can dominantly contribute the ambient light extinction and include anthropogenic chemical components such as nss-sulfate, nitrate, elemental carbon, organic carbon, etc. [Turpin et al., 1991; Y. J. Kim et al., 2001].
Table 3. Air Mass Pathways, Measured Mass Concentrations of Fine and Coarse Particles, and Light Attenuation Coefficients Observed at the Kwangju Site During the ACE-Asia IOP
 Average fine mass, coarse mass, and fine mass fraction were measured to be 37.8 ± 7.2 μg m−3, 144.2 ± 107.9 μg m−3, and 0.21 ± 0.09, respectively, when air mass pathways came from northwestern China during the Asian dust events (22 March (N), 23 March (D), 10 April (N), 11 April (D), 12 April (N), and 13 April (D)). They were measured to be 22.1 ± 7.8 μg m−3, 26.3 ± 14.0 μg m−3, and 0.46 ± 0.18, respectively, when air mass pathways came from Chinese continent during non-Asian dust days, while they were measured to be 22.4 ± 3.5 μg m−3, 15.2 ± 16.3 μg m−3, and 0.60 ± 0.21, respectively, when air mass pathways came from marine areas (7 April (D), 7 April (N), 28 April (N), and 4 May (N)). They were measured to be 26.0 ± 7.9 μg m−3, 9.0 ± 7.1 μg m−3, and 0.74 ± 0.10, respectively, when air mass was stagnant (8 April (N) and 10 April (D)). It can be concluded that mass concentration of coarse particles increased when air mass pathways came from the Chinese continent, while fine mass concentration showed less variation. It indicated that ambient air at Kwangju in spring was impacted frequently by long-range-transported continental aerosol.
 On 26 March, fine mass concentrations were very low: 10.8 μg m−3 and 13.0 μg m−3 during daytime and nighttime, respectively. However, relatively high coarse mass concentrations of 28.8 μg m−3 and 42.5 μg m−3 were measured during daytime and nighttime, respectively. The average ambient light extinction coefficient was observed to be 98 Mm−1 and 125 Mm−1 during daytime and nighttime, respectively. Under those conditions, visibility can be theoretically as long as 30–40 km [Larson et al., 1988]. The relatively high concentration of coarse particles on that day did not significantly affect the visibility impairment because of its low mass extinction efficiency of 0.6–0.8 m2 g−1 [Malm et al., 1996; Li et al., 1996; K. W. Kim et al., 2001].
 From Table 3, average PM10 mass extinction efficiency (σext,PM10), fine mass fraction, and RH except Asian dust events were calculated to be 7.8 ± 4.5 m2 g−1, 0.51 ± 0.19, and 51 ± 11%, respectively. A lower value of average PM10 mass extinction efficiency, 4.8 ± 0.9 m2 g−1, was calculated during Asian dust periods under average RH of 49 ± 11% and low average fine mass fraction of 0.22 ± 0.08 conditions. Relatively low fine mass fraction and relative humidity contributed to the decrease in PM10 mass extinction efficiency during the Asian dust periods. Figure 2 shows the scatterplot between relative humidity and PM10 mass extinction efficiency for all sampling periods in Table 3. PM10 mass extinction efficiency increased significantly with relative humidity under fine-dominated conditions because of the hygroscopic properties of fine particles. However, the relative humidity effect decreased under coarse-dominated conditions and decreased even further during the Asian dust events. It has been known that hygroscopic fine aerosol at high-RH conditions can contribute to strong light extinction [Tang, 1996].
 The combined effect of fine particles and relative humidity on light extinction resulted in a diurnal difference in PM10 mass extinction efficiency. The PM10 mass extinction efficiency was calculated to be 5.8 ± 2.3 and 8.5 ± 5.4 m2 g−1 during daytime and nighttime, respectively. The average fine mass fraction and RH were measured to be 0.31 ± 0.17 and 46.6 ± 8.5% during daytime and 0.51 ± 0.21 and 55.0 ± 12.3% during nighttime, respectively. Relatively high fine mass fraction and relative humidity during nighttime contributed to the increase of PM10 mass extinction efficiency during the ACE-Asia IOP. On the night of 28 April the average extinction efficiency was computed to be 21.1 m2 g−1 under a high RH of 81.5% and a high fine/PM10 ratio of 0.77 as shown in Table 3.
3.2. Chemical Composition and Air Mass Pathways
3.2.1. Chemical Composition
 Concentrations of major aerosol components in the fine and coarse modes were calculated from the results of carbon particle and elemental analyses as summarized in Table 4. It has been reported that the mineral dust concentration has been relatively large when an Asian dust storm has occurred [Uematsu et al., 1983; Iwasaka et al., 1988]. The available equations for dust concentration for fine and coarse particles in the open literature were used in this study. Atmospheric dust aerosol concentrations can be computed from the oxides of Al, Si, Ca, K, Fe, and Ti using the IMPROVE equation as shown in Table 4 for fine particles. Dust concentrations for fine particles (FS) were calculated using the IMPROVE program [Malm et al., 1996], and mineral dust aerosol concentrations for coarse particles (MD) were calculated from Al mass concentration assuming a mineral dust to Al mass ratio of 12.5 (Al mass fraction of 8%) [Duce et al., 1980]. However, IMPROVE measurement was only made for PM2.5 fine aerosol. There is a lack of such an equation for coarse mineral dust mass. A scatterplot between IMPROVE-based dust concentration and Al-based dust concentration is shown in Figure 3. The slope, offset, and R2 were calculated to be approximately 0.79, 3.1 μg m−3, and 0.97, respectively. It shows a good correlation between them, but Al-based dust concentrations were generally higher than the IMPROVE-based ones. The average relative error (∣Al-based – IMPROVE-based∣/IMPROVE-based) due to the use of a simplified assumption in dust concentration was calculated to be approximately 31%.
Table 4. Chemical Composite Variables for Particulate Matter Used in This Study
Table 4 summarizes the chemical composite variables used in this study. EC and OC in the fine mode were measured directly with a carbon analyzer. However, no measurement of EC and OC were made for the coarse particles. Thus aethalometer-based BCa was used in Table 4. Non-sea-salt sulfate was calculated assuming a seawater SO42−/Na mass ratio of 0.25 [Chow et al., 1996].
Table 5 summarizes the concentration of aerosol chemical components, including elemental carbon (EC), organic carbon (OC), nss-sulfate, nitrate, fine soil (FS), sea-salt (SS), mineral dust (MD), and aethalometer-based BC (BCa) observed at the Kwangju site during the ACE-Asia IOP. Average values of EC, OC, nss-SO42−, NO3−, and FS in the fine mode were calculated to be 2.6 ± 0.8, 8.1 ± 2.8, 6.9 ± 3.4, 3.0 ± 1.7, and 8.9 ± 10.8 μg m−3, respectively. Coarse mode particles were dominated by mineral dust with an average value of 37.1 ± 54.5 μg m−3 during the sampling period, and even the fine mass fraction of FS was significant. This result indicates that dust aerosol that originated from the Chinese continent and possibly from the local area contributed to local haze conditions in Kwangju during the spring season.
Table 5. Aerosol Chemical Components of Aerosol Observed at the Kwangju Site During the ACE-Asia IOPa
 Pathways of air masses that arrived at Kwangju show several patterns during the ACE-Asia IOP as shown in Figure 4. Air mass pathway characteristics were further categorized in Table 3: northwestern China, northeastern China across Korea, northern Korea across western marine, western marine, southwestern marine, southeastern China, Korea, eastern China, southern marine, marine across Japan, southeastern China across marine, and marine. Figure 4 shows the air trajectories for six selected episodes. Figure 4a shows an air mass back trajectory that moved from northwestern China to the Korean peninsula. The case is named “clean continental” (CC), which represents clean conditions perturbed by continental aerosol. Figure 4b shows the “northwestern China/Korea” (NK) case, where air masses were perturbed by the mixture of continental and regional aerosols. A mixture of local and marine aerosols, “southeastern marine” (SM), is shown in Figure 4c. Hazy conditions impacted by local pollution, “stagnant local” (SL), is illustrated in Figure 4d, Asian dust particles originated from northwestern Chinese desert regions; “continental with Asian dust” (CD) is shown in Figure 4e. A mixture of Asian dust particles originated from northeastern Chinese sandy areas and regional aerosol, “northeastern China/Korea with Asian dust” (NKD), is shown in Figure 4f.
 Chemical composition and optical properties of the six types of air mass pathways shown in Figure 4 are summarized in Table 6. Under CC conditions, mineral dust was a major contributor to total PM10 mass with a mass fraction of 0.57. Lower mass fractions of carbonaceous particles, 0.03 for EC and 0.06 for OC, were observed under CC while they were 0.09 and 0.28, respectively, under stagnant air mass conditions (SL). A mixture of continental and regional aerosols (NK) exhibits a reduced mineral dust mass fraction of approximately 0.5. The sum of mass fraction of acidic and carbonaceous particles increased to 0.36, which suggested that the aerosol were impacted by the regional sources in Korea. In particular, the mass fraction of carbonaceous particles in NK increased to 0.22, more than twofold compared to that in CC, which was 0.09. Under hazy conditions impacted by local pollution (SL) the mass fraction of mineral dust decreased to 0.18, and that of the sum of acidic and carbonaceous particles increased to 0.65. The mass extinction efficiency of PM10 particles increased to 9.5 m2 g−1, approximately 4 times higher than that of CC, which was 2.4 m2 g−1. It was the highest for SM under relatively higher RH conditions. It is believed that this high value of the mass extinction efficiency resulted from the combined effects of higher fine mass fraction and the relative humidity. It can be concluded that the chemical composition of aerosol varied with local air conditions as well as air mass pathway characteristics.
Table 6. Chemical Composition and Optical Properties of Aerosols Observed at the Kwangju Site During the ACE-Asia IOP
Fine, μg m−3
Coarse, μg m−3
PM10, μg m−3
Fine Particles, μg m−3
Coarse Particles, μg m−3
σext,PM10, m2 g−1
CC (clean continental): clean conditions perturbed by continental aerosol, two events (26 March (D) and 26 March (N)).
NK (northwestern China/Korea): mixture of continental and regional aerosols, two events (4 April (D) and 22 April (D)).
SM (southern marine): mixture of local and marine aerosols, four events (7 April (N), 8 April (D), 28 April (D), and 4 May (N)).
SL (stagnant local): hazy conditions impacted by local pollution, two events (8 April (N) and10 April (D)).
CD (continental with Asian dust): Asian dust particles that originated from northwestern Chinese desert regions. First CD: two events (22 March (N) and 23 March (D)). Second CD: four events (10 April (N), 11 April (D), 12 April (N), and 13 April (D)).
NKD (northeastern China/Korea with Asian dust): mixture of Asian dust particles that originated from northeastern Chinese sandy areas and regional aerosol, one event (25 April (D)).
 Three major Asian dust storm (AD) events were observed during the ACE-Asia periods, which were classified as CD and NKD in Figure 4. The first two AD events (CD) peaked on 22 March and 11 April and were originated from the northwestern Chinese desert regions, while the third AD event (NKD) on 25 April consisted of a mixture of dust aerosol that originated from the northeastern Chinese sandy areas and regional aerosol in Korea. Therefore they had different characteristics of chemical composition and optical properties. The mineral dust mass fraction of NKD was lower than that of CD (0.68). The sum of the mass fraction of nss-sulfate and nitrate decreased from 0.9 for CD to 0.6 for NKD. However, the mass fraction of carbonaceous particles in the fine fraction increased from 0.04 for CD to 0.09 for NKD. It can be concluded that the NKD aerosol was perturbed by the anthropogenic local and regional pollutants, while the air mass was transported to Kwangju through the industrial area in northeastern China and urban areas in Korea.
3.3. Characteristics of Carbonaceous Particles During Asian Dust Events
 During the Asian dust storm periods, severe visibility impairment can result from the enormous amount of coarse particles, although their mass extinction coefficient is lower than that of fine particles. The contribution of major chemical components of fine and coarse particles to visibility degradation was studied in the urban atmosphere of Kwangju from May 1999 to December 2000 [Y. J. Kim et al., 2001]. In this previous study, the fine mass fraction of carbonaceous particles (EC + OC) during the Asian dust storm periods was observed to be less than that of other days. During the ACE-Asia IOP the extensive optical monitoring was therefore partially modified to investigate the chemical changes in coarse particles during the Asian dust event periods.
 During the Asian dust event periods, optical properties were measured with a nephelometer and an aethalometer by switching their inlets every hour between PM10 and PM2.5 modes in order to differentiate the contributions by each mode to aerosol optical properties. Aethalometer measures the absorption of light and calculates the BC mass concentration using the specific formula of attenuation cross section. Liousse et al.  have shown that the specific mass attenuation coefficient (σae) may vary with region depending on aerosol chemical composition. Figure 5 shows unusual characteristics of aethalometer output in the PM10 mode. It shows that aethalometer-based BC mass concentration drastically increased every hour when switched to a PM10 inlet to the aethalometer. It could be expected that there would be some artifacts involved with measuring absorption by dust particles on a filter because dust particles could not only scatter light but also absorb light although their mass absorption efficiency was smaller than that of black carbon particles. Thus aethalometer-based BC concentration was defined as BCa in this study. A scatterplot of mass concentration for fine carbonaceous particles between BCa measured by an aethalometer and EC measured by an ambient particulate carbon monitor is shown in Figure 6. It shows that BCa concentration was generally higher than EC concentration, although they did not show good correlation. In particular, BCa was measured to be significantly higher than EC during the AD events. Slope, offset, and R2 were calculated to be 1.13, 1.08 μg m−3, and 0.25. From these results, mass absorption efficiency of BCa can be calculated to vary from 2.8 to 13.3 m2 g−1 when mass absorption efficiency of EC is assumed to be 10 m2 g−1. Moreover, EC cannot be responsible for total light absorption by fine particles because absorbing dust particles can exist in the fine mode, such as oxides of Fe during the Asian dust periods. Light absorption coefficients in the fine (babs,fine) and coarse (babs,coarse) modes were calculated from BCa concentration. The uncertainty bound of light absorption coefficient was estimated considering a variation in mass absorption efficiency of BCa (2.8–13.3 m2 g−1), uncertainties in dust particle concentration, and mass absorption efficiency (Table 8).
 The mass absorption coefficient of various kinds of dust (σdust) was reported to be 0.03–0.1 m2 g−1, which is more than 2 orders of magnitude smaller than that of BC, 3.8–17 m2 g−1 [Patterson et al., 1977; Horvath, 1993] and 2.8–13.3 m2 g−1 in this study. From the published values of dust absorption in the literature, it is possible to show that the absorption in the coarse mode could not be due to dust alone. If there were no strongly absorbing aerosol in the coarse mode, the average absorption coefficient of coarse particles collected on the filter in the aethalometer is calculated to be ∼6–27 Mm−1 during the first CD. Then average BCac for the first CD is calculated to be ∼0.3–1.1 μg m−3 using the attenuation cross section specified for the system (19 m2 g−1). The average BCac for the second CD and NKD are calculated to be ∼0.1–0.4 and ∼0.1–0.4 μg m−3, respectively. However, the actual BCac measured by the aethalometer were 2.3, 1.5, and 2.0 μg m−3 for the first CD, the second CD, and NKD, respectively. These values were higher than the calculated ones under the assumption of no strongly absorbing aerosol. Consequently, these results indicate that strongly absorbing aerosol such as EC or BC exist on the filter in the aethalometer during the Asian dust storm events.
 Therefore calculation of the uncertainty bound of BCac was considered in this study. Possible error in BCac due to absorption by dust particles in the coarse mode can be calculated as
where [BCac] is BC mass concentration in the coarse mode measured by an aethalometer, [MD] is calculated mass concentration of mineral dust, and σae is the mass attenuation coefficient of BC, 19 m2 g−1 used by the aethalometer. However, other uncertainties still need to be considered. They are the response of the aethalometer to scattering aerosol and the uncertainty in the dust concentration due to the use of a simplified assumption from the literature. The former, so-called scattering artifact of the aethalometer can be estimated using the relationship obtained for the integrating plate absorption method [Horvath, 1997]. It was reported that the “true” attenuation cross section of an aethalometer varied with the BC fraction of total mass [Petzold et al., 1997]. Correlation factors from the subtraction method by Horvath  can be applicable to estimate the overestimation of light absorption due to the response of the aethalometer in this study. It was calculated to be 15, 8, and 5 Mm−1 for the first CD, second CD, and NKD, respectively. It can be concluded that the scattering artifact cannot account for the increased absorption in the coarse mode. Variation of aethalometer-based BCaf and mass fraction for BCac during the three Asian dust storm event periods is shown in Figure 7. Possible uncertainties were calculated to be 1.8–2.9, 0.8–1.4, and 0.8–1.0 μg m−3 on 22 March (N), 11 April (D), and 25 April (D), respectively, when the relative error (31%) in dust concentration due to the use of simplified assumption from the literature was included. Each bar represents the uncertainty bound (UBCa) of possible error in BC to dust particles. All possible uncertainty sources were considered.
 Nevertheless, in order to investigate whether the unusual variation of BCa resulted from loading of only BC or mixture of BC and dust particles, single-particle analysis was carried out on some samples of fine and coarse particles collected during the Asian dust storm event using scanning electron microscope/energy dispersive X-ray analysis (SEM/EDX). The images in Figure 8 are SEM images that show soot particles stuck to a mineral dust particle collected on a polycarbonate Nuclepore filter. The color of those dust particles is also somewhat opaque, which suggests internal mixing with other chemical components during long-range transport.
 The mass fraction of BCac, defined as [BCac]/[BCa], was calculated to be 0.41, 0.50, and 0.24 during the three Asian dust events on 22 March (N), 11 April (D), and 25 April (D), respectively. A lower mass fraction value of 0.24 was calculated for the third Asian dust event, which had had lower coarse mass concentration, 82.2 μg m−3. Higher mass fraction values of 0.41 and 0.50 were calculated for the first and second Asian dust events, respectively. Coarse mass concentrations for the first and second Asian dust events were measured to be higher value of 350.5 and 154.0 μg m−3, respectively. Hitzenberger and Tohno  studied the size distribution of black carbon aerosols in two urban areas, Vienna, Austria, and Uji, Japan. BC concentration was analyzed with an optical technique (integrating sphere technique). The BC concentration was calculated to convert the optical signal to BC mass. Its analytic algorithm was similar to that of the aethalometer used in this study. It can be estimated that average mass fractions of BC in the coarse regime were calculated to be 0.15 in Uji and a smaller value in Vienna when their average PM10 mass concentration were measured to be 35.0 and 46.7 μg m−3, respectively. According to Parungo et al. , Asian dust particles consisted of many particles of different origins externally mixed to form loose conglomerates and some particles of internal mixture of various compounds. From these results it can be expected that those BC particles were found in the coarse mode as large dust particles abundant during the Asian dust storm period collided with fine BC particles and the BC particles were externally mixed with them or agglomerated themselves.
3.4. Optical Properties of Aerosol During Asian Dust Events
 Single-scattering albedo ω depends for the most part on particle composition. BC particles are the most absorbent aerosol substance with ω < 0.5 at all wavelengths, whereas all other kinds of particles have ω > 0.85 in the visible wavelength range. Single-scattering albedo is the controlling indicator for aerosol radiative forcing in the atmosphere. Positive forcing (ω < 0.75) indicates a warming effect by aerosol radiative forcing, and negative forcing (ω > 0.85) indicates a cooling effect [Charlock and Sellers, 1980]. In Tables 7 and 8, optical aerosol parameters for particulate matter used in this study are defined and summarized, respectively. The atmospheric light extinction coefficient and scattering coefficient were measured using a transmissometer and a nephelometer, respectively. A scatterplot of light attenuation coefficients between measured bext and the sum of bscat and babs is shown in Figure 9. The slope, offset, and R2 were calculated to be approximately 0.92, −12 Mm−1, and 0.97, respectively. It shows a good correlation between them, but measured bext was consistently higher than the sum of bscat and babs because bext was measured over an open path.
Table 7. Aerosol Optical Parameters for Particulate Matter
Measurement Method or Composite Equation
Light absorption coefficient is calculated with the equation of babs = 10[BC]; 10 m2 g−1 is the absorption efficiency of BC [Hansen et al., 1993].
Fractional attenuation coefficient
ambient light extinction coefficient
transmissometer with a path length 1.91 km
ambient light scattering coefficient of PM10
nephelometer with a PM10 inlet
ambient light scattering coefficient of fine particle
nephelometer with a PM2.5 inlet
ambient light scattering coefficient of coarse particle
bscat,PM10 − bscat,fine
ambient light absorption coefficient of fine particle a
aethalometer with a PM2.5 inlet
ambient light absorption coefficient of coarse particle
aethalometer with a PM10 inlet
babs,PM10 − babs,fine
Fractional mass attenuation efficiency
mass extinction efficiency of PM10
mass scattering efficiency of fine particle
mass scattering efficiency of coarse particle
mass absorption efficiency of fine particle
mass absorption efficiency of coarse particle
Fractional single-scattering albedo
bscat,PM10/(bscat,PM10 + babs,PM10)
single-scattering albedo of fine particle
bscat,fine/(bscat,fine + babs,fine)
single-scattering albedo of coarse particle
bscat,coarse/(bscat,coarse + babs,coarse)
Table 8. Aerosol Optical Properties Observed at the Kwangju Site During the ACE-Asia Asian Dust Periodsa
Air Mass Type
Factional Attenuation Coefficient, Mm−1
Fractional Mass Attenuation Efficiency, m2 g−1
Fractional Single-Scattering Albedo
In parentheses: uncertainty bound due to the sum of absorption by dust particles, variance of BC mass absorption efficiency (2.8–13.3 m2 g−1), and dust concentration due to the use of a simplified assumption from the literature in σ and ω calculation.
CD (continental with Asian dust): Asian dust event that originated from northwestern Chinese desert regions. First CD: two events (22 March (N) and 23 March (D)). Second CD: four events (10 April (N), 11 April (D), 12 April (N), and 13 April (D)).
NKD (northeastern China/Korea with Asian dust): mixture of an Asian dust event that originated from northeastern Chinese sandy areas and regional aerosol, one event (25 April (D)).
 Atmospheric light absorption was not measured directly in this study. The aethalometer measured the light absorption of collected BC particles on the filter, whose output data were not their light absorption coefficients but BC concentration converted from a specific formula of attenuation cross section. Then the atmospheric light absorption coefficient by BC particles was calculated again using a mass absorption efficiency of 10 m2 g−1, which was suggested by Hansen et al.  and Huffman [1996a, 1996b]. However, there is no single value for the conversion factor from BC mass to light absorption; published values vary typically between 3.8 and 17 m2 g−1 [Horvath, 1993], and it is estimated to be 2.8 to 13.3 m2 g−1 in this study. Furthermore, an optical measurement method such as an aethalometer may respond to slightly scattering particles such as dust particles. It turned out that the light absorption coefficient determined by the optical method (integrating plate method) was always higher that the value obtained by the reference method [Horvath, 1997]. Consequently, there were several possible errors in the calculation of babs from the aethalometer such as not well known mass absorption efficiency of BC particles, response of the aethalometer to scattering aerosol, light absorption by dust particles, and dust concentration estimation. Single-scattering albedo was calculated in this study. Its uncertainty (USSA), which included the errors due to the sum of absorption by dust particles, variance of BC mass absorption efficiency, response of aethalometer to scattering aerosol, and dust concentration due to use of simplified assumption from the literature was estimated and summarized in Table 8. Single-scattering albedos ω increased to 0.93, 0.90, and 0.84 for three Asian dust events, respectively, from 0.85 for mean ω except for Asian dust events. Uncertainty bounds in single-scattering albedo ω were calculated to be 0.006–0.032, 0.004–0.019, and 0.004–0.016 for three Asian dust events, respectively. However, it was estimated that its total uncertainty bound for each case would be higher than the above values if uncertainty due to the response of the aethalometer to scattering aerosol were considered. Zhou et al.  estimated ω to be 0.98 during a dust storm period in the source region in the Gobi Desert, China. It is believed that the effect of pollution on the dust particles during long-range transport caused a decrease in ωcoarse as in Table 8.
4. Summary and Conclusions
 One of main objectives of the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) was to characterize the physical, chemical, and optical properties of Asian dust aerosols to reduce the uncertainties in estimating radiative forcing by them. Aerosol measurement at the Kwangju ground site provided information on temporal variability of aerosol chemical, physical, and radiative properties. To characterize the long-range transport of aerosol (mineral dust, organic, ionic, and element) in the region of interest, air mass pathways were investigated. The major air mass pathway of Asian dust storms was from either northwestern Chinese desert regions or northeastern Chinese sandy areas during the ACE-Asia IOP. The local atmospheric conditions at Kwangju were greatly impacted by continental dust aerosol transported by prevailing wind during the spring season. To quantify the direct radiative effect of the combined natural and anthropogenic aerosol during the ACE-Asia IOP, optical properties of atmospheric aerosols have been analyzed. The Asian dust particles were perturbed by the mixture of both continental and regional aerosols; thus single-scattering albedos were measured to be relatively low. High loading of Asian dust particles resulted in the forming of externally mixed dust particles with agglomerated BC particles. The mass fraction of BCac was averaged to be 0.27 during three Asian dust events. Its uncertainty bound was calculated to be 0.15–0.26. In addition, it was revealed that the quantification of coarse mass scattering efficiency should be investigated chemically as well as optically to improve the chemical transport models and to refine radiative transfer models.
 This work was supported in part by the Korea Science and Engineering Foundation (KOSEF) through the Advanced Environmental Monitoring Research Center at Kwangju Institute of Science and Technology and the Brain Korea 21 Project of Ministry of Education.