Characteristics of Asian aerosol transport simulated with a regional-scale chemical transport model during the ACE-Asia observation

Authors


Abstract

[1] The transport and optical thickness of tropospheric aerosols (dust, sulfate, carbonaceous aerosols, and sea salt) during the ACE-Asia intensive observation period (spring 2001) were simulated using a CFORS chemical transport model coupled with a regional meteorological model. Simulated aerosol fields were examined intensively with surface monitoring stations (PM10, sulfate, and total carbonaceous aerosol), Mie Lidar, and satellite observation data. It was shown that CFORS aerosol fields agree with observations and reproduced many observed characteristics including the several high concentration levels associated with the continental outflow, aerosol vertical profiles and strong correlation between dust and sulfate transports. We found the presence of the latitudinal gradient of aerosol concentrations from these comparisons. The two-month (March and April) averaged aerosol concentration and AOT fields show this latitudinal gradient more clearly, and indicated that the main dust field is located between 30°N and 45°N, while sulfate and carbonaceous field are mainly dominant from their main sources in central China and Southeast Asia to northern Japan (between 25°N and 45°N). Analyses of aerosol horizontal fluxes were also performed. We found that these distributions are closely related to characteristics of wind field of springtime and that each aerosol has the following transport route; the main dust flow is eastward along the 45°N parallel and is located in the free atmosphere; sulfate and carbonaceous aerosols within the boundary layer has a clockwise and divergent flow pattern over central China, which produce the strong outflow associated with anthropogenic emissions at northern latitudes and constrain the continental outflow at the southern latitude; and carbonaceous aerosols at the upper level (2–6 km) have another transport pathway that is along about 30°N from Thailand and Laos. Regional budgets of tropospheric aerosols showed that total emissions were 105 Tg for dust, 8.3 Tg-SO2 for sulfur (73% from human activities and 27% from volcanic activities), and 3.07 Tg for carbonaceous aerosols. Dry deposition, gravitational settling, and northward outflow of dust accounted for 33%, 27%, and 14% of total emissions, respectively. Wet deposition, eastward outflow, and dry deposition of sulfur accounted for 33%, 27%, and 21%, respectively. Regarding carbonaceous aerosols, the outflow to the east has the highest fraction (49%), followed by dry deposition (16%) and the outflow to the north (14%).

1. Introduction

[2] Atmospheric aerosols from natural and anthropogenic sources have important effects on the global and regional climate system because they scatter and absorb solar and thermal radiation (direct effect), modify the cloud optical properties by acting as cloud condensation nuclei (CCN) (indirect effect), and change atmospheric radiative budgets [e.g., Twomey, 1974; Charlson et al., 1992; Kaufman et al., 2002]. Previous aerosol model studies specifically addressed the evaluation of the role of tropospheric aerosols (i.e., mineral dust, sulfate, carbonaceous aerosols, and sea salt) for the global direct radiative forcing. However, the estimated radiative forcing for individual aerosol components differs greatly among the studies. Main reasons for these differences include the inhomogeneous distribution of aerosols caused by their short lifetimes in the atmosphere, wide range of size distributions, and complicated chemical and optical properties. Therefore it is important to develop a high-resolution three-dimensional aerosol transport model to represent spatial distribution of tropospheric aerosols and to increase our knowledge for various aerosol properties based on a detailed comparison of intensive observation data.

[3] Aerosol transport and studies of its radiative impact also showed high aerosol concentration and radiative impacts in several regions. Liousse et al. [1996], Takemura et al. [2000] and Chin et al. [2002] simulated the essential contribution of carbonaceous aerosols to the aerosol optical thickness (AOT) over biomass-burning regions (e.g., Africa, Brazil, and Southeast Asia) using the global model. Moreover, Novakov et al. [1997] reported a large amount of carbonaceous aerosols over industrial regions using results of aircraft measurement. Carbonaceous aerosols roughly consist of black carbon (that is a primary absorbing aerosol) and organic carbon (that mainly scatters solar radiation) and are considered to be an important aerosol that controls the climate system through complicated interactions. Along with carbonaceous aerosols, sulfates, which have a negative radiative effect, are another predominant aerosol over industrial regions. Both developing countries (e.g., south and east Asia) and developed countries (e.g., eastern North America and Europe) have reported high anthropogenic sulfate aerosol concentrations because anthropogenic sulfates arise primarily from oxidation of sulfur dioxide generated by fossil fuel combustion. Aside from these anthropogenic sources, natural tropospheric aerosol sources are dispersed worldwide. Tegen and Fung [1995] and Chin et al. [2002] simulated that a large amount of dust is emitted into the atmosphere from Australia, northern Africa, the Middle East, and Asia. However, it is difficult to accurately quantify the emission strength of the aerosol sources complexly dispersed, which consequently leads to the improper aerosol transport simulation. Therefore the aerosol transport study that focuses on the regional scale is required.

[4] Under that situation, recent economic growth in east Asia has sharply increased the emission strength of anthropogenic aerosols and has made this region an important anthropogenic aerosol source in the world. Moreover, east Asia includes not only industrial regions, but also biomass burning regions, desert regions, volcanoes (the main source of natural SO2) and the ocean (the sea salt aerosol source). Although a number of aerosol transport model studies have been performed, there is relatively little observation of physical, chemical and optical properties of aerosol focus on this region, which largely limits our knowledge for aerosol. Therefore it is important to perform a simulation applying intensive observation over this region, and to perform comprehensive analysis based on observational facts to understand aerosols long-range transport, their transport processes and their radiative impacts.

[5] The Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) was designed to elucidate physical, chemical and optical properties of aerosols and characteristics and mechanisms of aerosol transport, and radiative forcing [Huebert et al., 2003]. Intensive observation was conducted using three aircraft, research vessels, surface stations and numerical models during the spring of 2001. The ACE-Asia intensive observation period corresponded to the period when aerosols are frequently transported from the Asian continent to the northwest Pacific region. Many physical, chemical and optical properties of aerosols over the ACE-Asia regions such as Japan, Korea, and China were reported. Information obtained from these experiments improves the model capability and also help us understand more detailed characteristics of aerosol transport. However, it has remained difficult to capture a clear overview of aerosol distribution, its vertical profile, its transport processes and its radiative impact upon east Asia because of spatial and temporal limitations of observation.

[6] This paper reports aerosol simulation results based on the CFORS model [Uno et al., 2003b]. Here, CFORS was developed as a regional chemical transport model coupled with the regional meso-scale meteorological model to overcome above-mentioned difficulties and clearly illustrate aerosol transport processes and radiative impacts based on the hind-cast mode calculation to the ACE-Asia intensive observation period. This study focuses on producing an overview of aerosol transport and the radiative impact of tropospheric aerosols over east Asia. It is important to reproduce the continental outflow from the Asian continent; therefore observation sites under the large-scale continental outflows are chosen for model validation. Then, we show the very detailed results of CFORS simulation and provide significant features of the aerosol transport including the vertical profiles and latitudinal gradients, based on comparisons of model results with observations. Finally, our approach achieves a detailed analysis of aerosol transport flux in east Asia and regional budgets for individual aerosols, as one unique point of this study. This approach gives us an overview of aerosol transport during spring 2001 through the outflow pattern of aerosol, its main transport altitude and its prominent transport processes.

2. Outline of CFORS

2.1. Regional Meteorological Model

[7] The Regional Atmospheric Modeling System (RAMS) developed at Colorado State University [Pielke et al., 1992] was used to simulate the regional-scale meteorological field. The RAMS contains numerous options that make it amenable to use in a wide range of applications. This study used RAMS options identical to that for CFORS [Uno et al., 2003b] (e.g., horizontal and vertical mixing coefficients are evaluated as by Smagorinsky [1963] and the level 2.5 turbulent closure of Mellor and Yamada [1974, 1982], respectively). RAMS includes Kuo-type cumulus parameterization to represent subgrid-scale convective cumulus and the Kessler-type microphysics model [Walko et al., 1995]. The microphysics module is capable of simulating meso-scale cloud and precipitation. A more detailed general model description of RAMS and its capabilities have been given by Pielke et al. [1992].

[8] The simulated domain encompassed east Asia; its rotated polar stereographic mapping center was set at 25°N and 115°E (Figure 1). The Arakawa C grid system is employed [Mesinger and Arakawa, 1976] for the grid structure and all thermodynamic and moisture variables are defined at the center of a grid volume with velocity components staggered 1/2 of a grid space in their normal direction. The horizontal grid consists of 100 by 90 grid points, with resolution of 80 km. The vertical model domain is divided into 23 layers (the top level is 23 km) with the 23 nonuniform grids varying from 150 m thick at the surface to 1800 m thick at the top of the model layer in the terrain-following sigma-z coordinate. This simulation domain includes many industrial and rural regions, dramatic variation of the land usage type and topography, covers ACE-Asia region, and is capable of forecasting the most of important chemical environment.

Figure 1.

Total emissions of (a) anthropogenic sulfur (g-SO2/m2), (b) anthropogenic BC + OC (g/m2), (c) biomass BC + OC (g/m2), (d) mineral dust (g/m2), and (e) sea salt (g/m2). Triangles in Figure 1d indicate observation sites (A, Amami; B, Beijing; G, Gosan; H, Hachijo; R, Rishiri; and S, Sado), and four gray colored regions are used to compare simulation results with TOMS AI in section 3.3 (see Figure 7).

[9] The RAMS requires initial and boundary meteorological conditions. For long term simulations, a four-dimensional data analysis (FDDA) option using the nudging technique was included based on RAMS/Isentropic Analysis Package (ISAN) output. The ISAN package converts the longitude-latitude grid with specified pressure coordinate data to the RAMS rotated polar stereographic terrain following the vertical coordinate system. European Centre for Medium-Range Weather Forecasts (ECMWF) global analysis data (6 hour intervals with 1° resolution) was used as the input for RAMS/ISAN step. This RAMS/ISAN output was used for model initialization and continuous nudging calculation. For the surface boundary condition, we used the data set of 1-km resolution land cover characteristics produced by the U.S. Geological Survey (USGS) (based on AVHRR data obtained in 1992–1993), topography (10-min resolution) and sea surface temperature (SST) taken from the National Center for Atmospheric Research (NCAR) database (1° resolution).

2.2. Chemical Transport Model

[10] A chemical transport model simulates transport and optical thickness of major tropospheric aerosols, i.e., sulfate, carbonaceous (black and organic carbons), mineral dust, and sea salt aerosols. It is one of unique points that this model was developed using RAMS' additional scalar transport options so that all meteorological information calculated in RAMS is used directly by the chemical transport model at the same time step. Detailed description of most of the model components has already been presented by Uno et al. [2003b]; here we give brief summary. The model includes main aerosol transport processes such as emission, advection, diffusion, chemical reaction/conversion, and deposition. The deposition scheme is divided into the following three removal processes in this model; wet deposition, which accounts for the below-cloud scavenging based on the scheme of Westphal et al. [1988]; dry deposition, which is caused by turbulent mixing in the first model layer; and gravitational settling, which is considered for mineral dust and sea salts because of the efficiency for large particles. Cumulus convection also plays an important role for vertical distribution of biomass burning sources in subtropical regions, such as southern China, Thailand, and Myanmar. Vertical redistribution of tracers by cumulus activities was treated as enhanced vertical turbulent diffusivity from the bottom to top of cumulus cloud layers as identified by the RAMS simplified Kuo-cumulus scheme.

[11] The CFORS simulation was performed with the hind-cast mode from 20 February to 30 April with an output interval of 3 h. This period contains the ACE-Asia intensive observation period; it corresponds to the period when biomass burning in Southeast Asia and dust storms in Chinese desert regions are relatively frequent. Moreover, we conducted sensitivity simulation experiments of (VolcOFF) without volcanic SO2 emissions to investigate volcanic impact on the atmospheric environment and radiation.

2.3. Emissions for Tropospheric Aerosols

[12] Figure 1 shows total emissions between 20 February and 30 April 2001 for anthropogenic sulfur (Figure 1a), anthropogenic total carbonaceous aerosols (BC + OC) (Figure 1b), total carbonaceous aerosols (BC + OC) from open biomass burning (Figure 1c), mineral dust (Figure 1d), and sea salt (Figure 1e), respectively. In this study all of the anthropogenic emission inventories (sulfur, black carbon and organic carbon) are calculated using the data set taken from Streets et al. [2003], which includes industry, residential, transportation, power generation and agriculture. Anthropogenic sulfur emission is very high in eastern China, especially in major cities in China such as Beijing (39.9°N, 116.4°E) and Xi'an (34.3°N, 108.9°E) (Figure 1a). For the SO2 emission, volcanic eruptions are another important source. Therefore we assigned volcanic SO2 emissions estimated by Fujita et al. [1992] and Streets et al. [2003]. Moreover, emission of SO2 from the Miyakejima volcano in Japan (34.08°N, 139.53°E) was renewed based on the latest observation taken from the Japan Meteorological Agency (JMA) because of the huge volcanic eruption since 8 July 2000. Consequently, estimated emission of volcanic SO2 from the Miyakejima volcano is 10 Mg-SO2/year. Emission from Miyakejima is highest among the east Asia countries. Its emission intensity is almost equivalent to 10 times the total anthropogenic SO2 emission in Japan.

[13] Biomass burning is an especially important source of BC and OC in spring in east Asia because their emission rates change day by day as a result of daily variation of biomass burning. Therefore we assigned the daily averaged emission data set taken from Streets et al. [2003]. The daily emission rates of BC and OC are estimated using the AVHRR satellite fire count and the TOMS AI data set. Figure 1c shows that biomass emission of carbonaceous aerosols is especially high in southeastern Asia. Total carbonaceous aerosol emissions in spring in east Asia comprise about 47% from human activities and 53% from biomass burning.

[14] Mineral dust source region is defined as the desert and semi-desert area from the USGS land coverage data set. By this approach, most parts of the Gobi and Taklimakan Desert are allocated as source regions. In present calculation, the Loess Plateau and Nei Mongol's small desert regions are also defined as source regions [Uno et al., 2003b]. The uplifted dust from these source regions is distributed uniformly within the mixing layer. The emission flux of mineral dust is calculated using the simple vertical dust deflation scheme proposed by Gillette and Passi [1988] and 12 dust particles bin (ranged from 0.1 to 20 μm in diameter; the effective radii are also used for calculation of mineral dust AOT) are assigned in the model. As shown in Figure 1d, most of mineral dust is emitted around the Taklimakan and Gobi. It is important to point out that dust emission from Taklimakan is remarkably strong when compared with that of the Gobi desert. Comparisons of lidar and TOMS AI observations with the CFORS results were performed after the ACE-Asia campaign. We found that the Taklimakan emission rate is underestimated. The dust module then, is tuned after several careful sensitivity simulations. The tuned dust module explains successfully why much dust is transported from the Taklimakan. More detailed description for this will follow in a separate paper.

[15] Sea salt transport is also addressed by CFORS. Sea-salt aerosols are injected mainly into the atmosphere by bubble bursting at the ocean surface. Therefore sea-salt emissions from the ocean are highly dependent on the surface wind speed. We simulated sea-salt emission with a 12 bin mode (from 0.005 to 20.48 μm diameter). The emission flux of sea-salt aerosols is calculated using the empirical relationship from Gong et al. [1997] and Monahan et al. [1986]. Figure 1e shows that the sea-salt emission is greatest at the east side of Japan because of the high surface wind speed. Sea salt aerosol transport processes are also simulated with a 2 bin mode (fine mode, from 0.005 to 2.56 μm diameter and coarse particle mode, from 2.56 to 20.48 μm diameter) in this model.

2.4. Aerosol Optical Thickness

[16] Aerosol optical thickness (AOT) for sulfate, black carbon, organic carbon, mineral dust, sea salt and total value at a wavelength 550 nm are calculated based on the Takemura et al. [2000] to illustrate the impact of aerosols on atmospheric radiation. The relationship between AOT and the mixing ratio of mineral dust and sea salt aerosols is given as

equation image

where τ is the aerosols optical thickness, Qext is the extinction efficiency factor as a function of the particle size bin i, qa is the aerosol mixing ratio, Δpk is the layer thickness by pressure between the kth vertical layer and k + 1th vertical layer in our model, ρ is the particle density, and reff is the effective radius of the aerosol. Table 1 lists optical parameters of mineral dust and sea salt used in the simulation. In evaluating AOT for mineral dust and sea salt, the same particle size and effective radius are used as in the transport processes. The radii of the sulfate and organic carbon particles are sensitive to ambient relative humidity, so we take hygroscopic growth of their aerosols into account in the same manner as Takemura [2001]. In this method, lognormal size distribution is assumed and the mass extinction coefficient of each particle (σ) is estimated as a function of relative humidity. The relationship between AOT and the mixing ratio of sulfate and carbonaceous aerosols is given as

equation image
Table 1. Modeled Effective Radii, Extinction Efficiency Factors, and Scattering Efficiency Factors of Dust and Sea Salt in the Model
 Bin Number
123456789101112
Mineral Dust
Mode radius, μm0.130.200.330.520.821.272.023.205.068.0212.720.13
Extinction efficiency factor0.862.384.102.592.732.282.342.112.172.092.072.05
Scattering efficiency factor0.822.303.952.352.361.841.791.411.361.201.201.20
 
Sea Salt
Mode radius, μm0.565.62          
Extinction efficiency factor3.812.27          
Scattering efficiency factor3.812.27          

[17] Table 2 also shows optical parameters of sulfate and carbonaceous aerosols. Each particle size under the relative humidity of 95% is nearly twice the size of that under the dry mass condition. The aerosol extinction coefficient and single scattering albedo are also calculated with the evaluated AOT.

Table 2. Relative Humidity and Modeled Mode Radii and Extinction and Scattering Coefficients of Sulfate and Carbonaceous Aerosols
 Relative Humidity, %
050708090959899
Sulfate
Mode radius, μm0.070.0850.0950.1030.1220.1570.1950.231
Extinction coefficient, m2/g4.287.2110.1212.6020.0935.0444.9547.38
Scattering coefficient, m2/g4.287.2110.1212.6020.0935.0444.9547.38
 
Organic Carbon
Mode radius, μm0.10.1080.110.1440.1690.1960.2740.312
Extinction coefficient, m2/g3.835.055.3812.8320.6230.5057.8860.76
Scattering coefficient, m2/g3.754.965.3012.7520.5330.4157.7860.67
 
Black Carbon
Mode radius, μm0.0118       
Extinction coefficient, m2/g8.339       
Scattering coefficient, m2/g1.796       

3. Comparisons With Observations From ACE-Asia Field Experiments

[18] A detailed analysis based on information obtained from ACE-Asia field experiments and CFORS results has already been presented in ACE-Asia heavy dust storms [Uno et al., 2004], surface black carbon analysis [Uno et al., 2003a], transport of dust and pollutants [Matsumoto et al., 2003b] and Mie scattering lidar data analysis [Sugimoto et al., 2002; Shimizu et al., 2004]. As a whole, it was shown that CFORS results have very good agreement with observation results and play important roles in understanding the aerosol transport structure and pathway as follows: Matsumoto et al. [2003b] elucidate characteristics of several air masses that bring high aerosol concentrations to the Rishiri site during the three outflow events, and Uno et al. [2004] found the relationships between dust transport and potential temperature and accounted for detailed structures of several dust storms during spring 2001.

[19] The purpose of this study is to characterize the aerosol transport and its radiation impacts during the spring in east Asia, so we will widely examine the model performances using the various observation results. Lidar observation at Beijing, China, and Nagasaki, Japan, are used first to examine the simulated dust field. Particularly the onset timing and its vertical distribution are compared with the model. The interaction between dust and anthropogenic air pollution is also discussed based on the lidar depolarization signal. Second, surface observation data of PM10, sulfate and total carbonaceous aerosols (TC) are targeted to show the typical time variation during large-scale continental outflow from the Asian continent. Here, surface observation sites from remote islands are selected to avoid unnecessary contamination from local pollution. Finally, we will compare the model results with TOMS aerosol index data. Time variation of TOMS AI and CFORS aerosol AOT will be discussed.

3.1. Vertical Profiles of Dust and Air Pollution by Lidar and CFORS

[20] The National Institute for Environmental Studies (NIES) has made continuous observations of atmospheric aerosols at Tsukuba, Nagasaki, and Beijing with a continuous polarization Mie-scattering lidar system during the ACE-Asia 2001 observation period. The aerosol extinction coefficient is calculated by the method proposed by Fernald [1984] and the boundary condition of the calculation was set at 6 km. The extinction coefficient was split to dust (nonspherical aerosol) and nondust (spherical aerosols such as air pollution) fractions based on the aerosol depolarization ratio. Sugimoto et al. [2002] and Shimizu et al. [2004] described details of lidar observation and splitting method details.

[21] Figure 2 shows the observed time-height cross section of aerosol extinction coefficients and simulated ones at Beijing, China from 1 March to 30 April 2001. Beijing is located at 39.9°N and 116.3°E and is very close to the dust source area (see Figure 1). In addition, Beijing city itself is a huge urban air pollution source and is also surrounded by industrial areas in China. Therefore high anthropogenic air pollutant concentration is expected as well as dust concentration.

Figure 2.

Time-height cross sections of (a) mineral dust extinction coefficient observed by lidar (1/km), (b) air pollution extinction coefficient observed by lidar (1/km), (c) simulated mineral dust extinction coefficient (1/km), and (d) simulated sulfate extinction coefficient (1/km) at Beijing.

[22] Lidar observation captured several high dust extinction levels. For example, Figure 2a shows that dense dust layers were transported almost every 10 days. The model also simulated all high extinction coefficients except for arrivals of dust on day 70, shown in Figure 2c. Simulated depths of dust layers occasionally reached 2–3 km indicating good agreement with lidar observation. For dust on day 100, the model shows the arrival of a vertically huge dust cloud. This large dust cloud is referred as the ‘Perfect Dust Storm (PDS)’ during the ACE-Asia observation. It was transported to North America while maintaining its relatively dense dust concentration. Our model also successfully captured this PDS. Details of this PDS have been already reported [Gong et al., 2003; Liu et al., 2003; Seinfeld et al., 2004; Uno et al., 2004].

[23] For anthropogenic pollution, several high values were also observed on day 70, after 90 days, and before 100, 110, and 120 days. The model explains that sulfate aerosols largely contribute to these peaks. Both regions of high extinction coefficients by lidar and the model are located within the lower boundary layer. These vertical profiles are understandable because the air pollutants are emitted mainly from the surface level. Careful examination of the interaction between dust and air pollution reveals an interesting relationship between dust and sulfate. Figure 2 shows that high extinction level regions of spherical aerosols intermittently appeared through spring 2001. They correspond to a period in which larger dust extinction coefficients are observed. Model results also capture this tendency. Therefore we infer that dust and sulfate at Beijing are strongly correlated.

[24] In general, the model explains most dust and sulfate events over Beijing, but in some cases the model cannot reproduce the high extinction profile. Typical examples are found around days 70 and 90 in Julian days (Figure 2). The TOMS aerosol index data did not retrieve even these high concentrations (not shown). One reason for this is that the lidar signal captured the Beijing area's local air pollutants. Horizontal grid size of our transport model is 80 km, so this grid resolution cannot capture such local air pollutants. More detailed comparison between lidar and CFORS output will be discussed by Uno et al. [2004].

[25] Figure 3 shows observed and simulated two-month-averaged vertical profiles of the total extinction coefficient at Beijing and Nagasaki. The dashed lines indicate the averaged profile from lidar observations and solid lines indicate simulated total extinction coefficients. The fractions from dust, sulfate, and carbonaceous aerosols are shaded for clear illustration. Contributions of sea salts are negligible, so we do not distinguish the contribution of sea salt in this figure. For both Beijing and Nagasaki, the lidar-observed extinction coefficient has a sharp peak near the ground. Its vertical profile sharply decreases up to the altitude of about 3 km. Above this level, the profile becomes almost constant. The vertical profile of the simulated extinction coefficient has a small peak at about 1 km and decreases continuously with altitude after the peak.

Figure 3.

Two-month averaged vertical profile of the modeled (solid line) and observed (dashed line) total extinction coefficients (1/km) at (a) Beijing and (b) Nagasaki. Shaded parts inside the solid line indicate dust extinction for the red, sulfate extinction for green and carbonaceous extinction for yellow. Impacts of spherical aerosols by lidar are also shown as the difference between extinction coefficients of total aerosols and dust.

[26] This underestimation of model extinction coefficients at the surface and differences in gradients at higher levels occur because the observed extinction coefficient is sometimes affected by local air pollution and by uncertain cloud screening. However, the temporal averaged vertical profiles of modeled and observed extinction coefficients are closely correlated with coefficients of 0.84 at Beijing and 0.94 at Nagasaki. We can see that the model reproduces the main characteristics of the observed vertical profile.

[27] Relative contributions to the total extinction coefficient are also estimated from the model results. Beijing was found to be 59% from dust, 28% from sulfate, and 13% from carbonaceous aerosols. For Nagasaki, 32% was from dust, 40% from sulfate, 27% from carbonaceous aerosols, and 1% from sea salt. Dust at Beijing has a stronger impact on the atmospheric radiation than sulfate and carbonaceous aerosols. This is true because Beijing is located near the dust source region, and large amounts of dust are often transported during the spring [Shimizu et al., 2004; Uno et al., 2004]. Shimizu et al. [2004] also reported that the dust has a strong contribution to observed extinction, which is above 60% between March and May 2001. We can see that the model captures this tendency in general. However, we must note that sulfate and carbonaceous aerosols in Beijing are also important in annual cases because many strong anthropogenic sources are located near Beijing. They may have a larger radiative impact than that of dust through a year. For the Nagasaki case, sulfate has the largest fraction. Nagasaki is located in western Japan and its air quality is frequently affected by continental outflow. A typical example is seen in the high extinction levels on days 78 and 79 [Uno et al., 2003b]. On these days, a high-pressure system moved eastward from southeastern China (near Shanghai) to the southern side of Japan. Strong southwest continental outflow at the backside of high-pressure systems is a typical feature in spring and it exports an air mass that contains many anthropogenic pollutants to Japan. Therefore we find that the strong contribution caused by sulfate at Nagasaki occurred as a result of these continental outflows.

3.2. Time Variation of Tropospheric Aerosols From Ground Observation and CFORS

[28] Figures 4, 5, and 6 show observed and simulated time variations of surface concentration of particulate matter (PM10), sulfate and total carbon (TC), respectively. Table 3 shows their mean values. In Figure 5, the dotted line indicates the simulated sulfate concentration without volcanic emission (experiment VolcOFF). All dates are Julian days and all times are Japan Standard Time (JST) in these figures. For PM10, we compare the summation of the simulated aerosol concentrations (PMmodel, i.e., sum of dust, sulfate, BC, OC, and sea salts). Five remote islands (locations are shown by the triangles in Figure 1d) are selected because of the absence of local air pollution sources. It is therefore expected that the observed surface concentrations primary reflect transported aerosol concentrations. Comparisons at these islands also yield valuable information about whether or not the model properly simulates aerosol transport. Measurements at Rishiri (sulfate and TC), Sado (TC), and Hachijo (sulfate and TC) are taken from the VMAP measurement networks [Matsumoto et al., 2003a]. Amami (sulfate and TC) observations are taken from Ohta et al. (personal communication, 2003). Gosan (sulfate and PM10) Jejyu Korea is by the National Institute for Environmental Research, Korea. PM10 observation data at Rishiri, Sado, and Amami are taken from the east Asia Acid rain monitoring network (EANET).

Figure 4.

Time variation for simulated total aerosol concentrations (PMmodel) (by solid line) (μg/m3) and for observed PM10 concentrations (μg/m3) (open circle and dot line) for (a) Gosan, (b) Rishiri, (c) Sado, and (d) Amami.

Figure 5.

Same as Figure 4, but for the sulfate concentration (μg/m3) for (a) Gosan, (b) Rishiri, (c) Hachijo, and (d) Amami. The bold dotted line indicates VolcOFF results.

Figure 6.

Same as Figure 4, but for carbonaceous aerosol concentration (μg/m3) for (a) Rishiri, (b) Sado, (c) Hachijo, and (d) Amami.

Table 3. Monthly Mean Surface Concentration for Aprila
 Carbonaceous (BC + OC)Sulfate (Anthropogenic)PM10 and PMmodel
  • a

    Monthly mean surface concentration is given in μg/m3.

Rishiri (45.1°N, 141.1°E)
OBS1.693.5248.50
Model1.743.97 (86%)46.77
 
Sado (38.1°N, 138.2°E)
OBS2.5852.73
Model1.866.68 (60%)41.25
 
Gosan (33.3°N, 126.2°E)
OBS5.0042.72
Model2.515.50 (71%)66.67
 
Hachijo (33.2°N, 139.8°E)
OBS0.773.67
Model1.594.48 (64%)45.73
 
Amami (28.4°N, 129.5°E)
OBS2.014.5155.46
Model2.673.55 (80%)46.41

[29] Among these five islands, Rishiri, Sado and Hachijo are located almost exactly 140°E, while Amami and Gosan are near 130°E (Figure 1). It should be noted that these chosen islands are not included as land in our model because their areas are much smaller than our model grid resolution. The modeled surface concentrations will therefore differ from those at the real surface level and therefore, vertically averaged concentrations (below ca. 400 m) are used for comparison with observations.

3.2.1. PM10 and PMmodel

[30] In Figure 4, PM10 concentration (aerosols less than 10 μm in diameter) shows a rapid increase, especially when the main components are mineral dust. A salient feature during spring 2001 is a huge dust storm (PDS) observed between 96–104 days over the model region. According to Uno et al. [2004], this dust storm can be divided mainly into two major outbreaks of dust. The first outbreak is from the Gobi Desert and Loess Plateau. This dust was transported at relatively high latitude (about 40°N) and arrived at Rishiri on day 99. On day 98, a second dust storm was generated by the arrival of a low-pressure system over Mongolia. One feature of the second storm is that it transported at lower latitude than the first one. The second storm, in turn, reached Gosan on day 101, Amami on day 102, and Sado on day 103. We can see their features clearly in Figure 4. The overall correlation coefficient between the PMmodel and PM10 is 0.67, indicating that the model captures characteristics of PM10 time variations at the four sites.

[31] Comparison among the sites also provides important information. For example, the observed time variation at Sado is relatively similar to that at Amami, which differs from that at Rishiri. The time variation at Rishiri shows high PM10 level during the PDS period, while observations at Sado and Amami detect high levels not only during the PDS period, but also on days 96 and 114–115. Gosan also shows a concentration peak at 114–115 days. Figure 1 shows that Rishiri, Sado, and Amami are located in the northern, central, and southern parts of Japan, respectively. These differences among observations suggest that dust is transported at low latitude.

[32] More detailed comparisons show several discrepancies between PMmodel and PM10. For example, although observation shows high PM10 values on day 111 at Gosan and on days 95 and 114–115 at the Sado and Amami, this is not shown by PMmodel. Chin et al. [2002] found that relatively high Ca2+ level are observed over the Yellow Sea and the Japan Sea around these days, and their source is mainly a desert region located just north of Shenyang city (41.8°N and 123.5°E). In addition, the weather report produced by the Japan Weather Agency (JWA) reported that small-scale low-pressure systems passed over that region just before these high PM10 values were shown. Further, TOMS AI detected relatively high AI values over its region at the same time. These facts indicate that these PM10 values are caused mainly by dust emitted from the desert near the Shenyang city and that the model tends to underpredict dust emission from that region.

[33] It is important to point out that the model underpredicts PM10 values at Rishiri during the PDS period. At the same time, the model shows the presence of a dense dust layer around the boundary layer level. This underestimation may therefore be caused by difference in transport altitudes between the model and observations. Another possible explanation is the underestimation of dust emission strength in several of the desert regions. As shown in Figure 1d, the main dust sources in this model are the Taklimakan desert, the Gobi desert, and the Loess Plateau, which is a common characteristic with other dust modeling studies [e.g., Wang et al., 2000; Gong et al., 2003; Xuan and Sokolik, 2002]. However, we found the emission strength of the Gobi desert and Loess Plateau to be smaller than in other models. Shao [2001] reported from numerous experimental results that saltation, which represents the horizontal dust emission flux, is one of the most important processes in dust emission. These indicate that further improvement of the dust emission scheme is needed.

[34] Table 3 shows a significant difference of PM10 concentration among three Japanese sites, which is largest between Amami and Rishiri (about 7 μg/m3). This result indicates that PM10 is spatially transported on a much larger scale than Japan, on average. Another important feature is that all the mean PMmodel values are lower than the mean PM10 values. The largest difference between the model and observation is seen at Gosan: The mean PM10 is about twice as large as the mean PMmodel, whereas about 10 (μg/m3) differences at Amami and Sado are seen among the Japanese sites. These differences are mainly attributable to the underestimation of the background level of PM10 and of high PM10 level on days 96 and 115 at Amami and Sado, as mentioned above.

3.2.2. Sulfate and Carbonaceous Aerosols

[35] Figures 5 and 6 show comparisons of sulfate and carbonaceous aerosols observations and model results. The figures show that model results agree well with the observations and depict the absolute concentration level and its time variations. Another important feature is that the observed and simulated sulfate and TC at the Hachijo exhibit similar temporal variation with correlation coefficients of 0.72 for model and 0.80 for observation. This feature is also seen at Rishiri and Amami (with correlation coefficients of 0.76 and 0.36 for the model and 0.47 and 0.54 for observation, respectively). This is because the sulfur emission distribution is closely related to the TC emission, as shown in the Figure 1. The sulfur and TC emitted from the same region are consequently carried by the same air mass. Several previous aerosol transport studies have reported that the air mass transported from the east Asia simultaneously includes high concentrations of sulfate and carbonaceous aerosols [e.g., Kaneyasu et al., 2000; Kaneyasu and Murayama, 2000; Matsumoto et al., 2003a; Uno et al., 2003a].

[36] Detailed analysis of sulfate and BC at Rishiri and Hachijo has been presented by Uno et al. [2003a, 2003b]. They suggested that modeled wet deposition process of BC and sulfate may be weak with the simulated precipitation amount. In addition, the sulfate and TC surface concentrations are consequently overestimated at Rishiri between Julian days 95–96, 98–100, and 101–102 and at Hachijo before day 95.

[37] The results of the VolcOFF experiment are also shown in Table 3 and Figure 5 (dotted line) to illustrate the impact of the volcanic and the anthropogenic sulfate. Generally a strong contribution caused by anthropogenic sulfate can be seen in Table 3. However, volcanic sulfate can occasionally exert a larger influence. For example, the observations capture the sharp peaks at Gosan, Amami, and Hachijo between Julian days 100–104. During these periods, the weather report presented by JWA and the RAMS model show that the traveling low-pressure systems passed over southern Japan. The VolcOFF result shows that a sharp increase of the sulfate concentration level occurred as a result of the large-scale westerly transport of the Miyakejima volcanic mass by their cyclonic flows [Uno et al., 2003b; Zhang et al., 2003]. Among these four sites, Gosan and Amami are far from Miyakejima (ca. 1200 km distant). This fact demonstrates the importance of the large-scale volcanic sulfate under special synoptic weather conditions. Zhang et al. [2003] discussed this case study in more detail.

[38] Table 3 shows that simulated sulfate and TC overpredict the observed level at Hachijo. Obvious overpredictions are visible between Julian days of 95–97 and after Julian days of 110 as shown in Figure 5. On the other hand, VolcOFF results clearly capture the observed sulfate level during these periods (Figure 5). This suggests that the sulfate overestimations are caused by transport of volcanic sulfate. As shown in Uno et al. [2003b], the wind direction is always from north to south just before these overestimations, implying that they occur when the air masses that include the Miyake volcanic sulfate and the TC emitted from Japan are coming from the northern side of Hachijo. The horizontal distances separating Hachijo-Miyake volcano and Hachijo-Honshu Island, Japan, are roughly 120 km and 300 km, respectively. The horizontal grid size of 80 km may be too large to simulate precisely the influence of Miyake volcanic sulfate and Japanese TC on Hachijo. Therefore detailed analysis of the Hachijo site simulation requires finer resolution.

[39] In contrast to the Hachijo, the observed TC at Sado greatly exceeds the simulated TC on days 96–102, 108–110, and 116–120 (Figure 6), as already been mentioned by Uno et al. [2003b]. One explanation for this is underestimation of Japanese domestic emission intensity as a result of local biomass burning before rice planting.

[40] Along the 140°E meridian, the distribution of the monthly mean surface concentration both for the observed and simulated TC has a peak at Sado Island (Table 3). This indicates the presence of a north-south gradient of TC. According to the observation, the gradient increases from northern Japan to central Japan and sharply decreases from central Japan to southern Japan. Model results also capture this general tendency, however, the concentration gradient is about 80% smaller than the observations. Presence of the latitudinal gradient for TC concentration also leads us to expect it for sulfate because temporal variations of sulfate and the TC have a close relationship as mentioned above. The monthly mean observed concentration for sulfate shows the highest sulfate level at Gosan. The model also simulates higher levels at Gosan and Sado than at Amami and Hachijo (Table 3). Westerly winds are prominent around Japan during the spring, so air masses including anthropogenic air pollutants are often transported to central Japan and the Japan Sea regions. This reflects the stronger contribution of continental outflow on the Japan Sea side (Gosan and Sado) compared to the Pacific Ocean side (Amami and Hachijo). Consequently, it indicates the presence of a latitudinal gradient for sulfate surface concentration. We will make a further examination in section 4 to more clearly elucidate this latitudinal gradient and explain more detailed characteristics of tropospheric aerosols.

3.3. Comparison of TOMS Satellite Observation With the CFORS AOT Field

[41] Total Ozone Mapping Spectrometer aerosol index (TOMS AI) provides effective information to understand aerosol transport mechanisms and paths. Over both land and ocean surfaces, TOMS AI operates at UV wavelength and detects UV-absorbing aerosols such as mineral dust, volcanic ash, and soot from biomass burning. It must be noted that TOMS AI has the observational limitation of detecting aerosols within 1–2 km above the surface [Herman et al., 1997]. Therefore AI values are not always proportional to the total AOT of the air column.

[42] Figure 7 depicts the simulated optical thickness of tropospheric aerosols and TOMS AI over the gray-colored regions in Figure 1d. Model results are shown for 1200 JST because TOMS observation times are scheduled at local solar noon. Table 4 summarizes relative contributions from individual aerosols for total AOT during the simulated period. A total of four regions are specified: northeastern China (NC), northern Japan (NJ), Japan Sea (JS), and southern Japan (SJ). Model aerosol AOT and TOMS AI are averaged over the gray regions in Figure 1d.

Figure 7.

Time variation of the model AOT and TOMS AI (unitless) at (a) northeastern China (NC), (b) northern Japan (NJ), (c) Japan Sea (JS), and (d) southern Japan (SJ). Model aerosol AOT and TOMS AI are averaged over the gray regions in Figure 1.

Table 4. Relative Contribution of Each Aerosol to the Total AOT and the Mean Total AOTa
 Northeastern Part of China (NC)Northern Part of Japan (NJ)Japan Sea (JS)Southern Part of Japan (SJ)
  • a

    The total AOT is given in % and the mean total AOT is unitless.

Sea salt0.412.45.89.9
Sulfate23.640.041.837.0
Dust64.232.636.612.3
Carbonaceous11.815.015.840.8
Total AOT0.350.340.370.32

[43] Figure 7 shows a close relationship between the simulated dust and TC and high AI values. For example, AI values are relatively higher in the north, and their tendencies captured by the model. In addition, three large dust storms occurred in northeastern China (NC), northern Japan (NJ) and Japan Sea (JS) during spring 2001 (between Julian days 60–70, 70–85 and 95–115). Both TOMS AI and modeled total AOT show high values on these days.

[44] Figure 7 also shows several interesting features for aerosol transport. The first feature is that dust AOT is high when sulfate has a large AOT fraction over NC, NJ, and JS. In section 3.1, we mentioned this relationship at Beijing. In here, we can also see the same relationship over NJ and JS except for Julian days 85–95. This result provides important evidence of the correlation between dust and sulfate transports. Another feature is that a large amount of dust over NC, NJ, and JS and carbonaceous aerosols over SJ are transported almost daily; in particular, as spring proceeds. We investigated the number of dust AOT exceeding 0.1 over NJ and JS using the simulation results. Results were 8 days for March and 15 days for April over NJ, and 9 days for March and 20 days for April over JS. This result suggests that dust transport during April is more frequent compared to March.

[45] More detailed analysis of model results also gives some important information concerning aerosol radiative impact. The modeled AOT shows that most of the observed high TOMS AI values over NC are associated with dust transport. This value is prominent compared with the dust contribution at other sites. In contrast to dust, carbonaceous aerosols show the strongest contribution over SJ, whereas the sulfate fraction is highest over JS. We can find that dust AOT has a strong fraction to the total AOT in the north of the model domain. The sulfate fraction is almost the same even throughout three Japanese regions. It is also important that carbonaceous aerosols have an opposite gradient to the dust as shown in Table 4.

4. Asian-Scale Transport and Regional Budgets

[46] The preceding data illustrated the idea that each tropospheric aerosol has a latitudinal gradient of concentration and AOT within the boundary layer and in the free atmosphere. In this section, their characteristics will be more clearly described with the mean concentration and AOT fields. In addition, their distributions are closely related to different meteorological conditions, aerosol source regions, emission strengths, and transport processes. Especially, variation of the surrounding wind pattern plays an important role in transport from the Asian continent to downwind countries in east Asia. These distributions further demonstrate their average transport. Our approach ultimately yields an overview of tropospheric aerosol transport in east Asia during spring 2001.

4.1. Horizontal Distribution of Boundary Layer Aerosol Concentration Field and AOD Field

4.1.1. Concentration Field

[47] Figure 8 shows the two-month (March and April) and vertically (below 1000 m) averaged concentration field for individual aerosols in the left column and the two-month-averaged AOT field from each aerosol type and total AOT in the right column, respectively. The dashed line indicates the results from the VolcOFF experiment. For the AOT case the same contour level as for the total AOT is used to represent each aerosol fraction as part of the total AOT. Maximum values for each aerosol are generally located near their source regions. For example, mineral dust has its highest level over the inland desert area of China. High concentration of sulfate and carbonaceous aerosol are located along the east coast of China to south China. One high TC region is Southeast Asia, reflecting the strong contribution from biomass burning in spring. These results indicate that serious air pollution originates inside China and affects air quality in Japan and over the Pacific Ocean through long-range transport [Takemura et al., 2002]. Finally, high sea salt aerosol regions are located at high latitudes, resulting from high emission rates caused by strong surface winds at these locations.

Figure 8.

Horizontal distribution of boundary layer (below 1000 m) averaged aerosol concentration field (μg/m3) (left column) and horizontal distribution of the averaged model AOT field (unitless) (right column) during spring 2001 for mineral dust, sulfate (solid line) and VolcOFF sulfate (dotted line), carbonaceous aerosols, sea salt, and total aerosols AOT.

[48] This concentration field clearly depicts the presence of the latitudinal gradient of sulfate, which has a ridge extending from southern China to central Japan and which mainly dominates the central part of the modeling domain (between 25°N and 45°N). Differences between the VolcOFF and the CNTL (standard) simulations exhibit the impact caused by the volcanic sulfate, which is visible over the northern side of Japan and the west side of Japan Sea. Sensitivity analysis shows that the contribution of volcanic sulfate is about 13% of the total sulfate concentration, implying that sulfate from anthropogenic emissions is dominant within the boundary layer.

[49] The averaged TC concentration field has a large gradient between industrial and remote regions. As well as sulfate, its distribution is dominant between 30°N and 45°N. This typical distribution is consistent with results reported by Liousse et al. [1996]. On the other hand, the dust concentration field is located mainly in the northern part of the modeling domain (between 30°N and 45°N) and its concentration decreases exponentially proportional to downwind distances as reported by Tsunogai et al. [1985].

4.1.2. AOT Field

[50] As shown in Figure 8, the AOT fields from each aerosol type are almost identical to the simulated the boundary layer concentration fields. However, we can see slight differences between their distributions. For example, a ridge of dust AOT is located slightly to the north. Wind speed is stronger and its direction is more even in the free atmosphere because of the reduced effects of surface friction. Therefore most of the dust particles that are lifted above the boundary layer level are transported directly to the east by the influence of the prevailing westerly winds. These differences between the two distributions suggest that a large amount of dust is transported in the free atmosphere and that the dust transport directions are slightly shifted to the north with increasing altitude. In section 4.2 a more detailed structure of dust transport will be presented.

[51] The highest carbonaceous AOT region is located between south China (as a result of high relative humidity) and Southeast Asia (as a result of intensive biomass emission). The same result is seen in the simulated boundary layer field, but the distribution differs greatly in contrast to dust and sulfate. It is predominant around the line of 30°N. This fact indicates that the carbonaceous transport direction is mainly eastward in the free atmosphere. Kaneyasu and Murayama [2000] reported that a great increase in the black carbon level was observed routinely around 30°N in research cruises between 1993 and 1996. This observational fact is consistent with our results.

[52] It is shown that the total AOT has a very complex distribution. However, the model analysis presented above indicats that this complex distribution is the result of different meteorological conditions, aerosol source regions, emission strengths, and transport processes. Consequently, dust is the most predominant aerosol over the northern part of the modeling domain. Sulfate also dominates between the industrialized regions in China to the eastern side of Japan. Carbonaceous aerosols caused by intensive biomass burning events are the most important aerosol type over Southeast Asia. They are dominant between that region and the southern side of Japan. These characteristics are confirmed by latitudinal variations of averaged AOT for each component at 130°E as shown in Figure 9. AOT for dust, sulfate, and carbonaceous aerosols have peaks at 41°N, 38°N, and 27°N, respectively. Carbonaceous AOT is higher in the southern part of the modeling domain, whereas the mineral dust fraction is higher in the northern domain. Sulfate's high AOT region is located mainly between 25°N and 45°N.

Figure 9.

Latitudinal variation of the averaged model AOT for main tropospheric aerosols at 130°E.

[53] Finally, mean contributions of each component to total AOT were evaluated. Contributions to total AOT were 36% for carbonaceous aerosols, 25% for sulfate, 24% for dust, and 15% for sea salt aerosols. Importantly, carbonaceous aerosols and sulfate of which the emissions are almost dominated by the human activity contribute largely to total AOT. Consequently, they exert a strong impact on the atmospheric environment and its radiation field.

4.2. Aerosol Transport Pattern

[54] Figure 10 shows the two-month averaged aerosol transport flux fields. The left column and middle column in Figure 10 show the vertically averaged horizontal flux (by vectors) of dust, sulfate and carbonaceous aerosols and their magnitudes (by color) for the boundary layer and for the total column, respectively. Here the horizontal transport flux (HFLXx and HFLXy) is calculated as

equation image

where u and v are the wind speed in the x, y directions, Caero is the aerosol concentration, and T is the averaging period. Thus the transport flux (HFLX) magnitude is given as

equation image

The mean cross section of the longitudinal transport from the west to east across longitude 130°E is also shown in Figure 10 in the right column.

Figure 10.

Boundary layer (below 1000 m) averaged horizontal mass flux (left column), the column averaged horizontal mass flux (middle column) and the latitudinal cross section of the eastward horizontal mass flux at 130°E (right column) during spring 2001 for mineral dust, sulfate and carbonaceous aerosols.

[55] Horizontal fluxes of dust are highest in the Chinese desert regions because dusts are emitted into the atmosphere by high surface winds. The main export pathway for dust is to the Pacific along the 45°N parallel in westerly flow (middle column in Figure 10). The latitudinal cross section of the horizontal flux shows that the region of highest dust flux ranges from 2 to 4 km altitude and is centered at 43°N, indicating that the active height for dust transport is the lower free atmosphere. Iwasaka et al. [1983] reported that the altitudinal range of mineral dust transport over Japan has been observed to be from 2 to 6 km in most cases. This fact concurs with our results. Another feature for dust transport is that surface level dust shows two main transport directions from their source regions: eastward transport and southeastward transport. In the latitudinal cross section at 130°E, this characteristic is also confirmed, as shown by R1 and R2. As spring comes, the strength of the Siberian high and Aleutian low pressure systems decays, while a high-pressure system is established over the central Pacific Ocean, bringing warmer tropical air from the south. During March and April, the Pacific anti-cyclone increasingly develops and frequency of the incursion of the warm air mass to the north increases. Therefore this occurs because the mean wind direction from the Yellow Sea to the Japan Sea shifts northward by enhancement of southerly inflows of air as a result of the Pacific anti-cyclogenesis. Thus these two ridges indicate representative transport directions for March and April: southeastward during March and eastward during April.

[56] Surface horizontal fluxes for sulfate and carbonaceous aerosols show a clockwise and divergent flow pattern over central China. This flow pattern in central China is closely related to the outflow of sulfate and carbonaceous aerosols. The northern edge of its divergent zone is located in northeastern China and its southern edge is located around southern China and Southeast Asia. The location of the north edge also corresponds to the high anthropogenic emission area, producing a strong westerly outflow pattern associated with anthropogenic pollutants (left column in Figure 10). We can clearly confirm the presence of its outflow for the latitudinal cross section at 130°E. We also find that its outflow is mainly dominant within the boundary layer. Kim et al. [2001] found that sulfur transport to the Yellow Sea is confined mainly in the boundary layer, which concurs with our analysis. The divergent zone southern edge also plays an important role in aerosol transport. It constrains the Asian continental outflow because this edge comprises easterly flows from the Pacific region. Therefore we conclude that the divergent wind pattern in central China is an important factor for aerosol transport within the boundary layer, driving a strong outflow containing anthropogenic pollutants at the northeastern China, while driving an inflow from the Pacific at more southerly latitudes.

[57] For the total column averaged horizontal carbonaceous flux field, a prominent westerly flow pattern is shown at the southern latitude in the modeling domain. This contrasts sharply with the boundary layer case. Its outflow corresponds to the southern peak in the latitudinal cross section in Figure 10. Its magnitude is comparable to or larger than that within the boundary layer. The active convection over the subtropical region produces an upward flux that lifts the pollution into lower free atmosphere where it is caught by the strong westerly, resulting in such a prominent flow pattern. In addition, Uno et al. [2003b] reported that biomass burning emission in the Southeast Asia also plays an important role for formation of this flow pattern; its contribution in this outflow exceeds 80%. We infer that carbonaceous aerosols have two main transport paths bounded by the boundary layer. One in the boundary layer between 35°N and 45°N is associated with anthropogenic aerosols. The other, in the lower free atmosphere between 20°N and 30°N, is dominated by biomass burning aerosols.

4.3. Regional Budgets

[58] Model results can be integrated to provide further insight into the transport of dust, sulfur, and carbonaceous aerosols during spring 2001. Figure 11 illustrates the regional budgets for each component schematically. It is notable that the outflow directions are considered in the RAMS/CFORS polar- stereographic projection and that SO4 concentration level is converted into g-SO2. Total emissions during the simulation period are 105 Tg for mineral dust, 8.3 TgSO2 for sulfur, and 3.07 Tg for carbonaceous aerosols. Total emission of mineral dust is equivalent to 3.5–10.5% of the global annual production of windblown dust reported by the Intergovernmental Panel on Climate Change (IPCC) [1996]. It is also an important feature that anthropogenic emissions of sulfur and carbonaceous aerosols constitute more than half of their total emission.

Figure 11.

Regional budgets for (a) mineral dust, (b) sulfur, and (c) carbonaceous aerosols. Emission, deposition, and outflow are shown by arrows in Tg (Tg-SO2 for sulfur cycle). Numbers in parentheses under each deposition and outflow indicate the ratio to emission.

[59] Dry deposition, gravitational settling, and northward outflow of mineral dust occupy 33%, 27%, and 13% of the total emission, respectively. Wet deposition, eastward outflow, and dry deposition of sulfur occupy 32%, 27%, and 21%, respectively. In the case of carbonaceous aerosols, the outflow to the east has the highest fraction (49%), followed by dry deposition (16%) and the outflow to the north (14%). Dust and sulfur total deposition processes play a major role in their transport processes, whereas carbonaceous deposition processes comprise a small fraction of the total emission in this simulation because of a set low deposition rate and of the largely transport in the free atmosphere. Overall, eastward and northward outflows are relatively prominent for all components when compared to the westward and southward, which reflects their transport direction. Particularly, the eastward outflow for carbonaceous aerosols has a larger fraction because of its characteristic transport pattern shown in Figure 11. This result indicates that a large amount of carbonaceous aerosols are exported into northern Pacific regions and cause a larger radiative impact.

5. Conclusions

[60] The transport of the main tropospheric aerosols (mineral dust, sulfate, black carbon, and organic carbon) in east Asia during spring 2001 was studied using the chemical transport model coupled with a regional-scale meteorological model (RAMS/CFORS). To understand the impact of aerosols on atmospheric radiation, AOTs for each component are also calculated. This simulation period contains an ACE-Asia intensive observation period and also corresponds to the period in which aerosols are frequently transported over east Asia. Sensitivity simulation without volcanic emission was performed during the same period to clarify impacts resulting from volcanic and anthropogenic sulfates on the atmospheric environment.

[61] Numerous comparisons of model results with observations clarified model performance and aerosol transport characteristics. Further analyses for aerosol transport with its concentration, AOT, horizontal flux, and budget analysis were performed to produce an overview of aerosol transport and its radiative impact in the springtime 2001. Our main findings are the following:

[62] 1. Comparisons of lidar observation with simulation results indicated that the model captured characteristics of dust and sulfate transported to Beijing. Most of the high extinction regions for sulfate are limited to relatively lower altitude levels than that for dust. It was also shown that dust and sulfate transports have a strong correlation. The model reproduced episodic dust transport, on the average, contributes to about 60% of the total extinction coefficient, indicating a strong impact of dust transport on atmospheric radiation at Beijing.

[63] 2. Further comparisons with observed surface concentration of PM10, sulfate, and TC show that the simulation has general agreement with them and reproduces many prominent features of aerosol transport over east Asia. In that comparison, temporally averaged values for both model and observation results during April indicated the presence of a latitudinal gradient of aerosol concentration field along the 140°E meridian, except for dust.

[64] 3. The modeled total AOT were compared with the TOMS AI over both land and ocean: close relationships between two temporal variations were then shown at all regions (northeastern China, northern Japan, Japan Sea, and southern Japan). From these results we found that dust transport is more frequent as the spring proceeds and that AOT field of dust and carbonaceous aerosols has an opposite gradient each other.

[65] 4. Analysis of the two-month averaged concentration field below 1000 m shows the latitudinal gradients more clearly. The main dust concentration field is located in more northern latitudes (between 30°N and 45°N), while sulfate and carbonaceous field are mainly dominant from their main sources in central China and Southeast Asia to northern Japan (between 25°N and 45°N). The volcanic sulfate fraction is also evaluated and contributed about 13% of the total sulfate concentration, indicating that the anthropogenic sulfate mainly dominates over its concentration field within the boundary layer.

[66] 5. Comparisons of the concentration field within the boundary layer with the AOT field show that carbonaceous AOT have largely different distribution; carbonaceous AOT dominant from Southeast Asia to southern Japan along the 30°N parallel. Finally, mean relative contributions to the total AOT are estimated to be 36% for carbonaceous aerosols, 25% for sulfate, 24% for dust, and 15% for sea salt aerosols. We conclude that carbonaceous aerosols and sulfate, which mainly emitted by the human activity, largely contribute to the total AOT.

[67] 6. Analyses of two-month averaged aerosol horizontal mass flux clearly show aerosol transport pattern and we found that they are closely related to characteristics of wind field of springtime. Dust has two main transport directions from its source region (northeastward and eastward) within the boundary layer; this transport direction shifts to a more northerly latitude as the strength of the Pacific High increases. On the whole, the main transport path of dust is eastward along the 45°N parallel and its main transport altitude corresponds to the 2–6 km from latitudinal variation of the eastward horizontal flux. Transport patterns for sulfate and carbonaceous aerosols in the boundary layer show a clockwise and divergent flow pattern, with the flow center located over southern China. Those patterns produce the strong outflow associated with anthropogenic emissions at northern latitudes (they are the main transport pathway for sulfate and carbonaceous aerosols) and constrain the continental outflow at the southern latitude. Further analysis shows that carbonaceous aerosols have another transport pathway at an upper level (2–6 km), which is along about 30°N from Thailand and Laos.

[68] 7. We performed a regional budget analysis for dust, sulfur, and carbonaceous aerosols. Total emissions during the simulation period are 105 Tg for mineral dust, 8.3 Tg-SO2 for sulfur, and 3.07 Tg for carbonaceous aerosols. Dry deposition, gravitational settling, and northward outflow of mineral dust occupy 33%, 27%, and 13% of the total emission, respectively. Wet deposition, eastward outflow, and dry deposition of sulfur occupy 32%, 27%, and 21%, respectively. In the case of carbonaceous aerosols, the outflow to the east has the highest fraction (49%), followed by dry deposition (16%) and the outflow to the north (14%).

Acknowledgments

[69] This work was partly supported by Research and Development Applying Advanced Computational Science and Technology (ACT-JST) and Core Research for Evolution Science and Technology (CREST) of Japan Science and Technology Corporation (JST). This work was also supported in part by grants from the NSF Atmospheric Chemistry Program, NASA ACMAP and GTE programs, and the NOAA Global Change program. EANET PM10 observation data were provided from the Acid Deposition and Oxidant Research Center, Niigata, Japan. This research is a contribution to the International Global Atmospheric Chemistry (IGAC) Core Project of the International Geosphere Biosphere Program (IGBP) and is part of the IGAC Aerosol Characterization Experiments (ACE).

Ancillary

Advertisement