THE LONG RANGE TRANSPORT MODEL WITH DEPOSITION AND TRANSFORMATION COMPONENTS AND APPLICATION TO THE EAST ASIAN REGION

-----A long-range sulfur transport model consisting of two submodels - a meteorological submodel and a dispersion submodel - has been developed to evaluate the extent of acid deposition in the East Asia. The Japan Meteorology Agency's operational weather forecasting model was adapted and improved upon, and employed to predict meteorological variables. A Lagrangian particle method was used as the basis for the dispersion model, and a random walk calculation was used to describe the diffusion process. Dry deposition, below-cloud scavenging and chemical transformation processes are also included.

A numerical simulation of the transport of sulfur oxides in the East Asia region in 1985, and the annual deposition evaluated. The results of the simulation showed that the wet deposition of sulfur oxides originating from neighboring countries was roughly 0.05gS/m²/year, this value was about 10 times lower than wet deposition of sulfate values observed in Japan. In an evaluation of the impact on deposition in Kita-Kyushu (in western Japan), it was shown that emission from Korea, Taiwan and south-eastern region of China affected the Kita-Kyushu values. However, in an evaluation of the impact on deposition in Niigata (on the Sea of Japan coast), it was shown that emission from China dominated.

Energy is being consumed in enormous quantities in East Asia as a result of rapid industrialization in the area. More than 23 million tonnes/year of sulfur oxides are emitted into the atmosphere, thus acid deposition is becoming a serious problem. In East Asia, however, no international monitoring network for acid deposition have been developed yet. Therefore it is very difficult to estimate the total amounts of acid deposition. One way to circumvent this difficulty is by using numerical models. The first objective of this study was to develop a numerical model of long-range transport, including deposition and chemical transformation of pollutants, which was applicable to the East Asia region. The second objective was to estimate the extent of dry and wet deposition of acid substances in the region using the numerical model. Numerous transport models which include deposition processes have been developed (Venkatram et al., 1982, Liu and Stewart, 1982, Shin et al., 1992), however, most of these models employ only observed meteorological data. One transport model which includes a module to actually predict meteorological variables, is that of Kimura et al. (1988). It was used to simulate the transport of radioactive pollutants from the accident at the Chernobyl nuclear power plant using global wind fields obtained from the Japan Meteorological Agency's operational-use Global Scale Spectrum Model (GSM). The EURAD (European Acid Deposition) Model includes a model which simulates meteorological fields, the NM4 (Mesoscale Model Version 4) (Mˆlders et al., 1994). Also Rolph et al. (1993) used preprocessed meteorological fields simulated by the U.S. National Meteorological Center's Nested Grid Model.

In the present study, a numerical model, instead of observed data, is employed to predict meteorological variables. This is especially important in East Asia because the greater part of the East Asian region is the ocean, where there are no available meteorological data.

The MRI (Meteorological Research Institute) long-range transport model consists of two main submodels; a meteorological model which predicts meteorological variables, and a dispersion model which includes processes of advection, diffusion, deposition and chemical transformation of SO₂ to SO₄^2-. Each component is explained bellow.

2.1. Meteorological Submodel

The meteorological submodel consists of 73 grids in the east-west direction and 55 grids in the north-south direction using a polar stereographic projection plane at 60°N. The horizontal model domain with topography is shown in Fig. 1.

Fig. 1. Horizontal model domain with topography.

The grid size is about 127 km at 60°N. The model was the s-coordinate system with 16 layers in vertical as shown in Fig. 2.

Fig. 2 Vertical model domain showing s-coordinates of layers.

The model is formulated in terms of the primitive equations for momentum, mass, specific humidity and virtual temperature using flux forms. The equation of motion consists of advection, vorticity, pressure gradient force, and subgrid scale horizontal and vertical diffusion terms, and is expressed as:

where, m is the map scale factor defined on a polar stereographic projection plane at 60°N as

and u^*=(pi)u/m, v^*=(pi)v/m, f is the Coriolis parameter, f is the geopotential, q is the potential temperature, C_p is the specific heat of air at constant pressure, and t is the Reynold's stress due to vertical wind shear.

The thermodynamic equation is written as:

where, Q is the heat budget due to diabatic heating or cooling.

The equation of specific humidity is:

were, M is the reduction of water vapor per unit mass by condensation and precipitation. H in the thermodynamic equation, and E in the equation of specific humidity, are vertical fluxes of the sensible heat and water vapor due to turbulent eddies smaller than the grid size, and F_v, F_T, F_q are subgrid scale horizontal diffusion for momentum, potential temperature, and humidity, respectively.

The continuity equations can be represented as follows:

Integrating the above equation from s=0 to s=1, and assuming as a boundary condition that the vertical velocity is zero at the top and bottom of the vertical domain of the model, then we can simplify the equation as follows:

This equation gives the time variation of the surface pressure. Integrating Eq. (5) from s=s to s=s+ds, we can derive the vertical velocity for any layer of the model atmosphere:

The hydrostatic equations are:

where, a is defined by the equation of state for the atmosphere as follows:

where T and R are temperature and the gas constant, respectively, and q, s, and p are defined as follows:

were, P_s and P_t are the pressure at the ground surface and top model layer of the atmosphere, respectively. To ensure that air does not pass across the planes at the top and bottom layers of the model atmosphere, the following boundary conditions are given:

The Monin-Obukhov similarity theory and the turbulent closure model of level 2 (Mellor and Yamada, 1974) are used to model physical processes in the surface layer and the Ekman layer, respectively. The Japan Meteorology Agency operational global objective analysis data sets (GANL) are used for initial and boundary conditions.

2.2. Dispersion Submodel

The dispersion model for sulfur oxides is based on the mass conservation of sulfur oxides as follows:

where, CSO₂ and CSO₄^2- are the ambient air concentrations of each chemical species in the atmosphere, QSO₂ and QSO₄^2- are the emission terms, At is the transformation rate of SO₂ to SO₄^2-, D_d2 and D_w2 are the dry and wet deposition rates for SO₂, and D_d4 and D_w4 the dry and wet deposition rates for SO₄^2-.

2.2.1. Advection and Diffusion

A random walk method applied to a Lagrangian particle is employed to describe vertical diffusion in the dispersion model. The three-dimensional movement of a particle is calculated using the following equations:

where, X, Y and s are positions of a particle in a three-dimensional coordinate system. A random force R is defined as:

where, K_z is the vertical diffusivity derived from the meteorological submodel, dt is the time step in the random walk model, and the sign is chosen randomly for each particle and each time step. The vertical diffusivity can be derived under the assumption that:

where, K_m is the vertical turbulent diffusivity for momentum, K_o is assumed to be equal to 1.0 m²/sec, and l denotes the mixing length scale which is determined according to Blackadar's (1962) interpolation formula:

where, k is Karman's constant, and the asymptotic mixing length l_o is an adjustable parameter. Mellor and Yamada (1974) define l_o by taking the ratio of the first to the zeroth moment of the turbulent energy profile q(z) as follows:

where, the turbulent energy is defined as

Louis (1979) takes l_o=100 m. In the present model l_o is assumed to be equal to 300 m. S_m is a function of flux Richardson number, R_f, which is defined by Mellor and Yamada (1974) as:

where, V=u²+v², q is the potential temperature, and g the acceleration due to gravity. The vertical diffusivity is calculated in the meteorological submodel.

The random walk model gives a much higher resolution than a grid method having the same grid size as the Japan Meteorology Agency's older version of their weather forecasting model, FLM (Fine-mesh Limited-area Model). If the number of particles in an averaged volume of air is small, the statistical error becomes large, therefore a large number of particles must be released from the emission source. The vertical position of a particle at level z and time t is z(t), the position of a particle after dt is extrapolated by the equation:

If the vertical diffusivity is constant, the number of particles after a long enough time integration will be equal to the analytical solution of the diffusion equation. When the vertical diffusivity depends on height, however, an artificial vertical flux of concentration appears even though the vertical gradient of concentration is zero. Even if the time step is very small, this error will not become small. To avoid this error, Diehl et al. (1982) added a correction term to Eq. (13) so that the time extrapolation gives an exact flux when K_z is a linear function of height. However, when K_z depends on height, the computation error here tends to be large. Thus, the following scheme is used in this study for time integration of the vertical diffusion. An estimated position of a particle at t+dt is made:

The height of a particle at t+dt is estimated by using z^* as follow:

In using this scheme, the error will become small if dt is sufficiently small.

2.2.2. Dry Deposition

A particle is assumed to be dry deposited if following conditions are satisfied- the height of the particle is lower than a prescribed critical level H_dep=0.99s in the vertical coordinate, and a number given randomly for each particle and each time step is smaller than a value P_ddep defined as:

where, V_d is the dry deposition velocity.

The dry deposition velocity varies with meteorological condition and underlying ground surface. Stewart et al. (1983) suggest a value of 5.0*10^-3 m/sec for a lake or ocean surface, 3.0*10^-3 m/sec for a metropolitan city, and 2.0*10^-3 for cropland, woodland, or grazing land. On the other hand, Waleck et al. (1986) estimate and parameterize the sublayer and surface resistance for the dry deposition of particulate sulfate by eddy correlation measurements which were carried out by Wesely et al. (1985). They show that the dry deposition velocity for SO₂, SO₄^2-, and NO₃ varies with type of ground surface.

We, therefore, make the simple assumption that the dry deposition velocity of SO₂ and SO₄^2- varies only with two kinds of the ground surface conditions, as shown in Table 1.

2.2.3. Wet Deposition

Only below-cloud scavenging is considered for wet deposition, because of the inability of FLM to simulate cloud processes. Wet deposition is calculated every hour, and particles are deposited on the ground surface with a probability of:

where, Dw is the wet deposition rate of either SO₂ or SO₄^2-, dt is the duration over which the wet deposition evaluated, and RR is a precipitation index, which is defined as:

• for particles at grid nodes
  RR=1; if the predicted precipitation rate exceeds a threshold value, defined as 0.1 mm/hour.
  RR=0; otherwise,

• for particles between grid nodes
  RR is interpolated horizontally using the RR values at the grid nodes around the particle.

Below cloud scavenging rates have been suggested in several studies (Eliassen, 1982; Fisher, 1984), however, values are scattered between 10^-5/sec and 10^-3/sec. Therefore, we assumed them to be D_w=3*10^-5/sec for SO₂, and D_w=1*10^-4/sec for SO₄^2- for precipitation rates larger than 0.1mm/hour.

2.2.4. Chemical Transformation

The only chemical transformation in this model is that of SO₂ to SO₄^2-. The transformation rates in the ambient atmosphere derived from the literature spread over a range of over an order of magnitude. For example, Cox (1974) estimates 0.01~0.1/hour in an urban plume for photo-oxidation with NO_x and hydrocarbons, or thermal oxidation with ozone and olefin. Eliassen and Saltbones (1975) estimate a transformation rate of about 0.007/hour. The estimation by Alkezweeny and Powell (1977) from atmospheric concentration data measured by aircraft yielded 0.10~0.12/hour. A rate of 0.01/hour was adopted in the present model, and the production of SO₄^2- due to the transformation of SO₂ to SO₄^2- is restricted to 90% of the original SO₂ concentration in a given Lagrangian particle. The mixing ratio of both species in an unpolluted atmosphere, such as the Antarctic, sub-Antarctic, and south Pacific ambient atmosphere are 0.08, 0.11, and 0.10, respectively (Nguyen et al., 1974), so it is assumed the maximum ratio of SO₄^2- to SO₂ is 0.9 if there is no more supply of SO₂ in the ambient atmosphere.

In order to valid the transport model, a simulation of sulfur oxide transport and deposition in North America was performed. Emission inventory and the observed wet deposition data were provided by U. S. Environmental Protection Agency. The emission inventory came from the National Acid Precipitation Assessment Program (NAPAP). A calculation was carried out for 94 days in winter (i.e. the winter quarter defined as 4 December 1984 through 5 March 1985). The calculated wet deposition for SO₄^2- was compared to observed data from the National Atmospheric Deposition Program and National Trend Network (NADP/NTN).

Fig. 3. Observed SO₄^2- wet deposition in g/m² during the winter quarter defined as the period 4 Dec. 1984 through 5 Mar. 1985.

Figure 3 shows the observed distribution of wet deposition of SO₄^2- during the winter quarter. The symbols on the map are the location of NTN stations. Large deposition values of SO₄^2- are in Ohio, Indiana and Kentucky, where the largest emissions are located. Figure. 4 shows the calculated distribution of wet deposition of SO₄^2- during the same period. The patterns between the observed and calculated distributions are roughly similar except for deposition in Oklahoma and the northern part of Texas.

Fig. 4. Same as Fig. 3, but for calculated SO₄^2- wet deposition.

Because pollutants are scavenged under a probability related to precipitation larger than a selected threshold value in the model, as described above, the area of precipitation becomes the most important factor in wet deposition distribution. Severe local storms break out frequently at this area, however, the precipitation caused by these local storms is not readily captured by the FLM with a grid size of 127 km. Figure 5 shows a scatter diagram for the observed and calculated wet deposition of SO₄^2- at each observation station. Here, the calculated deposition is taken as the average in a quarter of a grid (approximately 31.8*31.8 km). The model underestimates SO₄^2- deposition by about 30%. The significant underestimation in the Oklahoma-Texas area is readily apparent as the square symbols in Fig. 5. The reason for underestimation is thought to be due primarily to an underestimation in predicted precipitation. The threshold of precipitation was set to 0.5 mm/hour in the North American validation run. However, the model threshold value was reset for East Asia to 0.1 mm/hour after making score analysis.

Fig. 5. The scatter diagram of observed and calculated S042- wet deposition at each station. The square symbol shows the data for Oklahoma and Texas.

4.1. The Emission Inventory of Sulfur Oxides

The anthropogenic emissions for sulfur oxides for 25 Asian countries east of Afghanistan and Pakistan have been calculated for the years 1975, 1980, 1985, and 1987 based on the fuel consumption (Kato et al., 1991), Kato and Akimoto (1992), and the emission transformed to 1**1* inventory by Akimoto and Narita (1994). In addition to the 25 whole countries, province and region-based calculations were made for China and India. The emission inventories for sulfur oxides in the East Asia used in present study are based on the 1985 inventory in Kato et al. (1991). These emissions were redistributed to 80 point sources based on industrial activities and population of the cities. The redistributed emissions, excluding Japan and the Far East of Russia, are shown in Table 2. (Note: Table 2 contains only 66 out of 80 point sources relevant to the present simulation runs, i.e., the Japanese and Russian point sources are not include). Volcanic and other natural emissions are also not included.

4.2. Simulation and Results

Simulation was performed for the year 1985. In order to evaluate the degree of deposition of sulfur oxides originating from neighboring countries, the calculation is carried out excluding Japanese domestic emissions. Since, there are no reliable sulfur oxides emission inventories for Far East of Russia, it was not possible to include these in the calculations.

Figure 6 shows the location in Japan of wet deposition observation stations and receptors area used for the comparison. The black squares with code number ranging from OB1 to OB14 express the observation stations, and the shadowed areas with code numbers from CA1 to CA10 expressed the receptors in which calculation depositions are averaged. The shadowed square corresponds to a quarter of a grid and the black circles in the receptor area indicates the representative cities. The wet deposition of SO₄^2- was observed at 14 stations in 7 pairs, one urban, the other rural or suburban. The calculated annual dry and wet depositions of sulfur oxides at 10 areas are shown in Fig. 7(a).

Fig. 6. Location of the observation stations and the receptors for evaluation of calculated deposition.

These are the average in a quarter of a grid. The deposition of SO₂ and SO₄^2- are expressed in terms of elemental sulfur. The calculated annual dry deposition tends to be smaller toward the receptor in the eastern part of the country. The calculated wet deposition has no tendency to vary with the location of receptor. Observed wet deposition of SO₄^2- expressed as sulfur equivalent is shown in Fig. 7(b). The amount of annual precipitation decrease from west to east, however the observed wet deposition does not vary with the location of observation station. This difference can be inferred to be due to the greater concentration of domestic emissions on the eastern coast of Japan. Four representative cities in which paired observation station over-lap model receptor areas are Sapporo, Tokyo, Nagoya and Osaka. By comparing the calculated wet deposition of SO₄^2- at CA5(Osaka), CA6(Nagoya), CA8(Tokyo) and CA10(Tomakomai), and the observed wet deposition at OB5, OB6(Osaka), OB7, OB8(Nagoya), OB9, OB10(Tokyo) and OB12, OB13(Sapporo) - the two stations of OB12 and OB13 are located completely inside of the receptor area of CA10 - it is clear that the calculated wet depositions of sulfur at these four receptors is around 0.05 gS/m²/year. Because these locations are widely separated it can be concluded that significant regional differences do not appear in the calculated wet deposition for these areas.

Fig. 7. Calculated annual deposition of sulfur oxides (SO₂+SO₄^2-) expressed as elemental sulfur at 10 receptors (shown in Fig. 6) in 1985 (a). The observed wet deposition of nss-SO₄^2- expressed as elemental sulfur at 14 observation stations from April 1985 through March 1986.

The observed wet depositions of sulfur are 0.6~1.3 gS/m²/year for these stations. If the model, as validated against North America data, is assumed to be modeling long-range transport in East Asia reasonably accurately, then it may be inferred that the differences between the observed and calculated deposition may be due to contribution by domestic emissions. If this is true then it can be concluded that the deposition originating from continental anthropogenic emission is on the order of only several percent.

Fig. 8. Contribution of sulfur oxides to dry deposition at the Kita-Kyushu receptor originating from continental emissions. (a) January and (b) July.

The Lagrangian model can also be used to determine source-receptor relations. The source origin of sulfur oxides affecting deposition at Kita-Kyushu located in western Japan was investigated. The Kita-Kyushu receptor was chosen because it had the largest total calculated deposition. Figure 8 shows the relative contribution of continental emission sources to dry deposition at Kita-kyushu in January (a) and July (b) of 1985, and Fig. 9 shows the same for wet deposition. The emission number of abscissa is the same as Table 2. The atmospheric pressure system is quit different between winter (January) and summer (July) in the East Asia. Migratory anticyclones frequently pass across the east Asian continent during winter season, and strong northwesterly wind prevail. Because of these conditions, the contribution from South Korean emission sources to dry deposition during the winter was calculated to be remarkably high. The contribution to wet deposition from Pusan, Teagu (South Korea) and several emission sources in China was high during the winter.

Fig. 9. Same as Fig. 8, but for wet deposition.

Usually, an anticyclone persistent in the Pacific Ocean in summer, and weak southerly winds prevail over Japan. However, in the first decade of July in 1985, a cyclone passed slowly through the Sea of Japan, and the northwesterly winds continued. Thus, the contribution to the dry deposition from Pusan and Teagu (South Korea) was calculated to be high during the summer. As seen from Fig. 9(b), the major contributive emission sources to wet deposition during the summer were Pusan and Taegu (South Korea), Taipei and Gaoxing (Taiwan), and Changsha, Nanking and Shanghai (China). It is surprising that a slight contribution from Uttaradit (Thailand) is also seen.

Fig. 10. The relative contribution from each country at Kita-Kyushu receptor, (a) and (b) for dry deposition in January and July, (c) and (d) for wet deposition, respectively.

The relative contribution of each country to dry deposition at Kita-Kyushu is shown in Fig. 10. The contribution from South Korea for dry deposition was remarkably high in both January and July, but a high contribution from China was seen for wet deposition.

The source origin of sulfur oxides affecting deposition at Niigata located central Japan and facing the Sea of Japan was also investigated. The Niigata receptor was chosen because the contribution from mainland Asia emission sources is considered to be largest on the Sea of Japan coast, especially in winter. Figure 11 shows the relative contribution of continental emission sources to dry deposition at Niigata in January (a) and July (b) of 1985. The contributive emission sources are quite different from the case of Kita-Kyushu, where most of the contribution to dry deposition was from South Korea. At the Niigata receptor, the emission sources in China dominate the dry and wet deposition contributions except dry deposition in July. The contribution from emission sources in South Korea (Pusan, Taegu and Taejon) is notable to dry deposition, and contribution from Taiwan (Gaoxing) and several emission sources in the southeast portion of China (Nanking and Wuxi) are observed in July.

Fig. 11. Contribution of sulfur oxides to dry deposition at Niigata receptor originating from continental emissions. (a) January and (b) July.

As seen in Fig. 12, the contribution to wet deposition from South Korea is relatively low in both January and July. There are contributions from several emission sources in the southeast part of China in January. this pattern is almost the same as the case of dry deposition in January, but the source location shifts from Jinan to Chengdu. A significant contribution from Nanking and Shanghai is seen in July.

Fig. 12. Same as Fig. 11, but for wet deposition.

The major contribution to dry deposition came from South Korea, and the major contribution to wet deposition came from China at Kita-Kyushu (see Fig. 10). On the other hand, at Niigata receptor, most of the contribution is from China (see Fig. 13).

Fig. 13. Same as Fig. 10 but for the Niigata receptor.

A numerical model of long-range transport have been developed to evaluate acid deposition in East Asia, and a numerical simulation of the impact of continental emissions on Japan's dry and wet deposition was performed. The results of the simulation showed that the wet deposition of sulfur oxides originating from neighboring countries was roughly 0.05 gS/m²/year, and the dry deposition 0.02~0.17 gS/m²/year. This value was about 10 times lower than wet deposition of sulfate values observed in Japan. Calculated wet deposition values at four representative receptors were found to be much lower than observed values. Therefore, it was inferred that the contribution from foreign emission was extreme low, on the order of several percent. In an evaluation of the impact on deposition in Kita-Kyushu (in western in Japan), it was shown that emissions from South Korea, Taiwan and the south-eastern region of China affected the Kita-Kyushu values. However, in an evaluation of the impact deposition in Niigata (on the Sea of Japan coast), it was shown that emissions from China dominated.

Acknowledgments

This study was performed as a part of "Study to clarify the behavior of acidic and oxidative components in east Asia" under the Global Environment Research Program Budget of the Environment Agency of Japan.

The authors wish to express their gratitude to Dr. Terry L. Clark of Environmental Science Laboratory, U. S. Environment Protection Agency for providing the emission and deposition data in U.S.A.. They also wish to express their particular thanks to Prof. F. Kimura of the University of Tsukuba for providing the basis of the random walk model, and Dr. M. Nagata of the Numerical Forecast Division, Japan Meteorological Agency for providing the improved meteorological prediction model (FLM). Special thanks are extended to Dr. Kenneth Wilkening of the National Institute for Environmental Studies for his valuable comments on the original manuscript. The computation was performed on the HITAC M280/S810 system in Meteorological Research Institute.

Akimoto, H. and H. Narita, 1994: Distribution of SO₂, NO_x and CO₂ emissions from fuel combustion and industrial activities in Asia with 1**1* resolution. Atmos. Environ., 28, 213- 225.

Alkezweeny, A. J. and D. C. Powell, 1977: Estimation of transformation rete of SO₂ to SO₄ from atmospheric concentration data. Atmos. Environ., 11, 179-182.

Blackadar, A. K., 1962: The vertical distribution of wind and turbulent exchange in a neutral atmosphere. J. Geophys. Res., 67, 3095-3102.

Cox, R. A., 1975: Particle formation from homogeneous reactions of sulfur dioxide and nitrogen dioxide. Tellus, 26, 235-240.

Diehl, S. R., D. T. Smith and M. Syder, 1982: Random-walk simulation of gradient-transfer process applies to dispersion of stack emission from coal-fired power plant. J. Appl. Met., 21, 69-83.

Eliassen, A. and J. Saltbones, 1975: Decay and transformation rates of SO₂, as estimated from emission data, trajectory and measured air concentrations. Atmos. Environ., 9, 425-429.

Eliassen, A. F. Hov, I. S. A. Isaksen, J. Saltbones and F. Stordal, 1982: A Lagrangian long range transport model with atmospheric boundary layer chemistry. J. Appl. Met., 21, 1465- 1661.

Kato, N. Y. Ogawa, T. Koide, T. Sakamoto, S. Sakamoto and Research Group on the Energy Consumption in Asia and global Environment, 1991: Analysis of the structure of energy consumption and the dynamics of emission of atmospheric species related to the global environmental change (SO_x, NO_x, and CO₂) in Asia. NISTEP report No. 21, 4th polisy-oriented Research Group, the National Institute of Science and Technology Policy (NISTEP), Science and Technology Agency of Japan.

Kato, N. and H. Akimoto, 1992: Anthropogenic emissions of SO₂ and NO_x in Asia:Emission inventories. Atmos. Environ., 26A, 2997-3017.

Kimura, F. and T. Yoshikawa, 1988: Numerical simulation of global scale dispersion of radioactive pollutants from the accident at the Chernobyl nuclear power plant. J. Met. Soc. Japan, 66, 489-495.

Liu, M. K. and D. A. Stewart, 1982: A mathematical model for the analysis of acid deposition. J. Appl. Met., 21, 859-873.

Louis, J. R.. 1979: A parametric model of vertical eddy flux in atmosphere. Boundary-Layer Met., 17, 187-202.

Mellor G. L. and T. Yamada, 1974: A hierarchy of turbulence closure models for planetary boundary layer. J. Atmos. Sci., 31, 1791-1806.

M*lders, N., H. Hass, H. J. Jakobs, M. Laube and A. Ebel, 1994: Some effects of different cloud parameterization in a mesoscale model and chemistry transport model. J. Appl. Met., 33, 527-545.

Nguyen, B. C., B. Bonsang and G. Lambert, 1974: The atmospheric concentration of sulfur dioxide and sulfate aerosol over antarctic subantarctic areas and oceans. Tellus, 26, 243-249.

Rolph, G. D., R. R. Draxler and R. G. Pena, 1993: The use of model derived and observed precipitation in long-term sulfur concentration and dispersion modeling. Atmos. Environ., 27A, 2017-2037.

Shin, Woo-Chul and G. R. Carmichael, 1992: Analysis of wet deposition in the eastern United States. Atmos. Environ., 26A, 465-484.

Stewart, D. A., R. E. Morris, M. K. Liu and D. Henderson, 1983: Evaluation of an episodic regional transport model for a multi-day sulfate episode. Atmos. Environ., 17, 1457-1473.

Venkatram, A. B., B. E. Ley and S. Y. Wong, 1982: A statistical model to estimate long-term concentration of pollutants associated with long-range transport. Atmos. Environ., 16, 249-257.

Waleck, C. J. R., R. A. Brost and J. S. Chang, 1986: SO₂, sulfate and HNO₃ deposition velocities computed using regional landuse and meteorological data. Atmos. Environ., 20, 949-964.

Wesely, W. L., D. R. Cok and R. L. Hart, 1985: Measurement and parameterization of particulate sulfur dry deposition over grass. J. Geophys. Res., 90, D1, 2131-2143.