Atmospheric Deposition (Proceedings of the Baltimore Symposium, May 1989). IAHS Pub], No. 179. MODELING THE FORMATION AND DEPOSITION OF ACIDIC POLLUTANTS Chris J. Walcek Atmospheric Sciences Research Center, State University of New York, 100 Fuller Rd., Albany, NY 12205, USA Julius S. Chang Atmospheric Sciences Research Center, State University of New York, 100 Fuller Rd., Albany, NY 12205, USA ABSTRACT A sophisticated mathematical model of the chemistry, transport, and deposition of tropospheric trace gases provides a useful tool for assessing the relationship between emission and deposition of atmospheric pollutants. We have developed a three-dimensional Regional Acid Deposition Modeling system (RADM) which calculates shortterm concentrations and wet and dry deposition of trace species over the northeastern U.S. and Canada. Emissions of S0 2, nitrogen oxides, organic compounds, ammonia, and carbon monoxide are specified using a comprehensive pollutant emission inventory. The model calculates the transport and chemical transformation rate of these compounds and other secondary pollutants (e.g. ozone, sulfuric acid, nitric acid) using meteorology data and a detailed gas-phase chemical reaction mechanism. A cloud chemistry and scavenging model computes trace species aqueous chemistry and wet deposition rates during cloudy periods, and dry deposition rates to underlying surfaces are calculated for many chemical species. The model has been evaluated using precipitation chemistry observations, and several studies have been performed demonstrating the interaction between meteorology and chemistry of acid rain formation. MODEL OVERVIEW The RADM is an Eulerian trace-species transport, transformation, and removal model, described fully in Chang et al. (1987). The model subdivides the atmosphere over the northeast United States into six vertical levels, each with 900 horizontal grid cells approximately 80 x 80 km 2 in size. Figure 1 shows schematically the various components of RADM. The model is initialized with a set of chemical conditions, and provided with boundary conditions for these species during a simulation period of several days. For each grid cell, hourly meteorology and trace gas emission data are specified for a simulation period. From this information, the model calculates timevarying three-dimensional distributions of trace gases and particles, as well as temporal and spatial distributions of dry and wet deposition of numerous atmospheric trace species. The model solves a set of chemical species conservation equations ( \ ( \ ± = -VVC + V(K e VC)- r P chm -L chm +E+ -^ + < L (l) I /cloud I /dry where C is the species volume mixing ratio, V is the three-dimensional 21
22 Chris J. Walcek and Julius S. Chang velocity vector at each grid point in the model domain, K e is the eddy diffusivity used to parameterize subgrid-scale fluxes of trace species due to non-cloudy turbulent motions, P chm and L chm are the production and loss rates due to chemical interactions, E is the emission rate, (<9C/<9t) clouds is the time rate of change of concentration due to cloud effects (including subgrid-scale vertical redistribution, aqueous chemical interactions, and scavenging), and (<9C/<9t) dry represents the change in concentration due to dry deposition. In order to accurately model the formation and deposition of acidity, RADM solves Eq. (1) for 24 atmospheric chemical species needed to calculate the formation and deposition of tropospheric acidity. Chemical initial & i boundary conditions Regional Acid Deposition Model Trace gas..concentration & deposition patterns l Transport 3-d advection vertical diffusion I I I Cloud effects Dry deposition Gas chemistry» vertical redistribution > wet removal aqueous chemistry turbulent 1 sublayer surface resistances 36 species 77 reactions diurnal, seasonal, latitudinal, height varying photolysis rates with cloud effects Emissions nitrogen oxides sulfur oxides organics ammonia CO diurnal, seasonal, weekday/weekend variations point and area sources Figure 1 Overview of Regional Acid Deposition Model. MODEL APPLICATIONS The RADM has been used extensively for the analysis of acidic deposition in the northeast U.S. Middleton et al. (1988) discuss model calculations of trace species deposition in the northeast U.S. together with comparisons with observations. Walcek and Chang (1987), and Walcek (1987) present results of model calculations of dry and wet deposition of acids and oxidants (ozone, hydrogen peroxide) over the northeast U.S. Figure 2 shows the calculated forms of acidic deposition in the states of Ohio and New York during a three-day spring period surrounding the passage of a midlatitude cyclone. With RADM, it is possible to assess the relative importance of sulfuric and nitric acids in their contributions to acidic deposition. As Figure 2 shows, approximately 75% of acidic deposition in New York and Ohio can be attributed to sulfur compounds during this simulation period. Since the dominant sulfur emission regions over the northeast US are located near Ohio, the relative importance of sulfur deposition is more important there. The relative contribution of wet and dry nitric acid deposition was calculated to be slightly higher in New York than in Ohio. Figure 2 also shows that during this relatively rainy period, wet
Modeling Formation and Deposition 23 H NO. HNO. OHIO NEW YORK Figure 2 Model-calculated acidic deposition (equivalents m 2 ) of sulphate and nitrate in wet and dry forms over Ohio and New York during 22-24 April 1981. deposition of acidity dominated the total acidic deposition (wet + dry) for Ohio and New York. The relative contribution of wet deposition increases as one moves further from the S0 2 emission regions. MODEL EVALUATION Models of atmospheric chemistry must be evaluated using observations of precipitation and cloudwater chemistry to establish their credibility. Unfortunately, only limited field data are available to stringently evaluate a model's ability to accurately simulate a particular storm event. In any storm environment, there are numerous meteorological and chemical processes interacting in a complicated fashion, making it difficult to even measure those parameters necessary for a credible model evaluation. The Oxidation and Scavenging Characteristics of April Rains (OSCAR) field study is used for the model evaluation discussed here. During this study, a network of ~36 aqueous chemistry samplers (shown in Figure 3) collected precipitation over the northeast U.S. and Canada during four storm events in April 1981. In addition, a fine-resolution network of ~42 precipitation chemistry samplers collected rainfall in a ~100 x 100 km 2 area in northern Indiana to measure trace species deposition to an area comparable to the size of a single grid area in RADM. Figure 3 shows the calculated distribution of sulfate wet deposition over the northeast U.S. during 22-24 April 1981. Also shown on Figure 3 are the locations of the regional and fine-resolution network of precipitation chemistry samplers used for this evaluation. Before presenting results of the comparison of the model predictions over the entire model domain, we will first discuss the measured deposition for the gray area of Figure 3. Figure 4 shows a frequency distribution of the sulfate, nitrate and water deposition for the fine-resolution network of precipitation samplers in
24 Chris J. Walcek and Julius S. Chang Figure 3 Calculated sulfate wet deposition (yumol m~ 2 ), 22-24 April 1981. Triangles denote locations of precipitation chemistry samplers used to evaluate model performance. Shaded box denotes area where 42 samplers were located to study high-resolution deposition behavior. northern Indiana. Both the means and the medians of the sampled data are shown on the plots, and in all cases the median deposition is 10-20 % below the "average" deposition to the sampling area. This implies that for this storm event in this sampling area, a randomly selected site would be more likely to receive less deposition than the average deposition to the area. Also shown on Figure 4 are the limits encompassing 50% and 75% of the samples about the median. These data, although specific to this site for this storm event in northern Indiana, suggest that numerous sampling sites are required to measure deposition to a -100 x 100 km area. A single site in a -100 x 100 km area can only provide a rough estimate (75% chance of measuring between 0.6-1.5 times the average deposition) of the deposition to that area. In the light of the inherent subgrid-scale variability observed at one grid area, we now present comparisons between RADM calculations and observations over the northeast U.S. and Canada. Figure 5 shows the calculated and observed sulfate and nitrate wet deposition rates over the OSCAR regional precipitation sampling network. The regional-scale sampling network consisted of a single precipitation sampler at each location. Therefore, as demonstrated at the fine-resolution Indiana network, these data may not be representative of deposition to a larger area, as RADM is calculating. The gray areas of Figure 5 denote the region
Modeling Formation and Deposition 25 120 240 360 480 600 Deposition ((xmol m' 2 ) Deposition ( j.mol m" ) 0) \ *- mean ± standard deviation 0) o CD Si 3 C 8 12 16 rainfall (mm) 20 h@ EB9-4 T -median -50% 75% Figure 4 Frequency distributions of wet sulfate, wet nitrate, and water deposition for 42 precipitation chemistry sampling sites (representative of one model grid area) in northern Indiana during 22-24 April 1981. that should encompass 75% of the observations, assuming the subgrid-scale variability observed over the Indiana network is the same for all sites in the regional-scale sampling network. This assumption cannot be verified until further high-resolution precipitation sampling is performed. Comparisons between RADM and observations shown in Figure 5 demonstrate that RADM is capable of calculating observed short-term deposition rates to within the relatively large variability of the observed data. CONCLUSIONS We have developed a comprehensive mathematical model of the chemistry, transport, and deposition of tropospheric trace gases. This model provides a useful tool for assessing the relationship between emission and deposition of atmospheric pollutants, and also allows one to perform numerous diagnostic analysis of the formation and deposition of acidity during shortterm periods. The model has been evaluated using precipitation chemistry observations, showing that the model is capable of calculating deposition of acidity to within the observed variability of trace species deposition during individual precipitation events.
26 Chris J. Walcek and Julius S. Chang 1200 -.1000 I 800 ' 600 en a> 400- _eg 3 «200 m O (a) : i / w I/! I > m i <6 ; ; JL I ±t L! 9^ I..Q.-^p- j i i i i i i i ' 200 400 600 800 1000 1200 Observed SO^Cumolm" 2 ) 1200 E1000 o E o o Z S 400 - o 800-600 H 200 to O 200 400 600 800 1000 1200 Observed NO3 (jimol m" 2 ) Figure 5 Calculated and observed sulfate and nitrate wet deposition for the network of precipitation samplers shown in Figure 3 during 22-24 April 1981 storm event. Shaded area denotes limits encompassing 75% of the observed variability of deposition in a ~100 x 100 km 2 area using a single precipitation sampler as determined from the high resolution precipitation sampling in northern Indiana. REFERENCES Chang, J.S., Brost, R.A., Isaksen, I.S.A., Madronich, S., Middleton, P., Stockwell, W.R., & Walcek, C.J., (1987) A three dimensional Eulerian acid deposition model: physical concepts and formulation, /. Geophys. Res. 92, 14681-14700. Middleton, P., Chang, J.S., del Corral, J.C., Geiss, H., Rosinski, J.M., (1988) Comparison of RADM and OSCAR precipitation chemistry data, Atmos. Environ. 22, 1195-1208. Walcek, C.J. (1987) A theoretical estimate of 0 3 and H 2 0 2 dry deposition over the northeast United States. Aim. Environ. 21, 2649-2659. Walcek, C.J. & Chang, J.S. (1987) A theoretical assessment of pollutant deposition to individual land types during a regional-scale acid deposition episode. Attn. Environ. 21, 1107-1113.