Environ. Sci. Technol. 1993, 27, 2772-2702
Mathematical Modeling and Control of the Dry Deposition Flux of Nitrogen-Containing Air Pollutants Armlstead G. Russell
Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania
15213
Darrell A. Winner, Robert A. Harley, Kenneth F. McCue, and Glen R. Cass'
Environmental Engineering Science Department, California Institute of Technology, Pasadena, California 9 1125 ~~~
An Eulerian grid-based air quality model has been modified to include a resistance-based dry deposition code. The magnitude and spatial distribution of the dry deposition flux of nitrogen-containing pollutants to the surface of the Los Angeles area was calculated as a function of land use. For August 1982 base case conditions, the dry deposition flux was 247 t of N per day (5 from NO, 49 from and 26 NOz, 7 from PAN, 101 from HN03, 59 from "3, from NH4N03),which corresponds to more than half of the daily NO, emissions to the local atmosphere. The effects of emission controls on NO, and hydrocarbon sources in Southern California as they existed in 1982were examined. At the highest level of control studied (37% reactive hydrocarbon reduction, 61 % NO, reduction), the nitrogen dry flux would be 174 t of N per day after control (2 from NO, 20 from NOz, 7 from PAN, 58 €rom "OB, 75 from NHR,and 1 2 from NHdNOd. 1. Introduction
On a daily basis, approximately 1120t of nitrogen oxides is emitted to the atmosphere of the South Coast Air Basin that includes Los Angeles (I). Direct emissions to the atmosphere consist principally of nitric oxide (NO). Atmospheric photochemical reactions subsequently convert this original NO burden into an entire family of nitrogen-containing pollutants, including nitrogen dioxide (NOz), nitric acid ("031, peroxyacetyl nitrate (PAN), and aerosol nitrates. Eventually, these pollutants are removed by wet or dry deposition to the earth's surface. In the Los Angeles area, the dry flux of nitrogen-containing pollutants has been estimated to be the largest contributor to the local deposition flux of inorganic acids (2). The delivery of acids to surfaces has the potential to damage materials, degrade water quality, and may affect economically valuable forest areas and crops. The actual magnitudes of these effects are not well defined at present. Methods are needed that help to quantify the chemical nature and quantity of the pollutant deposition flux. For the past several years, research has been underway on the control of atmospheric concentrations of nitrogencontaining air pollutants. As an important step in that research effort, computer-based mathematical models have been developed and then tested in Southern California that compute the atmospheric concentrations of NO, HN03,PAN, "3, aerosol nitrate, and 0 3 from input data on reactive hydrocarbons, NO,, and NH3 emission patterns (1,3-6). Both trajectory models that track individual air parcels over time and grid-based airshed models that view all points in the air basin simultaneously have been created. In the present study, the grid-based version of the Caltech airshed model has been modified to compute the 2772
Environ. Sci. Technol., Vol. 27, No. 13, 1993
spatial distribution of the pollutant fluxes. The treatment of the surface resistance to dry deposition has been modified such that the deposition flux to particular types of ground cover (e.g., urban areas, grasslands, forests) can be computed for the various pollutant species for each type of ground cover. The model has been modified to read gridded land use data from input data files, and the new land use data have been used to improve both the deposition flux calculations and to calculate the surface roughness parameter values used within this model. A detailed data set on land uses within the Los Angeles area has been assembled, based on United States Geological Survey land use maps. Historical emissions, meteorological, and air quality data available in the Los Angeles area for the period August 30-31,1982, have been used along with the new land use data to test the characteristics of the modified airshed model against field observations. The spatial distributions of the deposition velocity and ofthe dry deposition flux have been computed. The air quality model discussed here utilizes information on the emissions from each air pollution source in the air basin. The model, therefore, can be used to predict the effect of candidate emission control measures in advance of their adoption. Following the determination of the historically observed dry flux of nitrogen-containing pollutants to surfaces, a study of the effect of a wide variety of emission control measures on achieving altered levels of pollutant. dry flux was undertaken. Mobile source controls considered include the effects of a vehicle inspection and maintenance plan, recent U.S. EPA proposals for future NO, reductions from heavy-duty vehicles, plus the possible conversion of the light-duty vehicle fleet to reflect attainment of either a0.7 g mi-l or 0.4 g mi-' NO, emission rate. Stationary source controls examined include future hydrocarbon and NO, controls expected as part of the Air Quality Management Plan (AQMP) for the South Coast Air Basin that was in force at the start of this research project, plus the possible addition of catalytic or noncatalytic NH3 injection technology for NO, control at large stationary sources. 2. Formulation of Improved Dry Deposition Module for Use in Airshed Model
2.1. Formulation of Dry Deposition Module. The dry deposition velocity, vg, is defined as ug = F/c@,>
(1)
where F is the flux of a pollutant and c(z,) is the local ambient pollutant concentration at a chosen reference height, zr(typically z, = 10 m). The dry deposition velocity as calculated in the original Caltech airshed model has been described previously ( 3 , 4 , 7 ) .Deposition is assumed to be a one-dimensional,steady-state, constant flux process 0013-936X/93/0927-2772$04.00/0
0 1993 Amerlcan Chemical Society
occurring without reentrainment within the surface layer (0 Iz Izr). By equating fluxes it is possible to show, using Monin-Obukhov similarity theory, that
where k is von Karman’s constant, u(zr)is the wind velocity at reference elevation, zr, U I is the friction velocity, L is the Monin-Obukhov length, zo is the surface roughness height, and Zd is the pollutant sink height. The expressions 4 m and 4p used in eq (2) are experimentally derived functions that account for the influence of atmospheric stability on turbulent transport (8). I t is assumed by analogy to heat-transfer experiments for surface-transfer rates (9, 10) that (3) where Sc and Pr are the Schmidt and Prandtl numbers for the pollutant in air. This assumption leads to the actual expression used to evaluate ug in the model:
Zr down to a thin fluid layer very near the surface, Q is the
resistance due to molecular scale diffusive transport through the thin atmospheric sublayer near the surface, and r, is the resistance due to the chemical interaction between the surface and the pollutant of interest once the gas molecules have reached the surface. The derivation of eq 4 includes the fluid mechanical aspects of pollutant transport through the atmosphere to the surface, although it should be noted that Monin-Obukhov similarity theory, while as good an approach as any available, may not be strictly correct for an urban area with a nonuniform environment. In the case of a pollutant for which the surface is a perfect sink rise., rs = 0 and C(Zd) = 01, the resistance to dry deposition is due solelyto fluid mechanical transport considerations, and the deposition velocity reaches a maximum value for the specified meteorological conditions:
and r, + rb = l/ugmax
An improved approach to capturing the effect of varying surface type on pollutant dry deposition is used. Gridded land use data are introduced into the model, and a table lookup procedure is adopted to specify the surface resistance, rs, in eq 6 directly as a function of the surface type, solar intensity, season of theyear (which is important for vegetative surfaces), and pollutant of interest. The completed dry deposition model then becomes:
+ + r,)-’
ug = (ra rb
The treatment of the surface resistance to dry deposition in eq 4 is captured in the factor 1- [c(zd)]/[c(zr)], which will be called the “surface affinity” for each pollutant. As implemented in the original version of the Caltech airshed model, the surface affinity is read a8 a single constant value for each pollutant species, ranging from 1.0 for pollutants that show no appreciable surface resistance to deposition, like HN03,to essentially 0 for pollutants like CO that do not readily react with surfaces. For many of the most important acid gases (e.g., S02, NO21 and oxidants the recommended values of the surface affinity (e.g., 03), lie between 0 and 1. These values should depend on the chemical nature of the surface as well as the pollutant, but in the original formulation of the airshed model, only the relative reactivity of the pollutant of interest is considered in setting these surface affinity values. The improved dry deposition module developed for use in the present study is represented in the form of an electrical resistance analog in which the nature of the chemical interactions at the surface can be specified in greater detail (10-12). The total resistance to pollutant deposition, rt, is taken as the inverse of the deposition velocity:
rt = l / v g (5) That total resistance to deposition is then composed of several parts: rt = ra + rb + r, (6) where ra is the resistance to deposition due to turbulent transport through the atmosphere from reference height
(8)
(9)
where the value of ra + rb is calculated from eq 7 and 8 and where the value of r, is obtained from the table lookup. The deposition flux to each of the various land uses that occurs within a particular grid cell within the model is computed separately and then summed to obtain the total flux from the atmosphere to the surface of that grid cell at each time step. Flux calculations are corrected for the dimensions of the ground-level cell in the model (3, 4). 2.2. Land Use Data. The procedure used for specifying surface characteristics within the model has been geared to a readily available source of land use data. The United States Geological Survey (USGS) has produced extremely detailed maps in which the surface characteristics of large areas of the United States have been specified according to the categories listed in Table I. In order to use the revised deposition model, land use data organized according to the area shown in Figure 1 were prepared. A grid system first was drawn over USGS land use maps, and then each 5 km X 5 km grid cell was subdivided into 100 smaller squares. Each of the 0.5 km X 0.5 km cells was assigned to the appropriate land use type, chosen from the 31 land use categories given in Table I. The results for these small 0.5 km X 0.5 km cells then were used to assign the percentage of each 5 km X 5 km grid cell that falls into each land use category. Examples of gridded land use maps that result from this process are shown in Figure 2. The area shown is the same as that in Figure 1, and the percentage of each grid square devoted to a particular land use is indicated by the density of the dot pattern in each cell. A cell that was 50%covered by residential uses in Figure 2 would be shown Environ. Scl. Technol., VoI. 27, No. 13, 1993 2773
Table I. Land Use Categories and Surface Roughness Values
land use 11 12 13 14 15
16 17
21 22 23 24 31
32 33 41 42 43 50 51 52 53 54
61 62 71 72 73 74 75 76 77
Urban or Built-up Land residential commercial and services industrial transportation, communications, and utilities industrial and commercial complexes mixed urban or built-up land other urban or built-up land Agricultural Land cropland and pasture orchards, groves, vineyards, nurseries, and ornamental horticultural areas confined feeding operations other agricultural land Rangeland herbaceous rangeland shrub and brush rangeland mixed rangeland Forest Land deciduous forest land evergreen forest land mixed forest land Water ocean streams and canals lakes reservoirs bays and estuaries Wetland forested wetland nonforested wetland Barren Land dry salt flats beaches sandy areas other than beaches bare exposed rock strip mines, quarries, and gravel pits transitional areas mixed barren land
roughness height zdm)
ref
3
4 4 4
0.3
a
3 2.5 2.5
4
2
3
0.1 0.45
b b
20 C
0.1
d 20
0.1 0.25 0.25
4 4
1.0 1.0 1.0
4, 20 4, 20 4,20
0.3
0.0001 0.002 0.0001 0.0001 0.0001
e
4
f
20 20 20
1.0 0.15
20 20
0.00004 0.0004 0.0004 0.10 0.10 0.002
4 4 4 20 g
0.002
h 20
Estimated as open land with nursery plants or cars and fences. Roughnesselement height = 2 m. ZO= 0.15 X 2 m zz 0.3 m. Estimated to be a mixture of commercial and residential areas. Orchard roughness elements estimated to be 4 m high; nursery roughness elements estimated to be 2 m high; average of roughness elements is thus about 3 m high. zo = 0.15 X 3 m = 0.45 m. d Roughness elements are cows (1.5 m), fences (1.5 m), and a few buildings (4 m). Roughness element height = 2 m. zo == 0.15 X 2 m = 0.30 m. e Midsummer rangeland given ZQ = 0.05 m in ref 20, but higher value used here because most of the herbaceous rangeland left in Southern California is located in hilly areas. f Streams in this area were treated as being dry in the summer, zo estimated as the same as mixed barren land. 8 Highly variable topography: some deep pits, some rocky and sandy land with shrubs, range; sand would have zo = 0.0004 m; depth of pits = 30 m, indicating zo = 4.5 m; value of zo == 0.1 m is chosen. h Estimated to be similar to mixed barren land.
with a pattern covering 50% of its surface area with black dots. Residential areas (use 11)are distributed in a pattern similar to that seen for vehicular-derived pollutants. Cropland, pasture land, orchards, and vineyards (uses 21 and 22) are found largely in the eastern portion of the South Coast Air Basin and in Ventura County. Most of the hillside areas of the South Coast Air Basin are covered with shrub and brush rangeland (use 32). Evergreen forests (use 42) are found on top of some of the mountainous areas. 2774
Envlron. Scl. Technol., Vol. 27, No. 13, 1993
2.3. Surface Resistance Data. Summaries of deposition velocity data are contained in several reviews of the technical literature (4, 13-16). In some cases, these reviews were intended simply to show the full range of values over which experimenal results have been obtained, while in other cases, the purpose of the study was to select typical values for use in an air quality modeling study. is removed at the There is general agreement that "03 highest observed rates, consistent with inferences (17) suggesting that the surface resistance for HN03 is essentiallyo. Most of these surveys are roughly consistent with the relative deposition velocity ordering seen in the experiments of Hill and Chamberlain (18): diffusionlimited acids > SO2 > NO2 = 03 > PAN > NO > CO. This suggests that surface resistance values, r,, should be ordered approximately as: CO > NO > PAN > 0 3 = NO2 > SO2 > "03 = 0.0. The notable exception to this summary is provided by the treatment which has been used with the model that has been assembled for the National Acid Precipitation Assessment Program, the Regional Acid Deposition Model (RADM) (13). They place the surface resistance values for both NO and NO2 at a level equal to that of SO2 and do not calculate depositional losses for PAN. Based on personal experience gained with the modeling of indoor NO concentrations (191, we believe that the NO dry deposition velocity is likely to be quite low and find it unlikely that NO2 and NO are removed at the same rate. Therefore, the surface resistance for NO should be set at a higher value than for NO2. Recommended values for the surface resistance for $02 deposition to urban, agricultural, range, deciduous forest, coniferous forest, forested swamp, water, swamp, and mixed agricultural-range land uses are found in ref 13,20, and 21. The recommended values for the SO2 surface resistance taken from RADM (20) are used in the revised Caltech deposition module. The smaller number of generic land use categories used in these prior studies has been extrapolated to match the 31 land use types that are available from USGS maps for Southern California by assigning their single recommended urban surface resistance value to all urban land uses, by assigning their single agricultural value to all agricultural uses, and so forth. Surface resistance values used in the present model for pollutants other than SO2 must be set in large part based on engineering judgment. The values chosen should be consistent with the existing experimental values for vegetative surfaces and should preserve the apparent rank ordering between the pollutant species that was just discussed. In the case of 03, surface resistance values as a function of land use type have been recommended (201, and these recommendations will be used in the present H202, HCHO, study. For the pollutants NO, NOz, "3, and higher aldehydes (RCHO), the RADM approach (13) will be followed in which the surface resistances will be set for each land use in fixed proportion to the values for SO2. Following the rationale explained previously, the surface resistance for NO is set to a value much higher than that for S02. For the remaining pollutant species, typical deposition rates to a few surfaces may be known, but there is no basis for trying to distinguish between the surface resistance values for different land use types. There are simply insufficient experimental data on the subject, and there are no systematic expert recommendations for drawing distinctions between land use types. For HN03 and for
....... - .-. -. .
ComputationalRegion Boundaty
Air quality monitoringsite
County Boundaty
Flgure 1. Southern California, showing a computational region that corresponds closely to the South Coast Air Basin that surrounds Los Angeles. Emissions and meteorological data fields are developed over the entire 150 km X 400 km grldded area. Air monitoring sites at whlch “Os, NH,, and aerosol nitrate data are available are shown by (0).
the highly reactive species NO3 and NzOs, it is assumed that removal at surfaces occurs at a diffusion-limited rate. Values of rs for these pollutants are set to nearly zero over all land uses. For PAN, the surface resistance value is set to 4.5 s cm-1, except that deposition of PAN over the ocean is suppressed, based on the findings of Garland and Penkett (22). The surface resistance for CO is set to a high enough value (rs = 50 s cm-1) to prevent substantial CO loss to surfaces. For the remaining gaseous pollutant species, no deposition data exist, and values of rs are set by analogy to related pollutants for which some data are available. All hydrocarbon species (alkanes, olefins, aromatics, ethene) are treated by analogy to CO; it is assumed that they are removed by dry deposition at a negligible rate. RON0 and RN04 are assumed to behave like PAN and are assigned a surface resistance of 4.5 s cm-1. Nitrous acid (HONO) is assigned a low surface resistance of 0.4 s cm-1 on the expectation that it will be removed by dry deposition at a fairly rapid rate. The surface resistance values are summarized in Table 11. Review of RADM’s recommended ozone surface resistance values (20) for urban areas shows that these resistances are as high as for bare dryland. It would appear that they envision residential areas that are entirely built upon or paved over. In the Los Angeles area, however, aerial photographs show that a very large fraction of most residential, commercial, and even some industrial areas consists of lawns and landscaping. By analogy to the surface resistance values given for grassy pasture land, a residential neighborhood with much lawn area should have a much lower surface resistance for many pollutants than would be true for an asphalt and building-covered urban core. For this reason, aerial photographs were examined to determine the fraction of urban land in each land use category that was not paved over and was not built upon. The results of this survey are shown in Table 111. Approximately 44% of the residential land in these photographs was not built upon or paved over. Similar amounts of open area were found for most other suburban land uses in Southern California. Therefore, in the modeling calculations that follow, deposition fluxes to ”urban” areas are computed using the urban surface resistance values only for that fraction of the “urban” land that is actually built upon or paved over. The flux to the remainder of the suburban landscape that is not built upon or paved over is computed as if that remaining land is
landscaped with a surface resistance like that of cropland and pasture. This representation of the suburban character of Southern California as a combination of high density urban coverage plus landscaping will be referred to as the “suburbannversion of the dry deposition model. In addition to gaseous pollutants, aerosol nitrate concentrations also are predicted by the Caltech photochemical airshed model. Aerosol nitrate predictions are based on the upper limit to the product of the gas-phase concentrations of NH3 times HN03that can be sustained in equilibrium with pure aerosol NHdN03. If coarse particle nitrates are present, then the deposition velocity that results from sedimentation and impaction could be as high as that for a gas such as “ 0 3 that is removed at a diffusion-limited rate, while if all aerosol nitrates are in the fine particle mode, then the deposition velocity could be very low. As aerosol nitrate size distribution predictions are not produced by the model, it is not possible to calculate aerosol nitrate dry deposition fluxes from first principles based on aerosol mechanics considerations. Instead, the following empirical approach is taken, based on experimental data. During the 1985 Nitrogen Species Methods Comparison Study, the deposition velocities of nitric acid vapor and aerosol nitrate to surrogate surfaces were measured simultaneously at Claremont, CA (23). The average of the ratio of the deposition velocity for aerosol nitrate to the deposition velocity for “ 0 3 is found to be 0.182. Nitric acid vapor is assumed to be removed at a diffusion-limited rate that can be calculated accurately by the present model. Therefore, in the present model, aerosol nitrate will be removed from the atmosphere with a deposition velocity that is 18.2% of that calculated for “03 at the same time and place. 3. Model Application: Determination of Base Case 1982 Dry Deposition Fluxes
3.1. Selection of Case Study. Nitric acid, aerosol nitrate, PAN, and ammonia are unregulated pollutants. As a result, their concentrations have not been measured routinely by governmental air monitoring networks. We seek the most comprehensive data set available on these pollutants that can be used for checking the scientific aspects of the air quality model’s performance. Such comprehensive data sets are very rare, and the choice of a case study for model evaluation must face many Environ. Sci. Technol., Vol. 27, No. 13, 1993 2775
LANDUSE 11 RESIDENTAL
UTM E.shng (km)
LANDUSE 21 CROPLAND AND PASTURE
UTM Easbng (km)
LANDUSE 32 SHRUB 8 BRUSH RANGELAND
3800
UTM Easting (km) Flgura 2. Maps showlng ihe spatial dlshlbutlon of selected residential. agrlcukural, and shrub. and brushcovered land uses
constraints. In particular, there is a tradeoff between use of the newest data sets available versus use of the most complete data set available. During the period August 30-31,1982, a field experiment was conducted to measure the concentrations of “08, NH3* and aerosol nitrates over consecutive 2-h periods at 10 monitoring sites in the South Coast Air Basin (24,25). South Coast Air Quality Management District air monitoring stations provided concurrentdataon NO, NOz, and 03concentrations, while PAN concentrationswere measured at Pasadena (Caltech) 2776
m&on. sci. TechMI.. Val. 27. No. 13. 1993
and at the University of California at Riverside. Further, this 1982 study period occurs during one of the few times when field measurements of actual nitrogenouspollutant deposition were being made in Southern California (26). In terms of ambient data and dry deposition flux data, the 1982datasetis in factthe only complete datasetcontaining all of the information needed for the present analysis. The next important consideration is the availability of emission data. The most recent NO, emissions estimates for the South Coast Air Basin, compiled for the year 1987,
field experimental data set has been used previously to evaluate the ability of the Caltech photochemicaltrajectory model and photochemical airshed model to accurately species r, (s/cm) predict NO2, total inorganic nitrate, "03, "3, PAN, aerosol nitrate, and O3concentrations in the South Coast os rot NO 13.9 X rg0,b Air Basin ( I , 6 ) . To apply these modeling procedures, a NOz 1.0 X rso; grid system is laid down over the area mapped in Figure NHs 0.2 x ?-so; 1. An inventory of pollutant emissions for hydrocarbons, H2Oz 0.1 X rSO; NO,, and NH3 is developed for each of the many source HCHO 0.5 X rgo; RCHO 2.0 X rSO; types in the air basin. Then, given the meteorological "03 0.01 conditions for the time period of interest, a simulation of NO3 0.01 pollutant transport with the prevailing winds, combined N~OS 0.01 with a detailed description of atmospheric photochemical PAN 4.5 HONO 0.4 and thermochemical reactions, proceeds to compute the RN04 4.5 spatial distribution and temporal pattern of the concenRON0 4.5 trations of the atmospheric pollutants of interest. other 50 Air quality model predictions from the previous study u = 0.182 X U H N O ~ aerosol nitrate (6) of the atmospheric concentrations of 0 3 , NO2, PAN, a Note that rOa is the surface resistance for 0 3 as a function of land and total inorganic nitrate were shown to be quite close use type, season of the year, and solar insolation as given by Sheih etal. (20). N~tethatrso~isthesurfaceresistanceforSOzasafunction to the observations-both in absolute value and in a relative sense over time. Predicted HN03 and aerosol of land use type, season of the year, and solar insolation as given by Sheih et al. (20). See Chang et al. (11). nitrate levels were within the equivalent of a few ppb NO, of the observed values. 3.3. Use of Revised Deposition Module. The modTable 111. Fraction of Urban Land Not Paved Over and ified photochemical airshed model and deposition module Not Built Upon with the suburan deposition scheme just described was land use fraction tested by application to the South Coast Air Basin over 0.25 highrise urban the period August 30-31, 1982. The emissions data, 0.44 residential meteorological fields, initial conditions, and boundary 0.34 commercial conditions are identical to those used by Russell et al. (6). 0.47 industrial The spatial distribution of deposition velocity values 0.55 utilities 0.40 industrial/commercial predicted by the model for several key pollutants in the 0.43 mixed urban mid-afternoon of August 31 are shown in Figure 3 over the 0.43 other urban 17 025 km2 computational region shown by the dashed line on the map of Figure 1. Nitric acid vapor is removed are believed to be approximately correct, based on at a rate limited only by atmospheric transport and, comparisons of emissions and ambient data (27). Inventherefore, has the highest deposition velocity values tory procedures used in 1982 would produce total NO, predicted by the model at any time. In the middle of the emissions that differ from those calculated today by less afternoon of August 31, the deposition velocity for "03 than 1576, so net NO, emissions contained in the 1982 reached 10 cm s-1 over southwestern Los Angeles County data set should be represented reasonably. The most in an area of high wind speed and surface roughness, falling recent hydrocarbon inventory for the South Coast Air to less than 5 cm s-1 in the less urbanized areas of the air Basin is throught to underestimate the reactive organic basin at that time. Ozone and NO2 with their significant gas emissions (27-29). Since the 1982 inventory contains surface resistances to dry deposition show lower deposition less optimistic assumptions about the effectiveness of velocities, in the range of 0.7-1.0 cm s-1 over much of the motor vehicle emissions control than have been used in air basin at that time (see Figure 3). The surface more recent years, the underestimation problem in 1982 resistances for NO and PAN are much higher than for 03, should be less severe. The best way to judge whether the and their deposition velocities equal about 0.1 cm 5-1 in 1982 estimates of hydrocarbon emissions were realistic mid-afternoon. The spatial distribution of the aerosol for the August episode is to compare the relative production nitrate dry deposition velocity mirrors that of "03 since of organic versus inorganic nitrate species by the model the deposition velocity is calculated as a constant ratio to As has been shown to measured ambient PAN and "03. that of "03 based on a rough analogy to field experipreviously ( I ) , if the hydrocarbon inputs to a photochemmental data, as discussed earlier. The values shown in ical model are estimated incorrectly at a fixed NO, level, Figure 3 represent peak deposition velocities that occur the predicted PAN versus "03 ratio will depart from in the presence of the highest observed wind speeds. the ambient ratio. As shown in previous grid modeling Averaged over all times on August 31 and all on-land studies of the August 1982 episode (6),the ambient PAN computational grid squares, the average deposition veand HNO3levels are predicted quite closely. This supports cm s-l for 03, 0.28 locities were 1.71 cm s-l for "03,0.31 the assessment that the hydrocarbon inputs for that cm s-l for NO2, and 0.66 cm s-1 for NH3. episode were satisfactory. The most recent inventory of A comparison of statistical model performance measures NH3 emissions for the South Coast Air Basin that is is presented in Table IV which contains values computed publicly available today is the 1982 inventory (30). before and after the changes were made in the dry Therefore, model applications will focus on calculations deposition calculations. Concentration predictions after for the South Coast Air Basin over the period August 30the changes to the dry deposition module in the model 31, 1982. generally are higher than from the previous version of the 3.2. Review of Previous Modeling Studies of the model. Thisis aconsequence of lower estimated deposition August 30-31,1982, Data Set. The August 30-31,1982, rates computed using the new surface resistance formuTable 11. Surface Resistance as a Function of Pollutant Species
Environ. Sci. Technol., Vol. 27, No. 13, 1993
2777
Flgure 3. Spatial distributions of predicted deposition velocities, atmospheric concentrations, and pollutant fluxes at 1400 PST August 31, 1982. The NOn, “03, and NH3 fluxes are given in mg of nitrogen m-2 h-I, while the O3 fluxes are in mg of O3 m-2 h-I.
Table IV. Statistical Comparison of Model Performance for Atmospheric Pollutant Concentrations: Former Modele vs Revised Deposition Module*
species (units) ozone (ppb) nitrogen dioxide (ppb) ammonia (ppb) nitric acid (ppb) aerosol nitrate (pgim3)
model former revised former revised former revised former revised former revised
mean of residualsc 10
28 -2 19 0.7
-0.3 4.2 0.6
-1.5 10.9
RMS error about mean 37 55 30 33 16 12 7.8 4.1 12
24
0 Russell et al. ( 6 ) .b Present study. e Residual = (prediction observation) at a monitoring site for each hour when observations are available.
lation of the deposition module. In spite of the increases in pollutant concentration predictions, the present model is a better model because it is better grounded in theory and because it is better able to accept data on land use characteristics and surface resistance values from field experiments. In order to compute the pollutant dry deposition fluxes, the deposition velocities are multiplied by the atmospheric concentrations at each time step in the model. Deposition flux maps for the mid-afternoon of August 31 are shown in Figure 3. NO,-derived species are removed predomHN03 fluxes to the surface deliver inantly as “03. nitrogen equivalent to 1.5-3 mg of N m-2 h-I throughout the central portion of the air basin at that time. NO2 also makes a significant contribution to the flux of NO,-derived nitrogen. Aerosol nitrate and ammonia fluxes are highest in the Rubidoux area, to the east of the zone of highest 2778
Environ. Sci. Technol., Vol. 27, No. 13, 1993
Table V. Dry Deposition Flux to Surface of Modeling Region: Base Case Conditions for August 31,1982
pollutant
flux (t of N/day) 4.8 49.1 7.2
101.4 58.7 25.9 247.1
HNOB and NO2 dry flux at that time. NO and PAN concentrations make a smaller contribution to the dry deposition flux in the mid-afternoon because of their low deposition velocities (and, in the case of NO, its low concentration at that time of day). Ozone fluxes to the surface are as high as 25 mg of O3 m-2 h-1 in portions of the eastern area of the air basin in mid-afternoon. Under base case August 31,1982, conditions, 247 t day1 of nitrogen is deposited by dry processes within the boundaries of the modeling region. As shown in Table V, that dry flux is dominated by nitric acid, ammonia, and NOz. The ammonium nitrate flux includes nitrogen from both the ammonium and nitrate ions. If the ammonia dry flux is removed from the totals along with the ammonium content of the aerosol nitrate, then the flux of NO,-derived species totals 176 t of N day-l, which is equivalent to 577 t day1 if stated at the molecular weight of NOz. Since the NO, emissions to the atmosphere of the air basin total about 1120 t day-l stated at the molecular weight of NO2, the prediction is that 52% of the NO, emitted will have been removed by dry deposition within the modeling region during a 24-h period like that studied here. This dry flux and NO2, of NO,-derived species is dominated by “ 0 3 the sums of which considerably exceed the NHs dry flux.
Field measurements of dry deposition fluxes are extremely rare. Riggan et al. (26)measured ammonium and nitrate ion deposition due to canopy throughfall and precipitation at the San Dimas Experimental Forest in the San Gabriel Mountains to the north of the zone of Figure 1 that lies between Azusa and Upland. For experiments from 1981 to 1983, the average flux due to canopy throughfall was 23.3 kg of N ha-1 yr-l (16.1 from nitrate ion, 7.2 from ammonium ion). That canopy throughfall flux of 16.1 kg of N ha-l yr-l from nitrate can be broken down into 4.3 kg of N ha-l yr-I delivered in bulk precipitation (wet deposition) plus 11.8 kg of N ha-l yr-l that was washed from dry deposition collected on canopy surfaces. The latter value of 11.8 kg of N ha-l yr-l represents the best available estimate of the local dry deposition flux of oxidized nitrogen species. Using our model results for the specific site of Riggan et al.'s experiments, the total oxidized nitrogen and ammonium nitrate flux to surfaces on August 31, 1982, was calculated to be 0.180 kg of N ha-l day1. If such deposition fluxes occurred every day of the year, the annual dry deposition flux would be 65.6 kg of N ha-l y r l (55.7 from oxidized nitrogen, 9.9 from ammonium). However, August 31, 1982, was a day of high photochemical air pollution; an average day would have had a lower flux of the photochemically derived pollutants. In particular, the oxidized nitrogen contribution from nitric acid ("03) should be much lower on an average day than on this heavily polluted day. This is important since "03 accounts for 62 % of the total N flux prediction and 73 % of the oxidized nitrogen flux prediction for August 31 at the location of Riggan et ala'smeasurements. Solomon et al. (31)measured the annual average "03 concentration at Tanbark Flats in the San Dimas Experimental Forest during 1986. These measurements show an annual mean "03 concentration of 2.7 ppb and a maximum 24-h average HN03concentration of 8.2 ppb. The 24-h average "03 concentration from our model prediction for the August 31,1982, photochemical smog episode is 12.6 ppb at this location. For comparison to annual average dry flux data, the dry flux of NO,-derived species predicted by the model for August 31,1982, can be scaled downward by a factor of 0.21, which accounts for the ratio of annual mean to August 31, 1982, HN03 concentrations. Using this scale factor, the predicted dry flux of NO,-derived speciesto the San Dimas Experimental Forest extrapolated to an annual basis is 12 kg of N ha-l yr-l. The extrapolation from the present model is in good agreement with Riggan et al.'s watershed experiments and is further supported by Riggan et al.'s summer season canopy washing experiments. It must be noted that there is little discussion in the technical literature of the deposition rates for ozone and most other pollutants to urban areas. All of the surface resistance parameter values used for urban landscapes in photochemical models at present are essentially educated guesses. What is needed at this point to increase confidence in the model is a program of field measurements necessary to determine the surface resistances for the large collection of pollutants and surface types that must be tracked by the airshed model. An experimental determination of the effect of the surface area of buildings and other roughness elements on the effective surface area for removal of pollutants also needs to be undertaken.
4. Evaluation of the Effect of Emission Controls
4.1. Introduction. Emission control measures that already have been examined for their effect on ambient pollutant concentrations by Russell et al. (32)now will be evaluated for their effect on the dry deposition flux of nitrogen-containing species. The objective is not to attempt to simulate the exact effect of a particular air quality management plan, but rather to examine the magnitude of the change in the dry deposition flux as emissions are lowered using control measures that have been discussed in recent years. 4.2. Emission Control Opportunities. Five groups of NO, and hydrocarbon emission control measures will be evaluated as part of this study, and a separate evaluation of the effect of ammonia emissioncontrol will be conducted. These control measures have been described in detail previously (32) and will be summarized here briefly. In the control calculations, the historical motor vehicle fleet is replaced by hypothetical fleets with clearly specified emissions levels. Even if current emissions from the motor vehicle fleet are uncertain (27-291, there is no ambiguity about the emissions represented by the control cases studied. Group 1 consists of controls that are a subset of the measures contained in the 1982 Air Quality Management Plan (AQMP) for the South Coast Air Basin (33)that had not yet been applied at the time of the base case simulation. Hydrocarbon emission controls in group 1include reducing solvent vapor emissions from painting and surface coating operations, reducing fugitive emissions from landfill gas leaks and oil and gas fields, suppressing solvent losses from cleaning operations and pesticide application, and capturing certain industrial process emissions using incineration, activated carbon adsorption, or other vapor recovery methods. Group 1controls on stationary sources of oxides of nitrogen involve relatively straightforward modification of combustion system design, but without the use of ammonia injection or selective catalytic reduction technology. Also included in group 1is the effect of a mandatory vehicle inspection and maintenance program involving a no-load idle test, followed by repairs to the vehicle designed to correct defects observed. Groups 2 and 3 controls simulate the effect of reducing emissions from the motor vehicle fleet to target levels that have been discussed by state and federal regulatory agencies (34, 35). At the group 2 level, the entire lightduty vehicle fleet is assumed to be reduced to an emission rate of 0.7 g mi-l NO, and 0.41 g mi-' total hydrocarbons (THC), and the NO, emissions from heavy-duty trucks are assumed to be reduced to 10.7 g bhp-1 h-I. Two ways exist to view the 0.7 g mi-l NO, emission rate from the light-duty vehicle fleet. Until recently, new cars sold in California were required to meet a 0.7 g mi-l NO, emission rate. Therefore, this emission rate could represent the successful completion of conversion of the entire vehicle fleet to meet such a regulatory objective for new cars, in combination with a high level of catalyst system durability and maintenance. In the absence of high durability and maintenance, catalyst system deterioration can be expected to increase actual on-road emissions to levels above legal objectives. The second way to view the 0.7 g mi-1 NO, fleet-wide emission rate employed here is that it closely approximates the introduction of a fleet of cars initially set to achieve a regulatory requirement of 0.4 g mi-' NO, when new, followed by a typical degree of control system deterioration in the hands of the final consumer. Environ. Sci. Technoi., Vol. 27, No. 13, 1993
2778
Table VI. Combinations of Mobile and Stationary Source Controls Examined for the South Coast Air Basin BASE CASE 1982 E m s ~ i o n(lonrlday) ~ THC 12416 RHC = 1224 NOx = l i 2 0 NH3 = 164
9
s
Ei:
Evaporative Controls and Combwtion Mcdiflcation
1
I
MOBILE SOURCE CONTROLS LlRht Duty Fleet MeeU
Vehicle Inspection atid Maintenance
::
Control Group I Effect on Emssmnr. NOx
-54%
,
1
0.41 glm THC 0.7 glmi NOx HW nuty FIW ~ 2 65 glmi THC 10 7 gmhp-hr NO,
Lteht Duly Fleet Meeu
041 glmi THC ~
Control Group I, 2 Effect on Emissions RHC-372% NOx -36 6% NH3 1 3 9 %
0.7 glm NOx Duf) l Flwf Me&
~ Hear) t
2 65 pimi THC 5 I 0 h p - k NOx
Effecr on Emissions RHC -37 2% KOx -47 6% KH3 1 3 9 %
bi
Non.cataiytic
&mmonis Injection
>
Control Group 1, 4 Effect an Emissions RHC - 9 3 % NOx -100% NH3 ~ 8 7 %
Control Group I,2 , 4 Effect on Emwons RHC -37 2% NOx -41 2% S H j +127%
Control m a u p 1 , 3 , 4 Effect on Emissions RHC -37 2% VOX -52 2% UH3 -12.7%
4
3
2e
Seiectiw Catalytic Reduction
controi Group 1,5 Effecr on Emissions. RHC - 9 3 % NOx -184%
Control croup I. 2 , s Effeci on Emiswmr
NH) 1 0 7 %
KH3 1 4 7 %
RHC -37 2% NOx -496%
contra1 croup I, 3,s Effect on Emmions RHC -37 2% VOX -600% KH3 + 4 7 %
Ammonia Reduetion Effect on Emissions RHC 4 0 % NOX 40% NH3 -42 4%
Mobile source controls in group 3 reflect the emissions that would result if a 0.4 g mi-' NO, and a 0.41 g mi-' THC emission rate for light-duty vehicles in fact were achieved and maintained by the vehicle fleet. Increased control system durability or maintenance would be required for this to occur. Further NO, reductions from heavy-duty vehicles have been added to group 3, at a stringent level previously discussed by the federal government (35). Group 4 controls would reduce the NO, emissions from stationary sources through the use of noncatalytic ammonia injection into the exhaust of the largest stationary combustion sources (36). Group 5 controls would use selective catalytic reduction technology on these large stationary sources instead of noncatalytic ammonia injection. Selective catalytic reduction (SCR) technology involves NO, abatement by injection of NH3 into stationary source exhaust in the presence of a catalyst. Control efficiencies are generally higher than in the case of the direct noncatayltic NH3 injection systems cited in control group 4, and NHB bleedthrough into the atmosphere is reduced. Table VI shows a matrix of control opportunities constructed by applying the controls in various combinations for nine cases. These cases represent the tradeoff between increasinglystringent stationary source control versus increasingly stringent mobile source control. Beginning near the upper left corner of Table VI, the base case 1982 emission inventory first will be perturbed by applying the group 1 controls to the emissions sources. Moving from left to right across the top of the table, increasingly stringent mobile source controls are added to the group 1 stationary source controls. Moving from top to bottom along the left edge of the table, increasingly demanding stationary source NO, controls are added to a minimal motor vehicle control program. At the lower right corner of that table, all of the most stringent mobile and stationary source controls are applied. In the far lower left corner of Table VI, a case is described in which base case hydrocarbon and NO, emissions remain unchanged but the ammonia emissions from livestock waste decomposition and from chemical fertilizer use are eliminated. This emission change could occur if increasingurbanization displaces agricultural activities from the eastern portion of the air basin. Environ. Scl. Technol., VoI. 27, No. 13, 1993
Vehicle Inspection and Maintenance
BASE CASE
il:
2
Evaporative Controls and Combustion Modifleation
NO NO2 PAK
E?
UH4UO3
-7% -5%
-2% -3% +?% -4%
KO
-42%
YO2 PAA
-39%
ma,
ii% -28%
hH3 NH4NO3
+17%
NO
-46% -41%
-30%
hO "IO? PAY
-54% -51% -9%
HKO,
36%
NH3 "@IO?
+2?% -43%
KO NO2 PAh
-55%
5
I
23 5
xm-catllytie Ammonia Injection
> z 4
c 5
6
Selective cataiytic Reduction
KO KO2 PAN
-11% -75%
"03 "3 MI4NO3
-78 +7% -i%
NO NO2 PAN
-26% -14%
-i"r
KO2 PAN
m03 "3 NH4'403
+2%
PAN
"3
-10% 16%
KH)
KHOO)
-;I%
mo3
11%
-32% -23% -29%
-.Mor
K\03
-39'70
NHJ
-28%
NH4VO?
-43%
PAN 127'1
Ammonia Reduction NO 0% NO2 0% PAS 0% HXO3 +I%
"3 NHflO?
L
2780
r--
MOBILE SOURCE CONTROLS
z
G
23
Table VII. Changes in Flux to Surface of Modeling Region on August 31, 1982 ( W )
-33% -48%
4.3. Effect of Emission Controls. The grid-based model described in Section 3 of this study was used to determine the effects on the dry deposition flux from each of the combinations of emission control measures defined in Table VI. The base case 1982 emission inventory for the South Coast Air Basin was modified to reflect the use of each set of control measures. These modified emission inventories were used in model calculations which were executed over two days of simulation (August 30-31,1982) using the meteorological conditions from the base case model. The effect of initial conditions and boundary conditions supplied to the air-quality model on emission control calculations has been studied previously by Russell et al. (34). That study aleady has shown that predicted changes in air quality on the second day of simulation are determined predominantly by changes in emissions to the model and not by altered initial or boundary conditions. The effect of emission controls now can be examined. The upper left-hand corner of Table VI shows that completion of the vehicle inspection and maintenance program plus evaporative hydrocarbon controls and combustion modifications would lower reactive hydrocarbon emissions by 9.3% and would lower NO, emissions by 5.4 % . In response to those controls, the dry flux of NO,derived species would decline by 2-7 %, as shown in the upper left corner of Table VII. The ammonia flux would increase slightly because less aerosol nitrate is formed while NH3 emissions remain unchanged; thus, NH3 concentrations will increase and so will the NH3 dry flux. Table VI11 shows the computed dry flux to the surface of the modeling region in that case. The dry flux in the presence of the group 1 controls totals 242 t of nitrogen day-', of which 170 t day1 N is derived from NO, emissions. Moving to the right across the top row of Tables VIVIII, the effect of progressively more stringent controls on motor vehicles is seen in the presence of a minimal stationary source control program. In response to a 37 76 reduction in both reactive hydrocarbon emissions and NO, emissions (top row, center column), dry fluxes of NO and NO2 decline by amounts that are slightly greater than proportional to the gross emissions change (-42% and -39%, respectively). HN03 and NH4N03fluxes decline by 28-30 % . The NH3 flux again increases as aerosol
Table VIII. Calculated Flux to Surface of Modeling Region on August 31,1982 (t of N/day) M O B I L E SOURCE C O N T R O L S Light Duly Fleet Meet!
Vehicle Inspeclion and Mainlmance
NO
a
8 *5
Evaporative ControB and Combustion Modilialion
I
homcala~ytic Ammonia Injection
X
6F 6
selective CBtsiytie Reduction
NO2
PAK m03 NH3 "003 NO NO2 PAN
m03
5 47 1 98 60 25
4 46 7 94
NH3 NH4N03
63
NO NO2 PAN "03 N>!3 "003
4 42
26
7 01
62
23
Lleht Duty Fleet Meets
041 @miTHC 0.7 s/mi NOx
0 41 gimi THC 0.7 gimi NOx
Htsvy Duly Fled MIeU
I b a v ) Duty Fled Meet3
2 65 gim THC 107gibhp-hrNOx
2.65 gimi THC 5 Igibhp-bNOx
NO NO2 PAS m 0 3 "3 "003
3
NO
30
NO2
6 73
PAS
69 18
NH3 "003
3
NO NO2 PA\ "03 NH3
NO NO? PA; m03 NH3 NHeUOj
29 6 69 72 18
NO NO2 PAS "03 NH3 NH4303
2 26 7 66 72 16
Hs0j
NHeVO) NO
NO2 PAN NH3 Ku03 "603
2 24
1 65 12
I5 2
23 7 62 75 15 2 20 7 58 75 I2
nitrate formation in the atmosphere is lowered due to the lower atmospheric HN03 concentrations. With a further decrease in NO, emissions from motor vehicles (far right column of Table VI), basin-wide NO, emissions would be reduced by close to 48%. The dry fluxes of NO and NO2 would decline by more than 50%, while the HN03 dry flux declines by 36 % relative to the base case. Moving down the left-hand column of Tables VI-VIII, the effect of progressively more stringent NO, controls on the stationary sources is examined. With highly effective selective catalytic reduction (SCR) systems applied to the largest stationary NO, sources plus evaporative hydrocarbon controls and a minimal motor vehicle control program NO, emissions would decline by 18.4% and reactive hydrocarbon emissions would decline by 9.3 % . The results of such a control program would include a 26% decline in the NO dry flux, a 14% decline in the NO2 dry flux, and a 10-11 % decline in the dry flux of "03 and NHdN03. If the most stringent combination of stationary source controls and mobile source controls on hydrocarbons and NO, studied here is applied, then NO, emissions would decline by more than 60%, accompanied by a 37% reduction in reactive hydrocarbon emissions. The effect of this maximum control case is shown in the box in the lower right-hand corner of Tables VI1 and VIII. The total flux of nitrogen-containing species in that case is predicted to decline from 247 t of nitrogen day' in the base case to 174 t of nitrogen day1 in the maximum control case studied. Of that total, 87 t of nitrogen day' is derived from acid gases plus PAN, and 75 t of nitrogen day1 is dry The remaining 1 2 t of nitrogen day' deposited as "3. is deposited as ammonium nitrate. Although the balance between deposited NH3 and acid gases is nearly equal in this case, the reader is cautioned to remember that the spatial distribution of the dry flux of these species is different, with the NH3 dry flux occurring predominantly concentration areas of the eastern portion in the high 3" of the air basin. Dry flux reductions in response to emission controls are largest in an absolute sense in the middle of the day when the deposition velocities are at their highest values. The spatial distribution of the dry flux values for
the maximum emission control case studied is given in Figure 3 during the high flux afternoon hours of August 31 and can be compared to the precontrol case. The final case considered involves elimination of ammonia emissionsfrom agricultural and livestock husbandry operations in the presence of base case hydrocarbon and NO, emissions. A 42 % reduction in basin-wide ammonia emissions greatly reduces ammonium nitrate aerosol concentrations, resulting in a 48% decline in the aerosol nitrate dry deposition flux. Ambient gas-phase ammonia concentrations also decline, accompanied by a 33 5% reduction in the dry deposition flux of ammonia. The reduction in ammonia emissions causes a shift from aerosol-phase to gas-phase species such that ambient gasphase concentrations and dry deposition fluxes do not decline as quickly as the change in ammonia emissions. As aerosol formation is suppressed, ambient gaseous "03 concentrations increase resulting in a 19% increase in the FINO3 dry deposition flux. In the presence of ammonia controls, the dry deposition flux of all nitrogenous species decreases slightly to 235 from 247 t of nitrogen day1 in the base case.
Conclusions The approximately 1120 t of nitrogen oxides emitted per day to the atmosphere of the South Coast Air Basin are converted by chemical reactions into an entire family of nitrogen-containing pollutants, including NOz, HN03, PAN, and aerosol nitrates. Eventually, these pollutants are removed by wet or dry deposition at the earth's surface. The dry deposition flux of nitrogen-containing pollutants to the surface of the 17 025 km2modeling region centered over Los Angeles wab calculated €or 1982 base case conditions. On a daily basis, NO, emissions accounted for 276 t of the total 247 t of deposited nitrogen, while the remaining deposited nitrogen originated from ammonia emissions. That 176 t of nitrogen would be equivalent to 577 t of NO, emissions if stated at the molecular weight of NO2. This indicates that approximately half of the local NO, emissionswere removed by dry deposition within the modeling region during the period studied. Gas-phase species delivered 90% of the nitrogen to the surface. The revised dry deposition model was employed to examine the nature of the effects that would occur if emissioncontrols were applied to the NO, and hydrocarbon sources in the South Coast Air Basin as they existed in 1982. At the highest level of control studied (37% reactive hydrocarbon reduction, 61 % NO, reduction), the daily nitrogen dry flux would be reduced from 247 t of nitrogen in the precontrol base case to 174t of nitrogen after control. Of that 174 t, 87 t would be derived from deposition of acid gases plus PAN, while 76 t would be deposited as "3. The remaining 1 2 t would be deposited as ammonium nitrate. In general, as emission controls are applied to reactive hydrocarbons and NO,, the dry flux of acid gases declines while the dry flux of NH3 increases slightly (dueto greater NH3 emissions from certain control devices and due to higher NH3 concentrations that result from lowered aerosol nitrate formation). Control of ammonia emissions from agricultural and livestock husbandry operations also was examined. In the presence of ammonia control, inorganic nitrate is shifted from the aerosol phase concentrations increase while to the gas phase. As "03 NH3 concentrations decrease, the dry deposition flux of total nitrogen to the surface of the air basin decreases marginally. Environ. Scl. Technol., Vol. 27, No. 13, 1993 2781
Further experimental data on the surface resistance of pollutants to dry deposition are needed to support application of the model. Field measurements of dry deposition to surfaces common in an urban environment would be especially valuable. Acknowledgments
This research was supported by the California Air Resources Board under Agreement A6-188-32and by the Center for Air Quality Analysis at Caltech. Literature Cited (1) Russell, A. G.; Cass, G. R. Atmos. Environ. 1986,20,20112025. (2) Liljestrand, H. M. Ph.D. Thesis, California Institute of Technology, Pasadena, CA, 1979. (3) McRae, G. J.; Goodin, W. R.; Seinfeld, J. H. Atmos. Enuiron. 1982, 16, 679-696. (4) McRae, G. J. Ph.D. Thesis, California Institute of Technology, Pasadena, CA, 1981. (5) Russell, A. G.; McRae, G. J.; Caw, G. R. Atmos. Environ. 1983, 17, 949-964. (6) Russell, A. G.; McCue, K. F.; Cass, G. R. Environ. Sci. Technol. 1988,22, 263-271. (7) Russell, A. G.; McRae, G. J.; Cam, G. R. Acid Deposition of Photochemical Oxidation Products-A Study Using a Lagrangian Trajectory Model. In Air Pollution Modeling and Its Application III; de Wispelaere, C., Ed.; Plenum Publishing: New York, 1984; pp 539-564. (8) Businger, J. A.; Wyngaard, J. C.; Izumi, Y.; Bradley, E. F. J. Atmos. Sci. 1971, 28, 181-189. (9) Brutsaert, W. J. Atmos. Sci. 1975, 31, 2028-2031. (10) Wesely, M. L.; Hicks, B. B. J. Air Pollut. Control Assoc. 1977,27, 1110-1116. (11) Bennett, J. H.; Hil1,A. C.; Gates, D. M. J. Air Pollut. Control ASSOC. 1973,23, 957-962. (12) Garland, J. A. In Atmosphere-Surface Exchange of Particulate and GaseousPollutants;Engelmann, R. J.,Sehmel, G. A,, Eds.; NTIS CONF-740921; National Technical Information Service: Springfield, VA, 1974. (13) Chang, J. S.; Brost, R. A,; Isaksen, I. S. A.; Madronich, S.; Middleton, P.; Stockwell, W. R.; Walcek, C. J. J. Geophys. Res. 1987, 92, 14681-14700. (14) Derwent, R.; Hov, 0.J. Geophys.Res. 1988,93,5185-5199. (15) McRae, G. J.; Russell, A. G. Dry Deposition of NitrogenContaining Species. In Acid Precipitation Series-Volume 4: Deposition Both Wet and Dry; Hicks, B. B., Teasley, J. I., Eds.; Butterworth Publishers: Boston, 1984; Chapter 9. (16) Sehmel, G. A. Atmos. Environ. 1980, 14, 983-1011. (17) Huebert, B. J.; Robert, C. H. J . Geophys. Res. 1985, 90, 2085-2090. (18) Hill, A. C.; Chamberlain, E. M., Jr. The Removal of Water Soluble Gases from the Atmosphere by Vegetation. In
Atmosphere-SurfaceExchange of Particulate and Gaseous Pollutants; Engelman, R. J., Sehmel, G. A., Eds.; NTIS CONF-740921; National Technical Information Service: Springfield, VA, 1974.
2782
Envlron. Scl. Technol., Vol. 27, No. 13, 1993
Nazaroff, W. W.; Cass, G. R. Enuiron. Sci. Technol. 1986, 20,924-934. Sheih, C. M.; Wesely, M. L.; Walcek, C. J. A Dry Deposition Module for Regional Acid Deposition; EPA/600/3-86/037; U.S. Government Printing Office: Washington, DC, 1986. Walcek, C. J.;Brost,R.A.; Chang, J. S.; Wesely,M. L. Atmos. Environ. 1986, 20, 949-964. Garland, J. A.; Penkett, S. A. Atmos. Environ. 1976, 10, 1127-1131. Pierson, W. R.; Brachaczek, W. W.; Japar, S. M.; Cass, G. R.; Solomon, P. A. Atmos. Environ. 1988, 22, 1657-1663. Hildemann,L. M.;Russell, A. G.; Cass,G.R. Atmos.Environ. 1984,18, 1737-1750. Russell, A. G.; Cass, G. R. Atmos. Enuiron. 1984,18,18151827. Riggan, P. J.; Lockwood, R. N.; Lopez, E. N. Environ. Sci. Technol. 1985, 19, 781-789. Fujita, E. J.; Croes, B. E.; Bennett, C. L.; Lawson, D. R.; Lurmann, F. W.; Main, H. H. J . Air Waste Manage. Assoc. 1992,42, 264-276. Harley, R. A.; Hannigan, M. P.; Caw, G. R. Environ. Sci. Technol. 1992,26, 2395-2408. Harley, R. A.; Russell, A. G.; McRae, G. J.; Caw, 6. R.; Seinfeld, J. H. Enuiron. Sci. Technol. 1993, 27, 378-388. Gharib, S.; Cass, G. R. Ammonia Emissions in the South Coast Air Basin 1982. Open file report 84-2; Environmental Quality Laboratory, California Institute of Technology, Pasadena, CA, 1984. Solomon, P. A,; Salmon, L. G.; Fall, T.; Cass, G. R. Environ. Sci. Technol. 1992,26, 1594-1601. Russell, A. G.; McCue, K. F.; Cass, G. R. Environ. Sci. Technol. 1988,22, 1336-1347. South Coast Air Quality Management District. Final Air Quality Management Plan 1982Revision-FinalAppendix No. VII-A-Short Range Tactics for the South Coast Air Basin; South Coast Air Quality Management District: El Monte, CA, 1982. (34) California Air Resources Board. Report to the Legislature on the Feasibility of a 0.4 Gram per Mile Oxide of Nitrogen Exhaust Emission Standard for Passenger Cars and Light Trucks. California Air Resources Board: Sacramento, CA. (35) Environmental Protection Agency. Control of Air Pollution from New Motor Vehicles and New Motor Vehicle Engines; Gaseous Emission Regulations for 1987 and Later Model Year Light-Duty Vehicles, and for 1988 and Later Model Year Light-Duty Trucks and Heavy-Duty Engines; Particulate Emission Regulations for 1988 and Later Model Year Heavy-Duty Diesel Engines. Fed. Regist. 1984, 40 CFR, parts 86 and 600. (36) Dziegiel, H. T.; Aure, T. B.; Anderson, D. W. The Thermal DeNO, Demonstration Project. Presented at the Joint Symposium on Stationary Combustion NO, Control, Dallas, TX, NOV1-4, 1982.
Received for review February 9, 1993. Revised manuscript received May 19, 1993. Accepted June 28, 1993." Abstract published in Advance ACS Abstracts, September 1, 1993.
