Environ. Sci. Technol. 2002, 36, 2614-2629
Estimates of the Atmospheric Deposition of Sulfur and Nitrogen Species: Clean Air Status and Trends Network, 1990-2000 R A L P H E . B A U M G A R D N E R , J R . , * ,† THOMAS F. LAVERY,‡ CHRISTOPHER M. ROGERS,‡ AND SELMA S. ISIL‡ U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, and Harding ESE, Inc., Jacksonville, Florida 32250
The Clean Air Status and Trends Network (CASTNet) was established by the U.S. EPA in response to the requirements of the 1990 Clean Air Act Amendments. To satisfy these requirements CASTNet was designed to assess and report on geographic patterns and long-term, temporal trends in ambient air pollution and acid deposition in order to gauge the effectiveness of current and future mandated emission reductions. This paper presents an analysis of the spatial patterns of deposition of sulfur and nitrogen pollutants for the period 1990-2000. Estimates of deposition are provided for two 4-yr periods: 1990-1993 and 19972000. These two periods were selected to contrast deposition before and after the large decrease in SO2 emissions that occurred in 1995. Estimates of dry deposition were obtained from measurements at CASTNet sites combined with deposition velocities that were modeled using the multilayer model, a 20-layer model that simulates the various atmospheric processes that contribute to dry deposition. Estimates of wet deposition were obtained from measurements at sites operated by the National Atmospheric Deposition Program. The estimates of dry and wet deposition were combined to calculate total deposition of atmospheric sulfur (dry SO2, dry and wet SO42-) and nitrogen (dry HNO3, dry and wet NO3-, dry and wet NH4+). An analysis of the deposition estimates showed a significant decline in sulfur deposition and no change in nitrogen deposition. The highest rates of sulfur deposition were observed in the Ohio River Valley and downwind states. This region also observed the largest decline in sulfur deposition. The highest rates of nitrogen deposition were observed in the Midwest from Illinois to southern New York State. Sulfur and nitrogen deposition fluxes were significantly higher in the eastern United States as compared to the western sites. Dry deposition contributed approximately 38% of total sulfur deposition and 30% of total nitrogen deposition in the eastern United States. Percentages are similar for the two 4-yr periods. Wet sulfate and dry SO2 depositions were the largest contributors to sulfur deposition. Wet nitrate, wet ammonium, and dry HNO3 * Corresponding author phone: (919)541-4625; fax: (919)541-1138; e-mail:
[email protected]. † U.S. Environmental Protection Agency. ‡ Harding ESE, Inc. 2614
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 12, 2002
depositions were the largest contributors to nitrogen deposition.
Introduction Atmospheric deposition takes place via three pathways: wet deposition, cloud deposition, and dry deposition. Wet deposition is the result of precipitation events (rain, snow, etc.) that remove particles and gases from the atmosphere. Cloud or fog deposition occurs when pollutant-laden cloud or fog droplets impact vegetation or other surfaces. Dry deposition is the transfer to the landscape of particles and gases through a number of atmospheric processes in the absence of precipitation. Wet deposition rates of acidic species across the United States have been documented since 1978 by the 200-site National Trends Network of the National Atmospheric Deposition Program (1). Estimates of cloud deposition are available from the Mountain Cloud Chemistry Program (2) and the Mountain Acid Deposition Program (3), and fog deposition programs are documented in work by Weathers et al. (4) and Collett et al. (5). Comparable dry deposition data on a national scale have been unavailable. However, the Clean Air Status and Trends Network (CASTNet) is now providing estimates of dry deposition at sites across the United States. The direct measurement of dry deposition is not straightforward, but a number of investigations suggest that it can be reasonably inferred by coupling air concentration data with routine meteorological measurements (6-8). Shieh et al. (6) showed that submicrometer particle and sulfur dioxide (SO2) deposition rates for the eastern United States were strongly dependent on wind speed, solar radiation, and condition and type of ground cover. For example, rapidly growing vegetation and forests were found to experience generally higher deposition rates than senescent vegetation, short grass, or snow. This approach for estimating dry deposition has been expanded by Wesely and Lesht (9) to calculate deposition rates for various additional atmospheric species using site-specific meteorological data. In 1986, the U.S. Environmental Protection Agency (EPA) established the National Dry Deposition Network (NDDN). The objective of the NDDN was to obtain field data at approximately 50 sites throughout the United States to establish patterns and trends of dry deposition. The approach adopted by the NDDN was to estimate dry deposition using measured air pollutant concentrations of SO2, particulate sulfate (SO42-), particulate nitrate (NO3-), nitric acid (HNO3), and particulate ammonium (NH4+) with modeled deposition velocities (Vd) estimated from meteorology, land cover, and site characteristic data. Passage of the Clean Air Act Amendments (CAAA) in 1990 required implementation of a national network to monitor air emissions, atmospheric deposition, and air quality; to determine the effects of emissions on water quality, forests, and other sensitive ecosystems; and to assess the effectiveness of the CAAA’s emission reduction requirements through the operation of a long-term monitoring program. In response to these requirements, the EPA, in coordination with other Federal agencies, created CASTNet. The network became operational in mid-1991 and incorporated the NDDN program. CASTNet was designed to support the investigation of the relationships between emissions, atmospheric concentrations, and deposition. Assessments of the sensitivity of ecosystems to acidic deposition have shown that large areas of potentially sensitive terrestrial and aquatic ecosystems exist in the eastern United 10.1021/es011146g CCC: $22.00
2002 American Chemical Society Published on Web 05/18/2002
FIGURE 1. Locations of CASTNet sites as of December 2000. States and that limited areas exist in the western United States. These findings, coupled with the expected changes in emissions of sulfur oxides (SOx) and nitrogen oxides (NOx), dictated the distribution of sites in the network. This paper provides a spatial analysis of deposition of SO2, SO42-, NO3-, HNO3, NH4+ from CASTNet sites for 19902000. As a result of the CAAA of 1990, emissions of SO2 decreased by almost 19% during 1994-1995 (10). Because of the large decrease in SO2 emissions during 1994-1995, the estimates of deposition provided herein are divided into two 4-yr periods, 1990-1993 and 1997-2000. Wet deposition estimates at National Atmospheric Deposition Program (NADP) sites are used along with modeled dry deposition fluxes to estimate total deposition of sulfur (SO2, SO42-) and nitrogen (NO3-, HNO3, NH4+). Although this paper contrasts deposition estimates at CASTNet sites from two distinct emission periods, these analyses are not designed to determine long-term trends in dry deposition. Robust statistical analyses of trends of sulfur species using CASTNet data are available in other recent publications (11-13). Trends in wet deposition data from the NADP may be found in articles by Lynch et al. (14, 15). Additional analyses of CASTNet data may be found in the CASTNet Deposition Summary Report (1987-1995) (16), the CASTNet 1999 Annual Report (17), and the CASTNet 2000 Annual Report (18).
Experimental Methods Network Description. Figure 1 shows the locations of CASTNet sites as of December 2000. At that time, the network consisted of 79 sites, two of which were collocated for the purpose of assessing network precision. Further information on the current status of CASTNet may be found at http:// www.epa.gov/castnet/. In addition to CASTNet, another dry deposition network consisting of 13 sites, the Atmospheric Integrated Research Monitoring Network (AIRMoN), operated in the eastern United States from 1985 to 2001. AIRMoN uses a similar approach to obtain estimates of dry deposition but is more research oriented in its focus. Further information
regarding the network is available at http://www.arl.noaa.gov/ research/programs/airmon.html. Site Selection. The eastern U.S. monitoring sites were selected based on the following criteria: appropriate inputs to the dry deposition model, regional representativeness, long-term availability, and accessibility. In the western United States, the limited number of sites combined with the regions of greater ecological diversity precluded determination of spatial patterns. Therefore, site selection focused primarily on locations where specific research issues could be addressed and/or where natural resources were at risk (e.g., national parks). These locations included calibrated watersheds in which dry deposition information was needed to close geochemical cycles for sulfur, nitrogen, and alkalinity. Regional representativeness refers to the overall similarity of the site to a characteristic area, typically 80 by 80 km, surrounding the site. First, the ambient concentrations at the site had to be representative of the region. Thus, sites near major sources of SO2 and/or NOx were avoided to reduce the likelihood of locally perturbed concentration fields. Second, land cover near the site had to match as much as possible the dominant regional land cover to make appropriate use of meteorological data in Vd calculations. Finally, monitoring sites needed to be available for extended periods (15-20 yr) to assess concentration and dry deposition trends. Site-specific criteria used during the site-selection process relate to conditions in the immediate vicinity of a prospective monitoring site that may perturb air quality and meteorological observations. Local sources of air contaminants and local features that may influence wind speed, wind direction, etc. were the focus of these criteria. By design, the network attempted to capture gradients in atmospheric concentrations of pollutants (especially SO2). Thus, sites in the Ohio River Valley were closely spaced around emission sources and more distantly spaced throughout New England and the Atlantic Coast states. Sites were located further apart in the Southeast and the West, where the density of emission sources is considerably lower than in the VOL. 36, NO. 12, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2615
Midwest. A recent analysis of SO2 and SO42- data by Baumgardner et al. (19) has shown that distinct gradients exist for both species moving outward from the Ohio River Valley in both northeasterly and southeasterly directions. Field Operations. Measurements of ambient concentrations of SO2, SO42-, NO3-, HNO3, and NH4+ and of meteorological variables required for dry deposition calculations are performed at each site. Meteorological variables are recorded continuously and reported as hourly averages. Variables measured include wind speed, wind direction, σθ (standard deviation of the wind direction), solar radiation, ambient temperature, temperature lapse, relative humidity, and surface wetness. Atmospheric sampling for sulfur and nitrogen species consists of week-long integrated sampling using a three-stage filter pack (16, 20). Filter packs contain three types of filters in sequence: a Teflon filter for SO42-, NO3-, and NH4+ aerosols; a nylon filter for HNO3 and SO2 (as SO42-); and dual K2CO3-impregnated cellulose filters for SO2. The filter pack system is “open faced”, meaning that a sizeselective inlet is not used. Sulfate particles are almost exclusively in the fine fraction (2.5 µm) and therefore collected less efficiently because of the configuration of the inlet and meteorological variables (i.e., wind speed). Filter packs are prepared and shipped to the field weekly and exchanged at each site every Tuesday. Blank filter packs are collected quarterly to evaluate passive collection of particles and gases as well as contamination during shipment and handling. Filter pack sampling and meteorological measurements are made at 10 m. Filter pack flow is maintained at 1.50 L/min at eastern sites. The flow is 3.00 L/min at western sites to better detect the lower concentrations of the analytes. Various observations are made periodically at the CASTNet sites to provide input variables for the model estimates of dry deposition. Vegetation status and land-use information are collected to characterize the distribution and condition of plant species that could influence deposition rates for gases (especially SO2 and particles). Vegetation data are obtained to track the evolution of the dominant plant canopy from leaf emergence (or germination) to senescence (or harvesting). Once a year, site operators also provide information on major plant species and land-use classifications within 1 km of the site. Additional land-use data are obtained by digitization, analysis, and interpretation of aerial photographs obtained from the U.S. Geological Survey (USGS) National Cartographic Information Center in Reston, VA (21). Leaf area index (LAI) measurements are taken at all sites. LAI is the one-sided leaf area of the plant canopy per unit of ground. All field equipment is subjected to semiannual inspections and multipoint calibrations using standards traceable to the National Institute of Standards and Technology (NIST). Calibration files are reviewed for completeness and adherence to standard operating procedures (SOPs). In addition, independent equipment audits are performed annually. Results of field calibrations are used to assess sensor accuracy and to flag, adjust, or invalidate field data. Laboratory Operations. Filter pack samples are loaded, shipped, received, extracted, and analyzed by a central laboratory. Following receipt from the field, exposed filters and blanks are extracted and then analyzed for SO2 (as sulfate), SO42-, NO3-, and HNO3 (as nitrate) by micromembrane-suppressed ion chromatography (IC). Teflon filter extracts are also analyzed for NH4+ by the automated indophenol method using a Technicon II or TRAACS-800 autoanalyzer system. All analyses are completed within 72 h of filter extraction. Atmospheric concentrations are calculated based on the volume of air sampled following validation of the hourly flow data. Atmospheric concentrations of particulate SO42-, NO3-, and NH4+ are calculated 2616
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 12, 2002
from analysis of Teflon filter extracts; HNO3 is calculated based on the NO3- found in the nylon filter extracts; and SO2 is calculated as the sum of SO42- found in nylon and cellulose filter extracts. Quality Assurance and Data Analyses. Quality Assurance. The CASTNet QA program is comprehensive in that it addresses all major aspects of project operations. Quality assurance plans and operational procedures were prepared for each network component. A detailed description of the CASTNet quality assurance and quality control procedures are available in the CASTNet Quality Assurance Program Plan (22). Estimates of precision and accuracy are calculated for all measurement variables. Precision is based on the difference between measurements from separate instruments and samplers at collocated sites. Collocated sites were selected to be representative of the observed range of pollutant concentrations and environmental conditions that exist within the network. The overall precision of atmospheric concentration and meteorological data is assessed quarterly by calculating the mean absolute relative percent difference (MARPD) of values for simultaneous measurements at collocated sites. Historically, two or three sites of the network were collocated. For the period 1987-1995 (all collocated sites-for all years), the MARPDs for all meteorological sensors were less than 10% (16). For the period 1990-2000 (all collocated sites-for all years), the MARPDs for measured filter pack concentrations for all analytes were less than 13% with SO2 and HNO3 at or below 8% and SO42- and NH4+ below 6% (18). The accuracy of field measurements is determined by challenging instruments with standards that are traceable to the NIST. The accuracy of laboratory measurements is determined by analyzing independently prepared reference samples. For the years 1987-1995, the reference sample recovery mean for all analytes was between 99 and 100% (16). Additional precision and accuracy information are available in the CASTNet 1999 and 2000 Annual Reports (17, 18). Data Analyses. Hourly dry deposition fluxes are calculated as the product of modeled hourly deposition velocities and hourly concentrations. Hourly concentrations of the analytes are determined from the weekly integrated value with the assumption that all hourly concentrations during the weekly sampling period are equal and constant. Weekly fluxes are calculated as the mean of the valid hourly fluxes for a deposition week scaled to represent a 168-h period. The weekly value is valid if at least 70% of the available hourly values are valid. Quarterly fluxes are calculated as the mean of associated weekly values scaled to represent a 13-week period and are valid if at least 69% of the weeks are valid. Annual deposition fluxes are the mean of at least three valid quarters scaled to represent a full year. To calculate annual total deposition values, both dry and wet deposition values for the year must be valid. Wet deposition values were obtained from the NADP/NTN site closest to the respective CASTNet site and were determined to be valid using the NADP/NTN criteria. Detailed descriptions of the NADP operations and completeness/validity criteria may be found in Sisterson et al. (1). Table 1 lists the CASTNet sites that met all completeness criteria and are used for this analysis. For 1990-1993, 47 sites met the dry deposition criteria, and 43 met both the dry and wet deposition criteria. For the period 1997-2000, 62 sites met the dry deposition criteria, and 59 sites met the total deposition criteria. For the purposes of this study, total deposition is defined as the sum of wet and dry deposition. The Appalachian Mountains of the eastern United States are subjected to cloud deposition at elevations at or greater than approximately 800 m. Cloud deposition may be a significant portion of the total deposition in these areas (23). Measurements of cloud
TABLE 1. CASTNet Sites Used in This Analysis site id
station
ABT147 ALH157 ANA115 ARE128 ASH135 BBE401 BEL116 BFT142 BVL130 BWR139 CAD150 CAN407 CAT175 CDR119 CHA467 CKT136 CND125 CNT169 COW137 CTH110 CVL151 DCP114 DEV412 EGB181 ESP127 GAS153 GLR468 GRB411 GRC474 GTH161 HOW132 JOT403 KEF112 LAV410 LCW121 LRL117 LYE145 LYK123 MCK131 MEV405 MKG113 MOR409 NCS415 OXF122 PAR107 PED108 PIN414 PND165 PNF126 PRK134 PSU106 RCK163 ROM406 SAL133 SAV164 SHN418 SND152 SPD111 STK138 SUM156 UIN162 UVL124 VIN140 VPI120 WEL149 WFM105 WPB104 WSP144 WST109 YEL408 YOS404
Abington Alhambra Ann Arbor Arendtsville Ashland Big Bend NP Beltsville Beaufort Bondville Blackwater NWR Caddo Valley Canyonlands NP Claryville Cedar Creek Chiricahua NM Crockett Candor Centennial Coweeta Connecticut Hill Coffeeville Deer Creek Death Valley NM Egbert Edgar Evins Georgia Station Glacier NP Great Basin NP Grand Canyon NP Gothic Howland Joshua Tree NM Kane Exp. Forest Lassen Volcanic NP Lilley C. Woods Laurel Hill Lye Brook Lykens Mackville Mesa Verde NP M. K. Goddard Mount Rainier NP North Cascades NP Oxford Parsons Prince Edward Pinnacles NM Pinedale Cranberry Perkinstown Penn State Reynolds Creek Rocky Mtn NP Salamonie Reservoir Saval Ranch Shenandoah NP Sand Mountain Speedwell Stockton Sumatra Uinta Unionville Vincennes Horton Station Wellston Whiteface Mountain West Point B Wash. Crossing Woodstock Yellowstone NP Yosemite NP
state latitude CT IL MI PA ME TX MD NC IL MD AR UT NY WV AZ KY NC WY NC NY MS OH CA ON TN GA MT NV AZ CO ME CA PA CA KY PA VT OH KY CO PA WA WA OH WV VA CA WY NC WI PA ID CO IN NV VA AL TN IL FL UT MI IN VA MI NY NY NJ NH WY CA
41.841 38.869 42.416 39.923 46.604 29.31 39.028 34.884 40.051 38.445 34.179 38.458 41.941 38.879 32.01 37.921 35.264 41.372 35.061 42.401 34.003 39.636 36.508 44.232 36.039 33.179 48.51 39.005 36.059 38.957 45.216 34.071 41.598 40.536 37.13 39.988 43.05 40.917 37.704 37.198 41.425 46.761 48.539 39.531 39.091 37.166 36.485 42.921 36.104 45.206 40.721 43.21 40.277 40.816 41.29 38.523 34.289 36.469 42.287 30.107 40.55 43.614 38.741 37.33 44.224 44.39 41.35 40.313 43.944 44.565 37.711
longitude
elevation (m)
land use
terrain
-72.0111 -89.6219 -83.9019 -77.3078 -68.4142 -103.177 -76.8175 -76.6213 -88.3719 -76.1115 -93.0989 -109.821 -74.5514 -80.8479 -109.389 -83.0658 -79.837 -106.242 -83.4302 -76.6536 -89.7989 -83.258 -116.838 -79.784 -85.734 -84.4053 -113.995 -114.215 -112.182 -106.985 -68.7092 -116.39 -78.7683 -121.572 -82.99 -79.2522 -73.0634 -82.9981 -85.0483 -108.49 -80.1447 -122.121 -121.446 -84.7231 -79.6614 -78.3068 -121.155 -109.787 -82.0448 -90.5978 -77.9319 -116.75 -105.545 -85.6608 -115.86 -78.4364 -85.9704 -83.8272 -90 -84.9938 -110.32 -83.3597 -87.4844 -80.5573 -85.8186 -73.86 -74.05 -74.8728 -71.7017 -110.4 -119.704
209 164 267 269 235 1052 46 2 212 4 71 1814 765 234 1570 455 198 3178 686 501 134 267 125 251 302 270 976 2060 2073 2926 69 1244 622 1756 335 615 730 303 353 2165 384 421 109 284 510 150 335 2388 1219 472 378 1198 2743 250 1873 1073 352 361 274 14 2500 201 134 920 295 570 203 61 258 2469 1605
urban/agric agric forested agric agric forested range agric agric forested/marsh forested desert forested forested range agric forested forested forested forested forested agric desert agric forested agric forested forested forested range forested desert forested forested agric forested forested agric agric forested forested forested forested agric forested forested forested forested forested agric agric range forested agric forested forested agric agric agric forested range agric agric agric forested forested forested range forested agric forested
rolling flat flat rolling flat complex flat flat flat coastal complex complex complex complex complex rolling rolling complex complex rolling rolling rolling complex rolling rolling rolling complex complex rolling complex rolling complex rolling complex rolling complex mountaintop flat rolling complex rolling complex complex rolling complex rolling complex rolling mountaintop rolling rolling rolling complex flat rolling mountaintop rolling rolling rolling flat complex flat rolling mountaintop flat complex complex rolling complex rolling complex
1990-1993 1997-2000 summary dry total dry total tables x x x x
x x x x
x
x
x
x
x
x
x x
x
x x x x x x
x
x x x
x x x
x x
x x
x
x
x x
x x
x x
x x
x
x
x x x
x x x
x
x
x x x
x x x
x x x x
x x x x
x x x x x x x x x x
x x x x x x x x x x
x x
x x x x x x x x x x x x x x x x x x x x x x x x x
x x x x x x x x x x x x x x x x x x x
x x
x x x x
x
x x x x x x x x
x x x x x x x x
x x x x x x x x x x x x x x x
x x x x x x x x x x x x x x x
x x
x
x
x x x x x
x x x x x
x x
x x x
x x x
x x x
x x x x
x x x x
x x
VOL. 36, NO. 12, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
x x x x x x x x x
x x x x
x x x x x x x x x
x
9
2617
deposition were not a routine part of CASTNet. The impact of cloud deposition on high elevation sites is discussed in a number of recent publications (3, 24, 25). Total sulfur deposition is defined as the sum of gaseous SO2 and wet and dry SO42-. Total nitrogen deposition is defined as the sum of gaseous HNO3 plus wet and dry NO3- and NH4+. Ammonia (NH3) was not measured as a part of CASTNet but may be a significant portion of total nitrogen, especially in agricultural areas (26). Dry Deposition Estimation. Dry Deposition Model. CASTNet was designed based on the assumption that dry deposition or flux could be estimated as the linear product of ambient concentration (C) and a deposition velocity (Vd):
flux ) CVd
(1)
The influence of meteorological conditions and vegetation on the dry deposition process is simulated by Vd. Vd is calculated using an inferential model that uses resistances to deposition to model the naturally occurring dry deposition processes (7, 8, 27):
Vd ) (Ra + Rb + Rc)-1
(2)
Ra, the aerodynamic resistance, is inversely proportional to the atmosphere’s ability to transfer material downward from the planetary boundary layer to the surface layer by turbulent processes. Rb is the quasi-laminar resistance to transport through the thin layer of air in contact with the surface. Rb depends on the aerodynamics of the surface and the diffusivity of the pollutant being deposited. Rc, the canopy or surface uptake resistance, contains several terms (represented as parallel resistances) that account for the direct uptake/absorption of the pollutant by leaves, soil, other biological receptors within and below the canopy, and other surfaces such as rock and water. Rc contains parametrizations for vegetation type and density, solar radiation penetration of the canopy, and wetness of the surface. Rc is difficult to treat theoretically, and the system of equations for estimating Rc is normally empirically adjusted based on direct measurement of dry fluxes. The multilayer model (MLM) described by Meyers et al. (27) is used to estimate deposition velocity. The MLM divides a canopy into 20 layers in contrast to earlier versions in which the canopy was treated as a single layer. For each layer of the canopy, leaf boundary layer and stomatal resistances are determined using radiative transfer and wind profile models. The MLM requires wind speed and standard deviation of wind speed at 10 m, temperature, relative humidity, solar radiation, surface wetness, and presence or absence of precipitation. The model also requires LAI, type and height of vegetation, and percent leaf out. The deposition velocity is computed on an hourly basis. The site Vd is calculated using area-weighted vegetation types within 1 km of the site. Hourly Vd and weekly integrated concentrations of SO2, SO42-, HNO3, NO3-, and NH4+ were used to produce a weekly deposition flux. Model Uncertainty. Uncertainties in MLM calculations can be considered in the context of the model formulation, input errors, and representativeness. The MLM itself is an imperfect representation of the large number of complex atmospheric processes the model simulates. Consequently, the MLM calculations will never be able to match observed deposition velocities perfectly. Instrument errors and incomplete or inaccurate characterization of other input data (e.g., evolution of LAI during the year) also produce uncertainties in model calculations. No standards or standard methods are available to determine the accuracy of the MLM. Comparison to a direct measurement of deposition is considered the most accept2618
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 12, 2002
able. The eddy correlation method is the most widely used to measure SO2 flux, and the gradient method is used for HNO3. Two recent studies have compared the MLM with direct measurements of SO2 and HNO3 dry deposition. Meyers et al. (27) compared the MLM with the eddy correlation and the gradient method measurements at three sites with different vegetation (pasture, corn and soybeans). Finkelstein et al. (28) compared the MLM at two forested sites (deciduous and mixed coniferous-deciduous). These studies were conducted during the growing seasons of 1994-1995 and 1996-1998, respectively. Data were gathered during times of fast plant and slow plant growth and day and night periods. The results of the comparison were presented as model bias. Bias was defined as the observed value Vd minus the simulated model value Vd. The precision of the comparison was the standard deviation of individual bias determinations. The results from Meyers et al. (27) indicated that the MLM underestimated the Vd for SO2 during the day in periods of fast-growing vegetation (soybeans and corn) and more nearly estimated the measured value at night. If all values are averaged, the bias was low: -0.05 cm/s for corn, -0.04 cm/s for grass, and -0.15 cm/s for soybeans as compared to an averaged Vd of 0.47, 0.82, and 0.58 cm/s, respectively. For HNO3, the comparisons were made only during the day. Across the three types of vegetation, the overall bias was low. However, the model overestimated some values and underestimated others. Also, considerable scatter was evident between individual values. The bias for corn was 0.09 cm/s, 0.47 cm/s for grass, and 0.22 cm/s for soybeans. The average HNO3 Vd was 2.5, 3.2, and 2.8 cm/s, respectively. In summary, the MLM underestimated dry SO2 flux by approximately 10% and overestimated HNO3 flux by about 10%. The results from the work by Finkelstein et al. (28) on dry deposition to forest vegetation showed that the MLM underestimated SO2 Vd for both deciduous and mixed coniferous-deciduous. For deciduous canopies, the model underestimated the SO2 Vd to a greater degree during the day with an average bias of 0.29 cm/s as compared to a nighttime average bias of 0.14 cm/s. For the mixed coniferousdeciduous canopy, the average bias was 0.36 cm/s during the day and 0.07 cm/s for the night. The average Vd for deciduous forests for all values was 0.60 cm/s and for mixed coniferous-deciduous was 0.72 cm/s. Also, there was significant scatter between observations for both types of vegetation. Finkelstein et al. (28) did not report comparisons of direct measurements and models estimates for HNO3. In summary, the MLM significantly underestimated the SO2 deposition velocity, i.e., from about 25 to 50% for the deciduous forest setting and from about 10 to 50% for the mixed forest. Meyers et al. (27) and Finkelstein et al. (28) did not compare modeled Vd with direct deposition measurements for particulate NO3-, SO42-, or NH4+. The MLM calculates a deposition velocity for fine particles (
