United States Environmental Protection Agency Environmental Monitoring Systems Laboratory Research Triangle Park NC 27711 Research and Development EPA/600/S4-88/027 Sept. 1988 Project Summary Comparison of the RADM Dry Deposition Module with Site - Specific Routines for Inferring Dry Deposition M. L. Wesely and B. M. Lesht The computer module for calculating dry deposition velocities in the Regional Acid Deposition Model (RADM) has been modified to operate with site-specific data provided by measurement stations such as CORE (COre Research Establish- ment, for dry deposition) satellite sites. Data collected during 1986 at seven widely separated sites in the eastern United States were used to estimate weekly averages of deposition velocities for SO2, 03, HN03, and S042'. Similar calculations were made with the inferen- tial technique that was developed at At- mospheric Turbulence and Diffusion Division of the National Oceanic and At- mospheric Administration's Air Re- sources Laboratory. Comparison of results obtained with the two techniques indicate that some systematic dif- ferences exist, even when the module uses distributions of landuse types that match as closely as possible the observ- ed vegitation coverages used in the in- ferential technique. When one ignores the systematic differences that could be removed by minor changes in the algorithms for computing resistances to deposition, the relative uncertainties for S02and O3 deposition velocities are ap- proximately ±30%. Likewise, the relative uncertainties corresponding to non- systematic differences in the deposition velocities for HN03 and SO42' are about ±30% and ±50%, respectively. Use of the landuse map to extrapolate to areas as large as RADM qrid cells (approx- imately 8O km square) around the measurement sites produces weekly averages of deposition velocities for SO2, O3, and SO42' that are within ±20% of those computed for the local site and approximately ±30% for HNO3, if one avoids landuse types such as ur- ban and water areas that are nonrepresentatlve and have very dif- ferent characteristics from the measure- ment sites. These estimates do not represent true uncertainty or accuracy because of unknown sources of error possible in the input data and deposition velocity algorithms for nonuniform surfaces. This Project Summary was developed by EPA's Environmental Monitoring Systems Laboratory, Research Triangle Park, NC, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction The dry deposition module of the Regional Acid Deposition Model (RADM) is used to compute the dry deposition velocities (downward flux divided by con- centration at a specified height) for SO2, SO42', O3, HNO3, and other compounds. With its computerized landuse map, the module can provide estimates of the deposition velocities for a given set of meteorological and surface conditions that are representative of an area located anywhere in the contiguous United States and nearby locations. As part of this proj- ect, the module uas been modified to ac- cept data on meteorological and surface conditions observed at measurement sta- tions rather than inputs computed with ------- RADM. The purpose of these modifications was to make site-specific estimates of dry deposition velocities, and compare them with results from the site-specific inferen- tial technique that has been developed at the Atmospheric Turbulence and Diffusion Division (ATDD) of NOAA's Air Resources Laboratory. Goals of the comparisons in- clude estimating the relative uncertainties of the two techniques, suggesting im- provements, and examining the ability of the modified module to provide estimates of dry deposition for expanded areas around the measurement sites. The observational data used in these analyses were collected with ATDD in- struments during 1986 at the seven sites listed in Table 1. The sites identified are either CORE (COre Research Establish- ment, for dry deposition) stations or CORE satellite stations. The data set itself consists of hourly averages of wind speed, the standard deviation of the horizontal wind direction, solar irradiation, air temperature, relative humidity, and surface wetness from a sensor that indicates the presence of moisture from dewfall and rainfall. Other data include one observation per week of the fraction of full leaf cover, from which the seasonal categories used in the module are derived, and the information given in Table 1 describing the types and amounts of vegetative species present at the local site. The landuse distributions for the local site are determined from the relative amounts of these species, while the distributions for the RADM grid squares (approximately 80 km square) encompassing each site are derived entirely from the computerized RADM landuse map. Although the deposi- tion velocities (vd) are computed for every hour, the results considered here are averages over periods of at least one week. Results and Conclustions The dry deposition module of the RDAM has been successfully adapted to use site- specific data obtained at dry deposition sta- tions such as the CORE satellite sites. The module was originally written to use more complete local micrometeorological infor- mation than is included in the parameters listed above, so considerable manipulation of the coding was needed in the RADM module in order to accept these inputs. Perhaps the least exact formaulations in the modified module are those that deal with estimating atmospheric stability, but uncer- tainties introduced by the compromised equations are probably overshadowed by the fact that most of the sites are located in areas with surface properties insufficient- ly uniform, or with terrain insufficiently flat, to allow very precise micrometeorological calculations. The algorithms in the modified module that are used to compute crucial parameters such as aerodynamic resist- ance above the surface and friction veloci- ty are quite similar to those used in the in- ferential technique. Some notable differ- ences exist, however. One is that the modi- fied module attempts to mimic the RADM module by computing changes in the gas- phase resistances for surfaces identified by landuse types different from the dominant landuse type for the local site. At most of the sites, this alteration produces smaller aerodynamic resistances and larger friction velocities averaged over the areas sur- rounding the site, and thus increases the deposition velocities for HNO3. The deposi- tion velocities for other substances do not seem to be significantly affected, because their surface resistances are usually much larger than the aerodynamic and gas- phase sublayer resistances (which depend on friction velocity). The alogrithms used in the modified module to compute the bulk surface resist- ance, usually the most important term for determining deposition velocities, are the same as those used in the RADM module. Some changes had to be made in the Table 1. Site names, locations, predominant surface vegetation (within one kilometer), and landuse types assumed with the modified module General Location (deg lat., long.) Site ID Local Surface Vegetation and Landuse Types* (percent coverage) Argonne National Laboratory, IL 41.TON, 87.98 W) Bondville (Champaign), IL (40.05 N, 88.37 W) Oak Ridge, TN (35.96 N, 84.28 W) Panola State Park, GA (33.63 N, 84.18 W) Pennsylvania State Univ., PA (40.72 N, 77.92 W) West Point, NY (41.35 N, 74.05 W) Whiteface Mountain, NY (44.39 N, 73.86 W) ARG BON OAK PAN PSU WST WHT grass (SO), white oak (SO) local site: 2(50), 4(50) RADM square: 1(23), 2(62), 4(2), 7(13) maize (50), soybeans (50) local site: 2(100) RADM square: 1(2), 2(98) white oak (70), loblolly pine (30) local site: 4(70), 5(30) RADM square: 1(3), 2(30), 4(64), 7(3) Loblolly pine (50), chestnut and red oak (50) local site: 4(50), 5(50) RADM square: 1(13), 2(22), 4(28), 5(6), 6(29), 7(2) maize (30), white oak (30) local site: 2(50), 4(50) RADM square: 1(4), 2(20), 4(73), 5(3) maple (60), white oak (30) local site: 4(100) RADM square: 1(41), 2(3), 4(28), 6(2), 7(26) white birch (70), maple (10) local site: 4(100) RADM square: 1(1), 2(6), 4(28), 5(10), 6(43), 7(12) 'Landuse types specified are [1] urban land fresh water. [2] agricultural land, [4] deciduous forest, [5] coniferous forest, [6] mixed forest including wetland, and [7] water, both salt and ------- parameterizations of the SO<2 sublayer resistance in air, which strongly controls SO42" deposition velocity, but the changes conform with the interpretation of the field experiments on which the parameteriza- tions are based. The inferential technique is configured somewhat differently and ap- pears to overestimate the deposition veloci- ty for SO42- in some situations. The mod- ule, however, tends to underestimate when the standard deviation of the horizontal wind direction is not measured very well. The most striking difference between the results of the module and the inferential technique is the much smaller estimates made with inferential technique for SO2 and 03 deposition velocities during nonsummer conditions. These are probably underesti- mates caused by overly restrictive assump- tions on the surface resistances of bare soil and sparse vegetation during nonsummer seasons, which could easily be adjusted by changing the values of a few numerical coefficients used in the inferential technique. While the algorithms for surface resistance in the modified module are bas- ed on a more complex and versatile set of multiple surface resistances to various por- tions of the surface than the algorithms us- ed in the current version of the inferential technique, the module is strongly limited by the fact that the surface can be described by at most 11 landuse types. The inferen- tial technique employs description of vegetation by species and tailors the sur- face resistances more precisely than the module to describe the physiological re- sponses to environmental parameters such as solar irradiation and temperature. An ad- ditional feature that may be desirable is a factor to take into account soil moisture stress on the stomatal resistances of healthy vegetation during the daytime, because the current version of the inferen- tial technique assumes no soil moisture stress and the module is effectively tuned to a small amount of moisture stress. As a result, the deposition velocities of substances such as SO2 and O3 that are strongly influenced by stomatal resistances are slightly larger when computed by the inferential technique than with the modified module. For the weekly averages of deposition velocities computed for 1986 at the seven sites, the inferential technique and the modified module produce estimates within about ±3O% of each other for SO2 and O3, if we neglect strong systematic differences such as those that occur during nonsum- mer conditions at surfaces with bare soil or sparse vegetative canopies. Likewise, the differences for HNO3 and SO42' deposition velocities are usually about ±30% and ±50%, respectively. These relative uncertainties could probably be reduced somewhat if the suggestions made above were adopted. Estimates of deposition velocities over large areas are easily computed with the modified module by use of the appropriate landuse distributions. Of course, if the areas are very large, the observations of meteorological conditions at the measure- ment stations might not be sufficiently representative. Some trial runs involving the 20-km squares of the landuse map cell and the 80-km RADM cells at the seven measurement sites show that the estimates of weekly averaged deposition velocities for SO2 ,O3 and SO42' are fairly sensitive to the amounts of urban and water areas in- cluded in landuse distributions. Aside from this effect, the deposition velocities caiculated for these three substances over the RADM grid cells are usually within ±20% of those for the local site descrip- tions. These limits should not be taken as generally valid at sites other than the seven examined in this study, because different meteorological conditions and altered com- binations of landuse types could cause larger changes in deposition velocities. For HNO3, minor changes in the assumed lan- duse distributions often cause rather large changes, easily ±30%, in deposition velocities. We conclude that the differences in deposition velocity estimates caused by scaling from local site conditions to RADM grid cells are often slightly smaller than the relative uncertainties of the deposition velocities estimated with the modified module and the inferential technique. Hence, inferences of deposition velocities for the local conditions at carefully chosen sites appear to be reasonably represen- tative of larger areas extending at least to sizes of RADM grid squares. Nevertheless, adjustments for changes in surface condi- tions are clearly desirable, and extrapola- tion to types of surfaces that have notably different properties can be made only at the cost of considerably greater uncertainty. By use of a computerized landuse map or, preferably, a more detailed description of surface conditions, one should be able to estimate weekly averaged dry deposition amounts for sulfur, nitric acid, and ozone to within about ±30% over some fairly large areas. The above number of 30% is not the true uncertainty or accuracy. An assessment of accuracy would require experiments designed specifically for dry deposition, because a number of factors can cause ad- ditional uncertainty. For example, one must determine how representative the local meteorological measurements actually are. Relatively low wind speeds measured at OAK and PAN suggest that proper ex- posure of instruments in a forested, hilly area might be difficult. In addition, measurements taken at sites not represen- tative of any one surface or influenced by local aerodynamic obstacles might not be suitable for application of the algorithms used in the inferential and the module ap- proaches. Finally, these algorithms do not directly consider the effects of surface nonuniformities in the area considered, in- cluding hills, isolated surface irregularities, and edges associated with changes in height of surface cover. The aerodynamic resistances could be strongly altered by surface nonuniformities, and stomatal re- sistances could be altered by varying degrees of shading of sunlight by hills and other vegetation. ------- M. L Wesely and B. M. Lesht are with Argonne National Laboratory, Argonne, IL 60439. Steven M. Bromberg is the EPA Project Officer (see below). The complete report, entitled "Comparison of the RADM Dry Deposition Module with Site-Specific Routines for Inferring Dry Deposition," (Order No. PB 88- 238 191/AS; Cost: $19.95, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Environmental Monitoring Systems Research Laboratory U.S. Environmental Protection Agency Research Triangle Park, NC 27711 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 Official Business Penalty for Private Use $300 EPA/600/S4-88/027 0000329 AGENCY ------- |