Acid Deposition: Atmospheric Processes in Eastern North America (1983)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Appendix Atmospheric Deposition Processes 1. INTRODUCTION In this appendix we present an overview of current scientific understanding about deposition phenomena, with the objectives of identifying key literature sources on this subject and providing the reader with the technical basis necessary for effective evaluation of the available literature. There are several important features of this subject, which should be noted at the outset. First, the ultimate deposition processes of interest are the end products of a complex sequence of atmospheric phenomena (cf. Figure 2.1). Deposition processes tend to reflect these preceding events strongly. Much of the material presented in this appendix therefore necessarily deals with the predeposition processes, which may act as important rate-influencing steps in the overall source-deposition sequence. A second important feature of initial interest is the relative difference in states of our current understanding of wet- and dry-deposition phenomena. Wet deposition is comparatively simple to measure. As a consequence there exists a substantial and growing base of data on wet deposition from a variety of networks and field studies. Precipitation processes tend to be rather complicated, however, and currently a high level of uncertainty exists regarding their mathematical characterization. Dry deposition, on the other hand, tends to be extremely difficult to measure, and the corresponding data set is relatively meager. Partly because of this fact most mathematical characterizations of dry- deposition processes have been quite simple in form. The tendency toward simplicity in most mathematical char- acterizations of dry deposition should not be taken to 213

214 imply that the physical processes themelves are simple. As a consequence of these differences the following sec- tions on dry and wet deposition have somewhat different formats, with emphasis in each placed on areas of current major activity. Finally, it should be noted that very little of the material presented in this appendix is new. A number of reviews of both wet and dry deposition have been presented during recent years, and the current treatment is merely an attempt to consolidate these efforts.* In view of this tendency toward redundancy, it is strongly recommended that the reader proceed directly to the indicated journal literature if more detailed pursuit of this subject is desired. 2. DRY-DEPOSITION PROCESSES 2.1 MECHANISMS OF DRY DEPOSITION 2.1.1 Introduction The rate of transfer of pollutants between the air and exposed surfaces is controlled by a wide range of chemical, physical, and biological factors, which vary in their relative importance according to the nature and state of the surface, the characteristics of the pol- lutant, and the state of the atmosphere. The complexity of the individual processes involved and the variety of possible interactions among them combine to prohibit easy generalization; nevertheless, a Deposition velocity," Vd, analogous to a gravitational falling speed, is of considerable use. In practice, knowledge of vd enables fluxes, F. to be estimated from airborne concentrations, C, as the simple product, vd. C. *Much of the material presented in this appendix was prepared by Drs. B.B. Hicks and J.M. Hales as a contribution to the Critical Assessment Document on Acidic Deposition being prepared by North Carolina State University under a cooperative agreement with the U.S. Environmental Protection Agency. These contributions are published here with permission of the authors and the concurrence of the editors of the Critical Assessment Document, Drs. A.P. Altschuller and R.A. Linthurst.

215 Particles larger than about 20-pm diameter will be deposited at a rate that is controlled by Stokes law, although with some enhancement due to inertial impaction of particles transported to near the surface in turbulent eddies. The settling of submicrometer-sized particles in air is sufficiently slow that turbulent transfer tends to dominate, but the net flux is often limited by the presence of a quasi-laminar layer adjacent to the surface, which presents a considerable barrier to all mass fluxes and especially to gases with very low molecular diff.,civi~= ~ a. The concept of a gravitational settling velocity is inappropriate in the case of gases, but transfer is still often limited by diffusive properties very near the receptor surface. Sehmel (1980b) presents a tabulation of factors known to influence the rate of pollutant deposition upon exposed surfaces. Figure C.2-1 has been constructed on the basis of Sehmel's list and has been organized to emphasize the greatly dissimilar processes affecting the fluxes of gases and large particles. Small, submicro- meter-diameter particles are affected by all the factors indicated in the diagram; thus, simplification is especially difficult for deposition of such particles. In reality, Figure C.2-1 already represents a consider- able simplification, since many potentially important factors are omitted. In particular, the emphasis of the diagram is on properties of the medium containing the pollutants in question; a similarly complicated diagram could be constructed to illustrate the effects of pol- lutant characteristics. For particles, critical factors include size, shape, mass, and Nettability; for gases, concern is with molecular weight and polarization, solubility, and chemical reactivity. In this context, the acidity of a pollutant that is being transferred to some receptor surface by dry processes is a quality of special importance that may have strong impact on the efficiency of the deposition process itself. Figure C.2-2 summarizes particle size distributions on a number, surface area, and volume basis. In this way, the three major modes are brought clearly to attention. The number distribution emphasizes the transient (or Aitken) nuclei range, 0.005-0.05-pm diameter, for which diffusion plays a role in controlling deposition. The area distribution draws attention to the so-called accumulation size range formed largely from gaseous precursors (0.05-2-pm diameter, affected by both diffusion and gravity). The remaining mode (2-50-~m

216 | AIRBORNE SOURCE | LARGE PARTICLES AERODYNAMIC | SETTLING l FACTO RS GASES I> TO R8U LENCE ~-~ TO R8U LENCE TH E RMOPH O R ESIS NEAR-SU RfACE PRORETIC ELECTROPHORESIS EFFECTS Dlf FUSI OPHO RESIS & STEFAN FLOW | I MPACTI ON QUASI-LAMINAR ~ LAYER I INTERCEPTION l f ACTO RS it= | 3ROWNIAN DlfFUSION SU R FACE PROPE RTI ES ~:r ~ STEFAN FLOW I l ~ MOLECULAR DlffUSION 1 | ORIENTATION| ~ STOMATA | | WETNESS | 1 . . | FLEXIBILITY | | WAXINESS | I CHEMISTRY I (SMOOTHNESS | | VESTITURE | | EMISSIONS | | MOTION 1 | EXUDATES 1 . 1 RECEPTOR l  FIGURE C.2-1 A schematic representation of processes likely to influence the rate of dry deposition of airborne gases and particles. Note that some factors affect both gaseous and particulate transfer, whereas others do not. However, submicrometer particles are affected by all the factors that influence gases and large particles, and hence it is these "accumulation-size-range" aerosols that present the greatest chal- lenge for deposition research.

217 -- ' ' 1 ' 1 1 ' ' 1 ' ' 1 '-- ' 1 FINE PARTICLES ~ COARSE PARTICLES | ACCUMULATION I MECHANICALLY I SIZE I GENERATED RANGE I PARTICLES A V , , 1 , , 1 , , 1 1 1 0.001 O.Ot 0.' 1.0 PARTICLE DIAMETER {pm) 10 100 FIGURE C.2-2 A hypothetical particle-size spectrum, such as might be found down- wind of an industrial complex. The smaller aerosols have gaseous precursors and are formed by condensation of exhaust gases and by atmospheric chemical reactions (typically oxidation), followed by growth due to particle coagulation. The larger particles are partly soil-derived, suspended by natural erosion and agricultural practices, and partly the direct result of the combustion of fossil fuels. Acidic aerosols are pn- marily in the smaller mode of the particle-size spectrum, whereas the larger mode contains material that might tend to neutralize the acidic deposition of the smaller particles. In evaluating the net input of acidity to a surface, it is critical that both size fractions and gaseous contributions be included. diameter, most evident in the volume distribution) is the mechanically generated particle range for which gravity causes most of the deposition. In most literature, 2-pm diameter is used as a convenient boundary between ~fine" and "coarse" particles. Atmospheric sulfates, nitrates, and ammonium compounds are primarily associated with the accumulation size range. Figure C.2-2 demonstrates that very little acidic or acidifying material is likely to be associated with the coarse particle fraction in background conditions. However, the larger particles include soil-derived minerals, some of which can react chemically with airborne and deposited acids. Moreover, it has been

218 suggested that some of these larger particles may provide sites for the catalytic oxidation of sulfur dioxide (for example when the particles are carbon; Chang et al. 1981, Cofer et al. 1981). Little is known about the detailed chemical composition of large particle agglomerates. However it is accepted that their residence time is quite short (i.e., they are deposited relatively rapidly), that there are substantial spatial and temporal variations in both their concentrations and their composition, and that their contribution to acid dry deposition should not be ignored. To evaluate deposition rates, several different approaches are possible. Field experiments can be conducted to monitor changes in some system of receptors from which average deposition rates can be deduced. More intensive experiments can measure the deposition of particular pollutants in some circumstances. Neither approach is capable of monitoring the long-term, spatial- average dry deposition of pollutants. To understand why, we must first consider in some detail the processes that influence pollutant fluxes and then relate these consid- erations to measurement and modeling techniques that are currently being advocated. The logical sequence illus- trated in Figure C.2-1 will be used to guide this discussion. 2.1.2 Aerodynamic Factors Except for the obvious difference that particles will settle slowly under the influence of gravity, small particles and trace gases behave similarly in the air. Trace gases are an integral part of the gas mixture that constitutes air and thus will be moved with all the turbulent motions that normally transport heat, momentum, and water vapor. However, particles have finite inertia and can fail to respond to rapid turbulent fluctuations. Table C.2-1 lists some relevant characteristics of spherical particles in air (based on data tabulated by Davies 1966, Friedlander 1977, and Fuchs 1964). The time scales of most turbulent motions in the air are con- siderably greater than the inertial relaxation (or stopping) times listed in the table. These time scales vary with height, but even as close as 1 cm from a smooth, flat surface, most turbulence energy will be associated with time scales longer that 0.01 see, so that even 100-pmrdiameter particles would follow most turbulent

219 TABLE C.2-1 Dynamic Characteristics of Unit Density Aerosol Particles at STP, Corrected for Stokes-Cunningham Effectsa Particle Stopping Settling Radius Dif fusivity Time Speed (~m) (cm2/s) (s) (cays) 0.001 1.28 x 10-2 1.33 x 10-9 1.30 x 10-6 0.007 3.23 x 10-3 2.67 x 10-9 2.62 x 10-6 0.005 5.24 x 10-4 6.76 x 10-9 6.62 x 1-6 0.01 1.35 x 10-4 1.40 x 10-8 1.37 x 10-5 0.02 3.59 x 10 5 2.97 x 10-8 2.91 x 10-5 0.05 6.82 x 10-6 8.81 x 1o~8 8.63 x 10-5 0.1 2.21 x 10-6 2.28 x 10-7 2.23 x 10-4 0.2 8.32 x 10-7 6.87 x 10-7 6.73 x 10-4 0.5 2.74 x 10-7 3.54 x 10-6 3.47 x 10-3 1.0 1.27 x 10-7 1.31 x 10-5 1.28 x ~o~2 2.0 6.10 x 10-8 5.03 x 10-5 4.93 x 10-2 5.0 2.38 x 10-8 3.08 x 10-4 3.02 x 10~ 10.0 1.38 x 10-8 1.23 x 10-3 1.2 x 10° _ ... . aData are from Fuchs tl964), Davies (1966), and Friedlander (1977) fluctuations. However, natural surfaces neither smooth nor flat, and it is clear that In many circumstances the flux of particles will be limited by their inability to respond to rapid air motions. Naturally occurring aerosol particles are not always spherical, although it seems reasonable to assume so in the case of hydroscopic Particles in the submicromet. r size range. are normally ~ _ ~ - ~ ~ lo.. ~ _ ~ ~ _ ~ _ Chamberlain (1975) documents the ratio of the terminal velocity of nonspherical particles to that of spherical particles with the same volume. In all cases, the nonspherical particles have a lower terminal settling speed than equivalent spheres. The settling speed differential is indicated by a dynamical shape factor," a, as listed in Table C.2-2. Thus, trace gases and small particles are carried by atmospheric turbulence as if they were integral come portents of the air itself, whereas large particles are also affected by gravitational settling, which causes them to fall through the turbulent eddies. In general, however, the distribution of pollutants in the lower atmosphere is governed by the dynamic structure of the atmosphere as much as by pollutant properties. .

220 TABLE C.2-2 Dynamic Shape Factors as by which Nonspherical Particles Fall More Slowly than Spherical (from Chamberlain, 1975) Ratio Shape of axes ~ Ellipsoid 4 1.28 Cylinder 1 1.06 Cylinder 2 1.14 Cylinder 3 1.24 Cylinder 4 1.32 Two spheres touching, vertically 2 1.10 Two spheres touching, horizontally 2 1.17 Three spheres touching, as triangle - 1.20 Three spheres touching, in line 3 1.34-1.40 Four spheres touching, in line 4 1.56-1.58 In daytime, the lower atmosphere is usually well mixed up to a height typically in the range 1 to 2 km, as a consequence of convection associated with surface heating by insolation. Pollutants residing anywhere within this mixed layer are effectively available for deposition through the many possible mechanisms. However, at night, the lower atmosphere becomes stably stratified and vertical transfer of nonsedimenting material is so slow that, at times, pollutants at heights as low as 50 to 100 m are isolated from surface deposition processes. Thus, in daytime, atmospheric transfer does not usually limit the rate of delivery of pollutants to the surface bound- ary layer in which direct deposition processes are active. The fine details of turbulent transport of pollutants remain somewhat contentious. Notable among the areas of disagreement is the question of flux-gradient relation- ships in the surface boundary layer. It is now well accepted that the eddy diffusivity of sensible heat and water vapor exceeds that for momentum in unstable (i.e., daytime) but not in stable conditions over fairly smooth surfaces (see Dyer 1974, for example). However, it is not clear that the well-accepted relations governing either heat or momentum transfer are fully applicable to the case of particles or trace gases; some disagreement exists even in the case of water vapor. The situation is

221 even more uncertain in circumstances other than over large expanses of horizontally uniform pasture. When vegetation is tall, pollutant sinks are distributed throughout the canopy so that close similarity with the transfer of more familiar quantities such as heat or momentum is effectively lost. There is even considerable uncertainty about how to interpret profiles of tempera- ture, humidity, and velocity above forests (see Garratt 1978, Hicks et al. 1979, Raupach et al. 1979). 2.1.3 The Quasi-laminar Layer In the immediate vicinity of any receptor surface, a number of factors associated with the molecular dif- fusivity and the inertia of pollutants become important. Large particles carried by turbulence can be impacted on the surface as they fail to respond to rapid velocity changes. The physics of this process is similar to the physics of sampling by inertial collection. Inertial impaction is a process that augments gravi- tational settling for particles that fall into a size range typically between 2- and 20-pm diameter (q.v. Slinn 1976b). Larger-sized particles tend to bounce, and capture is therefore less efficient, while smaller-sized particles experience difficulty in penetrating the quasi- laminar layer that envelops receptor surfaces. From the viewpoint of acidic particles, inertial impaction is a process of questionable relevance since most acidic species are associated with smaller particles (see Figure C.2-2), which are not strongly affected by this process. However, Figures C.2-2 and C.2-3 show that many airborne materials exist in the size range likely to be affected by inertial impaction. Since many of the chemical constituents of soil-derived particles are capable of neutralizing deposited acids, inertial impaction may have important indirect effects on acidic deposition. To illustrate the role of molecular or Brownian diffusivity, it is informative to consider the simple case of a knife-edged thin plate, mounted horizontally and with edge normal to the wind sector. As air passes over (and under) the ~late, a laminar layer develops, of thickness ~ = c(vx/u) /2, where v is kinematic vis- cocity, x is the downwind distance from the edge of the plate, and u is wind speed. According to Batchelor (1967), the value of the numerical constant c is 1.72. Thus, for a plate of dimensions 5 cm in a wind speed of

222 ~.~.` 1 1 1 1 1 1 111 1 1 1 1 111] CD 10-4 10-5 I. · - o to to to \ 1 i \ 1 1 1 1 1 1 1 111 1 1 1 1 1 1 111 1 1 1 1 i 111 - 2 ~03 104 105 SO FIGURE C.2-3 Laboratory verification of Schmidt-number scaling for particle trans- fer to a smooth surface. The quantity plotted is B _ vd/u*, evaluated for transfer across a quasi-laminar layer of molecular control immediately adjacent to a smooth surface. Data are from Harnott and Hamilton (1965; open circles), Hubbard and Light- food (1966 ; triangles), and Muzushinz et al. (1 971; solid circles), as reported by Lewellen and Sheng (1980~. The line drawn through the data is Equation (C.2-1), with exponent al = -2/3 and constant of proportionality A_ 0.06. 1 m/s, we should imagine a boundary-layer thickness reaching about 1.5-mm thick at the trailing edge. Over nonideal surfaces, the internal viscous boundary layer is frequently neither laminar nor constant with time. The layer generates slowly as a consequence of viscosity and surface drag as air moves across a surface. The Reynolds number Re (_ ux/v, where u is the wind speed, x is the downwind dimension of the obstacle, and v is kinematic viscosity) is an index of the likelihood that a truly laminar layer will occur. For large Re, air adjacent to the surface remains turbulent: viscosity is then incapable of exerting its influence. In many cases, it seems that the surface layer is intermittently turbulent. For these reasons, and because close similarlity between ideal surfaces studied in wind tunnels and natural surfaces is rather difficult to swallow, the term "quasi-laminar layer" is preferred. Wind-tunnel studies of the transfer of particles to the walls of pipes tend to support the concept of a limiting diffusive layer adjacent to smooth receptor

223 surfaces. Transfer across such a laminar layer is conveniently formulated in terms of the Schmidt number, Sc = v/D, where v is viscosity and D is the pollutant diffusivity. The conductance, or transfer velocity vl, across the quasi-laminar layer is proportional to the friction velocity u*: v1 = Au* Sca, (C.2-1) where A and ~ are determined experimentally. Most studies agree that the exponent a is about -2/3, as is evident in the experimental data represented in Figure C.2-3. However, a survey by Brutsaert (1975a) indicates exponents ranging from -0.4 to -0.8. The value of the constant A is also uncertain. The line drawn through the data of Figure C.2-3 corresponds to A ~ 0.06, yet the wind-water tunnel results of Moller and Schumann (1970) appears to require A ~ 0.6. These values span the value of A ~ 0.2 recommended for the case of sulfur dioxide flux to fibrous, vegetated surfaces (Shepherd 1974, Wesely and Hicks 1977). ~ _~: _ ~ _ =_. _ ~. . ~ _ _,= _ _ _ =~'l~a~ w unuary-layer theory Imposes the expectation that particle deposition to exposed surfaces will be strongly influenced by the size of the particle, with smaller particles being more readily deposited hv diffusion than larger. It is clear that many artificial Furnaces or structures made of mineral material will have characteristics for which the laminar-layer theories might be quite appropriate. However, the relevance to vegetation can be questioned. Microscale surface roughness elements can penetrate the barrier presented by this quasi-laminar layer and should be suspected as sites for enhanced deposition of both particles and gases (see Chamberlain 1980). 2.1.4 Phoretic Effects and Stefan Flow Particles near a hot surface experience a force that tends to drive them away from the surface. For very small particles (<0.03-pm diameter, according to Davies 1967), this "thermophoresis" can be visualized as the consequence of hotter, more energetic air molecules impacting the side of the particle facing the hot sur- face. For larger particles, radiometric forces become important (Cadre 1966). In theory, thermal radiation can

224 cause temperature gradients across particles that are not good thermal conductors, resulting in a mean motion of the particle away from a hot surface. In summary, the thermophoresis depends on the local temperature gradient in the air, on the thermal properties of the particle, on the Krudsen number En _ \/r (where ~ is the mean free path of air molecules and r is the radius of the particle), and on the nature of the interaction between the particle and air molecules (see Derjaguin and Yalamov 1972). As a rule of thumb, the thermophoretic velocity of very small particles (<0.03-pm diameter) is likely to be about 0.03 cm/s (estimated from values quoted by Davies 1967). For particles exceeding 1-pm diameter, the velocity will be about four times less. The process of diffusiophoresis results when particles reside in a mixture of intermixing gases. In most natural circumstances, the principal concern is with water vapor. Close to an evaporating surface, a particle will be impacted by more water molecules on the nearer side. Since these water molecules are lighter than air mole- cules, there will be a net "diffusiophoresis" toward the evaporating surface. In essence, these "phonetic" forces result from the flow of molecules of some special kind through the gas mixture and the "drag" exerted on par- ticles. Since diffusiophoresis and thermophoresis depend on the size and shape of the particle of interest, neither can be predicted with precision, nor can safe generaliza- tions be made. These subjects are sufficiently compli- cated that they constitute specialities in their own right. Excellent discussions have been given by Friedlander (1977) and Twomey (1977). These phonetic forces vary with particle size but are generally small, and their influence on dry deposition can usually be disregarded. Many workers include Stefan flow in general discussion of diffusiophoresis, but because of the conceptual dif- ference between the mechanisms involved it seems better to consider them separately. Stefan flow results from injection into the gaseous medium of new gas molecules at an evaporating or subliming surface. Every gram-molecule of substrate material that becomes a gas displaces 22.41 liters of air, at STP. Thus, for example, a Stefan flow velocity of 22.41 mm/s will result when 18 g of water evaporates from a 1-m2 area every second. Generaliza- tion to other temperatures and pressures is straight- forward. Daytime evaporation rates from natural

225 vegetation often exceed 0.2 g/m~2 s for considerable times during the midday period, resulting in Stefan flow of more than 0.2 mm/s away from the surface. Detailed calculation for specific circumstances is quite simple. For the present, it is sufficient to note that Stefan flow is capable of modifying surface deposition rates by an amount that is larger than the deposition velocity appropriate for many small particles to aero- dynamically smooth surfaces. Electrical forces have often been mentioned as possible mechanisms for promoting deposition (as well as the retention, see Section 2.1.5) of small particles, particularly through the "viscous quasi-laminar layer immediately above receptor surfaces. Wason et al. (1973) report exceedingly high rates of deposition of particles in the size range 0.6 to 6 um to the walls of pipes whenever a space charge is present. Chamberlain (1960) demonstrated the importance of electrostatic forces in modifying deposition velocities of small particles, when fields are sufficiently high. Plates charged to produce local field strengths of more than 2000 V cm~1 experi- enced considerably more deposition of small particles than uncharged plates, by factors between 2 and 15. However in fair-weather conditions, field strengths are typically less than 10 V cm~1 so that the net effect on particle transfer is likely to be small. Further studies of the ability of electrostatic forces to assist the transfer of particulate pollutants to vegetative surfaces were conducted by Lange r (1965) and Rosinski and NanomoEo ,, ~ ~ ~ ~ ~ _ _ ~ . . . . ~ _ _ . ` ~ . according to navy (-), a series of experiments was conducted using single conifer needles and conifer trees. "For single needles or leaves, electrical charges on 2-pm-diameter ZnS dust with up to eight units of charge had no detectable effect at wind speeds of 1.2 to 1.6 m/s. The average collection efficiency was found to be 6% for edgewise cedar or fir needles, with broadside values an order of magnitude lower. Bounce-off after striking the collector was not detected, but re- entrainment could take place above 2 m/s wind speed. Tests on branches of cedar and fir by Rosinski and Nagomoto (1965) suggested similar results as for single needles.. It should be noted, however, that the electrical mobility of a particle is a strong negative function of particle size, ranging from 2 cm/s per V/am of field strength for O.001-pm-diameter particles, to 0.0003 cats per V/cm for O.l-pm particles (Davies 1967).

226 2.1.5 Surface Adhesion Most workers assume that pollutants that contact a surface will be captured by it. For some gases, this assumption is clearly adequate. For example, nitric acid vapor is sufficiently reactive that most surfaces should act as nearly perfect sinks. Less reactive chemicals will be less efficiently captured. The case of particles is of special-interest, however, because of the possibil- ity of bounce and resuspension. The role of electrostatic attraction in binding deposited particles to substrate surfaces remains something of a mystery. The process by which particles become charged and set up mirror-charges on the under- lying surface is fairly well accepted. The resulting van der Waals forces are often mentioned as the major mech- anism for binding particles once deposited. For large, nonspherical particles, dipole moments can be set up in natural electric fields, and these can help promote the adhesion at surfaces. These matters have been conveni- ently summarized by Billings and Gussman (1976), who provide mathematical relationships for evaluating the electrical energy of a particle on the basis of its size, shape, dielectric constant, and the strength of the surrounding electrical field. For smaller particles, the principal charging mechanism is thermal diffusion, leading to a Boltzmann charge distribution. Condensation of water reduces the effectiveness of electrostatic adhesion forces, since leakage paths are then set up and charge differentials are diminished. However, the presence of liquid films at the interfaces between particles and surfaces causes a capillary adhe- sive force that compensates for the loss of electrostati .c attraction. These "liquid-bridge" forces are most effective in high humidities and for coarse particles (<20 ~m, according to Corn 1961). Billings and Gussman (1976) draw attention to the effect of microscale surface roughness in promoting adhesion of particles to surfaces. Much of the experi- mental evidence is for particle diameters much greater than the height of surface irregularities (e.g., Bowden and Tabor 1950). It is the opposite case that is likely to be of greater interest in the present context, as will be discussed later.

227 2.1.6 Surface Biological Effects The efficiency with which natural surfaces ~capture. impacting particles or molecules will be influenced considerably by the chemical composition of the surface as well as its physical structure. The Plead candle. technique for detection of atmospheric sulfur dioxide is a historically interesting example of how chemical substrates can be selected to affect the deposition rates of particular pollutants. Uptake rates of many trace gases by vegetation are controlled by biological factors such as stomata! resistance. In daytime, this is known to be the case for sulfur dioxide (Shepherd 1974, Spedding 1969, Wesely and Hicks 1977) and usually for ozone in most situations (Wesely et al. 1978). The similarity between sulfur and ozone is not complete, however, because the presence of liquid water on the foliage will tend to promote SO2 deposition and to impede uptake of ozone; the former gas is quite soluble until the solution becomes too acidic, whereas the latter is essentially insoluble (q.~. Brimblecombe 1978). Pubescence of leaves has received considerable attention. Chamberlain (1967) tested the roles of leaf stickiness and hairiness in a series of wind-tunnel tests. He concludes that With the large particles (32 and 19 Am) the velocity of deposition to the sticky artificial grass was greater than to the real grass, but with those of 5 um and less, it was the other way round, thus confirming . . . that hairiness is more important than stickiness for the capture of the smaller particles.. The importance of leaf hairs appears to be verified by studies of the uptake of 210Pb and 210po particles by tobacco leaves (Fleischer and Parungo 1974, Martell 1974) and by the wind-tunnel work of Wedding et al. (1975), who report increases by a factor of 10 in deposition rates for particles to pubescent leaves, compared with smooth, waxy leaves. It remains to be seen how greatly biological factors of this kind influence the rates of deposition of airborne particles to other kinds of vegetation. 2.1.7 Deposition to Liquid-Water Surfaces Trace-gas and aerosol deposition on open water surfaces is of significant practical interest, especially consid

228 Bring the acidification of poorly buffered inland Air blowing from land across a coastline will slowly equilibrate with the new surface at a rate that is strongly dependent on the stability regime involved. the water is much warmer than upwind land, dynamic instability over the water will cause relatively rapid adjustment of the air to its new lower boundary, but if the water is cooler, stratified flow will occur and adjustment will be very slow. In the former (unstable) case, dry-deposition rates of all soluble or chemically reactive pollutants are likely to be much higher than in the latter. Clearly, air blowing over small lakes will be less likely to adjust to the water surface than when blowing over larger water bodies. Thus, during much of the summer, inland water surfaces will tend to be cooler than the air, and hence protected from dry deposition, because of the strongly stable stratification that will then prevail. This phenomenon will occur more frequently over small water bodies than larger ones (see Hess and Hicks 1975). Following the guidance of chemical engineering gas-transfer studies, workers such as Kanwisher (1963), Liss (1973), and Liss and Slater (1974) have considered the role of Henry's law constant and chemical reactivity in controlling the rate of exchange of trace gases between the atmosphere and the ocean. In general, acidic and acidifying species like S°k are readily removed on contact with a water surface. Thus Hicks and LiSs (1976) neglected liquid-phase resistance and derived net deposition velocities appropriate for the exchange of reactive gases across the air-sea interface. The work of Hicks and Liss is intended to apply to water bodies of sufficient size that the bulk exchange relationships of air-sea interaction research are applicable. Their considerations indicate that deposition velocities for highly soluble and chemically reactive gases such as NH3, HC1, and S02 are likely to be between 0.10 and 0.15 percent of the wind speed measured at 10-m height. The analysis leading to this conclusion assumes that the molecular and eddy diffusivities can be combined by simple addition. This assumption has been shown to approximate the transfer of water vapor and sensible heat from water surfaces. However, for fluxes of trace gases the validity of this assumption is questionable. Slinn et al. (1978) argue that it is better to introduce molecular diffusivity through a term analogous to the Schmidt number of Equation (2-1), with the exponent

229 ~ ~ -2/3. (In contrast, the linear assumption used by Hicks and Liss implies ~ = -1.0.) In view of the uncertainties mentioned in discussion of Equation (C.2-1), further comment on the implications and ramifi- cations of these alternative assumptions is not warranted. In the limiting case of a trace gas of low solubility, the deposition velocity is determined by the large liquid- phase resistance, which is essentially mron~rtimn~1 -^ the Henry's law constant. It is probable that breaking waves will modify the simple gas-transfer formulations derived from chemical engineering pipe-flow and wind-tunnel work. It is not clear to what extent such features account for the apparent discrepancy between the various Schmidt number dependencies of the kind expressed by Equation (C.2-1). However, the fractional power laws are basically extensions of laboratory work, whereas the unit-power, additive-diffusivities result is an approximation to field data. It is to be hoped that the two approaches produce results that will converge in due course. Figure C.2-4 previews the discussion of wind-tunnel particle deposition results that will be given later. Such wind-tunnel work indicates exceedingly low deposi- tion velocities for particles in the size range of most acidic pollutants. As in the case of gas exchange, there are conceptual difficulties in extending these results to the open ocean. The role of waves in the transfer of small particles between the atmosphere and water surfaces remains essentially unknown. Not only does engulfment by breaking waves provide an alternative path across the quasi-laminar sublayer where molecular (or Brownian) diffusion normally controls the transfer, but also waves are a source of droplets that can scavenge particulate material from the air (see, however, the study of Alexander 1967, which indicates otherwise). Hicks and Williams (1979) have proposed a simple model of air-sea particle exchange that extends smooth-surface, wind- and water-tunnel results (as in Figure C.2-4) to natural circumstances by permitting rapid transfer to occur whenever waves break. This results in very low deposition velocities in light winds, but rapidly increasing when winds increase above about 5 m/s. Slinn and Slinn (1980) also suggest that particle transfer is more rapid than the wind-tunnel studies of Figure C.2-4 might indicate but present an alternative hypothesis for this more rapid transfer: that hydroscopic particles grow rapidly when exposed to high humidities such as are ~_.z =_ _~_^ _ ~ ~TV

230 cry - o JO Oh o llJ C) WATER 0.1 0.01 ·t 0.01 L ·~.~. o ·~ 1 1 1 1 1 /1 1 o.1 PAR T ICLE ~ 10 DIAMET E R ( Am ) /° Jo / 0' 1 1 1 FIGURE C.2-4 Results of wind-tunnel studies of particle deposition to water surfaces. Solid circles are due to Moller and Schumann (1979), open circles to Sehmel and Sutter (1974). The dashed line at the right represents the terminal settling speed for 1.5 g cm~3 particles. found in air adjacent to a water surface, resulting in increased gravitational settling and impaction to the water surface. 2.1.8 Deposition to Mineral Surfaces  Acid deposition is an obvious source of worry to architects, historians, and others concerned with the potentially accelerated deterioration of structures. Many popular building materials react chemically with acidic air pollutants, generating new chemical species

231 that can contribute directly to the decay process even if they are rapidly and efficiently washed off by precipita- tion. Furthermore, in some cases the chemical product causes a visual degradation that cannot easily be rectified, such as the blackening of metal work exposed to hydrogen sulfide. The presence of water at the surface is known to be a key factor in promoting the fracturing and erosion of stone. Water penetrates pores and cracks and causes mechanical stresses both by freezing and by hydration and subsequent crystallization of salts (see Fassina 1978, Gauri 1978, Winkler and Wilhelm 1970). The earlier discussion of surface effects that influence dry deposition indicated that surface scratches and fractures will cause accelerated dry-deposition rates in localized areas. Moreover, phoretic effects are likely to be more important than in the case of foliage (because dry surfaces exhibit wider temperature extremes than moist vegetation). Stefan flow associated with dewfall is also probably more important than for vegetation. Hicks (1981) has summarized a number of relevant points as follows: 1. In daytime, particle fluxes will be greatest to the coolest parts of exposed surfaces. 2. Both particle and gas fluxes will be increased when condensation is taking place at the surface, and decreased when evaporation occurs. 3. If the surface is wet, impinging particles will have a better chance of adhering, and soluble trace gases will be more readily "captured. n 4. The chemical nature of the surface is important; if reaction rates with deposited pollutants are rapid, then surfaces can act as nearly perfect sinks. 5. Biological factors can influence uptake rates, by modifying the ability of the surface to capture and bind pollutants. 6. The texture of the surface is important. Rough surfaces will provide better deposition substrates than smoother surfaces and will permit easier transport of pollutants across the near-surface quasi-laminar layer. 2.1.9. Fog and Dewfa~_ The processes that cause aerosol particles to nucleate, coalesce, and grow into cloud droplets are precisely the same as those that assist in the generation of fog.

232 Whenever surface air supersaturates, fog droplets form on whatever hydroscopic nuclei are available. These small droplets slowly settle onto exposed surfaces or are deposited by interception and impaction. The character- istics of the liquid that is deposited are much the same as those of cloud liquid water (see Section 3). The conditions under which low-altitude surface fogs form are the cases of strong stratification in which vertical turbulent transport is minimized. The frequency of occurrence of fogs varies widely with location and with time of year. The depth is also highly variable. However, it must be assumed that fogs constitute a mechanism whereby the lower atmosphere (say the bottom hundred meters or so) can be cleansed of particulate and some gaseous pollutants. At higher elevations, fog droplets are precisely the same as the cloud droplets that in other circumstances would grow and finally precipitate in substantially diluted form. The importance of cloud droplet inter- ception has been demonstrated recently by Lovett et al. (1982) at an altitude of 1200 m in New Hampshire. Most of the net deposition of acidic species is by cloud droplet interception. The presence of liquid water on exposed surfaces will obviously help promote the deposition of soluble gases and wettable particles. This surface water arises through the action of three separate mechanisms. Some plants expel fluid from foliage, usually at the tips of leaves, by a process known as guttation. Moisture can evaporate from the ground and recondense on other exposed surfaces, a mechanism known as distillation. However, these mechanisms are frequently confused with dewfall, which is properly the process by which water vapor condenses on surfaces directly from the air aloft. In practice, the origin of the surface moisture is immaterial to pollutants that come in contact with it. However, dewfall and distillation are processes that assist pollutant deposition through Stefan flow, whereas guttation does not. According to Monteith (1963), the maximum rate of dewfall is of the order of 0.07 moth, so that the maximum Stefan flow enhancement of the nocturnal deposition velocity is about 8 cm/in (see Section 2.1.4).

233 2.1.10 Resuspension and Surface Emission Deposited particles can be resuspended into the air and subsequently redeposited. The mechanisms involved are much the same as those that cause saltation of particles from the beds of streams and from eroding croplands. These subjects are of great practical importance in their own right and have been studied at length. Concern about resuspension of radioactive particles near sites of accidents or weapons tests injected a note of some urgency into related studies during the 1950 s and 1960s, as evidenced in the large number of papers on the subject included in the volume "Atmosphere-Surface Exchange of Particulate and Gaseous Pollutants" (Engelmann and Sehmel 1976). The momentum transfer between the atmosphere and the surface is the driving force that causes surface particles to creep, bounce, and eventually saltate. There is a minimum frictional force that will cause particles of any particular size to rise from the surface. Bagnold (1954) identifies u2 as a controlling parameter, so that it is the few occurrences of strongest winds that are the most important. While most thinking seems to center on widespread phenomena like dust storms, Sinclair (1976) points out that dust devils provide a highly efficient light-wind mechanism for resuspending surface particles and carrying them to considerable altitudes. Clearly, very large particles will not be moved frequently, or far. Very small particles are bound to the surface by adhesive forces that have already been discussed and tend to be protected in crevices or between larger particles. Chamberlain (1982) has provided a theoretical basis for linking saltation of sand particles and snowflakes and for relating these phenomena to the generation of salt spray at sea. It is not clear how saltation and related phenomena affect acid deposition. Surface particles that are injected into the air by the action of the wind do not normally move far, nor do they offer much opportunity for interaction with other air pollutants (firstly because they are confined in a fairly shallow layer near the suface and secondly because they have a very short r - At:: is - no" t" i m" ~ an,. Their effects are largely local. Much smaller particles (in the submicrometer size range) are generated by reactions between atmospheric oxidants and organic trace gases emitted by some vegetation, especially conifers (see Arnts et al. 1978).

234 Once again, it is not obvious how these should best be considered in the present context of acid deposition. This is but one of many natural surface sources that provide a conceptual mechanism for injecting particles and trace gases into the lower atmosphere. 2.1.11 The Resistance Analog Discussion of the relative importance of the various factors that contribute to the net flux of a particular atmospheric pollutant and determination of which process might be limiting in specific circumstances is simplified . . . · . ~ ~ ~ _ ~ ~ ~ . by considering a resistance morel analogous co unm s law. Figure C.2-5 illustrates the way in which the concept is usually applied. An atmospheric resistance, ra, is identified with the transfer of material through the air to the vicinity of the final receptor surfaces. This resistance is defined as that associated with the transfer of momentum; it is dependent on the roughness of the surface, the wind speed, and the prevailing atmo- spheric stability. The aerodynamic resistance can be written as ra = Cfn-1 ~ (c/k)/u*, where Cfn is the appropriate friction coefficient (the square root of the familiar drag coefficient) in neutral stability, u* is the friction velocity (a scaling quantity defined as the root mean covariance between vertical and longitudinal wind fluctuations), k is the van Karman constant, and Tc is a stability correction (C.2-2) function that is positive in unstable, negative In stable, and zero in neutral stratifications (see wesely and Hicks 1977). Equation (C.2-2) is obtained by straightforward manipulation of standard micrometeoro- logical relations, as given by Wesely and Hicks, for example. The value of k is usually taken to be about 0.4. Table C.2-3 lists typical values of the friction coefficient for a range of surfaces. The surface boundary resistance, rb, is that which accounts for the difference between momentum transfer (i.e., frictional drag) at the surface and the passage of some particular pollutant through the near-surface quasi-laminar layer. In the agricultural meteorology literature, a quantity B~1 is frequently employed for this purpose (Brutsaert 1975a). The relationship between

235 ~ ra 1 ~ ~ rcf ~ FIGURE C.2-5 A diagrammatic illustration of the resistance model frequently used to help formulate the roles of processes like those given in Figure C.2-1. Here, ra is an aerodynamic resistance controlled by turbulence and strongly affected by atmo- spheric stability, rbf and rbS represent surface boundary-layer resistances that are determined by molecular diffusivity and surface roughness, and ret and rCs are the net residual resistances required to quantify the overall deposition process, to the eventual sink. The second subscripts f and s are intended to indicate pathways to foliage and to soil, respectively. There are many other pathways that might be important; the diagram is not intended to be more than a simple visualization of some of the important factors. these quantities can be clarified by relating both to the micrometeorological concept of a roughness length, zO (the height of apparent origin of the neutral logarithmic wind profile). Then the total atmospheric resistance, R. between the surface in question and the height of measurement, z, can be written as R = (ku*) l [ln(z/zoc) ~lc] = (ku*)-1 [ln(z/zO) +ln(zo/zoc) ~lc] = ra + (kU*)-l- ln(Zo/Zoc) ~ where Zinc is a roughness length scale appropriate for the transfer of the pollutant. The residual boundary-layer resistance, rb = R - ra, is then rb = (ku* 1 ln(Zo/Zoc)' which alternatively is written as rb = (u*B) 1. (C.2-3) (C.2 -4) (C. 2-5)

236 B is, therefore, a measure of the nondimensionalized limiting deposition velocity for concentrations measured sufficiently close to a receptor surface that the resistance to momentum transfer can be disregarded. It should be noted that some workers refer to rb as the aerodynamic resistance and use the symbol ra for it (e.g., O'Dell et al. 1977). Shepherd (1974) recommends using a constant value kB-1 = ln(zO/z~c) = 2.0 for transfer to vegetation, on the basis of results obtained over rough, vegetated surfaces. However, the role of the Schmidt number in accounting for diffusion near a surface needs to be taken into account. Wesely and Hicks (1977) advocate the use of a Schmidt number relationship like that of Equation (C.2-1), so that the surface boundary-layer resistance would then be written as rb ~ 5 Sag /3u*. (C.2-6) Equation (C.2-6) implies a value of 0.2 for A in the boundary-layer relationship given by Equation (C.2-1), as was mentioned earlier. The final resistances in the conceptual chain of processes represented diagrammatically by Figure C.2-5 TABLE C.2-3 Estimates of Roughness Characteristics Typical of Natural Surfacesa Approx. Canopy Roughness Neutral Fr lotion Surface Height (m) Length (cm) Coefficient, Cfn Smooth ice 0 0.003 0.042 Ocean 0 0 0 0 5 0 0 4 5 Sandy desert 0 0.03 0.055 Ti lied soil 0 0 . 10 0.0 66 Thin grass 0.1 0.70 0.09S Tall thin grass 0.5 5. 0.16 Tall thick grass 0.5 10. 0.21 Shrubs 1.5 20. 0.25 Corn 2.3 30. 0.29 Forest 10. 50. 0.23 Forest 20. 100. 0.24 aValues of the friction coefficient Cfn(~ u*ju) are evaluated for neutral conditions, at a height 50 cm above the surface or top of the canopy.

237 are those that permit material to be transferred to the surface itself. For many pollutants, it is necessary only to consider the canopy foliage resistance rcf' but for some it is also necessary to consider uptake at the ground by invoking a resistance ~o Fr~n~f~r ~^ ami ~ t^' F~c~ ~ ~1 ~_` _ -of l, Tics In concept, it is also appropriate to differentiate between boundary-layer resistances rbf and rbS, for transfer to foliage and soil, respectively, as is shown in the diagram. Many other r-ni.~=n~= -=n be identified and might often neea co He considered, but further complication of Figure C.2-5 is unnecessary. Its main purpose is illustrative. Transfer of many trace gases to foliage occurs by way of stomata! uptake, which, because of stomata! physiology, imposes a strong diurnal cycle on the overall deposition behavior. Following initial work by Spedding (1969), studies of foliar uptake of sulfur dioxide have repeatedly confirmed the controlling role of stomata! resistance. Chamberlain (1980) summarizes results of experiments by Belot (1975) and Garland and Branson (1977), who compared surface conductances of sulfur dioxide with those for water vapor over a broad range of stomata! openings (which largely govern stomata! resistance). The conclusion that stomata! resistance is the controlling factor when stomates are open appears to be well founded. However, once again, it is necessary to apply corrections to account for the diffusivity of the trace gas in question; the higher the molecular diffusivity of the gas, the lower the stomata! resistance. Fowler and Unsworth (1979) point out that SO2 deposition to wheat continues, even when stomates are closed, at a rate that suggests significant deposition at the leaf cuticle. Thus, it is not always sufficient to compute the canopy-foliage resistance rat on the assumption that SO2 uptake is via stomates alone (although this may indeed be a sufficient approximation in most circumstances). Instead, it is more realistic to estimate rat from its component parts via a ~an. . rat ~ (rst + rout 1) 1/(LAI) (following Chamberlain 1980), where rst is the stomata! resistance, and rout is cuticular resistance. LAI is the leaf area index (the total area of foliage per unit horizontal surface area). Note that in most literature the LAI is assumed to be the single-sided leaf area index. However, sometimes both sides of the leaves are counted. (C.2-7)

238 ~EPIDERMIS J (am/-~ SPONGY, MESOPHYLLIC IC ,' >~ CELLS ~ / ':r-~` in. / ~t ~ r ~] 1 ~ rm rout / ~ rb PALISADE CE LLS 'I ~ GUARD CELLS FIGURE C.2-6 An illustration of the roles of different resistances associated with trace gases uptake by a leaf. Material is transferred along several possible pathways, of which two are shown. These involve cuticular uptake via a resistance, rem, and trans- fer through stomata! pores (via racy into substomatal cavities, with subsequent transfer to mesophyllic tissue (via rm). The way in which the various resistances are combined to provide the best visualization of the overall transfer process is not clear. The resistance analogy permits a closer look at the mechanisms that transfer gaseous material to leaves. Figure C.2-6 illustrates the pathways involved: via stomata! openings into the interior of the leaf (involving stomata! and mesophyll resistances, rst and rm) or _ _ through the epidermis (involving a cuticular resistance, rout) ~ The resistance money Is somewnac mu By ~1lC Itl"~- ~_. . ~a lo,, ~ ~-rim- in which it structures the chain of relevant processes, each being represented by a resistance to transfer that - ~'r`;~= ~ r~r^~-rih-~1 location in a conceotua1 network. v--"H ~ ~ ~ ~ mu_ _ ~ The structure of this network is sometimes not clears and furthermore, there are important processes that do not conveniently fit into the resistance model. Mean drift velocities (e.g., gravitational settling of particles) are not easily accommodated in the simple resistance

239 picture, and it is doubtful whether some of the bio- logical factors are relevant to the question of particle transfer. Studies of leaves show that stomates are typically slits of the order of 2-20 Am long. For stomata! uptake of particles to be a controlling factor of deposition, we would need to hypothesize spectacularly good aim by the particles. 2.2 METHODS FOR MEASURING DRY DEPOSITION 2.2.1 Direct Measurement There is little question that the deposition of large particles is accurately measured by collection devices exposed carefully above a surface of interest. Deposit gauges and dust buckets have been important weapons in the geochemical armory for a long time. They are intended to measure the rate of deposition of particles that are sufficiently large that deposition is controlled by gravity. In studies of radioactive fallout conducted in the 1950s and 1960s, these same devices were used. In the case of debris from weapons tests, the major local fallout was of so-called hot particles, originating with the fragmentation of the casing of the weapon and its supporting structures, and the suspension of soil in the vicinity of the explosion. These large particles fall over an area of rather limited extent downwind of the explosion. This area of greatest fallout was the major focus of the work on fallout dry deposition. It was largely in this context that dustfall buckets were used to obtain an estimate of how much radioactive deposition occurred. It was recognized that collection vessels failed to reproduce the microscale roughness features of natural surfaces. However, this was not seen as a major problem, since the emphasis was on evaluating the maximum rate of deposition that was likely to occur, so that upper limits could be placed on the extent of possible hazards. Nevertheless, efforts were made to "calibrate" collection vessels in terms of fluxes to specific types of vegetation, soils, etc. (see Hardy and Harley 1958). Much further downwind, most of the deposition was shown to be associated with precipitation, since the effective source of the radioactive fallout being deposited was typically in the upper troposphere or the lower stratosphere. The acknowledged inadequacies of collection buckets for dry deposition were then of only

240 little concern, since dry fallout composed a small fraction of the total surface flux. In the context of present concerns about acid depo- sition, we must worry not only about large, gravita- tionally settling particles but also about small "accumulation-size-range" particles that are formed in the air from gaseous precursors and about trace gases themselves. All of these materials contribute to the net flux of acidic and acidifying substances by dry processes. _ . ~ it IS known that collection vessels do indeed provide a measure of the flux of large particles. However, accumulation-size-range particles, typically of less than 1-um diameter, do not deposit by gravitational settling at a significant rate. These small particles are transported by turbulence through the lower atmosphere and are deposited by impaction and interception on surface roughness elements, with the assistance of a wide range of surface-related effects (e.g., electrophoresis, Stefan flow) many of which will be influenced by the detailed structure of the surface involved. Early work on the deposition of radioactive fallout made use of collection vessels and surrogate surface techniques that were frequently "calibrated" in terms of fluxes to specific types of vegetation, soils, etc. Studies of this kind were relatively easy, especially in the case of radioactive pollutants, since very small quantities of many important species could be measured accurately by straightforward techniques. Most of the radioactive materials that were of interest do not exist in nature, and so experimental studies benefitted from a zero background against which to compare observed data. Moreover, major emphasis was on the dose of radioactivity to specific receptors, a quantity that is strongly influenced by contributions of large, "hot" particles in situations of practical interest. Such circumstances included deposition of bomb debris, fission products, and soil particles from the radioactive cloud downwind of nuclear explosions. In such cases, highest doses were incurred near the source and were due co these larger particles. The applicability of collection vessels and surrogate surfaces in studies of the dry deposition of acidic pollutants is in dispute. Principal among the conceptual difficulties concerning their use is their inability to reproduce the detailed physical, chemical, and biological characteristics of natural surfaces, which are known to control (or at least strongly influence) pollutant uptake

241 in most instances. Furthermore, the continued exposure of already-deposited materials to airborne trace gases and aerosol particles undoubtedly causes some changes to occur, but of unpredictable magnitude and unknown significance. A recent intercomparison between different kinds of surrogate surfaces and collection vessels has indicated that fluxes derived from exposing dry buckets are more than those obtained using small dishes, which in turn exceed values obtained using rimless flat plates (Dolske and Gatz 1982). This provides a tantalizing tidbit of evidence for an ordering of performance characteristics according to the total exposed surface area per unit horizontal projection. In this context,- the similarity with arguments concerning leaf area index seems especially attractive. Micrometeorological data obtained during the same experiment fall between the extremes represented by the buckets and the flat plates. Dasch (1982) reports on a comparison between many different configurations of flat-plate collection surfaces, pans, and buckets. The results indicate that glass surfaces provide the greatest flux estimates for almost-all chemical species considered and Teflon the lowest. Bucket data generally fall midway in the range. Tracer techniques that were developed in the radioecology era for investigating fluxes to natural surfaces offer some promise. A 6-emitting isotope of sulfur, 35S, lends itself to use in studies of SO2 uptake by crops since measurements of low rates of sulfur accumulation are then possible. Garland (1977), Garland and Branson (1977), Garland et al. (1973), and Owers and Powell (1974) report the results of a number of studies of 35SO2 uptake by various vegetated surfaces ranging from pasture to pine plantation and bv nonveaetated surfaces such as water. In concept, it is feasible to extend studies of this kind to the deposition of sulfurous particles, but as yet no such experiment has been reported. However, analogous studies of particle deposition using nonradioactive aerosol tracers have been carried out. In wind-tunnel experiments, Wedding et al. (1975) employed uranine dye particles in conjunction with lead chloride particles to study the influence of leaf microscale roughness on particle capture characteristics; uranine particles are relatively easily measured by fluorimetry, whereas measurements of lead deposition require far more paint staking chemical analysis of the deposition surface. The particle sizes used by Wedding et al. were in the range of 3- to 7-pm diameter.

242 Considerably larger particles have been used in many studies. In detailed wind-tunnel studies, Chamberlain (1967) used lycopodium spores (~30-pm aerodynamic diameter). Workers at Brookhaven National Laboratory extended these wind-tunnel techniques to real-world circumstances by conducting a series of experiments employing pollen grains in the same general size range (Raynor et al. 1970, 1971, 1972, 1974). In general, these methods of tracer measurement have not been applied to natural circumstances for the particle sizes of major interest in the present context of acid deposition. An important exception concerns studies of deposition on snow surfaces. The retention of deposited material at the top of or within a snowpack has been studied in some detail and continues to be an intriguing area of research. Particulate materials such as sulfate were considered by Dovland and Eliassen (1976), who studied the accumulation upon snow surfaces during periods of no precipitation and found average deposition velocities in the range 0.1 to 0.7 cm/s depending on the assumption made regarding the contri- bution by gaseous so2 deposition. Similar work by Barrie and Walmsley tl978) yielded average sulfur dioxide deposition velocities to snow in the range 0.3 to 0.4 c~/s, with standard error equivalent to about a factor or 2. Dillon et al. (1982) and Eaton et al. (1978) present examples of the use of calibrated watersheds to estimate atmospheric deposition. Dry-deposition fluxes are estimated as a residual between measured fluxes out of a conceptually closed system and measured wet deposition into it. Considerable effort is required to document annual chemical mass balances for specific watersheds. Once the effort is made, it appears possible to draw fairly well-founded conclusions regarding dry deposition, although obviously such estimates might result as the difference between fairly large numbers. According to Eaton et al., the annual dry-deposition flux estimate obtained at the Hubbard Brook Experimental Forest in New Hampshire is accurate to about +35 percent (one standard error). The data do not permit apportionment between gaseous and particulate sulfur inputs, but the total sulfur flux corresponded to a deposition velocity of about 0.6 cm/s.

243 2.2.2 Laboratory Studies ~ _ _ _ ~ ~_ ~. . Figure C.2-1 illustrates the overall complexity of the problem of dry deposition. While it is indisputable that no indoor experiment can provide a comprehensive evaluar tion of pollutant deposition that would be applicable to the natural countryside, laboratory studies provide the unique attraction of controllable conditions. It is reassure to study the relative importance of various factors thought to be of importance, as in Figure C.2-1 and especially as in Figure C.2-6, and to formulate these processes in a logical manner. In this general category, we must include the extensive wind-tunnel work referred to earlier, the pipe-flow and flat-plate studies con- ducted in experiments more aligned to problems of chemical engineering, and the chamber experiments favored by ecologists and plant physiologists. Distinction between these kinds of experiment is often difficult. Many exposure chambers and pipe-flow studies have features of wind tunnels. The utility of chamber studies is well illustrated by the series of results reported by Hill (1971). By comparing the rates of deposition of various trace gases to oat and alfalfa canopies exposed in loran ~h~mh~r~ To: ~ ~ _ ___ ~ ._a ~ ~ ~ uea anal solUDlllty was a critical parameter in determining uptake rates of trace gases by vegetation. The ordering of deposition velocities was: hydrogen fluoride > sulfur dioxide > chlorine > nitrogen dioxide > ozone > carbon dioxide > nitric oxide > carbon monoxide. Furthermore, the chamber studies indicated a wind-speed dependence of the kind predicted by turbulent transfer theory and demonstrated a physio- logical effect of chlorine and ozone on stomata! opening: exposure to high concentrations of either quantity caused partial stomata! closure, thus limiting the fluxes of all trace gases that are stomatally controlled. Experiments conducted by Judeikis and Wren (1977, 1978) yielded valuable information on the deposition of hydrogen sulfide, dimethyl sulfide, sulfur dioxide, nitric oxide, and nitrogen dioxide to nonvegetated surfaces (Table C.2-4). The values listed were derived prom Nag aeposz~on rates obtained before surface accumulation limited uptake rates. For comparison, surface resistances derived from Hill's (1971) studies of trace-gas uptake by alfalfa are also listed. On the whole, the ordering of deposition velocities suggested by

244 Hill's work appears to be supported, providing some justification for extending the ordering to CO, H2S, and (CH3)2S in the manner indicated in the table. Residual surface resistance to uptake of soluble gases by solid, dry surfaces appears to be substantially greater than for vegetation, which is as would be expected. The values listed in Table C.2-4 represent resistances to transport very near the surface, to which atmospheric resistances must be added to obtain values representative of natural, outdoor conditions. The reciprocals of the tabulated numbers provide upper limits of the appropriate deposition velocities. Similarly informative data have been obtained about particle deposition on surfaces that can be contained in wind tunnels. Studies of this kind are an obvious extension of pipe-flow investigations by workers such as Friedlander and Johnstone (1957) and Liu and Agarwal (1974), which provide strong support for theories involving particle inertia and Schmidt number scaling. Wind tunnels provide a means to extend chamber and TABLE C.2-4 Resistances to Deposition of Selected Trac Gases, Measured for Solid Surfaces in a Cylindrical Flow Reactor (Judeikis and Stewart, 1976) and for Alfalfa in a Growth Chamber (Hill, 1971); Solid-Surface Data are Derived from Table 2 of Judeikis and Wren (1978); the Alfalfa Values are Obtained from Table 1 of Hill (1971) e Substrate Surface Pollutant Adobe Clay Sandy Loam Alfalfa CO ~ H2S 62 67 - (CH3)2S 3.6 16 _ NO 7.7 5.3 10. CO2 - _ 3 3 O3 ~ - 0.7 NO2 1.3 1.7 0.5 C12 ~ - 0.5 SO2 1.1 1.7 0.4 HF - - 0.3

245 1 1 1 1 1 1 GRAVE L 10 Z Of 0 cn o CL IN ~ 0.01 ~ 0 '- i ·/ // ~ / 1 1 1 1 1 ~ 1 1 - 0.01 .~/ · /° / 0.1 PARTICLE DIAMETER ~ ~ m TO FIGURE C.2-7 Results of wind-tunnel studies of particle deposition to 1 .6-cm- diameter dry gravel (after Sehmel et al. 1 973a). Solid circles were obtained at about 2.4 m/s, open circles at about 16 m/s. pipe-flow investigations to situations more closely approximating natural conditions. Results obtained in studies of particle deposition to dry gravel (Sehmel et al. 1973a) are shown in Figure C.2-7. The wind-speed effect evident in these data is fairly typical and applies also in the case of vegetation (Figure C.2-8). Experiments on deposition to wet gravel were also conducted. These indicated deposition velocities some 30 percent less than the values evident in Figure C.2-7 (for particles in the 0.2- to 1.0-pm

246 1 1 1 1 1 1 1 1 1 ~ 40 1 cat a_ - o J lo > 0 0-1 CJ) o lo' IN C) GRASS \ 0.01 ./^ // ~//// , / / ~ ~ / . / - 1 1 /1 1 1 1 1 0.01 0.1 ~ lO PARTICLE DIAMETER (~m) FIGURE C.2-8 Results of wind-tunnel studies of particle deposition to grass as reported by Chamberlain (1967; dots; natural grass; u* 70 cm/s) and by Sehmel et al. (1973; the curve; artificial grass; u* _ 16 cm/s). size range), as might be expected from considerations of Stefan flow and diffusiophoresis. When surface roughness was increased, deposition velocities also increased. Chamberlain (1967) extended his earlier (1966) wind-tunnel studies of gas transfer to "grass and grasslike surfaces" by considering particle deposition to rough surfaces. Sehmel (1970) conducted similar wind-tunnel experiments, employing monodisperse particles ranging from about 0.5- to 20-pm diameter. Figure C.2-8 combines results from Chamberlain (1967), and Sehmel et al. (1973b). The Chamberlain data refer to live grass, but the Sehmel et al. data were obtained

247 using 0.7-c~high artificial grass. Moreover, the two sets of data were obtained at different wind speeds (Chamberlain, u* ~ 70 cm/s; Sehmel et al., u* ~ 19 cm/s). Further tests conducted by Chamberlain (1967) indicated that deposition velocities to natural grass exceeded those to artificial grass by a factor of about 2 for particles smaller than about 5 Am. This appears to be contrary to the indication of Figure C.2-7, where vd (natural) of Chamberlain is seen to be about half the Vd (artificial) of Sehmel et al. for sizes less than a few micrometers. The difference in u* between their experiments amplifies this discrepancy, rather than resolving it. However, both data sets provide evidence for the deposition velocity particle-size dependency that is predicted by theory and supported by all such laboratory investigations. 2~2.3 Micrometeorological Measurement Methods The factors that control pollutant fluxes are frequently surface properties such as stomata! resistance and soil moisture (for soluble gases), cuticular resistance and the available leaf area (for strongly surface-reactive gases like HF and HN03), and microscale roughness (for submicrometer particles). Any measurement technique that interferes with a controlling property is likely to yield erroneous results, and hence there has been considerable effort expended to develop and apply methods of measure- ment that impose no surface or environmental modification. In concept, if an area is sufficiently homogeneous, flat, and contains no areas of strong sources or sinks, pol- lutant fluxes can be assumed to be constant with height. Therefore questions regarding dry deposition can be addressed by measuring the flux of material through a horizontal layer of air at some more convenient height above the surface. The intent of any such study is to investigate dry-deposition fluxes in carefully documented natural situations in order to identify and quantify controlling properties. The results of these investiga- tions are formulations of surface mechanisms, surface boundary-layer resistances stomata! resistances. it_ The demanding site criteria are required to enable these results to be obtained from the experiments; the surface parameterizations that are derived are far more widely applicable.

248 Several micrometeorological methods are suitable for measuring dry-deposition fluxes in intensive case studies. The flux can be measured directly by eddy correlation, a process that evaluates instantaneous products of the the vertical wind speed, w, and pollutant concentration, C, in order to derive the time-average vertical flux Fc as FC = pW'C', (C.2-8) where p is air density and the primes denote deviations from mean values. The over-bar indicates a time average. This is an extremely demanding task and constitutes a specialized field of micrometeorology in its own right. Details of experimental procedures are given by, for example, Dyer and Maher (1965), Kaimal (1975), and Kanemasu et al. (1979). Figure C.2-9 shows some examples of sensor output signals that are fundamental to the eddy-correlation technique. Fast-response sensors of any pollutant concentration can be used; the trace shown for CO2 in the diagram is an interesting example of considerable agricultural relevance. As a basic requirement, sensors suitable for eddy-correlation applications should have response times shorter than 1 sea for operation at convenient heights on towers. For application aboard aircraft (Bean et al. 1972, Lenschow et al. 1980), considerably faster response is required. Eddy-correlation methods have been used in field experiments addressing the fluxes of ozone (Eastman and Stedman 1977), sulfur (Galbally et al. 1979, Hicks and Weselv 1980) nitrogen oxides (We sely et al. 1982b), carbon dioxide (Desjardins and Lemon 1974, Jones and Smith 1977), and small particles (Wesely et al. 1977). Rates of transfer through the lower atmosphere are governed by intensities of turbulence generated by both mechanical mixing and convection. In this context, there are three atmospheric quantities that cannot be separated: the vertical flux of a material, the local gradient of its concentration (aC/az), and its corresponding eddy diffusivity (K). Knowledge of any two of these quantities will permit the third to be evaluated. Often, when sensors suitable for direct measurement of pollutant fluxes are not available, assumptions regarding the eddy diffusivity are made to

249 330 CO2 (plum) 319 308 MA ~ -0. 8 1.8 u (m so) 0.6 0.2 27.0 T °C 25.5 24.0 ~ v_ ~ ut >'~ W¢ W1' 11 I- ~ · . 12 hr:35 min 12:36 12:37 Tl ME FIGURE C.2-9 An example of atmospheric turbulence near the surface. These traces of CO2 concentration, vertical velocity (w), wind speed (uJ, and temperature (T) were obtained over a corn canopy by workers at Cornell University. provide a method of vertical concentration gradients: for estimating fluxes from measurements Fc = PK(aC/aZ) (C. 2-9) Droppo (1980) and Hicks and Wesely (1978) have summarized a number of critical considerations. In particular, with a typical value of u* = 40 cm/s and neutral stability, the concentration difference between adjacent levels differing in height by a factor of 2 is about 9 percent, for a 1 cm/s deposition velocity. In unstable (daytime) · . ~ conditions, smaller gradients would be expected tor the same vd; in stable conditions, they would be greater. The demands for high resolution by the concentration measurement technique are obvious. Nevertheless, a substantial quantity of excellent information has been obtained, especially concerning fluxes of SO2 (Fowler 1978, Garland 1977, Whelpdale and Shaw 1974). It should be emphasized that the stringent site uniformity requirements mentioned above for use of

250 eddy-correlation approaches are also relevant for gradient studies. The detection of a statistically significant difference between concentrations at two heights is not necessarily evidence of a vertical flux and can only be interpreted as such after the appropriate siting criteria have been satisfied. Gradients of particle concentration present special problems since it is often not possible to derive internally consistent results from alternative measure- ments. Droppo (1980) concludes that "The particulate source and sink processes over natural surfaces cannot be considered as a simple unidirectional single-rate flux.. Thus, the proper interpretation of gradient data in terms of fluxes might not be possible for airborne particles, even in the best of siting circumstances, because of the role of the surface in emitting and resuspending par- ticles. In this case, eddy-correlation methods will still provide an accurate determination of the flux through a particular level, but this flux will be made up of a downward flux of airborne material and an upward flux of similar material of surface origin. Disentang- ling the two is likely to present a considerable problem. None of the various micrometeorological methods has yet been developed to the extent necessary for routine application. Rather, they are research methods that can be used in specific circumstances, requiring considerable experimental care and the use of sensitive equipment and fairly complicated data analysis. They are more suitable for investigating the processes that control dry deposi- tion than for monitoring the flux itself. Nevertheless, some new techniques that might be appropriate for dry- deposition monitoring are currently under development. A Modified Bowen ratio" method is being developed in the hope that it might permit an accurate determination of vertical fluxes without the need for rapid response or great resolution (Hicks et al. 1981). Highrfrequency variance methods are also being advocated but have yet to be fully investigated; for these, sensors having rapid response are required. An eddy-accumulation method that bypasses the need for rapid response of the pollutant sensor is of long-standing interest (e.g., Desjardins 1977) but has yet to be applied to the pollutant flux problem with significant success.

251 2 .3 FIELD INVESTIGATIONS OF DRY DEPOSITION 2.3.1 Gaseous Pollutants Table C.2-5 summarizes a number of recent field experiments on trace gas deposition. The listing is drawn from a variety of sources (especially Chamberlain 1980, Garland 1979, Sehmel 1979, 1980a); it is not meant to be exhaustive but is intended to demonstrate that much of the available data on surface fluxes of trace gases refers to daytime conditions, when "canopy" resistances are usually the controlling factors. Extrapolation of these values to nighttime conditions is dangerous on two grounds; first, because of the large changes that might accompany stomata! closure and, second, because of the much greater influence of aerodynamic resistance in nighttime, stable conditions. Figure C.2-10 illustrates the large diurnal cycle that is typical of the dry-deposition rates of most pollutants. These observations were made over a pine plantation in North Carolina, using eddy correlation to measure each quantity (Hicks and Wesely 1980). The diurnal cycle of sensible heat flux meshes well with expectations based on the available heat energy (i.e., on net radiation), and the friction velocities determined by direct measurement conform to expectations based on the known roughness characteristics of the site. The eddy fluxes of total sulfur demonstrate a diurnal cycle that appears to be as strong as for the meteorological properties, a result that is not surprising when it is remembered that many of the causative factors are common (e.g., vertical tur- bulent exchange). The quality of the data appears to be quite good--sensible heat fluxes and friction velocities are difficult to measure, and the ability to measure each with the run-to-run smoothness evident in the diagram instills considerable confidence in the sulfur data, measured with the same apparatus. Nevertheless, it would be unwise to place too much confidence in the fine details of the sulfur-flux data. While the strong (downward) fluxes of sulfur evident during the midday periods are probably accurately represented, it is possible that the indicated nightime fluxes are contaminated by a water- vapor interference or some other factor that is especially significant in the calmer, more humid nighttime situation. Thus, some caution must be associated with interpreting the negative (upward) fluxes of sulfur evident on two periods as evidence of emission or resuspension from the

252 U] {n As o o Ad .,, A o a U] a) EN o Q a) Al x a) c) At: Up C) At: U] o U1 In  s a, a: o : · - to . fir ~ O · - U) to · ·~1 lo, 21 H O U) Q la ~ S o 3 ~ Cal o ~ o  en US _ ~ HI h . - GS Cal o V) l _ 21 1 a) '. - on U] C) ~ Cal O hi . 21 - 21 a ~S~ :5 ~ U] U] a .,1 , O cn O U] O O _ ~ a' _ _ ~ 3 o ~5 54  U] ~ .,1 U] t) · ~ ~ o _ 21 21 21 ~n ~O rJ1 U] a' o 3 U] - ~5 s" S ~4 a) 3 s o U] 3 o U. ~ 3 ~5 .^ ~n ? 21 0 U] {R [Q > - tQ ~ O O 21 21 {) ~n a) ~n O tQ .,. ' a, ~ 3 o ~Q ~Q o a) ~O tQ a ,' ~ U] `4 ~ a,1 o ~ U] o U] - - a) -  . ~ ~ ~ _I a) _ s u~ ~n · ~ o ~ tl t U1 .,. ~S ~ 1 o cn 0D - 3 a) o U] 1

253 Q' e~ .,' ~Hi. en ~ ~ t,, en 'A ._, ~ ~· - ~·rl ~21 (Q· en USA toto e ~ O ~ ~ ~ ~ ~ ~ ~` ~ U 1 ~ ~ `` ~ U] eg d. · . . . . . ~ ~ . -,1 · O O ~O O O O · ~O ~ O 21 21 V 21 21 21 21 21 21 21 ~21 21 b to a' o v ~ 3 s ~ ~ 3 b 3 03 ~a u' O c: ~O ~ ~3 O O O = .O i-. 0- ~ v C ~ ~o ~ c ~ c O O o ~O ~O V ~V O ~U O O :^ ~ U. U] O U~ U] ~ #¢ ~V - _ Q ~o \9 a _ _ ~ ~ _1 0 N o _I {,q 0) ClO - - eq ~ ao ~ [Q -I a~ ~,\' . ~_ O V ~= ~ ~ ~V ~V _ ~ _ C ~-~ C) ~ > _ ~ _ [Q a) ~ O ~a, v m v ~ m ~c' 3 0 Z O

254 - 1 An ~ 100 Can 111 me on Coy Lo ,, 400 ~ 200 50 o Lo I LIJ cr Lo en -- 0.05 ,= Con cn _ A 1: . , 1 , 1 "it, i, 17 JULY 18 JULY 20 JULY 1 1 1 1 1~ ~- 1 ,~ 0 12 0 12 LOCAL STANDARD TIME (h) 0 12 0 FIGURE C.2-10 Records of sulfur flux, sensible heat flux, and friction velocity through 3 days of an intensive study of dry deposition to a pine plantation (Hicks and Wesely 1980~. The darker portions of the sulfur data indicate periods when gaseous sulfur could not be detected. At all times, the data refer to total sulfur, usually made up of gaseous and particulate contributions.

255 canopy. However, attempts to explain the phenomenon in terms of some interference have so far failed. Similarly, large diurnal cycles of SO2 deposition are reported by Fowler (1978), who introduces the further complexity of enhanced SC2 deposition to wheat covered with dewfall. Using the notation of Figures C.2-5 and C.2-6, Fowler finds typical daytime values to be ra = 0.25 s/cm, rb = 0.25 s/cm, rst = 1.0 s/cm, rout = 2.5 s/cm. For deposition to dry soil, Fowler suggests the use of rCs = 10.0 s/cm, and ran = 0 when the soil is wet. It might be noted, in passing, that the aerodynamic resistance ra influences the deposition of all nonsedimenting pollutants, and thus it is not possible for any trace gas to have a deposition velocity greater than 1/ra, i.e., about 4 cm/s in the daytime conditions of Fowler's experiment. Because of stability effects, the maximum possible deposition velocity at night would be considerably lower. Many of the exceedingly large deposition velocities reported in the open literature appear to exceed the limits imposed by our knowledge of the aerodynamic resistance. Thus, several of the results included in the exhaustive tabulation presented by Sehmel (1980b) should be viewed more as indications of experi- mental error than as determinations of a Dhvsica quantity. , ~_ Figure C.2-11 addresses the question of the time variation of the aerodynamic resistance, rat Values plotted are the maximum deposition velocity permitted by the prevailing aerodynamic resistance, evaluated directly from eddy fluxes of heat and momentum determined during the pine plantation experiment of Figure C.2-10. The reciprocals of the plotted velocities provide evaluations of rat It is seen that in daytime, deposition velocities could be as much as 20 cm/s if the surface resistance is zero, implying ra ~ 0~05 s/cm during midday periods. At night, however, ra can increase to 10 s/cm on infrequent occasions, but often exceeds 0.5 s/cm. The recommendations of Fowler are probably representative of the long-term average. The importance of diurnal cycles in pollutant depo- sition and the close relationship with other meteoro- logical quantities is further illustrated by Figure

256 100.0 'u, 10.0 -I ~ 1 1 1 1 1 1 1 1 ' 1 1 ~1.0 o lo 0.1 0.4 14 15 .. - 1 t ~6 ~ f ~7 ~8 Z o ~loom 1 ' 1 1 1 1 1 o ~`,_~\ 18 ~ 9 20 21 1 1 1 , 1 ,1 , 1 , 1 , 1 , 1 1 1 00 00 00 00 HR (lest) FIGURE C.2-11 Values of the maximum possible deposition velocity of trace gases,  determined as the inverse of the aerodynamic resistance ra for the pine plantation experiment of Hicks and Wesely ( 1980~. C.2-12, which provides examples of the trend from nighttime, through dawn, and into the afternoon of the residual canopy resistance rc for ozone and water vapor determined using eddy correlation (Wesely et al. 1978) These data were obtained over corn (Zea mays) in July 1976. The upper sequence shows good matching between rc for ozone and water vapor, with the former exceeding the latter by a small amount, on the average. As the day progresses, to increases gradually, presumably as a consequence of increasing water stress and eventual stomata! closure. The lower data sequence has two features of considerable interest. First, the gradual initial decrease of rc for O3 corresponded to a period of evaporation of dewfall (note the relatively low value of rc for H2O during the same period), sug- gesting that the presence of liquid water on the leaf surfaces might inhibit ozone deposition (much as might be expected on the basis of ozone insolubility in H2O). This would not be the case for SO2 deposition (Fowler 1978). Second, the peak in both evaluations of rc at

257 about 10 a.m. is associated with the passage of clouds, which caused a rapid and strong decrease in incoming radiation and lasted for about an hour. The peak is seen as further evidence for stomata! control, since some stomata! closure would be expected with reduced insolation. The above discussion of both SO2 and O3 deposition confirms the generalization made by Chamberlain (1980) that the deposition of such quantities might be modeled after the case of water-vapor transfer with considerable confidence. Recently, Wesely et al. (1982b) have reported a field 2 - _' 1 1 1 1 1 1 1 1 1 1 1 1 4 I , to_ A\ I 1 ·\. ~ ~e \~/ ~ ·-` · b' o~/ x ~\ \~ _ ~ ~ ,, ~ . Am-- ~2~ / ~ . 4 1 o ·\ \ O3-~y ° A /~ 1 1 1 1 1 1 1 1 1 1 1 air/ . 6 8 10 1 2 14 16 ~ 8 HR (C.S.T.) FIGURE C.2-12 Evaluations of the residual "canopy resistance," rc, to the transfer of ozone and water vapor, based on eddy fluxes measured above mature corn in  central Illinois on 29 July 1976 (upper sequence) and 30 July 1976 (lower sequence). Data are from Wesely et al. (1978~.

258 study in which both O3 and NO2 fluxes were measured. Bulk canopy resistances to ozone uptake for a soybean canopy exceeded water-vapor values by about 0.5 cm/s during daytime, with rc for NO2 still greater by a similar amount. 2.3.2 Particulate Pollutants No technique for measuring particle fluxes has been developed to the extent necessary to provide universally accepted data. In every case, research and development are continuing at this time. Use of gradient methods, for example, is limited by the inability to resolve concentration differences of the order of 1 percent. Turbulence methods require rapid response, yet sensitive chemical sensors that are not often available. In both cases, practical application is hindered by the need for a site meeting stringent micrometeorological criteria. Nevertheless, several applications of micrometeorological flux-measuring methods have been published. Table C.2-6 provides a list that illustrates the narrow range of available information. ~ ~ ~ The evidence points to a differ- ence between the deposition characteristics of small particles and sulfate; the latter seems to be transferred with deposition velocities somewhat greater than the value of 0.1 cm/s that has been assumed in most assess- ment studies and greater than the values appropriate for small particles, on the average. At this time, the possibility that sulfate fluxes are promoted by the strong effect of a few large particles cannot be dismissed. As must be expected, taller canopies are associated with higher values of Vd, on the average. Figure C.2-13 shows how small particle fluxes varied with time of day over a pine plantation in North Carolina during 1977 (Wesely and Hicks 1979). These eddy-correlation results display a run-to-run smoothness that engenders considerable confidence; moreover, they are supported by the finding that simultaneous eddy fluxes of momentum and heat closely satisfied the usual surface roughness and energy balance constraints. There is little doubt that the surface under scrutiny (or at least the air below the sensor) did indeed represent a source rather than a sink for substantial periods (Arnts et al. 1978). A basic ques- tion then arises about the significance of the measured deposition rates, since these probably represent a net

3 U) o . - . - U] o a) S ~Q 4J a: C a JJ Ul o S A: ~5 C N  -a' U] a, ~_ U] .,' CQ ED 03 a) m ~ A: - E~ I on A: U. 259 a) .,, :3 [a ~ U) ~ ~ ~JJ O ., V . - - - ~ U) ~ ~ ~:' o ~ ~· U) _I ~O ~ Al h t ~a A ~1 ~ ~ U] - ~0 · · a,' ~dP L' ~4J ~ 0 L4 0 ' - ~ ~ ~ ~ ~ D ~ ~ ~ ~O ~ O lo I ~ ~ ~ ~ A JJ ~a ~ aC 4~ ~ ~ ~ 1 ~ ^ ~ ~ S O ~ ~ ~ U] ~1 - - , ~ ~ ~ C n ` - 0 ~o 0 cm c) ~0 ~ == I_~ ~ ° ~ ~ C ~ O ~ ·- ~ 0 l., ( U3. - ~ te s ~ ~ U) C ~ ' ~° ~: ~ ~ ~ ~ £ ~ ~ ~ 3 ~ e u~ ~ ~ ~ . ~ ~ ~ ~ 0 28 E · - · b? ·- C ~ ~ ~ ~1 L4 ~ C) ) ~O u~ ~ ~ a) ~q ~U) 0 ~ `: ~, - - .- n5 X 14 ~ ~ ~, C a~- ·- ~ ~ ·- ~ O c) ~a ~ ~ ~ ~a, ~ c: ~ ~ ~ ~ · :' ~ 3 4 ~. - ~ ~ ~ ~ ~ · - -I O ~k4  a, ~ ~v U] ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ O V · a) Ll O Z 3 U) o U] O Le .,~ a,, - ~a U) U2 C ~ O J, · - Q. ~ o O 1 ~5 U, O . o eq Ul ~: a, o · - · ~5 1 ~1 . o c.) c · - o -,-, ·-i s~ ~ QJ ~ o ~ u . o ~ 1 u' o ~ . o u) ~ c ~ o 4J · - L4 ~  ~ Ll o . o ~ 1 o ~ o m u, ~5 o s s . 3 aJ U) U] a.) a,, - U) .,~ · - JJ ~, Q. o ~_ ~ a, CO ~a, n _ __ ~_ [_ _ a u ~· u ~ Y ~. y _ ·_ ~I ~·,1 3: ~:= c ,0 _ c~ ~_ c) ~: ~R5 .- a, .- ~0 ~ ._ ~._, 0 a u, a 33 ~0 3 u' 3 . a u' :~ o ~ ·,' r- -' · ~ 0 · - 4 a a) ° · - a) .,' ·- Q  a C a) U] a, s V] U] 8 JJ o o x tn Ul

260 -a ~ a)4 ~a) E~ ~E- ~ O a, E ~ O ~ E n5.c .- ~ ~ ~ ~ ~ 4 ~ a, ~ O :' L4 ,Q to ~ 1 S O ~ -l up ·_l ~ U] ~ In ~ ' ~I:: ~ 3 3 03 q5~ ~ O ~ ~ ' 4J ~U1 ^ ~ 0 ~A up ~ al U) u, ~ ~ ~ ~ ~ E u' ·~ 3 ~ · ~ ' - o en U U] · ~ - - ~ `o ~ X ~ o ~ ~+1 ~ ~ ~ ~ ~ O 3 ~ ~ ~ ~ · ~ ~ E · ~ e . ~ ~ At ~ ~- ~0` 0= C) ~ As O s ~ O ~ ~ ~ ' ~ .- ~t1 ~ ~ ~ ~ ~ U) ~ ~ ~ ~ Ma ~° ~ ~ ~ i., U) O ~ ~s Hi ,~ ~ ~ 0 to o ~ E u~ o ~ ~ ~o 00 V eq c ~0 0, ,,o. .,, O ,, tt5 e~ ~ ~ ~, s 4 ~0 0 ~JJ (v . - 0, ~ ~ ~ ~ . D 3 21 ~, ~V o ~ 3 ~ ~o ~ ~ O ~ ° ~:> S JJ ~: a, N .,1 U] Ul ~5 ' - O ~ ~O ~ 3 ~a) .a ~ ut3 ~ ~ ~ ~3 O ~ ~ O · ~D ~ O '' - a, ~ a, C u] ~ ~u] 3 C) ~ ~ V E ta £ 0 ~5 ~ t~ a, a) .- O ~ .- to ' - ~ O ~ ~ ~ ' - u~ ~ 0 ~ ~s~ a~ e ~ ~ Ll ~- ,., a,= :>, >, 54 ~ 4 ~{0 3 ~ c: N ~ '-I ~ ~ ~ ~ . - ~ O ~ ~ ~ ~ O ~ ~ ~ ~ ~  4~ ~ ~c: ~ E ul ~ ~ 0 h S ·) O 4) O s ~ a ~ ~ s O ~o u~ ~ ~a~ ~ E u~ 3 s. JJ X ~ ~ ~ U] ~ · ~ JJ ~ ~4~ ~ 4~ :^ 4J >1 JJ 3 3 3 ~ ~ ~ =~0 ~ O ~ ' == Q ~ n tl~ J~ ,Q-- O U) ~ ~ V I ~ t) ~ O ~ ~ ~ O G5 - - 3 V 4) ~ ~ ~ ~ '-/ · 3 3 ·- - {0 ~V (U 4~ ( X ~ 0 4~ 0 1 ~ O ~o ~ o ~ ~ ~ J~ ~ ~ ~ ~ 3 ~ ~ ~ O ~ ~ I ~. - ~ I ~ ~ ~ ~ ~ ~ O O J~ S 3 _I ~G5 · O h u-) ~ 3 11~ kl ~ ~ O ~4 O (5 J ~ (_) O ~U) ~4 0 ·~ O ~U) ~3 C) 3 0 O· ~: ~O ~O 0O h r- al a, ~a _ ~cn ~O ~,~ ~ ~ O - ~3 ~t, - ~_ C) , ~,{ ~ ~{~ a) 0 CX) 0, 4 ~c a, c) ~ e <, ~ ~ ~u 3 S :, ~ ~:> ~ E3 u~ C) ~ 3 ~u ~3 m

261 1 ~ O ~ O ~ JJ ~ O C S ~ ~ ~ ~ ~1 a, ~0 ~o ~ ~ ~ ~ ·3 Ed ~ ~oJ ~ ~ ~ ~ ~ ~ e ~ ~ ~ O ~° -~ Jo .- ~ ~ ~ net s GS Q ~ 4) ~ N A ~ ~ O ~ - ~ ~ ~C V ~ O · -I · - ~ ' - ~ 1 ~ ~ ~ ~ ~ o O C I ~ ~ U ·- S lo, ~ C ~ O .o ~ ra ~ ~ ~ ~ e ~ ~ ~ ~ ~ ~ ~ ~ ~ lo ~ ~ lo, ~ 3 ~ U C ~ C ~ ~ ~ U ~ ~ ~ S ~ ~ ·c ~ C up ~ rot C ~ ~ V . · - ~ ~ · O ~ · ~ c ~ a' 0 ~ a, 0 Ll 3 Ql ~ /D _/ ·- a, ~, ~ ta a) O t:~` x tl ~ `e tl Ul o ~ · - ~ O C o ~ Ll 4, C ~ O ~ ~ ~ ' - O ~ ~ A ~ C · - O O ~' - 4 ~J U] U' ~q tQ >', >, ~ C U]~ ~U) . - O ~ U ~ '. ~ 0) ~.0 O (L)u) QU o ~],~0 o C J~ '-I N~ ~4d '- ~ ~J a, ~, ~ ~- ~Ll ~· C - U) ~ to ~ ~ ~a, - ~=- e ~ c ~- ~s ~ c ·e ~ 0 ~ tq ~ O ~m e~= ~ ~es ~·~-,' 4J `) ~ ~ C ~a ~ ~ 3 ta ~ c ° a, ~ ~O u, ~ ·~' ~/ ~ U U ~ U ~ ~ U ~ N O >1 a, 4J (V a) o >, ~ (V G) ~ ·a ~ 1 ~ C ~ ~ C 1 ~ ~ ~ ~ 4' ~ ~ Ln ~ a) ~ o ~u, ~O ~ O o ~ ~ U ~0 ~` - . ~· D ~o U] o ~, U]  U] o a' r~ a, t_ =, _ _ ,- a _ a U] ~U) ~ Y ~X . _ U ^ ~ o . - ,~ O ~a~ oc ~:= {X) 3 a, a C ~ ~a ~a co ~41} U) ai O U) O - U] C U] a) ~._ ~a 3 5: 3 ~:~ U] E~

262 result of continuing but varying surface emission and a deposition flux that is also varying with time. The relevance of the answers obtained is then unclear. In particular, it is not obvious how to relate such results to the common situation in which we wish to evaluate the atmospheric deposition rate of some particulate pollutant that is not emitted or resuspended from the surface. Figure C.2-13 identifies periods of the 1977 pine plantation study during which no gaseous sulfur was detectable. These occasions were used by Hicks and Wesely (1978) to evaluate residual canopy resistances for particulate sulfur that averaged about 1.5 cm/s (with a standard error margin of about +15 percent) for July 17 and about 1.1 cm/s (+25 percent) for July 18. Results of two tests of sulfate gradient equipment over arid grassland, reported by Droppo (1980), yield values of 0.10 and 0.27 cm/s for Vd, in very light winds (~1 m/s). The residual surface resistances evaluated from his data are 7.7 and 3.3 cm/s, respec- tively. These values are considerably higher than the pine plantation results quoted above but might not be wholly discordant when the nature of the surface present in the gradient studies is taken into account. Results of an extensive series of eddy-correlation measurements of particulate sulfur fluxes to a variety of vegetated surfaces have been summarized by Wesely et al. (1982a). In daytime conditions, deposition velocities to grass range from about 0.2 to 0.5 cm/s. Values for a deciduous forest in winter (few leaves) are not significantly different from zero. In general, somewhat lower values are appropriate at night. In almost all of the cases summarized by Wesely et a'., normalization of surface transfer conductances by u* appears to reduce the residual variance. Hicks et al. (1982) present supporting data from another study of the same series, also over grassland. Considerable controversy remains concerning the value of Vd appropriate for formulating the deposition of sulfate aerosol (and presumably all similar particles). Garland (1978) advocates the continued use of values oF O.1 cm/s or less, since experiments conducted over grass in England failed to detect a significant gradient. However, some of the experiments listed in Table C.2-6 indicate quite high deposition velocities for sulfate particles. A possible explanation in terms of a strong contribution by particles much larger than the usual accumulation size mode has been discussed (Garland 1978), and different deposition velocities (0.025 and 0.56 cm/s)

263 2 o -2 0: In 1 -2 cn o O ~ ~- -2 o -2 - I I T I I I- ~9 1 1 1 1 Y ~ ~ 20 ~ A. A ' ~: - ~ ., ..'. Hi._ . ~ ,,: 15 1 1 1 1 1 1 1 1 1 1 1 1 . 0 4 8 12 46 20 24 H R ( est ) FIGURE C.2-13 Deposition velocity of small particles (~0.1 lam) measured by eddy  correlation above a pine plantation in North Carolina in 1977 (Hicks and Wesely 1978 Wesely and Hicks 1979~. Note the strong diurnal cycle, with frequent extended periods of emission rather than deposition (as indicated by the negative "deposition velocities"~.

264 have been postulated for the submicrometer and larger particles, respectively. There are great uncertainties about results obtained by deposition plates or other surrogate collection surfaces. All such methods assume that the collection characteristics of some artificial surface are the same as those of the natural surface of interest. Clearly, this assumption will be valid when particles are suf- ficiently large that gravity is the controlling factor. However, small particles are transferred predominantly by turbulence, with subsequent impaction on the surface of microscale surface roughness elements; these latter factors are not easily reproduced by commonly used artificial collecting devices. The use of collection vessels to monitor the accumulation of particles in them continues to be a widespread practice; however, relating the data obtained to natural circumstances is difficult (q.v. Hicks et al. 1981). In a special category of its own, however, is the method of foliar washing, as used by Lindberg et al. (1979). As applied in careful studies of particle dry deposition at the Walker Branch Watershed in Tennessee, this method of removing and analyzing material deposited on vegetation has succeeded in demonstrating long-term average values of vd larger than the usually accepted values for several elements. 2.4 MICROMETEOROLOGICAL MODELS OF THE DRY-DEPOSITION PROCESS 2.4.1 Gases Almost all models of dry deposition of trace gases have as their foundation either the resistance analogy illus- trated in Figures C.2-5 and C.2-6 or some equivalent to it. The convenience of this approach is obvious: it permits separate processes to be formulated and combined in a manner that mimics nature, while providing a clear- cut mechanism for determining which processes can be omitted from consideration in specific circumstances. The relevance of the resistance approach to the matter of particle deposition is not so obvious, especially when gravitational settling must be considered. It is useful to start by identifying the gaseous properties of interest for gases and to identify possible controlling processes.

265 so2 NO2: Uptake by plants is largely via stomates during daytime, with about 25 percent apparently via the epidermis of leaves 1978). At night, stomata! resistance will increase substantially, but cuticular resis- tance should be unchanged. When moisture condenses on the depositing surface, associated resistances to transfer should be allowed to decrease to near zero (Fowler 1978, Murphy 1976). To a liquid-water surface, water-vapor appears to provide an acceptable analogy to SO2 flux. Behavior is like SO2, but with significant cuticular uptake at night (rout ~2-2.5 cm/s at night, see rc quoted by Wesely et al. 1982) and with surface moisture effectively minimizing uptake. Deposition to water surfaces, in general, is very slow. Similar to ~(Fowler o3 in overall deposition characteristics, but with a significant additional resistance (possibly mesophyllic, see Wesely et al. 1982b) of about 0.5 cm/s. Even though NO2 is insoluble in water in low concentrations deposition pm wafer ellrf=~= _, _, ~ ~ ~ ~ ~ ~ might be quite efficient. (Table C.2-4) ~ resistances for SO2 and NO2. NO: Typical canopy resistances are in the range 5-20 c~/s, as indicated by chamber studies (Table C.2-4) and field experiments (Wesely et al. 1982b). NO appears to be emitted by surfaces at times, possibly as a consequence of NC deposition and of the intimate linkage with ozone concentrations (Galbally Chamber studies Indicate similar overall surface and Roy 1980). HNC3: No direct information is available; however, on the basis of its high solubiliEv and chemical reactivity, substantial similarity to HE should be expected. Consequently, the use  of rc = 0 appears to be a reasonable first approximation. Again, no direct measurements are available but in this case similarity with SO2 appears likely. Natural surfaces may be emitters of NH3 because of a number of biological processes occurring in and on soil. NH3

266 Variations in aerodynamic resistance must be expected to modulate all the behavior patterns summarized above. In many circumstances, deposition rates at night will be nearly zero solely because atmospheric stability is so great that material cannot be transferred through the lower atmosphere. The evaluations given in Figure C.2-12 are especially informative, since even over a pine forest whose surface roughness operates to maximize Vd, an occasion was encountered on one evening out of eight in which atmospheric stability was sufficient to constrain the deposition velocity of all airborne material to less than 0.1 cm/s (with the exception of gravitationally settling particles). Some models focus on micrometeorological aspects of the pollutant transfer question, others on the biological. Meteorological models tend to follow the lead of agri- cultural workers. Shreffler (1976) considers profiles of pollutant concentration above and through a canopy, making use of the familiar concepts of zero-plane displacement, leaf area density, and aerodynamic roughness. Inter- facial transfer is formulated as recommended by Brutsaert (1975a). The results are shown to agree, in general form, with the experimental data of Chamberlain (1966, for Th-B). Roth (1975) also uses micrometeorological relations to investigate gaseous dry deposition, adopting the general resistance format and emphasizing the time dependency of the aerodynamic resistance and surface (i.e., canopy) resistance. Brutsaert (1975b) applies theory to cases of sensible heat and water vapor and extends them to consider a general scalar quantity. He emphasizes that the common micrometeorological assumption that the roughness length is the same for all quantities can lead to considerable error. O'Dell et al. (1977) consider details of leaf uptake mechanisms, with emphasis on stomata! and mesophyllic resistances. The model addresses questions of plant physiology in detail and is intended to permit comparison of uptake rates of different gaseous species, once air r,mn~-ntr At icons near leave s are known. None of these above models considers all the processes that have been discussed above, nor is it likely that any model will have sufficient generality to permit its use for all pollutants in all circumstances. To circumvent many of the difficulties involved, Sheih et al. (1979) prepared a land-use map of North America and coupled this with information regarding surface conditions in order to

267 derive a spatial distribution of surface resistances appropriate for formulating the deposition of sulfur dioxide. By coupling these values with aerodynamic resistances characteristic of different stability conditions and for different seasons, they produced a map of So2 deposition velocities suitable for use in numerical models and for interpreting concentration data. This approach has not been extended to other gaseous pollutants. 2.4.2 Particles Modeling of particle deposition is complicated by three major factors: (1) gravitational settling, which causes particles to fall through the atmospheric turbulence that provides the conceptual basis for conventional micro- meteorological models (Yudinge 1959); (2) particle inertia, which permits particles to be projected through the near-surface laminar layer by turbulence but also prohibits them from responding to the high-frequency turbulent motions that transport material near receptor surfaces; and (3) uncertainty regarding the processes that control particle capture. These three factors are interrelated in such a manner that clear-cut differentia- tion of their separate consequences is not possible. The problem has attracted the attention of many theoreticians, and many numerical models have been developed. Each model represents a selected combination of processes, chosen for consideration on the basis of the modeler's understanding of the problem. Without adequate consideration of all the mechanisms involved, none of these models can be considered as a simulator of natural behavior. This is not to question the worth of such models, but rather to emphasize that each should be applied with caution and only to those situations commensurate with its own assumptions. The many numerical models can be classified in several different ways. Some extend chemical engineering results to surface geometries that are intended to represent plant communities. Others extend agrometeorological air-canopy interaction models by including critical aspects of aerosol physics. Both approaches have benefits, and the final solution will probably include aspects of each. An excellent review of model assumptions has been given by Davidson and Friedlander (1978). They trace the evolution of models from the 1957 work of Friedlander and Johnstone, which concentrated on the mechanism of inertial

268 impaction and assumed that particles shared the eddy diffusivity of momentum, to the canopy filtration models of Slinn (1974) and Hidy and Heisler (1978). Early work concerned deposition to flat surfaces and made various assumptions about the surface collection process. Friedlander and Johnstone (1957) permitted particles to be carried by turbulence to within one free-flight distance of the surface, upon which they were assumed to be impacted by inertial penetration of the quasi-laminar "viscous" sublayer. Beal (1970) introduced viscous effects to limit the transfer of small particles, while retaining inertial impaction of larger particles. Sehmel (1970) assumed that all particles that contact the surface will be captured and used empirical evidence obtained in his wind-tunnel studies to determine the overall resistance to transfer, assumed to apply at a distance of one particle radius from the surface. Sehmel's work has been updated recently to provide an estimate of deposition velocities to canopies of a range of geometries in different meteorological conditions (see Sehmel, 1980a). The above models are based largely on observations and theory regarding the deposition of particles to smooth surfaces, usually of pipes. More micrometeorologically oriented models have been presented by workers such as Chamberlain (1967), who extended the familiar meteoro- logical concepts of roughness length and zero plane displacement to the case of particle fluxes. Much of this work was considered as an extension of models developed for the case of gaseous deposition to vegetation, which in turn were based on an extensive background of agricultural and forest meteorology, especially concerning evapotranspiration. A recent development of this genre is the canopy model of Lewellen and Sheng (1980), which utilizes recent techniques in turbulence modeling to reproduce the main features of suboanopy flow and combines these with particle- deposition formulations like those represented in Figure C.2-1. Lewellen and Sheng emphasize their model's omission of several potentially critical mechanisms, especially electrical migration, coagulation, evolution of particle size distributions, diffusiophoresis, and thermophoresis. To this list we can add a number of other factors about which little is known at this time such as suboanopy chemical reactions, interactions with emissions, and the effect of microscale roughness elements.

269 Although outwardly simpler than the case of particle deposition to a canopy, that to a water surface has given rise to a similar variety of models. Once again, however, different models focus on different mechanisms. That of Sehmel and Sutter (1974) is based on their wind-tunnel observations and lacks a component that can be identified with wave effects. Slinn and Slinn (1980) invoke the rapid growth of hydroscopic aerosol particles in very humid air to propose rather rapid deposition to open water; deposition velocities of the order of 0.5 cm/s appear possible in this case. On the other hand, Hicks and Williams tl979) propose negligible fluxes unless the surface quasi-laminar layer is interrupted by breaking waves. At present, none of these models has strong experimental evidence to support it. However, experi- mental and theoretical studies are proceeding, and a resolution of the matter can certainly be expected. 2.5 CONCLUSIONS TO SECTION 2 All the many processes that combine to permit airborne materials to be deposited at the surface have aspects that are strongly surface dependent. While broad generalities can be made about the velocities of deposition of specific chemical species in particular circumstances, there will be wide temporal and spatial variabilities of most of the controlling properties. The detailed nature of the vegetation covering the surface is often a critical consideration. If depositional inputs to some special sensitive area need to be estimated, then this can only be accomplished if characteristics specific to the vegetation cover of the area in question are taken into account in an adequate manner. Recent field studies investigating the fluxes of small particles have confirmed wind-tunnel results that point to a surface limitation. Studies of the rate of deposi- tion of particles to the internal walls of pipes and investigations of fluxes to surfaces more characteristic of nature, exposed in wind tunnels, tend to confirm theoretical expectations that surface uptake is controlled by the ability of particles to penetrate a quasi-laminar layer adjacent to the surface in question. The mech- anisms that limit the rate of transfer of particles involve their finite mass. Particles fail to respond to the high-frequency turbulent fluctuations that cause transfer to take place in the immediate vicinity of a surface. However, the momentum of particles also causes

270 an inertial deposition phenomenon that serves to enhance the rate of deposition of particles in the 10- to 20-pm size range. The general features of particle deposition to smooth surfaces are fairly well understood. Studies conducted so far support the theoretical expectation that particles smaller than about 0.1 Am in diameter will be deposited at a rate that is largely determined by Brownian dif- fusivity. In this instance, the limiting factor is the transfer by Brownian motion across the quasi-laminar layer referred to above. On the other hand, particles larger than about 20 Am in diameter are effectively transferred via gravitational settling, at rates deter- mined by the familiar Stokes-Cunningham formulation. Particles in the intermediate size ranges are transferred very slowly. The minimum value of the "well" of the deposition velocity versus particle size curve is approximately 0.001 cm/s. However, natural surfaces are rarely aerodynamically smooth. Wind-tunnel studies have shown that the "well" in the deposition velocity curve is filled in as the surface becomes rougher. Although studies have been conducted, in wind tunnels, of deposition fluxes to surfaces such as gravel, grass, and foliage, the situation involving natural vegetation such as corn, or even pasture, remains uncertain. It is well known that many plant species have foliage with exceedingly come plicated microscale surface roughness features. In particular, leaf hairs increase the rate of particle deposition; studies of other factors, such as electrical charges associated with foliage and stickiness of the surface, indicate that a natural canopy might be con- siderably different from the simplified surfaces suitable for investigation in laboratory and wind-tunnel investigations. Caution should be exercised in extending laboratory studies using artificially produced aerosol particles to the situation of the deposition of acidic quantities. Special concern is associated with the hydroscopic nature of many acidic species. Their growth as they enter a region of high humidity and their liquid nature when they strike the surface are both potentially important factors that might work to increase otherwise small deposition velocities. Moreover, there is evidence that acidic particles, especially sulfates, might be carried by larger particles; the rates of deposition of such complicated particle structures are essentially unknown. However, the shape of particles can have a considerable

271 influence on their gravitational settling speed and probably on their impaction characteristics as well. It is not clear to what extent special considerations appropriate for acidic species, such as those mentioned above, contribute to the finding of unexpectedly high deposition velocities for atmospheric sulfate particles, as reported in some recent North American studies. European work has been fairly uniform in producing deposition velocities close to 0.1 cm/s, while North American experience has generated larger values. It is informative to consider the flux of any airborne quantity to the surface underneath in terms of an electrical analog, the so-called resistance model developed initially in studies of agrometeorology. In this model, the flux of the atmospheric property in question is identified with the flow of current in an electrical circuit; individual resistances can then be associated with readily identifiable atmospheric and surface properties. While the electrical analogy has obvious shortcomings, it permits an easy visualization of many contributing processes and enables a comparison of their relative importance. Micrometeorological studies of the fluxes of atmospheric heat and momentum show that the aerodynamic resistance to transfer (i.e., the resistance to transfer between some convenient level in the air and a level immediately above the quasi-laminar layer) ranges from between 0.1 s/cm in strongly unstable, daytime conditions, to more than 10 s/cm in many nocturnal cases. There are several resistance paths that permit gaseous pollutants to be transferred into the interior of leaves. An obvious pathway is directly through the epidermis of leaves, involving a cuticular resistance. An alternative route, known to be of significantly greater importance in many cases. in ViA Why ^^r^= ~F ~- ~.r ~~ ~ ~eaves, Involving a stomata! resistance that ~ntrml.~ t ranster to within stomata! cavities and a subsequent mesophyllic resistance that parameterizes transfer from substomatal cavities to leaf tissue. Comparison between resistances to transfer for water vapor, ozone, sulfur dioxide, and gases that are similarly soluble and/or chemically reactive shows that in general such quantities are transferred via the stomata! route, whenever stomates are open. Otherwise, cuticular resistance appears to play a significant role. Cuticular uptake of ozone and of quantities like NO and NO2 appears to be quite significant, whereas for SO2 this pathway appears to be less important. When leaves are wet, such as after heavy dewfall, uptake of sulfur dioxide is exceedingly efficient

272 until the pH of the surface water becomes sufficiently acidic to impose a chemical limit on the rate of absorp- tion of gaseous SO2. However, the insolubility of ozone causes dewfall to inhibit ozone dry deposition. The same conceptual model can be applied to the case of particle transfer with considerable utility. While the roles of factors such as stomata! opening become less clear when particles are being considered, the concept of a residual surface resistance to particle uptake appears to be rather useful. Studies of the transfer of sulfate particles to a pine forest have shown that this residual surface resistance is of the order of 1 to 2 s/cm. It appears probable that substantially larger values for residual surface resistance will be appropriate for nonvegetated sufaces, especially to snow, for which the values are more likely to be approximately 15 s/cm. At this time, an exceedingly limited quantity of field information is available; however, it appears that in North American conditions the surface resistance to uptake of sulfate particles will be in the range 1.5 to 15 s/cm. While sulfate particles have received most of the recent emphasis, the general question of acid deposition requires that equal attention be paid to nitrate and ammonium particles. There is little information regarding the deposition velocities of these particles. However, there is no strong indication that they are different from the case of sulfate. Regarding trace-gas uptake, sulfur dioxide has received the majority of recent attention. Chamber studies and some recent field work indicate that highly reactive materials such as hydrogen fluoride (and presumably iodine vapor, nitric acid vapor, etc.) are readily taken up by a vegetative surface, whereas a VIA `3~ ^F r~1 1 ',+=nt. ¢: i n~1 Call ina Sol. NOR and O3 seem to be easily transferred via stomates, and a third category of relatively unreactive trace gases are poorly taken up. Transfer to water surfaces presents special problems, especially when the surface concerned is snow. As mentioned above, surface resistances to particle uptake by snow appear to be of the order 15 s/cm. Soluble gases will be readily absorbed by all water surfaces, and so equivalence to transfer of water vapor might be expected. An important exception occurs in the case of SO2, in which case absorbed SO2 can increase the acidity of the surface moisture layer to the extent that further SO2 transfer is cut off. Trace-gas transfer to liquid-water surfaces is influenced by the Henry's law constant. = By_ V1 1 ~-- ~- ~ ~-J - - ~ ~ ~

273 Wind-tunnel studies of particle transfer to water surfaces all show exceedingly small deposition velocities for particles in the 0.1- to 1-pm size range. Several workers have suggested mechanisms by which larger depo- sition velocities might exist in natural circumstances; for example, the growth of hydroscopic particles in highly humid, near-surface air can cause accelerated deposition of such particles, and breaking waves might provide a route that bypasses the otherwise-limiting quasi-laminar layer in contact with a water surface. Once again field observations are lacking. While larger deposition velocities of so uble trace gases to open water surfaces appear quite likely, water bodies are frequently sufficiently small that air-surface thermal equilibrium cannot be achieved. Air blowing from warm land across a small cool lake, for example, will not rapidly equilibrate with the smooth, cooler surface. Flow will then be stable and largely laminar, with the consequence that very small deposition velocities will apply for all atmospheric quantities. In many circuit stances, especially in daytime summer occasions, deposition velocities are likely to be so small as to be disregarded for all practical purposes. On the other hand, during winter when the land surface is frequently cooler than the water, the resulting corrective activity over small water bodies will cause the air to come into fairly rapid equilibrium with the water, and rather high deposition velocities, in agreement with the open water surface expectations, will probably be attained. An associated special case concerns the effect of dewfall, which can accelerate the net transfer of trace gases and particles in some circumstances. The velocities of deposition involved are small, however they permit an accumulation of material at the surface in conditions in which the atmospheric considerations are likely to predict minimal rates of exchange (i.e., limited by stability to an extreme extent). When surface fog exists, the highly humid conditions will permit airborne hydroscopic particles to nucleate and grow rapidly. This process provides a mechanism for cleansing the lower layers of the atmosphere of most acidic airborne par- ticles. The small fog droplets that are formed around the hydroscopic acidic nuclei are transferred by the classical process of fog interception to foliage and other surface roughness elements. Conclusions of recent workshops (e.g., Hicks et al. 1981) have indicated that it is not possible to measure

. 274 the dry deposition of acidic atmospheric materials using exposed collection vessels, since they fail to collect trace gases and small particles in a manner that can be related in a direct fashion to natural circumstances. However, surrogate surface methods appear to be useful in indicating space and time variations of deposition in some cases. It is possible to measure the flux of some airborne quantities by micrometeorological means, without interfering with the natural processes involved. These studies, and laboratory and wind-tunnel investigations, provide evidence that the controlling properties in the deposition of many gaseous pollutants are associated with surface structure, rather than with atmospheric prop- erties. The exception to this generalization is the nocturnal case, in which atmospheric stability may often be sufficient to impose a severe restriction on the rate of delivery of all airborne quantities to the surface below. The conclusions presented above can be summarized as follows: . Drv deposition of small aerosol particles and trace gases is a consequence of many atmospheric, surface, and pollutant-related processes, any one of which may dominate under some set of conditions. The complexity of each individual process makes it unlikely that a comprehensive simulation will be developed in the near future. · m e convenient simplicity afforded by the concept of a deposition velocity (or its inverse, the total resistance to transfer) makes it possible to incorporate dry-deposition processes in models in a manner that is adequate for modeling and assessment purposes. The simplicity of the deposition velocity approach imposes obvious limitations on its applications. For example, the use of average deposition velocities is inappropriate when it is desired to look at time- or space-resolved details of deposition fluxes. - Sufficient is known about the processes that control the deposition of trace gases that in many instances deposition velocities can be considered to be known functions of properties such as wind speed, atmospheric stability, surface roughness, and biological factors such as stomata! aperture. Important exceptions are for the case of insoluble (or poorly soluble) gases and for deposition to nonsimple surfaces such as forests in rough terrain.

275 · The deposition of particles larger than about 20-pm diameter is controlled by gravity and can be evaluated using the Stokes-Cunningham relationship. These large particles might contribute to the deposition of acidic and acidifying substances. The deposition of small particles remains an issue of considerable disagreement. On the whole, model predictions agree with the results of laboratory and wind-tunnel studies, at least for test surfaces that are usually smoother than pasture, but field experiments provide data that indicate greater deposition velocities. The reasons for the apparent disagreement are not yet . clear. · Over water surfaces, there are almost no field data on the deposition of small particles. Different models have been put forward, predicting a wide range of deposition velocities. At this time, there is little evidence that would permit us to choose between them. The situation for trace cases like sulfur ~ i ski ~" And ammonia is much better. ~ _ _ · ~. ~. J ~ . ~& On the whole, models agree with One evade f ield data, although there is disagreement between the models on how factors such as molecular diffusivity should be handled. Dry deposition to the surfaces of materials used in the construction of buildings and monuments, for example, can be measured in many instances by taking sequential samples of the surface over extended periods. Particulate material at the surface can creep, bounce, and eventually resuspend under the influence of wind gusts. The large particles entrained in this way can cause a local~modification of the acid deposition . . . phenomenon that is associated with accumulation-size aerosol particles and trace gases of more distant origin. For both case-study measurement purposes and for long-term monitoring, accurate measurements of pollutant air concentrations are necessary. For monitoring pur- poses, measurement of airborne pollutant concentrations in a manner carefully designed to permit evaluation of dry-deposition rates by applying time-varying deposition velocities specific to the pollutant and site in question appears to be the most attractive option. · Micrometeorological methods for measuring dry- deposition fluxes have been developed from the techniques conventionally used to determine fluxes of sensible heat, moisture, and momentum. These methods are techno- logically demanding, and their use in routine monitoring applications is not yet possible.

276 2.6 REFERENCES Alexander, L.T. 1967. Does salt water spray trap strontiu~90 from the air? U.S.A.E.C. Health and Safety Laboratory Quarterly Summary Report HASL-181, I-21-I-24. Arnts, R.R., R.L. Seila, R.L. Kuntz, F.L. Mowry, K.R. Knoerr, and A.C. Dudgeon. 1978. Measurement of a-pinene fluxes from a loblolly pine forest. Proceedings Fourth Joint Conference on Sensing of Environmental Pollutants. Washington, D.C.: American Chemical Society, pp. 829-833. Bagnold, R.A. 1954. The Physics of Blown Sand and Deser t Dunes. London: Methuen and Company. Barrie, L.A., and J.L. Walmsley. 1978. A study of sulphur d ioxide deposition velocities to snow in northern Canada. Atmos. Environ. 12:2321-2332. Batchelor, G.K. 1967. An Introduction to Fluid Mechanics. New York: Cambridge University Press. 615 pp. Beal, S.K. 1970. Deposition of particles in turbulent flow on channel or pipe walls, Nucl. Sci. Eng. 40:1-11. Bean, B.R., R.F. Gilmer, R.L. Grossman, and R. McGavin. 1972. An analysis of airborne measurements of vertical water vapor flux during BOMEX. J. Atmos. Sci. 29:860-869. Belot, Y. 1975. Etude de la captation des pollutants atmospheriques par les vegetaux, C.E.A., Fontenay-aux-Roses, France. Billings, C.E., and R.A. Gus sman. 1976. Dynamic behavior of aerosols, pp. 40-65 of Handbook on Aerosols. R.E. Dennis, ed. U.S. ERDA TIC-26608, 142 pp. Bowden, F.P., and D.Tabor. 1950. The Friction and Lubrication of Solids. Oxford: Clarendon Press. Brimblecombe, P. 1978. Dew as a sink for sulphur dioxide. Tellus 30:151-157. Brutsaert, W.H. 1975a. The roughness length for water vapor, sensible heat, and other scalars. J. Atmos. Sci. 32:2028-2031. Brutsaert, W.H. 1975b. A theory for local evaporation (or heat transfer) from rough and smooth surfaces at ground level. Water Resources Res. 11:543-550. Cadle, R.D. 1966. Particles in the Atmosphere and Space. New York:Reinhold. 226 pp. Chamberlain, A.C. 1960. Aspects of the deposition of radioactive and other gases and particles. Int. J. Air Pollut. 3:63-88. Chamberlain, A.C. 1966. Transport of gases to and from grass and grass-like surfaces. Proc. Roy. Soc. London A 290:236-265.

277 Chamberlain, A.C. 1967. Transport of Lycopodium spore s and other small particles to rough surfaces. Proc. Roy. Soc. London A 296:45-70. Chamberlain, A.C. 1975. The movement of particles in plant communities. Pp. 155-203 in Vegetation and the Atmosphere, Volume 1, Principles. J.L. Monteith, ed. London: Academic Press. Chamberlain, A.C. 1980. Dry deposition of sulfur dioxide. Pp. 185-197 in Atmospheric Sulfur Deposition. D.S. Shriner, C.R. Richmond, and S.E. Lindberg, eds. Ann Arbor, Mich.: Ann Arbor Science Publishers. Chamberlain, A.C. 1982. Roughness length of sea, sand and snow. Boundary-Layer Meteorol. Chang, S.G., R. Toossi, and T. Nova kov. 1981. The importance of soot particles and nitric acid in oxidizing SO2 in atmospheric aqueous droplets. Atmos. Environ. 15:1287-1292. Cofer, W.R., D.R. Schryer, and R.S. Rogowski. 1981. The oxidation of SO2 on carbon particles in the presence of O2, NO2 and N2O. Atmos. Environ. 15:1281-1286. Corn, M. 1961. The adhesion of solid particles to solid surfaces. Dannevik, W.P., S. Frisella, L. Granat, and R.B. Husar. 1976. SO2 deposition measurements in the St. Louis region, pp. 506-511 of Proceedings, Third Symposium on Atmospheric Turbulence, Diffusion, and Air Quality, Raleigh, N.C. Boston, Mass.: American Meteorological Society. Dasch, J.M. 1982. A comparison of surrogate surfaces for dry deposition collection, Proceedings, Fourth International Conference on Precipitation Scavenging, Dry Deposition, and Resuspension, Santa Monica, California, 29 November-3 December. Davidson, C.I., and S.K. Friedlander. 1978. A filtration model for aerosol dry deposition: application to trace metal deposition from the atmosphere. J. Geophy. Res. 83:2343-2352. Davies, C.N. 1966. Deposition from moving aerosols. Pp. 393-446 in Aerosol Science. C. N. Davies, eds. New York: Academic Press. 468 pp. Davies, C.N. 1967. Aerosol properties related to surface contamination. Pp. 1-5 of Surface Contamination. B.R. Fish, ed. New York: Pergamon Press. 415 pp. Derjaguin, B.V., and Yu.I. Yalamov. 1972. The theory of thermophoresis and diffusiophoresis of aerosol particles and their experimental testing. Pp. 1-200 of Topics in Current Aerosol Research, Part 2. G.M. Hidy and J.R. Brock, eds. New York: Pergamon Press. 384 pp.

278 Desjardins, R.L. 1977. Energy budget by an eddy cor- relation method. J. Appl. Meteorol. 16:248-250. Desjardins, R.L., and E.R. Lemon. 1974. Limitations of an eddy-correlation technique for the determination of the carbon dioxide and sensible heat fluxes. Boundary-Layer Meteorol. 5:47S-488. Dillon, P.J., D.S. Jeffries, and W.A. Scheider. 1982. The use of calibrated lakes and watersheds for estimating atmospheric deposition near a large point source, Water, Air Soil Pollut. Dolske, D.A., and D.F. Gatz. 1982. A field intercoms parison of sulfate dry deposition monitoring and measurement methods: preliminary results, Proceedings, American Chemical Society Acid Rain Symposium, Las Vegas, Nevada, 30 March 1982. Dovland, H., and A. Eliassen, 1976. Dry deposition on a snow surface. Atmos. Environ. 10:783-785. Droppo, J.G. 1980. Experimental techniques for dry deposition measurements. Pp. 209-221 of Atmospheric Sulfur Deposition. D.S. Shriner, C.R. Richmond, and S.E. Lindberg, eds. Ann Arbor, Mich.: Ann Arbor Science Publishers. 568 pp. Dyer, A.J. 1974. A review of flux-profile relationships. Boundary-Layer Meteorol. 7:363-372. Dyer, A.J., and F.J. Maher. 1965. The "Evapotron". An Instrument for the Measurement of Eddy Fluxes in the Lower Atmosphere, CSIRO (Australia) Division of Meteorological Physics Technical Paper Number 15. 31 pp. Eastman, J.A., and D.H. Stedman. 1977. A fast response sensor for ozone eddy-correlation measurements. Atmos. Environ. 11:1209-1212. Eaton, J.S., G.E. Likens, and F.H. Borman. 1978. The input of gaseous and particulate sulfur to a forest ecosystem. Tellus 30:546-551. Engelmann, R.J., and G.A. Sehmel, eds. 1976. Atmosphere- Surface Exchange of Particulate and Gaseous Pollutants, U.S. ERDA CONF-7409Z1. 988 pp. Everett, R.G., B.B. Hicks, W.W. Berg, and J.W. Winchester. 1979. An analysis of particulate sulfur and lead gradient data collected at Argonne National Laboratory. Atmos. Environ. 13:931-943. Fassina, V. 1978. A survey of air pollution and deterioration of stonework in Venice. Atmos. Environ 121:2205-2211. Fleischer, R.L., and F.P. Parungo. 1974. Aerosol particles on tobacco trichomes. Nature 250:158-159.

279 Fowler, D. 1978. Dry deposition of SO2 on agricultural crops. Atmos. Environ. 12:369-373. Fowler, D., and M. H. Unsworth. 1979. Turbulent transfer of sulphur dioxide to a wheat crop. Q. J. Roy. Meteorol. Soc. 105:767-784. Friedlander, S.K. 1977. Smoke, Dust, and Haze. New York: John Wiley and Sons. 317 pp. Friedlander, S.K., and H.F. Johnstone. 1957. Deposition of suspended particles from turbulent gas streams. Ind. Eng. Chem. 49:1151-1156. Fuchs, N.A. 1964. The Mechanics of Aerosols. New York: Pergamon Press. 408 pp. Galbally, I.E., and C.R. Roy. 1980. Ozone and nitrogen oxides in the southern hemisphere troposphere. Pp. 431-438 in Proceedings, Quadrennial International Ozone Symposium. 4-9 August 1980, Boulder, Colo. (available from IAMAP). Galbally, I.E., J.A. Garland, and M.J.G. Wilson. 1979. Sulfur uptake from the atmosphere by forest and farmland. Nature 280:49-50. Garland, J.A. 1977. The dry deposition of sulphur dioxide to land and water surfaces. Proc. Roy. Soc. London A 354:245-268. Garland, J.A. 1978. Dry and wet removal of sulphur from the atmosphere. Atmos. Environ. 12:349-362. Garland, J.A. 1979. Dry deposition of gaseous pollutants. Pp. 95-103 in WHO No. 538. Papers presented at the WHO Symposium on the Long-Range Transport of Pollutants and Its Relation to General Circulation Including Stratospheric/Tropospheric Exchange Processes, Sofia, 1-5 October 1979. Garland, J.A., W.S. Clough, and D. Fowler. 1973. Deposition of sulphur dioxide on grass. Nature 242:256-257. Garland, J.A., and J.R. Branson. 1977. The deposition of sulphur dioxide to a pine forest assessed by a radioactive tracer method. Tellus 29:445-454. Garratt, J.R. 1978. Flux profile relations above tall vegetation. Q. J. Roy. Meteorol. Soc. 104:199-212. Gauri, K.L. 1978. The preservation of stone. Sci. Am. 238:196-202. Hanna, S.R., and R.P. Hosker. 1980. Atmospheric removal processes for toxic chemicals, NOAA Technical Memorandum, ERL-ARL-102. 34 pp. Hardy, E.P., Jr., and J.H. Harley, eds. 1958. Environmental contamination from weapons tests. U.S. AEC Health and Safety Laboratory Report, HASL-42A.

280 Harriott, P., and R.M. Hamilton. 1965. Solid-liquid mass transfer in turbulent pipe flow. Chem. Eng. Sci. 20:1073. Hess, G.D., and B.B. Hicks. 1975. The influence of surface effects on pollutant deposition rates over the Great Lakes. Pp. 238-247 in the Proceedings of the Second Federal Conference on the Great Lakes, ICMSE. Hicks, B.B. 1981. Wet and dry surface deposition of air pollutants and their modeling. Pp. 183-196 in Conservation of Historic Stone Buildings and Monuments. National Materials Advisory Board. Washington, D.C.: National Academy Press. 365 pp. Hicks, B.B., and P.S. Liss. 1976. Transfer of SO2 and other reactive gases across the air-sea interface. Te flus 28:348-354. Hicks, B.B., and M.L. Wesely. 1978. An examination of some micrometeorological methods for measuring dry deposition. U.S. EPA report EPA-600/7-78-116. 27 pp. Hicks, B.B., and M.L. Wesely. 1980. Turbulent transfer processes to a surface and interaction with vegetation. Pp. 199-207 in Atmospheric Sulfur Deposition. D.S. Shriner, C.R. Richmond, and S.E. Lindberg, eds. Ann Arbor, Mich.: Ann Arbor Science Publishers. 568 pp. Hicks, B.B., and R.M. Williams. 1979. Transfer and deposition of particles to water surfaces, Chapter 26, pp. 237-244 in Atmospheric Sulfur Deposition. D.S. Shriner, C.R. Richmond, and S.E. Lindberg, eds. Ann Arbor, Mich.: Ann Arbor Science Publishers. 568 pp. Hicks, B.B., G.D. Hess, and M.L. Wesely. 1979. Analysis of flux-profile relationships above tall vegetation, an alternative view, Q. J. Roy. Meteorol. Soc. 105:1074-1077. Hicks, B.B., M.L. Wesely, and J.L. Durham. 1981. Critique of methods to measure dry deposition; concise summary of workshop, presented at the 1981 National ACS Meeting, Atlanta, Georgia. Ann Arbor, Mich.: Ann Arbor Science Publishers. Hicks, B.B., M.L. Wesely, R.L. Coulter, R.L. Hart, J.L. Durham, R.E. Speer, and D.H. Stedman. 1982. An experimental study of sulfur deposition to grassland Proceedings, Fourth International Conference on Precipitation Scavenging, Dry Deposition, and Resuspension, Santa Monica, California, 29 November-3 December. Hidy, G.M. 1973. Removal processes of gaseous and particulate pollutants, Chapter 3, pp. 121-176 in Chemistry of the Lower Atmosphere. S.I. Rasool, ed. New York: Plenum Press. 335 pp.

281 Hidy, G.M., and S.L. Heisler. 1978. Transport and deposition of flowing aerosols, in Recent Developments in Applied Aerosol Technology. D. Shaw, ed. New York: John Wiley and Sons. Hill, A.C. 1971. Vegetation: a sink for atmospheric pollutants. J. Air Pollut. Control Assoc. 21:341-346. Hubbard, D.W., and E.N. Lightfoot. 1966. Correlation of heat and mass transfer data for high Schmidt and Reynolds numbers. I/Ec Fundam. 5:370. Hudson, J.G., and P. Squires. 1978. Continental surface measurements of CON flux. J. Atmos. Sci. 35:1289-1295. Jones, E.P., and S.D. Smith. 1977. A first measurement of sea-air COk flux by eddy correlation. J. Geophys. Res. 82:5990-5992. Judeikis, H.S., and T.B. Stewart. 1976. Laboratory measurement of SO2 deposition velocities on selected building materials and soils. Atmos. Environ. 10:769-776. Judeikis, H.S., and A.G. Wren. 1977. Deposition of H2S and dimethyl sulfide on selected soil materials. Atmos. Environ. 11:1221-1224. Judeikis, H.S., and A.G. Wren. 1978. Laboratory measurements of SCt and NCt depositions onto soil and cement surfaces. Atmos. Environ. 12:2315-2319. Kaimal, C.J. 1975. Sensors and techniques for the direct measurement of turbulent fluxes and profiles in the atmospheric surface layer, Atmos. Technol. Number 7, 7-14. Kanemasu, E.T., M.L. Wesely, B.B. HickS, and J.L. Heilman. 1979. Techniques for calculating energy and mass fluxes. Chapter 3.2 in Modification of the Aerial Environment of Plants. B.J. Barfield and J.F. Gerber, eds. ASAE Monograph No. 2, pp. 156-182. Xanwisher, J. 1963. On the exchange of gases between the atmosphere and the sea. Deep-Sea Res. 10:195-207. Langer, G. 1965. Particle deposition on and re-entrainment from coniferous trees, Part II. Kolloid Z. 204:119-124. Lenschow, D.H., A.C. Delany, B.B. Stankov, and D.H. Stedman. 1980. Airborne measurements of the vertical flux of ozone in the boundary laver Rm',nH=rv-r.a`?^ Meteorol. 19:249-308. ~, ~,, ~ ~ Lewellen, W.S., and Y.P. Sheng. 1980. Modeling of Dry Deposition of SO2 and Sulfate Aerosols, Electric Power Research Institute Report EA-1452, 46 pp. (Available from EPRT, Research Reports Center, Box 50490, Palo Alto, Calif. 94303.)

282 Lindberg, S.E., R.C. Harris, R.R. Turner, D.S. Shriner, and D.D. Huff. 1979. Mechanisms and Rates of Atmospheric Deposition of Selected Trace Elements and Sulfate to a Deciduous Forest Watershed. Oak Ridge National Laboratory Report ORNL/TM-66 74, 514 pp. Liss, P.S. 1973. Processes of gas exchange across an air-sea interface. Deep-Sea Res. 20:221-238. Liss, P.S., and P.G. Slater. 1974. Flux of gases across the air-sea interface. Nature 247:181-184. Liu, B.Y.H., and J.K. Agarwal. 1974. Experimental observation of aerosol deposition in turbulent flow. Aerosol Sci. 5:145-155. Lovett, G.M., W.A. Reiners, and R.K. Olson. 1982. Cloud droplet deposition in subalpine balsam fir forests: hydrological and chemical inputs. Science 218:1303-1304. Martell, E.A. 1974. Radioactivity of tobacco trichomes and insoluble cigarette smoke particles. Nature 249:215-217. Meezaros, A., and K. Vissy. 1974. Concentrations, size distribution and chemical nature of atmospheric aerosol particles in remote oceanic areas. J. Aerosol Sci. 5:101-109. Mizushina, T., F. Ogino, Y. Oka, and H. Fukoda. 1971. Turbulent heat and mass transfer between wall and fluid streams of large Prandtl and Schmidt numbers. Int. J. Heat and Mass Transfer 14:1705-1716. Moller, U., and G. Schumann. 1970. Mechanisms of transport from the atmosphere to the earth's surface. J. Geophys. Res. 75:3013-3019. Monteith, J.~. 1963. Dew, facts and fallacies. In The Water Relations of Plants. A.J. Rutter and F.H. Whitehead, eds. New York: John Wiley and Sons. Pp. 37-56. Murphy, B.D. 1976. The influence of ground cover on the dry deposition rate of gaseous materials, Oak Ridge National Laboratory Report UCCCND/CSD-19. 28 pp. O'Dell. R.A., M. Taheri, and R.L. Kabel. 1977. A model for uptake of pollutants by vegetation. J. Air Pollut. Contr. Assoc. 27:1104-1109. Owers, M.J., and A.W. Powell. 1974. Deposition velocity . of sulphur dioxide on land and water surfaces using a S35 method. Atmos. Environ. 8:63-67. Raupach, M.R., J.B. Stewart, and A.S. Thom. 1979. Comments on "AnalYsis~of ~ flux-profile relationships above tall vegetation an alternative view" by Hicks et al. Q. J. Roy. Meteorol. Soc. 105:1077-1078.

283 Raynor, G.S., E.C. Odgen, and J.V. Hayes. 1970. Dispersion and deposition of ragweed pollen from experimental sources. J. Appl. Meteorol. 9:885-895. Raynor, G.S., E.C. Ogden, and J.V. Hayes. 1971. Dispersion and deposition of timothy pollen from experimental sources. J. Appl. Meteorol. 9:347-366. Raynor, G.S., E.C. Ogden, and J.V. Hayes. 1972. Dispersion and deposition of corn pollen from experimental sources. Agronomy J. 64:420-427. Raynor, G.S., J.V. Haynes, and E.C. Ogden. 1974. Particulate dispersion into and within a forest. Boundary-Layer Meteorol. 7:429-456. Rosinski, J., and C.T. Nagomoto. 1965. Particle deposition on and reentrainment from coniferous trees, Part I. Kolloid Z. 204:111-119. Roth, R., 1975: Der vertikale Transport van Luftbeimengungen in der Prandtl-Schmidt und die Deposition-Velocity. Meteorol. Rundsch. 28:65-71. Sehmel, G.A. 1970. Particle deposition from turbulent airflow. J. Geophys. Res. 75:1966-1781. Sehmel, G.A. 1979. Deposition and Resuspension Processes, Battelle, Pacific Northwest Laboratory Publication PNL-SA-6746. Sehmel, G.A. 1980a. Model Predictions and A numm~rv of ...~. . dry deposition velocity data. Pp. 223-235 of Atmospheric Sulfur Deposition. D.S. Shriner, C.R. Richmond, and S.F. Lindberg, eds. Ann Arbor, Mich.: Ann Arbor Science Publishers. 568 pp. Sehmel, G.A. 1980b. Particle and gas dry deposition: a review. Atmos. Environ. 14:983-1012. Sehmel, G.A., and S.L. Sutter. 1974. Particle deposition rates on a water surface as a function of particle diameter and air velocity. J. Rech. Atmos. 3:911-920. Sehmel, G.A., W.H. Hodgson, and S.L. Sutter. 1973a. Dry Deposition of Particles, Battelle Pacific Northwest Laboratory Report BNWL-1850. Part 3, pp. 157-162. Sehmel, G.A., S.L. Sutter, and M.T. Dana. 1973b. Dry Deposition Processes, Battelle Pacific Northwest Laboratories Report BNWL-1751. Part 1, pp. 43-49. Sheih, C.M., M.L. Wesely, and B.B. HickS. 1979. Estimated dry deposition velocities of sulfur over the eastern United States and surrounding regions. Atmos. Environ. 13:1361-1368. Shepherd, J.G. 1974. Measurements of the direct deposition of sulphur dioxide onto grass and water by the profile method. Atmos. Environ. 8:69-74. Shreffler, J.H. 1976. A model for the transfer of gaseous pollutants to a vegetational surface. J. Appl. Meteorol. 15: 144-746.

284 Sievering, H. 1982. Profile measurements of particle dry deposition velocity at an air-land interface. Atmos. Environ. 16:301-306. Sievering, H., M. Dave, D.A. Dolske, R.L. Hughes, and P. McCoy. 1979. An experimental study of loading by aerosol transport and dry deposition in the southern Lake Michigan basin. U.S. Environmental Protection Agency Report EPA-905/4-79-016. Sinclair, P.C. 1976. Vertical transport of desert particulates by dust devils and clear thermals. Pp. 497-527 of Atmosphere-Surface Exchange of Particulate and Gaseous Pollutants. R.J. Engelmann and G.A. Sehmel, eds. U.S. ERDA CONF-7409Z1. 988 pp. Slinn, W.G.N. 1974. Analytical Investigations of Inertial Deposition of Small Aerosol Particles from Laminar Flows into Large Obstacles--Parts A and B. PNL Ann. Report to the USAEC, DBER, 1973; BNWL-1850, Pt. 3, Battelle-Northwest, Richland, wash. Available from NTIS, Springfield, Va. Slinn, W.G.N. 1976a. Formulation and a solution of the diffusion, deposition, resuspension problem. Atmos. Environ. 10:763-768. Slinn, W.G.N. 1976b. Dry deposition and resuspension of aerosol particles--a new look at some old problems. Pp. 1-40 of Atmospheric-Surface Exchange of Particulate and Gaseous Pollutants--1974. R.J. Englemann and G.A. Sehmel, Coords. Available as ERDA CONF-740921 from NTIS, Springfield, Va. Slinn, S.A., and W.G.N. Slinn. 1980. Predictions for particle deposition and natural waters. Atmos. Environ. 14:1013-1016. Slinn, W.G.N., L. Hasse, B.B. Hicks, A.W. Hogan, D. Lal, P.S. Liss, K.O. Munnich, G.A. Sehmel, and O. Vittori. 1978. Some aspects of the transfer of atmospheric trace constituents past the air-sea interface. Atmos. Environ. 12:2055-2087. Spedding, D.J. 1969. Uptake of sulphur dioxide by barley leaves at low sulphur dioxide concentrations. Nature 224:1229-1234. Twomey, S. 1977. Atmospheric Aerosols. Amsterdam, Netherlands: Elsevier Scientific Publishing Company. 302 pp. Wason, D.T., S.K. wood, R. Davies, and A. Lieberman. 1973. Aerosol transport: particle charges and re-entrainment effects. J. Colloid Interface Sci. 43:144-149. Wedding, J.B., R.W. Carlson, J.J. Stiekel, and F.A. Bazzaz. 1975. Aerosol deposition on plant leaves, Environ. Sci. Technol. 9:151-153.

285 Wesely, M.L., and B.B. Hicks. 1977. Some factors that affect the deposition rates of sulfur dioxide and similar gases on vegetation. J. Air Pollut. Control Assoc. 27:1110-1116. Wesely, M.L., and B.B. Hicks. 1979. Dry deposition and emission of small particles at the surface of the earth. Pp. 510-513 in Proceedings Fourth Symposium on Turbulence, Diffusion and Air Quality, Reno, Nev., 15-18 January 1979. Boston, Mass.: Am. Meteorol .~- ... _ _ _ . ., ~ watery, M.~. , .. BACKS, W.P. Dannevik, S. Frisella, and R.B. Husar. 1977. An eddy correlation measurement of particulate deposition from the atmosphere. Atmos. Environ. 11:561-563. Wesely, M.L., J.A. Eastman, D.R. Cook, and B.B. Hicks. 1978. Daytime variation of ozone eddy fluxes to maize. Boundary-Layer Meteorol. 15:361-373. Wesely, M.L., D.R. Cook, R.L. Hart, B.B. Hicks, J.L. Durham, R.E. Speer, D.H. Stedman, and R.J. Trapp. 1982a. Eddy-correlation measurements of dry deposition of particulate sulfur and submicron particles. In Proceedings, Fourth International Conference on Precipitation Scavenging, Dry Deposition, and Resuspension, Santa Monica, California, 29 November-3 December. Wesely, M.L., J.A. Eastman, D.H. Stedman. and E.D, zalvac. 1982b. An eddy-correlation measurement of NCZ flux to vegetation and comparison to C3 flux. Atmos. Environ. 16:815-820. Whelpdale, D.M., and R.W. Shawl 1974. SulPhur dioxide t- Ames 1 W~ ~ ~ at_ ~ = ~ ~ _ ~=lvv"~ my curoulenc transfer over grass, snow, and other surfaces. Tellus 26:196-204. Whitby, K.T. 1978. The physical characteristics of sulfur aerosols. Atmos. Environ. 12:135-159. Williams, R.M., M.L. Wesely, and B.B. Hicks. 1978. Preliminary eddy correlation measurements of momentum, heat, and particle fluxes to Lake Michigan. Argonne National Laboratory Radiological and Environmental Research Division Annual Report, January-December 1978, ANL-7865 Part III, pp. 82-87. Winkler, E.M., and E.J. Wilhelm. 1970. Saltburst by hydration pressures in architectural stone in urban atmosphere. Geol. Soc. Am. Bull. 81:576-572. Yudinge, M.E. 1959. Physical considerations on heavy-particle diffusion. Pp. 185-191 of Advances in Geophysics, Volume 6. H.E. Landsberg and J. van Mieghem, eds. New York: Academic Press.

286 3 . PRECIPITATION SCAVENGING PROCESSE S 3.1 STEPS IN THE SCAVENGING SEQUENCE 3.1.1 Introduction Precipitation Scavenging is defined generally as the composite process by which airborne pollutant gases and particles attach to precipitation elements and thus deposit to the Earth's surface.* This process typically contains many parallel and consecutive steps, and as an introduction to this section it is appropriate to provide a brief overview of these intermeshing pathways. In a general sense there are four major events in which a natural or pollutant moleculet may participate, prior to its wet removal from the atmosphere; depicted pictorially in Figure C.3-1, these are as follows: 1 - 2. 2-3 . The pollutant and the condensed atmospheric water (cloud, rain, snow, . . .) must intermix within the same airspace. The pollutant must attach to the condensed-water elements. *One should note that this definition pertains to removal from the gaseous medium of the atmosphere combined with deposition to the ground. An alternative definition, employed often throughout the open literature, pertains to the simple attachment of airborne pollutants to liquid water elements, without regard to whether the material is subsequently conveyed to the Earth's surface. Which of these definitions is used is unimportant so long as the precise definition is understood. The definition of "scavenging" adopted here will be utilized consistently throughout this text. When specific reference to the alternative situation is made, the terms "attachment" and "capture" will be employed essentially interchangeably. "Initial portions of this section will treat precipitation scavenging in a general sense, with limited reference to specific types of atmospheric material. The reader should continue to note, however, that the "natural or pollutant molecules" of primary concern in the present context are species associated with acid-base formation, such as SO2, HNO3, NH3, sulfate, chloride, and metallic cations.

287 ~ _ 4< G ~ i - You at: an JO _:-~ 1~1 i_;  ~I 1 _ _ ~ o Ct CO Q) ~4 . - o C) . . C) C) C) :- Cal on Ct C) 5~ \ _ 3 -

288 3-4. The pollutant may react physically and/or chemically within the aqueous phase. 3-5 or (4-5). The pollutant-laden water elements must be delivered to the Earth's surface via the precipitation process. The interaction diagram of Figure C.3-2 gives a somewhat more detailed portrayal of these four major events. Here the individual steps are represented as transitions of the pollutant between various states in the atmosphere, and one can note that a multitude of reverse processes is also possible; thus a particular pollutant molecule may experience numerous cycles through this complex of pathways prior to deposition. Indeed, - Figure C.3-2 indicates that this cycling process may ~ By pollutant off-cassinu and other resuspension processes the deposited continue even atter ultimate aeposz~zon. material can be re-emitted to the atmosphere, with the possibility of participating in yet another series of cycles throughout the scavenging sequence. Another important feature of Figure C.3-2 is its indication that, while physicochemical reaction within the aqueous phase is potentially an important step in the This contrasts - scavenging process, it is not essential. to the remaining forward steps that must take place it scavenging is to occur. Despite its nonessential nature, this step is often of utmost importance in influencing scavenging rates, owing to its role in modifying reverse processes in the sequence. An example of this effect is the devolatilization of dissolved sulfur dioxide via wet oxidation to sulfate. This effectively eliminates gaseous Resorption from the condensed water and thus has a strong tendency to enhance the overall scavenging rate as a result. From Figure C.3-2 one can note also that precipitation scavenging of pollutant materials from the atmosphere is intimately linked with the precipitation scavenging of water. If one were to replace the word "pollutant" with "water vapor" in each of the steps, Figure C.3-2 (with the exception of box 4) would provide a general descrip- tion of the natural precipitation process. In view of this intimate relationship, it is not surprising that pollutant wet-removal behavior tends to mimic that of precipitation. Pollutant-scavenging efficiencies of storms, for example, are often similar to water-extraction efficiencies. This relationship is useful in the practical estimation of scavenging rates and will

289 m I ;~ ~1 POLLUTANT IN CLEAR AIR cot _ ~{.r~ em <=l ~ cat x 0 POLLUTANT AND CONDENSED WATER INTERMIXED IN COMMON AIRSPACE POLLUTANT ATTACHED TO CONDENSED WATER ELEMENTS l . o o _ 4 ATTACHED POLLUTANT MOD I F I ED BY AQUEOU S - PHA SE PHY S l OCOCHEM I CAL REACT I ON S  1 POLLUTANT DEPOSITED ON EARTH'S SURFACE FIGURE C.3-2 Scavenging sequence: interaction diagram. _ v v 0 _ . 0 Ohio ,~,~c 1

290 reappear continually in the ensuing discussion of wet-removal behavior. Figure C.3-2 is interesting also because of its indication that, if some particular step in the diagram occurs particularly slowly compared with the others, then this step will dominate behavior of the overall process. This is similar to the "rate-controlling step" concept in chemical kinetics and has been applied rather extensively in practical scavenging calculations (Slinn 1974a). Finally, it is important to note that Figure C.3-2 presents a framework for development and evaluation of mathematical models of scavenging behavior. Successful scavenging models must emulate these steps effectively and tend to reflect the structure of Figure C.3-2 as a result. This point will re-emerge later when scavenging models are examined specifically. The remainder of the present subsection will address qualitative aspects of the scavenging sequence, in the order of their forward progress to ultimate deposition. 3.1.2 Intermixing of Pollutant and Condensed Water (Step 1-2) On first consideration one often is inclined to dismiss pollutant/condensed-water intermixing as an unimportant, or at least trivial, step in the overall scavenging sequence. It is neither. In a statistical sense it usually is neither cloudy nor precipitating in the immediate locality of a freshly released pollutant molecule; and typically this molecule must exist in the clear atmosphere for several hours, or even days, before it encounters condensed water with which it may commingle. This in itself establishes step 1-2 as a potentially important rate-influencing event. Moreover, this extended dry period typically presents the pollutant with signifi- cant opportunities to react and/or deposit via dry processes; thus the chemical makeup of precipitation is influenced profoundly by this preceding chain of events. Significant insights into the behavior of step 1-2 can be gained via past analyses of storm formation (e.g., Godske et al. 1957) and the atmospheric water cycle (Newell et al. 1972). Several statistical analyses of precipitation occurrence (Baker et al. 1969; Junge, 1974; Rodhe and Grandell 1972, 1981; Slinn, 1973b) have been applied as general interpretive descriptors of this step. These will not be examined in detail here; rather,

291 we shall concentrate on the mechanisms by which step 1-2 can occur, from a more pictorial viewpoint. Two types of mixing processes exist in which pollutant and condensed water can ammo ~^ Allot Amman ~ _ these are as follows: and ~, ~-~. ~ .~=- 1. Relative movement of the initial ~ v unmixed pollutant and condensed water, in a manner such that they merge into a common general volume; 2. In situ phase change of water vapor, thus producing condensed water in the immediate vicinity of pollutant molecules. The relative importance of Type-1 and Type-2 mixing processes will depend to some extent on the pollutant. If a particular pollutant is easily scavengable, and if precipitation is occurring at the pollutant's release location, then Type-1 processes are likely to contribute significantly. If these two conditions are not met, the pollutant will usually mix intimately with makeup water vapor for some future cloud. and Tvme-2 nr^~"ccmc w; ~ ~ . predominate. Based on in-cloud versus below-cloud scavenging estimates (Slinn 1983) it is not unreasonable to estimate that, as a global average, roughly 90 percent of all precipitation scavenging occurs as the ~^n~'l~n-" of a Type-2 Process. - --~ _- ~- ~v1~cI1-C _, ~ _ _ ~ ~ As indicated in Figure C.3-2, reverse processes exist that can serve to reseparate pollutant and condensed water. Evaporation, for example, can reinject pollutant from cloudy to clear air, and relative motion such as precipitation ~fall-through" can remove hydrometeors from contact with elevated plumes. Cloud formation-re- evaporation cycles are particularly significant in this respect. Junge (1964), for example, estimates that a single cloud condensation nucleus is likely to experience of the order of ten or more evaporation-condensation cycles before it is ultimately delivered to the Earth's surface with precipitation. The rate-influencing effect of such cycling on precipitation scavenging is obvious. Additional types of cycles will be described below, in conjunction with succeeding steps of the scavenging sequence.

292 3.1.3 Attachment of Pollutant to Condensed-Water Elements (Step 2-3) The microphysics of the pollutant-attachment process has been the subject of extensive research, and numerous ~ ~ , _ ~ _ ~ _ reviews of this area have been prepares (e.g., Davies 1966, Dingle and Lee 1973, Hales 1983, Junge 1963, Pruppacher and Klett 1978, Slinn 1983, Slinn and Hales 1983). In the context of Figure C.3-1, this process is complicated somewhat in the sense that, depending on the particular attachment mechanism, Step 2-3 may occur either simultaneously or consecutively with Step 1-2. Simultaneous comixing and attachment occur in the case . of cloud-particle nucleation. This is a phase- transformation (Type-2) process wherein water molecules, thermodynamically inclined to condense from the vapor phase, migrate to some suitable surface for this purpose. Pollutant aerosol particles provide such surfaces within the air parcel, and the consequence is a cloud of droplets (or ice crystals)* containing attached pollutant material. Different types of aerosol particles possess different capabilities to nucleate cloud elements and grow by the condensation process. As a consequence there is typically a competition for water molecules among the aerosol and associated cloud particles; some will capture water with high efficiency and grow substantially in size. Others will acquire only small amounts of water, and still others will remain essentially as "dry" elements. In addition, some particles may nucleate ice crystals, while others will be active only for the formation of liquid water. The nucleating capability of a particular aerosol Particle is determined by its size, its morphological , _ cnaracter~stics, and its chemical composition. Various *At this point it is important to note that aerosols can participate in several types of phase transitions in cloud systems. These include vapor-liquid, vapor-solid, and liquid-solid transitions, in addition to a subset of interactions between numerous solid phases. Particles active as ice-formation nuclei are generally muon less abundant than those active as droplet (or "cloud- condensation") nuclei. As will be demonstrated later, the relative abundance of ice nuclei can have a profound effect on precipitation-formation processes and related scavenging phenomena. .

293 aspects of this subject are discussed at length in standard cloud-physics textbooks (e.g., Mason 1971, Pruppacher and Klett 1978) and in the periodical literature (e.g., Fitzgerald 1974). An additional important aspect of the cloud-droplet nucleation and growth process is the fact that once initiated, cloud-droplet growth does not proceed instantaneously to some sort of thermodynamic equilibrium. Because of diffusional constraints on delivering water molecules from the surrounding atmosphere, the growth in droplet diameter slows appreciably as droplet size increases (cf. Slinn 1983). Superimposition of this lag on the continually fluctuating environment of a typical cloud results in a dynamic and complex physical system. Finally, the competitive nature of the cloud-nucleation process results in significant impacts by the pollutant on the basic character of the cloud itself. If the local aerosol were populated solely by a relatively small number of large, hydroscopic particles, for example, would expect any corresponding cloud to be composed chiefly of low populations of large droplets. If on the other hand the local aerosol were composed of large numbers of small, nonhygroscopic particles, the corresponding cloud should contain 1 freer my an smaller droplets. ~. ~. ~, _ _ . ~ ~_~ ^ ~ ~ Inks is precisely what is observed in practice. Unpolluted marine atmospheres, for example, contain large sea-salt particles as a primary component of their aerosol burden. Warm marine clouds are noted for their wide drop spectra containing large drop sizes and their corresponding capability to form precipitation easily. Continental clouds, on the other hand, are typically composed of larger populations of smaller droplets. Figure C.3-3, which was prepared on the basis of results published by Squires and Twomey (1960), provides a good example of this point. Here measured convective-cloud droplet spectra are compared for two different cloud systems. The continental air-mass cloud exhibits a distinct tendency toward smaller drop sizes and larger populations, as compared with its maritime counterpart. It is interesting also in this context to note the estimates of Junge (1963) with regard to relative amounts of aerosol participating in the nucleation process. Junge suggests that while 50 to 80 percent of the mass of continental aerosols can be expected to participate as cloud nuclei, as much as 90 to 100 percent of maritime aerosols can become actively involved. one

294 U1 Hi: ~ 41: - 6 J ~ a: a: Z Z At o . o o  £ Urn/ 513~dO~G 'NOIlV~lN3ON03 13~do~a Gnome U. U) CO Ct ·Ct - o Cal Ct C) £ ._ ~4 _~ a:- LL~ ~ Cue LL~ A~ O> _._ LL~ C] C}~ ~o OCal sol ° Ct ~ _ ~ ;^ ~ £ o ~ _ ¢) o ~ o . - C~ ~ V, o ~L. ¢

295 As a concluding note in the context of nucleating capability and water competition it should be pointed out that acid-forming particles, by their very nature, are chemically competitive for water vapor and thus tend to participate actively as cloud-condensation nuclei. This attribute tends to enhance their propensity to become scavenged early in storm systems and has a significant effect on the nature of the acid-rain formation process. There are numerous mechanisms by which pollutants can attach to cloud and precipitation elements offer the elements already exist, and thus in a manner that is consecutive with Step 1-2. These mechanisms are itemized in the following paragraphs; they are typically active for both aerosols and gases, although the relative importances and magnitudes vary widely with the state of the scavenged substance. Diffusional attachment, as its name implies, results from diffusional migration of the pollutant through the air to the water surface. This process may be effective both in the case of suspended cloud elements and falling hydrometeors. It depends chiefly on the magnitude of the pollutant's molecular (or Brownian) diffusivity; and since diffusivity is inversely related to particle size, this mechanism becomes less important as pollutant elements become large. Diffusional attachment is of utmost importance for scavenging of gases and very small aerosol particles. For all practical purposes it can be ignored for aerosol particle sizes above a few tenths of a micrometer. In concordance with Fick's law (Bird et al. 1960), diffusional transport to a water surface is dependent as well on the pollutant's concentration gradient in the vicinity of this surface. Thus if the cloud or precipi- tation element can accommodate the influx of pollutant readily, it will effectively depopulate the adjacent air, thus making a steep concentration gradient and encour- aging further diffusion. If for some reason (e.g., particle "bounce off n or approach to solute saturation) the element cannot accommodate the pollutant supply, then further diffusion will be discouraged. If the cloud or precipitation element, through some sort of outgassing mechanism, supplies pollutant to the local air, then the concentration gradient will be reversed and diffusion will carry the pollutant away from the element. Mixing processes inside of cloud or precipitation elements play an important role in determining the accommodation of gaseous species. If mixing is slow, for

296 example, it is likely that the element's outer layer will saturate with pollutant and thus inhibit further attach- ment processes. This is quite often a limiting factor in cases involving gas scavenging by ice crystals. Internal mixing occurs as a consequence of diffusion and fluid circulation and has been analyzed at length by Pruppacher and his co-workers (cf. Pruppacher and Klett 1978). In general diffusional attachment processes are sufficiently well understood to allow their mathematical description with reasonable accuracy, and numerous references are available as guides for this purpose (e.g., Hales 1983, Pruppacher and Klett 1978, Slinn 1983). Inertial attachment processes are directly dependent - on the size of the scavenged particle and thus are unimr portent for gaseous pollutants. In a somewhat general sense this class of processes depends on motions of pollution particles and scavenging elements relative to the surrounding air, which arise because both have finite volume and mass. The most important example of inertial attachment is the impaction of aerosols on falling hydrometeors. Here the hydrometeor (because of its mass and volume) falls by gravity, sweeping out a volume of space. Some of the aerosol particles (because of their mass) cannot move sufficiently rapidly with the flow field to avoid the hydrometeor and thus are impacted. In principle impaction could occur, even if the aerosol particles were point masses with zero volume. Assigning a volume to a particle further increases its chance of collision, simply on the basis of geometric effects. The inclusion of aerosol volume has been generally referred to in the past literature as accretion. A second example of inertial attachment is turbulent collision. In this case the particles and scavenging elements, subjected to a turbulent field, collide because of dissimilar dynamic responses to velocity fluctuations in the local air. This scavenging mechanism is thought to be of secondary importance and has received compara- tively little attention in the past literature, although some recent theoretical analyses have suggested it to be significant for specific drop-size/particle-size ranges. While the mechanisms of diffusional and inertial attachment are efficient for capturing very fine and very coarse particles, respectively, a region of low efficiency should exist in the 0.1-5 Em range, where the mechanism is effective. This effect is shown schematically in Figure C.3-3. Because its importance to scavenging was first recognized by Greenfield (19S7) it has become known

297 generally as the "Greenfield gap. n Depending on circuit stances there are several additional attachment mech- anisms (including the two-stage nucleation-impaction mechanism mentioned earlier) that can serve to "fill" the Greenfield gap. Some of the more important of these are itemized in the following paragraphs. Diffusiophoretic attachment to a capturing element can occur whenever the element grows via the condensation of water vapor. In effect the flux of condensing water vapor "sweeps" the surrounding aerosol particles to the element's surface. In a competitive cloud-element system where some droplets grow while others evaporate, diffusio- phoresis can be a rather important secondary attachment mechanism. This is particularly true when the cloud contains mixtures of ice and liquid. Under such conditions the ice crystals have a pronounced tendency, owing to their lower equilibrium vapor pressure, to gain water at the expense of the droplets. Known as the Bergeron-Findeisen effect, this process is important in precipitation formation as well as in diffusiophoretic enhancement. Thermophoretic attachment results from a temperature gradient in the direction of the ca~turina Pluming H-~" ~ ~ _, ~ ~= Serene aces essentially as a miniature thermal precipitator. Warmer gas molecules on the outward side of the aerosol particle impart a proportionately larger amount of momentum, resulting in a driving force toward the capturing element.* Thermophoresis depends directly on the temperature gradient in the vicinity of the capturing element. In cloud and precipitation systems local temperature gradients are caused most often by evaporation/ condensation effects; thus thermophoresis is usually strongly associated with diffusiophoresis,l and in fact these two processes often tend to counteract each other. *One should note that the precise mechanisms of thermal transport differ radically. denen~ina an n~rhi~l" a; ~" (cf. Cadle 1965). 1 ~ --=~a-^~ ~ H~= ~= Off= tAs noted by Slinn and Hales (1983) inappropriate treatment of this relationship has caused erroneous conclusions to be drawn in some of the past literature. The reader should be cognizant of this if more detailed pursuit is intended.

298 10° 10 - UJ 10_2 1 0-3 10-4 GREENFIELD GAP ~ Z I I t ''~<i l,,,,l , , , I,,,,l -I-~-li,,,,l , , , i,,,,l , I , 1,... l `~` A, ~_ _ ~ , 1 ~ 10-3 10-2 10-1 1.0 10 RADIUS OF AEROSOL PARTICLE 1,um) FIGURE C.3-4 Theoretical scavenging efficiency of a falling raindrop as a function of aerosol particle size. Adapted from Pruppacher and Klett ( l 978 ). Dashed lines cor- respond to contributions by electrical and phoretic effects under chosen humidity and raindrop-charge conditions (see original reference for details). Phoretic processes are unimportant in the case of gaseous pollutants, owing to the overwhelming con- tributions of molecular diffusion. At present the theory of diffusiophoretic/thermophoretic particle attachment is at a state where reasonably quantitative assessments can be made for simple systems such as isolated droplets (Pruppacher and Klett 1978, Slinn and Hales 1971) (cf. Figure C.3.4). Rough estimates are possible also for more complex and interactive cloud/precipitation systems, but much remains to be done to bring our knowledge of this area to a really satisfactory state. Electrical attachment of aerosol particles to cloud and precipitation elements has been the subject of continuing study over the past three decades. Under- standing of this process is currently at a state where

299 relationships between aerosols and isolated drops can be quantified with reasonable accuracy (cf. Wang and Pruppacher 1977). In general, electrical charging of cloud andVor precipitation elements must be moderately high for electrical effects to become competitive with other capture phenomena, although such charging is certainly possible in the atmosphere--particularly in convective-storm situations. Understanding of electrical deposition in clouds of interacting drops is still at a relatively unsatisfactory stage of development. As a conclusion to this discussion of attachment processes it is appropriate to note that, while the mechanisms have been presented here on an individual basis, they tend in actuality to proceed in a simultaneous and competitive manner. Insofar as atmospheric cleansing is concerned, this is a fortunate ~ ircum~n~- hat ~, ~ ~ _ ~ ~ ~ ~ some mechanisms tend to be operative in physical situa- tions where others are ineffective. Figure C.3-4 gives an excellent illustration of this point. Here theo- retical attachment efficiencies appropriate to a 0.31-mm radius raindrop are presented for various electrical and relative-humidity conditions, demonstrating the capability of phoretic and electrical mechanisms to "bridge" the Greenfield gap. This simultaneous and competitive interaction of mechanisms serves to complicate profoundly the mathematics of the scavenging process and lends an additional degree of difficulty to the problem of scavenging calculations. This aspect will continue to emerge throughout this section, especially during the discussion of scavenging models. 3.1.4 Aqueous-Phase Reactions (Step 3-4) Aqueous-phase conversion phenomena have been discussed in some detail in Appendix A and will not be examined further here except to note their general importance within the framework of the overall scavenging sequence. Or ; _ ~ u~ ___s ~ As observed =~` . _~2 ~ ll ~= ~Ul1~=X~ or figure ~.~-~, aqueous-phase reactions are not essential to the scavenging process. Depending on the pollutant material, however, these reactions often can have the effect of stabilizing the captured material within the condensed phase and thus enhancing the scavenging efficiency appreciably. There is much to be learned before this important aspect is brought to a satisfactory stage of understanding.

300 3.1.5 Deposition of Pollutant with Precipitation (Steps 3-5 and 4-5) Although a variety of mechanisms exist (e.g., impaction of fog on vegetation), the predominant means for deposit- ing pollutant-laden condensed water to the Earth's surface is simply gravitational sedimentation. Sedimentation rates depend on hydrometeor fall velocities, which depend i n turn on hvdrometeor size: thus the processes by which · · ~ · ~ ~ the pollutant-laden cloud droplets grow to precipitation elements emerge as major determining factors in this final stage of the scavenging sequence. Once attached to condensed water, a pollutant molecule has several alternative pathways for action (Figure C.3-2). If the captured pollutant possesses some degree of volatility, it may desorb back into the gas phase. Reverse chemical reactions may occur. Evaporation of the condensed water may, in effect, ~free" the pollutant to the surrounding gaseous atmosphere. This multitude of pathways results in an active competition for pollutant. If the precipitation stage of the scavenging sequence is to be effective, it must interact successfully within this competitive framework. Besides competing actively for pollutants, the above interactions produce a vigorous competition for water. This parallel relationship between pollutant scavenging and water scavenging, apparent in some of the preceding discussion regarding attachment processes, can be drawn even more emphatically when considering precipitation processes. The following paragraphs provide a brief overview of some of the more important mechanisms in this regard. Once initial nucleation has occurred, cloud particles mav crow further by condensation of additional water vapor. Net condensation will occur to the surface or a cloud element whenever water vapor molecules can find a more favorable thermodynamic state in association with it; and because clouds contain varieties of makeup elements having different thermodynamic characteristics, a competition for water vapor usually exists. Such interactions are discussed at length in standard text- books (Mason 1971, Pruppacher and Klett 1978). Slinn (1983) has developed a conceptual scavenging model in which condensational growth is an important rate-limiting step. Thermodynamic affinity for water-vapor molecules depends on the cloud-element's size, its pollutant

301 burden, and its physical structure. These latter two factors often influence precipitation characteristics profoundly. In particular, the favored thermodynamic state of a water molecule in association with an ice crystal (as compared with a supercooled water droplet) results in rapid competitive growth of ice particles in mixed-phase clouds. This Bergeron-Findeisen process has been mentioned already in the context of diffusiophoretic and thermophoretic transport. Growth of large cloud elements via this process is the primary reason that ice-containing clouds tend to be so strongly effective generators of precipitation water. A further mechanism by which suspended cloud droplets can grow to form precipitation elements is coagulation. This process occurs via the collision of two or more cloud elements to form a new element containing the total mass (and pollutant burden)* of its predecessors. Coagu- lation occurs over size-distributed systems of cloud elements by a variety of physical mechanisms, and because of this it is a rather poorly understood and mathemati- cally complex process. Comprehensive analyses of coagulation processes have been performed by Berry and Reinhardt (1974). Coagulation can be considered to be an important initiator of precipitation in single-phase clouds (water or ice). In mixed-phase clouds the Bergeron-Findeisen process can be expected to enhance the coagulation process by widening the droplet size distribution, as well as contributing to precipitation growth in a direct sense. Once a moderate number of precipitation-sized elements have been generated, the process of accretion ration ~ as L~t:41nS C0 (lOmlnAt~ == :~ macro Be_ ~~_A__= ~ ~ I ~ Van ye~=Laul~ly preclplraclOn water. As noted previously, this process occurs by the "sweeping" action of large hydrometeors falling through the field of smaller elements, attaching them on the way. As was the case with coagulation, the accretion process tends to accumulate the pollutant burden of all collected elements. *Coagulation is often referred to as autoconversion in one c~oud-physics literature. It is interesting to notice in this context that while coagulation tends to accumulate nucleated pollutants, the Bergeron-Findeisen process tends to re-liberate nucleated pollutants to the air.

302 Accretion can occur via drop-drop, drop-crystal, and crystal-crystal interactions. Drop-crystal interactions are particularly important in mixed-phase clouds; when supercooled droplets are accreted by falling ice crystals, the process is usually referred to as riming. Although the above discussion has been confined primarily to deposition in conjunction with rain and snow, it should be emphasized that fog deposition* often is an important secondary process for conveying pol- lutants to the Earth's surface. Classification of fog-bound pollutant deposition is problematic for two major reasons. The first of these is that no sharp demarcation exists between n fog droplets" and "water- containing aerosols~; thus the choice of considering fog deposition as simply the dry deposition of wet particles, or the wet deposition of contaminated water, depends primarily on personal preference. Secondly, there is no real distinction between fog droplets and precipitation. Cloud physicists often find it convenient to categorize condensed atmospheric water into "precipitation" and "cloud" classifications, with the presumption that aloud water has a negligible sedimentation velocity. Such a classification is of limited use when considering fog deposition, however, owing to the fact that fog droplets do have significant gravitational fall speeds. A 50-pmrdiameter fog droplet, for example, will fall at a rate of about 10 cm/s. This, combined with the fact that typical fogs and clouds contain droplet-size distributions ranging between O to 100 Am (cf. Pruppacher and Klett 1978), suggests that gravitational transport of fog droplets will indeed be a significant pollution- deposition pathway under appropriate circumstances. In addition to purely gravitational transport, fog droplets have a strong tendency to impact on projected surfaces. The rates of fog impaction depend in a complex fashion on drop size, wind velocity, and geometry of the projected object. The common observations of rime-ice accumulation on alpine forests and on power-transmission lines give direct testimony to the effectiveness of this process. Chemical deposition by fogs is directly proportional to fog-bound pollutant concentration, and this fact often acts to enhance substantially the pathway's overall *A n fogs is (rather pragmatically) defined here as any cloud that is in the proximity of the Earth's surface.

303 effectiveness. Owing to their proximity to the Earth's surface, fogs typically form in conjunction with high pollutant concentrations. Attaching particles and gases via the variety of mechanisms described in Section 3.1.3, the droplets typically accumulate extremely high burdens of material. It is not difficult to find evidence in support of this point. Scott and Laulainen (1979), for example, reported sulfate and nitrate concentrations approaching 500 liter in water obtained near the bases of clouds over Michigan, while the SUNY group has reported (e.g., Falconer and Falconer 1980) numerous similar concentrations (as well as extremely low pH measurements) in clouds sampled at the Whiteface Mountain, New York, observatory. Recently Waldman et al. (1982) have reported nitrate and sulfate concentrations in Los Angeles fogs ranging up to and beyond 5000 um/liter. This compares with typical precipitation-borne concentrations of about 35 um/liter for the northeastern United States. Recently Lovett et al. (1982) have applied a simple impaction model to estimate fog-bound pollutant depo- sition to subalpine balsam fir forests and have concluded that chemical inputs via this mechanism exceed those by ordinary precipitation by 50-300 percent. This is undoubtedly an extreme case, and it would be more meaningful to possess a re tonal assessment indicating the general importance of fog deposition on an areal basis. This requires substantial effort however, involving climatological fogging analysis (cf. Court 1966) as well as numerous additional factors, and no really satisfactory evaluation of this type is currently available. Regardless of this it is appropriate to conclude that fog-deposition processes probably play an important, if secondary, role in pollutant delivery on a regional basis. In the future more effort should be addressed to this important research area. 3.1.6 Combined Processes and the Problem of Scavenging Calculations The preceding discussion of individual steps in the scavenging sequence has been presented intentionally on a highly visual and nonmathematical basis, with appropriate references given for the reader interested in more detailed pursuit. Despite the qualitative nature of this presentation, however, it should be obvious that the most

304 direct and expedient approach to model development is first to formulate mathematical expressions corresponding to each of these steps and then to combine them in some sort of a model framework that describes the composite process. This subject will be examined in greater detail in Section 3.4, which is addressed specifically to scavenging models. 3.2 STORM SYSTEMS AND STORM CLIMATOLOGY* 3.2.1 Introduction From the preceding discussion it is not difficult to imagine that scavenging rates and pathways will be dictated to a large extent by the basic nature of the particular storm causing the wet removal to occur. Storms containing water that is predominantly in the ice phase, for example, will provide little opportunity for attachment mechanisms associated with droplet nucleation, accretion, or phonetic processes. The abundance of liquid water and the temperature distribution in a given storm will have a direct bearing on the degree to which aqueous-phase chemistry can occur. Storms containing no ice phase whatsoever will be generally ineffective as generators of precipitation and thus will tend to inhibit the scavenging process. An interesting indication of the importance of storm type in this regard is presented in Figure C.3-23, which presents estimated scavenging efficiencies that vary extensively with storm classifi- cation. Different storm types differ profoundly with regard to inflow, internal mixing, vertical development, water-extraction efficiency, and cloud physics; and consequently it is appropriate at this point to consider briefly the major classes and climatologies of storm systems occurring over the continental United States. Two major points should be stressed at the outset to this discussion. The first is the essential fact that all storms are initiated by a cooling of air, which leads to a condensation process. Such cooling may occur by the *In the present text the term "storms is intended to denote any system in which precipitation occurs. This definition thus encompasses all occurrences, ranging from mild precipitation conditions up to and through the major and cataclysmic events.

305 transport of sensible heat, such as when a comparatively warm, moist air parcel flows over a cold land surface. The dominant cooling mode for most storm systems, however,\ is expansion, which occurs via vertical motion of the air parcel to elevations of lower pressure. The second note- worthy point in this context is that the overwhelming majority of storm systems is strongly associated with fronts between one or more air masses. The primary reason for this fact, of course, is that thermodynamic per- turbations and discontinuities associated with the frontal surfaces provide the opportunity for vertical motion (and thus expansion processes) to occur. This relationship is an essential component of storm- classification systems and will emerge repeatedly in the following discussion. Overlaps in the characteristics of different storm types render a strict classification largely impossible. For practical purposes, however, it is convenient to segregate mid-latitude continental storms into two classes, which are usually described as being "convective" and "frontal" in nature. These two major categories then can be subdivided further as deemed expedient for the purpose at hand, although it should be noted that sig- nificant overlap among storm types occurs even at this major level of classification. Frontal storms, for example, often possess significant convective character in their basic composition, and true convective storms often occur as the consequence of fronts. Because of this the following discussion will utilize storm classification primarily as a descriptive aid and will not belabor taxonomical detail beyond this rather pragmatic end. 3.2.2 Frontal-Storm Systems Much of what is understood today regarding mid-latitude frontal-storm systems stems from the pioneering work of the Norwegian meteorologist Bjerknes, who conducted a systematic survey of large numbers of storm systems and from this developed a conceptual model of frontal-storm development and behavior. Characterized schematically in Figure C.3-5, the Bjerknes model can be understood most easily by considering a cool northern airmass, separated from a warm southern air mass by an east-west front, as indicated in Figure C.3-5A. The progression of figures represents a typical result of the atmosphere's natural \ \

306 ~ I . ~ Van T 1 ' r

307 tendency to exchange heat from southern to northern latitudes. This is often referred to as a "tongue" of warm air intruding into the cold air mass. In the northern hemisphere this wave will tend to propagate in an easterly direction; thus the intrusion is bounded by two moving fronts--a warm front followed by a cold front, as shown in Figure C.3-5C. =, ~ ~ mow assoc~acea wits the wave system occur in a manner such that a depression in atmospheric pressure occurs at the vertex of the warm~air intrusion, and as a consequence a general counterclockwise or "cyclonic" circulation pattern emerges. Because of this feature Bjerknes's conceptual model is often referred to as the "Bjerknes cyclone theory, n and frontal storms associated with this pattern are termed "cyclonic" storms. A typical feature of storms of this type is the tendency for the cold front to overtake the warm front and, ultimately, to annihilate the wave. The "occluded" front created as a consequence of this behavior is shown schematically in Figure C.3-5D. In view of this birth-death sequence of the Bjerknes cyclone model, the progression depicted in Figure C.3-5 often has been termed the "life history" of a cyclone. Some idea of spatial scale and the general cyclonic flow pattern of a mature cyclone are given in Figure C.3-6. In viewing these indicated flow patterns r however, the reader should note carefully that con- siderable vertical structure exists in such systems, and marked deviations of the wind field with elevation are typical. In particular, one should take care not to confuse the indicated general circulation patterns with corresponding surface winds. Although created from the limited observational base available during the early twentieth century, the fundamental precepts of the Bjerknes theory have proven valid even as more sophisticated observational and analytical facilities have become available. Certainly nonidealities and deviations from this model occur; but its general concepts have proven to be immensely valuable as a conceptual basis and as an idealized standard for the assessment of actual storm systems. Comprehensive descriptive and theoretical material pertaining to such systems is available in the classic text by Godske et al. (1957), and more elaborate and modern extensions are given in the periodical literature (e.g., Browning et al. 1973, Hobbs 1978).

308 . ~ U. 3 o o ~ 3 c) ._ (, o Cal C> Cal a' _4 C> ¢, ¢) Q. Cal .= o G) ._ ._ ._ C) ._ an Cal o - C~ no v V o to I: Cal Cal sr Cal o ._ - ._ C) cd a' C) o

309 Warm~Front Storms It is important to note that the plan views exhibited by Figure C.3-6 are gross simplications, since they do nothing to characterize the three-dimensional nature of the cyclonic system. If one were to construct a vertical cross section of the warm front (A-A' in Figure C.3-6), then typically one would observe an inclined frontal surface as shown in Figure C.3-7.* In this situation the presence of warm air aloft creates a relatively stable environment, which inhibits vertical mixing of air between the two air masses. The warm, moist air moves up over the cold air wedge, expanding, cooling, and ultimately forming clouds and precipitation. _ . . . Typically the warm air supplying moisture tor this purpose has been advec ted from deep within the southern air mass, carrying water vapor and pollutant over extensive distances. This trans- port trajectory has been aptly compared to a "conveyor belt" for moisture by Browning and his co-workers (1973). It is appropriate to note that this moisture conveyor belt is a conveyor belt for pollution as well. Warm-front storms often are associated with long periods of continuous precipitation, although significant structure can exist within such systems. An important structural element in this regard is the occurrence of prefrontal rain bonds, which take the form of concen trated areas of precipitation imbedded within the major storm system. At present the factors contributing to rain-band formation are not totally understood, although mechanisms such as seeding from aloft by ice crystals and nonlinearities of the associated thermodynamic and flow processes undoubtedly contribute to a major extent. Waterfront storms usually can be expected to be rather effective as scavengers of pollution originating from within the warm air mass, especially if temperatures in the feeder region are sufficiently high to allow the presence of liquid water and the nucleation-accretion process. Scavenging of pollutants from the underlying cold air mass usually will be less effective, owing to the relative scarcity of clouds and generally less definitive flows in this sector. Scavenging in both regions will of course depend on the physiochemical nature of the pollutant of interest and the microphysical attributes of the cloud system in general. Methods for *See Table C.3-1 for definition of cloud abbreviations.

310 1 1 1 ~ ' 1 . _~ il~r _ ~ 1~1 _ ~ 1- _ ~ 1~ l _ ~ 1 _ ITS ._: ~ _ ~ Lf , _ ~ HAT it CHIT de J) it _ ~ ~ .- ~ ~ _ ~ t · ~ 1 t I -~cxc I ~ ACHE . -a-... ~ ~ ~1 ~ . ~ ~ * it TIC TIC ~ ~ ARC * ~ XC_ ; q -, . IC TIC ~ ** ~ \~C XC' *~C c 1 - tp~c~C~C c l~c ~c, `,4~ ~ [ l . _ , . ~ : . : : ~ cz I V' w ~ ~S  J L~ C] ~Z C ~ C ~ C~ Z C~ o , Z ~ , o L~ CY ~S LL o C~ ~° ~LLI o U' - C~ o - ~: C} o o o J > L~ ~: - C~ - . Ct o C: ~ l== C~ ¢  3 C~ C~ ·~ cd o o · - c) v) c~ c~ ~o o - ~ cy L~ ~2 ~ CO _ o _ cO C_) > Z =) _ ~ <: ... o _ r t ~

311 TABLE C.3-1 Summary of Cloud Types Appearing in Figures C.3-7-C.3-9 Type Cirrus Cirrostratus Cirrocumulus Altostratus Atlocumulus Stratus Stratocumulus Nimbostratus Cumulus Cumulonimbus Abbreviation ~_ - c Cs Cc As Ac St Sc Ns Cu Cb estimating scavenging rates in such circumstances are discussed in Section 3.4. Cold-Front Storms A typical vertical cross section (B-B' in Figure C.3-6) of a cold-front storm is shown in Figure C.3-8. This differs substantially from the warm-front situation in the sense that, instead of flowing over the frontal surface, the warm air is forced ahead by the moving cold air mass. This action produces a more steeply inclined frontal surface, which, combined with the presence of low-elevation warm air, creates a relatively unstable situation leading to convective uplifting and the formation of clouds and precipitation. Although discussed here in a frontal-storm context, this precold-front situation composes an important class of convective storms, which will be discussed in some detail later. Scavenging rates and efficiencies associated with such storm systems will again depend on the pollutant and the physical attributes of the particular cloud system involved.

312 ~1 1 . 1 1 · 1 1 L\ ~ ~ u - l / 1 1 ~ . ~ l 1 1 , t := oo l I,) ~1 Rut, 8 Lo At in / 1 1 1' ~1 1 1 ~ om ~ - of Do t a: Far

313 Occluded-Front Storms Owing to the fact that occluded fronts are formed via merger of warm and cold fronts, it seems reasonable to expect that storms associated with occlusions should share characteristics of the respective elementary systems. Figure C.3-9, which shows a typical vertical cross section (Section C-C' on Figure C.3-6) of an occluded system demonstrates this point. Typically the Awry resow or warm air alort maintains a relatively stable environment to the east of the occlusion, and clouds and precipitation occur in this region largely as a consequence of ascending flow from the south. Much more detailed accounts of occluded systems can be found in standard references such as the book by Godake et al. (1957). _ _ _, _ _, ~ . _ ~ . _ . . . . 3.2.3 Convective-Storm Systems An idealized cross section of a typical convective storm is shown in Figure C.3-10. Such storms depend on atmospheric instabilities to induce the necessary vertical motions and concurrent cooling and condensation processes; and as such they are most likely to occur under warm, moist conditions where the eneraetics Are most conducive to this process. Often convective storm systems occur as "clusters" of cells such as that shown in Figure C.3-10 and exhibit a marked tendency to exchange moisture and pollutant between cells; thus the flow dynamics and ~uPnain~ Bra - for; a~;~. ~ =..~h ~, . . ~ ~- -~ ^ ^ ~-~ ~ ~= ~ ~ 1 yes Rena co oe extremely complex. Typically the moisture and pollutant input to a ~ . . . . Nave cell occurs primarily through the storm's updraft region (cf. Figure C.3-10), although entrainment ~~^ ~~ ~~ Dynamics of this Prom upper regions Is possible as well. process are such that violent updraft velocities often occur; these are capable of lifting "ntrAin-~ Fir wager vapor, and pollution to extremely high elevations (sometimes breaching the stratosphere). Along this course entrained pollutant is subjected to a large variety of environments and scavenging mechanisms; as will be noted in Section 3.4, convective storms tend to be highly effective scavengers of air pollution. As was stated earlier, convective storms often are associated with frontal systems, although frontal influence is not absolutely necessary for their presence ~_ .~= ~& ~& ~"~eL A W" "=L .

314 ~ - f ~ ~ ;/1! ~1 1 / t w' _ ,/ t~ ~ _ ~, ~ ~ ~ ~ ~ ~ r i ~ ~ ~ / ~ ~ ~ r _ I ~1 ~1 1 ~ 1 _ _ ~ _ r _ _ ~ ~C~J O - ~C to to CY ° ~ ° ~ <s ~ 8 On 80 I ~ I ~ ~ ~ ~I i8 (8 z o ~o - Lo a _ ~ z - , - z _ ~ .c - Z _ Z _ _ _ _ J or o ._ U: o C) c) o ._

15000 m 10000 m 315 ANVIL OUTFLOW REGION A ~ x- * _4 ok * _~ i, ~*,*. ~ .. ~ ·,'-,,: : · ·:~` OF ~ W: i'' ~ '~/ _ -60C v ~ ~ * . . . ~OUTF ~PRIMARY INFLOW REGION 25C OC 30C FIGURE C.3-10 Idealized cross section of an isolated convective storm. An isolated air mass, for example, is totally capable of acquiring sufficient energy and water vapor to induce a convective disturbance on its own accord. Perturbations arising from fronts, however, often contribute to the creation of convective activity--if for no other reason than supplying a n trigger" to initiate convection in a conditionally unstable atmosphere. 3.2.4 Additional Storm Types: Nonideal Frontal Storms, Orographic Storms, and Lake-Effect Storms As noted previously, the Bjerknes cyclone model represents something of an idealized concept, and numerous features can contribute to deviations from this "textbook" behavior. Orographic effects are highly important in this regard. Considering a cyclonic disturbance approaching the North American continent from across the

316 Pacific Ocean, for example, the frontal patterns typically lose much of their original identity after impacting with the mountainous regions of the west. In addition to the physical distortion of flow patterns, the lifting induced by the terrain encourages further precipitation, result- ing in large spatial variability in rainfall patterns and pronounced local phenomena such as n rain shadows and chinooks. Precipitationrformation and precipitation- scavenging processes associated with such systems tend to be highly complex. Frontal systems often tend to reconstitute their structure after crossing the Rocky Mountains; but continental effects still impart a marked impact on their basic makeup. In the midwest-northeast region, for example, there is a tendency for the fronts to orient themselves in an east-west direction and become stationary for extended periods, often punctuated by several minor low-pressure areas. Even under relatively ideal conditions continental frontal storms tend to possess more convective flavor in their basic makeup than do their oceanic counterparts. As indicated above, terrain-induced or "orographic" effects are usually most important in augmenting major storm systems, although relatively isolated orographic storms (such as oceanic n island-induced" storms) certainly do occur. Orographic effects obviously will tend to be most pronounced in regions where radical terrain changes occur; but even the small elevation changes typical of the Midwest can contribute signifi cantly at times. Orographic effects also are suspected to influence storm behavior over substantial downwind distances. Lee waves from the Rocky Mountains, for example, have been suggested to trigger thunderstorm formation at extended distances. Lake-effect storms are yet another example of a somewhat nonideal phenomenon that often is superimposed with more major meteorological patterns. Typically such storms occur during fall and early winter periods when land surfaces tend to be cooler than their adjoining water bodies. Considering an air parcel moving on an easterly course across Lake Michigan, for example, the warm lake surface tends to supply both heat and water vapor as it proceeds. As this parcel is advec ted across the downwind shore, however, two important things will occur. Firstly, the cold land mass will act to extract the heat from the air, and secondly the orographic lifting (of the order of a few tens of meters) will result in ascent, expansion, and further cooling. The

317 net result is a lake-effect storm. Such storms are capable of inducing highly variable precipitation patterns in specific areas around the Great Lakes region. Although confined largely to this portion of the United States, these storms are accountable for a majority of the snow- fall accumulated in specific cities, such as Muskegon, Michigan, and Buffalo, New York. Some appreciation for the magnitude of this effect can be gained by looking at the climatological precipitation map given in Figure C.3-11. 3.2.5 Storm and Precipitation Climatology The subject of storm climatology is exceedingly complex and will be discussed here only to the point necessary to describe some key attributes and indicate references for more detailed pursuit. Factors especially important in the context of precipitation scavenging are temporal and spatial precipitation patterns, storm~trajectory behavior, and storm-duration statistics. These will be discussed in order in the following paragraphs. Precipitation Climatology Figure C.3-12 provides climatological averages of monthly precipitation amounts at various stations throughout the United States. This figure was taken directly from the Climate Atlas of the United States (1968) and requires little elaboration at this point. It is interesting to note, however, that precipitation amounts do not~vary radically throughout the year at most northeastern U.S. stations; this contrasts especially with the western and arid stations, whose seasonal variabilities tend to be pronounced. It should be noted as well that actual precipitation amounts for a given single month can vary appreciably from the climatological averages presented here. Storm Tracks Because of the difficulties noted previously with regard to precise classification or definition of storms, a really concise climatological summary of storm-pathway behavior is largely impossible. Some useful information can be generated, however, by observing the tracks of the

318 cyclonic (low-pressure) centers associated with major storm systems. Klein (1958), for example, has conducted a systematic survey of cyclonic centers in the northern hemisphere and from this has constructed monthly climato- logical maps of low-pressure tracks. Figure C.3-13, taken from the book by Haurwitz and Austin (1944), presents the combined results of the analyses by several previous authors. On the basis of the previous discus- sion it should be re-emphasized that, owing to the complex flow processes associated with cyclonic systems, one should not interpret the motion of these low-pressure centers as being identical with feeder trajectories for the storms themselves. Careful and skilled meteoro- logical guidance is mandatory for the successful interpretation of such information in the context of source-receptor analyses. Several additional points should be emphasized in the context of Figure C.3-13. Firstly it should be noted that this presents a long-term composite average and that marked deviations from this pattern can be expected to occur with season. Secondly the statistical variability of storm tracks is such that substantial departures from the long-term averages can be expected for any particular year. Finally, there is substantial evidence for longer- term shifts in average stormrtrack distributions (Zishka and Smith 1980); thus presentations (such as Figure C.3-13), which are based on historical data may vary considerably from storm patterns to be observed over the next 20 years. The implications of this with regard to long-term acid-deposition forecasting are obvious. Additional features of cyclonic storm climatology can be found in standard climatological textbooks (e.g., Haurwitz and Austin 1944). Convective-storm climatology, which tends to be much more region-specific, can be evaluated from such references as well, although more recent weather modification programs such as METROMEX, HERE, and HIPLEX have generated a considerable amount of new information in this area. StormrDuration Statistics In the preparation of regional scavenging models it often is desirable to create some sort of statistical average of storm characteristics so that ~average" wet-removal behavior can be defined. Although little activity has been devoted to this area until very recently, the

319 1' GREEN BAY MIL] 8U ~ 90 100 \ 1 1 no _;~ ~ - .. .S~ ~ ! r. ~^ '60 (LUDI NGTON~ 50 L_T UTH BEND ~I Or FIGURE C.3-11 Average annual snowfall pattern (inches) over Lake Michigan and environs. Adapted from Changnon (1968~.

320 NORMAL MONTHLY TOTAL j Espy city 1 a! ~ '-2 "\~` ,, I j a. .= ~ 1.. I ~ ~1.~ . / ~°rd it!/ - ~iDenver '[lo ~ I' ~.! `,,/,1.~!1!. i, _ ~/j ~`-~_ _ r ~4!! / ~r ~- ~ | ~0 North Platte . _. -... 1 W Browna_le '] , - 1-1 . ~_ _ . _ _ . .

321 PRECIPITATION (Inches) FIGURE C.3-12 Climatological summary of U.S. precipitation. From U.S. Climato- logical Atlas.

322 ad\ - ?\\N ~< ~ \ \~4 \ \ \ ~ ~,:~ 5t,;~f 1~`,~ ~ ~ - _ ;_g _ - FIGURE C.3- 13 Major climatological storm tracks for the North Amencan Continent. Adapted from Haurwitz and Austin (1944~. Dashed line denote tropical cyclone centers, and solid lines denote those of extratropical cyclones. usefulness of such an approach to regional model develop- ment suggests accelerated effort during future years. The analysis by Thorp and Scott (1982) provides an example of one such effort. These authors compiled data from hourly precipitation records from northeastern U.S. stations to obtain seasonally stratified duration statis- tics, which were expressed in terms of probability plots as shown in Figure C.3-14. As can be noted from these plots, "average" storm durations during summertime are significantly less than their wintertime counterparts, reflecting relative influences of short-term convective behavior. Some of the references given in Section 3.4 suggest potential modeling applications for these statistical summaries.

323 uo!lel!d!oaJd leuo!6a~ lelo1 paz!lewJoN ~o uo!~oeJ~ aA!lelnwn~ o ~ ~ _ o o ~, o o o o ,] ' Z ~n ~ ll $\ 1! c - C C o . Q `° .> ~ ~ E o o i CO t., E 'c  o cn ~ o o E E _ _ \ \ ; \. \ \ - o - o c C o o _ ·- {t, ~ ._ ~ .Q _ ~ .> Q ~ C E .-° c c C .C ._ 0, oo o - o t1) _ C} CC o: - Z~ ~ E E c o - ~ c ~ . ° ~ _ ._ C.l - ~ CL C, _ aC ~ ,o ~o a, , ~ . _ -o  ~, o o ~D C: ° C - o - ~o ~AO~ \ ~ o uo!lel!d!oaJd leuo!6a~ wnS pueJ~ ~o uo!loeJ~ e se sssel~ Uo!leJna W~°S Jad U°!lel!d!~8Jd o ._ _ :> .. - ._ C~ ~ C' C> C~ ~ ~ g ~ C> Ct ~ O O o U. ~ 3 ~ o ~t C~, ._ - o Ct Ct ~  4_ ·C~ ) o ~ · - o~ ~ a~ ~ ~ o cd (v o 4- C{2 o p~ c: <; o ~ · - c) 'e ~ :> o ·~ cd ct: 3 ^ o ~ C~ G~) ·- ¢ ° {d C-_~ ~ ~ o _ : - ~ ^ o ~ C~ ~ - , o _' 3, CO C) ~ ~ 3 U, ~ ~ ·~ ~ C> Ce V~ ~D C~ . o >,~ C) . s o ~, ~ . .~, . C) ~ o ~ C~ ~ o ~ ~ .~ Z~ 3 ~ ~_ CO _ C~ o ~ o ~

324 3 . 3 SUMMARY OF PRECIPITATION-SCAVENGING FIEL D I NVESTIGATI ONS For the purposes of this document n field investigations" of precipitation-scavenging mechanisms will be differ- entiated from routine precipitation-chemistry network measurements, which are intended primarily for charac- terization purposes. There is of course a great deal of overlap between these two classes of measurements, and significant reciprocal benefit is generated as a conse- quence of each. There are some essential differences between the two, however, and it is convenient for present purposes to differentiate them accordingly. The primary distinguishing feature of a scavenging field investigation is that the study usually is designed around the basis of some sort of conceptual or interpre- tive model(s) of scavenging behavior, which is tested on the basis of the field data. If the model predictions and data disagree, then some basic precepts of the model must be invalid, and additional mechanistic insights must be generated to rectify the situation. In the event that predictions and data agree, then this may be taken as evidence that the precepts may be correct. Regardless of whether positive or negative results are obtained (and assuming that the field study has been well designed and well interpreted), an advance in understanding has been achieved. The importance of such input cannot be over- emphasized. Examples exist wherein field investigations have demonstrated then-accepted models to be in error by several orders of magnitude (e.g., Hales et al. 1971). Field studies have been essential in keeping the models "honest. n Field studies of precipitation scavenging began in earnest during the early 1950's to gain an understanding of radioactive fallout. Pioneering studies in this area were performed in England by Chamberlain (1953), which pertained to radioactive pollutant releases from point sources in anticipation of reactor accidents and related phenomena. These constituted the basis for the washout- coefficient approach to scavenging modeling (see Section 3.4). Other studies focused primarily on nuclear- detonation fallout and thus approached the scavenging problem from a more global point of view. Following the English lead, nuclear-oriented studies were conducted by the United States, Canada, and the Soviet Union. These included studies of tracers as well as those of the radionuclides themselves, and although

325 some of this material still remains in the classified literature, it may be stated with certainty that most of what we know today regarding scavenging processes has been generated as a consequence of the nuclear era. The review "Scavenging in Perspective" by Fuquay (1970) presents a comprehensive account of this early stage of scavenging field studies. During the late 1960s field-experiment emphasis shifted to more conventional pollutants, with the general recognition of precipitation scavenging's importance in preserving atmospheric quality and its potential adverse impacts of deposition on the Earth's surface ecosystem. Since that time a variety of large and small field studies have been conducted. These are summarized in Table C.3-2, which provides a logical classification in terms of source type, pollutant type, and geographical scale. Although field studies have been focused strongly on quantitative aspects of precipitation scavenging, they have provided important qualitative information regarding acid precipitation processes as well. The ensemble of studies listed in Table C.3-2 presents a rather cohesive base of evidence in this regard; and although some conflicting results and uncertainties do exist, a generally coherent picture can be constructed in several important areas. Although there is considerable overlap of source-receptor distance scales among these studies, they tend to group rather conveniently into three classes of areal extent: 0-20 km, 0-200 km, and 0-2000 km. These classes shall be termed loosely as "local, n "intermediate, and ~regional" scales in the following discussion, where key qualitative features are illustrated by considering the fate of specific acidic- precipitation precursors (SOk, NOk, and HC1) as they are transported over these increasing scales of time and distance. On a local scale (0-20 km) field studies have generally demonstrated the precipitation scavenging of sulfur and nitrogen oxides from conventional utility and smelting sources to be minimal. -~ ~ The virtual absence of excess nitrate or nitrite ion in precipitation samples collected beneath such plumes (Dana et al. 1976) provides strong evidence that direct uptake of primary nitric oxide and nitrogen dioxide by precipitation and cloud elements is a negligibly slow process. Nonreactive scavenging of plumeborne sulfur dioxide is solubility dependent and tends also to be a rather

326 U] o ·rl .,. U] H · - · - V U) o . - ·rl .,' ~U o U] a ·,' CQ U) a' v U) a E~  a, U aJ U. E~ l O C-) V] E~ o 4J ~ r ~ _ ~_ a ~ C u ~a ~a ~ ~ r ~> _ ~ ~ a~- o ~c _o~ C.~ ~Cr ~. , -_ ~._ ,~ ~ _ ~vo,- _ 431 - C ~C7\ ~C _~ ~. ~. a ~^ ~ c ~-v ~ _ ~ ~- ~a ~~ r ~ a, -I ~· ~ ^ · ~ ~ C ~ ~V · ~_ U) ,~, _~_ ~V _ ~ ~_ ~ C ~ ~ ~_ ~ ~a ~ ~ c o ~ 0 C ~ ~ 0 ~- ~ 0 _~ V· ~ 0 U~ 0 ~ ~ ~ ~ ._ _ ~C ~ Ul ~ CO Q) _ ~ ~ JJ ~ C - ~ ' u, o' JJ ~a C C ~ U] C ~ ^ ~ _ ~U] _ ~ ~ c c V~ ~ s ~ · ~ ~ ° ~.- C E E _ ~ v ~ a~ ~t ~ t ~0 ~15 a~ rL: ~tr .^ 4) ~03 ~ ~ 3 ~ a ~s ~ ~- 3 ~ a) ~ ~ c :e -v ~u: ~ I ~a ~ ~a' c: ~ a, ~- ' ~r~ C Cns :~: - ~ _ ~ C ~ ~_ 0 ~ c ~o a~ ~u ~~ {,q ~ Y ~ ~ t- U] o~ ~_ O ^ ^^ ^ ~ ~_m v ~ ~ ~ v ~ ~ eo -~ ~ ·~ -I aJ ~ ~ ~ ~· ti~ U~ [_ d~ ~ t~ _ ~_ - a) ~c ~ a~ a,o' a ~r~ ~ ~ . ~·- ~.- ~ ~ ~a'~ ~ ~ . y ~ ~ _ . ~ ~ ~_ . - ~ _ --_ ~10 _ _ ~ ~ _ ~ ~ C · - - ~3 _~ (~ Il5 ~ ~ ~ ~ ~ ^ ~· ~V c c. . a' ~ p: I_ V 0` ~ ~ ~oo ~v O? ~ ~ ~·~ a, s ~ a~ ~ ~v - ~ OD =, ~U] ,<~ ~ , ~1C,- ~ 3, ~ ,- _ ~ C ~ o _ _ C _ E C n, ~_ v u, a) i,, ~V V ' ~ 0 e, O ~ `£) ~O ~ _ ~ (1) kD ~ ~_ i,~ C N ~- ~I C ~4 ~O ttS ' - ~ ~ ~ ~ ~ ~ V ~ ~V ~ ~ V ~ E E ~ nl L' ~ C ~ ~ ~ ~ ~ - ~ ~ ~v ~v ,~ ro C c a' ~ ~ ~ t: - ~ ~ ~ - Ll ~ ~ ~ O U) ~3 E S S ~ n' ~J ~. S 3 0 05 0 C) ~t ~O ~V <~ ~a a ~m ~ ~m a: ~ x u, ~z u, ~s V a, ~ v ~ U) V ~L) 1 ~C O _i la 4. ~C) ~1 C ~ ~ ~ ~ ·- O I - 1 ~O · - E s ~ Cr o Ul v U) C) 3 ~C `0 ·.4 3 ~ V dd C Q, 0) U) O ' - E v rU ~ O O C · - ~·~. ~5 . - r~ ~ V C ~ ~ ~ · - a~ ~ ~n O ~ ~u U] .- C) ~S  V' V O O · O V U] ~ C V C ~ ~I Q, 3 ~ C ~ ~U] O ' - O O O C ~ C .,. o O ~0 0 0 E ~Q C ~ ~ ~ ·,' `ts ~ ~ O ~ 3 U) U) U) ~ ~tt V ~ V O ~- V C u~ U) u) u~ u u) ~ ~ ~0~ O) ~· (~) 3 q~ 3 · U~ L4 ~0) ~ 0~ G ~0~ (V )- (0 ~0) 0~ 0 ~u~ V 0) O -1 0~ ~l u) ° G ~O r ~ 03 u ~03 u ~o 00 E ~ O ~ ~ 3 £ ' ~ 1 ~ . ~ ~ ~ ,~ q' ~tt ~ ~, ~V 3 0) ~ ~ Cq U, ~ ~ 0 ~ G, . - ~ a~ a~ v ~ 41) td ~C ~ ~ ~ V ~ a~ ~ Y ~ ~ ~ ~Q , ~ ~1 ~ ~ V~ ' ~ ~ ~ C - - - ~E (~ ~ a~ ~, a~ ~u ~ u~ ~V k4 a~ ~ ~ ~ ~ ~ ~ ~ ~ ~ c 0 0 1 ~ ~ V v ~ ~ O C C O C V C V ~ ~ ~ ~ ~ ~, ~V ~ Ll ~ V ~ ~ ~ O V O U) O U] O a, a, ~ ~ ~ ~ ~ ~ ~ ~ ~ ~y ~ ~ . - C . - ~ · - E 3 3 tt v 3 3 3 3 ~3~ ~ . ~ m.- O O O o 0 0~ ·- 0 ~ ~ 0 ~0 3 P4U) ~0: ~ cn ~ o ~:~; a, O 3 ~,~, O ~ U] U) C a) C U] ~ ~O O V ~.,' C . - ~ O ~ o 3 ~u, L, v O ~U] QJ U) C C ~C ~ 3 to . - 3 ~ u, c tc o V ~O ~ ~ 3 c o u, ~ o o C v E C " ~` ~a' O n O c c 0 n `: u2 0 u, ° ~a' _ : : ~ D

327 inefficient process, although it is definitely detectable in field studies conducted in relatively clean environs meets (Dana et al. 1973, 1976; Hales et al. 1973). This phenomenon, which is suppressed under conditions involv- ing high rain acidity, is relatively well understood at present (Drewes and Hales, 1982). Nonreactive scavenging of sulfate aerosol can be an efficient removal process. The preponderance of relevant field tests of Table C.3-2, however, has demonstrated that wet deposition of sulfate from local power-plant and smelter plumes occurs rather slowly. This is undoubtedly a consequence of the small amounts of primary sulfate available for scavenging under such circumstances. Field tests conducted under situations wherein sulfur trioxide was intentionally injected into the stack of a coal-fired power plant (Dana and Glover 1975) show correspondingly high sulfate scavenging rates, and it has been suggested that under certain operating conditions some types of power plants (especially oil-fired units) will produce sufficient primary sulfate to account for appreciable local deposition. To date, however, there has been no really strong field evidence in support of this point. Hogstrom et al. (1974) reported the observation of substantial sulfate scavenging from the local plume of an oil-fired power plant in Sweden, but these results are rather dependent on the interpretation of background contributions. Granat and Soderlund (1975) performed a similar investigation in the vicinity of a second Swedish oil-fired plant and found a comparatively small scavenging rate. Reactive scavenging of plumeborne sulfur dioxide to form rainborne sulfate is difficult to differentiate from primary sulfate removal. The previously noted findings of low excess sulfate in below-plume rain samples, however, suggests that neither process is particularly effective in near-source plume depletion. The scavenging of hydrochloric acid to produce chloride and hydrogen ions in precipitation will most certainly be a highly effective process, depending on the quantities of hydrochloric acid available. Considerable theoretical and laboratory work has been conducted in this area for Space Shuttle impact assessment, and there are limited data suggesting that hydrogen chloride is scavenged in measurable amounts from power-plant plumes (Dana et al. 1982). With the exception of studies conducted under rather clean ambient conditions (e.g., Dana et al. 1973, 1976)

328 the influence of background contributions has made the interpretation of plume scavenging a difficult task. Typically the sulfate and nitrate concentrations in precipitation collected adjacent to the plume are quite variable, and subtracting this influence to determine source contributions involves substantial levels of uncertainty. ~ the local scale is compounded appreciably as greater Phi c Hi ffi~,,1 Ev of "worse attribution" at scales of time and distance are considered. On a more intermediate scale (0-200 km) an enhancement of sulfate and nitrate precipitation scavenging seems co occur, presumably because the precursors have had more opportunity to dilute and to react under these circuit stances. Hogstrom (1974) reported substantial scavenging rates of sulfur compounds using an extended network of samplers in the vicinity Of Uppsala, Sweden. Hales and Dana (1979a,b) observed summertime convective storms to remove appreciable fractions of urban NOk and SOx burdens in the vicinity of St. Louis, Missouri. Although both of these studies were subject to the usual uncertainties with regard to background contributions, there is little doubt about their general conclusions of significant scavenging under such circumstances. On a regional scale (0-2000 km) there are relatively few data from intensive field experiments. Precipitation- chemistry network data are of some utility in this regard, however, and several analyses have applied these measurements to specific ends. One result of these analyses is the suggestion that, in the northeastern quadrant of the United States, roughly one third of the emitted NOX and SOX is removed by wet processes (Galloway and Whelpdale 1980). Network data for the northeast (MAP3S/RAINE, 1982) show also that the molar wet delivery rates of NOk and S°k are roughly equivalent. Combining this result with regional emission inventories suggests that nitrogen compounds begin to wet deposit with a significantly enhanced efficiency as distance scales become regional in extent. The above changes in behavior with increasing scale seem to be a logical consequence of current understanding regarding the atmospheric chemistry of S°k and NOk. On local scales neither is scavenged very effectively owing to the chemical makeup of the primary emissions. On intermediate scales both groups have had some oppor- tunity to react into more readily scavengable substances. Depending on ambient conditions, the nitrogen oxides will have participated to some extent in initial photolysis

329 reactions and proceeded on to form scavengable products such as nitric acid, peroxyacetyl nitrate, and nitrate aerosol. Sulfur dioxide also will have reacted homo- geneously to a limited extent; more importantly, however, this compound will have diluted to levels where limited reactants (and possibly catalysts) will facilitate its oxidation in the aqueous phase. On a regional scale this progression continues with the relative acceleration of NOx scavenging. Present field-study indications that NOX scavenging may occur primarily through the attachment of gas-phase reaction products, while the scavenging of S°k may depend much more heavily on aqueous-phase oxidation processes are also reflected in precipitation-chemistry data. A possible consequence of this difference in mechanisms is illustrated in Figure C.3-15, which is a time series of daily precipitation-chemistry measurements for a northeastern U.S. site. The decidedly periodic* behavior of sulfate-ion concentrations in contrast to the largely disorganized behavior of nitrate-ion concentra- tions has been suggested to occur as a consequence of an aqueous-phase oxidation of sulfur dioxide, which proceeds more rapidly during summer months. Whatever the cause, it is readily apparent from this figure that scavenging mechanisms for these two species differ appreciably. *One should note in Figure C.3-16 that the periodic functions are fit to the total data, whereas the linear regressions are fit only for the period January 1, 1977-December 31, 1979; thus the cyclic functions are not exactly symmetric about the linear regression curves. Some idea of statistical improvement in fit may be obtained using the expression ~2 ^2 linear regression - periodic fit r = - 2 a linear regression where the o2's pertain to variances of the data points over the three and one-half period. For sulfate in Figure C.3-16, r2 equals 0.22, indicating a significant reduction in variance; the corresponding r2 value for nitrate is 0.01, suggesting that no significant annual periodicity exists in this case.

330 uo!lel!d!oa~d ieU0!698 elo1 paz!lewJoN ~o uo!loe~ aA!lelnwn~ o o o 1 1 1 1 1 1 o c ~c,o . ~ .- a' Q .' Q .1 a~al !, ~ ~ ~1 ·1 1` i'' \ 1 1 l - z Z cu C) - 1 1 1 1 o o o O a, c0 o o o o o o . C) - - ~ o _ Q O ~ ,~, a c _ 0 _ 0 1 \ \ I \ c cr c c ~ 0 Q  ,> _ ~ C O C - ~ ~: ~ O 1 ~voo 1 ° ~ ~ ~ 1 O O ~ - ' ' Z G C} ~- E ~ ~ E _ _ ct ~ ~1 oo 'm { r - \ - \ `\ ~ / \ '\N `\: \ o o o . O ~0 UOD ~- \ - O C - - u) O W.\ O O o o uo!lel!d!oaJd JawwnS leuo!6a~ 1elol ~o uo!loe~ e se ssel~ uo!lelna wJols Jed Uo!~el!d!o9Jd 0 ~: O ~ ~ O ce ¢) ~ c,~ <, 3 C~ C) ._ ~ C~ _4 ~ ~ ~ . o ~} o ~ C~ >., o L~ cd a.>  . .= E~ c) : c3a, o o v: ·oC ~ - ~ 'e o o o . - c~ £ C~ o Ct £ o C~ ._ ·~_ C~ ~: 5 £ o . £ ,= o C~ s~ ~ ~ o .= o Ct ^ P ~ C~ Ct C5 C~ ~ ~: ~ - £ c, ~ o - o . - ._ ~ C~ ,_ .= 5 .~' . O ._ 4_ ~ .~ O ._ ~ ~ ._ ¢o 3 ~ - :, _ ~ o ~ o ~ · - ~V .> ~ _ ~ Ct C) o o

331 As noted a~Ve,~-most past field experiments have experienced difficulty in resolving precisely which source(s) of pollution has been responsible for material wet-deposited at sampled receptor sites, and this problem is typically amplified as time and distance scales increase. Source attribution is particularly uncertain on a regional scale, and the basic data obtainable from standard precipitation-chemistry networks are of limited help in this regard. Combined with the lack of data from well-designed regional field studied, this aspect poses one of the most important and uncertain questions facing the acid deposition issue at present. As a consequence of this need, a major regional field experiment has recently been designed and conducted in the northeastern United States (Easter 1982, MAP3S/RAINE 1981). Known as the Oxidation and Scavenging Character- istics of April Rains (OSCAR) study, this field experi- ment was based on the concept of characterizing, as completely as possible, the dynamical and chemical features of major cyclonic storm systems as they traverse the continent. Specific objectives were as follows: To assess spatial and temporal variability of precipitation chemistry in cyclonic storm systems and to test the adequacy of existing networks to characterize this variability; ~ . ~ . . and to provide a comprehensive, high-resolution data base for prognostic, regional deposition-model development; 3. To develop increased understanding of the transport, dynamical, and physicochemical mechanisms that combine to make up the composite wet-removal process and to identify source areas responsible for deposition at receptor sites. The data collected and assembled by the OSCAR project are summarized in Table C.3-3. These are being made avail- able to the general user community in a computerized data base. A general layout of the OSCAR precipitation chemistry network is shown in Figure C.3-16. The points and triangles on this map represent locations of sequential precipitation-chemistry stations on an n intermediate- density" network, and the open square, overlapping Indiana and Ohio, depicts a concentrated network of 47 additional sites. Specific design criteria for this

332 TABLE C.3-3 Base Summary of Data Collected for the OSCAR Data Meteorolog ical Data . . North American standard 12-h upper-air observations ( r awinsondes) OSCAR special rawinsonde data North Amer ican 3-h standard surf ace observations North American hourly precipitation amount data Trajectory forecast data (Limited Fine Mesh and Global Spectral Models) Gr idded forecast data (Limited Fine Mesh Model) Satellite observations Precipitation-Chemistry Data OSCAR network: Sequential measurements of rainfall, field pH, laboratory pH, conductivity, NOT, NO2, SO4, SOT, C1-, NHt, Ca++, Mg++, 1~, Na+, Al ++, PO{, total P b Additional networks: Time-averaged data as available from sources such as NADP, CANSAP, CCIW, and APN Special rainborne H2O2 measurements Aircraf t Data Trace gases: O3, NO/NOx, SO2, HNO3, NH3 Aerosol parameters: scattering coefficient (bScat), Aitken nuclei, aerosol sulfur, sulfate size distribution, aerosol size distribution, aerosol acidity Cloud water chemistry: NO-3, NO-2, SO4, Sit, pH, NET, conductivity, C1-, Ca++, Mg++, K+, Na+, total Pb Meteorological parameters: Temperature, humidity, liquid-water content, wind speed and direction, cloud droplet size di str ibution Position parameters: Latitude, longitude, altitude, time Surface Air Chemistry Data Emissions . OSCAR SAC site (Fort Wayne 40° 49.8' N. 85° 27.6' W): H202, peroxyacetyl nitrate, sulfur aerosol size distribution, NH3, SO2, SOT, O3, NO/NOx, HNO3, aerosol composition versus particle size, aerosol acidity Selected air -uality data from specific surface monitoring sites throughout eastern North America MAP3S/RAINE standard inventory

333 ~ __ 14~\ ,- ,_4 \ ~- . I 1 '1: ~ 't in hat ~ Z - ~ 2 Z _ On ~< X IS eL Cat Z ~ ~ 1~ UJ Cow Z Hi_ '' __'-' I ~ . ~I ~ / ~I \ ~ ~'\~ .J I · "a I L I'm ~ I / `____r I . _- ~-_ ''1 Hi-' 'en i\ I ~ of a: 3 o At: o 3 C) Cat Cat o ._ Ct .= .o CD  ._ :: A U. ¢ ~ ~ CO o Cot .o o o o >, _ ^ C<~= V .= Cat C: .o C> ~ So V _ ;^

334 configuration are discussed in the supporting literature (MAP3S/RAINE 1982). The OSCAR data set is currently under intensive analysis, and only preliminary results are available. It is of interest to consider some of these results at this point, however, to evaluate the potential future utility Of this material. One early result, presented by Raynor (1981), is primarily of qualitative interest. These are the first-sample/last-sample pa data obtained by the sequential rain samplers for individual storms and are tvoified by the plots shown in Figures C.3-17 and . . ~ , _ C.3-18. It is interesting to note that Figure C.3-17 is strongly reminiscent of annual- or multiyear-average clots for the northeastern United States in the sense that it shows the familiar acid "core" region centered upon Pennsylvania.* The final-sample distribution in Figure C.3-18 is quite different. Besides indicating a much cleaner sample set, very little structure exists in this final distribution. ...__ . . ants relative cleanliness of late-storm precipitation is consistent with the general OSCAR finding that most of the pollutant is scavenged comparatively early in a storm's life cycle (Easter and Hales 1983a). Substantial source-receptor analysis is currently being conducted in conjunction with the Indiana-Ohio concentrated network. One early analysis, conducted for the April 22-24, 1981, storm, is presented in Figure C.3-19. Backtrajectories of this type are currently being combined in diagnostic scavenging models with aircraft and surface data to evaluate source-receptor relationships in greater detail (Easter and Hales 1983a,b). *It should be noted in this context that field studies having higher spatial resolution (e.g., Hales and Dana, 1979b, Semonin 1976) indicate that significant fine structure typically exists in spatial pH distributions. Much of this fine structure can be expected to be hidden within the relatively coarse sampling mesh shown in Figures C.3-18 and C.3-19.

.: 1 (it' o At o a: At c~ I: o . - 'e ~ - / ~ 'e ~ - · .= Hi o a: o ~ - · 50 cO - J ~ 1 ~ Am-\ ,-r' "a l 1 ~ _ _ j ~ 1 ~<_~1~ 1

336 i,,,. 1 _~\ I i,,/''/' - ~ / 1 1 ~ Do ax I_ ¢ o o C: O . - ¢) Cat ~3 o ._ Ct ._ .= C,  P" c: Cot TIC o o U. oo Cat C:

337 - - ~ \ G R B .~ ~ l i; a\\/ FNT . ~-\ _ BUF TEA ! INS SAL __ ~ CLE J ~ PIT I ~ FIGURE C.3-19 Loci of points contributing pollution to the high-density network near 1400 EST on April 22, 1981. Contour intervals 3, 6, 9 represent travel times in hours from source regions. The large arrow represents the likely path of air originating from points 9 hours upwind of the receptors.

338 3 . 4 PREDICTIVE AND INTERPRETIVE MODELS OF SCAVENGING 3 . 4.1 Introduction A precipitation-scavenging model can be defined as any conceptualization of individual Figure C.3-2, in a manner that allows their expression in mathematical form. Often such models take the form of submodels or Modules within a larger calculational framework, such as a composite regional pollution code. When considered in a modular sense the lines connecting the boxes of Figure C.3-2 can be considered as channels for information exchange within the overall framework, whereas the boxes (or clusters of boxes) can be identified with the modules themselves. Scavenging models are currently in a rapidly evolving state, and a profusion of associated computer codes and computational formulas is currently available. Indeed, one of the major problems in precipitation-scavenging assessment is determining precisely which model to select from the large number of available candidates. A major aim of the present subsection is to guide the reader in this pursuit. There are a number of potential uses for precipitation-scavenging models, and the intended use will to a large extent determine which model should be employed. Some of the more important potential uses are itemized as follows: . . . and Prediction of the impact on precipitation chemistry of proposed new sources, source modifications, and alternate emission-control strategies; · Prediction of long-range trends in precipitation chemistry; Estimation of the relative contributions of specific sources to precipitation chemistry at a chosen receptor point; Estimation of transport of acidic-precipitation precursors across political borders; Estimation and prediction of air-quality modifications occurring as a consequence of the scavenging process; Site selection for precipitation-chemistry network  sampling stations; Design of field studies of precipitation scavenging;

339 . Elucidation of mechanistic behavior of the scavenging process on the basis of field measurements. In selecting an appropriate model, the user should review his intended application carefully with regard to the pollutant materials of interest, the time and distance scales, the processes in Figure C.3-2 covered, the source configuration, the precipitation type, and the mechanistic detail required. The question of pollutant materials is particularly important when precipitation acidity is of interest. Acidity in precipitation is determined by the presence of a multitude of chemical species, and in principle one must compute (via a model) the scavenging of each species and then estimate acidity on the basis of an ion balance: [H+] = £ Anions - (£ cations other than H+). (C.3-1) Inorganic ions usually important in precipitation chemistry are itemized in Table C.3-4. Organic species play a secondary role in the acidification process, which appears to vary widely with region. Modeling of all of these species simultaneously requires substantial effort, and all ~acid-precipitation" models up to the present have focused on only one or just a few of the more impor- tant species, with contributions of the others estimated on the basis of empiricism. Currently there is a tendency for newer models to accommodate larger numbers of these species; but complete modeling coverage will not be achieved in the foreseeable future. Mechanistic detail is another important feature determining the basic composition of a scavenging model. A comprehensive mathematical description of the scavenging process can become rapidly overwhelming, and there is usually a need to represent these relationships in a comparatively simple, albeit approximate, manner. The process of consolidating complex behavior in this fashion is often referred to as lumping the system's parameters. The resulting simplified expressions are termed Parameter- izations. Consolidating the effects of nonmodeled species in empirical form, described in the preceding paragraph, is one example of lumping. Numerous other examples will arise throughout the remainder of this section. This section will not attempt to provide the reader with a detailed treatise on how models should be

340 TABLE C.3-4 Some Inorganic Ions Important in Precipitation Chemistrya Cations Anions H- Nat Na+ K+ Ca++ Mg++ C1 NO3 SO3 SO4 po-4 CO3 aAll ions are presented here in their completely dissociated states. The reader should note, however, that various states of partial dissociation are possible as well (e.g., HSO3, HCO3). formulated and applied.* The approach, rather, will be to develop a basic understanding of the fundamental elements of a scavenging model and then provide a systematic procedure for choosing and locating appropriate models from the literature. The following subsection discusses the basic conservation equations, which constitute the conceptual bases for scavenging models in general. This is followed in turn by two simple applications of these relationships, which are presented to illustrate usage and to define some terms commonly *For the reader interested in more detailed pursuit of this area, the works by Hales (1983) and Slinn (1983) are recommended. The Hales reference is something of a beginner's primer, while Slinn's treatment delves substantially deeper into mechanistic detail. Together they constitute a reasonable starting point for under- standing and modeling basic scavenging phenomena.

341 used in scavenging models. The final subsection attacks the problem of model selection, using a flow-chart approach, which is designed to guide the user to a valid choice in a systematic manner that avoids many of the pitfalls normally encountered on such endeavors. 3.4.2 Elements of a Scavenging Model 3.4.2.1 Material Balances . In Figure C.3-3 the various arrows between boxes cor- respond physically to streams of pollutant and/or water, and from this it is not difficult to realize that any characterization of this system must include material balances. Material balances thus form the underlying structure for all scavenging models. To formulate a material balance one simply visualizes some chosen volume of atmosphere and sums over all inputs and outputs of the substance in question. Two basic types of material balance are possible: 1. and 2. Microscopic material balances, based on summation over a limiting small volume element of atmosphere; "Macroscopic" material balances, based on summation over a larger volume element of atmosphere (e.g., a complete storm system). Microscopic material balances invariably lead to differential equations, which must be integrated over finite limits to obtain practical results. Macroscopic balances result in mixed, integral, or algebraic equations. Again the choice of material-balance type depends on the specific modeling purpose at hand. An important general form of the differential material balance for some chosen pollutant (denoted by subscript A) is given by the equations* (cf. Hales 1983) Equations (C.3-2) and (C.3-3) are quite general in the sense that the velocity vectors denote velocity of pollutant (rather than that of the bulk media) and thus provide for all modes of transport (convective, diffusive, ...) without yet specifying how this transport is to occur. These equations are not yet time-smoothed; thus no closure assumptions have been applied at this point.

342 and aC a Y = -v cA vA WA + rAy (gas phase) (C.3.2) ac Ax a t AXVAx + WA + rAx (aqueous phase). (C.3.3 Here cAy and COAX denote concentrations of pollutant in the gaseous and condensed-water phases, respectively. The time rate of change of these concentrations within the differential volume element is related to the sum of inputs by (1) flow through the walls of the element, (2) interphase transport between the gaseous and condensed phases. and (3) chemical (and/or physical) reaction within the element. The v terms in Equations (C.3-2) and (C.3-3) denote velocity vectors, while V. is the standard vector divergence operator. The interphase transport term wA accounts for all "attachment" processes (impaction, phoresis, diffusion, . . .) as well as any reverse phenomena such as pollutant-gas Resorption, while the r terms denote chemical conversion rates in the usual sense. To formulate a usable model from these equations one needs to specify values for the functions v, w, and r and then solve differential equations (C.3-2) and (C.3-3) (subject to appropriate initial and boundary conditions) to obtain the desired concentration fields cay and cAx. A simple example of this procedure is given in Section 3.4.2. Energy Balances Many terms in Equations (C.3-2) and (C.3-3), especially vAx, WA, and TAX, depend strongly on the amount, state, and interconversion rates of condensed water; and it is important at this point to note that atmospheric water itself obeys material-balance expressions of this form. In selecting a scavenging model one often is confronted with the problem of deciding whether to estimate precipitation attributes and these related terms independently on the basis of assumptions or previous information or to attempt to compute the desired entities directly by solving appropriate forms of Equations (C.3-2) and (C.3-3).

343 If the latter of these alternatives is chosen, then the inclusion of an energy-balance equation is mandatory. This need arises because the evaporation- condensation process influences, and is influenced by, a variety of energy-related considerations. These include temperature influences on vapor pressure and latent-heat effects and can be incorporated in the model via an energy balance performed over the same element of atmosphere as that of the associated material balances. In microscopic form, a general expression of the energy balance (cf. Bird et al. 1960), is aT ~ ~- PCV a t = ~ V.h - pV.v + r - D . (C.3-4) Here the time rate of change of temperature is related to the sum of inputs by (1) flow through the walls of the element and (2) generation via (a) compression work, (b) latent heat effects, and (c) frictional dissipation. The vector terms h and v denote sensible heat flux and fluid velocity, respectively, while r and D pertain to latent heat and dissipation. P and Cv denote fluid density and specific heat in the usual sense. A straightforward example of the incorporation of Equation (C.3-4) for scavenging modeling purposes is given by Hales (1983). Momentum Balances Solutions to Equations (C.3-2)-(C.3-4) depend on the existence of some previous description of fluid velocity v [or vAy in the case of Equation (C.3-2)]. As was the case for the preceding parameters associated with the energy balance, velocity may be specified for the model on the basis of previous measurements or assumptions. Flow patterns in storm systems may be sufficiently complex to defy empirical specification, however, and the modeler may wish to compute the associated fields on the basis of a modeling approach. If this is to be done, a momentum-balance equation must be employed. In microscopic form the general momentum balance may be expressed (cf. Bird et al. 1960) as a t Pv = -verve - Vp - F + pg. (C.3-5)

344 Here the time rate of change of momentum (pv) is expressed as the sum of inputs by (1) flow through the walls of the element, (2) pressure forces, (3) viscous drag forces, and (4) gravitational forces. To apply Equation (C.3-5) for modeling purposes one specifies frictional, pressure, and gravitational terms and solves the differential equation subject to appropriate initial and boundary conditions to obtain fields of the velocity vector v. An example of application of Equation (C.3-5) for scavenging modeling purposes is given by Mane (1978). Incorporation of energy and momentum balances Equations (C.3-4) and (C.3-5) into a scavenging model is a rather challenging exercise, and a relatively small number of models exist that apply these equations for this purpose. The usual tack is simply to "prespecify" the required parameters and proceed with material-balance calculations alone. Numerous examples of both types of models will be presented in Section 3.4.5. 3.4.3 Definitions of Scavenging Parameters Four key parameters often arise in the context of scavenging models, and it is appropriate at this point to define these terms and indicate their general application. Reference to these entities as "parameters" is consistent with the usage applied in the previous section, in that they serve to ~lump" the effects of a number of mechan- istic processes in a simple formulation. These will be discussed sequentially in the following paragraphs. The first parameter to be defined is the attachment efficiency. Also known as the capture efficiency, this term can be visualized most easily by considering a hydrometeor falling through a volume of polluted air space, as shown in Figure C.3-20. This hydrometeor sweeps out a volume of air during its passage; and attachment efficiency is defined as the amount of collected pollutant divided by the amount that was initially in this volume. The efficiency can exceed 1.0 if pollutant from outside the swept volume becomes attached to the drop. From the discussion of attachment mechanisms in Section 3.2 it is seen that the attachment efficiency Usually the efficiency is less than 1; but mechanisms such as diffusion, electrical effects, and interception can give rise to larger values, especially when the collecting accounts for a multitude of processes.

345 dz ~1 1 OR ICY FIGURE C.3-20 Schematic of a scavenging Hydrometeor falling through a volume element. element's fall velocity is small. Efficiencies can be negative if the element is releasing pollutant to the surrounding atmosphere, such as in the case of pollutant-gas Resorption. Typical efficiencies for aerosol particles collected by raindrops are shown in Figure C.3.4. Another important parameter is the scavenging coefficient. This entity is basically an expression of the law of mass action, and is defined by the form w A CAY (C. 3-6)

346 . where [in a manner consistent with Equations (C.3-2) and C.3-3)] wA is the rate of depletion of pollutant A from the gaseous phase by attachment to the aqueous phase in a differential volume element. This is similar to a rate expression for a first-order, irreversible chemical reaction, and as such it applies strictly only to irreversible attachment processes (e.g., aerosols or highly soluble gases). A can be related to the attachment efficiency E by the form (which assumes spherical hydrometeors) Ata) = -nNTi R vz(R)E(R,a)fR(R)dR, (C.3-7) where a and R denote aerosol and hydrometeor radii, respectively, vz is the hydrometeor fall velocity, and NT and fR are the total number and probability- density functions for the size-distributed hydrometeors residing in the volume element of Figure C.3-20 at any instant in time. From this one can note that ~ essentially extends the paramaterization over the total spectrum of hydrometeor sizes. Atmospheric aerosol particles are typically dis- tributed over extensive size ranges, and because of this it is often desirable to possess some sort of an effective scavenging coefficient, which represents a weighted average over the aerosol size spectrum. Figure C.3-21 presents a family of curves corresponding to such aver- ages, which are based on assumed log-normal particle- size spectra, with different geometric standard deviations. From these curves one can observe that for the same geometric mean particle size, changes in spread of the size distribution can result in dramatic changes in the effective scavenging coefficient. Inclusion of reversible attachment processes in a scavenging model usually involves utilization of the mass-transfer coefficient. This parameter can be defined in terms of the flux of pollutant moving from the scavenging element as K YFlux = - - (c - h c ). Here ~ is the mass-transfer coefficient and q~ is the concentration, within the scavenging element, of (C.3-8)

347 \~ \ In\ \ \ ~ ~ - . Us \~ Cal ,, ~ en \ ~ \~= `^ - \\ a \ 1. I. Ll:: - l J4 'V - - Ct3 - Ct et Ct o  . C> £ Ct Hi? ._ Cal U3 _ o so it: _ e ._ , ~ _ ._ _ 1 o ~ At: U) - Cal o ._ CO ;^ ~ A_ c: <¢ ·C) a ~ o so: ~ , .= Ct Cal o o _ ~ 3 . ~ ._ ._ o Cal o 43 Cal - o o - C~ o a> Cal to Cal o .~ .> :> C) in; Cal C) ~ o it:  C) ~ ._ Hi, o ~ o.c A- ~ ~ ._ ~ 3

348 collected pollutant. h' is essentially a solubility coefficient, which, when multiplied by CA, produces a gas-phase equilibrium value. c is the molar concentration of air molecules, which appears in Equation (C.3-8) because of the manner in which Ky has been defined. Thus the flux can be either to the drop or away from it, depending on the relative magnitudes of the parenthetical terms. Equation (C.3-8) can be integrated over all drop sizes in a manner similar to that used in Equation (C.3-7) (cf. Hales 1972) to form the following expression for we: 4nN Jo R fR(R)Ky(R)(cAy ~ h cA) dR. (C.3-9) 1 ~ The final scavenging parameter to be described here is the scavenging ratio. This entity is usually the of a model calculation, rather than an input, and defined by the form CA i= , CAy result is (C.3-10) where CA is the concentration of pollutant contained in a collected precipitation sample. ~ is a term that is immediately usable for a number of pragmatic purposes, because once its numerical value is known it can be applied directly to compute precipitation-chemistry concentrations on the basis of air-quality measurements. Tables of measured (Engelmann 1971) and model-predicted (Scott 1978) scavenging washout ratios have been published, although caution is advised in the application of these values. A simple example of scavenging-ratio application is given in the following section. It is useful for the sake of visualization to discuss briefly the qualitative features of the scavenging parameters noted above. The parameter E is easy to visualize in the context of Figure C.3-20; it is, simply, the collection efficiency of an individual cloud or precipitation element and as such should be expected to fall numerically in the approximate range between zero and one. The scavenging coefficient ~ can be visualized as a first-order removal rate, in much the same manner as that of a first-order reaction-rate

349 coefficient. As such it may be utilized roughly as a characteristic time scale for wet removal. A = 1 h 1, for example, would imply that the scavenging process will cleanse 100(1 - 1/e) percent of the pollutant in 1 h if conditions remain constant and competitive processes do not occur. From this one can note that 1 h~1 is a moderately large scavenging coefficient. As ranging from zero to 1 h~1 and beyond have been reported in the literature elf hi mild" C.3-21). ~_. ~ _ _, a. _ = ~- ~ fine mass-transfer coefficient RV is essentially a normalized interracial flux of pollutant between the atmosphere and an individual droplet. Little needs to be said here regarding magnitudes of K , except to note that a variety of different definitions of Ky exist, and one must be cognizant of these definitions when employing values obtained from outside sources. The washout ratio, i, is essentially a measure of the concentrating power of precipitation in its extraction of pollutant from the atmosphere. As will be noted in the next section, precipitation often has the ability to concentrate airborne pollution by a factor of a million or more. is ranging from below 100 up through 108 and higher have been reported in the literature. The expected magnitudes and uncertainty levels associated with the scavenging parameters listed in this section depend strongly on the substance being scavenged and the environment in which the scavenging takes place. Large aerosol particles in below-cloud environments, for example, are characterized by scavenging efficiencies in the range of 1.0 (cf. Figure C.3-4), which can be esti- mated with relatively high precision. Smaller particles, especially those in the "Greenfield-gapn region, are much more difficult to simulate, and associated errors in estimated efficiencies may approach an order of magnitude or more. Errors in these efficiency estimates will of course be compounded by uncertainties in raindrop size spectra, if extended to scavenging coefficients via Equation (C.3-7). In the case of gases, the mass- transfer coefficient usually can be estimated to within a factor of 2 or less; again this error can be expected to compound when integrated over assumed raindrop size-spectra. In the case of in-cloud scavenging of aerosols our capability for estimating transport parameters is seriously impeded, owing to the profusion of mechanisms and the complex environments involved. Typical

350 uncertainties in both A and ~ can be expected to approach an order of magnitude in some cases. Some appreciation for the factors influencing in-cloud scavenging coefficients can be obtained from the work of Slinn (1977), who attempts to evaluate theoretical, n storm~averaged" values for A. An idea of the _ . magnitudes and uncertainties of ~ is given in Figure C.3-23. In all cases involving reactive gases the values of E, A, and ~ are heavily contingent on the aqueous-phase chemical processes involved. Much remains to be accost plished in our understanding of aqueous-phase chemistry before a meaningful assessment of associated uncertainties is possible. - As a final note in this context it should be emphasized that uncertainties in scavenging parameters dictate uncertainties in scavenging calculations in a complex fashion and that errors associated with the microscopic phenomena can be either amplified or attenuated by their applications in macroscopic models to produce practical results. Uncertainties associated with macroscopic modeling applications will be discussed at some length in a later section. 3.4.4 Formulation of Scavenging Models: Simple Examples of Microscopic and Macroscopic Approaches As noted previously, the description given in this docu- ment will refrain in general from deriving and applying -~ an: ~ ~1 ~ ~v~1; ha;-1 w scavenging lllo"=l~ =~l`~l~^y. This is too broad and complex a topic to be discussed in detail here, and the reader is referred to the previously cited literature for more detailed pursuit of this subject. For purposes of illustration, however, it is worthwhile to consider two simple examples of scavenging-model formulation, which demonstrate microscopic and macroscopic approaches to the problem. The present subsection addresses this task. The microscopic material balance approach will be considered first. For this example it is useful to visualize an idealized situation where rain of known characteristics is falling through a stagnant volume of atmosphere, which contains a well-mixed, nonreactive pollutant with concentration cay. The air velocity is known (v = 0) so solution of the momentum equation (C.3-5) is not required. The raindrop size distribution is presumed to remain constant; thus evaporation

351 condensation and other energy-related effects are immaterial, and the energy equation (C.3-4) may be disregarded. Since the pollutant is well mixed, no concentration gradients occur; thus the divergence term in Equation (C.3-2) is zero. Because of nonreactivity the reaction term is zero as well. Now presume that the pollutant is an aerosol, whose attachment can be characterized in terms of the known scavenging coefficient A, using Equation (C.3-6). The corresponding reduced form of Equation (C.3-2) is, then, tic Ay at -Ac Ay (C.3-2a) Given some initial pollutant concentration chino, Equation (C.3-2a) can be integrated to obtain the form cay(t) = cAyo exp(-At), (C.3-11) which expresses the decrease of the gasphase pollutant concentration with time. Counterpart expressions for rainborne concentrations may be derived by subjecting Equation (C.3-3) to a similar treatment. The reader is cautioned to consider this treatment as an example only and to recognize that actual atmospheric conditions seldom conform to the idealizations invoked above. Gas-phase concentrations are usually not uniformly distributed in space, raindrop characteristics are usually not invariant with time, wind fields are usually not well characterized by v = 0. A is usually not a time- independent constant, and many pollutants are usually not well characterized by the washout coefficient approxi- mation, anyway. The pollutant often is not unreactive. Examples of existing models where these constraints are relaxed in various ways are presented in the following subsection. Figure C.3-22 illustrates the formulation of a macroscopic type of scavenging model. Here, in contrast to the differential-element approach, the material balances are formulated around a large volume element, in this case a total storm. If one denotes concentrations and flow rates of water and pollutant as follows: cAy = airborne concentration of pollutant, H = airborne concentration of water vapor into cloud'

352 ~ ~ 1l / in ~,~5, <'l If W\ ~ - ~C Z _ _. ~ ~ Z o _ ~ Z ~ o 3 o ~ ~,~ ~ e' ~en, _ _ ~- - - l t Ct . - C) At -  c 3 13 - Cat C) -

353 8 107 -- ~' 1 1 ' 1 ~, , 1, ~ ~ '1 , - -1 ~ ~ 1 1 ~ 1 \ 2 105 0.01 - ,,, ,,,,,1 ,,, 1 1.0 PRECIPITATION RATE (mm h 1) \ ,, ,,1 1 1 1 1, ,,, 0.1 10 100 FIGURE C.3-23 Scott's scavenging ratio curves: 1, convective storms; 2, warm, non- convective storms; 3, cold storms, where Bergeron-Findeisen process is active (Scott 1 978). CA Pw win wOut fin fout W F concentration of scavenged pollutant in rainwater, density of condensed water, flow rate of water vapor into the storm, flow rate of water vapor out of the storm, flow rate of pollutant into the storm, flow rate of pollutant out of the storm, low rate of precipitation out of the storm, flow rate of scavenged pollutant out of the storm,  then extraction efficiencies for water vapor and pollutant can be defined, respectively, as W P Win and (C. 3-11)

354 FIn (C.3-12) If one further performs material balances over this storm system for pollutant and water vapor, and then combines the two, the following form is obtained: C £ p A P w CAY (C.3-13) where the scavenging ratio, [, was defined earlier in Section 3.4.3. Equation (C.3-13) is an important result in the sense that it demonstrates once again the strong linkage between water-extraction and pollutant-scavenging processes. If both occur with equal efficiency (up = £),* for example, then [= -. ~ 10 -10. (C.3-14 Experimentally measured scavenging ratios often fall in this range, although wide variability often may be observed. Utilizing a rather involved series of arguments pertaining to cloud-physics processes and attachment mechanisms, Scott (1978) has created a family of curves expressing aerosol scavenging ratio as a function of precipitation rate. Shown in Figure C.3-23, curves 1, 2, . _ *There is no direct reason to expect that ~ should be similar to ~ in magnitude. In the absurd circumstance where all the pollutant were concentrated into one particle, for example, then scavenging of that Pollutant by a very light rainfall would yield ~ ~ 1.0 >> sp. Conversely' a large storm processing an insoluble gaseous pollutant bang, say' would provide ~ ~ O << ~ . For practical conditions involving aci8-forming aerosols, however, the scavenging of water vapor and pollutant appear to be sufficiently related to allow sp ~ ~ to be employed as an approximate rule of thumb.

355 and 3 pertain, respectively, to convective storms, nonconnective warm~rain process storms, and cold storms where the Bergeron-Findeisen process is active. A major assumption of Scott's analysis is that the pollutant is ingested by the storm in the form of aerosol particles that are active as cloud condensation nuclei. The analysis also assumes a steady-state storm system and complete vertical mixing of Pollutant between the storm height and the surface. Under such conditions Scott's curves can be considered as reasonably good estimators of actual scavenging behavior. More elaborate systems, involving reactive pollutants, gases, and nonhomogeneous systems are discussed in references given in the following section. 3.4.5 Systematic Selection of Scavenging Models: A Flow-Chart Approach Hales (1983) has suggested a flow-chart approach to aid in the process of scavenging-model selection. Presented as a decision tree in Figure C.3-24, the user proceeds by answering a series of questions that relate to the model's intended use, the temporal and geographical scales, the pollutant characteristics, the choice between macroscopic and microscopic material balances, and the type of conservation (i.e. mater tat . enerav . momentum ) equations involved. · . ~ ,~ ~ , Various pathways through this Recision tree are discussed in the original reference. Proceeding through Figure C.3-24 in this manner the user can arrive at simple or complex end points, depending on the nature of his particular application. A trivial example is pathway 1-5-6, which instructs the user to disregard modeling totally and rely solely on past measurements. The simple microscopic-balance example of Section 3.4.4 can be traced through n~Ehw~v 1-2-7-8-21-23-15-16. _, ,= _ _ ,, _ ~ cache u.~-u presents an itemization of some currently available models, which can be related directly to the . . . pathways of Figure C.3-24. This provides the reader with a rapid and efficient means of access to current modeling literature, while minimizing the chance of pitfall encounters that can arise from the inadvertent invocation of inappropriate physical constraints. For a more definitive description of this model selection process, the reader is referred to Hales's original reference.

356 r--~ l 1 0 flu W O ~ 4 < \/ L_ r--~~1 ~ Jon Hi :: ~ · I T ~_: ~ T: _# ~ I,,| ,,9]._ ': .--. i___ ': h---1 :__ CO o . - Ct ._. A> Ct to o - rut C) _d

357 3 . 5 PRACTICAL ASPECTS OF SCAVENGING MODELS: LEVELS AND SOURCES OF ERROR UNCERTAINT Y Quantitative assessment of the predictive capability of present wet-removal models is a complex task and is well beyond the scope of this document. There are, however, number of general statements that are highly useful in focusing in on this question and in providing insights pertaining to model reliability. These are itemized sequentially below. . . · The predictive capability of a scavenaina morph ~: By contingent on its desired application AS notes in section 3-4~1, there exists a variety of different applications of scavenging models, and some are much more difficult to fulfill than others. One can, for example, employ existing regional models to reproduce distributions of annually averaged, wet-deposited, sulfate ion in eastern North America with moderate success. If one is charged with the task of relating specific sources to deposition at a chosen receptor site, however, our predictive capability can be expected to be relatively imprecise. Similarly, if one is expected to forecast the change in deposition that would occur in response to some future change in emissions, then the associated uncertainty level would be very high indeed. The question of nonlinear response is of paramount importance in this last application. A large component of our uncertainty in predicting source attribution and transient response is based simply on the fact that we do not have adequate data bases for testing model performance for these applications. Our present models may in actuality be better predictors in this respect than anticipated; but because we have no immediate way of confirming this our uncertainty level remains high. Regardless of the above considerations it should be emphasized strongly that the first step in scavenging model evaluation must be the precise definition of the intended uses of the model. All subsequent efforts will be confounded in the absence of this focal point. ~ ~ ~ . The predictive canabilitv of A .~AV"n~; net m^A" 1 is dependent on the choice of model. , At first sight this appears to be a self-evident and trivial statement. A profusion of scavenging models exists, however, and it is not at all difficult to choose

358 TABLE C.3-5 Pertinent Literature References for Wet-Removal Models Model 1. Classical washout coefficient 2. Distributed washout coefficient Two-stage nucleation- accretion Nonreactive gas scavenging 5. Reactive gas scavenging 6. In-cloud aerosol scavenging In-cloud aerosol scavenging 8. In-cloud reactive gas and aerosol scavenging 9. In-cloud reactive gas and aerosol scavenging 10. Composite analytical 11. Composite trajectory 12. Composite grid 13. Composite statistical 14. Nonreactive 15. Reactive Type of Balance Equations Material (differential) Material (differential) Material (differential) Material (differential) Material (differential) Material (differential)  Material (integral) Material (differential) Material (integral) Material (differential) Material (differential) Material (differential) Material Material energy and momentum (differential) Material and energy (differential) Mechanisms Irreversible attachment Irreversible attachment Irreversible attachment Reversible attachment Reversible attachment with aqueous-phase reaction Irreversible attachment Irreversible or reversible attachment Transport, reaction, and deposition Irreversible or reversible attachment with chemical reaction Transport, reaction, and deposition Transport, reaction, and deposition Transport, reaction, and deposition Transport, reaction, and deposition Irreversible attachment, nonreactive All modes of scavenging including chemical reaction

359 Typical Application Be logic loud scaveng i ng of aerosols and reactive gases Below-cloud scavenging of size-d istr ibuted aerosols Condensation-enhanced be low c loud scaveng i ng of aerosols Be low-c loud scaveng i ng of nonreactive gases Below-cloud scavenging of reactive gases Scavenging in storm systems (nonreactive ) Scaveng ing in storm sys tems Scoping studies Interpretation of study data Reg ional-s cafe deposition Reg tonal-scale depos ition Regiona 1-scale depos ition Scoping studies and life-t ime assessment In-cloud scaveng ing analysis I n-c loud scaveng i ng analys is Pertinent References Chamberlain (1953), Engelmann (1968), Fisher (1975), Scriven and Fisher (1975), Wangen and Williams (1978) Dana and Hales (1976), Slinn (1982) Radke et al. (1978), Slinn (1982) Barrie (1978), Hales et al. (1973), Slinn (1974b) Adamowitz (1979), Drewes and Hales (1982), Durham et al. (1981) , Hill and Adamowitz (1977), Overton et al. (1979) Dingle and Lee (1973), Junge (1963), Klett (1977), Lange and Knox (1977), Slinn (1982), Storebo  and Dingle (1974) Engelmann ( 19 7 1), Gat z ( 19 7 2 ), Hale s and Dan a (1979), Scott (1978), Slinn (1982) Gravenhorst et al. (1975), Omstedt and Rodhe ( 1978) Scott (1982) Astarita et al. (1979), Fay and Rosenzweig (1980) Bass (1980), Bhumralker et al. (1980), Bolin and Persson (1975), Eliassen (1978), Fisher (1978), Hales (1977), Heffter (1980), Henmi (1980), Kleinman et al. (1980), McNaughton et al. (1981), Patterson et al. (1981), Sampson (1980), Shannon (1981), Voldner (1982) Carmichael and Peters (1981), Lamb (1981), Lavery et al. (1980), Lee (1981), Liu and Durran (1977), Prahm and Christensen (1977), Wilkening and Ragland (197 8) Rodhe and Grandell (1972) Hane (1978), Kreitzburg and Leach (1978), Molenkamp (1974) Hales (1982a)

360 an inappropriate candidate inadvertently. Such inappropriate selections have on occasion resulted in reported calculations that have been in error by several orders of magnitude. This component of error may of course be totally eliminated by selection of the most appropriate model for the intended application. The flow chart presented in Figure C.3-24 is a useful guide for this purpose, especially for those only casually familiar with the field. · The Predictive capability of a scavenging model depends strongly on the processes modeled. As noted in the context of Figure C.3-Z a scavenging model mav encompass one. several, or all of the steps in the composite wet-removal sequence. If only a small portion of this sequence is being considered, the model depends heavily on information supplied from the remaining components. This information may originate from assumptions, from empirical measurements, or from the output of other models. . . . Assuming that all input information Is error-rree, then it may be stated generally that, the more steps in Figure C.3-2 encompassed by a given model, the greater will be its predictive uncertainty. This is simply a consequence of Propagating errors and must be considered as a primary factor when addressing the validation of wet-removal calculations. . The predictive capability of a scavenging model is dependent on its areal range. This statement is largely a corollary of the one immediately above. As a scavenging model is extended to, say, a regional scale it is forced to include essentially all the components of Figure C.3-2. As noted previously, this is likely to increase uncertainty levels appreciably. · The predictive capability of a scavenging model is contingent on its temporal averaging time. _ ~ - I_ Owing to the propensity of stochastic phenomena to average out to mean values, the predictive capabilities of (especially regional) scavenging models can be expected to improve somewhat as averaging times increase. This improvement is, of course, gained at the expense of sacrificing temporal resolution, and a value judgment is necessary (again requiring a precise

361 definition of intended model application) at this juncture.* This observation should be tempered by the fact that, in addition to random errors, scavenging models can be expected to possess substantial systematic biases. In general these biases do not decrease with averaging time and in fact many lead to cumulative discrepancies on occasion. Examples of systematic errors are biases in trajectory calculations and artificial offsets induced by the superimposition of random events on nonlinear processes. Again the seriousness of such factors is heavily contingent on the intended model application. In general summary it may be stated that several important factors lead to widely varying levels of uncertainty in scavenging-model predictions. ~ ~ ~ ~ ; _ ~ ~. . . . One may I, ~ w~ example, One scavenging of SO2 from a local power-p'ant plume using existing models and expect to match measured results within a factor of 2. On the other hand, similar predictions of, say, the fraction of sulfate at a given receptor that originated from some particular source can be expected to have orders-of- magnitude associated uncertainty. Both a comprehensive model-evaluation effort and a substantially improved data base will be required before this situation can be remedied to any appreciable extent. 3.5 CONCLUSIONS TO SECTION 3 This section has provided an overview of meteorological processes contributing to wet removal of pollutants and has summarized the current state of our capability to describe these complex phenomena in mathematical form. Because of the magnitude of this problem it has been necessary to refrain from detailed descriptions of models and modeling techniques; rather, we have chosen to describe the general mathematical basis for wet-removal modeling, to give two simple examples of direct application, and then to sunolv the r-~-r wi Ah a means LO he'- =~^ W ~ ~11 - *This issu" i ~ "=n"mi =1 1~ ~_~:~_= ~ ^- We ~ arty per In Vlew cuff the contention, often voiced by some scientists within the acid-precipitation effects community, that temporally averaged results (averaging times of a few months or more) are totally adequate for assessment purposes.

362 for efficiently pursuing the available literature for specific applications of interest. In conclusion to this discussion it is appropriate to summarize the state of these calculational techniques by asking the following questions: · Just how accurate and valid are current wet-removal modeling techniques as predictions of precipitation chemistry and wet deposition; that is, how well do they fulfill the needs itemized in Section 3.4.1? · What must be accomplished before the present capabilities can be improved? The answers to these questions are somewhat mixed. Certainly the techniques discussed in this section, if used appropriately, are capable of order-of-magnitude determinations in many circumstances; and under restricted conditions they can even generate predictions having factor-of-2 accuracy or better. Moreover, there is ample explanation in existing theories of wet removal to account easily for the spatial and temporal variabilities observed in nature. These capabilities, however, cannot be considered to be satisfactory in the context of current needs. The noted ability to explain spatial and temporal variability on a semiquantitative basis has not resulted in a large competence in predicting such variability in specific instances. Moreover, we possess little competence in identifying specific sources responsible for wet deposition at a given receptor site. Finally, the order-of-magnitude predictive capability noted above can hardly be judged satisfactory for most assessment purposes. In reviewing the discussions of this section against the backdrop of these deficits, several research needs become apparent. The most important of these are itemized in the following paragraphs. · Much more definitive information is needed with regard to the scavenging efficiencies of submicrometer aerosols, for both rain and snow. Especially important in this regard is the effect of condensational growth of such aerosols in below-cloud environments. · We need to know much more about aqueous-phase conversion processes, which are potentially important as alternate mechanisms resulting in the presence of species such as sulfate and nitrate in precipitation. Since virtually nothing is known at present regarding the

363 chemical formation of such species in clouds and precipitation, there is a tendency to lump these effects with physical removal processes In most modeling efforts, expressing them in terms of pseudo scavenging coefficients or collection efficiencies. Such phenomena must be resolved in finer mechanistic detail than this before a satisfactory treatment is possible, and this requires a knowledge of chemical transformation processes that is much more advanced than existing at present. ~ Much more extensive understanding of the competitive nucleation capability of aerosols in in-cloud environments is needed, especially for those substances that do not compete particularly well in the nucleation process. The influence of aerosol-particle composition-- especially for "internally mixed aerosols"*--is particularly important in this regard. · The identification of specific sources responsible for chemical deposition at a given receptor location requires that we possess a much more accomplished capability to describe long-range pollution transport. Progress in this area during recent years has been encouraging, but much more remains to be achieved before we have a proficiency that is really satisfactory for reliable source-receptor analysis. · We still need to enhance our understanding of the detailed microphysical and dynamical processes that occur in storm systems. Besides providing required knowledge of basic physical phenomena, such research is important in providing valid parameterizations of wet removal for subsequent use in composite regional models. As a final note, it is useful to reflect once again on the fact that scavenging modeling research--as treated in this section--has been in a rather continuous state of development over the past 30 years. while progress has been indeed significant during this period, a number of important and unsolved problems still exist. Accordingly, one must use this perspective in assessing our rate of advancement during future years. Reasonable progress in resolving the above items can be expected over the next decade; but the complexity of these problems demands that a serious and sustained effort be applied for this purpose. *Those containing individual particles composed of a mixture of chemical species.

364 3.7 REFERENCES Adamowitz, R.F. 1979. A model for the reversible washout of sulfur dioxide, ammonia, and carbon dioxide from a polluted atmosphere and the production of sulfate in raindrops. Atmos. Environ. 13:105-122. Astarita, G., J. Wei, and G. Iorio. 1979. Theory of dispersion transformation and deposition of atmospheric pollution using modified Green's functions. Atmos. Environ. 13:239-246. Baker, M.G., H. Harrison, J. Vinelli, and K.B. Erickson. 1969. Simple stochastic models for the sources and sinks of two aerosol types. Tellus 31:1-39. Barrie, L.A. 1978. An improved model of reversible SO2 washout by rain. Atmos. Environ. 12:402-412. Barrie, L.A., and J. Kovalick. 1978. A wintertime investigation of the deposition of pollutants around an isolated power plant in northern Alberta. Atmospheric Environment Service, Environment Canada, REP ARQT-4-78. Bass, A. 1980. Modeling long range transport and diffusion. In Proceedings Second Conference on Applied Air Pollution Meteorology. AMS/APCA, New Orleans. Berry, E.X., and R.L. Reinhardt. 1974. An analysis of cloud drop growth by collection. Part IV. A new parameterization. J. Atmos. Sci. 31:2127-2135. Bhumralkar, C.M., W.B. Johnson, R.H. Mancusco, R.H. Thuillier, and D.E. Wolf. 1980. Interregional exchanges of airborne sulfur pollution and deposition in eastern North Americae In Proceedings Second Conference on Applied Air Pollution Meteorology. . MS/APCA, New Orleans. Bird, R.B., W.E. Stewart, and E.N. Lightfoot. 1960. Transport Phenomena. New York: John Wiley and Sons. Bolin, B., and C. Persson. 1975. Regional dispersion and deposition of atmospheric pollutants with particular application to sulphur pollution over western Europe. Tellus 27:281-309. Browning, K.A., M.E. Hardman, T.W. Harrold, and C.W. Pardoe. 1973. The structure of rainbands within a midlatitude cyclonic depression. Q. J. Roy. Meteorol. Soc. 99:215-231. Burtsev, I.E., L.V. Burtsevva, and S.G. Malakhov. 1976. Washout characteristics of a 32P aerosol injected into a cloud. Atmospheric Scavenging of Radioisotopes. Symp. Proc. Palanga, USSR.

365 Cadle, R.D. 1965. Particle Size. New York: Reinhold Publishing Company. 390 pp. Carmichael, G.R., and L.K. Peters. 1981. Application of the sulfur transport Eulerian model (STEM) to a SURE data set. 12th International Technical Meeting on Air Pollution. Modelling and Its Applications. NATO. Palo Alto, Calif. Chamberlain, A.C. 1953. Aspects of travel and deposition of aerosols and vapor clouds. AERE Harwell Report R1261. London: HMSO. Changnon, S.A. 1968. Precipitation scavenging of Lake Michigan Basin. Illinois State Water Survey Report, Bull. 52, Urbana, Ill. Changnon, S.A., A. Auer, R. Brahm, J. Hales, and R. Semonin. 1981. METROMEX-A Review and Summary. Meteorological Monograph, Vol. 18. Boston, Mass.: American Meteorological Society. Climatic Atlas of the United States. 1968. Washington, D.C.: U.S. Dept. of Commerce. Court, A. 1966. Fog frequency in the United States. Geog. Rev. N.Y. 56:543-550. Dana, M. T. 1970. Scavenging of soluble dye particles by rain. In Precipitation Scavenging 1970. R.J. Engelmann and W.G.N. Slinn, eds. AEC Symposium Series. Dana, M.T., and D.W. Glover. 1975. Precipitation scavenging of power plant effluents: rainwater concentrations of sulfur and nitrogen compounds and evaluation of rain samples Resorption of SO2. PAL Annual Report to U.S. AEC, BNWL-1950. Dana, M.T., and J.M. Hales. 1976. Statistical aspects of the washout of polydisperse aerosols. Atmos. Environ. 10:45-50 Dana, M.T., J.M. Hales, and M.A. wolf. 1972. Natural precipitation washout of sulfur dioxide. Battelle- Northwest Report to EPA. BNW-389. Dana, M.T., J.M. Hales, W.G.N. Slinn, and M.A. Wolf. 1973. Natural precipitation washout of sulfur compounds from plumes. Battelle-Northwest Report to EPA. EPA-R3-73-047. Dana, M.T., D.R. Drewes, D.W. Glover, and J.M. Hales. 1976. Precipitation scavenging of fossil fuel effluents. Battelle-Northwest Report to EPA. EPA-600/4-76-031. Dana, M.T., N.A. Wogman, and M.A. Wolf. 1978. Rain scavenging of tritiated water (HTO): a field experiment and theoretical considerations. Atmos. Environ. 12:1523-1529.

366 Dana, M.T., A.A.N. Patrinos, E.G. Chapman, and J.M. Thorp. 1982. Wintertime precipitation chemistry in North Georgia. In Proceedings ACS Symposium on Acid Rain, Las Vegas, Nev. Davenport, H.M., and L.K. Peters. 1978. Field studies of atmospheric particulate concentration changes during precipitation. Atmos. Environ. 12:997-1008. Davies, C.N. 1966. Aerosol Science. New York: Academic Press. Dingle, A.N., and Y. Lee. 1973. An analysis of in-cloud scavenging. J. Appl. Meteorol. 12:1295-1302. Dingle, A.N., D.F. Gatz, and J.W. Winchester. 1969. A pilot experiment using indium as tracer in a convective storm. J. Appl. Meteorol. 8:236-240. Drewes, D.R., and J.M. Hales. 1982. SMICK: a scavenging model incorporating chemical kinetics. Atmos. Environ. 16:1717-1724. Durham, J.L., J.H. Overton, and V.P. Aneja. 1981. Influence of gaseous nitric acid on sulfate production and acidity in rain. Atmos. Environ. 15:1059-1068. Easter, R. C. 1982. The OSCAR Experiment. In Proceedings ACS Symposium on Acid Rain, Las Vegas, Nev. Easter, R.C., and J.M. Hales. 1983a. Interpretations of the OSCAR data for reactive gas scavenging. Proceed- ings Fourth International Conference on Precipitation Scavenging, Dry Deposition and Resuspension, Santa Monica, Calif. Easter, R.C., and J.M. Hales. 1983b. Mechanistic evaluation of precipitation-scavenging data using a one-dimensional reactive storm model. Battelle- Northwest Report to EPRI. EPRI RP-2022-1. Eliassen, A. 1978. The OECD study of long-range transport of air pollutants. Atmos. Environ. 12:479-487. Engelmann, R.J. 1965. Rain scavenging of zinc sulphide particles. J. Atmos. Sci. 22:719-724. Engelmann, R.J. 1968. The calculation of precipitation scavenging. In Meteorology and Atomic Energy 1968. D. Slade, ed. U.S. AEC. Engelmann. R.J. 1971. Scavenging prediction using ratios of air and precipitation. J. Appl. Meteorol. 10:493-497 Engelmann, R.J., R.W. Perkins, D.I. Hagen, and W.A. Haller. 1966. Washout coefficients for selected gases and particles. U.S. AEC Report. BNWL-SA-657. Enger, L., and U. Hogstrom. 1979. Dispersion and wet deposition of sulfur from a power-plant plume. Atmos. Environ. 13:789-810.

367 Falconer, R.E., and P.D. Falconer. 1980. Determination of cloud water acidity at a mountain observatory in the Adirondack Mountains of New York State. J. Geophys. Res. 85:7465-7470. Fay, J.A., and J.J. Rosenzweig. 1980. An analytical diffusion model for long-d~stance transport of air pollutants. Atmos. Environ. 14:355-365. Fisher, B.E.A. 1975. The long range transport of sulfur dioxide. Atmos. Environ. 9:1063-1070. Fisher, B.E.A. 1978. The calculation of long term sulphur deposition in Europe. Atmos. Environ. 12:489-501. Fitzgerald, J.W. 1974. Effect of aerosol composition of cloud-droplet size distribution: a numerical study. J. Atmos. Sci. 31:1358-1367. Fuchs, N.A. 1964. The Mechanics of Aerosols. Oxford: Pergamon Press. 407 pp. Fuquay, J.J. 1970. Scavenging in perspective. In Precipitation Scavenging 1970. R.J. Engelmann and W.G.N. Slinn, eds. AEC Symposium Series 22. Galloway, J.N., and D.M. Whelpdale. 1980. An atmospheric sulfur budget for eastern North America. Atmos. Environ. 14:409-417. Gatz, D.F. 1972. Washout ratios in urban and non-urban areas. In Proceedings AMS Conference on Urban Environment, Philadelphia, Pa. Gatz, D.F. 1977. A review of chemical tracer experiments onprecipitation systems. Atmos. Environ. 11:945-953. Godske, C.L., T. Bergeron, J. Bjerkness, and R.E. Bundgaard. 1957. Dynamic Meteorology and weather Forecasting. Boston, Mass.: American Meteorological Society. Graedel, T.E., and J.P. Franey. 1977. Field measurements of submicronaerosol washout by rain. In Precipitation Scavenging 1974. ERDA Symposium Series 41. Granat, L., and H. Rodhe. 1973. A study of fallout by precipitation around an oil-fired power plant. Atmos. Environ. 7:781-792. Granat, L., and R. Soderlund. 1975. Atmospheric deposition due to long and short distance sources with special reference to wet and dry deposition of sulphur compounds around an oil-fired power plant. MISU Report A-32, Stockholm University, Sweden. Gravenhorst, G., T. Janssen-Schmidt, D.H. Ehhalt, and E.P. Roth. 1975. Long-range transport of airborne material and its removal by deposition and washout. Atmos. Environ. 9:49-68.

368 Greenfield, S. M. 1957. Rain scavenging of radioactive particulate matter from the atmosphere. J. Meteorol. 14:115-123. Hales, J.M. 1972. Fundamentals of the theory of gas scavenging by rain. Atmos. Environ. 6:635-659. Hales, J.M. 1977. An air pollution model incorporating nonlinear chemistry, variable trajectories, and plume-segment diffusion. Battelle-Northwest Report to EPA. EPA-450/3-77-012. Hales, J.M. 1983. Precipitation chemistry: its behavior and its calculation. In Air Pollutants and Their Effects on the Terrestrial Ecosystem. S.V. Krupa and A.H. Legge, eds. New York: John Wiley and Sons. Hales, J.M., and M.T. Dana. 1979a. Precipitation scavenging of urban pollutants by convective star m systems. J. Appl. Meteorol. 18:294-316. Hales, J.M., and M.T. Dana. 1979b. Regional scale deposition of sulfur dioxide by precipitation scavenging. Atmos. Environ. 13:1121-1132. Hales, J.M., J.M. Thorp, and M.A. Wolf. 1971. Field investigation of sulfur dioxide washout from the plume of a large coal-fired power plant by natural precipitation. Battelle-Northwest Final Report to Environmental Protection Agency. NTIS PB 203-129. Hales, J.M., M.A. Wolf, and M.T. Dana. 1973. A linear , model for predicting the washout of pollutant gases from industrial plumes. AICHE J. 19:292-297. Hane, C.E. 1978. Scavenging of urban pollutants by thunderstorm rainfall: numerical experimentation. J. Appl. Meteorol. 17:699-710. Haurwitz, B., and J.M. Austin. 1944. Climatology. New York: McGraw~Hill Book Company. Heffter, J.L. 1980. Air resources laboratories atmospheric transport and dispersion model (ARL-ATAD). NOAA Tech. Memo. ERL-81. Henmi, J. 1980. Long-range transport model of SO2 and sulfate and its application to the eastern United States. J. Geophys. Res. 85:4436-4442. Hill, F.B., and R.F. Adamowitz. 1977. A model for reversible washout of sulfur dioxide, ammonia, and carbon dioxide from a polluted atmosphere, and the production of sulfates in raindrops. Atmos. Environ. 11:912-927. Hobbs, P.V. 1978. Organization and structure of clouds and precipitation on the mesoscale and microscale in cyclonic storms. Rev. Geophys. Space Sci. 16:741-755.

369 Hobbs, P.V. 1979. A reassessment of the mechanism responsible for the sulfur content of acid rain. Proceedings of Advisory Workshop to Identify Research Needs on Formation of Acid Precipitation. EPRI Report, EA-10 74, WS-78-98. Hogstrom, U. 1974. Wet fallout of sulfurous pollutants emitted from a city during rain or snow. Atmos. Environ. 8:1291-1303. Hutchenson, M.R., and F.P. Hall. 1974. Sulfate washout from a coal-fired power plant plume. Atmos. Environ. 8:23-28. Junge, C.E. 1963. Air Chemistry and Radioactivity. New York: Academic Press. Junge, C.E. 1964. The modification of aerosol size distribution in the atmosphere. Final Tech. Report, Meteor. Geophys. Inst., Johannes Gutenberg Universitat. U.S. Army Contract DA-91-591-EVC2979. Junge, C.E. 1974. Residence time and variability of tropospheric trace gases. Tellus 26:477-488. Klein, W.H. 1958. The frequency of cyclones and anticyclones in relation to the mean circulation. J. Meteorology 15:98-102. Kleinman, L.J., J.G. Carney, and R.E. Meyers. 1980. Time dependence on average regional sulfur oxide concentrations. Proc. Second Conf. on Applied Air Pollution Meteorology. AMS/APCA, New Orleans. Klett, J. 1977. Precipitation scavenging in rainout assessment: the ACRA system and summaries of simulation results. LASL Report to ERDA, LA6763. Kramer, J.R. 1973. Atmospheric composition and precipitation of the Sudbury Region. Alternatives 2:18-25. Kreitzberg, C.W., and M.J. Leach. 1978. Diagnosis and prediction of troposhperic trajectories and cleansing. Proc. 85th National Meeting AIChE, Philadelphia, Pa. Lamb, R.G. 1981. A regional scale model of photochemical air pollution. Draft Report, Meteorology and Assessment Division, ~Pa/~Cn~ D^~h -~ ; =~1 ~ D_ _ b N.C. Lange, R., and J.B. Knox. 1977. Adaptation of a three-dimensional atmospheric transport-diffusion model to rainout assessments. In Precipitation Scavenging 1974. R.S. Semonin and R.W. Beadle, eds. ERDA Symposium Series 41, CONF 741003. ~ar son, T.V. R.J. Charlson, E.J. Knudson, G.D. ~ --^~ r ^~- ~ ~ ~C~ll~ ~ ~ r~L ~ t , Shristian, and H. Harrison. 1975. The influence of a sulfur dioxide point source on the rain chemistry of a

370 Levich, V.G. 1962. Physicochemical Hydrodynamics. single storm in the Puget Sound region. Water Air Soil Pollut. 4:319-328. Lavery, T.L., et al. 1980. Development and validation of a regional model to simulate atmospheric concentra- tions of S°k and sulfate. In Proceedings Second Joint Conference on Applied Air Pollution Meteorology, New Orleans, La. Lee, H.N. 1981. An alternate pseudospectral model for pollutant transport. Diffusion and deposition in the atmosphere. Atmos. Environ. 15:1017-1024. Englewood Cliffs, N.J.: Prentice-Hall. 700 pp. Liu, M.K., and D. Durran. 1977. The development of a regional air pollution model and its application to the northern Great Plains. EPA Report EPA-908/1-77-001. Lovett, G.M., W.A. Reiners, and R.K. Olson. 1982. Cloud droplet deposition in subalpine balsam fir forests: hydrological and chemical inputs. Science 218:1303-1304. MAP3S/RAINE. 1981. Biennial Progress Report. NTIS PNL-4096, U.S. EPA/DOE. MAP3S/RAINE. 1982. The MAP3S/RAINE precipitation chemistry network: statistical overview for the periods 1976-1980. Atmos. Environ. 16:1603-1631. Mason, B.J. 1971. The Physics of Clouds. Oxford: Clarendon Press, p. 579. McNaughton, D., D. Powell, and C. Berkowitz. 1981. A User's Guide to RAPT. MAP3S/RAINE Report, PNL-3390. Millan, M.M., S.C. Barton, N.D. Johnson, B. Weisman, M. Lusis, W. Chan, and R. Vet. 1982. Rain scavenging from tall stacks: a new experimental approach. Atmos. Environ. 16:2709-2714. Molenkamp, C.R. 1974. A one-d~mensional numerical model of precipitation scavenging with application to rainout of radioactive debris. Lawrence Livermore Laboratory Report to U.S. AEC. UCRL-51627. Morgan, J.J., and H.M. Liljestrand. 1980. Measurements and interpretation of acid rainfall in the Los Angeles Basin. Cal Tech Final Report AC-2-80, Pasadena, Calif. Mosiac. 1979. Acid from the sky. Mosiac (National Science Foundation) 10:35-40. Newell, R.E., J.W. Kidson, D.G. Vincent, and G.J. Baer. 1972. The General Circulation of the Tropical Atmosphere. Vols. 1 and 2. Cambridge, Mass.: MIT Press. Omstedt, G., and H. Rodhe. 1978. Transformation and removal processes for sulfur compounds as described by

371 a one-dimensional time-dependent diffusion model. Atmos. Environ. 12:503-509. OSCAR. 1981. Chapter 4 of MAP3S/RAINE Biennial Progress Report. EPA Report PNL-4096. Over ton, J.H., V.P. Ane ja, and J.L. Durham. 1979. Production of sulfate in rain and raindrops in polluted atmospheres. Atmos. Environ. 13:355-367. Patterson, D.E., R.B. Husar, W.E. Wilson, and L.F. Smith. 1981. Monte Carlo simulation of daily regional sulfur d istribution. J. Appl. Meteorol. 20:404-420. Prahm, L.V., and O. Christensen. 1977. Long-range transmission of pollutants simulated by a tw~dimensional pseudospectral dispersion model. J. Appl. Meteorol. 16:896-910. Pruppacher, H.R., and J.D. Klett. 1978. Microphysics of Clouds and Precipitation. Dordrecht, Netherlands: D. Reidel Publishing Company. Radke, L.F., M.W. Eltgroth, and P.V. Hobbs. 1978. Precipitation scavenging of aerosol particles. In Proceedings Cloud Physics and Atmospheric Electricity. Boston, Mass.: American Meteorological Society. Raynor, G.S. 1981. Design and preliminary results of the intermediate density precipitation-chemistry experiment. Report BNL 29992. For presentation at Third Joint AMS/APCA Conference on Applications of Air Pollution Meteorology, J.anuary. San Antonio, Tex. Rodhe, H., and J. Grandell. 1972. On the removal time o f aerosol particles from the atmosphere by precipitation scavenging. Tellus 24:442-454. Rodhe, H., and J. Grandell. 1981. Estimates of characteristic times for precipitation scavenging. Atmos. Sci. 38:370-386. Saffman, P.G., and J.S. Turner. 1955. On the collision of drops in turbulent clouds. J. Fluid Mech. 1:16-30. Sampson, P.J. 1980. Trajectory analysis of summertime sulfate concentrations in the northeastern United States. J. Appl. Meteorol. 19:1382-1394. Scott, B.C. 1978. Parameterization of sulfate removal by precipitation. J. Appl. Meteorol. 1375-1389. Scott, B.C. 1981. Sulfate washout ratios in winter storms. J. Appl. Meteorol. 20:619-625. Scott, B.C. 1982. Predictions of in-cloud conversion rates of SO2 to SO4 based upon a simple chemical and kinematic storm model. Atmos. Environ. 16:1735-1752. Scott, B.C., and N.S. Laulainen. 1979. On the concentration of sulfate in precipitation. J. Appl. Meteorol. 18:138-147. J.

372 Scriven, R.A., and B.E.A. Fisher. 1975. The long range transport of airborne material and its removal by deposition and washout. Atmos. Environ. 9:49-68. Semonin, R.G. 1976. The variability of pH in convective storms. In Proceedings First International Symposium on Acid Precipitation and the Forest Ecosystem. USDA Tech. Rept. NE-23, pp. 349-361. Shannon, J. 1981. A regional model of long-term average sulfur atmospheric pollution, surface removal, and net horizontal flux. Atmos. Environ. 5:689-701. Shopauskas, K., B. Styra, and E. Verbal 1969. Spreading and rainout of passive admixture injected into a cloud. In Seventh International Conference on Condensation and Ice Nuclei, Vienna, Austria. Slinn, W.G.N. 1973a. Fluctuations in trace gas concentrations in the troposphere. J. Geophys. Res. 78:574-576. Slinn, W.G.N. 1973b. In-cloud scavenging studies. Annual Report to US AEC/DBER. Battelle-Northwest, BNWL-1751 pt. 1. Slinn, W.G.N. 1974a. Rate limiting aspects of in-cloud scavenging. J. Atmos. Sci. 31:1172-1173. Slinn, W.G.N. 1974b. The redistribution of a gas plume caused by reversible washout. Atmos. Environ. 8:233-239. Slinn, W.G.N. 1977. Some approximations for the wet and dry removal of particles and gases from the atmosphere. J. Water Air Soil Pollut. 7:513-543. Slinn, W.G.N. 1983. Precipitation scavenging. In Atmospheric Sciences and Power Production. D. Randerson, ed. U.S. DOE. Slinn, W.G.N., and J.M. Hales. 1971. A reevaluation of the role of thermophoresis as a mechanism of in- and below-cloud scavenging. J. Atmos. Sci. 28:1465-1471. Slinn, W.G.N., and J.M. Hales. 1983. Wet removal of atmospheric particles. EPA Monograph Series. Squires, P., and S. Twomey. 1960. The relation between cloud droplet spectra and the spectrum of cloud nuclei. In Physics of Precipitation. NAS/NRC Monograph No. 5. Washington, D.C.: American Geophysical Union. Storebo, P.B., and A.N. Dingle. 1974. Removal of pollution by rain in a shallow air flow. J. Atm. Sci. 31:533-542. Summers, P.W., and B. Hitchon. 1973. Source and budget of sulfate in precipitation from Central Alberta, Canada. JAPCA 23:194-199.

373 Thorp, J.M., and B.C. Scott. 1982. Preliminary calculations of average storm duration and seasonal precipitation rates for the northeast sector of the United States. Atmos. Environ. 16:1763-1774. Voldner, E.C., K. Olson, K. Oikawa, and M. Loiselle. 1982. Comparison between measured and computed concentrations of sulfur compounds in eastern North America. J. Geophys. Res. Waldman, J.M., J.W. Munger, D.J. Jacob, R.C. Flagan, J.J. Morgan, and M.R. Hoffman. 1982. Chemical composition of acid fog. Science 218:677-679. Wang, P.K., and H.R. Pruppacher. 1977. An experimental determination of the efficiency which aerosol particles are collected by water drops in sub-saturated air. J. Atmos. Sci. 34:1664-1669. Wangen, L.E., and M.D. Williams. 1978. Elemental deposition downwind of a coal-fired power plant. Water Air Soil Pollut. 10:33-44. Wilkening, K.E., and K.W. Ragland. 1980. Users Guide for the University of Wisconsin Atmospheric Sulfur Computer Model (UWATM-SOX). Report to EPA/Duluth Research Laboratory. Young, J.A., C.W. Thomas, and N.A. Wogman. 1973. The use of natural and man-made radionuclides to study in-cloud scavenging processes. PNL Annual Report for 1972 to U.S. AEC/DBER. BNWL-1751, pt. 1. Young, J.A., T.M. Tanner, C.W. Thomas, and N.A. Wogman. 1976. The entrainment of tracers near the sides of convective clouds. Annual Report to ERDA/DBER. Battelle-Northwest, BNWL-2000, pt. 3. Zishka, K.M., and P.J. Smith. 1980. The climatology of cyclones and anticyclones over North America and surrounding ocean environs for January and July, 1950-77. Mon. Weather Rev. 108:387-401.