the role of non-precipitating clouds in producing ambient sulfate during summer: results from...

12
Atmospheric Environment Vol. 26A, No, 6, pp. 1041-1052, 1992. 0004--6981/92 S5.00+0.00 Printed in Great Britain. © 1992 Pergamon Press pie THE ROLE OF NON-PRECIPITATING CLOUDS IN PRODUCING AMBIENT SULFATE DURING SUMMER: RESULTS FROM SIMULATIONS WITH THE ACID DEPOSITION AND OXIDANT MODEL (ADOM) PRAKASH KARAMCHANDANIa n d AKULA VENKATRAM ENSR Consulting and Engineering, Camarillo, CA 93012, U.S.A. (First received 25 January 1991 and in final form 30 August 1991) Abstraet--A comprehensive acid deposition model was used to investigate the importance of non- precipitating stratus clouds for the production of ambient sulfate. A comparison of model estimates of ambient sulfate and SO2 concentrations with corresponding observations for an episode in the summer of 1988 showed that the model underestimated ambient sulfate concentrations and overestimated ambient SO2 concentrations when non-precipitatingstratus clouds were ignored in the model formulation. When the model was modified to include non-precipitating stratus clouds, a distinct improvement in model performance was obtained. Key word index: Non-precipitating clouds, cloud physics, aqueous-phase chemistry, scavenging, hydrogen peroxide, comprehensive model, model evaluation, Eulerian Model Evaluation Field Study (EMEFS). 1. INTRODUCTION There is overwhelming evidence that aqueous-phase oxidation of SO2 to sulfate makes a major contribu- tion to sulfate in rain. The evidence consists of results from a variety of approaches to understanding the processes that govern sulfate in rain. Theoretical studies (e.g. Saxena and Seigneur, 1986; Hong and Carmichael, 1983; Fung et al., 1991) have demon- strated the importance of the oxidation of SO2 to sulfate in cloud water, where the primary oxidants have been shown to be hydrogen peroxide (H202) and ozone (O3). These theoretical suggestions are sup- ported by results from laboratory experiments (e.g. Penkett et al., 1979) and field measurements (e.g. Daum et al., 1984a, b). Daum et al. (1984a) conducted aircraft measure- ments of the composition of cloud liquid water and interstitial air near Charleston, SC. They found that the relative acidity of cloud water was much greater than that of the interstitial aerosol or of clear-air aerosol samples, indicating the occurrence of in-cloud acid formation. Daum et al. (1984b) measured the chemical composition of cloud liquid water and inter- stitial air in warm non-precipitating stratus and stra- to-cumulus clouds at various locations in the eastern United States. They observed a strong inverse rela- tionship between cloud water H202 and interstitial SO2 concentrations, suggesting that aqueous-phase oxidation of dissolved SO2 by H202 is an important process for the formation of cloud-water sulfate. They also found a greater fractional conversion of SO2 to cloud-water sulfate than of NO2 to cloud-water ni- trate, indicating more extensive in-cloud oxidation of SO2 than of NO2. Studies conducted with semi-empirical models (e.g. Venkatram and Pleim, 1985) have shown that it is necessary to incorporate an empirical parameteriz- ation for the aqueous-phase oxidation of SO2 to explain observations of sulfur in rain. Recently, convincing support has come from comprehensive models such as the Acid Deposition and Oxidant Model (ADOM; Venkatram et al., 1988) and the Regional Acid Deposition Model (RADM; Chang et al., 1987), which incorporate detailed modules for cloud physics and aqueous-phase chemistry. Estimates of sulfur in rain from these models compare well with corresponding observations averaged over events last- ing about 10 days. Sensitivity studies conducted with ADOM (Misra et al., 1989) indicate that in-cloud oxidation of SO2 contributes over 50% of sulfate in rain in summer; the primary aqueous-phase oxidant is believed to be H202. Several theoretical studies (Saxena and Seigneur, 1986; Seigneur and Saxena, 1988) have also demon- strated the importance of aqueous-phase oxidation of SO2 in determining ambient concentrations of sulfate. They show that sulfate formed in clouds, and not removed by precipitation, is left behind in the air when the cloud evaporates. This sulfate can make a contri- bution to ground-level concentrations when it is mixed down to the ground from cloud level. The significance of this pathway for sulfate production becomes clear when we realize that, over most of the year, close to 50% of the sky in the eastern United States is covered by clouds (Saxena and Seigneur, 1986). Over 30% of the sky is covered by non- precipitating stratus, which can convert SO2 to aqueous-phase sulfate, which then becomes airborne sulfate when the cloud evaporates. 1041

Upload: akula

Post on 03-Jan-2017

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Atmospheric Environment Vol. 26A, No, 6, pp. 1041-1052, 1992. 0004--6981/92 S5.00+0.00 Printed in Great Britain. © 1992 Pergamon Press pie

THE ROLE OF NON-PRECIPITATING CLOUDS IN PRODUCING AMBIENT SULFATE DURING SUMMER:

RESULTS FROM SIMULATIONS WITH THE ACID DEPOSITION AND OXIDANT MODEL (ADOM)

PRAKASH KARAMCHANDANI and AKULA VENKATRAM ENSR Consulting and Engineering, Camarillo, CA 93012, U.S.A.

(First received 25 January 1991 and in final form 30 August 1991)

Abstraet--A comprehensive acid deposition model was used to investigate the importance of non- precipitating stratus clouds for the production of ambient sulfate. A comparison of model estimates of ambient sulfate and SO2 concentrations with corresponding observations for an episode in the summer of 1988 showed that the model underestimated ambient sulfate concentrations and overestimated ambient SO 2 concentrations when non-precipitating stratus clouds were ignored in the model formulation. When the model was modified to include non-precipitating stratus clouds, a distinct improvement in model performance was obtained.

Key word index: Non-precipitating clouds, cloud physics, aqueous-phase chemistry, scavenging, hydrogen peroxide, comprehensive model, model evaluation, Eulerian Model Evaluation Field Study (EMEFS).

1. INTRODUCTION

There is overwhelming evidence that aqueous-phase oxidation of SO2 to sulfate makes a major contribu- tion to sulfate in rain. The evidence consists of results from a variety of approaches to understanding the processes that govern sulfate in rain. Theoretical studies (e.g. Saxena and Seigneur, 1986; Hong and Carmichael, 1983; Fung et al., 1991) have demon- strated the importance of the oxidation of SO2 to sulfate in cloud water, where the primary oxidants have been shown to be hydrogen peroxide (H202) and ozone (O3). These theoretical suggestions are sup- ported by results from laboratory experiments (e.g. Penkett et al., 1979) and field measurements (e.g. Daum et al., 1984a, b).

Daum et al. (1984a) conducted aircraft measure- ments of the composition of cloud liquid water and interstitial air near Charleston, SC. They found that the relative acidity of cloud water was much greater than that of the interstitial aerosol or of clear-air aerosol samples, indicating the occurrence of in-cloud acid formation. Daum et al. (1984b) measured the chemical composition of cloud liquid water and inter- stitial air in warm non-precipitating stratus and stra- to-cumulus clouds at various locations in the eastern United States. They observed a strong inverse rela- tionship between cloud water H202 and interstitial SO2 concentrations, suggesting that aqueous-phase oxidation of dissolved SO2 by H202 is an important process for the formation of cloud-water sulfate. They also found a greater fractional conversion of SO2 to cloud-water sulfate than of NO2 to cloud-water ni- trate, indicating more extensive in-cloud oxidation of SO2 than of NO2.

Studies conducted with semi-empirical models (e.g. Venkatram and Pleim, 1985) have shown that it is necessary to incorporate an empirical parameteriz- ation for the aqueous-phase oxidation of SO2 to explain observations of sulfur in rain. Recently, convincing support has come from comprehensive models such as the Acid Deposition and Oxidant Model (ADOM; Venkatram et al., 1988) and the Regional Acid Deposition Model (RADM; Chang et al., 1987), which incorporate detailed modules for cloud physics and aqueous-phase chemistry. Estimates of sulfur in rain from these models compare well with corresponding observations averaged over events last- ing about 10 days. Sensitivity studies conducted with ADOM (Misra et al., 1989) indicate that in-cloud oxidation of SO2 contributes over 50% of sulfate in rain in summer; the primary aqueous-phase oxidant is believed to be H202.

Several theoretical studies (Saxena and Seigneur, 1986; Seigneur and Saxena, 1988) have also demon- strated the importance of aqueous-phase oxidation of SO2 in determining ambient concentrations of sulfate. They show that sulfate formed in clouds, and not removed by precipitation, is left behind in the air when the cloud evaporates. This sulfate can make a contri- bution to ground-level concentrations when it is mixed down to the ground from cloud level. The significance of this pathway for sulfate production becomes clear when we realize that, over most of the year, close to 50% of the sky in the eastern United States is covered by clouds (Saxena and Seigneur, 1986). Over 30% of the sky is covered by non- precipitating stratus, which can convert SO2 to aqueous-phase sulfate, which then becomes airborne sulfate when the cloud evaporates.

1041

Page 2: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1042 P. KARAMCHANDANI and A. VENKATRAM

This paper illustrates the importance of the aque- ous-phase pathway for ambient sulfate production. Specifically, we show that the inclusion of SO 2 oxida- tion in non-precipitating stratus clouds within the framework of ADOM allows us to explain the sulfate concentrations observed during the Eulerian Model Evaluation Field Study (EMEFS) in the summer and fall of 1988. Neglecting the chemistry of these clouds results in an underestimation of ambient sulfate levels.

Before describing the application of ADOM to an EMEFS period, we will provide a brief description of ADOM, concentrating on the treatment of clouds.

2. THE COMPONENTS OF ADOM

ADOM is an Eulerian grid model with a horizontal spacing of approximately 100 km. The vertical grid consists of 12 unequally spaced levels between the surface and the top of the model domain at 10 km. The core of the model is the mass conservation equation, which is solved numerically for 30 species on this grid system. The model incorporates modules for the pro- cesses that govern the fate of acidifying pollutants and oxidants. These processes include transport and dis- persion, gas-phase chemistry, aqueous-phase chem- istry, cloud physics and dry deposition.

The meteorological information used to drive the transport and dispersion module is derived from a diagnostic model, which combines wind fields predict- ed by the Canadian Meteorological Center (CMC) large-scale numerical weather-prediction model with information from a high-resolution boundary-layer model (Scholtz et al., 1986).

The gas-phase chemistry module incorporates the photochemistry that yields the oxidants responsible for converting the emitted SO 2 and NOx to the corresponding acids. It is based on a "lumped mole- cule" approach, consisting of roughly 100 reactions among 50 species, 30 of which are transported by the advection module. The mechanism itself is a con- densation of a detailed mechanism consisting of nearly 300 reactions among 100 species (Lurmann et al., 1986, 1987). The condensed mechanism produces res- ults that compare well with those from the detailed mechanism and smog chamber data. The gas-phase chemistry module also produces the aqueous-phase SO 2 oxidants, hydrogen peroxide and ozone.

The dry deposition module calculates the dry- deposition velocity, which is expressed as the inverse of the sum of three resistances: the atmospheric resist- ance, the deposition-layer resistance and the surface resistance. The atmospheric resistance is a function of micrometeorology, and is thus independent of the species being deposited. The deposition-layer resist- ance is a function of the surface-friction velocity and the molecular properties of the gas or particle being deposited. The surface resistance is a function of many factors, which include biological processes at the receptor surface.

The cloud-physics and aqueous-phase chemistry module is perhaps the most important component of ADOM. In view of its relevance to the results pre- sented in this paper, it is described in some detail in the next section.

3. THE CLOUD-PHYSICS AND AQUEOUS-PHASE CHEMISTRY MODULE

The ADOM scavenging module assumes that all cloud systems consist of stratus and cumulus compon- ents. The physics of the stratus module is governed by large-scale vertical motion, while the cumulus module physics is controlled by subgrid-scale buo- yancy-induced motions in a conditionally unstable atmosphere. The stratiform and convective cloud mo- dules are run in parallel (i.e. independent of each other), using the individual stratiform and convective precipitation amounts, which are externally specified. Both modules are initialized with the same chemical profiles, and the final concentration profiles and wet depositions are the precipitation-weighted averages of the corresponding quantities calculated by the indi- vidual cloud modules. The following sections briefly describe the two cloud modules.

Although each cloud module has its own scav- enging/chemistry component, the basic aqueous- phase processes that are incorporated in the scaveng- ing modules are similar and are described in a separate section. Where appropriate, differences between the aqueous-phase components of the models are pointed out. A detailed description of the stratiform-cloud module is available in Karamchandani et al. (1985), while the cumulus-cloud module is described in Venkatram and Karamchandani (1989).

3.1. Stratiform-cloud module

The stratiform-cloud module is a simplified, one- dimensional, steady-state version of a two-dimen- sional, unsteady-state model developed by Rutledge and Hobbs (1983). The model solves the conservation equations for cloud water, rain water and snow, where cloud ice and snow are treated as a single frozen precipitating species. Air is assumed to be saturated within the cloud so that water vapor is continuously available for conversion to cloud water (via condensa- tion) or snow (vapor deposition), depending upon the temperature. The bulk-water technique (Kessler, 1969) is used for the microphysical formulation. A Marshall-Palmer-type size distribution is assumed for the precipitating fields, while cloud droplets are as- sumed to have a uniform size distribution. The follow- ing microphysical processes are included in the model:

• vapor condensation to cloud water • cloud water autoconversion to rain (collision and

coalescence) • cloud water collection by rain (accretion) • vapor deposition to snow

Page 3: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Non-precipitating clouds and ambient sulfate 1043

• cloud water collection by snow (riming) • evaporation of rain water below cloud base • sublimation of snow below cloud base.

The model is driven by surface precipitation rates. Other required inputs are vertical temperature and pressure profiles, cloud-base and cloud-top elevations and relative humidities below cloud base. The cloud is divided vertically into a small number (typically four) of physically representative zones--each zone is deter- mined by the major microphysical process occurring within it. Steady-state mass-balance equations in each zone lead to a set of coupled, nonlinear algebraic equations. These equations are then manipulated to reduce the system to one nonlinear equation in one variable (other variables are related to this variable). The nonlinear equation is solved numerically using the Regula Falsi method (Carnahan et al., 1969). The output from the model consists of mixing ratio profiles of cloud water, rain water and snow; microphysical process rates; updraft velocities; hydrometeor fall velocities; aerosol scavenging rates; and the mean sizes of raindrops and snow. This information is sub- sequently used in the aqueous-phase chemistry and wet-deposition computations.

Because the vertical resolution of the stratiform cloud module is coarser than the resolution of the regional model (4 vs 12 layers), a mass-conserving and profile-conserving scheme is used when going from one vertical grid system to the other. While this ensures that the vertical profiles of inert pollutants do not change, it can result in an artificial mixing of pollutants such as SOs or H202, which are destroyed by aqueous-phase reactions, or sulfate, which is pro- duced in the aqueous phase.

3.2. Cumulus microphysics and dynamics

The life cycle of the cumulus cloud is modeled in three stages, as shown in Fig. 1. First, the active region is formed by ingestion of air from the cloud sur- roundings. Second, chemical reactions take place in the cloud water formed in the active region. This stage, referred to as the dwell phase, is assumed to make up most of the life-cycle of the cloud. At the end of the dwell phase, scavenging occurs. Finally, in the third stage, the cloud dissipates and the remaining pollut- ants are ejected into the cloudy region. Redistribution of pollutants occurs during both the cloud formation and cloud dissipations stages.

The turbulent mixing processes in the active region of the cloud are modeled using the cloud-mixing model of Raymond and Blyth (1986). The active cloud area consists of a mixture of air lifted from below cloud base and air entrained into the cloud at levels between cloud base and top. The mixing is accomplished by creating a sample of air parcels at each level of the cloud. The thermodynamics and motion of these cloud parcels are determined using the following assumptions:

Cloud dwell phase

chemical reactions

Fig. 1. Schematic of ADOM convective cloud life cycle: am represents the area of one grid cell, and a c represents the area covered by clouds.

• Every cloud parcel consists of a combination of cloud-base air and environmental air entrained into the cloud at different levels within the cloud;

• Air from below cloud base is carried adiabatically to each level in the cloud, where mixing with environmental air in an assumed distribution of proportions occurs, resulting in a number of cloud parcels, each with a unique water content and temperature;

• Following entrainment, cloud parcels move adia- batically in the cloud under the influence of vertical buoyancy accelerations;

• Air parcels exit the cloud at the level where their density is equal to the environmental density.

The creation of air parcels results in an active cloud area that consists of 50% cloud-base air and 50% air entrained from the sides.

The cloud lifetime is proportional to the condensed water lifetime, and varies between 10 min and 1 h.

Page 4: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1044 P. KARAMCHANDANI and A. VENKATRAM

Therefore, between 1 and 6 cloud "cycles" can occur during a 1-h simulation. Clouds with a high ratio of precipitation rate to liquid-water content are assumed to repeat the above cycle more rapidly.

The ADOM mixing model uses cloud-base height and vertical profiles of temperature, pressure and relative humidity as input. The mixing model calcu- lates vertical distributions of the inflow and outflow of air. The inflow profiles are used to determine mixing of pollutants during the cloud-formation stage, while mixing during the cloud-dissipation stage is deter- mined by both the inflow and outflow profiles. The average thermodynamic properties and pollutant con- centrations in the active cloud region are determined from the properties of the environmental air and the cloud base air using the inflow of air into the cloud from each level as the weighting factor.

The average cloud-water mixing ratio is calculated from the equilibritlm properties of an air parcel at a pressure midway between the cloud base and cloud top and an average water vapor mixing ratio and wet equivalent potential temperature equal to that in the active region. The vertical distribution of total con- densed cloud water is estimated from the average cloud-water amount by assuming that the profile can be represented by a sine curve with a maximum at the middle of the cloud and zero cloud-water amounts at the cloud top and base.

The temperature profile within the cloud is de- scribed using a quadratic.polynomial knowing the average temperature and using the constraint that the temperatures at the cloud top and base are equal to the environmental temperatures at those levels. Cloud-ice contents are calculated from the temper- ature profile by freezing a portion of the condensed water load linearly between no ice at 0°C, and all ice at the homogeneous nucleation temperature ( -37°C) .

Note that a cumulus cloud is created only when lifting of an air parcel from cloud base can support convection. Otherwise, the stratiform cloud module, described in the previous section, is used. Because the cumulus cloud module only uses cloud-base height and environmental temperature, pressure and humid- ity profiles to create a cloud, it can treat both precip- itating and non-precipitating clouds. On the other hand, the stratiform-cloud module is driven by surface precipitation rates, and does not treat non-precipitat- ing clouds. Thus, the current version of ADOM ignores cloud processes in those grid cells where a cloud exists (as determined by the specified cloud base and top, and cloud cover), but the precipitation rate is negligible (below 0.005 mm h-1), and the specified environmental profiles cannot sustain a convective cloud.

3.3. Scavengino and chemistry

The scavenging/chemistry components of the strati- form and cumulus cloud modules treat the reversible mass transfer of soluble gases (SO2, H202, organic

peroxides, HNO3, NH3, 0 3 and CO2) to cloud water, irreversible soluble gas scavenging by ice, and irrevers- ible particulate [H2SO4, NH4HSO4, (NH4)2SO4, NH4NO 3 and soil dust] scavenging by cloud water and ice. Nucleation scavenging of soluble aerosols by cloud water is assumed to occur instantaneously.

Both the stratus and cumulus scavenging modules include the oxidation of S rv to S vl by H202, 03, organic peroxides and 0 2 in the presence of trace metals. The rate constant of Kunen et al. (1983) is used for the aqueous-phase oxidation of S w by H20 2. The rate constant for the oxidation of S w by Oa is deter- mined from Maahs (1983). The rate expression for the oxidation of S w by organic peroxides is approximately half of that used by Lind et al. (1987) for the reaction of S w with peroxyacetic acid. The rate constant of Ibusuki and Takeuchi (1987) is used for the catalytic oxidation of S w by ozone in the presence of trace metals such as iron and manganese.

For stratiform clouds, the steady-state vertical dis- tributions of cloud microphysical parameters calcu- lated by the stratus module are provided as input to a time-dependent, one-dimensional cloud chemistry/ scavenging module. The scavenging/chemistry mod- ule calculates the wet depositions and final concentra- tions of soluble and reactive species throughout the grid column. This is done by simulating the chemistry and scavenging in each microphysical zone, described earlier in Section 3.1. The following processes are treated explicitly by the stratiform cloud scavenging module:

• scavenging of aerosols and soluble gases by cloud water

• scavenging of aerosols and soluble gases by rain • transfer of pollutants from cloud water to snow

by riming • transfer of pollutants from snow to rain by

melting • transfer of pollutants from cloud water to rain by

accretion and autoconversion • transfer of pollutants below cloud base to the gas

phase by evaporation of precipitation.

Since no subgrid-scale vertical mixing is allowed in the stratiform cloud model, the concentration profiles of inert species remain unchanged during the stratus model execution.

For the cumulus-cloud model, scavenging in the active region is accomplished by dividing this region into two reactors: a liquid-water reactor and an ice reactor. The liquid-water reactor uses an average liquid-water content and a liquid-water-weighted average temperature. Both aqueous-phase chemistry and mass-transfer processes are allowed to occur in the liquid-water reactor. The ice reactor is created with an average frozen-water content and an average temperature equal to that of the liquid-water reactor. Nucleation scavenging and equilibrium dissolution are the only processes considered in the ice phase.

Page 5: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Non-precipitating clouds and ambient sulfate 1045

After the aqueous-phase processes are simulated for the life cycle of the cloud, pollutants in cloud water are removed via wet deposition. The concentration of a trace gas in rain (Cr) is assumed to be the water- weighted concentration in the cloud water and ice reactors. The deposited amount, D, is:

D = CrPrTcld, (1)

where Pr is the grid-averaged precipitation rate and zcm is the lifetime of air in a cloud, which varies between 10 min and 1 h. Deposition is never allowed to exceed the total mass of trace species initially present between the base and top of the cloud.

4. ADOM EVALUATION

ADOM is currently undergoing an extensive evalu- ation with the EMEFS data base. The EMEFS experi- ment, conducted in 1988, 1989 and 1990, provides a unique data base of surface and aircraft measurements to evaluate regional-scale acid-deposition models. In addition to routine daily surface measurements of air quality (particulate sulfate, nitrate and ammonium; gaseous nitric acid, ammonia, sulfur dioxide, ozone and nitrogen dioxide) and precipitation chemistry (rainfall amount, pH, major acidic and basic ions), surface and vertical measurements of a larger set of variables at finer temporal resolutions were collected for selected periods during the field studies. These include measurements of gaseous SO2, nitric acid, ozone, ammonia, NO2, hydrogen peroxide, speciated hydrocarbons, PAN, formaldehyde and other aldehy- des; particulate sulfate, nitrate and ammonium; and cloud and precipitation chemistry variables. A de- scription of the field-study design is provided by Hansen et al. (1989).

This paper describes results from simulations of a 12-day period (25 August to 5 September 1988). This period is a subset of the first EMEFS intensive meas- urement period, which was conducted to determine the regional redistribution of primary pollutants and to examine the gas-phase chemical and transport pathways and conditions leading to oxidant limita- tion. Summer was chosen because it is typically the period with the greatest photochemical activity (and thus, the highest sulfate and oxidant concentrations) as well as the largest deposition events.

4.1. Base-case results

We begin with a discussion of results from the base- case simulation of the 12-day period mentioned above. Figures 2a to 2d compare model estimates of event- averaged concentrations of sulfur species (ambient sulfate, ambient SO2, total ambient sulfur and sulfur in rain, respectively) with observations made at a height of 10 m. The model estimates of ambient con- centrations correspond to an average over a height of 60 m. Thus, one does expect the observations to

deviate from the model estimates just because of the spatial averaging effect. However, the deviation is not expected to be systematic unless the concentration always decreases or increases with height.

It is seen from Fig. 2 that ambient sulfate concentra- tions are underestimated, and ambient SO2 concen- trations are overestimated by the model. However, model estimates of sulfur concentrations in rain and total ambient sulfur concentrations indicate little bias.

The preliminary conclusions derived from these results were that there were unlikely to be any signific- ant errors in the SO2 emission inventory, the meteoro- logical inputs or the transport component of the model. It appeared that an important mechanism to convert SO2 to sulfate was missing in the model formulation. Possible pathways that were missing included: (a) aqueous-phase free radical reactions, (b) heterogeneous reactions (SO2 oxidation on wetted aerosols) and (c) reactions in non-precipitating stratus clouds. In addition to the above pathways not accoun- ted for in the model, errors and uncertainties in the model formulation could also explain the under- estimation in ambient sulfate concentrations. For example, the reaction rates used for the trace-metal- catalysed oxidation of SOe were uncertain. An under- estimation of OH and H202 concentrations would also lead to underestimates of ambient sulfate concen- trations.

Before we embarked on a detailed investigation of the underestimation of ambient sulfate concentra- tions, a preliminary analysis was conducted to elim- inate some of the possible causes and to prioritize the remaining factors. For example, the OH concentra- tions predicted by the model were examined and found to be reasonable--peak daytime gas-phase SO2 oxidation rates (a measure of the OH concentration) were of the order of 3 -4% h- 1. Because sulfur con- centrations in rain indicated little bias, it appeared reasonable to assume, at least for the conditions of this study, that (a) the model was not consistently under- estimating H202 concentrations, (b) errors due to uncertainties in trace-metal-catalysed oxidation rates were minor and (c) neglecting free-radical aqueous- phase reactions was not a significant source of error.

Based on the preliminary analysis, it appeared that the underestimation of sulfate concentrations was most probably related to the omission of either the heterogeneous pathway or the non-precipitating stratus-cloud pathway. Because of the uncertainties associated with the heterogeneous oxidation process, we deci~ied to investigate the non-precipitating cloud pathway first. An examination of the meteorological fields for the simulated episode revealed that non- precipitating stratiform clouds covered 10-40% of the grid, with an average coverage of about 25%. Further- more, the peak effective SO2 oxidation rates in the aqueous phase were estimated to be of the order of 100% h -1, significantly higher than the gas-phase oxidation rates. Thus, it seemed likely that neglecting non-precipitating stratiform clouds in the model for-

Page 6: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1046 P. KARAMCHANDANI and A. VENKATRAM

Ambient Sulfate Ambient SO 2

i ! o J o t o i

o ~ o ,.-,"~' o i i ,,....,,-" i

" i ................ r ' " " ~ ' ~ .................

• ........ o. ........ i .................... i

o 3 6 12 12 16 2o

Predicted C o n c e n v s t l o n ( l ~ / m 3) Pred ic ted Concen t ra t ion (~g/rn 3)

(a) (b)

Total Ambient Sulfur (as Sulfate) Sulfur in Rain

Predicted ~ n (l~g/m s) Predicted Concentra~on (mg/I)

(c) (d)

Fig. 2. Comparison of ADOM (base-case study) estimates of sulfur species concentrations with observations. Concentrations are averaged over the event (25 August 1988 to 5 September 1988).

mulation would lead to significant underestimates in sulfate formation. The sensitivity studies that were conducted to test this hypothesis are described in the following sections.

4.2. Sensitivity studies

Several sensitivity studies were conducted to invest- igate the possible reasons for the underestimation of ambient sulfate concentrations by ADOM for the EMEFS period. The strategy we adopted was to develop first a linear version of ADOM that only treated the transport and chemistry of SO2 and sulfate. The treatment of gas- and aqueous-phase oxidation of SO, in the model was linearized by specifying the concen- trations of the primary oxidants in the two phases (the OH radical and H20 2, respectively). The main reason

for developing the linear model was that it would require much lower computational resources than the full model and would thus provide the capability of conducting several simulations in a relatively small time period. The linear model could be used to determine the relative importance of the gas- and aqueous-phase SO2 oxidation pathways, and to in- vestigate the importance of non-precipitating stratus clouds. The main features and limitations of the linear model were:

• OH concentrations were specified for gas-phase SO2 oxidation. Diurnal variations in OH concen- trations were allowed. However, there were no daily or spatial variations;

• The H20 2 concentration and pH were specified

Page 7: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Non-precipitating clouds and ambient sulfate 1047

for aqueous-phase SO2 oxidation. There were no diurnal, daily or spatial variations in the H202 concentration or pH. Furthermore, H 2 0 , was not allowed to be depleted as a result of pre- cipitation scavenging or aqueous-phase oxida- tion of SO2.

Because of its limitations, the linear version of ADOM was expected to estimate generally higher gas- and aqueous-phase SO2 oxidation rates and thus, higher sulfate concentrations and lower SO2 concen- trations than the complete model. This was confirmed from the results of a base case study of the selected EMEFS period (25 August 1988 to 5 September 1988). Figures 3a to 3d compare the event-averaged sulfur-species concentration estimates from the linear

model with those from the full model. Note that the linear model estimates of total ambient sulfur concen- trations are in good agreement with the full model estimates (Fig. 3c), even though the linear model com- putes higher concentrations of ambient sulfate and sulfur in rain, and lower ambient SO2 concentrations than the full model.

The fact that the deviations between the full model and linear model results were systematic suggested that we could use the linear model results to predict the possible behavior of the full model for the cases considered in the sensitivity studies. In the first study, the role of gas-phase vs aqueous-phase SO2 oxidation was examined by conducting a study in which the gas- phase oxidation of SO2 was switched off. A com- parison of linear model estimates of ambient sulfate

Ambient Sulfate 10

I' y ,- o~ e ..

' 7 I /1 o , . £ . ~ . .

Full Model Prediction (j~l/msl

( a )

I

j

Ambient SO 2

i t

I

I / ) /~o~, ° t ,,./ '~o+o

o °d~ i b o i 0

+ - i -

5 10 15 2O

Full Model Prediction (,ug/rn "~)

( b )

i: ! :5

Total A m b i m t Sulfur (as Sulfate)

+ l i + 1 L/" I , ,yoo+O i j 1 , !

i 6 t

° '

i 1

• 1 10 20 30

Full Model Prediction (l£glm~

Sulfur in Rain

o ~ o o

• i i ' ,° " 7 ' l ' / 1 1 '

I o

......................... i ........................ i

°o i

i i i

i ....

i t

Full Model Pr~ll,~lon (rno/I)

(c) (d)

Fig. 3. Comparison of ADOM (base-case study) estimates of sulfur species concentrations with linear model estimates. Concentrations are averaged over the event (25 August 1988 to 5 September 1988).

AE(A)26:6-G

Page 8: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1048 P. KARAMCHANDANI and A. VENKATRAM

concentrations with and without gas-phase SO 2 ox- idation is shown in Fig. 4. As expected, estimates of ambient sulfate concentrations without gas-phase SO2 oxidation were lower than base-case estimates. However, we also see that, even in the absence of gas- phase SO2 oxidation, the model produced a signific- ant amount of ambient sulfate. Thus, these results show the importance of the aqueous-phase SO2 oxida- tion pathway for the production of sulfate.

To understand the relationship between the under- estimation of ambient sulfate concentrations and the contribution of aqueous-phase SO2 oxidation to sul- fate production, a scatterplot of the relative error in linear model estimates vs the aqueous-phase contribu- tion is shown in Fig. 5. The relative error is calculated as the difference between the computed and observed concentration normalized by the computed concen- tration. As shown in Fig. 5, 40-90% of the ambient sulfate was produced by aqueous-phase SO2 oxida- tion for the conditions of this study. Thus, SO2 oxidation in both the gas and aqueous phases were found to be important for the formation of sulfate. However, Fig. 5 also shows that aqueous-phase oxida- tion was a major contributor in those cases where ambient sulfate concentrations were greatly under- estimated. This suggested that the model performance for ambient sulfate concentrations could be improved if a mechanism could be found to oxidize more SO2 to sulfate in the aqueous phase.

In the next sensitivity study with the linear model, we examined the role of non-precipitating stratus clouds. As described in Section 3.2, ADOM does not simulate cloud physics and chemistry for grid cells having stratus clouds with negligible precipitation rates. A stratus cloud is assumed to be non-precipitat- ing when the grid-average precipitation rate is below 0.005 mm h - 1. Although this cut-off precipita- tion rate is somewhat arbitrary, model estimates of sulfur and nitrate concentrations in rain have com- pared well with observations (Venkatram et al., 1988),

i0

0

| / ~ 8 8~

°o o ®~ ~ o

o~

o o o o o

o o

o o

o

• 1 . i 4

With GasoPhase SO 2 Oxidat ion (pg/m 3)

/ o

Fig. 4. Comparison of linear model estimates of ambient sulfate concentrations with and without gas-phase SO 2 oxidation. Concentrations are averaged over the event (25 August 1988 to 5

September 1988).

I °

:,..n

.N -11111

w

0

i o j o I~

r r o / o o

! r o°j;° °

i 1

~ o ~o

o o o

c

'o o ~

o o

o

, o

8 o g o

A q u e o u s - P h u e Contribution to Ambient Sulfate (%)

Fig. 5. Relative error in linear model estimates of ambient sulfate concentrations vs the aqueous-phase contribution to ambient sulfate. Concentrations are averaged over the event (25 August 1988 to 5 September 1988).

suggesting that this assumption has a small impact on wet deposition predictions.

To investigate the importance of non'-precipitating stratus clouds for ambient sulfate concentrations, the code was modified by setting the precipitation rate to 0.005 mm h - t for grid cells covered by non-precipitat- ing stratus clouds. Removal by wet deposition was prevented by switching off rain scavenging and the transfer of pollutants from cloud droplets to pre- cipitation. Thus, the only processes allowed were cloud scavenging and aqueous-phase chemistry within clouds.

Figure 6a compares ambient sulfate observations and linear model estimates with and without non- precipitating stratus clouds. It is clear that non- precipitating stratus clouds play an important role, at least for the conditions of this study. The linear model underestimates ambient sulfate concentrations with- out non-precipitating stratus clouds, and overestim- ates sulfate concentrations when non-precipitating stratus clouds are included. Figure 6b shows that the converse is true for ambient SO2 concentrations--the linear model overestimates SO2 concentrations with- out non-precipitating stratus clouds and underestim- ates SO2 concentrations with non-precipitating stra- tus clouds. Notice that including non-precipitating stratus clouds in the model formulation has a rela- tively small impact on total ambient sulfur concentra- tions or sulfur concentrations in rain, as shown in Figs 6c and 6d, respectively.

A likely explanation for the small impact on sulfur concentrations in rain is that a precipitating cloud scavenges both SO2 and sulfate effectively (subject to the availability of aqueous-phase oxidants such as H202). Thus, sulfur concentrations in rain reflect the total (SO 2 +sulfate) ambient sulfur concentrations. When non-precipitating clouds are included, ambient SO2 concentrations decrease and ambient sulfate con- centrations increase, resulting in a relatively minor change in total ambient sulfur concentrations.

Page 9: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Non-precipitating clouds and ambient sulfate 1049

Ambient Sulfate

=1 ' " ~ ~ !]'+] ~ ../,~ s' o o • • I /

0

UnNr Model Prediotlon (l~/m =)

Ambient SO=

° ' ' ~ " I / / /

i o +~°~ , ............ -.. 4~_.+.;, L/y/- o

• # i > l i l ~ O o "L" I ..... • - . - . , : ~ i , , - ~ ~ - . ° . . . . - . . ~ . . . . . . . . . . .

°-" eZ.tl=g lt,,d,,.~."" o °

Unur k~xiol Prodlotlon (l~g/m s)

(a) (b)

Total Ambient Sulfur (as Sulfate) 40 10

II o - ' - . ~ - - I A I I _ t1" " ~ " " I / 1 I /

........................ I ................................................... +i + 0 Q ...................... 6.

iiiiiiiii ............ 4

l

0

Unear Model PredL-ton (ttg/m s)

Sulfur in Rain

• w n ~ o l p . ~ I

il

i~ - o~,~'L oo,

Unmr Modol Prediction ling/l) | m i " i " '

(c) (d)

Fig. 6. Comparison of linear model estimates of sulfur species concentrations with observations. The hollow circles represent model estimates without non-precipitating stratus clouds, while the filled circles represent model estimates with non-precipitating stratus clouds. Concentrations are averaged over the event (25 August

1988 to 5 September 1988).

Since the linear model generally computes higher ambient sulfate concentrations and lower ambient SO2 concentrations than the full model, these results suggested that including non-precipitating stratus clouds in the full model would result in improved model performance for ambient sulfate and SOz. This hypothesis could only be confirmed by conducting a similar simulation with the complete model.

4.3. Results from the complete model

The complete model was modified to include non- precipitating stratus clouds using the procedure de- scribed in the previous section. The model was then used to simulate the base-case study period (25

August 1988 to 5 September 1988). Figure 7 com- pares observed event-averaged sulfur species concen- trations with model estimates. Two model estimates are shown: (a) results from the base case study (with- out non-precipitating stratus clouds) and (b) results from the sensitivity study (with non-precipitating stra- tus clouds). As can be seen in Fig. 7a, there is a distinct improvement in model performance for ambient sul- fate concentrations. Figure 7b shows that including non-precipitating stratus clouds also improves the model performance for ambient SO2 concentrations, previously overestimated by the model. As with the linear model, including non-precipitating stratus clouds does not appreciably change ADOM estimates

Page 10: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1050 P. KARAMCHANDANI and A. VENKATRAM

Ambient Sulfate Ambient SO 1

'~T~-'~--~V-/ 7 i o---~'-- / i / / i

• o • i • ! , I . . IAL: i T ...................... ~ 7 " : ~ ' . ' g " ~ .............. i I 7_. 0°_ . / : . i" i t . ,:'3~. ~. " _"i . ~ , o ...................... I.' ............... i i ' / l ~ . ' " i - - ' - ' - °

.............. , | I / o / , , I i / ' i f ' . . i 1 • •

8 7. ~.'~-c" • i ° i o

°h 4 ~ l~ o s lo is ~o

Full Model Prediction O~/m 3) Full Model Prediction (j1g/m 3)

(a) (b)

Total Ambient Sulfur (as Sulfate)

4O

-7 ~ / i i / 1 ~ / i . / . ; i i ~ i° i

e~o !/ - ~. _.3,r i , t

t,o / _ ~ _ ' i l " " j...-, i i : < ' % " ~ , z

6 " l o I b " 3o " 4o

Full Model P r ~ ( j lg /m~

10

~ 8 .

1 4

0

S u l f u r i n R a i n

- - m ' - - I i / i i / i ............ I - o , . .......... g ................. ~ .................. L ............... i

. . . . . . . . . . . . . . . . . . . i . . . . . . . . . . . . . . . . . . . .

* i i i . i . . !

2 4 6 10 Fun I~kalel P r ~ n (m~)

( c ) ( d )

Fig. 7. Comparison of ADOM estimates of sulfur species concentrations with observations. The hollow circles represent model estimates without non-precipitating stratus clouds, while the filled circles represent model estimates with non-precipitating stratus clouds. Concentrations are averaged over the event (25 August 1988 to,

5 September 1988).

of total ambient sulfur concentrations (Fig. 7c), or sulfur concentrations in rain (Fig. 7d).

Although these results are encouraging, it is import- ant to remember that the nominal precipitation rate (0.005 mm h - 1) assigned to non-precipitating stratus clouds is arbitrary. To investigate the importance of the nominal precipitation rate used for non-precipitat- ing stratus clouds, two sensitivity studies were conduc- ted with the full version of ADOM. In the first study, a nominal precipitation rate of 0.001 mm h - 1 was speci- fied. In the second study, non-precipitating stratus clouds were assigned a nominal precipitation rate of 0.1 mm h - 1. These precipitation rates, which spanned the base-case rate of 0.005 mm h -1, allowed us to

investigate the effects of a factor of 100 variation in the nominal precipitation rate on model estimates. The results from these studies were very encouraging-- ambient sulfate concentrations computed by the model for the two precipitation rates were within 3% of each other. The average absolute relative change between model estimates of sulfate concentrations for the two studies was 0.5%. Similarly, the relative change in model calculations of ambient SOz concen- trations for the two studies was negligible--the max- imum relative change was less than 3%, and the average absolute relative change was 0.7%.

The results of the sensitivity studies indicate that the conversion of SO2 to sulfate in non-precipitating

Page 11: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

Non-precipitating clouds and ambient sulfate 1051

stratus clouds is only weakly dependent on the cloud- water content of the clouds; it is total cloud cover (or the presence of cloud) that matters. This is consistent with our understanding of the aqueous-phase conver- sion of SO2 to sulfate. Because H202 is the dominant aqueous-phase oxidant, the rate of conversion is rapid and independent of the pH of the cloud droplets. This means that the amount of sulfate formed in a cloud droplet, whose lifetime is large compared to the inverse of the reaction rate, is independent of the cloud-water content. The cloud water just brings the SO2 and H202 together to react. The amount of sulfate formed is proportional to the smaller of the SO2 and the H202 concentration. This hypothesis is also consistent with the measurements of Daum et al. (1984b), who found that H202 and SO2 were essen- tially mutually exclusive in non-precipitating, liquid- water, stratiform clouds, indicating that the reaction between the two species has proceeded to completion in such clouds.

5. SUMMARY

Base case simulations of an Eulerian Model Evalu- ation Field Study (EMEFS) period with a compre- hensive model, the Acid Deposition and Oxidant Model (ADOM), showed that the model was under- estimating ambient sulfate concentrations and over- estimating ambient SO2 concentrations. However, there was no apparent bias in model estimates of total ambient sulfur concentrations or sulfur concentra- tions in rain. These results indicated that an important mechanism to produce sulfate from the oxidization of SO2 was missing in the model. Alternatively, incorrect treatment of some of the processes in ADOM could have ied to the discrepancies in the sulfate estimates. While it might have been useful to investigate all the possible reasons for the underpredictions of ambient sulfate concentrations, we chose to examine the sim- pler hypothesis that a mechanism for converting SO2 to sulfate was missing in the model. Sensitivity studies with a linear version of ADOM suggested that the missing mechanism was SO2 oxidation in non-precip- itating stratus clouds.

These results were confirmed with simulations from the comprehensive model, which was modified to treat non-precipitating stratus clouds. The resulting model estimates of ambient sulfate and SO 2 concentrations were found to agree well with observed concentra- tions, Model estimates of total ambient sulfur concen- trations and sulfur concentrations in precipitation were very similar to the base-case estimates. These results suggest that non-precipitating clouds can play a crucial role in converting SO2 to sulfate during typical summertime conditions.

Acknowledgements--This work was supported by the Elec- tric Power Research Institute, Environment Canada and the Ontario Ministry of the Environment. The authors also

thank the referees for their comments and suggestions on improving the paper.

REFERENCES

Carnahan B., Luther H. A. and Wilkes J. O. (1969) Applied Numerical Methods. John Wiley, New York.

Chang J. S., Brost R. A., Isaksen I. S. A., Madronich S., Middleton P., Stockwell W. R. and Walcek C. J. (1987) A three-dimensional Eulerian acid deposition model: phys- ical concepts and formulation. J. oeophys. Res. 92, 14,681-14,700.

Daum P. H., Schwartz S. E. and Newman L. (1984a) Acidic and related constituents in liquid water stratiform clouds. J. oeophys. Res. 89, 1447-1458.

Daum P. H., Kelly T. J., Schwartz S. E. and Newman L. (1984b) Measurements of the chemical composition of stratiform clouds. Atmospheric Environment 18, 2671- 2684.

Fung C. S., Misra P. K., Bloxam R. and Wong S. (1991) A numerical experiment on the relative importance of H202 and 0 3 in aqueous conversion of SO2 to SO 2-. Atmo- spheric Environment 2.5A, 411-423.

Hansen D. A., Barnes H. M., Lusis M. and Puckett K. J. (1989) A North American field study to evaluate Eulerian models. In Air Pollution Modeling and its Application VII (edited by Han van Dop), pp. 297-306. Plenum Press, New York.

Hong M. S. and Carmichael G. R. (1983) An investigation of sulfate production in clouds using a flow-through chemical reactor model approach. J. geophys. Res. 88, 10733-10743.

Ibusuki T. and Takeuchi K. (1987) Sulfur dioxide oxidation by oxygen catalyzed by mixtures of manganese (II) and iron (III) in aqueous solutions at environmental reaction conditions. Atmospheric Environment 21, 1555-1560.

Karamchandani P., Lurmann F. and Venkatram A. (1985) ADOM/TADAP model development program, Volume 8---Central operator. ENSR Document PB866-450, ENSR Consulting and Engineering, Camarillo, CA.

Kessler E. III (1969) On the distribution and continuity of water substance on atmospheric circulations. Meteor. Monoor. 32, 1-84.

Kunen S. M., Lazrus A. L., Kok G. L. and Heikes B. G. (1983) Aqueous oxidation of SO 2 by hydrogen peroxide. J. oeophys. Res. 88, 3671-3674.

Lind J. A., Lazrus A. L. and Kok G. L. (1987) Aqueous phase oxidation of sulfur (IV) by hydrogen peroxide, methyl- hydroperoxide, and peroxyacetic acid. J. geophys. Res. 92, 4171-4177.

Lurmann F. W., Lloyd A. C. and Atkinson R. (1986) A chemical mechanism for use in long-range transport/acid deposition computer modeling. J. geophys. Res. 91, 10,905-10,936.

Lurmann F. W., Carter W. P. L. and Coyner L. A. (1987) A surrogate species chemical reaction mechanism for urban- scale air quality simulation models, Vols I and II. EPA Contract No. 68-02-4104, U.S. Environmental Protection Agency, Research Triangle Park, NC.

Maahs H. A. (1983) Kinetics and mechanism of the oxidation of S(IV) by ozone in aqueous solution with particular reference to SO2 conversion in nonurban tropospheric clouds. J. geophys. Res. 88, 10,721-10,732.

Misra P. K., Bloxam R., Fung C. and Wong S. (1989) Non- linear response of wet deposition to emissions reduction: a model study. Atmospheric Environment 23, 671-687.

Penkett S. A., Jones B. M. R., Brice K. A. and Eggleton A. E. J. (1979) The importance of atmospheric ozone and hydrogen peroxide in oxidizing sulfur dioxide in cloud and rainwater. Atmospheric Environment 13, 123-137.

Raymond D. J. and Blyth A. M. (1986) A stochastic mixing model of nonprecipitating cumulus clouds. J. atmos. Sci. 43, 2708-2718.

Page 12: The role of non-precipitating clouds in producing ambient sulfate during summer: Results from simulations with the Acid Deposition and Oxidant Model (ADOM)

1052 P. KARAMCHANDANI and A. VENKATRAM

Rutledge S. A. and Hobbs P. V. (1983) The mesoscale and microscale structure and organization of clouds and pre- cipitation in mid-latitude cyclones. VIII: A model for the seeder-feeder process in warm-frontal rainbands. J. atmos. Sci. 40, 1185-1206.

Saxena P. and Seigneur C. (1986) The extent of nonlinearity in the atmospheric chemistry of sulfate formation. J. Air Pollut. Control Ass. 36, 1151-1154.

Scholtz M. T., Weisman B., Mahrt L. and Christie A. D. (1986) Generation of meteorological data fields for the ADOM Eulerian regional model. Fifth Joint AMS/APCA Conf. Applications of Air Pollution Meteorology, Chapel Hill, NC, pp. 145-150.

Seigneur C. and Saxena P. (1988) A theoretical investigation of sulfate formation in clouds. Atmospheric Environment 22, 101-115.

Venkatram A. and Karamchandani P. (1989) The ADOM II scavenging module: incorporation of an improved cu- mulus cloud module. ENSR Document 0780-004-205, ENSR Consulting and Engineering, Camarillo, CA.

Venkatram A. and Pleim J. (1985) Analysis of observations relevant to long-range transport and deposition of pol- lutants. Atmospheric Environment 19, 659-667.

Venkatram A., Karamchandani P. K. and Misra P. K. (1988) Testing a comprehensive acid deposition model. Atmo- spheric Environment 22, 737-747.