CHAPTER 4 Contents



The Use of Atmospheric Modelling in Policy Development 130

Using Models to develop Air Toxic Reductions Strategies: Lake Michigan as a Test Case

4.3.1 Goals and Structure 131

4.3.2 Description of Applied Models 132

Dr. Mark Cohen

Dr. Frank Wania

Dr. Trevor Scholtz

Ying-Kuang Hsu and Dr. Tom Holsen

Dr. Keri Hornbuckle

4.3.3 Estimated Extent of Deposition of Selected Contaminants to Lake Michigan 135

4.3.4 Source Identification The Need for Enhanced Emission Inventories 144

4.3.5 Extent of Urban Areal Emissions for PCBs Chicago, Baltimore and the Chesapeake Bay 145

4.3.6 Actions Supported by the Lake Michigan Models 146

4.3.7 Options for Estimating Emissions 147




4.6.1 Core Findings 151

4.6.2 Recommendations 151

4.7 BOARD MEMBERSHIP FOR 1999-2001 153



I. Significant Activities since 1987 under Annex 15 of the Great Lakes Water Quality Agreement 155


1. Invited Presentations
Using Models to Develop Air Toxics Reduction Strategies: Lake Michigan as a Test Case

2. Enhanced Urban Concentrations of Atmospheric PCBs 142

3. Physical and Chemical Properties of Pesticides 143

4. Potential Sources and PCB Concentrations Upwind and Downwind 148

5. Estimates of the Percent of Lake Michigan Dioxin Loadings Attributable
to the Atmospheric Deposition Pathway

6. Critical Pollutants 155


a. The POPCYCLING-Baltic Model 133

b. Summary of inputs, outputs and main modules of PEM 134

c. The Potential Source Contribution Function (PSCF) 134

d. Stochastic Fractionation 135

e. Estimated 1996 Per-Capita PCDD/F Emissions (g TEQ/person-yr) 135

f. Total Dioxin Emissions for 1996 136

g. Estimated Contributions to the 1996 Atmospheric Deposition of Dioxin
to Lake Michigan (µgrams TEQ/km 2 -yr)

h. Percent of Total Emissions or Total Deposition of Dioxin (1996) from
Within Different Distance Ranges from Each of the Great Lakes

i. Approximate range of uncertainties in estimating
total 1996 PCDD/F deposition to Lake Superior

j. Effect of fate simulation variations on the geographical pattern of
deposition contributions to Lake Michigan

k. Effect of fate simulation variations on the relative contribution of different
source categories to 1996 atmospheric deposition of dioxin to L. Michigan
(for top 12 categories)

l. Net Gas-Exchange of GPCBs 140

m. Averaged PCB Concentrations over Time (ng/m 3 ), Chicago and
Two Representative Background Sites

n. Temperature Dependence of Gaseous PCBs 142

o. PCB `Footprints' 145

p. Monthly whole-lake gross deposition of PCB 31 + 28 and trans-nonacholor 146

q. PSCF Plots Showing Potential Source Regions as Squares 147

r. Comparison of Model Predictions with Ambient Measurements 149


The International Joint Commission (IJC) and its advisory boards have made major contributions during the past two decades to the understanding of atmospheric deposition of persistent toxic substances. Recently, they have focussed their efforts on the application of transport and deposition modelling for the identification of sources with emissions that must be reduced if Great Lakes water quality goals are to be met. Significant sources lie within and outside the basin and are a mix of large discrete point sources, such as waste incinerators and metallurgical processes, and areal sources, such as landfills, wastewater treatment plants and other small but cumulative urban and rural sources where quantification is in its infancy. This report, along with supporting documentation, provides guidance and encouragement to governments and others to extend the application of the described science to the identification and, as necessary, control of sources of those persistent toxics substances that threaten the health of the Great Lakes and its inhabitants.

During the 1999-2001 priority work cycle, the International Air Quality Advisory Board (IAQAB) and the Great Lakes Science Advisory Board (SAB) held two workshops, in cooperation with the Delta Institute, focussing on the capability of atmospheric models to support the development of policies, including source control strategies, by confirming deposition trends and identifying significant sources of persistent contaminants.

The breadth of science discussed at the workshops extended well beyond what is discussed in this chapter. A CD-ROM titled Atmospheric Deposition: Science and Policy , containing background papers, presentation materials and a record of the related discussions, is available on request from the Great Lakes Regional Office of the IJC.

The workshop held in Milwaukee, Wisconsin in November 2000, drew upon the findings of the first workshop in Ann Arbor, Michigan and focussed particularly on Lake Michigan. Background papers from leading scientists were reviewed by government policy personnel, academics, consultants and industrial and public interest groups. The Delta Institute also presented a draft strategy, which has since completed, on steps to reduce toxic air pollution to Lake Michigan. The strategy is available on the Delta Institutes web site.

Recommended actions from the IAQAB include: completion of the Lake Michigan Mass Balance Study for pathways other than atmospheric deposition; extension of that Mass Balance to other contaminants; mprovement of emission inventories, particularly for point and areal dioxin sources within 100 km of the Lake Michigan basin and for dominant areal, and largely unquantified sources of PCBs and other banned contaminants; development of a predictive, first estimate model for areal urban emissions of banned contaminants; use of models to estimate emissions of residual banned pesticides from agricultural practices; and the continuation and extension of enhanced ambient measurement schemes to better estimate areal and regional loading and support model verification.

The two workshops were held to further examine issues contained in the Great Lakes Water Quality Agreement. Annex 15: Airborne Toxic Substances, committed the governments of the United States and Canada, the parties to the Agreement, to the reduction of atmospheric deposition of toxic substances, particularly persistent toxic substances, to the Great Lakes basin ecosystem through research, surveillance and monitoring, and ultimately, the implementation of additional pollution control measures. Models to determine the significance of atmospheric loadings to the Great Lakes system, relative to other pathways, and the sources of such substances from within and outside the Great Lakes system, also were to be developed.

The Integrated Atmospheric Deposition Network (IADN) was established in the Great Lakes basin as a direct response to Annex 15 and built upon the expertise available under the IJC for its design. The network was designed to: 1. determine atmospheric loadings of toxic substances by quantifying their total and net atmospheric input to the Great Lakes system; 2. define the temporal and spatial trends in the atmospheric deposition in the basin; and 3. support development of Remedial Action Plans and Lakewide Management Plans pursuant to Annex 2 of the Agreement.

Additionally, the Parties, in cooperation with state and provincial governments, committed to develop and implement measures to control emission of toxic substances and eliminate the sources of persistent toxic substances in cases where the atmosphere is a significant contributor to the Great Lakes system. The governments were to review their progress in implementing this annex and report to the IJC biennially, commencing with a report no later than December 31, 1988.

Appendix I contains a history of significant activities that have taken place since 1987 under Annex 15: Airborne Toxic Substances of the Great Lakes Water Quality Agreement. This history provides helpful background information leading up to the current work of the IAQAB.


In July 2000, the Delta Institute, IAQAB and SAB collaborated on a workshop entitled The Use of Atmospheric Modeling In Policy Development in Ann Arbor, Michigan. Discussion focussed on the quantity and quality of model results necessary for use in initiating or modifying policy affecting the transport and deposition of persistent toxic substances. There was agreement that models provide a useful and necessary contribution to the scientific understanding of atmospheric deposition of toxic substances and, as reflected in the evolution of the U.S. ozone regulations, could play a prominent role in policy formulation. However, from the perspective of the decision makers, the issue of model capability has not been adequately addressed.

The Ann Arbor workshop reiterated several needs identified in earlier Delta Institute and IJC events including: chemical speciation of emissions and other improvements in emission inventories; further fundamental research into the physical and chemical properties of persistent toxic substances; use of appropriate meteorological data; and additional ambient measurements to verify model outputs. In some cases, finer spatial scale resolution in meteorological databases to account for urban and agricultural plumes and the subsequent volatilization and transport of previously deposited contaminants from the lakes to the atmosphere, the grasshopper effect, would be needed. More research on large particle transport and scavenging and deposition processes also were seen as necessary.

Policy makers noted that they operate in a complex environment where other models, such as those associated with risk analysis and socioeconomic factors, enter into the decision-making process. Localized air pollution modelling associated with the permitting of single sources was familiar, but further exposure to regional- and continental-scale modelling was necessary for such tools to gain acceptance in toxics management. It was felt that an understanding of what is behind an estimated deposition number and the strengths and weaknesses associated with that estimate would be required before model outputs would be prominent in the formulation of persistent toxic reduction strategies. Some further linkage among these larger-scale deposition models and risk analysis and socio-economic models would also be desirable.

Workshop participants concluded that clarifying and ranking questions with the most urgent policy implications, as well as establishing the degree of scientific certainty required of models used to support policy development, were necessary exercises among a broad base of stakeholders. Database developers, model developers, model users, policy makers and the affected communities should be brought together to determine together how to better link data collection, modelling and policy activities.


4.3.1 Goals and Structure

The second workshop, Using Models to Develop Air Toxics Reduction Strategies: Lake Michigan as a Test Case , sponsored by the Lake Michigan Forum, Delta Institute, IAQAB and the SAB, was held November 8 - 9, 2000 in Milwaukee, Wisconsin.

The objectives of the workshop were to:

• examine trends in air toxics deposition using several different models;

• identify strategic reduction opportunities based on these scientific findings; and

• develop a policy framework for reducing air toxics deposition to the Lake Michigan region as a test case.

The rich database available from the Lake Michigan Mass Balance Study was used by workshop participants to illustrate the use of models in support of policy strategies, explain the uncertainties associated with existing models, and identify policy strategies applicable to the Lake Michigan basin and relevant to other lake basins.

At the workshop, presentations from leading researchers and modelers were followed by discussion of the policy implications of their work. Participants included representatives of municipal, state and provincial governments, the U.S. and Canadian governments, universities, consultants, industry and environmental groups. Included were researchers, modelers, regulatory personnel, policy makers and advocates for industry and the public interest. Representatives from the Lake Michigan Forum, IAQAB, Great Lakes Commission and U.S. EPA staff working on lakewide management plans, the Great Waters program, the Lake Michigan Mass Balance Study and Total Maximum Daily Loads were among those participating.

To provide background in the current science, the workshop sponsors commissioned papers on various aspects of atmospheric deposition from prominent scientists in the field. Included were unpublished results of modelling studies applicable to the Great Lakes basin, particularly Lake Michigan, as well as relevant information from studies elsewhere. In presenting these papers, researchers were asked to emphasize the policy and source control implications of their model findings and to address uncertainties and resource requirements associated with the application of their models. A list of the presentations is provided in Table 1.

A second element of the workshop focussed on the development of a preliminary strategy to address atmospheric deposition of persistent toxic substances to Lake Michigan for use in the Lakewide Management Plan. Prior to the workshop, a draft strategy prepared by the Delta Institute was distributed to participants. The draft strategy evaluated the potential of existing state and federal control approaches to respond to information from atmospheric models, proposed a strategy for using policy tools in the Lake Michigan basin and identified unresolved policy questions to which models could respond.

Recommended actions from the Delta Institute strategy included:

• creation of an adequate monitoring network and comprehensive emission inventories; enhancement of regional modelling efforts;

• examination of the implications of urban air toxics initiatives;

• application of environmental management systems;

• extension of pollution prevention techniques to agricultural practices;

• consideration of a total maximum daily load (TMDL) calculation for Lake Michigan;

• targeted emission reductions from federal facilities; and,

• integration of reduction targets into energy policies.

Table 1

Invited Presentations: IAQAB/Delta Institute, Milwaukee Workshop

Using Models to Develop Air Toxics Reduction Strategies: Lake Michigan as a Test Case

The Transport and Deposition of Dioxin to Lake Michigan: A Case Study Mark Cohen NOAA Air Resources Research Laboratory, Silver Spring, Maryland
Lessons from Modeling Contaminants in Other Large Water Bodies: Identifying Origin and Time Responses of HCHs in the Baltic Sea Frank Wania Wania Environmental Chemists Corp and Division of Physical Sciences, University of Toronto at Scarborough Toronto, Ontario
A Modeling Assessment of the Impact of Pesticide Application Methods and Tilling Practices on Emissions to the Atmosphere M. Trevor Scholtz Canadian Global Emissions Interpretation Centre, Mississauga, Ontario
Exchange of Atmospheric Chemicals with Urban Surface Waters: Controls on Long-Term Response Times Joel Baker Chesapeake Biological Laboratory, University of Maryland, Solomons, Maryland
The Use of Receptor Models to Locate Atmospheric Pollutant Sources: PCBs in Chicago Ying-Kuang Hsu Department of Civil and Environmental Engineering, Clarkson University, Potsdam, New York
Polychlorinated Biphenyl Emissions to Urban Atmospheres: Enhanced Concentrations, Atmospheric Dynamics and Controlling Processes Steven Eisenrich Department of Environmental Sciences, Rutgers University, New Brunswick, New Jersey
The Impact of Chicago on Lake Michigan: Results of the Lake Michigan Mass Balance Study Keri Hornbuckle Department of Civil and Environmental Engineering, University of Iowa, Iowa City, Iowa

It was recommended that these and other approaches be considered under the Binational Toxics Strategy and during the reauthorization of the U.S. Clean Air Act.

During the workshop, facilitated small group discussions focussed on the following questions.

• What are the most important ideas (policy and science) learned or heard to that point in the workshop?

• What are initial reactions to strategy elements laid out in the draft strategy?

• Are there any key ideas or elements that are missing?

• What would be the best way to track progress on specific aspects of the strategy or on the strategy as a whole?

Based on feedback provided during the workshop, the Delta Institute refined its strategy. It focuses on generic recommendations for actions necessary to provide information to support policies and programs for the reduction of atmospheric transport and deposition of persistent toxic substances into the Great Lakes from local, regional, national and international sources. The strategy is available on the Delta Institute web site.

4.3.2 Description of Modellers

and their Applied Models

Dr. Mark Cohen

Dr. Mark Cohen outlined his use of a modified version of the NOAA HYSPLIT-4 (Hybrid Single Particle Lagrangian Integrated Trajectory) model to simulate atmospheric fate and transport of dioxin from sources in the U.S. and Canada to Lake Michigan and the other Great Lakes. In this model, puffs of pollutant are emitted from user-specified locations and advected, dispersed and subjected to destruction and deposition phenomena. Using meteorological output from NOAA's Nested Grid Model, the dioxin model simulates vapor and particle partitioning, wet and dry deposition, reaction with the hydroxyl radical and photolysis during the transport and deposition process.

With the support of the IAQAB and U.S. and Canadian environmental agencies, Dr. Cohen and his former coworkers at the Center for the Biology of Natural Systems, SUNY Queens College, developed a dioxin emission inventory based on U.S. EPA and Environment Canada inventories. This included additional source specific and contaminant speciation information and estimates on several other source categories including residential waste (backyard) burning. The current inventory contains over 5,700 point sources. Area sources, such as mobile sources and backyard burning, were estimated at the county level in the U.S. and within 50 or 100 km grids in Canada.

Dr. Frank Wania

The POPCYCLING-Baltic model used by Dr. Frank Wania describes an entire regional environment, including the atmosphere, marine system and terrestrial system, and estimates the fractions of persistent toxic substances in various `well mixed' environmental compartments and the fluxes or exchanges between them. Contaminants are associated with recent and past releases in the drainage basin and air masses advected into the region. The model integrates information on partitioning, transport and transformation to estimate distribution of the contaminant within this simplified system.

Key processes include exchanges between the atmosphere and aquatic and terrestrial surfaces, and runoff from soil via fresh water to the sea. The Baltic Sea environment is described in the model using four atmospheric, 16 marine and 15 marine sediment components, as well as 10 drainage basins composed of five compartments each (forest canopy, forest soil, agricultural soil, fresh water, fresh water sediment), as

shown in Figure a . As this model is meant to calculate long-term trends (on the scale of years to decades), it employs average monthly values for atmospheric transport rates, temperature, wind speed and hydroxyl radical concentration. Other environmental parameters, in particular those relating to water and organic carbon cycling in the Baltic Sea, are assumed fixed in time. The 85 differential mass balance equations are solved in a step-wise fashion with a finite difference approximation using a time step of six hours.

Dr. Trevor Scholtz

Dr. Trevor Scholtz used the Canadian Global Emission Interpretation Centre's Pesticide Emissions Model (PEM) to derive theoretical emissions, over a three-year period, of twenty pesticides applied to soil by various means including incorporation, surface spraying and in-furrow application at time of planting. PEM, which solves for moisture and pesticide concentration and the advection and diffusion of heat in agricultural soils, is driven by hourly meteorological data.

Figure b . shows the main modules and the input data requirements of the PEM model. A relatively large number of soil levels, 45 variably spaced levels over a one-metre soil depth, is utilized to define the pesticide concentration profile for computation of the volatilization rate. At the surface, PEM is coupled to the adjacent atmospheric layer through a surface energy balance. Sensible and latent heat fluxes are modeled using similarity theory for the atmospheric surface layer, while radiative heat fluxes are estimated using a simple model employing incoming solar radiation at the ground surface. PEM is also coupled to a modified `big leaf' canopy submodel that includes interception of post-emergent spray by the canopy as well as subsequent volatilization and/or wash off during precipitation events. The time dependent, one-dimensional governing equations for heat, moisture and pesticide concentration are solved using a finite element technique with a time step of 1,200 seconds.

Ying-Kuang Hsu and Dr. Tom Holsen

Ying-Kuang Hsu and Dr. Tom Holsen used the Potential Source Contribution Function (PSCF) to identify PCB sources at the southern end of Lake Michigan. Receptor models, such as the PSCF, focus on the behavior of the contaminant in the ambient environment at the point of impact or detection, as opposed to dispersion models that focus on transport, dilution and transformations between the source and receptor. The PSCF incorporates wind trajectories to attempt resolution of locations for unknown pollutant sources.

PSCF model statistics count each trajectory segment endpoint terminating within a particular grid cell to determine the probability that an event at the receptor site is related to that cell. The NOAA HYSPLIT-4 model was used to calculate backward trajectories. The PSCF value can be interpreted as the conditional probability that concentrations larger than a given criterion value are related to the passage of air parcels through that cell during transport to the receptor site, with high cell PSCF values indicating areas of high potential pollutant contributions. This is illustrated in Figure c ., where the source is located in cell C11 and the receptor is in C45. Solid curves represent high concentration wind trajectories and dotted lines represent low concentration wind trajectories.

Dr. Keri Hornbuckle

As part of the U.S. EPA's Lake Michigan Mass Balance Study, Dr. Keri Hornbuckle constructed a predictive model for gas-phase PCB congeners and trans-nonachlor over Lake Michigan using air temperature, wind direction and atmospheric PCB concentration data collected around and over the lake. A twofold approach is used to estimate the concentration of gas-phase chemicals over the lake, predicting the daily variation in gas-phase concentrations at each site, and then interpolating the discrete site predictions over the entire surface area of the lake.

As shown in Figure d ., a best fit equation of the temporal variation in gas-phase concentrations is derived for each site as a function of the number of hours that the land domain was sampled; the number of hours that the water domain was sampled; air temperature for the air flow over the land; air temperature for air flow over the water; and four fitting coefficients. After predicting the daily concentrations at each of the eight shoreline sites and 12 over-water sites, these concentrations were interpolated to predict concentrations at all of the approximately 2,300 cells over the lake, using an inverse distance weighting approach.

4.3.3 Estimated Extent of Deposition

of Selected Contaminants to Lake Michigan


Dr. Mark Cohen of the Air Resources Laboratory of the National Oceanic and Atmospheric Agency described his application of the NOAA HYSPLIT-4 (Hybrid Single Particle Lagrangian Integrated Trajectory) model to evaluate the atmospheric fate and transport of dioxin from U.S. and Canadian sources to the Great Lakes. Emission inventories used as input to the model are summarized in Figures e . and f . This source-receptor model estimated the total dioxin flux to Lake Michigan for the year 1996 to be on the order of 5 - 50 grams TEQ per year, with a central estimate of approximately 17 grams TEQ per year. Incineration, particularly of municipal, medical and hazardous waste, along with metallurgical processing, continue to be the dominant known source sectors contributing dioxin to the Great Lakes via the atmospheric pathway.

Sources from as far away as 2,500 km contributed dioxin to Lake Michigan; however, the model indicated that approximately 40 percent of all dioxin deposition to the lake originates from within 100 km. Figure g . illustrates the magnitude and geographic distribution of North American sources of dioxin deposition to Lake Michigan, based on the Cohen model.

Figure h . indicates the distribution of sources of total dioxin emissions in North America within various distances of each of the Great Lakes and the distribution of sources of dioxin deposition to each of the lakes with distance. The peak in total emission sources for Lake Michigan occurs at a distance between 700 - 1,000 km, while the peak in total deposition to Lake Michigan occurs from sources within 0 - 100 km. This suggests that the more proximate sources (those within 100 km of the basin), while their aggregate emissions are less, are linked to a very significant fraction, 40 percent, of the deposition in that lake.

Figure h . shows that this trend of greater deposition impact by proximate sources is not as significant for the other Great Lakes, as the majority of dioxin deposition arises from more distant sources.

In Figure i ., a summary of sensitivity analyses is presented for deposition to Lake Superior, in order to illustrate the relative magnitude and range of uncertainties in the overall simulation. The influence of six different aspects of the simulation on the model-predicted deposition is shown, including the number of standard source locations used to determine theoretical transfer (84 vs. 28), the interpolation methodology, the photolysis rate, the characterization of wet and dry deposition and the emissions themselves.

In some cases, the distribution was explicitly evaluated in the sensitivity analysis in only one direction or in a fairly limited manner. This was true for wet deposition, where the sensitivity analysis consisted of increasing the in-cloud particle washout ratio by a factor of four, photolysis , where the photolysis rate was decreased essentially to zero, and the number of standard source locations, where only two variations were evaluated (28 vs. 84 locations).

In these situations, where a more complete analysis of the influence of the methodological variation would have included variations in the other direction (e.g. a decrease in the in-cloud particle washout ratio) to determine the magnitude of the band of possible annual deposition, one of the endpoints of the uncertainty range was inferred by assuming that the influence of variations was approximately the same on either side of the central estimate. To estimate the impact of uncertainty in the emissions, the low end and high end of the estimated emissions range were used for each source.

It can be seen from Figure i . that uncertainties arising from the interpolation procedures appear relatively insignificant, those for the fate methodologies are moderate, and the uncertainty arising from the emission estimates appears relatively significant. While not shown here, results for the other Great Lakes are comparable to these for Lake Superior.

While uncertainties in the fate simulation methodology result in significant uncertainties in the magnitude of the model-predicted deposition to the Great Lakes, the estimates of the relative importance of different sources or source regions are not strongly affected. Examples of this are shown in Figure j . and Figure k . for Lake Michigan. In these figures, the effects of the most significant fate variations are examined, including the dry deposition algorithms used, wet deposition, and photolysis. Each set of estimates in these figures represents a complete model analysis with a given set of parameters and/or algorithms. Note that the default estimate is that for dry deposition algorithm "A."

Uncertainties in emissions have a pronounced affect on the overall predicted deposition and a direct effect on the estimates of the relative importance of different sources and source regions. For example, the relative significance of backyard burning and its contribution to emission estimate uncertainty remains a question. Cohen used a dioxin emission factor of approximately 250 grams TEQ per year for backyard burning, which would be approximately 10 percent of the total U.S. dioxin inventory. Burn barrel test data vary significantly, yielding orders of magnitude differences in estimated emission factors. Actual annual emissions may be as high as 1,000 or 2,000 grams or could also be significantly below the modeled estimate. Determining the significance of this source should be addressed promptly through appropriate research. Uncertainties in other source categories are also very significant. It was stressed that uncertainties in emissions are the controlling source of uncertainty in this type of analysis and further progress would require the directing of additional resources toward improving and updating emission inventories.

Polychlorinated Biphenyls (PCBs)

Deposition of gas-phase organic chemicals, including PCBs, is a major contributor to the loading of persistent toxic substances to Lake Michigan and, in the case of PCBs, up to 90 percent of the atmospheric concentration is associated with the free vapor phase. Not only is gas exchange the dominant process, it also occurs quite rapidly. Dr. Joel Baker emphasized the close linkage between the concentrations of PCBs and other volatile contaminants in the atmosphere and in the waters of the Great Lakes. Atmospheric gas exchange flux estimates indicate that somewhere near 20 or 40 nanograms of PCBs per square meter of lake surface per day (ng/m 2 /day) are being continually exchanged between the atmosphere and Lake Michigan. Up to 10 percent of the PCBs in the atmosphere exchange with the surface waters of Lake Michigan every day.

Baker estimates that 90 percent of the uncertainty in the gas exchange flux calculation is associated with the mass transfer coefficient and noted that focussed research to better understand air-water exchange physics would be critical to improvement of flux estimates. Gas exchange cannot be easily measured and must be estimated. As air and water PCB inventories approach steady state, it is increasingly difficult to calculate the air-water exchange rate.

Dr. Keri Hornbuckle discussed the application of a predictive model for persistent organic compounds, developed as part of the Lake Michigan Mass Balance Study, to the estimation of PCBs over Lake Michigan. Concentrations of 30 gas-phase PCB congeners and the sum of approximately 100 individually measured congeners or coeluting congener groups were estimated for every day of the 18-month field season and in the air over each of approximately 2,300 cells of lake surface area.

Measurements over such a fine scale grid allowed the estimation of variability of atmospheric deposition over space and time resulting in a model useful in the prediction of the spatial impact of the Chicago source area. A twofold approach is applied to the estimation of the concentration of gas-phase PCBs over the lake, first predicting the daily variation in gas-phase concentrations at each site based on wind direction and temperature, then interpolating the discrete site predictions over the entire surface area of Lake Michigan.

Hornbuckle determined net gas exchange of PCBs as a function of gas-phase concentrations, dissolved water concentrations, the physical and chemical properties of the compounds, wind speed and surface water temperatures. The resulting air/water exchange fluxes for PCBs exhibit intense variability over space and time, as well as in the direction of flux between net deposition and net volatilization.

Both air temperature and wind direction are strong factors in the variability in gas-phase PCB concentrations over Lake Michigan. Under low temperature and northern wind regimes, PCBs exhibit volatilization (transfer from the lake to the atmosphere) fluxes over most of the lake, primarily due to low air concentrations during cooler weather and negligible transport of air from the Chicago area over the lake. Figure l . illustrates four distinct daily scenarios.

For October 3, 1994, a day with primarily northeasterly winds and cool temperatures, volatilization is dominant. On the following day, the winds were primarily southeasterly and the Hornbuckle model predicted a small region of PCB deposition (transfer from atmosphere to lake) flux just north of the Chicago area. On October 5 and 6, the winds were predominantly southerly and the Chicago PCB plume causes a larger region of deposition to the lake. On October 6, the deposition zone covers almost the entire lake as a result of increased ambient temperatures and prevailing winds from the Chicago area. Although most of the southern basin of the lake experiences net gas deposition of PCBs, there are large regions along the southern coasts that still exhibit volatilization fluxes, primarily a result of gas-phase concentration variability across the lake. Water temperatures and the prevailing localized wind directions at each of the sampling sites also contribute to this complex pattern of air/water exchange. The model predicts large variations in gas-phase concentrations of PCBs at all of the Lake Michigan Mass Balance Study measurement sites, due principally to substantial changes in wind direction and temperature. Figure m . contrasts the extreme variability in predicted daily concentrations of gas-phase PCBs at Chicago with more stable values from open lake and the Sleeping Bear Dunes regional IADN monitoring site on the eastern shore of the lake. These significant changes in concentrations in the air mass coming from or through Chicago result in very large, often diurnal, variations in the area of Lake Michigan affected by the Chicago plume.

The Sleeping Bear Dunes site clearly registers lower ambient concentrations of PCBs than measurements taken elsewhere in the Lake Michigan basin, including measurements taken over water. The signal from this site is not directly correlated with temperature; suggesting other more complex factors must be considered. However, regardless of cause, this evident underestimation may preclude use of these data alone as representative of regional PCB concentrations without further corroboration with measurements from both other land based and over-water sites.

Also, due to the large daily variations in air concentrations and depositional fluxes, future studies should be integrated over larger time periods and larger areas, and should consider the impact of Chicago on deposition load, rather than attempting to define an area of impact, which can be highly variable.

Hornbuckle et al calculated the contribution of Chicago to the total gas-phase loading of PCBs to Lake Michigan as total gross gas deposition and net gas exchange. This approach removes the effect of water concentrations and associated transfers, which exhibit some seasonal and spatial variability, and allows for better integration of the contribution of each individual site.

For the whole lake, gross annual deposition of PCBs, the sum of the modelling results for all 98 congener groups, was approximately 3,200 kg. The results indicated the percent contribution of the Chicago site to the whole lake monthly gas-phase loads ranged from less than five percent to 20 percent, depending on the congener and the month. On an annual basis, Chicago is the largest single source of all 20 sites considered, contributing 10 percent of the total annual deposition load of gas-phase PCBs.

In his presentation, Dr. Steve Eisenreich showed that, as indicated for Chicago in the Lake Michigan research, over-water PCB concentrations are elevated downwind of the Baltimore and New York urban areas. On average, over-water concentrations were enhanced by a factor of four over regional background values. This pattern also was observed in the Chesapeake Bay near Baltimore and the New York harbor estuary.

Table 2 provides data from the Atmospheric Exchange over Lakes and Oceans Study (AEOLOS), Lake Michigan Mass Balance Study and New Jersey Atmospheric Deposition Network (NJADN) data sets, illustrating enhanced urban PCB concentrations for Chicago, Baltimore, New York - New Jersey and Camden - Philadelphia.

Table 2. Enhanced Urban Concentrations of Atmospheric PCBs

GPCBs (pg/m 3 )

Site Urban Over-Water or References

Range Background


Chicago- 270 - 14000 130 - 1200
AEOLOS 70 - 800 Simcik et al, 1997

1994 - 1995 Zhang et al , 1999

Chicago - LMMB 500 - 6800 100 - 500

1994 - 1995 (Sleeping Bear Dunes) Miller et al, 2001

Chicago - LMMB 100 - 16000 43 - 440 Green et al, 2000

1994 - 1995 (modeled - 24 hr. day) (modeled - 24 hr. day

Beaver Island)

Chicago - IADN Hites and Basu,

1998 460 - 6800 Unpubl. data

1999 335 - 7000

Baltimore - 380 - 3360 210 - 740 Offenberg and Baker, 1999

AEOLOS June 1996

Baltimore - 760 - 2280 290 - 990 Brunciak et al , 2000

AEOLOS June 1997

NY-NJ Area 100 - 3300 60 - 2340 Brunciak et al , 2000

1997 - 2000 (Jersey City) (Suburban)

(NJADN) 90 - 1600


Camden/Philadelphia 1020 - 16000 45 - 550 Eisenreich and

1999 - 2000 (Pinelands) Reinfelder, 2001

The Eisenreich presentation also stressed the dependence of atmospheric concentrations of PCBs, at any site, on temperature, which drives the magnitude of air-surface exchange. As illustrated in Figure n ., higher ambient temperatures generate higher atmospheric PCB concentrations particularly in areas where higher surface contamination would be anticipated, such as urban and industrial environments.

Concentrations of PCBs in the urban atmosphere range from 100-300 pg/m 3 in winter to 5,000 to 16,000 pg/m 3 on hot summer days, based on data from Chicago, Baltimore, Jersey City and Camden - Philadelphia. Eisenreich also concludes that high ambient temperatures in contaminated areas yield high atmospheric concentrations, whose inventory is then transported away from the urban area to nearby water bodies and terrestrial landscapes. As noted earlier during the discussion of Chicago PCB data, variability in concentration is linked to ambient temperature and the size of the contributing environmental reservoir of PCBs.

Agricultural Chemicals

For pesticides, delineation of emission sources and attribution of air concentrations to those sources are extremely complex. While the primary loading of pesticides to the atmosphere occurs during application to agricultural lands, emissions from residues in soil due to historical use can also be significant. The transport and deposition in North America of banned or restricted pesticides still used in other parts of the world are also of significance. The difficulty of estimation is further compounded by the confidential status of much of the current and historic pesticide sales data.

In his presentation, Dr. Trevor Scholtz described application of a Pesticide Emission Model (PEM) to development of estimates of pesticide emission inputs for chemical transport and deposition models. His methodology allowed an assessment of the effects of pesticide application methods and tilling practices on emission of pre-emergent pesticides to the atmosphere over a three-year period following application.

Three modes of pre-emergent application were examined: incorporation into the soil, spray application to the soil surface and in-furrow application during seed planting. While the in-furrow application limits the amount of pesticide initially available at the soil surface, evaporation of moisture and subsequent tilling of the soil may both expose the pesticide for subsequent volatilization. The percentages of applied pesticide theoretically lost from the soil to the atmosphere are compared for various combinations of application method, tilling of the soil in the fall or spring, or the effect of no tilling.

The 20 selected pesticides included in the theoretical model represent a wide range of physical and chemical properties and include some analogous to those currently used in North America, as well as some that have been banned or restricted on this continent, but may be in use elsewhere. Table 3 contains estimates of the extent of loss over three years in response to variation in tilling practices for the 20 pesticides studied.

Table 3 Physical-Chemical Properties of Pesticides

Source: Background paper from T. Scholtz, found on the CD-ROM Atmospheric Deposition: Science and Policy , available from the IJC.

# Pesticide Class* Diffusivity Diffusivity K oc - Soil K H - Henry's Soil

in Air in Water Sorption Law Constant Half Life

(m 2 /s) (m 2 /s) (m 3 /kg) (dim'less) (days)

1 2,4-DB H 4.97 x 10 -6 4.97 x 10 -10 0.5 2.36 x 10 -7 7

2 Aldrin I 4.97 x 10 -6 4.68 x 10 -10 5.01 3.74 x 10 -2 53

3 Atrazine H 4.97 x 10 -6 5.39 x 10 -10 0.1 1.19 x 10 -7 60

4 Chlordane I 4.97 x 10 -6 4.51 x 10 -10 20 3.70 x 10 -3 1205

5 DDT I 4.47 x 10 -6 4.54 x 10 -10 411 9.69 x 10 -4 1095

6 Dieldrin I 4.97 x 10 -6 4.67 x 10 -10 12 4.60 x 10 -4 2555

7 Endosulfan I 4.57 x 10 -6 4.72 x 10 -10 12.4 1.22 x 10 -3 50

8 Endrin I 4.97 x 10 -6 4.67 x 10 -10 10 1.35 x 10 -5 1825

9 Fenthion I 4.97 x 10 -6 5.21 x 10 -10 1.5 9.03 x 10 -6 34

10 -HCH I 5.41 x 10 -6 5.48 x 10 -10 2.59 3.57 x 10 -4 400

11 Heptachlor I 4.97 x 10 -6 4.76 x 10 -10 24 4.60 x 10 -2 219

12 Hexachlorobenzene F 5.56 x 10 -6 5.81 x 10 -10 411 2.92 x 10 -3 365

13 Lindane I 5.18 x 10 -6 5.48 x 10 -10 1.1 5.30 x 10 -5 400

14 Methoxychlor I 4.97 x 10 -6 4.97 x 10 -10 79.4 4.10 x 10 -4 120

15 Metolachlor H 4.51 x 10 -6 4.97 x 10 -10 0.2 9.10 x 10 -7 90

16 Metribuzin H 5.79 x 10 -6 4.97 x 10 -10 0.06 9.30 x 10 -8 40

17 Mirex I 4.97 x 10 -6 4.05 x 10 -10 3260 3.45 x 10 -1 365

18 Quintozene (PCNB) F 5.59 x 10 -6 4.97 x 10 -10 5 4.1 x 10 -5 250

19 Toxaphene I 4.97 x 10 -6 4.35 x 10 -10 100 1.70 x 10 -4 365

20 Triallate H 4.67 x 10 -6 4.71 x 10 -10 2.4 4.19 x 10 -4 82

* H - herbicide; I - insecticide; F - fumigant

Scholtz concluded that applying a pesticide in a furrow that is then covered results in the least loss (less than 26 percent) of applied pesticide to the atmosphere. The second best method is soil incorporation (less than 44 percent), while the highest losses of applied pesticide to the atmosphere (up to 92 percent) result from the use of a pesticide as a pre-emergent spray. Model results also suggest that spring tilling can cause releases of chemicals associated with previous applications, with subsequent transport and deposition

Atmospheric data for trans-nonachlor collected during the Lake Michigan Mass Balance Study support predictions from the PEM of Scholtz et al. Trans-nonachlor is a component of chlordane, a banned pesticide formerly in wide use in the U.S. Midwest. The Lake Michigan Mass Balance Study ambient monitoring data indicated a significant source of atmospheric trans-nonachlor occurring in May (Miller et al , 2001). Other ambient results correlated well with the modeled trans-nonachlor in the vapor phase, with the exception of this May signal. While there is no complete explanation for this finding, the PEM suggests this could be a spring till release of this banned chemical, which is residual in the tilled soil a release not directly correlated to temperature, but rather to activity.

4.3.4 Source Identification — Need for Enhanced Emission Inventories

Most workshop presenters stressed the need to improve and enhance current emission inventories for pollutants of concern. In his presentation, Dr. Cohen identified the major impediment to further upgrading of his model as lack of current, accurate, geographically-resolved emission inventories. Dr. Wania noted that the most likely explanation for model under and over predictions is linked to variability in the emission estimates. Dr. Eisenreich indicated that, although an appreciable amount of effort went into creating the emission inventory for the Great Lakes region for PCBs, it is not adequate to account for the concentrations, fluxes, deposition and accumulation observed around Lake Michigan.

There is a major inconsistency between the PCB source inventory and the extent of deposition in Lake Michigan. Total PCB emissions in the Great Lakes Air Toxics Emissions Inventory are estimated to be approximately 3.2 kg per year, while annual gross deposition to Lake Michigan is estimated at 3,200 kg per year. Similar concerns were raised regarding the dioxin inventory, especially related to open burning emissions that are very much first estimates of unknown quality.

Comparable questions could be raised about the inventories of other chemicals subject to long-range transport, particularly banned substances, such as chlordane, hexachlorocyclohexane (HCH) and toxaphene. Atmospheric deposition appears to be substantially higher than would be consistent with current total emissions in available inventories, and inventories are largely not accounting for areal or fugitive emissions. Internal inventory inconsistencies also were apparent from state to state among the Great Lakes Air Toxics inventory data, especially for areal sources of pollutants such as PCBs, dioxins and furans.

The Cohen presentation emphasized that improvement of the dioxin inventory is clearly necessary, both to ensure that all sources are accounted for and to assess their impact on the Great Lakes and other water bodies more accurately. In the case of Lake Michigan, he recommended that a thorough review of established and potential sources — point and areal — within 100 km of the lake should be given priority. As noted above, his results indicated that approximately 40 percent of the dioxin deposition arose from sources within 100 km of the lake. For Great Lakes other than Lake Michigan, the proportion of the loadings arising from within 100 km of the lake is somewhat less (see Figure h). Therefore, an emphasis on nearby sources is proportionately less important for these other lakes.

The available source testing data bank and the accuracy of the emission factor approach can be increased by:

• conducting source testing at point source facilities known or suspected to be major contributors, particularly those within 100 km of the shoreline;

• testing at particular facilities that also could be representative of source sectors for which few or no tests have ever been conducted;

• performing additional stack tests at facilities with an established source testing record to provide a more robust database for developing emission factors and to better understand the variability in emissions from individual facilities; and

• ensuring regular, accurate updates of basic information from significant sources on processes, including alterations in air pollution control equipment, activity factors and other parameters affecting emissions.

Although his model included over 5,000 North American point sources of dioxin, Cohen indicated that timely updates of the emission inventory for the 100 most significant dioxin sources in the U.S. and Canada would substantially improve model accuracy while providing some guidance and assurance to control program outcomes. In addition, as noted above, further information on significant areal sources, such as backyard burning, would also be very important.

In addition, the current U.S. dioxin inventory does not contain estimated emissions from residential or commercial coal combustion, magnesium manufacturing or small commercial incinerators. Neither the U.S. nor Canadian inventory includes emissions from open-burning of PVC-coated wires (e.g. structure and vehicle fires), asphalt production, landfill fires and landfill gas combustion, coke production, leaded gasoline combustion or petroleum refining.

4.3.5 Extent of Urban Areal Emissions

for PCBs Chicago, Baltimore

and the Chesapeake Bay

The hypothesis of the Atmospheric Exchange over Lakes and Oceans Study (AEOLOS) program is that elevated concentrations of air pollutants, such as PCBs in urban areas, result in enhanced deposition to nearshore areas. In 1994 and 1995, the study established a series of sampling locations in Southern Lake Michigan and Chicago, as well as Chesapeake Bay and Baltimore; sampling took place in the cities, along the shorelines, on the waters of the lake or bay and on the opposite shorelines. The study attempted to determine the extent to which pollutant plumes from these cities enhance the deposition and loadings of chemicals to nearby water bodies.

In both the Chicago and Baltimore areas, the concentrations in the urban atmosphere greatly exceed those measured downwind, over water or over land, often by a factor of five and sometimes by factors of 10 or even 100. For Lake Michigan, the high concentrations occurring over the lake were associated with winds from the direction of the shoreline between Evanston, Illinois and Gary, Indiana. A substantial amount of the decrease in concentration between the Chicago area and the eastern lake shoreline in Michigan has been linked to enhanced deposition to Lake Michigan.

It appears that the source of the urban PCB plume exhibited at Chicago and elsewhere is volatilization from open sites that can be affected by variations in ambient temperature, such as landfills, open spill sites and abandoned and unremediated industrial process sites. The extent of the transport of PCBs from a city such as Chicago indicates the presence of some hotspots, such as transformer yards, abandoned industrial sites and land fills, as well as a large general background signal coming from PCBs that have been sorbed onto various surfaces through volatilization and deposition cycles occurring with variations in ambient temperature. Identifying and quantifying these sources would be crucial to any effective deposition reduction strategy.

To gain a sense of the possible extent of the `footprint' of atmospheric emissions from Chicago and other urban areas in the transport and deposition downstream, Steve Eisenreich evaluated AEOLOS and NJADN (New Jersey Atmospheric Deposition Network) data and determined a downwind decay rate of approximately six percent per kilometer, resulting in a reversion to background levels at approximately 50 km downwind.

The Eisenreich scenario assumes that the decrease in atmospheric PCB concentrations away from Chicago and other source areas is due to dispersion and dilution under transport, and removal by deposition and atmospheric loss processes. Figure o . demonstrates this phenomenon for PCB concentrations from Chicago to an over-water sampling site 20 km away, and from Camden - Philadelphia to the Pinelands, New Jersey site 50 km away. The downwind zone of influence is on the order of 50 - 60 km in both cases although the diurnal variation in the magnitude of this zone can be substantial (See Figure l ).

Hornbuckle estimated that 2 - 20 percent of the gas-phase PCB atmospheric loading to Lake Michigan originates from Chicago, depending on the congener considered and the time of year. Overall, Chicago accounts for approximately 10 percent (about 300 kg per year) of the total annual PCB input to the lake. Figure p . illustrates the percentage contribution to the whole-lake gross deposition from the Chicago site for one PCB congener group (PCB #31+28), showing a highly seasonal variability, with the greatest deposition occurring during the summertime, when air concentrations are high. (Note that gross depositional loads are assigned a negative value, consistent with the equation used to calculate them.)

4.3.6 Actions Supported by the Lake Michigan Models

While the Great Lakes Air Toxics Inventory is one of the more comprehensive regional efforts, there is need for improvement. For example, as noted above, modelling and ambient and deposition measurements for PCBs and dioxin indicate that source data are not adequate to account for the loading to the lakes. The inventory for specific pollutants should be improved, as noted above, by focusing on a specific geographic area, such as a 100 km radius from Lake Michigan for dioxin, and on specific point source sectors and areal emissions, such as burn barrels or landfills. Various workshop presenters noted that, for banned contaminants, the reservoirs in the environment, such as sediment, tillage residue and landfill contents, are of much greater significance than the point sources currently included in the inventories. Emissions associated with such reservoirs must be better defined.

More effort is needed to refine areal emission factors generally in the U.S. and Canadian inventories. Several presentations highlighted the need to include PCB and emissions of other persistent toxic substances from landfills, as estimates of PCB and emissions of other persistent toxic substances from individual landfills in Chicago (Hsu) and New York (Eisenreich) tend to dwarf other sources currently included in emission inventories. In addition, wastewater treatment plants, sludge drying beds and sludge used as landfill cover may all be significant sources of atmospheric PCBs and emissions associated with them should be measured. Other source sectors identified for improvement or establishment of emission estimates include:

• off road and heavy duty vehicles;

• confined disposal facilities for contaminated sediments;

• transformer storage yards; and

• highly contaminated brownfield or former industrial sites.

Given that urban areas are significant sources of PCBs, as well as other persistent toxic substances, to the atmosphere and nearby water bodies, and that each urban area may contain many of the above sources, it is imperative to better quantify emission estimates as a basis for reduction programs.

4.3.7 Options for Estimating Emissions

Alternative emission estimation methods are needed, particularly in the urban setting, to assist in the identification of sources and the apportionment of the observed pollutant loadings to those sources. Source apportionment modelling using ambient monitoring data can be used to back-calculate emissions and thus identify or verify sources.

Several promising techniques were presented at the workshop. Dr. Tom Holsen and Ying-Kuang Hsu demonstrated that receptor models and factor analysis can be applied to ambient concentration and related meteorological data to zero in on localized, high emission factor sources of PCBs. Following such identification, sites can be further isolated and downwind measurements taken to confirm the source and estimate the quantity of the PCBs emitted.

In their work, the Potential Source Contribution Function (PSCF) was used to identify PCB sources to southern Lake Michigan. Major differences of this study from prior PSCF applications were the small local area considered and the modelling of semi-volatile organic compounds mostly in the vapor phase. Prior applications of PSCF were regional-scale and sources were resolved to within a hundred miles.

This PSCF modelling in the vicinity of Chicago resolved three PCB source directions: 1. the northwest direction pointing to Madison, Wisconsin; 2. the southwest direction between Joliet and Chicago; and 3. the neighborhood of Lake Calumet. Figure q . illustrates several PSCF model plots in which the alignment of the squares indicates the general directions of potential significant sources of PCB emissions.

To support and build upon the PSCF results, several five-hour, upwind and downwind air samples were taken near potential sources in areas identified as having high PSCF values, as well as other sites that were suspected of being sources of PCBs. Municipal sludge drying beds, a transformer storage yard and a landfill were verified as significant PCB sources based on downwind concentrations 1.5 to 5.3 times the upwind concentrations.

The transformer storage yard exhibited some of the highest downwind PCB concentrations measured during this study, yet it is not currently listed in U.S. EPA's PCB Transformer Registration Database. A closer investigation of the sludge drying operation suggested that its emissions could be three times greater than the 1996 estimated U.S. National Toxics Air Emission Inventory for PCBs in EPA Region 5. Results of upwind and downwind sampling are provided in Table 4.

Table 4 Potential Sources: Upwind and Downwind PCB Concentrations

Date Site PCB Concentration ng/m 3

Upwind Downwind

07/06/99 Calumet East Drying Beds 2.87 5.47

07/02/99 Stickney Drying Beds NA 2.17

08/16/98 NA 1.92

08/13/98 CID Landfill NA 5.13

07/04/99 1.93 3.99

08/16/99 1.23 2.47

08/14/98 ComEd Transformer Storage Yard NA 11.89

08/15/99, AM 1.41 2.11

08/15/99, PM 1.33 2.73

08/17/99, AM NA 3.29

07/20/00, AM 1.21 6.49

07/20/00, PM 1.53 8.07

Given that it would not be practical to use upwind and downwind measurement techniques to identify all the contributions of volatilizing PCBs, Eisenreich and Hornbuckle have proposed a method for predicting emissions from major source regions, such as urban areas, using historical infrastructure and census information, and data on local climate regimes. Using this broadly available information as a screening tool and direct ambient measurement for validation, this method or model could be used to calculate a first estimate of the extent of probable urban volatilization of contaminants such as PCBs, dichlorodiphenyltrichloroethane (DDT) and others.

Results to date indicate that use of the 1970 Cook County (Chicago) census data explained 60 percent of the variability in PCB concentration in the Lake Michigan Mass Balance Study data set and ambient temperature explained about 30 percent. Thus, in the case of PCBs, ambient temperature and population apparently can be used as inputs to a model that could provide a first estimate of the magnitude of the areal environmental reservoir.

Further development of this model requires more rigorous application in other urban and industrial centers for which there are good PCB concentration data and other locales for which there are very few or no measured values, particularly newer urban areas not strongly associated with PCB use. This model may offer a means to develop regional and continental emission inventories for PCBs and other persistent organic pollutants. When coupled with source-receptor models, emission inventories may be developed for compounds for which the usual inventory methodologies are inapplicable.

Another approach to supplementing emission inventories, particularly for banned and current use pesticides, is the application of the Scholtz Pesticide Emissions Model (PEM) to determine releases to the atmosphere. This modelling technique, in the absence of other data, can use recommended application rates or can be improved by the establishment of pesticide-use databases. It also can be applied to estimate emissions of residual banned pesticides if historical usage data of reasonable quality are available or can be derived.


The Great Lakes data set is among the best in the world in terms of temporal span and quantity and quality of measurements. The 20 years of PCB data in the Great Lakes are not available for any other ecosystem. In addition, the Lake Michigan Mass Balance Study has a very high quality database. The entire Great Lakes database provides a unique opportunity to further understanding of pollutant cycling and evaluate future pollutant levels. However, as these data are used in various models, and the complexity of the system is unveiled, some enhancements would be necessary to further understand pollutant dynamics. A key question is how to strategically obtain additional, more refined ambient data within existing program limitations.

Cohen indicated that, at the moment, there are only five ambient dioxin measurements from 1996 suitable for use in model calibration, Figure r . , and none of these is located in the Great Lakes basin, as dioxin is not included in the Integrated Atmospheric Deposition Network (IADN).

However, a new dioxin monitoring program, the National Dioxin Ambient Monitoring Network, composed of 29 sites nationwide, including some in the Great Lakes region, is under development in the United States. New dioxin monitoring sites should be located in rural areas, as measurements made in urban areas can't clearly determine the origin of the dioxin due to the pronounced complexity of the spatial variability of emissions and meteorological conditions within such areas. Also, much of human dioxin exposure appears to come from the food chain via agricultural products rather than inhalation, suggesting that ambient air concentrations of dioxin in urban areas are not of the same significance from a public health standpoint.

There is very significant variability in space and time in the concentrations of many of the pollutants of concern in Lake Michigan. As a result, at Chicago, a single measurement reflects only that sampling period and cannot be taken as representative of any longer time span. Rather than depending on a limited number of monitoring sites to determine the temporal and spatial variation in these over-lake concentrations efficient, appropriate ways to monitor, interpolate and interpret pollutant levels are needed in order to estimate concentrations over the entire lake with more accuracy. Long-term averages will not reflect the influence of meteorology or other factors driving changes in concentration. As illustrated by Dr. Hornbuckle's work, while averages may be adequate for some applications, modelling processes and determining the significance of sources requires data that are discrete over the shortest possible time frame.

The Sleeping Bear Dunes IADN site clearly registers lower PCB concentrations than every other site in the Lake Michigan basin, including over-water sampling locations. Hornbuckle suggested that this underestimation indicates that long-term studies using these data do not appropriately describe regional PCB concentrations and associated loadings. Efforts should be made to resolve this issue.

Hsu also indicated a need for ambient concentration data obtained over relatively short sampling periods; otherwise the sample may be influenced by numerous sources and therefore less useful for receptor modelling. In some cases, finer spatial-scale resolution in meteorological databases also is needed, particularly to account for the behavior of urban or agricultural plumes and to allow incorporation of the grasshopper or revolatilization effect into models.


Several workshop presenters suggested the need for more information on the non-atmospheric pollutant loading pathways to Lake Michigan in order to better know the relative contribution of pollutants of concern from atmospheric deposition. For example, Cohen summarized two crude estimates of the relative contribution of atmospheric deposition to the total loading of dioxin to Lake Michigan — both ranging between 50 and 100 percent (see Table 5). The overall contribution of the atmospheric pathway of PTSs relative to other loading pathways remains uncertain and needs to be better quantified.

The Lake Michigan Mass Balance Study should be completed to provide further information on non-atmospheric pathways. Using the wealth of data generated in 1994-1995 for PCBs, mercury, atrazine and trans-nonachlor, indicative estimates of loadings from tributaries, direct discharges, sediment, groundwater and other sources, should be possible. U.S. EPA is encouraged to make a firm commitment to the completion of the Mass Balance Study.

A Multimedia Fate and Transfer Model (MFTM), as described by Dr. F. Wania, may be used to understand the behavior of semivolatile chemicals that move among more than one environmental compartment (e.g. atmospheric reservoir, soils, vegetation, water, and sediments) over periods of a decade or more. While this type of model has relatively low spatial resolution compared to others, it allows the description of conditions in various compartments of the environment over long time periods, making it useful in working with persistent chemicals.

Table 5 Estimates of the Percent of Lake Michigan Dioxin Loadings Attributable

to the Atmospheric Deposition Pathway

Study Fraction of Current Loadings Contributed Through

Atmospheric Pathway

Cohen et al 3 PCDD/TEQ: 50-100
(central estimate: 88)

Pearson et al 2 PCDD: 50-100

PCDF: 5-35

2 Pearson, R.F., D.L. Swackhamer, S.J. Eisenreich, and D.T. Long (1998).

"Atmospheric Inputs of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans to the Great Lakes" Compositional Comparison of PCDD and PCDF in Sediments.

" J. Great Lakes Research 24(1):65-82.

3 Cohen, M., et al , 1995. Quantitative Estimation of the Entry of Dioxins, Furans, and Hexachlorobenzene into the Great Lakes from Airborne and Waterborne Sources.

Flushing, NY: CBNS. Final Report to the Joyce Foundation.


4.6.1 Core Findings

• Given the availability of appropriate data, the fate of several contaminants prominent in the Great Lakes, including dioxin, PCBs and pesticides, such as lindane, chlordane and others, can be effectively examined through the application of models.

• Similar to their application for the development of management strategies for ozone, models can now be effectively applied as necessary tools for identifying and broadly ranking sources of selected persistent toxic substances entering the Great Lakes by atmospheric transport and deposition.

• Source-receptor models can be used to quantify the contributions of specific point and areal sources of deposition to the Great Lakes within established bounds, given adequate information on 1. the physical and chemical properties of the contaminant; 2. emission inventories; and 3. ambient measurements for verification.

• Of these factors, an examination of several source-receptor models has established that the quality of their predictions is chiefly dependent on the accuracy of associated emission inventories.

• There is significant spatial and temporal variability in the concentrations of several pollutants of concern over Lake Michigan. Specific to the Sleeping Bear Dunes Regional IADN site, measurements taken there appear to underestimate concentrations of air toxics, particularly PCBs, likely due to site characteristics and local meteorology.

• The Lake Michigan Mass Balance (LMMB) and model applications in other urban locales indicate the dominance of PCB volatilization from uncharacterized areal sources rather than permitted point sources or established areal sources. It is likely that such sources also will prove significant for other banned or byproduct persistent toxic substances.

• Further application of source-receptor models to PCB transport and deposition must await the development of more comprehensive inventories, as existing inventories seriously underestimate PCB emissions. However, other modelling to date has identified several likely dominant areal sources.

4.6.2 Recommendations

The IAQAB makes the following recommendations to the IJC.

Sources and Loadings

There is a clear need to place the relative loading of persistent toxic substances from atmospheric deposition in context with the loading of these substances from all pathways to Lake Michigan and the other Great Lakes.

• Recommend to the Parties that the completion of the Lake Michigan Mass Balance Study for all pathways is crucial to development of an effective control strategy and that U.S. EPA should expedite its prompt conclusion.

The Lake Michigan Mass Balance Study targets only four pollutants. At the present time, little information is available regarding the relative importance of atmospheric deposition to the loadings of other pollutants to Lake Michigan, or for essentially all pollutants of concern to the other Great Lakes.

• Recommend to the Parties that basic mass balance information should be developed for other pollutants of concern in Lake Michigan, as well as pollutants of concern in other lakes.

Current emission inventories must be improved and extended to areal sources if more precise model outputs and an effective control strategy are to be developed.

Recommend to the Parties that the following immediate actions be taken.

a. For dioxin, review and compare the Great Lakes Air Toxics Emissions Inventory and the inventory developed by Dr. Cohen, with the goal of improving the dioxin inventory for the region, particularly for major point sources. The enhancement of emission factors, other parameters necessary to modelling, and production or process data should be significant elements of this effort. Specific to Lake Michigan, emissions from major sources within 100 km of the basin should be confirmed, preferably by a combination of source testing and data quality review.

b. Support further quantification of dioxin emissions associated with backyard residential waste burning, including refinement of areal emission factors and determination of the extent of this practice on a regional basis.

c. For a number of persistent toxics, including PCBs, chlordane, mercury and critical banned pesticides, perform a review of current and historical land-use records, along with targeted modelling and monitoring at urban centres using one or more of the techniques presented at the November 2000, IAQAB workshop in Milwaukee, Wisconsin to estimate potential areal loadings.

d. Undertake an air toxics monitoring and measurement program designed to identify open sources of PCBs, such as contaminated brownfield and storage and waste management sites. This monitoring program should have a mobile capability with simplified procedures for deployment and relocation, as well as for upwind and downwind studies or measurements. All measurements should be coordinated with modelling predictions. Immediate priority should be given to estimating emissions from individual landfills, wastewater sludge drying operations and open transformer storage facilities for inclusion in the inventory. Measurement of other banned contaminants should also accompany such programs, as feasible.


Recommend to the Parties that the models and strategies for Lake Michigan and its related urban area of Chicago, reviewed at the workshops, should be developed further and their application extended to other urban areas and other lakes within and outside the Great Lakes basin.

• Recommend to the Parties, that as a first step, the adequacy of information on contaminant physical and chemical properties, as well as available emissions and ambient concentration data, should be determined prior to any modelling application.

• Recommend to the Parties that the Lake Michigan Mass Balance dataset, including available sample extracts and related measurements, and appropriate model(s), be used for the prediction of the sources and transport of other air toxics, such as polycyclic aromatic hydrocarbons (PAHs), beyond the original LMMB target compounds of mercury, PCBs, trans-nonachlor and atrazine.

Recommend that the Parties explore the application of other multimedia non-steady state models as an effective method of determining the longer term trends in the deposition of persistent toxic substances to the Great Lakes.

• Recommend that the Parties apply models for pesticide volatilization from soils to fields within and outside the Great Lakes basin where significant concentrations of banned pesticide residuals are detected, both to estimate the possible contribution of continued cultivation and to develop a code of best practice for such areas.

• Recommend that the parties begin a predictive modelling effort to identify regions around the Great Lakes for which there is a high probability of substantial emissions of persistent toxics.

Derivation and verification of any such modelling technique should be focussed initially in major urban areas.

Ambient Sampling

Recommend that the Parties continue ambient air sampling over the surface of the lake to provide better estimates of representative regional concentrations of these pollutants and improve the characterization of their air/water exchange.

• Given the regional designation of the Sleeping Bear Dunes IADN ambient monitoring site, recommend that the Parties interpret data collected at this site with assistance from atmospheric models that address air/surface dynamics and include meteorological models.

• Ensure that any modelling effort be supported by adequate ambient measurements to provide verification for any model output.

Source Control Initiatives

• The recently completed Delta Institute Lake Michigan Regional Air Toxics Strategy identifies linkages and opportunities for further air toxics reductions via various, ongoing, specific programs and initiatives under state and U.S. federal legislation. The IJC, through their relevant advisory boards, should review this proposed strategy and comment on its applicability to deposition reductions in the Lake Michigan and other Great Lakes basins from a binational perspective.


FOR 1999-2001

Canadian Members

Dr. Don McKay, Co-Chair
Director, Air Quality Research Branch

Meteorological Service of Canada

Environment Canada

4905 Dufferin

Downsview, ON M3H 5T4

Dr. David I. Besner , P. Eng.

Assistant Deputy Minister

InterGovernmental and External Relations

New Brunswick Dept. of the Environment

and Local Government

364 Argyle Street, 2nd Floor

Fredericton, New Brunswick E3B 5H1

Dr. Michael Brauer

Associate Professor

School of Occupational and Environmental Hygiene, and

Department of Medicine — Respiratory Division

The University of British Columbia

2206 East Mall, Room 366A

Vancouver, B.C. V6T 1Z3

Mr. Edward W. Pichéé

Director, Environmental Monitoring and Reporting

Ontario Ministry of Environment

125 Resources Road, West Wing

Etobicoke, ON M9P 3V6

United States Members

Dr. Gary J. Foley, Co-Chair

Director, National Exposure Research Lab



Catawba Bldg., Progress Center

3210 Hwy. 54

Research Triangle Park, NC 27709

Mr. Richard S. Artz

NOAA Air Resources Laboratory

Room 3151, SSMC3, R/E/AR

1315 East West Highway

Silver Spring, MD 20910

Mr. Harold T. Garabedian

Deputy Director

Vermont Agency of Natural Resources103 South Main Street

Waterbury, VT 05671-0402

Dr. Paul J. Lioy

Environmental and

Occupation Health Sciences Institute

681 Frelinghuysen Rd., 3rd Floor

Piscataway, NJ 08855-1179

Dr. Kathy Ann Tonnessen

Research Coordinator

Rocky Mountains Cooperative Ecosystem Studies Unit

University of Montana

Missoula, MT 59812

Commission Liaisons

Edward A. Bailey

Engineering Adviser

International Joint Commission

Canadian Section Office

234 Laurier Ave. West, 22nd Floor

Ottawa, Ontario K1P 6K6

Joel L Fisher

Environmental Advisor

International Joint Commission

U.S. Section Office

1250 23rd Street N.W., Suite 100

Washington, D.C. 20440


John F. McDonald

Senior Engineer/Secretary, IAQAB

International Joint Commission

Great Lakes Regional Office

100 Ouellette Ave., 8th Floor

Windsor, ON N9A 6T3


Brunciak, P.A., J. Dachs, T.P. Franz, C.L. Gigliotti, E.D. Nelson, B.J. Turpin, S.J. Eisenreich. PCBs and particulate organic and elemental carbon in the Chesapeake Bay atmosphere. Atmos. Environ. 2000, In review.

Brunciak, P.A., J. Dachs, T.P. Franz, C.L. Gigliotti, E.D. Nelson, S.J. Eisenreich Air-water exchange of PCBs and PAHs in the BY-NJ Harbor Estuary, Environ. Sci. Tech., 2000, In review.

Cohen, M., et al , 1995. Quantitative Estimation of the Entry of Dioxins, Furans, and Hexachlorobenzene into the Great Lakes from Airborne and Waterborne Sources . Flushing, NY: CBNS. Final Report to the Joyce Foundation.

Eisenreich, S.J., J. Reinfelder. The New Jersey Atmospheric Deposition Network (NJADN). An Interim report to the New Jersey Department of Environment Protection. March 2001, 84 p.

Green, M.L., J.V. DePinto, C. Sweet, K.C. Hornbuckle. Regional spatial and temporal interpolation of atmospheric PCBs: Interpretation of lake Michigan mass balance data. Environ. Sci. Technol. 2000, 34 , 1833-1841.

Hites, R. and I. Basu, Indiana University, Bloomington, IN. Data forwarded to USEPA Great Lakes National Programs Office Unpubl

Miller, S.M., M.L. Green, J.V. DePinto, K.C. Hornbuckle. Results from the Lake Michigan Mass Balance Study: Concentrations and Fluxes of Atmospheric Polychlorinated Biphenyls and trans-Nonachlor. Environ. Sci. Technol. 2001, 35 , 278-285.

Offenberg, J.H. and J.E. Baker (1999) Influence of Baltimore's urban atmosphere on organic contaminants over the northern Chesapeake Bay. J. Air Waste Manag. Assoc ., 49 , 959-965.

Pearson, R.F., D.L. Swackhamer, S.J. Eisenreich, and D.T. Long (1998). Atmospheric Inputs of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans to the Great Lakes . Compositional Comparison of PCDD and PCDF in Sediments." J. Great Lakes Research 24(1):65-82.

Simcik, M., T. Franz, H. Zhang, S.J. Eisenreich (1998) Gas-particle partitioning of PCBs and PAHs in the Chicago urban and adjacent coastal atmosphere: States of equilibrium. Environ. Sci. Tech. , 32 , 251-257.

Simcik, M.F., H. Zhang, S.J. Eisenreich, and T.P. Franz (1997) Urban contamination of the Chicago/coastal Lake Michigan atmosphere by PCBs and PAHs during AEOLOS. Environ. Sci. Techno., 31 , 2141-2147.

Voldner, E. C. and L. Smith. Invited Paper Production, Usage and Atmospheric Emissions of 14 Priority Toxic Chemicals. Appendix 2 to the Proceedings of the Workshop on Great Lakes Atmospheric Deposition. Joint Report of the Water Quality Board, Science Advisory Board, and International Air Quality Advisory Board. International Joint Commission, Windsor, Ontario. 1991. 94 pp.

Zhang, H., S.J. Eisenreich, T.R. Franz, J.E. Baker, and J.H. Offenberg (1999) Evidence for increased gaseous PCB fluxes to Lake Michigan from Chicago. Environ. Sci. Technol. , 33 , 2129-2137.

Appendix I Significant Activities Under Annex 15 of the Great Lakes Water Quality Agreement


In the early 1980s, Dr. Steve Eisenreich and Dr. William Strachan, in one of the first estimates of the contribution of atmospheric deposition to the contamination of the Great Lakes, determined that approximately 90 percent of polychlorinatd biphenyls (PCBs) loading to Lake Superior could be attributed to deposition from the atmosphere.

In its 1985 report to the IJC, the Great Lakes Water Quality Board (WQB) presented a list of 11 (12) Critical Pollutants (Table 6) . For every listed pollutant, there was reason to believe or evidence to support the supposition of the atmosphere as a significant pathway. In response, the IJC's WQB, Science Advisory Board (SAB) and International Air Quality Advisory Board (IAQAB) began assembling an international emission inventory of persistent toxic substances. These and other developments contributed to the inclusion of Annex 15, Airborne Toxic Substances, in the 1987 Protocol to the Great Lakes Water Quality Agreement.

Table 6 Critical Pollutants

Total polychlorinated biphenyl (PCB)




DDT and metabolites

2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD)



Alkylated lead



Activities Since 1987

Annex 15 committed the governments of the United States and Canada, the parties to the Agreement, to the reduction of atmospheric deposition of toxic substances, particularly persistent toxic substances, to the Great Lakes basin ecosystem through research, surveillance and monitoring, and ultimately, the implementation of additional pollution control measures. Models to determine the significance of atmospheric loadings to the Great Lakes system, relative to other pathways and the sources of such substances from outside the Great Lakes system, also were to be developed.

The Integrated Atmospheric Deposition Network (IADN) was established in the Great Lakes basin as a direct response to Annex 15 and built upon the expertise available under the IJC for its design. The network was to: 1. determine atmospheric loadings of toxic substances by quantifying their total and net atmospheric input to the Great Lakes system; 2. define the temporal and spatial trends in the atmospheric deposition in the basin; and 3. support development of Remedial Action Plans and Lakewide Management Plans pursuant to Annex 2 of the Agreement.

Additionally, the Parties, in cooperation with state and provincial governments, committed to develop and implement measures to control emission sources of toxic substances and eliminate the emission sources persistent toxic substances in cases where the atmosphere is a significant contributor to the Great Lakes system. The governments were to review their progress in implementing this annex and report to the IJC biennially, commencing with a report no later than December 31, 1988.

Addressing the monitoring element of the annex, a 1989 report from the IJC's Integrated Atmospheric Monitoring Task Force established the basis for the international Integrated Atmospheric Deposition Network (IADN). The network continues to supply valuable data on concentrations of many, but not all, targeted persistent toxic substances.

Also in 1989, in response to the research element of the annex, the IAQAB compiled a first inventory of atmospheric sources of the Critical Pollutants (Voldner and Smith, 1991). The report concluded that "a larger undertaking ... on toxic chemical emissions will be required to provide the necessary information on atmospheric emissions and their subsequent deposition in the Great Lakes region." The Parties did not carry on this specific task, choosing rather to integrate some of the critical pollutants into their national and regional emission inventory efforts. Notwithstanding some observed progress in emission inventory activities, an identical statement could be made today with regard to the Level I and Level II contaminants from the 1997 Binational Toxics Strategy (BNS) list.

Heavy metals, organochlorines and pesticides currently used in North America were the focus of subsequent modelling efforts. Annual estimates of deposition to individual lakes and basins have been computed for sulphur and nitrogen during the period from 1980 to 1988; for toxaphene during approximately 1980; and for mercury during a thirty-day period in late 1980.

The IAQAB noted that the quality of these model-estimated depositions would be very dependent upon the quality of emission inventories, knowledge of the chemical and physical processes affecting their lifetimes in the atmosphere, and support for further model development. This statement was explicitly reinforced in the IJC's Seventh Biennial Report on Great Lakes Water Quality (December 1993).

This same report also noted the comments of the IAQAB on the inadequacy of emission inventories with regard to development of a Lake Superior Zero Discharge Demonstration Program and recommended that "federal governments provide coordinated national inventories of toxic air emissions ... " The Commission also noted, "a focus on research to improve understanding of the pathways, fate and effects of airborne toxic substances, required by Annex 15, has not occurred. Specifically a research program emphasizing atmospheric processes, transfer coefficients, and gas exchange processes is needed."

HYSPLIT Modelling by Dr. Mark Cohen, National Oceanic and Atmospheric Agency (NOAA)

In March of 1995, the Work Group on Parties Implementation of the SAB was among the first to preview a report by Dr. Barry Commoner and Dr. Mark Cohen from the Center for the Biology of Natural Systems, SUNY Queens College, titled Quantitative Estimation of the Entry of Dioxins, Furans and Hexachlorobenzene into the Great Lakes from Airborne and Waterborne Sources . Responding directly to one element of Annex 15, the report describes a first attempt to link, through modelling, the atmospheric deposition to the Great Lakes of dioxin and related compounds from major source sectors throughout the U.S. and Canada. Its outputs include a first analysis of the relative importance of distinct sources and source regions to overall deposition.

In subsequent months, the IAQAB contracted with Dr. Cohen to report on preconditions necessary to allow modelling of the atmospheric transport of the 1985 Critical Pollutants and other BNS contaminants. Specifically, a review of the adequacy of available information and programs in four areas was requested. The four areas are: 1. the capability or potential of the BNS pollutants to be transported for long distances in the atmosphere; 2. the availability and accuracy of emission inventories for these contaminants in the United States and Canada; 3. application of models to describe the atmospheric fate and transport of these compounds, with particular attention to studies involving deposition to the Great Lakes; and 4. the adequacy of existing ambient monitoring information to estimate the overall loadings of BNS contaminants to the Great Lakes and to serve as benchmarks against which to evaluate atmospheric models.

Dr. Cohen's work identified significant concerns over the incomplete data associated with physical and chemical properties of the contaminants, leading to uncertainties regarding their atmospheric fate and transport; the quality and comprehensiveness of emission inventories in the U.S. and Canada; the paucity of efforts to model transport and deposition, especially with respect to source-receptor relationships; and the absence of some critical pollutants from ambient monitoring programs.

Since 1997, the IAQAB and Dr. Cohen have continued to model atmospheric transport and deposition of a limited number of BNS contaminants. Due to the limitations outlined above, modelling could only be attempted for dioxin, cadmium and mercury. The dioxin project is largely finished and a mercury model should be available in late 2002. Completion of the cadmium modelling remains uncertain due to data and resource constraints. Modelling for atrazine, a non-BNS pesticide, also has been developed, but production of a final and more current version requires further support.

The IAQAB and WQB Workshop Romulus, Michigan (1997)

In May 1997, the IAQAB and WQB hosted a workshop in Romulus, Michigan to consider loadings, sources and pathways of persistent toxic substances. The focus was on the quantification of sources and pathways, particularly through the atmosphere, and further control of such substances, especially through `beyond compliance' initiatives. The boards, echoing the Cohen report to the IAQAB, noted that determining the significance of atmospheric deposition to the total burden of PTSs in the basin requires improved source and process inventories, a further determination of physical and chemical properties of the contaminants, further monitoring of their presence in the environment, and additional modelling to link sources and receptors.

The Delta Institute Workshops on Atmospheric Deposition (1999)

The first Delta Institute workshop, held in May 1999, focussed on the findings and implications arising from recent atmospheric research. The second workshop, held in October 1999, focussed on issues prompted by the research results of: setting priorities for research that address policy needs, integrating and making research results more readily available; coordinating a national strategy for estimating and controlling atmospheric deposition; and increasing international focus and coordination for this issue.

General conclusions reached at these workshops suggest that current research, data and models could identify source areas and source sectors, and, if they are substantial enough, certain large point sources of persistent toxic substances. To date, research continues to affirm that atmospheric deposition must be addressed if Great Lakes water quality goals for persistent toxic contaminants are to be met. It was acknowledged that modelers must present their work more effectively if it is to be used as an appropriate tool for policy formulation.

These workshops also reinforced the notion that the long-range transport phenomenon presents significant policy challenges. The Great Lakes region cannot address its contamination burden alone. National and international programs controlling or eliminating the use and release of persistent toxic contaminants would have a direct bearing on reducing loadings to the lakes. Further control of both domestic and international sources of atmospheric deposition depends on a strong, innovative strategy integrating regional, national, binational and international initiatives.

A key recommendation from the latter workshop advocated the development of an atmospheric deposition strategy using Lake Michigan as a case study. A strategy developed for a specific region using real data was seen as a useful step toward establishing policies for all the Great Lakes and in forming national and international policy-making on atmospheric deposition.