Thank you very much, Don.

Good Afternoon, everyone.

Today, I'm going to talk about three things {See Slide 2}; first, I'll give a little bit of context by discussing the atmospheric deposition path and where the work that I'll be discussing later fits; then I'm going to talk about dioxin and some work that has been going on for many years now (we are furthest along now on that compound), looking at the emissions inventories, fate and transport modeling and what it says about deposition to the lakes; and then I'm going to be talking about atrazine which is a chemical which we have been working on more recently. We have a little bit less to show for atrazine, but there are some interesting results.

This is a pretty big project and there is a number of people that have contributed either directly or indirectly over the course of the last several years {See Slide 3}, from project oversight and support, people at the IJC, people at NOAA, people on the Air Board here and the EPA. There's people that helped with developing the models -- Roland Draxler was the author of the model that I'm using for the study.

I'd particularly like to thank the people that helped with the mapping and the GIS analysis. Later on today you'll be seeing a lot of maps -- in fact, you might want to move up closer so you'll get the full pleasure of looking at them. Larissa, would you stand for a second? Larissa [Mathewson] of the Ontario Ministry of Natural Resources made all these beautiful maps you're going to see today, and it's not an easy thing to do, I'll tell you.

Then, of course, all the people that did all the emissions inventories -- as many of you know, it's very difficult to put an emissions inventory together; it takes many, many years and you're always improving them and they are always evolving; and then we have people who made measurements out in the field against which we can evaluate our models, and of course, the funders. The funders in the past that help us develop this methodology -- the Joyce Foundation, and the Alton Jones Foundation -- funded me when I was at Queen's College; now that I'm at NOAA, NOAA is supporting this work, along with the USEPA, Environment Canada and Ontario Ministry of Environment and Energy.

In my mind, I guess there are three main questions that we ask when you talk about the Great Lakes pollution {See Slide 4}; maybe I'm oversimplifying it but. . .

The first question is:

For a given lake, which pollutants are important? Which are causing harm to wildlife and humans?

The second question:

For a given lake and a given pollutant, which loading pathways are the most important? How does the pollutant get into the lake? Is it atmospheric deposition? Is it direct discharge through tributaries or runoff? Is it from contaminated sediment? Is it all three of those? Which are the important pathways?

The third question:

Once you've narrowed down the pollutants and pathways you are looking at and the pathways you are looking at, for a given lake and a given pollutant and a given pathway, which are the sources that are important? Ultimately we want to go toward the sources; try to reduce them; try to reduce the biggest sources in the most efficient way that we can.

With regard to the first question about which chemicals -- {See Slide 5} this is the list that's in the Binational Toxics Strategy Pollutants -- it includes heavy metals, some organochlorine pesticides, industrial miscellaneous chemicals, chlorobenzines, dioxins, PCBs, and PAHs. An open question is whether there are chemicals on this list that shouldn't be there anymore? And, are there chemicals that aren't on this list that should be there? These are other questions for another discussion.

With regard to the second question of which pathway is important for a given lake -- I've compiled here, {See Slide 6} some of the estimates I could find from previous studies that have looked at the percentage of the loading to a given lake for a given compound attributable to the air pathway. You can see that for Lake Superior and Lake Huron and perhaps Lake Michigan - for the compounds that have been studied - it looks like the air pathway was fairly important and maybe even predominantly important (except perhaps for atrazine). For Lake Erie and Lake Ontario, you see lower percentages; much more of the loading comes from direct discharge and tributaries. However, the atmospheric pathway is important - even if it is only 30 percent, its still a significant loading pathway. Unfortunately, for most of the pollutants that I showed you in the overhead just before this - the current list of pollutants of concern - we don't really have estimates yet of the loading percentages from different pathways. This is a problem when one tries to attribute the loading to different sources and tries to figure out how to develop strategies for reducing the loadings.

With regard to the third question - if you focus in on a lake and you focus in on a pollutant and the pathway -- can we estimate the various sources that are contributing to that pathway? For the air pathway, it's obviously very difficult because you don't have a pipe discharging material directly into the lake... the material is emitted in the air and it goes through all these very complicated processes and some small fraction gets deposited in the lake, it's very difficult to figure out exactly how much.

What I'm going to be talking about today is a methodology that is schematically represented here {See Slide 7} ... which I have called a "comprehensive atmospheric modeling methodology"... where we start with an emissions inventory - and you need to have the inventory as accurate as you can possibly make it - that is geographically resolved and temporally resolved. You combine that with meteorological data - wind speeds and directions, precipitation, and temperatures - and you combine that with your information about the behaviour of the pollutant in the atmosphere... Together these three elements are combined into an atmospheric model. The model that I'm going to be talking about today is called HYSPLIT, but there are other models that could have been used to do this analysis. You make predictions with the model... and you can predict, for example, what the concentration would be at a certain point, and you can compare that against ambient monitoring data and this tells you how valid your modeling is. If you can match the measurements that have been made then you know you haven't made horrible, horrible mistakes. You might have gotten lucky but. . .at least you are on the right track. You also can predict the deposition that would occur to each of the Great Lakes. What's difficult to do is to preserve the so-called "source-receptor relationships" -- the dotted red line on the schematic. It's difficult to keep track of all the different sources and what each one of them deposits in the lake. In the dioxin inventory that I'm going to be showing you in a few minutes, we have 46,000 sources in the United States and Canada and it's numerically difficult to keep track of all of them individually... This is why in most studies you don't really see the source-receptor linkage being investigated. In most studies, all of the emissions are thrown up into the air and an overall prediction of deposition is made... but they don't tell you where the deposition came from, and so it's not quite as useful from a regulatory or policy standpoint.

As Don (McKay) was saying, the Board was charged with trying to figure out where these pollutants were coming from and so we asked the question, which compounds on the IJC list could we do such an analysis for? There are several requirements {See Slide 8} -- you want it to be a pollutant of concern within the Great Lakes; we want there to be an emissions inventory that would be available or nearly available; there had to be ambient monitoring data for evaluation; and it had to be feasible to model.

It turns out that of all of those factors, the existence emissions inventories may be the single most important limiting factor. What I've listed here {See Slide 9} - and I think Harold's (Garabedian) going to talk about this as well -- are the compounds on the IJC priorities list and I've tried to show whether an inventory exists in the United States and/or Canada and I've given a subjective rating to the inventory - from my point of view (and we can certainly have a very interesting discussion about whether you agree with the ratings). The A's would be a very good inventory, fairly accurate geographically resolved inventories and then you go down all the way to D or E where the inventory doesn't exist. You can see there's very few A's on the list and there's very few compounds for which there's an A in both Canada and the United States. If you wanted to do a modeling study for the Great Lakes and you wanted to take into account both Canada and the U.S. you're sort of limited to maybe a few compounds -- mercury, dioxin, cadmium...that may be it...maybe hexachlorobenzine possibly if you can settle for a B inventory. {See Slide 10}

Two other compounds that we could probably study would be lead and atrazine - which are not on the IJC list - but for which inventories exist. In fact, I am going to be talking about atrazine today because we did have an inventory, we did have ambient monitoring data, and we did have the ability to study it.

What we determined is that with current knowledge, we could probably only look at a limited subset of the pollutants {See Slide 11} that we might be interested in... and I guess that's something to think about. We'd like to be able to do this kind of analysis -- not necessarily me, but all of us -- for all the pollutants of concern in the Great Lakes... and we're not there yet.

Let me go on now and begin talking about dioxin. Let me start with the emissions inventory for dioxin. What I have shown here {See Slide 12} is a US inventory on the top and the Canadian inventory on the bottom.

The US inventory was created at the Center for the Biology of Natural Systems at Queen's College (where I used to work before I joined NOAA). It's essentially based on a huge amount of EPA data and a huge amount of independent work that CBNS had done - literally hundreds, maybe thousands of phone calls to different facilities trying to verify their pollution control equipment, whether they had stack tests or not, their throughput, things like that. To the largest extent possible, the inventory is based on actual stack tests, but in general you don't have that many stack tests for most facilities, so it's generally based on emissions factors.

The Canadian inventory is from Environment Canada. You can download it from their website, and it was produced under the direction of Raouf Morcos and colleagues at Environment Canada, and the data were prepared for our use by David Niemi and Dominique Ratté at Environment Canada.

I've plotted the emissions inventory data here -- perhaps this is an unusual way to plot these data. The units on the Y-axis are nanograms dioxin TEQ per person per year. We wanted to compare the United States and Canada, and so, we normalized (that is, divided) by the populations. Canada is much smaller and so it would be very small compared to the U.S. if you didn't do that. You can see that for both the United States and Canada, municipal waste incineration -- the one on the far left -- is the largest source. It's a logarithmic scale, so be a little bit careful as you look at this graph. The quantities at the bottom are orders of magnitude less than the emissions quantities at the top. You can see there's basically three main categories of pollutants. I'm going to revisit those categories later: waste incineration, metals processing, and fuel combustion. For the most part, waste incineration is the largest source of dioxin in the U.S. and Canada and that is true in most countries. Metal processing maybe runs a close second and fuel combustion is third.

You can see that there are some holes and some gaps -- those question marks are sources for which we know emissions exist but which we didn't have enough information to make even the crudest of estimates. Some of the estimates on these graphs are pretty crude; some of the source categories aren't included in the EPA inventory because they didn't feel they had enough data, but we felt we had to try to include them because they were so big... sources like iron sintering and backyard burning.

Backyard burning - when people go out in their backyard and throw trash into a barrel and burn it - may be one of the largest sources of dioxin in the United States. We have it in here as actually relatively moderate although the EPA thinks it may be much larger than this. In the Canadian inventory, it's not included yet - so that's potentially a major omission in the Canadian inventory at this time. There are several other types of source categories that aren't included so I wouldn't say that we have, by any means, the definitive word on the emissions inventory yet for dioxin (or any of these compounds), but compared to some of the other pollutants, maybe we're further along for dioxin than most.

I'm going to show you some maps showing you this inventory in different ways. Here's a map {See Slide 13} showing the medical waste incinerators in the United States and Canada. Each facility (there's over 2000 in the United States and about 100 or so in Canada) has been mapped individually and the size and colour of the circle indicates the emissions from the facility. These are estimates, and I would say the medical waste incinerator emissions estimates are quite uncertain. These are almost all based on the emissions factors; based on only a few tests at a few facilities and so we're extrapolating from probably on the order of 10 facilities up to over almost 2500 facilities. There's quite a bit of uncertainty as to the degree of pollution control at medical incinerators today. These estimates have been made assuming a pretty high degree of control that I'm not sure exists.

Here is the same kind of picture {See Slide 14} for municipal waste incinerators; this was the largest emissions source... more than any of the other categories that we looked at, these estimates were based on stack tests... almost all of the emissions from Canada were based on a series of stack tests, and many of the emissions in the U.S. - especially from some of the bigger incinerators - were based on stack tests.

A few points to make about this -- one is that even though there's been tests on the given facility, nobody really believes (I don't think) that if you test once a year or if you test once every three years or every five years that you are necessarily getting average, representative emissions. Maybe you get lucky and you're getting a worst-case situation or an average case but a lot of times those tests are made under sort of best-case conditions. There's was a paper presented very recently at the World Dioxin Conference in Vienna showing that when long-term monitoring was performed at a facility -- I think it was in Germany -- that during start-up and shut-down operations, the incinerator emitted much greater quantities of dioxin, and that emissions during those periods were very significant even though the start-up and shut-down periods were relatively small compared to the overall operating periods of the incinerator... the emissions were high enough during these short periods that you would significantly underestimate the total emissions from the facility if you only measured during the continuous operation. These things [incinerators] are so big they don't shut down and start up right away like your stove; they take as much as a day to get up to temperature or a day to shut them down, and while that's happening you're getting a lot of inefficient combustion and a lot of different kinds of combustion than you would normally have.

Another thing is that there's been some significant changes. The inventories I'm showing you today are for 1995 - 1996. That big red dot up in Canada in Quebec - that incinerator actually is still operating, but last year, I've been told, it underwent a series of renovations and process changes and that its' emissions went from around 60 grams TEQ per year - which was how it's plotted here (in the 50 - 100 category) - down to less than 1 gram TEQ per year ... and so that's good. You could take one of those dots off the map right now or make it a small dot at this point. So, remember I'm talking about what occurred in 1996. I wish I could tell you exactly what is happening today but it's hard to catch up; the emissions situation is always changing.

Here's a map of cement kilns burning hazardous waste {See Slide 15}. Making cement is a very energy-intensive operation and most cement kilns use coal; and some use oil and about 20 or 30 in the United States and a few in Canada have gone to using hazardous waste for fuel over the last decade or so. The emissions of dioxin from those facilities are much greater per ton of cement produced than from facilities that don't use hazardous waste as a fuel.

Here's an iron sintering plant {See Slide 16}. To my knowledge, even though it's one of the largest sources, we think, in the United States, there have been no tests yet in the United States on these kinds of facilities. Our data come from tests in Europe and tests in Canada. There were two iron sintering plants in Canada. Both were tested; the one in Hamilton still operating; the one in Wawa, just up north of Lake Superior, has recently been closed down.

Here's a map of the overall inventory {See Slide 17}; we add everything together from all those sources. It's a little hard to make heads or tails of this but you certainly can see that there's a lot of dioxin coming from a lot of different places; mainly on the eastern half of the countries. We've plotted the emissions density here; I've divided the total emissions for a county by the area of the county (Larissa divided those by county when she mapped this) to get the emissions per kilometer squared. This was done so that if the county was very big, it wouldn't be penalized just because it was a big county. It's very interesting when you map these data - depending on the assumptions that you make and depending on how you map them, you don't necessarily come to different conclusions but you definitely can get a different picture....

This {See Slide 18} is the same data exactly but here we aggregated, instead of on the county basis, just on a province and state basis. We've also, instead of giving you the areal density of emissions we've just given you the percentage of emissions so, for example, the "brown" states are estimated to emit between 8-11 percent of the total dioxin emissions in U.S. and Canada. So Florida, Ohio, and New York are estimated to emit around 8-11 percent of the total emissions; the "red" states -- Pennsylvania, Illinois, Indiana -- between 5 and 8 percent etc. This picture is very interesting - you can see, actually, that the Great Lakes are surrounded by some of the worst emitting states and provinces based on our current information.

Here's yet another way to look at the inventory {See Slide 19}. Here we've taken the top 20 counties in the United States, the top 10 grid areas that we we're using in Canada, and mapped the different sectors of the emissions inventory that make up the emissions of those top emitting counties... You can see that most of the top are that "gold" color which is the MSW incineration although there's a couple, you see, of the aluminum processing and the copper processing (smelting) operations and also the "blue" of iron sintering. That is consistent with the initial graph I showed, a plot which showed that the municipal waste incinerators were the biggest source in the inventory.

Now, we don't want to know just where the dioxins are coming from; we also want to know where it's going. In particular, we want to know how much is getting into the Great Lakes and how much from each source is getting into the Great Lakes. In order to do this we have to connect the sources to the lakes with the model. This is a schematic {See Slide 20} of some of the processes that you have to consider when you doing this type of modeling. When material is emitted from a smoke stack, several things begin to happen all at the same time. One thing that happens is the material is blown downwind -- this grey plume is getting bigger and bigger as it goes further and further away from the source. You can see it takes a while for the plume even to hit the ground so sometimes, actually, right within a couple hundred feet of the incinerator would be the safest place to be because it the plume hasn't hit the ground yet. When the plume hits the ground you can get what's called dry deposition and that's just deposition to the ground without rain. If it happens to be raining through the plume, you can get wet deposition of the material. The material can react with hydroxyl radicals (not drawn to scale here!) and even if the material is deposited it can be re-emitted... I've shown the path of one hypothetical molecule that was deposited but then got re-emitted maybe a week later.

This is a similar schematic {See Slide 21} . This is how we do the modeling... We're using a Lagrangian puff model. Essentially, a puff is emitted from a source, say, every hour over the course of a year; you emit these puffs and you just follow these puffs everywhere they go. You let the wind blow them around; you let the various chemical mechanisms operate on the pollutants in the puff; you let wet and dry deposition occur; if the puff happens to be over one of the lakes and it deposits, then you count that as deposition to the lake. So you can follow, for a given source, where its emissions go if you do this kind of modeling. This type of model is specifically designed to keep track of the source-receptor relationships that I talked about earlier. You'll see the importance of this later when I talk about the results from this.

Before I get to the overall deposition amount, here are some of the basic modeling results. This map {See Slide 22} shows the "efficiency of transfer". Imagine, for a minute, that there were equal emissions sources everywhere in the entire United States and Canada. How much dioxin would get into - in this case Lake Superior - if there were sources emitting dioxin everywhere exactly equally. This shows you the isopleths (or the layers) of efficiency of transfer to Lake Superior. In the close-in layer -- the brown layer -- about between 5 and 10 percent of the dioxin emitted from a source (in that layer) would be predicted to be deposited in Lake Superior. In the red layer, about 2.5 percent of the emissions are deposited, etc. What happens is, of course the closer in you are the bigger the impact is [the higher the transfer coefficient is]; the further out you get, the impact gets less and less. But you get more and more sources as you go out further and further; so it's kind of an interesting balance. It's not really clear yet, at least from this map, whether the close-in sources will be more important or the far-away sources will be more important.

Also, before I get to the deposition results, I'd like to discuss the model evaluation procedures that we went through. We took all the emissions inventory information and all the modeling information and we tried to predict the ambient concentrations for the locations and times for which we had actual monitoring data for; it turns out there aren't at lot of data for dioxin {See Slide 23}. It's very expensive to make measurements.

In 1996, there were actually only five measurements that were suitable for this kind of work; and by suitable, there are two criterion I used to see if the data would be suitable.

One is, for the most part, it must be in a rural location. At this point, we're not really modeling urban areas. You would have to know the sources down to the block level and the meteorology down to a much finer scale than we are doing. Here, we have very large grid sizes so we're only able to look at - for model validation purposes - rural areas.

Second, we want long-term measurements. Each of these measurements is actually a one-month measurement - you actually take a continuous one-month sample of dioxin. What you find is, if you go out and take only a one-day sample of dioxin, you get mostly "non-detects". You get a lot of "non-detects" and these are hard to deal with. Sampling (and analysis of the samples) is very expensive and its frustrating when you spend a lot of money to get "non-detects." Also, the state of the modeling is not that accurate at the present time .... you can imagine that things are very sporadic out in the environment ... you can imagine a puff or a plume of material going past a sampling point and you could miss it by a few hours and yet your sample result would appear to be way off but perhaps you weren't that far off. The longer-time-period samples are more forgiving, but they are still a useful way to see if you are doing reasonably well.

We did OK. I wouldn't say this is fantastic agreement. The "green" is the model predictions in the air and the uncertainties in the model predictions are the error bars in these model estimates, and to the right of each model estimate is the measured data (in red). We're definitely within the uncertainty of the measurements (note that this is a logarithmic plot, there is quite a bit of variation). There's about a factor of 10 - you can see - in the modeling uncertainty. Most of that uncertainty comes from the emissions inventory. We can quantify that easier than the uncertainties in other parts of the modeling. The uncertainties in the measurements actually are pretty small in this scale. A lot of them had duplicate samples right next to each other, and essentially the same result was obtained, etc. I would say that we're on the right track here. This is encouraging, but we could probably still do better.

As an introduction to the results, results for two illustrative case-studies will be presented. {See Slide 24} In these case studies, emissions of dioxin from two different metropolitan regions, Hamilton, ON and St. Louis, MO-IL, are estimated, and their impact on atmospheric deposition to each of the Great Lakes is considered.

The total emissions from the Hamilton area are estimated to be on the order of 30 grams TEQ/year. For the St. Louis area, the emissions are estimated to be on the order of 133 grams TEQ/year. For both areas, metallurgical processes are the largest emissions category. Significant emissions also are contributed by waste incineration processes. Emissions from fuel combustion is relatively insignificant for both areas.

The estimated efficiencies of transfer of emissions from each of the case-study areas to each of the Great Lakes through the atmospheric deposition pathway are shown in this figure. For example, it is estimated that approximately 3.6% of the dioxin emitted into the air in the Hamilton area will be deposited in Lake Ontario over the course of a year. Due to the proximity of Hamilton to Lake Ontario, this transfer coefficient is higher relative to that for the other Great Lakes. For emissions from the St. Louis area, the largest transfer coefficient is 1.5%, to Lake Michigan, the closest Great Lake. Generally, the larger the lake, the higher the transfer coefficient will be, as there is a greater surface area for atmospheric deposition. However, the distance of the source from the lake and its orientation relative to the direction of prevailing winds play a major role as well.

The final step in this analysis is the use of the estimated transfer coefficients to assess the atmospheric deposition impacts of the emissions on the Great Lakes. This is done by multiplying the emissions by the model-estimated transfer coefficients. Results of this multiplication are shown in this figure. Thus, for example, of the 130 grams TEQ/year emitted from the St. Louis area, 1.5% - or about 2 grams TEQ/year - are deposited in Lake Michigan. Lesser amounts are deposited in the other Great Lakes from this source area. Analogously, the Hamilton area contributes approximately 1.1 grams TEQ/year to Lake Ontario, and lesser amounts to the other Great Lakes.

In general, dioxin emitted to the atmosphere from any source is subject to dispersion (dilution) in the atmosphere, transformation (e.g., chemical reactions), and deposition to the earth's surface. These phenomena all tend to reduce the concentrations at greater and greater distances from the source. However, for dioxin, these influences are not so significant that regional and long-range transport can be ignored. This example shows that emissions from outside the Great Lakes basin can be as or more significant than emissions from within the basin.

This map {See Slide 25} essentially shows the multiplication of that overall source inventory map that I showed you times the transfer coefficient map (the one with the orange and the yellow and the red and the brown on it). Essentially, you take each county or each grid in the U.S. or in Canada and you take the total dioxin emissions multiply it by what the model is telling you should be the percentage of that dioxin which gets deposited in a given lake. In this case, this map is for Lake Superior. This is what happens when you take emissions inventory multiplied by transfer coefficients for Lake Superior and this shows the contribution on a micrograms toxic equivalents per kilometer squared per year basis. Again, we divided by the area of the counties so that big counties wouldn't be penalized just for being big. You can see that there's a lot of materials coming from a fair distance away -- several hundred kilometers -- and also close in. There would be a different map for each lake. We have maps prepared for each lake but I just wanted to show you one here for Lake Superior. These {See Slide 26} are exactly the same data but here it's aggregated on a provincial and state basis, and instead of giving you the areal deposition contribution, I'm just giving you the percent of the total deposition that arises from within a given state (and again this is for Lake Superior). The highest fraction shown -- the "brown" fraction -- shows that 12-17 percent comes from just two states, Illinois and Indiana (for Lake Superior) and the "red" states contribute between 6-12 percent of the predicted deposition.

Another way to look at the data {See Slide 27} - and this kind of analysis you can only do by GIS... and you have to have a good GIS person to do this kind of analysis - we've done an analysis where we've looked at the emissions and deposition that arises at different distance ranges from the lake; the orange bars are the emissions that occur in, say, between 0 and 100 kilometers from the lake; 100 to 200 km; 200 to 400 km, etc., in these distance rings away from the lake, and the green bars are the deposition that arises from those distances... And so this is an attempt to answer the question I was asking earlier -- how important are the close-in sources relative to the further-away sources? Well, you can see for Lake Superior and perhaps Lake Huron, most of the deposition (the green) is coming from relatively far away -- 400 to 700; maybe 1000; even as much 1500 kilometers away from the lake. That's pretty far; we're talking 500 to 1000 miles away from the lake that you're getting significant deposition contributions from -- whereas for the bottom three lakes, Lake Michigan, Lake Erie, and Lake Ontario you're getting a pretty big hit close in; 30, 40, maybe 50 percent comes from within first 100 kilometers of the lake. Even for these three lakes you do get some from further away, for example for Lake Ontario, about 2/3 of the deposition comes from further away... But you can see the pattern looks a little bit different for the top two and the bottom three lakes.

You can make sense of this from {See Slide 28} what Ron Hites call the yellow flashlight map - where you give everybody in the Great Lakes basin a yellow flashlight and at midnight one night you ask them to turn it on and you take a satellite picture of that; at least I think that's how they make these maps (!) .. You can see that for Lake Michigan, Lake Erie and Lake Ontario there's really a lot more urbanization (as we all know) right around the lakes and thus perhaps it's not surprising that those three lower lakes had the highest percentage of their impact for dioxin coming from close in whereas for Superior and Huron there was not all that much that came from close in sources.

Another way to look at the data - and again you can really only do this with the GIS analysis -- here {See Slide 29} we plotted for each lake the relative proportion of the dioxin that comes from within the watershed vs. that which came from outside the watershed. Now, as far as the air deposition pathway goes, the watershed boundary is not really that important... but people do think in terms of the Great Lakes basin and the watershed so, perhaps, it's politically important. You can see that on the order of about 20 percent or so - and this is remarkably consistent from lake to lake - about 20 percent or so of the dioxin comes from within the Great Lakes watershed; about 80 percent comes from outside.

Here {See Slide 30} (NOTE: This overhead was shown in the Q/A period) is a graph showing the percent of deposition to each lake that comes from each of the Great Lakes states and provinces, and, that which comes from the rest of the U.S. and the rest of Canada. You can see that about 75% of the dioxin deposition to each lake comes from the Great Lakes states and provinces.

Here {See Slide 31}, I have plotted for the three main dioxin source categories -- remember when I showed the inventory at the very beginning, I said there were three main categories, waste incineration, metals processing, and fuel combustion -- the deposition occurring per capita for each category, for each lake, for each country -- in the United States and Canada. The U.S. is in blue and Canada is in red. For most of the lakes, U.S. waste incineration looks to be one of the largest impacts per capita on most of the lakes. For Lake Ontario, Canadian incineration also seems to be rather important as well.

Now we switch gears, rather quickly, to atrazine {See Slide 32}. For atrazine, we picked the year 1991 because that was the year we had an inventory available and monitoring data available. Atrazine is an herbicide; it is used on corn and sorghum largely, although it is used on other crops e.g. sweet corn, seed crops, and sugar cane. It is one of the most widely used current-use herbicides in the country, about 33 million kilograms per year is used of active ingredient in the United States and Canada. We based our analysis on atrazine use in agriculture and we didn't include things like use on golf courses and residential use - which are a fairly small percent of the total usage. It's mainly used on "field corn" (the corn that is fed to animals, not the corn that we eat). The kind of corn we eat is called "sweet corn". The way we did the estimate for emissions is quite a labor-intensive task. You have to look at the acres that are planted in each of these crops of sorghum and field corn and the other crops; you've got to figure out for each area, which percentage of those acres are treated with atrazine; you have to know what the rate of atrazine application is; and all these factors vary over geographical regions quite a bit depending on the weather and some other factors. Atrazine, even though it is not on the IJC list, was chosen as a pollutant of concern in the Lake Michigan LaMP process and it was also chosen as one of the four pollutants the EPA is looking at in the Lake Michigan Mass Balance study.

This {See Slide 33} shows the annual usage of atrazine in U.S. counties and Canadian grid zones. You can see that it is largely used in the midwest, and you can also see that if you have to guess which lake was going to get the worst impact you might guess Lake Michigan (later we'll see if that guess bears out).

Here is a map {See Slide 34} showing the geographical distribution of emissions factors for atrazine. These are the annual percent of the applied atrazine that are predicted to be emitted over the course of the year. These overall emissions factors are based on the modeling work of Trevor Scholtz of ORTECH, in Canada. The emission factors are based on the soil type and the soil moisture and the weather, etc. We also had to estimate the time course of emissions after application - not just the total fraction emitted - and to do this we looked at published field studies of atrazine volatilization. These studies show that the emissions are maximum in the first week after application, and decrease each week thereafter. Most of the emissions appear to occur during the first 8 weeks after application.

This {See Slide 35} is a map of the total atrazine emissions, where we've essentially multiplied the map of the usage times the map of the emissions factors to get the total emissions. Here we've divided by the area of the county to get the areal density of emission. So you see, for example, that even though the emissions factors were high in the Rocky Mountains area, there's hardly any atrazine used there and so there aren't a lot of emissions from that area.

Let's go to the movie now. {MOVIE} Here's a "movie" showing atrazine emissions during 1991. Atrazine is not applied on a regular basis; it is applied just before the crops are planted or just after the crops are planted, so it turns out that most of the atrazine is used and applied in the spring and the summer... and this situation requires a different kind of modeling approach. With dioxin you have much more of a continuous type of release; incinerators are more or less operating all of the time. There's certainly some temporal variations in dioxin but not nearly like what you have for atrazine. For atrazine, when they make the measurements you see huge spikes in May, June, July and then you don't see much else for most of the rest of the year. That was actually quite a more difficult modeling problem. We'll run through this twice, so you can get oriented. What you are going to see is, week by week, the estimated atrazine emissions in the United States and Canada. This is Week 1,2,3, 4, 5 - we're still in the winter right now. You see very little emissions are taking place -- essentially nothing is going on. You see a little bit in Florida, and southern Texas, in late winter... and you see that as we're getting into early spring, the growing season (and atrazine usage and emissions) is slowly moving northward until finally we see emissions from Canada... the growing season is relatively short in Canada. Then as we get into the end of the year, there are very little emissions. So you see - we'll run this one more time - there is a period of about 10 or 15 weeks in the middle of the year where the bulk of the emissions occur. In Florida and Texas they are growing sugar cane and a little bit of sweet corn during the winter (that's partly how you can get some fresh corn in your supermarkets in the middle of the winter).

Here is another view of what I showed you in the movie. This {See Slide 36} shows the weekly percentage of atrazine emitted over the course of 1991. I put the month down there, but actually each one of those data points is a week... You can see that for a period of about 10 or 15 weeks most of the atrazine is emitted... In fact, in order to save time in our modeling we actually only modeled a 20 week period - the weeks shown with the red line through them. Probably we should have continued out a little further, but we didn't realize that at the time. So, the results I'm going to show you are actually for only about a third of the year, but it's the third of the year where pretty much everything that's happening with atrazine is going to be happening.

This {See Slide 37} gives you a sort of a feel for some of the modeling and I wanted to show this because there are two points I wanted to make here. One is, look at the top graph -- this shows results for a particular point in central Iowa (one of the heavy usage areas and emissions areas for atrazine)... if you just had a source there emitting constantly for those 20 weeks in the spring and summer of 1991, what would the impact be on Lake Superior? It's quite interesting -- you can see that, first of all, some of the weeks the wind doesn't even blow from Iowa to Lake Superior so you get no predicted impact; some of the weeks -- the clear part of the bars is dry deposition and the black part is wet deposition - some of the weeks you don't have any rain (like in that 5th week) so there wasn't any wet deposition etc. One of the things that always strikes me when I do this modeling is how episodic things are, and the importance of doing the modeling for a long time period. You wouldn't want to do your modeling for just a week or two because you might really miss something... or if you did your modeling for a week or two, you would want to make sure you picked the right week or two to do the modeling.

The second graph shows, for a source in northeastern Nebraska - over the entire 20 week period -- its impact on different receptors in the Great Lakes region. You see the Great Lakes, Lake Erie, Lake Michigan, Lake Superior, Lake Erie, and Lake Ontario sort of in the middle of the graph here. In this study, we looked at some other smaller drinking water reservoirs and agricultural areas as well. You see a fair amount of difference, again, because the winds are blowing in different directions.

We applied the same kind of modeling and model evaluation procedure that we did with dioxin - with different parameters, of course - but the same approach was used. In the case of atrazine, we have a wonderful data set for model evaluation. It was taken in 1991 -- the same year that this modeling study was done for - by Donald Goolsby of the United States Geological Survey. He took the normal samples that we take in our country for acid deposition - he took those samples and also analyzed them for atrazine for about a year and a half (during 1991 and 1992). He did this at about 100 sites across the United States. We picked about 20 sites (the 20 shown on this map here {See Slide 38}) that were away from the most intense areas of usage. We are modeling regional and long-range transport. We're not doing the kind of modeling that you could do if you were sitting right in the middle of a corn field using atrazine and trying to model very local effects. That's not the kind of modeling we are doing; we were looking at bigger scales. All these points are areas where some atrazine is used in these counties, but not very much. They are generally away from the heavy, heavy usage areas.

This graph {See Slide 39} shows how we did, in terms of model evaluation. What I've done here is compared for each of 20 sites, for the 20 weeks, the actual vs. the predicted wet deposition of atrazine. I've plotted the deviation between the two... You can see that, for the most part, we did OK. There's definitely some outliers and those outliers can be important -- where we predicted or overpredicted quite a bit. On average, the standard deviation between the predictions and the measurements is about 25 percent of the mean. In some ways, you could say we were plus or minus at a factor of about 25 percent. Not too bad. If you take out the six worst points, this reduces to about 11 percent -- that is, the rest of the points were only off by an average of about 11 percent or so. This is not too bad, considering all the uncertainties involved here.

Here's a map {See Slide 40} showing the same kind of results that I showed you for dioxin; this is a map of the atrazine contributions (as a result of the modeling) to Lake Superior. You can see that there's quite a bit that is coming from fairly far away... so that even though atrazine is somewhat different that dioxin - it's not thought to travel as far in the atmosphere... certainly it reacts with hydroxyl radical much faster; its half-life is probably only about in the order of eight hours to a day... still, pollutants can travel about 400 kilometers in a day, and even if the half-life is only a day you still have half of the pollutant left at the end of the day, and there is half left that can be transported further distances. As you can see, there's a fair amount of atrazine that is predicted to make it to Lake Superior from reasonably far away.

This is a comparison {See Slide 41} of our preliminary results based on this modeling for each of the Great Lakes. I plotted also the results of Schottler and Eisenreich who published a paper in ES&T in 1997. They also based their estimates on Goolsby's measurements. We're fairly close (although this is a logarithmic plot), especially for Lake Erie. Lake Michigan looks like we're pretty far off; we had a number of around 10,000 - 12,000; they had a number around 2,000 or 4,000 so it's a factor or three or so. Actually, I'm sort of wondering about their numbers for Lake Michigan because when you look at the usage of atrazine and how big Lake Michigan is and how much closer it is to the atrazine usage than the other lakes, I'm wondering perhaps that answer may be somewhere in the middle or at least higher than Schottler and Eisenreich's estimates.

One of the things with modeling is that it helps you fill in the spaces between the measurements. Modeling can be a very nice adjunct to measurements and by the same token, without measurements you really can't do the modeling. You need the measurements to do your model evaluation.

Let me just close with a few conclusions {See Slide 42}. This is just from my viewpoint (I think the board is going to be offering some conclusions), and I hope in our discussion we'll have some more conclusions offered by some of you.

First of all, for many of the pollutants of concern in the Great Lakes we don't have really have enough information to determine the relative loading pathways; whether the air pathway or the water pathway or the contaminated sediment pathway is the most important.

Secondly, for a given significant loading pathway, it's important to determine the contributing sources and source regions, and for the air pathway, you need to use a model to do that linkage. In order to be able to do that type of modeling, you have to have accurate air emissions inventory. For most of the pollutants of concern in the Great Lakes, we don't really have an accurate emissions inventory yet. This is a problem.

When you do have an inventory, then this HYSPLIT-based modeling approach that I've shown you today seems to be a fairly useful tool. It gives you some idea of where the sources are that contribute to the Great Lakes deposition. But, I want to point out that other models could be used as well, as long as some effort is made to preserve the source-receptor information. There are many models out there today that you can theoretically do that with, but, it may be numerically challenging to do so.

For dioxin and atrazine, the two chemicals that I've shown you today, we're obtained some idea, I think, of the relative importance of the nearby and the faraway sources, and different source sectors for dioxin, but of course, this is dependent on the emissions inventory. If we had something wrong in the inventory, then clearly this error is going to propagate all the way through the calculation and lead to erroneous results. Hopefully, with 46,000 dioxin sources in the inventory - while I'm sure some small percentage are wrong - hopefully not enough are wrong to make the results dramatically in error. It looks like atmospheric transport from both inside and outside the Great Lakes basin is important. We wouldn't want to ignore either of those. I think that this point, certainly, is an important one for the LaMP process. That is, sources both inside and outside the basin should be considered for loading contributions.

This comprehensive atmospheric modeling approach that I've talked about today is an evolving field. We're not nearly as far along as we are, say, for acid deposition (where this kind of modeling has been going on for 20-30-40 years) or, say, for urban ozone, where again, modeling has been going on for quite a long time. You don't see much of this type of modeling for things like dioxin, mercury, and other toxic compounds like these, but it's starting to happen more and more these days, and the results are going to getting better and better... so stay tuned. I hope to be able to give you even more accurate results in the future.

Finally, I hope that the work today demonstrates the promise of this analysis and that we'll be encouraged to develop the information necessary to apply it to the other pollutants of concern in the Great Lakes.

Thank you for your attention.

Thank you very much, Mark, for an excellent presentation on the methodology and the work that you've done. It's been very informative and it's certainly been very time-consuming to produce what was done.

We have time for some discussion before we take a break and so I'd like to take questions and Mark, I think, will be pleased to answer them.

Q:

I'd like to know where . . .Ohio is on top of the list top for three air emissions and dioxins and yet you showed that Lake Erie - the discharges were more important than air deposition for dioxin to the lake - so am I to conclude that there's a whole lot of pipes stuck into the lake or is there some other source upstream?

A: (M. Cohen)

Probably both. I'm not sure of the accuracy of those estimates on that table. We tried to do, five years ago or so, an assessment of both the water and the air loading and if you think that the air pathway that I talked about today is complicated because you have to do the modeling stuff (it's very hard) but the water pathway is very complicated because there are no data. There's a lot of pipes and there's a lot of sewage treatment plants discharging things and industry discharging things into rivers but there are no data. What do we have for the percentage from Lake Erie? I think maybe it's 40 percent; I'm not quite sure. In any event, that's an open question and I think we need to do more. If you're on a LaMP or if trying to do something to reduce the pollution that's getting into these lakes, you really want to have a list of where it is coming from; it seems like an easy question. It seems like that's the first place we should start from but it is awfully hard to get there with the current data that's out there.

Q: How available will this information be?

A: (M. Cohen)

We're going to try to publish some of this on the web. It'll either be in the IJC and/or the NOAA website. I'll try to make it so you can download all the pictures and text and the like so there's a draft report like this which is on the website now; on the IJC website in the priorities chapter. You can download that right now. (There are copies of the report on the table back here -and if you'll just leave your name and your address and we can send you the report.)

Q: (Peter Wise)

First of all, thank you for an excellent presentation. Two questions - one, you showed a slide that showed a percentage of dioxin coming from within the Great Lakes basin. I must assume that in states like Illinois and Indiana, that that's a very small portion of those states and if you looked at the percentage coming from the Great Lakes states it would be a much higher percentage.

A: (M. Cohen)

Yes, it absolutely is. In fact, I have an overhead for that. It's about 75 percent of the dioxin comes from those Great Lakes states. Let me show you.

Followup question (inaudible)

M. Cohen -- That's exactly right. In fact, here about 75% comes from the Great Lakes states and provinces. That's a question that I might even pose to you. How do we define the Great Lakes basin? It's a question when you start doing the GIS stuff. When Larissa was doing this, the questions was "well, what do we use as the boundary for the Great Lakes basin to make our cutoffs for?" What we used was the watershed as defined by the USGS but when you do that, Chicago is not included.

Followup observation:

Yes, Illinois operates incinerators that are not in the Great Lakes basin. . .(balance inaudible)

A: (M. Cohen)

Yes, yes. Absolutely.

Q: (inaudible). . .in terms of your emission factors for atrazine, do you consider the affect of atrazine handling training programs in particular states and province or do you assume every farmer is using atrazine in the same fashion across the Basin?

A: (M. Cohen)

There are different emissions factors for atrazine depending on the type application; whether it's pre-emergent and put into the soil where it's plowed under; whether it's sprayed on top; whether they put starch encapsulation. Some attempt was made in the inventory to take into account those different factors. Surprisingly, in the field studies that I've seen there's not a huge difference in the emissions. You seem to get about 10 or 20 percent of the atrazine get's emitted and it seems to always happen about the first month or two. All the experiments I've seen, the first week you get a big spike of emissions and it slowly drops down over the next month or two until by eight to ten weeks after the application you are just getting a very slow, very small amount of atrazine.

Q:

(inaudible). . .Our farmers may be thinking that if a 3 lb. application is recommended, 6 lbs. may be much, much better. . .

A: (M. Cohen)

That's an interesting point. So it could be that when we got the atrazine application amounts we were using the recommended amounts from the company. The point you make is well taken. It could be underestimated.

Q: (Bruce Walker)

First of all, thank you for providing the updated information on the Quebec City garbage incinerator. I learned something from your maps -- that Newfoundland is a specific source of dioxin.

A: (M. Cohen)

Let me comment quickly on that. The people at Environment Canada that gave me the inventory are wondering about that. They think it's the TP burners up there that are burning municipal waste and they're not sure what the emissions factors should be for that. God bless them, they put a pretty high factor in there. They think they may have overestimated it. That's an open question. As are most of the points on the graph.

Q:

(inaudible) My question does relate to the atrazine emission inventories -- what I'd like to ask you - as a modeler, would it help you in future studies if several stationary sources were required to report under either the TRI or, in Canada, the NPRI -- their dioxin emissions; or for that matter other PTS substances like mercury. As a modeler, would that provide you with much better data?

A: (M. Cohen)

Absolutely. The TRI, the NPRI are one model of trying to collect the data. It's generally self-collected or self-estimated and so there are some issues with that. There are also issues, of course, in the TRI (United States) not all industries are covered so, for example, municipal waste incinerators aren't covered under the TRI in the U.S. and neither are medical waste incinerators. So, you start knocking off the two biggest sources and so it may not be quite as interesting. I also would point out that whether or not you want to do the modeling like I've shown today, you still need a good inventory and so, yes, better information would be useful to me but it would also be useful for anybody in this room who is trying to look at what's happening in the Great Lakes and where the sources are coming from and so forth.

Q:

(inaudible) You based your atrazine data on 1991, is that right? (Yes) Wasn't there a label change in 1992 so the farmers would be using less atrazine after that date?

A: (M. Cohen)

Yes. There may have been. I confess I'm not totally sure. I know the usage hasn't dropped that much. I've asked some people who helped me with this inventory just last year, "has the atrazine dropped dramatically in the last eight years?" and they don't think that it has but there very well may have been some changes. Are you thinking of acetachlor and atrazine specifically? [Atrazine] Are you aware of data recently that has shown any dramatic decrease in the usage of atrazine in the United States and Canada? The last data I have is from 1991. '91 OK. I've seen some numbers from '95 or '96 and they didn't drop that much.

[Chair asked that microphones be used as session was being recorded]

Q: (Daniel Green from Montreal)

Looking at studies -- much smaller studies; particularly Swan Hill hazardous waste facility, done by Schindler and ???, what they did (I don't know what model was used) is they modeled emissions and then they went to see what deposition and bio-uptake; if they found it in wildlife etc. and it didn't always concur. As environmentalists and amateur GISers we are using these things to predict uptake in wildlife. It should be real life situations in citing an incinerator or siting of a plant that would generate these things. Are you working with field people, toxicologists to find out if your deposition can explain bioaccumulation in wildlife?

A: (M. Cohen)

That's an excellent question. I did actually do a study a couple of years ago on dairy farms where we looked at the same kind of study and instead of the Great Lakes, we looked at eight farms in Wisconsin and eight farms in Vermont. We measured dioxin in the air; we measured dioxin in the crops that were growing; the corn and alfafa and pasture crops that were growing; we measured dioxin in the milk and in also the feed crops (the food that the cows were eating). So that was more of an ecological uptake-type of a study. There's very little data on dioxin uptake. You find that you can measure it in milk; you can measure it in beef; so you can measure it in the animals or road kill or however you get it but when you measure it in plants it hasn't bioaccumulated yet and there's very low levels and so it's a very difficult study to do. That question of "is the deposition really right?" -- it's very difficult and subtle to answer that. We don't really know how to measure dry deposition very well, for example. One way you can do it is to compare your predictions against what you find in the plants and we got a pretty reasonable agreement, actually, with what we thought should be going into the corn plants as they were growing, etc. as to what came up. It's very difficult to get a match. The people that worked on the Swan Lake study -- and this has happened all over the world -- when you try to model from a given incinerator (this happened in Vermont in the Rutland Study, too) and say "what should the dioxin be five miles away? it's very rare that you can get a good correlation between your predictions and the actual measurements. One of the things, I think, that causes that is that -- remember those maps that I showed you earlier of the dioxin coming from everywhere -- well, the background is actually coming from all those other sources out there; I showed you how far away the impact can be for dioxin. If you're standing at a certain point -- say three miles away from the Swan Lake incinerator -- and you ask the question "where does the dioxin come from that gets to you at that point or that sampling location?" well, on the days that the Swan Lake incinerator is emitting and blowing right in your direction, it may be the biggest impact by far but on the days when it's not blowing in that direction, and that may be 90 percent of the days depending on where you are, then you might be getting stuff from a lot of other sources all over the United States and Canada. In fact, you saw we weren't just seeing stuff right around the lake being the most important; for dioxin; we're seeing stuff for 500 to 1000 miles away making significant contributions. I think that could be part of the problem. You can't just look at one source in isolation and say "will that give me the answer?" at a particular sampling point.

Q: (Tom Dawson, Director of the Wisconsin Strategic Pesticide Information Project)

I have two questions. One is, has anyone attempted to do a mass-loading type of analysis as to the sheer amounts of atrazine that obviously are going into the environment, or, another way of putting it, being wasted by running off into streams or, going up into the air?

A: (M. Cohen)

I've seen a few papers where they tried to do that. We do a field study and you apply atrazine and you say, "how much goes up in the air and how much gets leached off, etc?" There's very few studies out there that have looked at that. About 20 percent or so gets volatilized; I don't know what the fraction is. It depends.

Q: (Tom Dawson)

I just wanted to know if it had been done because I think it would be a valuable exercise in form for a lot of reasons including providing some political impetus to do something meaningful here. The second point is that I just came from testifying at the IJC hearing basically to plug my project which is to create in the state of Wisconsin a pesticide-use data base that would be integrated with GIS so that we can create very detailed location and time-specific maps demonstrating where? what? how much? when? pesticides are applied for an assortment of reasons including the kinds of research you were talking about but also with regard to dealing with medical health, watershed pollution, groundwater pollution, integrating it with soil GIS and watershed GIS maps and health cluster maps in those kinds of things and if you see value in this kind of a database being created and it would be the first in the midwest of its kind. I would simply request that people in the scientific community, like yourself, also give us a plug over there. We are deadlocked in the state budget right now. It's a political issue. It's become a political football although it isn't bi-partisan -- a supported proposal, we are trying to set the precedent in the US for creating a database like this so we can provide good, reliable, scientifically-valid data to people like yourself.

A: (M. Cohen)

I would definitely support you. I'll tell you, California has one of the only databases that exists right now in the United States. It is surprising. We're talking about hundreds of different pesticides and herbicides. We're not talking about something bad like DDT or dioxin which is emitted in a few pounds per year. We're talking about chemicals that are sprayed routinely; put routinely in the environment at 20, 30, 40 thousand tons ...

Q: (Tom Dawson)

It's not just agriculture, it is also in the urban and suburban environment.

A: (M. Cohen)

and yet you cannot get information on the emissions and the usage, etc. You cannot get information, so, yes, I would absolutely encourage you. By the way, Keri, do you know if "was there a study done at the University of Iowa State on atrazine? Was there a big study done. A sort of mass balance approach of atrazine?" I think, a couple of years ago, one of the Iowa colleges, somebody did a big study looking at a mass balance for atrazine.

(Keri Hornbuckle)

There was one done on the Minnesota River estimated that about 1% of what was applied was found in the River.

Q: (Tom Dawson)

Well, it occurs to me that, if you are finding atrazine at levels of one part per million in the Mississippi River - that's a heck of a lot of atrazine. [Cohen - Yes, yes] Somebody aught to calculate that figure so we could use it to try and get the research dollars and get the regulatory action that's needed.

A: (M. Cohen)

We're finding, in New York, which is far from the heavy atrazine usage, we're getting more atrazine in the rain in New York; it violates the drinking water standard in the rain in New York.

Q:

Can you give us some vision of the potential for this tool to be used to quantify the contributions of various sources to the Arctic problem? You didn't talk about the revolatilization and this model just doesn't seem to address the grasshopper affect but is there a potential -- is the computer big enough to eventually do that?

A: (M. Cohen)

Absolutely. We can do the grasshopper affect. In fact, we're going to be doing mercury in the coming year and we're going to have to include the grasshopper effect for mercury. It is a much harder problem though. Luckily, the computers are getting bigger and bigger so. . . We could do something like this for the Arctic, although what you would want to have for the Arctic would be an emissions inventory for Russia, Europe, Asia, maybe all of the Northern Hemisphere, which is even more complicated. There are people doing this in the Arctic. Janos Putakeywitz is a scientist for Environment Canada who is doing work on the hexachlorocyclohexane, lindane (which is a pesticide); looking at sampling the particulates at Alert and then comparing a worldwide inventory of hexachlorocyclohexane with measurements there. He is doing pretty well on that. The Arctic is certainly the next place to look at.

Q:

(inaudible) . . .followup on the last one - we are now learning that various lakes are littered with PCBs and so you are talking about a large surface, low-dose emitter. Can that be modeled?

A: (M. Cohen)

PCBs are one of the hardest things, I think, these days to try to model because they are everywhere. We can't do an emissions inventory for PCBs in this country or in Canada today. The EPA was asked to do one by Congress and they literally couldn't because most of the PCBs are being emitted from past spills, contaminated landfills, things like current spills of transformers and things like that. You don't have a lot of quantifiable point sources or even area sources for PCBs. PCBs are very difficult. Nobody is even thinking that they will be able to model that. You could model for the Great Lakes if you knew the concentration and their doing that, I think, in the Lake Michigan Mass Balance Study. Right, Keri?

Q:

The issue is very important. According to the Agreement, the virtual elimination part of the Great Lakes Water Quality Agreement, and the binational toxic strategy. Environmentalists brought up the issue of sediment remediation, they're saying well, this is not the issue. We are now looking at targets for emission reductions and we argued, well, toxic sediment is a source and should virtually emit to, of course, move the agenda of remediating contaminated sites. It will be very, very useful for us to really start modeling these emissions. If we could demonstrate, to the regulators, that these ....in-place toxic sediment sites are important sources of emissions, then maybe get faster action for sediment cleanup.

A: (M. Cohen)

One place I would urge you to start looking at would be the work that has been done in the Great Lakes so far on the PCBs. You'll definitely get estimates of the amounts that are coming out. Now, whether you can find an estimate anywhere of the amount that's coming out from other places so you can compare the ultimate amount, I don't know. I don't know how much PCB is being emitted. The Great Lakes are a big reservoir source of PCBs and as we've turned down the air emissions of PCBs and now the stuff is coming back up they're going to be emitting PCBs for quite a long time.

Q: (Angela Bandemehr)

I just have a comment on that last question. My name is Angela Bandemehr. I am the Atmospheric Program Manager for EPA GLNPO program and I manage the atmospheric side of the Lake Michigan Mass Balance and ...while the data haven't been published yet, but I think what we're finding is different from what was found before. PCBs seem to be on an equilibrium with the water or actually depositing into the lake. We're not seeing that volatilization from the lake, largely because of water concentrations have decreased and what's happening is the air concentration seems to be driving more concentrations in certain areas, not all areas, but those are initial, unpublished findings but we find them to be very interesting . . .for the future.

A: (M. Cohen)

So maybe they're not as big of a source as we thought. Angela was saying that she's in charge of the air side of the Lake Michigan Mass Balance Study and she was saying that the data, which haven't been published yet for Lake Michigan, seem to suggest that there's not as much volatizing of PCBs, net volatization of PCBs from the lake as they thought so they may not be quite as big of a source, but these calculations are really hard and estimates are very difficult to do. There's also the papers that are published every couple of years where they take the IADN data and try to make estimates (the latest one was by Hoff a year or two ago; lead author by Hoff in Atmospheric Environment) and you'll see estimates for each of the pollutants. Some of them - it's summer time; an estimate of net volatilization and then winter time; generally a net deposition because of the temperature effects. You can get some numbers out of those that could be useful for you.

Don McKay

OK. If that's all the questions, again I would just want to take this opportunity to thank Mark for an excellent presentation (applause). I'd like to also recognize, before we break here, a few other people. We've heard one -- the excellent maps that you've seen by Larissa Mathewson before but also collecting all that emissions inventory data for the counties -- the work was done by Debra Meyer who is here and also her predecessor, Christina Caplan; that took monumental work; and also Scott Green who has done some work and I also want to acknowledge John McDonald who is the secretary for the Air Board who, due to illness, is not here. John has also done a lot of work in making sure that this work has moved as it has. We'd like to take a fifteen minute break and I don't want to lose you all so please come back because there's a lot more to have discussed and we'll have a presentation and discussion on how well the parties are doing and what still needs to be done. If we could get back at about 3:20 we'll proceed with the program and hope you will all return.