untitled

ILLINOIS POLLUTION CONTROL BOARD

June 15, 2006

IN THE MATTER OF

)

PROPOSED NEW 35 ILL ADM. CODE) R06-25

225 CONTROL OF EMISSIONS FROM) (Rulemaking - Air)

LARGE COMBUSTION SOURCES )

(MERCURY)

)

TESTIMONY OF DR. GERALD KEELER

PART I

BEFORE MARIE E. TIPSORD

HEARING OFFICER

The testimony of Dr. Gerald Keeler, a

witness called in the rulemaking proceeding before the

Illinois Pollution Control Board taken on June 15, 2006,

at 9:00 a.m., at the offices of the Environmental

Protection Agency, Springfield, Illinois, before Holly

A. Schmid, Notary Public and Certified Shorthand

Reporter, CSR No. 084-98-254587 for the State of

Illinois.

Page1

A P P E A R A N C E S

MEMBERS OF THE ILLINOIS POLLUTION CONTROL BOARD:

Ms. Marie E. Tipsord, Hearing Officer;

Dr. G. Tanner Girard, Board Member;

Ms. Andrea S. Moore, Board Member;

Mr. Thomas Johnson, Board Staff;

Mr. Tim Fox, Board Staff;

Mr. Nicholas Melas, Board Staff;

Ms. Alisa Liu, Board Staff.

COUNSEL FOR THE ILLINOIS

ENVIRONMENTAL PROTECTION AGENCY:

Mr. Charles Matoesian;

Ms. Gina Roccaforte;

Mr. John Kim;

Mr. Richard Ayres;

COUNSEL FROM SCHIFF-HARDEN

Ms. Kathleen Bassi;

Mr. Stephen Bonebrake;

Mr. Sheldon Zabel;

Mr. Jim Ingram, Dynegy, Inc.

For Dynegy Midwest Generation, Inc.

COUNSEL FROM JENNER & BLOCK

Mr. Bill Forcade;

For Kinkade;

COUNSEL FROM McGUIRE-WOODS:

Mr. James Harrington;

Mr. David Rieser;

Mr. Neal Cabral;

For Ameren;

Mr. Keith Harley,

Chicago Legal Clinic

Page2

Page3

E X H I B I T S

IDENTIFICATION

PG.

Page4

MADAM HEARING OFFICER: Good morning,

everyone. My name is Maria Tipsord, and I'm the hearing

officer in this proceeding entitled in the Matter of

Control of emissions from Large Combustion Sources,

Mercury, Docket No. RO6-25.

Again, this morning to my left is

Dr. Tanner Girard. To my right, Andrea Moore, the

presiding board members in this proceeding. At the far

right is Nicholas Melas, one of our board members. At

the far left, is Tom Johnson, also one of our board

members. Today from our technical unit we have Alisa

Liu and Tim Fox, Andrea Moore's assistant. Connie

Newman is in the audience and she is the point person

for press concerns. Erin Conley is with as today, as is

Kathy Griffen, and any procedural questions you may

address to Erin, Tim or I and, I we will try to answer

them.

We are I believe starting with

Dr. Keeler's testimony. He was sworn in when we entered

his prefiled testimony two days ago. In addition, we

have already sworn in Jim Ross, Jeffrey Sprague, Marcia

Willhite and Dr. Hornshaw, and I believe that's all that

have been sworn, and I would remind them that -- any

testimony or statements made today will be considered

Page5

sworn statements. With that, Mr. Kim.

MR. KIM: Thank you. As you just

stated, we will be starting with Dr. Keeler today. I

believe he is going to start with the questions posed to

him by Ameren Energy, and I think -- I can't remember --

one or two, or however many questions that were also

submitted by Prairie State that he will address after

that.

CROSS EXAMINATION BY MR. RIESER:

Q. Before the prefiled questions, I was

wondering if I could ask some questions just to clear up

some things from yesterday, just to confirm a couple of

things. Dr. Keeler, your expertise is as an atmospheric

scientist. Is that correct?

A. Yes, it is.

Q. And so you're an expert in the mechanics

of mercury deposition from the stack, to the ground, if

you will?

A. That, plus a whole lot more, yes.

Q. But the expertise doesn't include the

methylation of mercury in the aquatic environment. Is

that correct?

A. My formal training does not include the

biological processing of the chemicals, but I have been

Page6

working in this area for 16 years now, and have been

part of projects looking at the whole ecosystem cycling

of mercury, which includes, both, the deposition of

mercury from the atmosphere, as well as the air surface

exchange, the photoreduction (phonetic), the chemical

processing in the aquatic ecosystem, as well as the

methylation and demethylation processes. Over this time

period, I have kept up with peered-review literature. I

have worked with some of the best scientists in the

field on these things on projects including work in

Florida work, in the Great Lakes on Lake Superior and

elsewhere, so I have more than a passing understanding

of most of the processes involved with mercury and the

environment.

Q. But the papers you have prepared yourself

or that you participated in have dealt, almost entirely,

with atmospheric deposition. Isn't that correct?

A. No. That's not correct.

Q. What's an example of one that's not? I

have a couple papers where we actually made measurements

of methylmercury, a paper that we did with the

University of Wisconsin looking at methylmercury

deposition and sources of methylmercury. We have a

couple of other papers that look at vegetation uptake,

Page7

as far as exchange of mercury with enforced ecosystems.

We have papers looking at volatilization of mercury from

large bodies of water, such as the Great Lakes, Lake

Michigan, in particular. I have done work also in Lake

Champlane in Vermont, and in my vitae, is I believe --

John, is my vitae part of the record or no?

MR. KIM: Yes, I believe it is.

DR. KEELER: So if you look through

there, you will see that there are papers -- I can't

tell you how many, but there are papers where we have

actually published study results that deal with more

than just atmospheric deposition.

MR. RIESER CONTINUES:

Q. With respect to your testimony yesterday,

with regard to the study in Florida, is it correct that

your direct involvement in the Florida activities was

with respect to modeling the deposition of mercury?

A. Could you be more specific, like my --

Q. Let me be more specific. The paper that

you prepared, which was included as Appendix 1 of the --

what I understand to be the Florida study, which was the

document that was introduced into evidence yesterday,

dealt entirely with mercury deposition modeling issues?

A. That's correct. In terms of the report

Page8

writing responsibility, that was my sole role.

Q. The modeling of the methylation life cycle

within the water body, the EMCM model I think it was

called?

A. Yes.

Q. Florida study? That was done Tetricheck

(phonetic) check. Is that correct?

A. That sounds correct. Reid Harrison, Curt

Pullman were the people -- and they have switched

companies and alliances and other things, so I don't

know exactly if -- but I think Tetricheck was the

company they were working for at the time.

Q. So would it be correct that the analysis

in the methylation and the biologic uptake in the

Florida work was done by Tetricheck?

A. Yes, and Florida DDE (phonetic) has some

involvement, as well, but yes, that's correct.

Q. Do you know whether the EMCM model was

designed for the Everglades?

A. You know, the EMCM model was not designed

for the Everglades. One of the things -- and I'm glad

you brought this up -- is that models can be designed

for a specific water body or air shed or specific

application. The mercury cycling model was developed,

Page9

initially, as I understand it, for sea fish lakes in

Wisconsin, specific type of lake, and the model, though,

has the physics and chemistry of mercury and

interactions with BIODA (phonetic) that allow it to

simulate what happens in the real environment, and it

takes into account the processes that govern the

behavior of mercury. As such, in the model, it can be

adapted to another body of water so an extensive amount

of work was done by Tetricheck taking the extensive data

that was put together by the U.S. EPA, Region 4, State

of Florida, the South Florida Water Management District

to change the model, so that it would be applicable to

south Florida and the Florida Everglades, specifically.

Q. So the model at that particular model that

was used in the Florida study was, if not originally

designed, then, certainly, modified to apply solely to

the Everglades environment. Is that correct, to your

knowledge?

A. I think the questions would probably be

asked to Tetricheck, if you wanted to get to what the

purpose of their model was. The word "solely" is

probably not correct.

Q. Is it also true that there was a

conclusion of the Florida report that there were no coal

Page10

sources, coal-fired utility sources, that were

identified in the south Florida area?

A. Again, are you referring to a specific

statement that it says in the report?

Q. Yes, I am.

A. If you could point that out --

Q. On the bottom of page 76.

MADAM HEARING OFFICER: For the

record, that's Exhibit 20.

MR. RIESER CONTINUES:

Q. Thank you. I will point you to the last

sentence, "Although coal is the largest source (45

percent) 65 milligrams per year of mercury emissions of

the U.S. (144 milligrams per year) no coal combustion

occurs in South Florida and only oil and aquatic related

emissions occur."

A. Yes, I see that.

Q. Do you disagree with that?

A. In the sense of utility coal combustion,

that's correct?

Q. In any other sense is that incorrect?

A. I believe that they do use coal as a fuel

source in some of the cement dealings (phonetic).

Q. Why don't we turn to the questions that I

Page11

submitted, already.

MR. BONEBRAKE CONTINUES: I also had a

follow-up from yesterday, and I believe an open issue I

think that was left open from yesterday. Dr. Keeler, we

were talking a little bit about yesterday about Table 12

on the Florida Report, and whether -- which of the data

points on Table 12 were within the area that were

affected by the reduction of emissions in South Florida.

Do you recall that conversation?

A. Yes, I do, and I made an attempt to

contact Dr. Atchison at the State of Florida to get a

copy of a map or something that would help me show where

the sites were, and I did not receive anything from him

overnight, so I'm unable to provide any more information

to you on that.

Q. There was one statement that I thought was

relevant, very relevant to this question in the report,

and I wonder if we could direct your attention to that,

and I will ask you a follow-up question. It's on page

81 of the Florida report.

A. Okay.

Q. It's the bottom paragraph, and it's the

sentence that starts, "The three sites" -- it's about

two-thirds of the way down.

Page12

MR. KIM: Last paragraph.

DR. KEELER: "Three sites," okay.

MR. BONEBRAKE CONTINUES:

Q. It reads, "The three sites in water

conservation 3-A near Site 3-A dash 15 (located near the

so-called hot spot of high fish tissue concentrations in

WCA hyphen 3-A) also showed some cohorts with

significant declines although nearly as many site-cohort

combinations also showed no change." Do you see that,

Dr. Keeler?

A. I do, yes.

Q. Site 3A-15 was that the particular

modeling point that you were referencing yesterday?

A. It is, yes.

Q. And does this suggest to you that there

are, at least, three of the sites reflected on Table 12

that were in the immediate vicinity of that model site?

A. Based on what it says here, yes, that's

correct.

Q. So those three additional sites would have

been subject to the same kind of emission reductions

that affected Site 3A dash 15?

A. It says they are near Site 3A-15, so that

would be correct.

Page13

Q. Thank you.

MADAM HEARING OFFICER: Then I think

we are ready to begin with Ameren Question No. 1.

MR. KIM: Actually, I think if we

could do Dynegy Question No. 1, and before you begin, I

have some additional copies of what was provided in the

documents to the TSD, but these are color, and I believe

the originals were not. Actually, the Board's copies

were in color, so nobody else has those. I can give you

those now as an exhibit and then we'll hand these out.

MADAM HEARING OFFICER: Mr. Kim, just

for clarification, these are attachments to the TSD?

MR. KIM: Correct.

MADAM HEARING OFFICER: This is

Exhibit B to the TSD.

MR. KIM: Yes. It's just that the

copies provided to the Board were in color, but other

people may not have been able to access them in color.

MADAM HEARING OFFICER: Just for

purposes of the report, even though it's already part of

the record, I think I will go ahead and mark this as

Exhibit 25, if there's no objection. All right. We'll

mark this as Exhibit 25.

(Exhibit No. 25 was admitted.)

Page14

MS. BASSI: I'm sorry. Is this

Exhibit B to the TSD?

MADAM HEARING OFFICER: That's my

understanding.

MR. KIM: I believe it is, yes.

MADAM HEARING OFFICER: Just to clear

up, we are going with Dynegy first, not Ameren?

MR. KIM: Correct. I misspoke.

DR. KEELER: Question 1: "Mr. Keeler

states in his testimony that `Illinois coal-fired power

plants are the largest source of man-made mercury

emissions in the State.'" A: How large are these

emissions compared to natural mercury emission? B: How

large are these emissions compared to the total amount

of mercury emitted in Illinois? C: How large are these

emissions compared to global missions from mercury

emissions from all sources?" Taking the data that was

in the TSD, Illinois coal-fired utilities emit about 3

tons of mercury. To my knowledge, I assume that you are

asking about natural mercury emissions in the state of

Illinois. To my knowledge, there is no good emissions

inventory number for the state of Illinois. However, in

my best judgment, the natural emissions mercury in the

state of Illinois would not be very large. We've done

Page15

extensive measurements of mercury emissions from natural

sources, and Illinois soils, according to the U.S. GS,

do not have enriched mercury soils. Therefore, the

probability of natural mercury emissions from natural

soils in the state would not be very large. Natural

mercury emissions are thought to be about a third of the

emissions on a global basis with anthropogenic emissions

being a third in re-emission from previously deposited

mercury coming back off the surface being the other

third, so the 3 tons compared to the global amount of

mercury that's emitted is a relatively small amount of

the total. It's less than a percent. What's important,

though -- and we'll get to this a little bit later on --

is really not the total amount of emissions, but the

form of the mercury emissions, and by that, I mean the

chemical form that's emitted from these sources, and in

proximity to the sources to the ecosystem, and we'll get

into this a little bit later on, so the magnitude is not

really the most important thing to consider here when we

think about the impacts of mercury sources on the

environment.

Question no 2: "What country has the

largest mercury emissions from coal burning?" Over the

past few years, there's been a lot of work looking at

Page16

mercury emissions from all the countries in the world.

At this time, due to the booming economy in China, China

is thought to have the largest mercury emissions. I

happened to have been in Beijing last fall at a Mercury

Emissions from Coal-fired Utilities Workshop, and it was

amazing how much mercury is emitted into China in the

nation of China from coal combustion, and their

projections appear to be that it's going to increase

over time, so it is a significant amount.

"Is mercury a global problem."

MADAM HEARING OFFICER: Question No.

DR. KEELER: Question No. 3. Mercury

is certainly a global problem. I think, over the last

20 to 30 years, we recognize that mercury contamination

is an issue everywhere. Of the 50 states, Wyoming I

believe is the only one that doesn't have fish --

warning fish consumption advisories and I think that's

still correct. I believe it's because they haven't

tested any fish, so this is a problem worldwide, and it

really is an issue that everyone has to take into

account and is concerned about. The ubiquitous nature

of mercury in the environment is what makes it that

problem.

Page17

MS. BASSI CONTINUES:

Q. When you say you think Wyoming doesn't

have any fish advisories because they haven't tested any

fish, is that based on some examination of some data?

A. I'm sorry. You linked two of my

statements together that I didn't actually say. I said

that Wyoming doesn't have any fish warnings, and then I

said I don't believe that they have tested any fish. If

you were to look at the EPA website for fish

contamination, Wyoming is the only white state. I was

told that maybe it was because Vice-President Cheney ate

all the fish too quickly, but I'm not sure if that's

correct or not. That's hearsay, but in fact, when I

inquired into that, I was told that they don't have a

fish testing program.

Q. Who told you that?

A. An EPA person at a meeting because I

presented the table at a meeting that I was at. I

presented it at a long-range transport conference a few

years back, and it's pretty glaring when you show a map

with all the states, except for Wyoming.

MADAM HEARING OFFICER: For

clarification, when you say "EPA" do you mean U.S. EPA?

DR. KEELER: U.S. EPA, yes. To my

Page18

knowledge, based on the EPA, and that's something that's

available on the U.S. EPA website on the mercury site

that Wyoming was the only one.

DR. KEELER: Let's see. Question No.

4: "Does all the mercury emitted by Illinois coal-fired

power plants end up in Illinois?" One of the things

that we were taught very early on in science class is

that, to use the words "all" or "always" or "everyone"

or "solely," is usually not good practice. In science,

there's always an exception to every rule and so my

answer to that would be no. All the mercury from

Illinois power plants clearly does not end up in

Illinois, as does the emissions from any power plant

does not end up solely in the state that they are

emitted. I would say that's probably a fair statement.

MADAM HEARING OFFICER: Mr. Rieser.

MR. RIESER CONTINUES:

Q. Do you have a sense -- do you have any

gauges suggesting what percentage ends up in Illinois?

A. I do not.

Q. Do you have any knowledge, based on your

experience, as to how much ends up in Illinois?

A. I have not done any state of Illinois

specific quantification of how much mercury emissions

Page19

from Illinois power plants is deposited in the state.

The research that I performed has looked at the Great

Lakes region, and includes all of the Great Lakes

states, so I can't give you a quantitative number from

how much comes from the state of Illinois, itself.

Q. Thank you.

DR. KEELER: Question No. 5: "Does

some of the mercury in Illinois water bodies come from

outside of Illinois?" And A: "How much?" Yes. Some

of the mercury in Illinois water bodies clearly comes

from outside the state. Again, I have not done, or I

have not performed any state-of-Illinois specific

research. I have looked at the Great Lakes states,

which, obviously, includes Illinois. In our analysis of

mercury deposition in the Great Lakes, we assess that

40-some percent -- I think it was 43 percent -- of the

mercury deposition that fell within the Great Lakes came

from within the Great Lakes basin states, and the

remaining came from outside, so there's a significant

fraction of mercury that comes from outside of the Great

Lakes basin, itself. Now, that's the water basin, so

that's a fairly tight area, so in fact, most of Illinois

would be included in the out-of-the-basin estimates, so

our focus was really looking at mercury deposition into

Page20

the Great Lakes, themselves, those actually large bodies

of water, but that would include, say, the state of

Michigan, and any other parts of the states that are

actually in the basin, so mercury definitely is a local

and a regional and a global transport issue.

MR. RIESER CONTINUES: This is just a

clarification. I think you said that most of Illinois

is not in the Great Lakes basin?

A. The water basin. That's correct.

MS. BASSI CONTINUES:

Q. My question was do you have, or can you

give us an idea of how much of Illinois is not within

the Great Lakes water basin? Is there a line? Route

80?

A. I think it's probably --

MADAM HEARING OFFICER: Yes, Ms.

Willhite, you are sworn in, and are still sworn in, so

go ahead.

MS. WILLHITE: Sure. When you

consider what's in the basin and what's not in the

basin, it's really what actually drains to Lake

Michigan, and there's a very small sliver just right

around the edge of Lake Michigan that actually drains to

Lake Michigan. That's what's considered in the basin,

Page21

so I would probably say, like, 99 percent of Illinois is

not considered within the Great Lakes basin.

DR. KEELER: Question No. 6: "In

Dr. Keeler's testimony, he mentioned that "source

contributions from . . . motor vehicle emissions sources

were important in Detroit." Would it be reasonable to

suppose they would similarly be important in Illinois?"

We've done an extensive amount of work looking at runoff

of atmospherically deposited mercury in the city of

Detroit to try to understand the source of mercury that

makes its way to the waste water treatment plant, and as

part of that study, we actually made runoff measurements

on highways and streets in the city of Detroit, and from

areas which do not have traffic, and from that work, we

were able to hypothesize that motor vehicles actually

were a source of mercury. After about a decades's worth

of investigation and hard work, we've actually been able

to start to quantify that motor vehicles actually do

contribute to mercury problems, and it's primarily

through the fuels and the oil that are consumed in the

motor vehicles and that highway surfaces actually have

more mercury in the runoff than non-traffic used

streets, so through this, we did publish a couple of

papers, and I have made several presentations showing

Page22

that, in fact, motor vehicles do contribute to surface

runoff of mercury, and it's something that we're just

finishing a quantitative project looking at that to be

able to define emissions factors, and to be able to

better define the total contribution of mercury. At

this point, the emissions estimates suggest that it's

going to be some amount less than 10 percent compared to

the total agriculture inputs, but that number will be

better defined as we get the final emissions estimates.

MADAM HEARING OFFICER: Mr. Melas.

MR. MELAS: Just a quick clarification.

The emissions from the automobiles, the mercury is the

same qualitative type -- you mentioned that certain

emissions vary in chemical nature. Would the emissions

from the automobile type of mercury have the same

chemical nature as the kind from coal-fired generators?

DR. KEELER: Thank you for asking that

question. That's actually the most important question,

and I should have said that in my statement. Actually,

the forms of mercury -- we tried to look at the

speciation of mercury, so we made measurements of the

elemental form, reactive gaseous form, which is the form

that we think is important, in terms of going into the

ecosystem, and then being the potential form of mercury

Page23

that methylates in the ecosystem, and so elemental,

reactive, and particulate forms. We see a small amount

of particulate coming out of the motor vehicles. We see

most of the mercury coming out as elemental, and then we

do see a small percentage as reactive, as well. So

that's an excellent question, and really quite important

in terms of its impact on the environment.

MS. BASSI CONTINUES:

Q. Dr. Keeler, are you -- do you know where

the most traffic in Illinois would be located?

A. Do you know where the most traffic in

Illinois would be locate? I have not looked at the

traffic statistics for the state of Illinois. We have

done mercury work in the city of Chicago, and I would

hypothesize that that would probably be the highest,

having been stuck in traffic in Chicago myself.

Q. Have you driven in Springfield's rush

minute?

A. No. I have missed that, but perhaps

tomorrow morning.

MR. RIESER CONTINUES: Statistically

significant rush hours, aren't we?

MS. BASSI: Yeah. Thank you.

MR. RIESER CONTINUES:

Page24

Q. In the answer to the question, you

mentioned the studies you have done on runoff, but you

also did a study that you discuss in your report

regarding Detroit in which you identify, and that was,

as I recall, an air sampling study?

A. I wasn't clear on exactly which study you

were referring to in that, just so I'm 100 percent

clear. Is that the most recent paper we did?

Q. Correct. Lynam-Keeler, 2005, "Source

Receptor Relationships for Atmospheric Mercury in Urban

Detroit Michigan."

A. Yes.

Q. And in that, as I understand that paper,

you identify -- you were looking at atmospheric mercury

and doing the statistics regarding source receptor

relationships and I think you identify emissions,

automobile emissions as one of the sources of the

atmospheric mercury that you identified. Is that

correct?

A. Yes, that is correct.

Q. I think, also, in this study in Chicago,

"Atmospheric Mercury in the Lake Michigan Basin

Influence of the Chicago/Gary Urban Area," and that's a

study by Keeler and Vette?

Page25

A. Yes, Vette.

Q. One of your sampling locations was at IIT?

A. Yes, downtown.

Q. Illinois Institute of Technology, and that

would be considered a high traffic area?

A. Yes.

Q. Adjacent to the Dan Ryan Expressway,

especially now, and that also identified automobile

emissions as a source of atmospheric -- is that correct?

A. Yes.

MADAM HEARING OFFICER: I just have

one question. You referred to two different articles

and asking questions, specifically, about them. Could I

ask you to submit those for the record.

MR. RIESER: I can submit one of them

right now and I will submit the other one this

afternoon. And these were studies, specifically,

referenced in Dr. Keeler's report, which is Appendix B

to the TSD.

MADAM HEARING OFFICER: The one that's

been handed to me is "Atmospheric Mercury in the Lake

Michigan Basin Influence of the Chicago/Gary Urban

area." We'll mark that as Exhibit 26, if there's no

objection. Seeing none, it is marked as Exhibit 26.

Page26

MR. RIESER CONTINUES: I don't know if

you want to save Exhibit 27 for the Detroit study for

the next session?

MADAM HEARING OFFICER: We'll reserve

that Exhibit 27 for the Detroit study.

MR. RIESER CONTINUES:

Q. Let me ask one more question on the same

subject. I believe you mentioned that the estimated --

that the amount of mercury associated with automobile

emissions was less than 10 percent of the anthropogenic

inputs. Is that correct?

A. That's less than 10 percent. That's our

preliminary estimate based on today's estimated

anthropogenic mercury emissions, which I guess I should

make a clarifying point. One of the most difficult

things for us to try to convey to people that aren't

working in the mercury business is that emissions

inventories change constantly. Major source category

goes down and the emissions categories can change, so

when we're talking about emissions, I will try to be

careful about what year I'm basing my statements on

because, if you look at the `95-`96 database when

municipal waste and medical waste incinerators were

still emitting, the percentages of each source category

Page27

to the total changes. Now municipal waste and medical

waste have been controlled, and so now coal-fired

utilities look like a larger percentage of the total

just because major source category has gone down, so the

fractions change, so I'm referring to the -- making a

relative statement to the `99 or 2000 emissions

inventory, so 10 percent of the emissions from

anthropogenic sources we would suggest is coming from

motor vehicles.

Q. That 10 percent, just to finish that

particular question, that 10 percent is of the 33

percent of anthropogenic sources that you talked about

earlier, the one-third?

A. That's correct.

Q. And with respect to the change in mercury

emissions over time, and I will direct your attention to

figure 19, on page 76 of Exhibit 20, which is a graph

regarding annual mercury emissions in South Florida in

1980 to 2000.

A. Could you tell me what page that is?

Q. 76 is what I have.

A. Figure 19?

Q. So figure 19, page 76 of Exhibit 20 shows

significant reductions in emissions, annual mercury

Page28

emissions, in South Florida merely between 1991 and

2000, correct?

A. Yes. 1991 on this graph in figure 19

shows an emissions estimate of about 3,000 kilograms per

year.

Q. It reduces to less than 200 kilograms per

year by the year 2000?

A. About 200, yes.

Q. This reflects your testimony that the

control of municipal waste incinerators and medical

waste incinerators -- and municipal waste combusters in

South Florida over that same time period, correct?

A. Yes. That's what the graph says, yes.

Q. Would you -- is it your understanding that

there are similar reductions nationwide of your mercury

emissions as a result of those same controls being

imposed state by state?

A. That's correct.

Q. And is it of similar magnitude as that

reflected here?

A. The percentage reduction was required by

the EPA rule on incinerators, and I would say that's

probably similar to this.

Q. Thank you.

Page29

DR. KEELER: Question No. 7: I did

not finish the last part of the question. "Would it be

reasonable to suppose that they would be similarly

important in Illinois?" That's the end of Question 6.

That's referring, again, to motor vehicle contributions.

Because, as one of the questions alluded to, because of

the location of major urban areas in the Chicago region

on Lake Michigan, I suspect that the Chicago area, motor

vehicle traffic and mercury emissions would have an

impact on Lake Michigan and in the downwind areas

because of the fact that it's located kind of in the

northern quarter of the state. I would say that motor

vehicle emissions in the southern part of the state

probably would be less of an importance just because of

meteorological conditions. The wind just doesn't blow

as often from the north as it does from the southwest.

MR. RIESER CONTINUES:

Q. When you say "impact on Lake Michigan," is

that impact in terms of the volume of deposition on the

lake surface?

A. Do you mean the mass of deposition?

Q. Yes. Thank you.

A. Yes.

Q. And do you have any understanding as to

Page30

whether the deposition in Lake Michigan results in high

levels of methylmercury in fish within lake Michigan?

A. I'm not sure if this was covered or not.

The Great Lakes, themselves, most of the Great Lakes,

themselves, have only limited mercury fish consumption

advisories. There are only a few places, Lake Superior

and I believe elsewhere. Lake Michigan, the water body,

itself because, it is not anoxic. Methylation does not

occur in the water body, so that methylation and

methylmercury levels in the lake, itself, are not

significant.

MR. BONEBRAKE CONTINUES:

Q. I believe you mentioned that -- if I

understood your answer correctly, that winds from the

southwest were predominant. Was that across Illinois,

Dr. Keeler?

A. I think it's across the entire Great

Lakes. That's a statement. If you would like more

specific -- are you asking me for a specific location

what the predominant winds would be?

Q. I'm asking, generally, across the state.

If I understood correctly, you were saying, generally,

in Illinois the predominant wind direction is from the

southwest.

Page31

A. South, southwest. Of course, that varies

as a function of season of year. You know, we certainly

get more northwest and north, northerly winds during the

colder months.

MS. BASSI CONTINUES:

Q. In modeling that we have seen performed in

the context of nitrogen oxide, NOx, we saw that motor

vehicle emissions were not transported long distances.

Would that same principle apply to emissions of mercury

from motor vehicles or in whatever form the mercury

would come out of motor vehicles.

A. I think that because the emissions are

surface based, tail pipe emissions, as we were saying

the NOx comes out of the tailpipe, which is not very

high off the ground, I think a significant percentage of

the mercury would be more locally disbursed and

deposited, although we do see evidence of the gas phase

mercury being transported on larger scales.

Q. Thank you.

CROSS EXAMINATION BY MR. HARLEY:

Q. I'm an attorney here on behalf of the

Illinois Public Interest Research Group and Environment

in Illinois. Good morning. You had just stated that

methylmercury levels in Lake Michigan are not

Page32

significant. Were you speaking about the waters of Lake

Michigan, the sediments of Lake Michigan or both?

A. I was referring to just the water.

Q. Would you care to comment on mercury

levels in the sediments in Lake Michigan?

A. I, personally, have had very limited

experience looking at methylmercury concentration in the

sediments. I know that Ron Rossman, and the people from

the University of Wisconsin have done some sediment

sampling, and I know that -- I think you can get that

from Lake Michigan Mass Balance Study website, but there

is a significant amount of mercury in the sediments. I

don't recall how much methylmercury is, quantitatively,

how much methylmercury is there, but there is mercury in

the sediments, and it does vary, according to nearness

to shore, and based on the input of the tributaries,

various tributaries into the body of water, so there is

significant amounts of mercury in the sediments. My

comment was only talking about the water, itself.

Q. Thank you for clarifying that. One other

question that I have. You indicated that motor vehicles

were a source of elemental mercury emissions. Is that

correct?

A. The motor vehicle is a source of all the

Page33

forms of mercury, but predominantly, it was elemental.

Q. Would you care to comment at this point

about the relative water solubility of elemental mercury

by comparison to mercury in its gaseous form and its

particulate form?

A. Sure. We will get into this later, but I

think it's probably a good point. As you asked earlier,

the key with mercury really is understanding the

chemical form of the mercury in the atmosphere or in the

environment. Elemental mercury is sparingly soluble,

which means if you take mercury from the air as its

floating around in this room -- there's mercury here --

and put a glass of water out, there's not going to be

much mercury that will go into the water, itself, so you

could leave your glass of water out for days, and you're

not going to get much mercury in that glass of water.

If we sprayed reactive mercury into this room and left

your glass of water out, you would see a significantly

larger fraction of that mercury actually go into the

glass of water, so if you take that analogy, and go to

large bodies of water or surfaces, reactive mercury

likes to go into droplets, so it likes to go into cloud

water. It's a very sticky substance. It likes to

deposit to the surface, so if it comes out of a stack

Page34

and interacts with any type of surface, it's going to

want to stick to it much more quickly.

Elemental mercury, depending upon the

surface type, most of it is just going to keep getting

blown away. It's not going to stick. It's not going to

go into the water, so if the mercury comes out of a

stack or comes out of a tailpipe, or whatever source, in

the elemental form, it's going to have less of an impact

on the surrounding area, unless it's coming out in

extremely high concentrations. There's evidence near

fluoroalkali (phonetic) chemical manufacturing

facilities that very high concentrations of elemental

mercury come out of those plants, and you can see high

concentrations of elemental mercury in the surrounding

areas, but in most places, the impact of the elemental

mercury emissions on the local environment isn't as

great, so in the case of incinerators and sources that

put out almost all of the mercury in the elemental form,

that has a very large local impact. I'm sorry, in the

reactive form. I misspoke. So, if the emissions come

out in the reactive form, they are going to have a very

local impact. When it comes to particulate mercury,

which was the other question, if mercury comes out in a

particulate form out of motor vehicle because it's so

Page35

close to the ground, it, too, will stay fairly close

stay in that highway zone and will get run off with

precipitation as it gets washed off. The reactive

mercury, of course, if it comes out is going to stick

really close by, and it, too, will get washed off or

volatilized, depending on the conditions, so the motor

vehicle emissions, maybe a smaller fraction, will stay

in the local vicinity because it's in elemental form

than if it was in reactive form. Did I answer your

question sufficiently?

Q. Yes. One other follow-up question to what

you have commented on so far, testified to so far. That

is, you said that the relative percentage of mercury

entering the environment from anthropogenic sources has

changed as emissions from medical waste incinerators and

municipal waste combusters have reduced. Do you know

why mercury emissions from medical waste incinerators

and municipal waste combusters have decreased so

substantially over the past few years?

A. Well, there was legislation that was

passed by the EPA and went into effect I believe in `98.

The legislation might have been passed earlier than

that, but I think it had -- I think it went in the `98

to 2000 time frame where there had to be a 90 percent

Page36

reduction I believe the number was from that industrial

sector, and so they had to reduce their emissions by 90

percent, which from our analysis in Florida, and from

work that was done in Massachusetts, and I think, as

time goes on, it would come out in other places had a

profound impact on the local deposition in those areas,

and it appears to be an impact on the ecosystem, as

well, and I think the Florida Report bears that out, and

I think now the biological monitoring that was done in

Massachusetts beared out the fact that things seem to be

improving in those areas where the reduction took place.

MR. ZABEL CONTINUES:

Q. One question in follow-up on the

incinerators, Dr. Keeler. Do you know what most of them

did to comply?

A. It depends upon the type. Most of the

medical waste incinerators, in the past, medical waste

incinerators -- there were a small number of large

incinerators, hospitals in urban areas, like in South

Florida would send their waste to one large medical

waste incinerator and these facilities are very

unsophisticated, surprisingly. I actually went to one

of the large ones in South Florida and I was surprised

that it was pretty unsophisticated, and the economics of

Page37

controlling mercury emissions from that. It seemed like

they were burning almost pure mercury, when it came down

to it, so the economics suggested that they could not

continue to operate, so they actually shut that plant

down, and what happened is a lot of smaller plants that

emit less mercury, but are able to cap their emissions

in a better way. The municipal waste incinerators had

to add various control technologies and I'm not -- I

don't have at my disposal at this moment exactly what

all the different incinerators did, but there were a

variety of control technologies that they used to

control the mercury emissions from those plants.

Q. Did the majority, in terms of volume of

the medical waste incinerators, simply shut down?

A. I think, initially, that was the response.

Many of the medical waste incinerators shut down because

people began sending their waste to other solid waste

practices, other ways of dealing with medical waste, but

the trend now -- for example, I know in South Florida

that now there are, like, a dozen medical waste

incinerators that have reemerged, but they're smaller

units, but better controlled, but they are still putting

out mercury, but it's they are meeting the regulations

in smaller units.

Page38

Q. Are they new units?

A. I'm not sure. They are new, in the sense

that they are new in terms of new sources that have been

permitted to emit, but I couldn't answer that question

whether they were existing facilities that retrofitted

or not.

MADAM HEARING OFFICER: Mr. Rieser.

MR. RIESER CONTINUES:

Q. Just to follow-up on something that

happened some time ago. In terms of the automobile

emissions, I believe your answer was primarily about

emissions in the Chicago area, but wouldn't it also be

true that the automobile emissions would also be a

potential source in the East St. Louis, Metro East area

around St. Louis.

A. Yes. I'm really I'm glad you asked that

question. When we look for evidence of motor vehicle

emissions, we really see them in the large urban areas.

When we look in a town such as Springfield, we would

have a very difficult time seeing any mercury levels

that were increased by traffic in this local vicinity.

We have a hard time seeing any in the Ann Arbor, area

which is a little bigger than Springfield, but we do see

it in the larger metropolitan areas, so of course, the

Page39

St. Louis, East St. Louis area would be another example

of where you would expected to see some influence from

traffic, just as you do in the ozone issue downwind of

that corridor.

MR. BONEBRAKE CONTINUES:

Q. In response to a question from Mr. Harley,

you mentioned that there was a relationship between

mercury levels in sediment in Lake Michigan and in

proximity to the shoreline?

A. Yes.

Q. Can you explain that relationship to us or

perhaps describe that relationship to us?

A. Describe that relationship?

Q. What is the relationship between the

proximity to shoreline and mercury concentrations in

sediment?

A. From my recollection of the presentations

that were made at some of the Lake Michigan Mass Balance

meetings, there were places where there was higher

sediment retention in the southeast corner of Lake

Michigan where sediments tend to accumulate there were

fairly high concentrations of mercury. In some of the

areas where like the Chicago Ship Channel is, and in

that area along the shore of Chicago and Gary where all

Page40

the heavy industry is located, there were higher

concentrations of mercury in the sediments.

We, actually, over the past 15 years,

we've done an extensive amount of work with vessels

going out on the Great Lakes taking measurements of the

air out over the water when the plumes came off shore,

and that's the work that's written in that paper that

you were given, and one of the things we see is very

high concentrations of particulate mercury and other

forms of mercury when the flow is off from those

industrial areas. You can actually see shiny materials

in the surface waters. We actually took samples of

microlayer, which is the surface layer of the water on

Lake Michigan, and you can see the metallic particles

floating on the water, and those samples which we then

analyzed actually had fairly high concentrations and

other metals. Those particles are very large because

you can see them, and those are, basically, the ones

that got emitted and deposited very rapidly, and those

will then go to sediment, so in areas where there's

large tributary inputs, as well as areas with high

mercury deposition from the atmosphere, which tend to be

located near the shoreline of Lake Michigan, you would

see higher particulate loads and higher sediment

Page41

concentrations. Of course, the sediment patterns are

complicated in any large lake because of the patterns of

the reed suspension (phonetic), but they still see

patterns of high deposition in a southern part of the

lake.

Q. Are the higher levels of mercury in

proximity tributaries of Lake Michigan related in any

way to waste water discharges of mercury to those

tributaries?

A. My role in the Lake Michigan Mass Balance

Study was to look at the atmospheric inputs, and the

University of Wisconsin was in charge of tributary

inputs working with United States Geological Survey, so

that was not my role, so I can't testify to what they

found. I know that atmospheric inputs were 85 percent

of the total inputs to Lake Michigan with tributary

inputs being 15, but there was a large fraction of the

tributary inputs that they couldn't account for in terms

of where the mercury was coming from, so some portion of

the tributary inputs, clearly, would come from runoff

and other sources like you mentioned.

MS. BASSI CONTINUES:

Q. I didn't get my notes completed on this.

You were talking about measuring the plumes or measuring

Page42

plumes over Lake Michigan, and I got distracted by

looking down and seeing the metals on the surface. Did

you say there were higher particulate loads in those

particular measurement forays that you were doing?

A. Yeah. They tended to have a higher

particulate fraction.

Q. And this is the HGP that you're talking

about?

A. Yes, I am.

MR. ZABEL CONTINUES:

Q. This is not on Lake Michigan, but

something you said made me think of this. You mentioned

reed suspension. Is reed suspension of mercury in a

water body -- strike that. Is reed suspension of

mercury in a water body, mercury in the sediment, a

source of accelerated methylation.

A. Again, I guess I'm a reductive type of

person. You asked a lot of things in that sentence. I

guess there's not a simple answer to asking whether

there's anything that accelerates. That suggests that

the rate increases with time. I think that any time you

reed suspend a mercury in a form that's not available

for chemical reaction up into the water column, you

would provide it to be chemically reactive. I don't

Page43

think it's technically correct that it would accelerate,

but yes, reed suspension does provide a way to get

mercury into the water column.

Q. It would make it available for methylation

where it might not otherwise be?

A. If the form of mercury was such that it

was potentially methylated, yes.

Q. I guess let me be a little blunt, and say

what my thought was while you were mentioning reed

suspension made me think of that. You talked about the

modeling in Florida before. I'm wondering if the

modeling -- my understanding is the Everglades are a

relatively shallow body of water. Is that correct?

A. I can usually stand in the Everglades up

to my neck in water, yes, so it's a fairly shallow

place, except for when you fall into the gator pits.

Q. Which they must enjoy greatly, but that's

a different thing. My concern is does the modeling

address the possibility of severe weather, hurricanes in

Florida causing reed suspension, and if so, how?

A. I don't believe anyone's addressed that,

but that's an excellent question, and I think that

severe storms play a huge role in causing havoc to the

ecosystem in South Florida. Interestingly enough, I

Page44

think they tend to clean things out. The problem with

the Florida Everglades has been that we have kind of

dammed it up, and whenever you put in canals and dam

things up, it, basically, filled in the Everglades, so

we could put condos down there, so snow birds like us

can go down and live in South Florida. The Everglades

has been confined to a smaller area so less water is

flushing through that system, so anytime you do that

type of a thing, you are actually building up and

storing up more contaminants in that ecosystem, and the

big storms, basically, flush that out, and those have

been occurring continuously for a long time. We hear a

lot more about the hurricanes, but those have been going

on for a long time, so they -- actually, they are a

positive, I would say, in terms of the mercury and other

contaminant issues in South Florida.

Q. Could cause reed suspension?

A. I'm sure they do in the flushing.

Q. Flushing the mercury into the Florida

bays?

A. Yeah, and we actually published a paper

looking at contaminant levels over time in Florida Bay,

and you can nicely see the inputs of the atmospheric

over the last 100 years in Florida Bay sediments, so

Page45

even though there are these storms that go on that

change the pattern of deposition, you can still see

those inputs, so the atmospheric signal is still very

strong. We can see lead and lead gasoline in the

Florida Bay sediments going back for 100 years.

DR. KEELER: Question No. 7: "Would

Dr. Keeler consider the Steubenville, Ohio Study site to

be representative conditions across the country? A:

Representative of conditions anywhere in Illinois, and

B: If so, where?" Just to back up, Steubenville, Ohio,

is a community that's in Eastern Ohio on the Ohio River

Valley. Around 1998, `99, the EPA director of the

National Exposure Research Laboratory, Gary Foley, asked

the question why there was no mercury monitoring being

done in the state of Ohio and there was interest by

people in the state of Ohio to have a mercury monitoring

site. The EPA issued a request for proposals for a

competitive bid process where proposals were submitted

to the Office of Research and Development in Washington

and we submitted a proposal to that call for proposals,

and in that call for proposal, they asked that we choose

a site that was within 15 kilometers of a large

coal-fired utility. Again, that wording is from my

recollection, not from the exact language, but they,

Page46

basically, wanted us to identify a location in the

country that we thought would have a high probability to

have an impact from coal-fired utilities. I've been

working in the air quality area now for over 20 years. I

started doing atmospheric chemistry and measurements in

1982, and as part of my thesis work, I did work in

Southwestern Pennsylvania looking at the sources of acid

rain, so I have been looking at sources of air quality

or aerial problems in this part of the country for more

than 20 years.

As part of my thesis work, we actually

went and made measurements in that area and actually

went and looked at local sources, went and visited, and

went to many of the power stations and zinc smelters and

glass welding and manufacturing plants an everything

that was in that vicinity.

After I got out of graduate school, I

went to the Harvard School of Public Health and

continued my work where we were doing a health study

looking at the health impacts of particulate air

pollution on people, mainly, on children and as part of

that study, there was a site in the Steubenville area,

and previous analysis had shown that this was a location

that had clear coal-fired utility impacts, so as part of

Page47

our new study to look at mercury, because of the wealth

of data that had been going on at this location for the

past -- actually, started in the 70's, we felt this was

a good location to go back to, so we went to

Steubenville, Ohio, and performed a mercury study, and

we're going to hear more about this as time goes on.

The question is to its representativeness of conditions

of across the country. Again, I'm an atmospheric

scientist. I've had formal training in meteorology.

When we talk about representativeness, we are saying,

"Could you take the information garnered at one spot and

transfer it to another spot?" And Steubenville, Ohio,

is not representative of, really, of any place, other

than Steubenville, Ohio, because of the unique nature of

its sources and the meteorological conditions that

govern the transport of pollutants to and from that

area.

However, if one wants to look at the

representativeness of an area, such as the Great Lakes,

then the Steubenville, Ohio, area is I think a fairly

good representative site of many locations along the

Ohio River Valley. It's more representative of a

Midwest industrial area than it is not representative,

so is it a perfect one-to-one correlation with everyone

Page48

in the country? Of course not. Steubenville, Ohio,

conditions are not like Florida conditions, in terms of

the weather. Otherwise, we would all go to Steubenville

in the wintertime. Having been there in the wintertime,

I can tell you you would not want to go to Steubenville

in the wintertime. You don't want to be in Steubenville

in the summer, either, but it's not representative of

most places in the United States, but it is a good

representative site of the Midwest area that is

dominated by coal burning.

"Is it representative of conditions

anywhere in Illinois?" Again, the meteorology is

probably not all that much different in terms of the

predominant winds the amounts of precipitation. I know

that, as people here in Illinois say, Illinois is

elevation challenged. Most of Michigan is the same way.

We don't have much topography. Southern Ohio has a

little bit of topography. In the upwind region of

Steubenville, Ohio, it's fairly flat, as well, so it's

not a mountainess area. There are no topographic

differences. Illinois has, of course, the Lake Michigan

shoreline. That is a little bit different, but Ohio

also has Lake Erie, so there are many more similarities

I would say than not, but to say that it's

Page49

representative of any point, that's probably not a fair

statement. I don't think anyone would try to say that

it's a one-to-one correspondence, but it does have a

good general I would say representativeness of the

midwestern region.

Question 8: "In the Steubenville

Study, was there any attempt made to relate deposition

of inorganic mercury from anthropogenic sources to

methylmercury concentrations in fish?" To continue, the

Steubenville Study, as we were asked to propose, was a

study to look at the atmospheric -- basically, it was

asked to quantify the levels of mercury in the

environment, as well as the levels of mercury in the

deposition, and to quantify the contributions from

coal-fired utilities to the levels that we were seeing

in the air and in the wet and dry deposition at that

spot. It was not part of the scope of the work for us

to link the deposition to aquatic impacts or to cycling

or to air-water exchange or any of those issues. It was

purely an atmospheric deposition study, so that was

beyond the scope of our project.

MR. RIESER CONTINUES:

Q. Yeah, I'm sorry for this. I want to go

back to the representativeness. How many -- and I had

Page50

these questions in mind, and I might as well get them

out now. How many coal-fired power plants are located

within 50 kilometers of Steubenville?

A. Of Steubenville?

Q. Yeah.

A. I know that's one of our later questions.

I don't have a quantitative number, but there's a large

number probably within 50 kilometers. There might be

15.

Q. Do you know what the combined aquatic --

if you're organized to answer this later, we can answer

it later.

A. I have a table that lists -- I actually

have a utility provided table that has all the megawatt

-- I mean, it's actually the utilities that put that

data together as part of the information requested by

the EPA, and so I can add those up if you would like me

to, but it's a fairly large number.

Q. So would it be correct to say that -- I

believe you actually said that Steubenville was selected

because it was extended in a place where you would see

impacts from adjacent coal-fired power plants.

A. Yes. EPA, specifically, asked us to

identify a location where we could utilize tools, and

Page51

again, I will talk about this a little later in some of

the other questions, but we have developed tools over

the last 20 years that allow us to make measurements in

the environment and to work backwards to identify the

sources of the pollution that we measure, and these

tools -- some people call it environmental forensics.

It's kind of like you find clues to what the source of

the pollution is, and then you go and work backwards to

identify the source. EPA asked us, specifically, to

identify places where we could test those tools and

investigate the feasibility and accuracy of using these

tools in this type of application, specifically, to look

at coal-fired utilities, and this was really an obvious

outcome of the 1998 Mercury Report to Congress, which

identified municipal waste and medical waste

incinerators as the largest source of mercury to the

environment, with coal-fired utilities being the second

largest.

Having successfully reduced the

emissions from the first category, which is the

incinerator category, and seeing -- again, to see some

of benefit, they started asking the questions "Well,

what is the impact of coal-fired utilities?" because

there really was no data out there, and they asked the

Page52

question why, and so they, specifically, told us to go

to an area where we had a fairly good idea that we would

be impacted by coal-burning utilities, so yes, that was

the purpose for us going there.

Q. So for the purpose of developing that

analogy, it would be good to go to a place where you

knew you were going to be able to see the impacts that

you were going to test?

A. Yes, that's correct. We would not go to

South Florida to look for coal-fired utility impacts

because there aren't any, so --

CROSS EXAMINATION BY MR. AYERS:

Q. Following up on that question, isn't it

useful to go to a place where you have the phenomenon,

if you are going to do the science, to try to understand

the phenomenon?

A. Yes. That's exactly right. If you're

asked to study something, in particular. In that case,

we were. It was a very specific request and so we used

our, again, 15 years worth of experience at a location

to tell us that this was the right place to go, and we

actually identified a couple of other locations, as

well, not too far from here.

We could have gone from Steubenville

Page53

over into Pennsylvania. We could have gone down to

Athens, Ohio. There's a lot of places in the Ohio River

Valley corridor that would have met EPA's requirements,

in terms of being within 15 kilometers.

MADAM HEARING OFFICER: Could you

identify yourself for the record.

MR. AYERS: I'm Richard Ayres, Ayres

Law Group in Washington D.C.

MR. AYERS CONTINUES:

Q. The second question I guess isn't it true,

then, that what you learned in the study, like the one

at Steubenville may be transferable in terms of

understanding the phenomenon, even if Steubenville is

not exactly like another place?

A. Yeah. That's a great question. The

methods that have been developed -- I should say right

up front that what we call these methods are receptor

methods. In other words, a receptor is the place that

receives the pollution, and so we'll get into this a

little later in other questions, but instead of using a

model, which takes the emission from a stack, and then

models where those pollution emissions go and then fall

down to the earth, the receptor methods make

measurements at the place, so they are real

Page54

measurements, and then we work backwards to figure out

where they came from. And these methods have been under

development for the past 25 years at the U.S. EPA, and

independent of the work that we've been doing, EPA has

been working with some of the best scientists in the

field to develop a series of models, and to go through

and extensive testing and evaluation of these models to

look at their use and regulatory applications and

there's a whole literature that is out there with tens

of papers, 10, 20, maybe up to 100 papers where other

people have used these methods to do exactly what we

were doing, which is proportion the amount of pollution

that came from a particular source type, to how much was

received at a specific location, so that work has been

going on for a long time and parallel.

Our work was not to work on that, to

show that these methods work. EPA has an exhaustive

report of that. These methods have been tested. EPA

has developed these methods. They are publicly

available. They have had workshops where they have

looked at uncertainties, and both, in the methods and

calculations and all that is very well documented. It's

not something we did as part of our study. We just took

and used those developed tools that EPA had, and used

Page55

those to understand what we found in Steubenville, so

choosing a place like Steubenville was just a way so

that we could look, specifically, at coal-fired

utilities to see what are the complexities and

difficulties in looking at this type of impact because

it is not simple. With everything in science, we start

at the beginning, and we work up, and as we learn, we

improve what we're doing, and so when this study was

proposed, they said, "Okay, go where you think you have

the highest probability of findings an answer," so we

went there, got a good signal, and we've been successful

in working backwards to define the contributions from

the sources in that area.

Q. One final question, you did this work near

the location of a lot of power plants. If, as some have

said, that most of the emissions of mercury went up, and

far away, rather than being deposited in relatively

local areas, would you have seen the signature of the

power plants that were located in the relatively local

area of Steubenville?

A. Again, I guess this answer is a question

that is posed to me later on, but again, getting back to

the form of the mercury, there are a number of issues or

controversies out there, in terms of mercury transport

Page56

and deposition. The form of mercury really determines

the distance that mercury is going to get transported,

and so understanding the form of mercury that comes out

of coal-fired utilities is of great importance. The

Electric Power Research Institute working with U.S. EPA

and others, have done an excellent job of quantifying

the amount of mercury that is consumed by this industry

and submitted by this industry, and we do have some

information on the form of mercury that comes out of the

coal-fired utilities. The data -- again, this is data

that is in -- we gave reports suggesting that it's

somewhere around 67 percent, plus or minus 15 or 20

percent of the mercury that is submitted from coal-fired

utilities comes out in that reactive gaseous form,

although on any given day, at any given plant, that

number can vary depending on the blend of coal and the

type of control technologies and so forth.

In Steubenville, Ohio, there are

enlarge number of coal-fired utilities around the plant,

and the mercury that comes out of these stacks, based

on, again, on the EPA's database, would suggest there's

a significant fraction of the mercury that comes out in

the reactive form, and this mercury goes into water

that's in the atmosphere and gets precipitated out. We

Page57

also believe that this mercury is deposited through dry

deposition to the earth's surface in a fairly local way.

We do see a clear indication that coal combustions

dominates the deposition of mercury to the Steubenville,

Ohio, area. This is commensurate with our understanding

of the form of mercury that comes out of the power

plants.

MR. RIESER CONTINUES: Well, that was

the whole nut, wasn't it? Obviously, there's a lot of

questions on that. I don't know where you want to take

them just to keep the record sane in the --

MADAM HEARING OFFICER: Keeping the

record sane -- I've given up hope of that.

MR. RIESER CONTINUES:

Q. Fair enough. On the context we have

organized our questions, or we can start on it now.

A. I think, if we keep going down, I think

you will be able to have a chance to probe at what you

would like to probe at.

MADAM HEARING OFFICER: I would like

to point out that it is already 20 after 10, and we are

going to be taking a break at 10:30. Why don't we try

to get through Ameren's questions before 10:30 and we

will take -- I'm sorry -- Dynegy.

Page58

MR. RIESER: That's fine. I just want

the record to reflect that not asking question here

doesn't mean we don't get questions on this later.

MADAM HEARING OFFICER: Absolutely.

MR. RIESER CONTINUES:

Q. I do want to ask a question on Mr. Ayres'

second question, which had to do with how this operates.

When you look at testing and experimenting with respect

to phenomenon and the place you expect the phenomenon to

be, isn't the next step of that to develop some method

for extrapolating that determination to other

circumstances?

A. Yeah. I think we always, again, in kind

of a reductionist way of thinking about things, you try

to identify an approach and methodology that allows you

to answer a specific question, and then the next step is

to try to broaden that specific answer to a larger

context, and we've been doing similar type of work in

the state of Michigan, as well. It hasn't just been in

the state of Ohio, so we have similar monitoring sites

at multiple locations in Michigan, which I believe, in

my testimony, I alluded to, and so we were trying to do

that. We have been working very hard, and since one of

the things that -- you feel, after awhile, that maybe

Page59

your head has gotten softened up because I've been

banging it up against the wall so long, but we've tried

very hard to get both the states and the agencies to

expand the type of work we have done to a larger

geographical area, so we could answer these questions,

and to do that, to take the science from a spot and take

it and broaden it, so we can answer these questions in a

more scientific, definitive way.

Like everything, the mercury field is

fairly new. It's young relative to the acid rain field.

We've only been at it for 15 or 20 years, and so it's

something that's evolving, but we've made major progress

because we learned so much from the acid rain research

that was done, and we've learned so much from the acid

rain research that the growth period has gone so much

quicker, but we would still like to expand, so yes.

Q. Listening to that answer, can I say that

you are working on methodologies to extrapolate from

your findings of Steubenville, but have not yet

completed that work?

A. Well, as my major advisor told me when

started in on mercury, and after a decade, I told him I

felt like I had just been on to learn the topic. He

said to me, "You're going to retire working on mercury."

Page60

I have a long time to retire, a good another 20 some

years to, work and I believe this is something that --

mercury is an interesting component. It's very

difficult to get a handle on in many ways, and so I

think, in terms of the changing local, versus regional,

versus global concerns, that's changing all the time,

and so I think it's something that's going to take a

long time. However, as a scientist, I feel like I have

a fairly good handle on things that are important for

today, and yes, extrapolating -- I wish I had 25 sites.

There's no doubt about it, but the sciences, basic

chemistry and physics and basic meteorology, and those

things are always transferable. The reason that we do

work in Florida and Michigan is not because I'm a smart

guy, and I want to go to Florida for vacation in the

wintertime. In fact, all of our mercury work has been

in middle of summer, which I tell my relatives the

opposite just, so they don't think I'm a nut, but we

work in these different environments, but the laws of

chemistry and physics have to be the same in both places

. You can't explain mercury behavior in Florida

different than you explain it in Michigan. It has to be

transferable. There has to be changes in meteorology or

temperature or other things that make the behavior of

Page61

mercury explainable, and that's what we're finding. We

are finding repeatable behavior in all the environments.

I have done an extensive amount of

work in New England. We're doing work out west, work in

the Mediterranean Sea, in the Arctic. All these

environments keep telling us that we are better

understanding the chemistry and physics of mercury, and

that body of knowledge that we have acquired over the

past 20 years really lends to credence to what we are

finding. It doesn't make it more blurry. It makes a

more better defined picture.

Q. So in order to draw conclusions from the

Steubenville work to, say, Illinois, you would actually

have to do the work you described of testing the

deposition and doing the methodology you described. Is

that correct?

A. Again, I think that's a question that gets

back to my comment earlier. When we talk about

identifying the contributions of the specific sources or

source type to the state of Illinois, in particular, you

can't go back in time and collect the rain. You can't

go back in time and collect the air that was coming

through the state, so the only way that you could do

that would be to go and collect samples in the state at

Page62

the time that you wanted to do that contribution.

However, we actually have collected

samples in Illinois. We collected them as part of my

Lake Michigan Mass Balance Study, which is samples in

Chicago, at IIT, and we have done an extensive amount of

work around the Great Lakes Basin, so we have data that

allows us to extrapolate and extend our understanding of

what's going on to areas in Illinois that I think are

very relevant and accurate.

Q. Were there any conclusions in your

testimony based on that data specific to conditions in

Illinois?

A. I believe there were.

Q. Were those conclusions the same as the

conclusions -- the same type of conclusions, i.e, the

contribution of local sources, local coal-fired sources

to local impacts?

A. Again, this is one of the things that

we -- I hate to keep saying it again, but -- the data we

collected in those studies was collected in the `94,

`95, `96 time frame. At that time frame, municipal

waste and medical waste incinerators were the dominant

source of mercury, and in the air shed and the Great

Lakes, we had a number of very large incinerators.

Page63

Plus, the Chicago area has -- the Chicago-Gary area has

a number of very large iron and steel production

facilities, chemical manufacturing refineries, a number

of other sources, so we did make conclusions regarding

the sources of the mercury, but the blend of the sources

was different at that time.

At that time, incinerators and power

plants were about the same, and those were -- I don't

know what -- 30 something percent each, in terms of the

total mercury emissions, and now the one's gone, so now

coal utilities are the largest source with incinerators

being a much smaller, so the answer we got then were

applicable to the time that we were making the

measurements. The tools that we used are, generally,

applicable to any time, but even at that time, though,

the contributions we were seeing from coal combustion

were I would say, relatively speaking, so if you take

the one-third of the mercury emissions coming from coal

combustion at that time, we were seeing signals in the

20- to 40-percent range, in terms of the contribution of

coal combustion to the levels that we were seeing in the

atmosphere and in the deposition, so it was consistent

with the emissions that were occurring at that time. If

we went today and took at look, we would see a different

Page64

picture.

Q. So is the answer to that question no.

A. The long-winded answer was given because I

think it helped to relate to the conclusions we drew

there. We saw the influence of coal combustion on the

regional scale, local-regional scale, in those studies.

The relative importance of coal combustion was different

at that time largely due to the differences in

emissions.

Q. So you don't have any recent data with

respect to Illinois or the Chicago area that would allow

you to draw specific conclusions. Is that correct?

A. I have not made any measurements in

Illinois since the 90's. However, the conclusions I

believe that we made there are still consistent. We've

still been making measurements in Michigan, which is

downwind of Illinois, and so we do still have data that

helped us understand the sources that are in Illinois,

but no, we do not have Illinois-specific measurements.

Q. Thanks.

MADAM HEARING OFFICER: Having arrived

at 10:30.

CROSS EXAMINATION BY MR. MATOESIAN:

Q. Did you say that, even in the early 90's,

Page65

you were --

MADAM HEARING OFFICER: They cannot

possibly hear you in the back of the room.

MR. MATOESIAN CONTINUES:

Q. Did you say that, in the early 90's, 20 to

40 percent of deposition you were finding was coming

from coal-fired power plants in Chicago based on your

sampling?

A. I said that, from our earlier studies, we

were seeing 20 to 40 percent contributions from coal

combustion to the ambient concentrations in deposition

that we were measuring.

Q. Has there been a significant change in the

weather patterns or meteorology for Illinois in the last

10 years?

A. Wow. This is where the legal, versus

scientific, definition comes in. What's the definition

of -- no. I would say that the simple answer is that

the last decade has been interesting; one,

meteorologically, but is not that consistent with the 30

year climatology, so no, there's not a significant

change, although the weather over the last decade has

certainly been unusual in many cases.

Q. Aside from Algora, so you would expect to

Page66

see, at least, a similar contribution from coal-fired

power plants if you were to take samples?

MADAM HEARING OFFICER: You are going

to have to speak up.

MR. MATOESIAN CONTINUES:

Q. You would expect to see, at least, similar

contributions from coal-fired power plants today as you

did then?

A. My expectation actually would be that we

would see a larger fraction of the contribution from

coal-fired utilities because the incinerator sector has

been controlled, so that would mean that it could be a

similar amount of total mercury, but it would be a

larger fraction of the contribution from coal

combustion.

Q. Thank you.

A. I don't have any evidence to suggest that

it would have declined.

MADAM HEARING OFFICER: At this point,

we are going to have to take a break. The board meeting

is at 11 o'clock. As I stated earlier, there is a

pending motion in this rulemaking. If the Board rules

on that motion in the meeting, I will have copies of the

record available when we come back at one o'clock.

Page67

(At which point, the hearing was

adjourned.)

MADAM HEARING OFFICER: Before we go

back on the record, I want to let you know, in addition

to granting the motion to amend today, the Board

declined to offer any of the direction asked for by

Ameren, Kinkade and Dynegy to the Hearing Officer.

That being said, what I would

anticipate is that next week, as we approach the end of

the week, we'll see where we're at. As I indicated in

my Hearing Officer Order setting this hearing, we will

look at the schedules. We'll see where we're at, and

see what we need to do, as far as continuing, perhaps,

cross-examination, extend prefiling deadlines, and that

sort of thing, and we'll do that as we proceed here, and

we know where we're at.

That being said, I think we will

proceed with Dr. Keeler. I believe we are on Dynegy

Question No. 9.

MR. RIESER: Eight, I believe.

MADAM HEARING OFFICER: I thought we

finished with eight.

MR. RIESER: Eight related to methyl

relationship --

Page68

DR. KEELER: I did answer that. That

was beyond the scope of the project we conducted.

MADAM HEARING OFFICER: Question No.

DR. KEELER: Question No. 9: "Do

different types of emission sources for mercury have

different mercury deposition patterns? Yes. We would

have different deposition patterns, and again, I'm sorry

to keep repeating myself, but the deposition pattern is

going to be a function of the type of mercury emitting

from the stack, so from the type of stored source it is,

and clearly, the characteristics of the stack, the

height of the stack, the velocity at which the emission

comes out and so forth, so yes, emission patterns will

be different depending on the type of mercury and from

the different type of sources.

Question 10: "Do different types of

sources emit during species of mercury?" Yes. There's

quite a bit of information in the literature now showing

that different source types do emit different types of

mercury. As we have already mentioned, municipal waste

and medical waste incinerators emit a predominant amount

of mercury in the reactive or ready depositable form,

greater than 80 percent. Power plants, based on the

Page69

work that's been done by the utility industry, suggests

that it's closer to 67 or so percent, but that the

amount that's emitted from a specific source depends

largely on the exact amount of coal that's used, the

type of coal and the blending and so forth.

Motor vehicles will emit a different

blend, less reactive, more elemental. Chemical

manufacturing tends to emit more elemental mercury than

reactive, so yes, the form that's emitted varies

dramatically from one source to another. "What elements

or facts influence or" --

MR. RIESER CONTINUES:

Q. The 67 percent number, with respect to RGM

emissions from coal-fired power plants, you said that's

-- do you recall the source of that number.

A. Presbo, et al., if I recall.

Q. Presbo?

A. Presbo. He works at Frontier

Geoscientists, and I believe that's the citation that I

have from that.

Q. And then that's -- do you know if that

represents an average of all coal-fired power plants in

the country?

A. It was the ones that were tested using --

Page70

I, honestly, don't recall which technique was used, but

there were a number of plants that were tested, and that

was the average. It was 67, plus or minus 15 or 17

percent, something like that.

Q. That was my next question. Wasn't there a

range?

A. Yeah. There was, and again, my

recollection is that it's 15 or 17 percent.

Q. I think we talked -- we talked earlier

about different coals. Would that -- I think we talked

earlier that that would be affected, also, by the

different coals that were used as fuel, correct, certain

types of coals?

A. Yeah. My understanding of the state of

knowledge right now is the parameters that are most

important, which is the next question, if that's okay if

I continue answering that. The things that affect the

or influence the type of mercury that's emitted for coal

combustion, the chlorine content of the coal is very

important, not only the amount, but the nature of the

fly ash that's inherent in that type of coal, and people

have indicated things like the iron content also plays a

role, and the ash content can vary dramatically from one

type of coal to another, and then it can actually absorb

Page71

some of the elemental mercury that's emitted, or in

other cases, it might tend to have more particulate

mercury, so those things are very important, in terms of

how much mercury comes out, and again, that's not my

area of expertise. I'm just going based on what I have

been told.

Q. Do power plant configurations affect this,

as well?

A. Yes. Again, that's not my area of

expertise, and perhaps somebody else would be better off

handling that question, but I know control technology

will affect the speciation that comes out of the stack.

Q. Do you know in what way?

A. Well, again, it's a fairly complex

relationship because it depends on what the emissions

were, to start with, so if it's a high chlorine coal,

which has lots of reactive mercury. If a wet scrubber

is used, you will tend to use more mercury. It it's

elemental mercury, a wet scrubber is not going to be as

effective, but again, that's not area of expertise. I

recommend that one of the engineers answer those

questions.

DR. KEELER: So that was 11 --

MADAM HEARING OFFICER: I don't think

Page72

we've addressed nine or 10, yet.

DR. KEELER: 9, I answered, "Do

different types of emissions source --

MR. RIESER CONTINUES:

Q. Let me can ask more question, if I may. I

don't believe you addressed steel manufacturing, in

whether those are sources, in your view.

A. Yes. They are sources of mercury. Any

sources that uses fossil fuel for its combustion fuel is

going to have a mercury emission, and so iron and steel

industry definitely does and the manufacturing of steel

produces mercury. We find -- we found quite a bit of

particulate mercury, in fact, in the Chicago area, for

example, due to the iron and steel industry.

Q. Would the amount of mercury and the type

of mercury produced or emitted -- put it that way --

vary on the type of steelmaking process? For example,

wet furnaces, as opposed to other types of operations?

A. Again, I'm not an expert on the steel

manufacturing process, so I would rather not comment on

that.

Q. Thank you.

MR. BONEBRAKE CONTINUES:

Q. Do refineries also emit mercury?

Page73

A. Yes, they do.

Q. What forms of mercury are emitted by

refineries?

A. We see predominantly elemental mercury

when we see influence from refineries.

Q. And in connection with the work that you

had mentioned concerning Lake Michigan, were you finding

any elemental mercury from refineries associated with

the mercury in Lake Michigan?

A. I know that we saw refinery influences on

the particular matter, and I don't recall that we saw a

strong signal from refining in the gaseous mercury.

DR. KEELER: Now, Question 11: "Is

the mercury deposition pattern for incinerators

different than it is for coal-fired electric generating

utilities?" I would say yes, and for the two reasons

that the questions were asked before me. The form of

the mercury coming out of the two are -- can be

different and the height of the stacks of an

incinerator, versus a coal-fired utility would also be

quite different, although some of the electric

generating units in the state have relatively low stack

heights compared to some of the larger ones elsewhere.

Some down in the couple-hundred-feet range. That's more

Page74

typical of an incinerator stack height. In general,

those two things will cause a difference in the

deposition pattern.

MS. BASSI CONTINUES:

Q. Is the state you're referring to Illinois?

A. Yes, it was.

MR. RIESER CONTINUES:

Q. Is it correct that the differences are

associated with stack heights, and in what way is it

associated with stack heights?

A. Well, lower stack height putting out large

amounts of reactive mercury is going to really tends to

push things towards more localized deposition than a

high stack height with the same amount of reactive

mercury, and then if you have less reactive mercury with

less stack height, you have very localized, within a few

kilometers of the plant type deposition.

Q. When you refer to some power plants in

Illinois having stack heights of a certain height, which

plants were you referring to?

A. I have -- I guess this is a list of all

the power plants in the state sorted by stack height.

Q. We have extra copies, don't we?

MADAM HEARING OFFICER: We need to

Page75

admit that as an exhibit.

DR. KEELER: He's going to get some

copies.

MR. RIESER CONTINUES: Why don't we

hold on the issues on this issue, until I can get a copy

of this. Thank you.

DR. KEELER: Question 12: "Not all

inorganic mercury deposited to all water bodies from the

atmosphere become methylated. Is that correct?" Yes, it

is correct. Some of the mercury that's deposited to a

water body, depending upon the form that it's deposited

in, can actually be transformed into elemental mercury

and again be re-admitted from service, so some fraction

of the mercury gets immediately sent back up. Some of

the mercury that's deposited to a body of water will

attach to particles and sink to sediments, so that it

won't instantaneously be methylated. It has potential

to be methylated at a later time.

MR. BONEBRAKE CONTINUES:

Q. What percentage of mercury, as you just

described it that's deposited into water bodies is

transformed and re-admitted?

A. It's water-body specific.

Q. Is there any general range that results of

Page76

that process, recognizing that there's going to be

variability among water bodies?

A. I would say there's no general range that

I can say.

DR. KEELER: Question 13: "Some of

the mercury deposited into water bodies is reemitted.

Is that correct?" As I just mentioned, yes, it is. In

the Lake Michigan Mass Balance Study, for example, of

the mercury that was deposited to Lake Michigan, about

one-third was reemitted, so two-thirds of the mercury

went into the water body and one-third was reemitted.

Question 14: "The amount of

methylation" --

MR. BONEBRAKE CONTINUES:

Q. Just a follow-up to that answer, do you

have a reason to believe that that percentage would be

significantly different for other water bodies in the

Midwest?

A. I believe that it would be different than

inland lakes. It may be very typical for the large

lakes, but I think you would get a real range in

concentrations, depending on the aquatic chemistry.

Q. In the lakes, would you have thought it to

be higher or lower?

Page77

A. Again, it would vary quite a bit just as

the chemistry in the lakes varies quite a bit. The DOC

and the calcium carbonate and other basic chemicals that

are in the lakes, those concentrations vary over, like,

the whole state of Michigan, so I would think the

evasion rates would vary, as well.

MADAM HEARING OFFICER: Mr. Zabel.

MR. ZABEL CONTINUES:

Q. Just as a follow-up to that question, was

a similar analysis done on the deposition of the

Everglades?

A. We published a paper looking at the earth

surface exchange of mercury. It was not a mass balance,

so I can't put a quantitative number on how much mercury

was deposited versus how much was reemitted. The other

difference is that we were able to calculate the flux

from Lake Michigan based on measurements of the water

body. The situations clouded in the case of the

Everglades because of the vegetation that's there.

There's a strong tendency for the aquatic plants to

actually mediate some of that evasion, so some of the

mercury that's in the sediment actually gets sent into

the atmosphere via the plants, and so we didn't do like

an evasion from the marsh land as part of that study.

Page78

Q. So there would be two reemission

mechanisms in the case of the Everglades?

A. That's correct. Instead of just being

from water, there would be another one.

Q. Is the reemission process that we've been

discussing limited to elemental mercury deposited into a

water body?

A. Yes.

Q. So the higher the level of deposited

elemental mercury to a water body, generally speaking,

the higher the reemission rate for a water body?

A. I would not say that's correct.

Q. What about that statement is inaccurate?

A. I would say that there's probably not been

enough work to suggest that the more elemental that goes

in, the more elemental that comes out. That's not

necessarily consistent with some of the work that we've

done.

Q. Although the percentage remains constant,

and there's a greater volume of elemental going in,

that's going to suggest a greater volume of elemental

being reemitted, right?

A. That's if there were no other processes

affecting the elemental mercury once it was deposited to

Page79

the surface, which there are tremendous processes.

Mercury attaches to particles as soon as it transforms,

so it's not a correct statement.

MR. ZABEL CONTINUES:

Q. I guess a follow-up question of mine, did

the model used in the Everglades take reemission into

account?

A. The cycling model does, yes.

Q. How does it do this?

A. Again, I think it has biotic than

antibiotic, which means through chemically transforming

the mercury into elementally formed evasion (phonetic)

and also biologically changing the form of mercury into

-- the exact mechanism, you would have to refer to the

report in order to be able to get that.

Q. I guess what I was curious about was how

you indicated that it was more difficult in the

Everglades than in Lake Michigan, how that was

quantified in the model then.

A. It was modeled.

Q. It was just modeled?

A. Yes.

Q. There was no collection of data comparable

to Lake Michigan?

Page80

A. No, there was not. There was no

concurrent, long-term monitoring of that. The work that

we did in the Everglades was part of a specific dried up

deposition project that we did where we made

measurements through the winter and summer seasons to

get, for the first time, some answers on evasion and dry

deposition of mercury to that ecosystem, but there was

no long-term record that would be comparable that could

be used in the Florida TMDL work as we did with the Lake

Michigan Mass Balance Study.

Q. So when you did the modeling on the

Everglades, you had assumptions about reemission?

A. I didn't make any assumptions. The

Tetricheck people had to take the knowledge that was

gained from other people's work and use that in their

models, yes.

MR. RIESER: My question has been

asked an answered.

MADAM HEARING OFFICER: Mr. Harley.

MR. HARLEY CONTINUES:

Q. Would you elaborate on how the chlorine

content of coal effects the emission of reactive

mercury?

A. The question was on the chlorine content

Page81

of coal and its relationship to RGM emissions.

MADAM HEARING OFFICER: The RGM, is

that --

DR. KEELER: Reactive gaseous mercury.

The relationship suggests that the more chlorine that's

in the coal, the more reactive gaseous mercury you are

going to form, and that gaseous mercury would be in the

form of chloride coming out of the stack, so or some

other compound that would have chlorine in it, so more

chlorine, more reactive mercury has been the tendency.

Q. Is a higher concentration of chlorine in

coal associated with bituminous or sub-bituminous coal?

A. That's a good question. It's actually a

fairly complex question to answer. Just saying that

sub-bituminous coal would have less chlorine or more

chlorine is too simple an answer, I believe. It think

it's quite variable. In general, it may be that the

average for bituminous coal is slightly higher than for

sub-bituminous, but again, I think you can find pretty

wide variability within those two types of coal, and

then with other coals, so I think that's not that

straightforward.

MR. RIESER CONTINUES:

Q. So you don't agree that, on average,

Page82

sub-bituminous coal has significantly less chlorine than

on an average of bituminous?

A. No. I said I think it's possible that the

average might be less for sub-bituminous than for

bituminous, but I said there's a large variability in

that, and that I think you would have to look at the

specific type of coal to make sure that, in fact, that

was true. The important parts in that is that

sub-bituminous coal also will have less BTU's, so you

have to burn more of it, and so the relationship between

the chorine in the mercury and the energy content really

are important, in terms of whether you put out more

reactive mercury from that same type of -- or from that

same amount of energy generated, so again, coal and

chemistry of coal is, again, not my area of expertise.

I'm not a geologist, but I'm going based on, again, much

of the fine work that was done by the utility companies.

Q. Based on that fine work, focusing in on

one type of sub-bituminous coal, I believe is how it's

frequently used around here, with respect to Power River

Basin coal, do you have an understanding of what

chlorine levels we see in Power River Basin coal is

significantly less than chlorine levels seen in Illinois

bituminous coal?

Page83

A. I would say that statement is correct, on

average.

MADAM HEARING OFFICER: Just so you

know, you have to speak directly into that microphone,

especially when you turned your head. We started to

lose you. Mr. Harley, did you have a follow-up?

MR. HARLEY CONTINUES:

Q. So does it follow, therefore, that a coal

plant operator who switched from low chlorine

sub-bituminous coal that's from less, to higher chlorine

bituminous coal from Illinois, there would be a

co-benefit in doing that in reducing, potentially,

mercury emissions from that facility?

A. I understand your question. Let me answer

this. Probably the reason we're having these hearings

in first place is this is a complex issue. The control

technologies that will be used play a big role on being

able to answer that question. From going to many

meetings, such as the one I went to in china where they

talk about control technologies and so forth, the

vendors that I heard speak said that, if there was more

reactive mercury coming out of the coal-fired utility

because of higher chlorine content, that they would be

able to remove the mercury more easily, than if it was

Page84

all in the elemental form, which requires a little bit

more difficult control measures, so if, in fact, what

you said was true, and they had the proper controls,

then, perhaps, they could reduce the total amount of

mercury more easily than if it was more in the elemental

form, but again, there's a million assumptions in that

that there's controlled technology. The exact type of

coal, the ash content, all of those kinds of things, so

there are a lot of engineering variables that go into

that that make it a difficult question to answer.

MR. RIESER CONTINUES:

Q. Would it be accurate that a lot of these

questions that were answered or were, at least,

discussed, and if not, will be part of the technology

discussion that will probably take place next week?

MR. KIM: Yes.

MR. RIESER: That's the beauty of a

simple answer. Let me go back, if I may.

MADAM HEARING OFFICER: We're getting

some interference. I think we are going to switch it

out.

MR. RIESER CONTINUES:

Q. Just to go back to the issue of the

reemission of mercury, is it accurate that the way the

Page85

cycle works through the water body that some of both of

the RGM and the elemental mercury that are deposited to

a water body are reemitted as elemental mercury?

A. Yes. The mercury that deposits in either

form will be reemitted in the form of elemental mercury.

Q. And so to the extent it's reemitted, that

bad mercury is not available for methylation, correct?

A. That's correct.

MADAM HEARING OFFICER: Before we go

on to the next question, Mr. Kim, did you have those

exhibits?

DR. KEELER: The table that's being

submitted is a listing of the Illinois coal-fired EGU

stack heights by the facilities.

MADAM HEARING OFFICER: And we

reserved Exhibit No. 27 for the Detroit Study, so we

will mark this as Exhibit 28, if there's no objection.

Seeing none, it's Exhibit 28.

(Exhibit No. 28 was admitted.)

MR. RIESER: I have the Detroit Study,

if it would be a good time to throw it in.

MADAM HEARING OFFICER: Sure. We'll

go ahead and mark it. I have also been handed "The

Detroit, Michigan, Source-Receptor Relationships for

Page86

Atmospheric Mercury in Detroit, Michigan," which we

reserved Exhibit No. 27 for. If there's no objection,

we will mark that as Exhibit 27. Seeing none, that's

Exhibit No. 27. Thank you.

(Exhibit No. 27 was admitted.)

MADAM HEARING OFFICER: Mr. Rieser,

you had some questions based on the stack height I

believe that you wanted to ask.

MR. RIESER: You know what, why don't

I look at it at a break, and we will come back to it.

MADAM HEARING OFFICER: That's fine.

Then I believe we are on Question 14 from Dynegy.

MR. ZABEL CONTINUES:

Q. Just a quick question on Exhibit 28, it

says, "Ranked by gross load megawatts." I don't think

the Baldwin station has 13 megawatts. Is it megawatt

hours?

A. I believe that's correct.

Q. Then it must have some particular year. I

guess I don't know what the number represents, if you

could find out Mr. Kim and let us know?

MR. KIM: We can find out.

MR. BONEBRAKE CONTINUES:

Q. A follow-up, do you recall, Dr. Keeler,

Page87

the average height of the municipal waste combustion

units in question in the Florida Study?

A. The average weight?

Q. Average stack height.

A. I don't recall what the average stack

height of all those waste combusters are, but they tend

to be 150, or so, 150 or 200 feet or less, but I don't

know what the average is.

Q. Do you know what the average was with

respect to the stack height of the medical waste

incinerator unis at issue in the Florida Study?

A. It was definitely less than 150 and maybe

in the Florida report I -- I don't recall if that

information was in there or not.

DR. KEELER: Question 14: "The amount

of methylation that can occur in a water body depends on

site-specific conditions. Is that correct?" Yes.

Question 15: "Demethylation can also occur. Is that

correct?" Yes. Question 16: "Is it possible to

accurately predict the amount of methylmercury that will

be found in a fish based on atmospheric deposition of

the inorganic mercury to the water body that the fish

lives in?" As we have discussed, models have been

developed that will predict the amount of methylmercury

Page88

found in a fish based on the atmospheric deposition of

inorganic mercury to the water body that the fish lives

in, and these results appear to give very reasonable

answers. The models that apply to the Florida

Everglades as part of that TMDL gave results that

compare I think very favorably with the actual measure

data of mercury in the fish in those ecosystems, so I

would say that it is possible to predict.

MR. HARRINGTON CONTINUES:

Q. Was that model adopted for the

Massachusetts Study.

A. I'm not aware of the Massachusetts Study.

Q. Were there any -- was that model utilized

in any other studies?

A. I believe it has been used in other

studies and off the top of my head, I'm not sure where

else, but I think there were other TMDL's that

Tetricheck has worked on.

Q. Was there any further verification of a

model with actual field data to determine its accuracy?

A. The data has been continually collected in

Florid, and it seems to track the predictions very well,

so they continue to collect biological samples from the

Everglades area where the modeling was performed, and

Page89

yes, they are doing continued verification, and the

model predictions are tracking pretty nicely.

Q. That's with respect to Florida?

A. Yes.

Q. That's very peculiar and unique

environment, is it not, the Florida environment?

A. It is a unique environment.

Q. And would that model necessarily be

applicable anywhere else as adopted in Florida?

A. As I mentioned earlier, you would not use

the same parameters that you would use in the Florida

case. You would adapt them using the ecosystem

parameters that you would measure in that specific

ecosystem, and so the things that were adopted, as you

suggested, for Florida would not be appropriate to

another location. You would have to use correct aquatic

chemistry data, environmental conditions, climatology,

all those kinds of things, in other words, for that

model to work. I believe they did apply it where it was

developed, up in Wisconsin.

MR. BONEBRAKE CONTINUES:

Q. The model that was used in the Florida

study, Dr. Keeler, that model was used to predict

methylmercury levels in fish tissue at only one

Page90

location. Is that correct?

A. Again, I think it predicts at more than

one location, and again that was not my work. That was

work done by Tetricheck for the State of Florida. It is

detailed in that report.

Q. You recall this morning we looked at some

language in the Florida report, which indicated that

three sites in the vicinity of what was referred to as

the "hot spot" in half of the -- what was called the

cohorts, there was no change in methylmercury fish

tissues. Do you recall that?

A. Yes, I do.

Q. Does that, therefore, mean that the model

that was used in the Florida Study was, in half of those

cases, predicting no change in methylmercury levels?

A. The observation was that there was no

change, correct, in that table.

Q. That's what we talked about this morning.

A. Right, so the model predicted that you

would see a change, and in some of the fish, it was born

out that there was change, and some of them there was

not, that's correct.

Q. So the model did not accurately predict

methylmercury fish tissue levels, at least, at all

Page91

locations. Is that correct?

A. That would be a correct statement.

MADAM HEARING OFFICER: I think that

finishes Dynegy's questions. Mr. Kim, where are we going

next?

MR. KIM: I believe we are next going

to move to Ameren's questions to Dr. Keeler. I'm kind

of hopeful when we get to Prairie State's questions that

they have already answered them in the course of others,

but we will look and see.

DR. KEELER: Question 1: "According

to your report and written testimony, the scope of your

presentation to the IPCB is to describe the source of

the mercury deposition to the Great Lakes, and to

specifically discuss the importance of coal-fired

utilities to the region. A: Is it correct that you

were not asked to address the impact of Illinois coal

plants on mercury deposition within Illinois, and B: Is

it correct that you have not performed any

source-receptor studies which determine impact of

Illinois coal plants on mercury deposition in Illinois,

and C: Are you aware of any -- are you aware of whether

any such studies have been performed?" I was asked by

the State of Illinois to address the work that we have

Page92

done looking at the impacts of various sources on

mercury deposition to the region. They did not ask me

to not look at it, the State of Illinois, as the

question suggests. I was asked to look at all sources

and all of the sources of deposition to the Great Lakes,

including the state of Illinois. It is correct that I

have not performed an Illinois-specific source-receptor

study, but having concluded Illinois within the greater

context in the Great Lakes in the study that we have

done, we have made measurements, as I mentioned earlier,

in three times one in Kankakee, Illinois, as part of the

Lake Michigan Urban Air Toxic Study, and Chicago as part

of the Lake Michigan Mass Balance Study and in

Bonnville, Michigan, as part of that same study, but we

-- again, those were in the context of looking at Lake

Michigan Study, not specifically, looking at Illinois.

I'm unaware of any other studies that have been

performed, specifically on the state of Illinois for

this purpose.

MR. BONEBRAKE CONTINUES: Did you use

the phrase "Illinois-specific study" in that answer?

A. I believe I did, yes.

Q. What do you mean by "Illinois-specific

study"?

Page93

A. The question asked me if I -- "Is it a

correct that you were not asked to address the impact of

Illinois coal plants on mercury deposition with

Illinois?" That is incorrect. That's what I meant by

"specific," that specific study about coal plants in

Illinois on Illinois. That was a double negative.

MR. KIM: It's a funny-worded

question. I think he's -- I think he's tried to answer

it within the frame work of the question. If you would

like to rephrase it.

MR. BONEBRAKE: It's not my question.

DR. KEELER: 1-A that is no, that is

incorrect.

MR. RIESER CONTINUES:

Q. It's incorrect in that you -- start over.

What you were asked to do with your testimony was look

at all the studies that have been performed in the Great

Lakes region and summarize them, correct? That was one

of your tasks.

A. No. They asked me to look at the studies

that my laboratory has performed looking at the sources

of mercury to the Great Lakes region.

Q. Page two of the question -- excuse me --

page two of the testimony under "Purpose of Testimony"

Page94

says, "I was asked by the Agency to prepare a

state-of-the-art assessment of the sources of mercury

deposition to the Great Lakes, and specifically, discuss

the importance of coal-fired utilities on the deposition

of mercury to the region." Correct?

A. That's what I tried to paraphrase just

now.

Q. By "Great Lakes," did we discuss this

morning that the Great Lakes reference is to the Great

Lakes Basin which only includes a small portion of the

state of Illinois?

A. The Great Lakes Basin reference was only

in reference to the one modeling study that I referred

to where we were looking at within the Great Lakes

Basin, sources to those, which are outside the Great

Lakes Basin, but our body of work is larger than just

the Great Lakes Basin.

MADAM HEARING OFFICER: That would

have been Exhibit 26, the report.

MR. RIESER CONTINUES:

Q. Thank you. When you use the term "Great

Lakes" in this sentence, what did you mean by it?

A. I mean all the Great Lakes states,

basically.

Page95

Q. All the Great Lakes states?

A. And the province of Ontario.

Q. When you use the term "region" in this

context, does this mean the same thing?

A. Same thing, yes.

Q. But other than the studies you described

that had sampling stations in some points in Illinois,

you haven't done any studies within the state of

Illinois?

A. Only the studies that we collected in the

samples in Illinois are the ones I'm referring to.

MR. ZABEL CONTINUES:

Q. Just so I'm clear, Doctor, the Great Lakes

study was all five Great Lakes?

A. Yes.

Q. I don't count Lake Champlane, as some New

York congressman would.

A. Actually, it was a Vermont congressman,

but yes, that's a correct statement.

Q. So that was the region that ranged from --

A. From Minnesota, to New York in this case.

Q. Thank you.

DR. KEELER: D: "What is the basis

for your statement at the end of your testimony that,

Page96

`Areas with elevated mercury deposition due to emissions

related to coal combustion have been identified'"? This

statement is based upon the work that we've done in the

state of Ohio as part of the Steubenville Study where we

performed a receptor modeling study to determine that,

approximately, 70 percent of the mercury deposited via

wet deposition at that site was related to coal-fired

utility emissions. E: Are these areas in Illinois?"

How have they been identified?" The answer to that is

no. F: "Did you participate in drafting the technical

support document, TSD, prepared by Illinois EPA for

these proceedings?" Yes. I contributed to the

Technical Support Document that is found in the

appendices.

MR. RIESER CONTINUES:

Q. You added the report that's Appendix B to

the TSD. Is that correct?

A. I don't recall what the --

MADAM HEARING OFFICER: Yes, that's

correct.

MR. RIESER CONTINUES:

Q. Did -- were you involved in the drafting

of the section which I believe is Section 5 that deals

with atmospheric deposition?

Page97

A. No.

Q. Did you review it?

A. After it was submitted I looked at it,

yes.

Q. But not before it was submitted?

A. No, sir.

DR. KEELER: Question No. 2: "On page

81 of the TSD it states that `Thus it can be expected

that significant mercury emissions reductions in

Illinois will yield significant reductions of mercury

deposition in Illinois.' Did you author this

statement?" No. Do you -- B: "Do you make the

statement in your report in Appendix B to the TSD or in

your testimony?" I did not make that direct statement

in either of those. C: "Do you believe that there is a

factual basis for this statement?" And my answer is

yes. I do believe it's a factual basis in that

statement.

MR. RIESER CONTINUES:

Q. Could you please describe the factual

basis which may have been well where you were going.

A. Again, the factual basis for the agreeing

with that statement is that we have done a body of work

now over the past 15 years looking at the sources of

Page98

mercury deposition to areas all over the Great Lakes

Basin, including some sites here in the state of

Illinois, which suggest to me that coal-fired utility

emissions are having an impact on mercury deposition in

the state, and do have an impact on the mercury

concentrations in the air in the state of Illinois.

That's why I agree with those statements.

Q. Is there quantification that you have

performed that can measure that amount?

A. Again, the Lake of Michigan Mass Balance

Study we collected samples for 18 months at Bonnville,

Illinois, in wet deposition, as well as the gas and

particle phase. Again, that's gas and particle phase

mercury at that site and source proportion was done on

that data, as well as the data that was collected in

Chicago, Illinois, looking at the various sources, and

that work was actually in the doctoral thesis of Matt

Landis, and that can be looked at there. Some of that

work is in the peer-reviewed literature, and in the

paper that you suggested, and in that work, they found a

significant -- again, my memory is that it was about 20

to 30 percent of the ambient mercury in deposition was

related to coal combustion.

MADAM HEARING OFFICER: Mr. Harley.

Page99

MR. HARLEY CONTINUES:

Q. The work that you just described, is that

the basis of the statement in your conclusion that 21

coal-fired power plants in Illinois emit close to four

tons per year of mercury into the atmosphere?

A. No. That's based on the emissions

inventory.

Q. Thank you.

MR. RIESER CONTINUES:

Q. The reference you had to the Chicago

Study, is that the paper we talked about this morning

that I think is Exhibit 26, "Atmospheric Mercury in the

Lake Michigan Basin" --

A. Yes.

Q. Is it accurate that the sampling for that

study was primarily applicable to the lake?

A. No, that's incorrect. Correct. We did

have sampling over the lake, in addition to running five

sites, Bonnville, Chicago, a site on the border between

Illinois and Wisconsin, a site in South Haven, Michigan,

and a site in Sleeping Bear Dunes, Michigan. At the

same time, we were taking measurements at a site in Lake

Superior up at Eagle Harbor, as well as in Dexter, so

all of those sites were simultaneous, so the great

Page100

majority of the samples that are collected were not

collected over the water.

Q. And did you also say you've done a source

apportionment as part of that study?

A. Yes. Dr. Landis did a source

apportionment as part of his thesis work.

Q. That source apportionment consisted of

associating certain chemicals with certain types of --

certain types of groups of chemicals together? Would

that be correct?

A. It was a variance of the type of analysis

that we did in Ohio, yes. It was a receptor modeling

calculation that used both the elemental and chemical

composition of the precipitation or ambient samples

together with the meteorological data.

Q. When you do that type of study, does that

allow you to identify specific sources or groups of

sources?

A. Receptor modeling, in general, unless it's

done on a plant that's in isolation out in the middle of

nowhere really gives you an answer for the source

category. It does not give you an answer for a specific

source.

Q. So by doing that type of study, you

Page101

couldn't identify a particular power plant as the sole

source of coal combustion in the chemicals that you're

seeing. Is that correct?

A. The only way that that could happen is if

they were -- if it was distance wise separated from

other plants. If there were two plants next door to

each other, it would be difficult to do that, yes.

Q. You also make the statement that -- this

is on page 4515 of the report, that?

MADAM HEARING OFFICER: Specify --

MR. RIESER CONTINUES:

Q. Over the paper -- excuse me -- and this is

Exhibit 26. You done remember every page of every

report?

A. No, getting hard to do that.

Q. Right down at the bottom, the paragraph

begins --

A. 4511?

Q. 4515. At the bottom, it says, "In fact,

the urban air shed" -- and you're referring to the

Chicago-Gary open area -- "the urban air shed was a

complex system of numerous point sources in distinctive

meteorology." Do you see that?

A. On 45 --

Page102

MADAM HEARING OFFICER: It's on the

left side.

MR. RIESER CONTINUES:

Q. It begins "The spacial -- you see that?

A. I see that.

Q. We're in the same place. You still

believe that, I assume, that the urban air shed was a

complex system with numerous point systems in

distinctive meteorology?

A. Yes. I think urban areas are, generally,

that way.

Q. Does that mean that it matters a great

deal where you put your sampling point, in terms of what

findings you are going to get?

A. This paragraph, specifically, was

addressing where you put the reserve vessel over the

water in order to be able to see specific source

influences, yes.

Q. So you would agree with me that it matters

where you put your sampling point?

A. It matters really in a big way when

talking about over-water measurements, yes. It matters,

generally, yes, where you put your sampling sites, but

it really made a big difference in terms of over-water

Page103

measurements because of the strong stability over the

water during the summer months.

DR. KEELER: Question 3: "Is it

correct that the conclusions in your testimony regarding

the sources of mercury deposition were based on the work

you did in Steubenville Ohio?" Yes. "Has that work

been published in any peer-reviewed journal?" The work,

at this point, is now in the process of being published

in a peer-reviewed scientific journal. We were hoping

to submit the paper about a year ago, but it was held up

in that process as a request from the then EPA

administrators, since that is a cooperative agreement,

which means that we're working collaboratively with

scientists from the U.S. EPA of Research and Development

that our paper be subjected to, both, a strict and

rigorous internal EPA review by their scientists, as

well as an independent outside review by independent

scientists outside of the EPA before we submitted it to

the journal, largely, because of the sensitivity

surrounding the upcoming CAMR rule, so that delayed the

publication submission by almost a year, so the

publication is right now in the process of being

finalized, and we expect publication of peer-reviewed

literature within the next three months.

Page104

MR. RIESER: I'd just like to say

that, because we don't have the actual report with its

supporting data and sources and specific conclusions,

that, should this proceeding be continuing after that

study is published, that I reserve the right to ask

additional questions of Dr. Keeler after the time that

it's published.

MADAM HEARING OFFICER: So noted.

DR. KEELER: Question B: "Is it

possible that you will change any of your statements or

conclusions as a result of the peer review process?"

No. The reviews came back quite favorably, and only

made suggestive and clarifying comments regarding any of

the science that was performed, and so no, none of the

conclusions will have changed, and even the quantitative

statements will not change. C: "Is the underlying data

available for review?" EPA policy, as well as that of

most scientists is that, once the peer-reviewed

publication becomes in press -- I mean, not in press. I

mean has come out in peer review that the underlying

data can be made available, so that's the answer.

That's a fairly consistent way that everyone approaches

intelligible property rights when it comes to scientific

investigation.

Page105

Q. So the answer to C is no. Is that

correct?

A. The underlying data is not available today

for review. I should offer, though, that as part of the

peer review process that EPA undertook, it was

unprecedented, in terms of the rigor of the evaluation.

We provided them, not only the raw data that we

collected as part of the study, but we also provided

them with the exact model and model formalisms, together

with all the input parameters that we used in our

calculations to allow them to independently run the

models and come to their own conclusions, and in

addition, apply their own receptor models and

statistical approaches to look at the data, and came up

with the exact same answers that we did in that review

process, so in terms of the external review, that was

done.

Q. Is it accurate that, both, yourself and --

is it Mr. or Dr. Landis?

A. Doctor.

Q. Both, yourself and Dr. Landis, have

presented Powerpoint presentations that include some

portions of the data?

A. Do you mean that you don't have?

Page106

Q. Is it true that you presented Powerpoint

presentations that contain some portions of the data?

A. Yes.

Q. And that was selected by you, and I assume

Dr. Landis for the purposes of those presentations?

A. What was selected by me?

Q. The data that you chose to present at

those presentations.

A. Yes.

Q. And the answer to the question and then

the question in terms of what I have and don't have, of

course, is whether there's a copy available of the

presentation made to LADCO in February that I didn't

get.

A. Yes. There is a presentation copy

available that you can have, yes. All the public

presentations are available. Most of them -- I think

four or five --or publicly available on the website and

some of the utility groups already have the LADCO

presentation.

Q. But you understand that the LADCO

presentation has not been on LADCO's website, so to my

knowledge, it hasn't been available, so you can make it

available is part of the process?

Page107

A. Yes.

Q. Thank you. When can we expect to get a

copy of that? Can we get one today, so we can talk

about that tomorrow?

A. I would have to check to make sure that I

have that on my laptop and if I do, I can get a copy of

that to you.

MR. BONEBRAKE CONTINUES:

Q. You mentioned some feedback from U.S. EPA

to the data that you provided to the Agency. Has that

feedback been in writing?

A. The feedback from EPA regarding the data?

Q. Regarding the data that you submitted for

their review in the Steubenville Study?

A. We got a written review, yes, that was

provided from the peer reviewers, yes.

Q. Peer reviews at U.S. EPA?

A. Those were -- no, there was not. These

were external peer reviewers.

Q. Who were retained by U.S. EPA?

A. By an outside contractor working for U.S.

EPA.

Q. Is a copy of those comments publicly

available?

Page108

A. You would have to direct that question to

the U.S. EPA.

Q. Do you have a copy of those?

A. I do, yes, not with me. I have a paper

copy back in my office. I don't carry them with me. It

was probably about 200 pages worth of comments because

of the extensive review that they did.

DR. KEELER: Question D: "Have you

ever released, for public review, any description of the

methodology you used for source attribution?" We have

made presentations at a few forums that describe the

methodology for source attribution, and as I mentioned

earlier, all the methods that we used have been

published extensively in the peer-reviewed literature,

so we did not use any new techniques like, PMF and

Unmix, which are two statistical approaches developed by

Paatero and Hopke, as well as Ron Henry at USC, and it

has been extensively peer reviewed.

MR. RIESER CONTINUES:

Q. And how often have those methodologies

been applied to mercury?

A. Well, that's one of the novel parts of

what we have done. They have not been applied very

often for mercury.

Page109

Q. Have they been applied ever for mercury?

A. I'm not sure of the answer to that

question. There is a possibility that there's one

application, but I don't know.

Q. And we've talked about this methodology

for source attribution briefly in different parts. Is

this the same methodology you used in Florida?

A. Actually, the methodology we used in

Florida was similar. The paper deadvantage et al. 1999

that was in "Environmental Science and Technology" used

a similar approach. It was a factor analysis

multivaried technique. The techniques that we're using

now have are a decade later, and they have gone through

a whole tremendous amount of improvement, including the

statistical handling and uncertainty analysis, and so

forth, so it's a much better and much more robust

statistical approach to doing source apportionment than

what we used in Florida.

Q. This probably as good a time as any for

you to describe that method.

A. The Florida methodology?

Q. No, the source attribution methodology

that you used in Steubenville.

A. Sure . As I mentioned earlier, what the

Page110

receptor modeling does -- the receptor model methodology

utilizes measurements taken at a location, and those

measurements include -- in this case, we make

measurements of precipitation every time it rains, so

every day that rain falls, we collect that, and then we

take and analyze those samples very carefully in a Class

I clean laboratory at the University of Michigan for the

concentrations of trace elements by some fairly

sophisticated analytical techniques, ICPMS, ion-coupled

plasma, mass spirometry and for mercury and major ions,

so we have a list of about 40 trace elements. Ions and

chemical constituents in every single rain sample. The

statistical approach then takes all of those samples and

works backwards, statistically, to determine what were

the major factors that contributed to the variation that

was found in the precipitation mercury levels, so it

takes this very complex large database and kind of

mathematically determines what sources contributed to

mercury in each of those samples. That methodology can

be done -- or that approach is done in multiple ways.

In this case, in our paper, we have used both positive

matrix factorization, which is a program written by

Dr. Paatero and Hopke, which ensures that the factors

that are calculated -- or the source factors that are

Page111

calculated are mathematically stable, and also, allow

you to input the uncertainty in the measurements that

you make. And then another model that's independent,

but a similar mathematical concept is Unmix. Again

these are both factor analysis models is independent and

doesn't allow you to include the uncertainty

calculations into that. We ran both of these models

independently. In fact, one of the questions, the next

question is "What were your respective roles on the

Steubenville project?" One of the ways that we worked

on this project collaboratively is EPA has several

receptor models that are expert in using these models

and developing these models and at the University of

Michigan we independently ran the PMF and unmixed models

from the researchers that were doing the work at EPA, so

that we could then compare the results that we got. In

the process of doing the initial work, both models wound

up giving us comparable answers. One of the questions

later on was why did I say about 70 percent contribution

from coal-fired utilities? The reason for the

approximate was so that I could more precisely give the

answer for a combination of the two models. PMF gave a

contribution to coal-fired utilities of 70 percent, and

Unmix actually gave a contribution of 74 percent, but

Page112

there was an uncertainty of about 15 percent, about both

of those numbers, and therefore, the mean for the two I

could have put 72 percent, 72.1, or whatever it was, but

I put it short hand as 70 with an uncertainty about

those numbers, so those methods, as I said before, have

being rigorously compared and developed over the last 15

years down at EPA, and in the scientific community.

Q. Let's take this a step at a time to break

it down. In order to identify -- when you use the term

"source" in the phrase "source-receptor modeling" --

A. Yes.

Q. Again, as you said earlier, that refers

not to a specific source, but to a class of sources,

correct?

A. Yes. That's correct.

Q. So you got coal-fired power plants

automobiles, steel plants, whatever, and in order to

identify a given source, is it correct that you identify

certain chemical fingerprints, if you will, associated

with those types of emissions?

A. Yeah, and that's the key to the whole

thing is that we do identify what we call chemical

signatures. It 's like a fingerprint for a person.

Each major source category will have a specific

Page113

fingerprint, or signature, and sources that are in the

same source category have fairly similar signatures. If

they were dramatically different, then they would look

like two different sources, so two different coal plants

that very different signatures, so one that burned one

type of coal, versus burning another type of coal

exclusively, would look like two different sources. In

this case, we have signatures that have been developed

through the literature over the past, since the mid

70's, that give us ideas about what these signatures

should look like from coal burning, from motor vehicles,

from iron and steel production, from all kinds of

different sources. One of the things that sets my group

apart from the work of other receptor models is that we

do all of our own chemical analysis in-house, so we are

involved in collecting actually source-signature data

from plants, from motor vehicles out in the field at the

sources, and then also analyzing the samples that we

collect in the field, so that there's no possible

differences between analytical laboratories, so we've

been very instrumental in providing the actual source

signatures and using the advanced analytical techniques

that we have, we actually are pushing the forefront in

finding new elements and new chemical markers and new

Page114

signatures for the various sources that we're looking

at.

Q. I think you said in that answer that the

chemical signature for coal plants is the same across

all types of coal plants. Is that correct?

A. No. That's not what I said. I said that

if two plants were burning different types of coal that

they would have a different signature.

Q. So have you identified different

signatures for plants that burn bituminous, as opposed

to plants that burn sub-bituminous coal?

A. We have analyzed coal from that

sub-bituminous coal versus bituminous coal, and it has a

different characteristic signature. You -- it's getting

back to the question that was asked about the chlorine

content. The chemistry of the coal is different.

That's why there's different mercury, different reactive

mercury formation. We are working towards trying to

identify an ambient signal that would allow us to

separate out the two types of coal in the ambient air.

Q. But at this point, you don't have that

capability?

A. No, and it wasn't important, in terms of

the receptor modeling that we were doing. It's just a

Page115

goal that we have.

Q. And is it accurate that the

source-receptor studies you have performed to date have

been primarily in areas that burn in bituminous coal

such as Steubenville?

A. The type of coal that's mined in that area

is certainly sub-bituminous coal. I don't have a record

of what actually was burned in the plants in the

surrounding area on a day-by-day basis. I don't know if

that's available or not, but as I understand it, there's

blends of sub-bituminous and bituminous coals burned in

the power plants in the general region.

Q. And in basing the identification of

sources on a chemical signature, does that allow you to

make a determination as to the proximity of those

sources to your sampling point?

A. No. Just the observed data by itself does

not have implicit within a distance, so what we do is

employ hybrid models or meteorological information

together with the help from these receptor models to

tell us the distance scale and the specific source

locations that were contributing to the mercury

deposition, and basically, it's taking all the available

meteorological data we can get from surface data to

Page116

upper air data to numerical models to do a very detailed

analysis of every storm event, and then match that

together with the help from the receptor models.

Q. Is that similar to the type of analysis

that was done in -- was that the type of meteorological

analysis done at Steubenville?

A. Yes.

Q. Was that using hy-split and there was one

other model that I think was referenced in the Florida

Study.

A. In the Florida Study?

Q. I'm sorry. I've read too many studies.

A. I have done too many studies.

Q. The Ram Study (phonetic)?

A. No. We did not employ rams in the

Steubenville Study.

Q. What where are the meteorological models

that you used in Steubenville?

A. We employed hy-split and we -- because of

the fact that we've been developing our own version of

the chemical model on the C-MAQ, which is the EPA

mercury model, we've been using MM-5 as the

meteorological preprocessor.

Q. And so it's the use of the meteorological

Page117

models that allows you to make a statement with respect

to the proximity of the source?

A. Yes, sir, and actually, the observed data

is much more powerful than, actually, even the models.

Q. In what way? When you say "observed data"

what are you referring to?

A. I'm talking about using NEX-RAD, surface

maps, observable detailed meteorological records from

the Great Lakes region. Most of the work that we've

been doing, which is something that I've been,

personally, doing for 25 years now, is taking observed

meteorology and using that to understand aerial issues

in various parts of the country.

Q. So the observed data you were referring to

is the observed meteorological data?

A. Data provided by the National Weather

Service, but I should mention that we also use -- we

have on-site in Steubenville our own meteorological

tower and our own set of meteorological data that we

also utilize.

Q. Why don't we go on to four, then.

MS. BASSI CONTINUES:

Q. In these signatures, in the signatures

that you were talking about that you have developed or

Page118

that have been developed over a number of years, as I

understand it, there is a different signature for

different types of industrial source categories or

non-industrial sources of categories. How can you tell

the relative contribution from each type of contributor

that has a signature?

A. That comes down to the mathematical

process that's involved in the, both, PMF and Unmix.

Are you familiar with multivarious statistical

techniques?

Q. Oh, no. You have to make this simple.

A. It comes down to -- there are a number of

elements and ratios of one element to another that

define the fingerprint for a specific source, and what

it, basically, does -- last night after I ate all these

jellybeans I think I woke up with a stomach ache from

the jellybeans, and I was trying to figure how I could

use the jellybeans to draw out this analogy for you.

And I decided I shouldn't go there.

Q. But you knew the question was coming.

A. Yeah, because it really is a good

question, and it's one that I think, as you get into the

mathematics, that you can see the beauty of it. It's

basically an ID value problem for someone who's done

Page119

physics or engineering. It, basically, is an

optimization. If you have a series of unknowns, and you

have a series of things that you know, such as the

elemental composition, you can solve those equations to

get the best solution to make all the equations true,

and in doing that, the answer, basically, is the

Ion-deductor (phonetic) or the source factor that it

calculates, and in doing that, it tells you how much

variance is explained, or how much of the data can be

explained by that one particular source, and then you

take and relate the amount of mercury that you collected

to the amount of these other trace elements in that

source, and you can get a statistical relationship, and

around that, then it gives you goodness of fitness

statistic and a whole bunch of statistics that tell you

how well your model performed. If there isn't a

relationship there, your model will give terrible

results and your goodness of fitness statistics come out

lousy, and in this case our goodness of fitness

statistics came out very good and the relationships were

very clean. And so we were very, very confident, and

very happy with the outcome of the results from the two

years worth of analysis.

DR. KEELER: I was on question 4-A I

Page120

didn't answer. "What portion of the work was performed

by Mr. Landis?" As I mentioned earlier, this was a

collaborative project with U.S. EPA Office of Research

and Development, and Dr. Landis is our project manager

on the project, and he's involved in all facets of the

research, from site set-up, to sampling, to analytical

results, to modeling. And so he's been involved in all

facets of the project. "What were your respective roles

in the Steubenville work?" The RFP was written by EPA,

so they asked far specific targeted pieces of work, but

in a cooperative agreement, the EPA cannot ask us to do

anything. It's cooperative study, so all the work that

we are doing is the work that I initiated through my

direction through the approaches that I chose to use in

this project. I'm the project director and principal

investigator on the project and involved in the same,

from site set-up, to site sampling, to analysis, to

interpretation of the data.

C: "Has he published any of his

findings regarding Steubenville in any peer-reviewed

journal?" No. The intellectual property rights belong

to the University of Michigan, and the code belongs to

the U.S. EPA. Within a cooperative agreement, again,

the principal investigator is given the first right to

Page121

publish the scientific results. However, we are joint

partners in this endeavor.

"Are these data available for public

review?" Our data -- I assume that's what you're asking.

Again, it's a normal procedure to make available

scientific results, once the peer-reviewed journal and

article comes out.

E: "To what extent did you rely on

work performed by Mr. Landis?" Again, I'm not sure to

what extent or what context you are asking the question,

but Dr. Landis is an excellent scientist and we relied

upon his work flout the project, and value his judgment

in all aspects of the project.

MR. RIESER CONTINUES:

Q. Did the two of you come to the same

conclusions regarding the Steubenville Study."

A. We worked on this project. As I mentioned

earlier, EPA began doing some of the modelings, and we

were doing some of the modeling independently, and it

was hard not to come to the same conclusion. In this

case, the importance of coal-fired utilities on the

mercury deposition in Ohio was so significant that we

did come to the same conclusions. I think as our

peer-reviewers showed us, it's difficult to come to any

Page122

other conclusions.

Q. You make the statement in your testimony

in the second paragraph of your conclusion that, on page

five, "Source-receptor studies have recently been

completed that indicate the coal-fired utilities

contributed, approximately, 70 percent of wet deposition

measured at a site in Eastern Ohio over a two-year

period, from 2003, to 2004," which I assume is this

Steubenville Study we are talking about?

A. Yes, it is.

Q. In one of the presentations that

Dr. Landis gave, didn't he say, specifically, that it

was not coal-fired utilities, but coal and fuel

combustion, and not solely electric utility generation?

A. I would have to look at the presentation

to -- I know that one of the early briefings that we are

forced to give, one of the things that occurred is that

EPA, Tim Opelt, who was the acting -- or not the acting,

assistant administrator at EPA at the time, really

pushed EPA to try to get us to get our results out

because of the impending rules, and because they thought

it was pertinent to writing the rules, and in that

process, he urged Matt to go to Washington at a very

early date when we had preliminary results and present

Page123

those results, and so one of the interesting things in

the briefings is that you will see that there are

changes from one briefing to the next, in terms of some

of the quantity of numbers, and the number of decimal

places, and other ways that things are listed in there,

and if it's one -- if it's a briefing to Tim Opelt, I

don't know if it says on the front or not.

Q. It does. It's the briefing for Tim Opelt,

April 27, 2005. This is a Powerpoint presentation for

which I do not have copy, but will by tomorrow, so it

can be introduced as an exhibit before the Board.

A. That is the first mention of our

preliminary results, so in that presentation, Dr. Landis

would have been the one who would have made those

slides.

Q. Did you work with him in putting these

slides together?

A. I would have worked with him on the

results that are included in those slides, but I did not

work next to him putting those slides together.

Q. I'm reading from one of the slides

entitled "Results" and I have just the one copy, so I

can show this to you to verify, but if I can read from

it, and again, I will have copies for the exhibit

Page124

tomorrow. "Approximately, 70 percent of the mercury wet

deposition in the Steubenville site is attributable to

local slash regional fossil fuel coal and oil combustion

sources," and then there's a bullet under this that

says, "Not entirely attributable to electrical

utilities."

A. Okay.

Q. Do you agree with that statement?

A. I think, at that time, when we had gotten

those preliminary results we had not gotten to the point

where we had done a more definitive interpretation.

Q. What had you done in the intervening time

that allowed you to change the conclusion to the one

that you have in your testimony here?

A. We have done a considerable amount of more

work, one of which was to try to understand, as Matt

said here in this presentation, where he put coal and

oil combustion sources, we actually took a much closer

look at the sources that were in the area, and the

transport and meteorological conditions to go with the

just the elemental data that we got to make sure that,

in fact, we didn't feel the oil combustion was a major

contributor. This was -- I think Matt was trying to be

extremely careful in his presentation to Tim Opelt at

Page125

that time because, literally, the results popped out of

the computer within the same couple of days as when he

had to make his presentation, and interestingly enough,

this was -- I don't know if it still says -- so briefing

for Tim Opelt. I know the version that I had seen

actually said, "Not for outside consumption" because --

due to the preliminary nature of the work, so that

somehow is not on there anymore.

Q. No, it's not, and it's available on the

U.S. EPA website.

A. Actually, is it available on U.S. EPA or

on the E-Wire website?

Q. U.S. EPA. But I'm not answering the

questions here, but that's quite fine. Is the 70 -- the

number of, approximately, 70 percent, what were the

ranges that you had for that?

(A small break was taken.)

MADAM HEARING OFFICER: Let's go back

on the record. Dr. Keeler, you were looking up some

information to answer a question from Mr. Rieser.

MR. KIM: While he's looking that up,

Exhibit No. 28, which I believe was the table that shows

stack heights Mr. Zabel had asked a question about one

column of numbers, and I believe that that should be the

Page126

correct identification for those numbers is gross load

by megawatt hours, and those figures represented the

average of the highest three yearly gross loads as

reported by U.S. EPA, CAMD, for the period of 2001,

2005. So over that five-year period, the average of the

three highest.

MR. ZABEL: That makes a great deal

more sense. Thank you.

MR. RIESER CONTINUES: Looking this up

-- could you read back the last question to Dr. Keeler?

(The previous question was read by the

court reporter.

DR. KEELER: The PMF model has

predicted a value of 70 percent a d the Unmix model

predicted 74 percent. Again, this is -- I just picked

up an earlier version, so I'm not positive that these

are the final numbers, so I'll give you the range. We

did two years' worth of measurement. The entire year of

2003 and the entire year 2004, and we did the

apportionment on the two years combined, and estimated

using the two different models, how much mercury came

from coal-fired utility boilers to the deposition of

mercury for 2003 and 2004, separately. The main reason

for doing this is that the deposition from one year to

Page127

another can vary dramatically. In 2003, the measured

mercury deposition in Steubenville was about 13

micrograms per square meter. In 2004, it was 19.8

micrograms per square meter, so there's a fairly large

difference between the two years. In 2003, the mean

contribution from coal-fired utility boilers was 9.5

with the range of 7.2 to 15.8. That's a five to 95

percent confidence interval around that mean. What has

been developed over time in these models is the ability

to propagate uncertainty through the models, and

continue to re-run the models varying the concentrations

that we got in the samples to simulate uncertainty, and

you can run this over hundreds and hundreds of times,

and then you can get a mean, plus an uncertainty range

around those numbers to give you a better sense of how

robust your solution is, so that's what that refers to,

and for the Unmix model, it calculated a mean

contribution of 9.9 with a confidence interval of five

to 15. For 2003 and for 2004, the mean contribution

from PMF was 12 with a confidence interval of nine to

20, and the Unmix model predicted a mean of 15 with a

mean of 8 to 23. So both models give, if you were to

take the means, or whatever, they gave very similar

results, in terms of percentages. And the uncertainty

Page128

in the numbers is what I gave you, so that's the

confidence interval on those.

MR. RIESER CONTINUES:

Q. Are there estimated coal-fired utility

boiler percentages associated with those numbers, as

well?

A. I don't have them in front of me, but

that's what the 70 percent was, and the mean for the two

years is 70 and 74 percent, 70 for PMF and 74 for Unmix.

You were asking me previously about the Landis briefing

where it had fossil fuel, oil, plus coal, and again, I

just wanted to make sure it was clear that that briefing

was done, literally, a couple days after the initial

analysis was done only on the 2003 data, and at that

time, the statistical robustness of the solution was

such that we weren't very confident in the solution that

we were getting because we didn't have a large enough

sample size, and hence, why we went and incorporated in

the 2004 data into this analysis, and you asked about

the question about coal-fired utility boiler

contributions, versus coal combustion I believe is what

you were referring to, correct, all coal combustion

sources, and initially, when we --

Q. I think it was just combustion sources.

Page129

A. Okay. So we were able to separate out the

oil combustion from the coal combustion very cleanly, so

there wasn't any need to have that designation in the

follow-up briefings and in our final conclusions.

Q. I think you said, with respect to the

publication, that you were looking to begin or submit

for publication around October or so of `05, but then

you had to run it threw a peer-review process?

A. That's correct.

Q. So by October of `05, you had everything

pretty well nailed down?

A. Yes, sir?

Q. And I think you --

A. You say, "About October of `05." I can't

tell you whether it was November 1, or I mean, to be

honest, it was the end of October. I know it was before

Thanksgiving and I know -- but I don't know the exact

time.

Q. Plus or minus one holiday, in other words.

I think you said, also, that you participated in an

international workshop on mercury control for coal

combustion in Beijing. Is that correct?

A. Yes.

Q. And you were there with Dr. Landis?

Page130

A. I was, yes.

Q. And Dr. Landis gave a presentation at that

regarding your findings?

A. He did, yes. Had had some other

information.

Q. Did you work with him on putting that

presentation together?

A. Again, the results that went into that

presentation were ones that we co-put together, but I

did not help him actually put the numbers on the slides

or build the tables or anything like that.

Q. But --

A. I did not review his presentation, no.

Q. You didn't review his presentation?

A. No.

Q. Did you watch his presentation?

A. I did watch his presentation, yes.

Q. Did you agree with it at the time you

watched with it?

A. Did I agree with it? Since it was a

collaborative study, I didn't -- I concurred with his

conclusions.

Q. I would like to show you what's going to

be marked as exhibit --

Page131

MADAM HEARING OFFICER: We'll mark

them in the morning. That's probably the best to wait,

until tomorrow when you actually have them.

MR. RIESER CONTINUES:

Q. What I have, for ease and quickness, is a

copy of the title page and two pages within the slides

that Dr. Landis prepared for the Beijing presentation.

In the interest of time and getting this thing out

quickly within the period of the break, again, I will

get colored copies of the whole thing, and I will bring

those tomorrow, so the whole thing can be submitted to

the Board as an exhibit, but for purposes of the

question, we only need these couple of pages, so Doctor,

if you need to look at the whole thing in order to look

at these and verify that you remember what else was in

there, I have got a copy, but if I can work with these,

that would be great to start.

A. I don't believe the presentation varied

all that much, and I certainly believe that the

conclusions and results, or whatever, did not vary from

one presentation to the other. The exact numerical

numbers might be slightly different. I know we did

refine the way we did the uncertainty analysis from one

time to another, so that they were consistent between

Page132

the PMF and Unmix, so there might be some slight

variances in the numerics, but the conclusions and the

results don't change all that much. If you look on the

first page of this three-page exhibit, and again, this

is just a portion of the whole thing, the page entitled

"Preliminary Steubenville Source Apportionment Results."

A. Yes.

Q. 2003-2004 and heading across the top is

"Measured Mercury Wet Deposition; PMF Estimated CFUB

Contributions," CFUB is identified as Coal-Fired Utility

Boiler, and "Unmix Estimated CFUB Contribution." Do you

see that?

A. Yes, I do.

Q. So there are percentages provided in the

columns for PMF estimated CFUB contribution, in addition

to absolute numbers, and those percentages, as you can

see for PMF for 2003, 73 percent; 2004 is 62 percent.

The Unmix percentage contribution are 60 percent, and

for 2003 for 2004, it's 59 percent. Do you see that?

A. Yes, I do.

Q. Are these the same numbers that are in

your report that's being presented for publication?

A. No, they are not.

Q. What way have they changed?

Page133

A. The numbers I read to you are the numbers

that --

Q. I don't recall you reading a percentage

number.

A. Because I didn't have that calculation in

front of me, but I did not -- these numbers were

preliminary data, so this presentation that Matt gave in

China was based on the preliminary results from the

April-May time frame.

Q. So as of the date of this, October 31,

when this was given, I thought you said that around that

time you had submitted this for public for peer review

and publication.

A. Just to correct what you just said, no,

that's not what I said, and I was specific in saying

that we would hope to submit the paper for publication

at the end of October, and at that time, EPA asked us to

put into a peer-review process.

Q. So you submitted it for peer review around

this time, October?

A. Yes.

Q. And was the paper complete at the time you

submitted it for peer review?

A. Yes, it was. Perhaps, you don't give

Page134

presentations, but very often you put together a

presentation on the plane as you're going over, and

sometimes you lose the materials that you have from the

previous presentation to put together the new

presentation, and I suspect that that's what Dr. Landis

did on this account. As I said before, I didn't work

with him on putting this presentation together, and so I

can't verify that's what he did. He was in the back of

the plane. I was in the front, as it should be.

Neither of us in the First Class, however, but in fact,

this was a preliminary -- as it says on there

"Preliminary Steubenville Source Apportionment."

Q. Did you make any efforts to correct it, to

correct these numbers when they were presented?

A. You don't have an opportunity to correct.

MADAM HEARING OFFICER: Excuse me,

Mr. Rieser. I apologize for interrupting, but I just

have a couple questions I would like to ask for

clarification. Dr. Keeler, you're saying the numbers --

the numbers that are in your final Steubenville Study

results are different from these, that these reflect the

preliminary numbers from the preliminary data?

DR. KEELER: Yes that's correct. Let

me just back up for a second. We were asked to rush to

Page135

do this analysis by EPA, so that they could include this

in the mercury rulemaking. Hence, we did a 2003 quick

analysis. We told them that we felt this was not a

robust solution to the question they were asking, so we

included -- we went and we got the 2004 data and

included a two-year data set, so we could provide them

with a more robust analysis. They, basically, were

after us on a daily basis to try and come and provide

these answers. In April, I went and made a presentation

of very raw and preliminary analysis. Now, at the

university, we do this all the time where someone will

get a result, present that, and everyone will critique

them and give them comments and suggestions for ways to

improve things. Because of the importance of this

issue, it was done on a federal level because it was

cooperative agreement, and so EPA was doing this behind

the scenes in their own internal labs. These

presentations that were given, many of them were from

one EPA group to another EPA group or Dr. Landis to the

assistant administrator were meant for information

purposes, and give them some update as to what was going

on. These were never intended -- and that's why the

word "preliminary" is given on these, is that these were

never intended to be final. This was preliminary work.

Page136

At this time, we hadn't even finalized the database. We

were still working very diligently to get the final data

base put together, so that the data that went into this

analysis was not even finalized at this time. The

presentations that I have been asked to talk to you

about are just that. They are preliminary results that

came along with the presentations that were given in

different venues by Dr. Landis, but these do not

represent the final numbers.

MADAM HEARING OFFICER: Thank you.

Sorry. I was a little confused on all that.

MR. RIESER CONTINUES: Mr. Kim, I'm

trying to remember if you were here for the part about

asking for the LADCO slides. Do you have Dr. Keeler's

LADCO slides?

MR. KIM: We don't have those, at

least, to the best of my knowledge. I think the person

in control of those would be Dr. Keeler, so --

DR. KEELER: As I mentioned, I will

check to see. I didn't see it on this, but if I don't

have it, I can ask somebody to E-mail that for me.

MR. RIESER CONTINUES:

Q. If we could have those for tomorrow, I

would appreciate it. Let me -- while we have this lull

Page137

here, let me go back to an item about chemical

signatures. You have certain elements that you

associate with certain types of source emissions,

correct?

A. Again, I'm sorry. I couldn't hear what

you were saying.

Q. When you do the chemical signature

analysis, you have certain elements that you associate

with certain types of emissions, correct?

A. Yes.

Q. You also identify percentages associated

with those types of sources, correct, X percent coal, Y

percent steel?

A. Yes.

Q. Now, when you are looking at the analysis

-- the initial chemical analysis just has X amount of

mercury, X amount of selenium and X amount of sulfur,

etc., correct, among other elements that you analyze?

A. For all intents and purposes, yes.

Q. How do you derive the percentage of -- I

assume there's, like, one element of mercury, X

micrograms of mercury, correct, in the sample data that

you're looking at?

A. We make measurements of every time it

Page138

rains, how much mercury deposition is found in every

single sample.

Q. How do you derive the percentages of

mercury that are associated with a given source

category?

A. The statistical method, PMF model or Unmix

model, calculates for each sample the absolute amount of

contribution from each of the samples, and calculates

how much mercury is predicted to come from that source

factor.

Q. What is the derivation of that

calculation? I mean, not the mechanics of it,

obviously, but what is the model looking for when it

makes that decision?

A. Well, it's actually a fairly complex

answer to that, but it's basically looking for

covariation or elements that are changing with time

similarly and uses that to determine what sources are

contributing to those elements.

Q. And the time element, where does that come

into the analysis?

A. Because we collect a sample every single

time it rains, so on each day, it would be a different

sample on every single day that it rains.

Page139

Q. So it's not just within each individual

sample, but all samples taken together. Is that

correct?

A. Yes, that's correct.

Q. Thank you.

DR. KEELER: Question No. 5: "In your

testimony, you reference a study you performed regarding

apportionment of mercury sources in Detroit. A: How do

you define the terms of "regional" and "local" for

purposes of this study, and B: Did you identify the

coal-fired power plants that you considered to be local

for purposes of your study?" You're asking me how I

defined "regional" and "local" in the paper that I

submitted as the Lynam and Keeler paper. Is that

correct?

Q. Correct.

MADAM HEARING OFFICER: Exhibit 27 for

the record.

DR. KEELER: The terms "local" and

"regional" often get confused. In fact, recently, a

representative from the Electric Power Research

Institute incorrectly cited my presentation at LADCO

saying that I suggested that regional transport was

continental scale transport, which I did not say. Local

Page140

transport is actually the transport that's going to

occur within one semidiennial (phonetic) cycle, or less

than one half of a day. So basically what that means is

how far pollutants can travel within about a 12-hour

period. A typical transport speed is about five meters

per second, which means that -- again, this is on

average, that if you're talking about local scale

transport being kind of within an urban area. Somewhere

up to, depending on the size of the area, but anywhere

from very close to up to 150 to 200 kilometers. In

terms of EPA modeling, they refer to local scale as less

than 10 kilometers. I mean 10 miles in distance scale,

very, very close. So that's a very different reference

than what I'm referring to. We think of the term

"local" in terms of meteorological sense that an air

mass can stay fairly consistent through the course of

one semidiennial pattern, which is, like I said,

somewhere less than 200 kilometers in scale, but will

vary tremendously, depending upon the wind speeds, so on

a given day, local transport could be as far away as

only a few kilometers on a day where there's very heavy

winds. During a wintertime period, it might be even

longer than that. Regional would then take off from

that scale from that local scale to the transport that

Page141

is, typically, controlled by synoptic meteorological

forcing, which is several days, three to five days, so

that can be up to -- maybe a couple thousand kilometers,

at most, but no more than that, so transport from Ohio

to somewhere east of the Mississippi or west of the

Mississippi, possibly, but not to the coasts. That's

not regional scale transport. That would be more

continental scale transport, so when we're referring to

Detroit local, we're really referring to Southeast

Michigan in that paper. When we're referring to local

in the Steubenville, Ohio, area we're really referring

to -- in terms of miles, if people are more comfortable

with that, 50 kilometers to 100 miles or so distance

scale.

Q. So when you use the terms "local" and

"regional" in the Detroit paper, they have the same

meaning as they will in the Steubenville paper. Is that

correct?

A. For the most extent, yes.

Q. When you use the terms "local" and

"regional" you are typically referring not to air mass

movements, but to specific things, like sources, talk

about local and regional sources?

A. We try to talk about sources being local

Page142

if they are in the vicinity of the plant, so sometimes

the distance scale will be a little bit shorter than on

a meteorological sense, so a local source might be in

Detroit, for example, the Ambassador Bridge and diesel

traffic at the Ambassador Bridge if it's only one

kilometer away or kilometer and a half away from our

site, so that would be a local source.

Q. When you describe local sources -- well,

for example, in your testimony in the conclusion on page

five -- this is the third sentence of the second

paragraph of your conclusion -- "The deposition of

mercury is heavily influenced by a few large

precipitation events that contribute significantly to

the annual deposition and these events are associated

with emissions from local slash regional sources." So

when you refer to "local slash regional sources" in this

context, what types of distances do you have in mind?

A. When I refer to "local slash regional

scale," it's often more accurate to express that scale

when referring to precipitation events because of the

nature of the disbursion of pollutants that feed into

precipitating systems, so I mean, from sources that are

very close, within a few kilometers, to those that are

out hundreds of kilometers to even as many as a thousand

Page143

kilometers away.

Q. So the term "local and regional sources"

is used in your testimony means anything, from adjacent,

to something that's a thousand kilometers away, maybe

even further. Is that correct?

A. No, closer.

Q. Closer than adjacent?

A. No, adjacent to a thousand kilometers or

closer.

Q. Or closer, so a thousand kilometers is the

outer bound.

A. Yes. That would be a good --

Q. When you use the term "synoptic" what does

that mean?

A. I was asked this yesterday, and I

immediately started to want to give a meteorological

answer, and I realized that nobody will understand that.

"Synoptic" refers to the scale in which you can see a

high and low pressure system on the map, hence the word

synoptic, as in visual. It really refers to the scale

that you see weather systems on the weather station when

you all look at the weather man, and you see a low

pressure system or the front coming through, and then

you can see a high, it, generally, takes up a region

Page144

from Minnesota to New York, so that's a synoptic scale,

several days worth of transport.

Q. Thank you. Go ahead to B, please.

MADAM HEARING OFFICER: Mr. Harley has

a follow-up.

DR. KEELER: You asked me a number of

questions, and I gave you a number of answers, and I

want to make sure that it's clear I have to be fuzzy on

the definition of distances because, when it comes to

the meteorology, it varies from day-to-day, from hour to

hour, so the transport is not like putting it on a

train, and carrying the pollutants consistently away

from one place. It's going to vary and the winds don't

travel in straight lines and the wind doesn't just blow

horizontally. It blows up into the atmosphere, so when

we talk about distance scales and local scale transport,

versus regional scale transport, versus continental

scale, we have to understand that these things are

time-varying, and there is a fuzziness to it, so I'm

trying to give you a general answer to these questions

that give you a sense of what we're looking at, but I

don't want to give you the impression that these things

are written down, and everything more than a thousand

kilometers or 1,500 is somehow a different scale because

Page145

that's not the impression I'm trying to give you.

Q. So again, looking at your testimony,

sentence -- same paragraph on page five,

"Source-receptor studies have recently been completed

that indicate the coal-fired utilities contributed to,

approximately, 70 percent of the mercury wet deposition

measured at a site in Eastern Ohio over a two-year

period. This finding is not unexpected as the

Steubenville site was selected due to its close

approximation to a number of coal-fired power plants,"

and then we go on to the sentence that I read about

local and regional sources, so these large -- the number

of large coal-fired power plants are included in these

local regional sources that could include all sources

within a hundred kilometers -- excuse me, thousand

kilometer circle of Steubenville?

A. That's correct.

MADAM HEARING OFFICER: Mr. Harley.

MR. HARLEY CONTINUES:

Q. You testified that you were able to

acquire information about -- you testified that you were

able to assess the volume, the total volume of mercury

from every precipitation event that occurred over a

two-period, from 2003 to 2004.

Page146

A. Yes.

Q. And you were able to describe that, in

different precipitation events, you would find different

concentrations of mercury and other contaminants that

were present?

A. Yes, that's correct.

Q. In your study, were you able to estimate

the total volume of mercury that was deposited on

Steubenville, Ohio, through precipitation over the

two-year period?

A. Yeah. That's the -- those were the two

numbers -- again, my computer went blank, but it was 13

point -- I don't remember what the number was.

Q. 13 micrograms per square liter (sic)?

A. Square meter. If I wasn't clear, I

apologize for my mispronunciation.

Q. 13 micrograms per square meter would have

been the total volume over a two-year period or total

mass per square area that was deposited in 2003?

A. In 2004, an additional 19 point --

whatever the number was that I gave -- micrograms per

square meter was deposited at that site.

Q. Does your study then go on to consider the

fate and transport of mercury which would be deposited

Page147

on Steubenville, Ohio, through precipitation at these

levels?

A. I'm sorry?

Q. What happens to that mercury?

A. That was not something that we had as part

of our scope of work for that project.

Q. Do you have an opinion about what happens

to that mercury?

A. The mercury that's deposited in

Steubenville?

Q. Yes.

A. Well, again, we are making measurements in

a place that's primarily grass fields, and so the

mercury that deposits there, some small fraction would

likely oxidize, or I mean, get reduced and then come

back up from the surface. The majority of the mercury

that's deposited would probably attach to organic

material in the soils and stay there at our site.

Surrounding our site is the Ohio River, so mercury that

deposits into that, obviously, will get transported down

river, and again, some of that mercury is going to

attach to particles and go to sediment in the river.

Some of that mercury will become methylated and work its

way up the ecosystem, and some fraction of that mercury

Page148

deposited in the Ohio River will evade, come back out of

the water, but those are the main things that would

happen to the mercury that gets deposited there. I

should add the one part that we haven't talked about at

all, yet, is the mercury that's dry deposited, and

that's a very important part. Some of the mercury will

actually go into the plant material, the trees the

leaves, the plants, and actually stay there, and become

part of the organic-bound mercury in these plants, which

then depending upon the deciduous trees, when the leaves

fall, that mercury is deposited to the earth's surface,

and then is there for methylation and decomposition as

the plant material decomposes, so that's an additional

source of mercury to the surface.

MR. RIESER CONTINUES:

Q. Was deposition measured in Steubenville?

A. We have been collecting ambient data

continuously. We have been using instrumentation that

gives us the reactive gaseous mercury forms, the

elemental mercury forms and the particulate mercury

forms together with meteorological measurements that we

are doing on site, and we plan on modeling the dry

deposition. We also, in doing this area, did a series

of intensive studies, which we have not completed to

Page149

date that we will be doing surrogate surfaces, and other

new techniques that we have developed to be able to

validate and compare to the models that we have

developed to do the mercury dry deposition, so at this

point, I don't have solid estimates for dry deposition,

but it's something that we will have at the conclusion

of this study.

DR. KEELER: Question 5-B: "Did you

identify coal-fired power plants that you considered to

be local for the purposes of your study?"

MADAM HEARING OFFICER: Again, for

purposes of the record, we are back to talking about the

Detroit Study.

DR. KEELER: We are talking about the

paper that was submitted to the Board Lynam and Keeler

paper. We have done a considerable amount of work in

Southeast Michigan over the past 16 years on mercury

deposition and sources, and there are actually only a

handful of actual coal-fired utilities that are in

Southeast Michigan, and these include mineral power

plant, which is on Lake Erie, and there are two or three

others that are actually even closer to our site, and

unfortunately, I'm having a post-afternoon mental lapse

here, but there are -- there's only a hand full of

Page150

sources that we consider local, and again, those are

fairly close, certainly within 50 miles of our site in

Detroit.

Question 5-C: "Do you know what type

of coal was utilized in these power plants during the

duration of your study?" I'm not aware of data that

provides the actual type of mercury, type of coal that

was burned by these power plants on a day-by-day basis,

and so I don't have that information, but I do have, in

the Michigan Mercury Utility Work Report, which I think

is cited somewhere in these documents there is a table

listing which plants burned bituminous, which blended

bituminous and sub-bituminous, which I believe is most

of the plants that were in that five or six that I

mentioned were blenders. They used both, but then a few

only used sub-bituminous, so that data is available in

that report, so I am aware of it but I don't have the

day-by-day blending and how much was used of each type

and so forth.

Question D: "Do you know what type of

emissions controls these plants use?" Again, Table E to

that Michigan Mercury Electric Work Group provides the

mercury -- or the emissions control that's used on each

one of those plants. I believe that was utility

Page151

provided to the work group.

MR. RIESER CONTINUES:

Q. With the type -- and I think we probably

talked about this, but I have the same late afternoon

thing you have got going on. Would the type of emission

control and the type of coal used affect the chemical

signature that you rely on?

A. I believe that the signature that we look

at can be influenced by the type of control. An example

would be if there was no type of particulate control on

the source, that would have a profound effect on what

came out, in terms of the major trace elements, which

most of those are going to be in the particulate form.

If there's some type of a scrubber, a wet scrubber, of

course, that's going to remove a large degree of the

soluble gases, so yes, those things will have an effect

on the signatures. The interesting thing is that when

you look at the signatures that have been collected over

the years, you can see variability in certain elements,

and you try to not use those to define your sources, but

they do introduce some uncertainty in your profiles and

that's one of the reasons why these new models are able

to propagate through the uncertainty in the signatures

to give you that sense that, okay, what if you're off 10

Page152

percent in this element and this sample because of

something like a control or because of some other

chemical transformation?

Q. Is it your expectation that these power

plants, at the time you did this study, had no

particulate control?

A. Oh, no. I'm not aware of other plants out

there --

Q. I hope not.

A. Okay, so, sorry. I asked you a question

again. I apologize. No. That's not my expectation. I

was giving you an example.

Q. Well, the follow-up, if there are, send

them my number.

DR. KEELER: E: "What other sources

of mercury emissions did you identify as being local

with respect to the study site?" Detroit is very

similar to Southeast Chicago in that it has a very high

density of motor vehicle traffic, iron and steel

production. There is a municipal waste incinerator

about 10 kilometers away from our site. There's oil

refining and chemical manufacturing, together with a

very large sludge incinerator. Those are the major

sources that we identified in that area.

Page153

F: "What was your basis for

determining that RGM you identified was a result of the

local emissions from coal-fired power plants?" The

answer I wrote to myself is I'm not sure what you're

referring to. If you could maybe point to what you're

asking me, I could maybe answer the question.

MR. RIESER CONTINUES:

Q. Sure. On the top of page 3153 in the

left-hand column.

MR. KIM: Which document?

MR. RIESER CONTINUES:

Q. I'm sorry. We are looking at the Detroit

paper. I have another copy, if you need it. I'm

looking at page 3153, on the top. It's in the

paragraph, "The presence of sulfur dioxide and RGM." Do

you see that?

A. Yes.

Q. The third sentence -- excuse me -- fourth

sentence says, "Since, both, RGM and sulfur dioxide are

primary emissions, we conclude that this factor

describes local emissions of RGM from coal-fired utility

proximate to the monitoring site in Detroit."

A. So you are asking me what the basis is for

that conclusion?

Page154

Q. Yes.

A. We have five-minute average SO2

concentrations, as well as concentrations of NOx, CO,

fine particulates, five-minute concentrations of

developmental mercury. Hourly concentrations of

reactive gas and reactive and particulate mercury

together with the onsite meteorological information,

plus all the meteorological information that we

collected from all the sites from the National Weather

Service, and during times when we got transport from the

direction of a couple of the larger coal-fired

utilities, this is when we saw this SO2-RGM

relationship, and it also turned out when we did this

analysis that that factor identified the periods when we

had that type of specific flow from that source, so it

was done through a data analysis of looking at when the

spikes of these things occurred together. The Monroe

power plant is to the south, southeast of the site.

It's one of the larger facilities, and we could see that

on occasion at our monitoring site and this was very

different than when we saw --

Q. I'm sorry. You could see that --

A. In the data, and this was different when

one looked at the data when we got flow from the island

Page155

where the industrial source was coming from the coke

oven and iron and steel facility and the sewage sludge

incinerator and refineries. It had a very different

characteristic to it, so this confirmed in our mind that

this analysis, which was done independently, was telling

us that this was coming from coal-fired emissions.

Q. Not only coal-fired emissions, but

coal-fired emissions from a certain facility?

A. Again, if I didn't put a facility here, I

guess it was not one facility. It could have been from

more than one because I think that was more than one

episode that we are talking about in that paragraph.

Q. So local emissions in that sentence I

assume also means this 50 kilometer range that we have

been talking about?

A. That's right, during one day.

Q. And the approximate also refers to that

50-kilometer distance?

A. Where it's approximate.

Q. It says, "Describes local emissions of RGM

from coal-fired facility proximate to monitoring site."

A. That's right, approximate.

Q. Thank you.

DR. KEELER: Question No. 6: "Is it

Page156

correct that the Steubenville study is designed to be a

four-year study and completed in 2006?" No. The study

was initially designed and put out on the street as an

RFP to be a two-year study. The study has been -- was

extended to become a three-year study because of the

additional equipment that was given to us, so we

requested an additional year of measurement and had

been, since then, extended one more year because of the

importance of this site, in terms of the results that we

were seeing, so now the study is intended to --

measurement collection is going to run through 2006.

We'll get done collecting data at the end of 2006, and

there will probably be a subsequent year of analyzing

the samples, doing the analysis and doing the modeling

that goes along with that. So originally, it was a

two-year study.

MR. RIESER CONTINUES:

Q. When do you -- this is sort of Question B,

but we might as well get to it. When do you expect the

2005 data to be available for review?

A. That is Question B, "Is the data for 2005

available for review?" The 2005 database is being built

at this time. We are working on the last two quarters

of the year. It takes quite a while to analyze all

Page157

these different samples because it's not just

precipitation aerosol samples we are collecting, as well

as the other gaseous data and meteorological data. All

that data has to be processed and put together, so we

are hoping that the 2005 data will be completed shortly.

What does that mean? It's June. We are going to be in

the field for the next two months. It's likely not to

happen, until the fall, so October-November time frame.

Q. Is that going to have to go through a peer

review process, as well, or subject to publication, or

is that something you expect to be released to the

public?

A. Our intentions, at this point, are to put

all of the data together into a complete analysis of the

entire data and submit that for per review. My guess is

it will probably go through another exhaustive peer

review internally and through EPA, but I don't know that

for a fact, so it will be released after that, that

process.

Q. Thank you.

DR. KEELER: "Is it accurate that the

information you present in your testimony is based upon

the first two years study 2003, 2004?" The answer is

yes.

Page158

Question 7: "You indicated that you

served on a Michigan Electric Utility Work Group, which

studied ways to reduce emissions from coal-fired power

plants. A: Did you participate in the final draft of

the report dated June 20, 2005?" The answer is yes.

B: "Did you participate in the

drafting of the chapter on mercury emissions and

deposition, which is chapter 3.3 on page 50 of the

report?" The answer is yes.

C: "Do you agree with the statement

on page 50 that, `The concern over mercury in the

environment stems from its eventual deposition at the

earth's surface and subsequent conversion to methylated

mercury'"? The answer is yes.

D: "Do you agree with the statement

on page 56 that `oxidized mercury, or Hg0, that is

deposited on the surface of the Great Lakes would not

likely enter the reaction pathway that would lead to the

production of methylmercury in the lakes; although

tributaries and surrounding wetlands would support

methylation activities?" I did not write that sentence.

For the most part, it's correct, except I believe that

there was some type of mistake made in that. I'm not

sure that they were correct in their oxidized mercury or

Page159

Hg0. I don't know if they meant -- Hg0 is not oxidized

mercury, so the statement is factually incorrect as it's

written, but the intent of the statement is, for the

most part, correct that mercury in, both, the elemental

form and in the reactive form that deposits to the Great

Lakes water body, itself, is -- some of that mercury is

going to evade and come back out, and I think that's the

intent of the sentence that was in that report.

MR. RIESER CONTINUES:

Q. Also, the reaction to the methylation

reaction pathway is not commonly active in the Great

Lakes?

A. Yeah. As we discussed earlier, the most

of the Great Lakes are oxygenated fairly well down to

the -- so there's no methylation that's occurring in

that oxygenating water. The big lakes, themselves, are

not where the methylation is occurring. The methylation

is occurring when that precipitation and pollution

deposition hits the forested areas around the Great

Lakes and the wetlands, and then the rain washes that

pollution off, or the mercury that's in the rain. Then

it runs off and goes into the wetlands where it then can

subsequently undergo the methylation, and then that runs

off into the tributaries, and feeds into the shoreline

Page160

of the lakes, but the question is correct. It doesn't

happen in the Big Lakes in open water.

E: "Why does not the report discuss

your work at Steubenville?" As I mentioned before, the

work in Steubenville, we were forced to rush and do some

of this analysis in the spring of 2005. The mercury

work group that I was a part of had started quite a bit

earlier than that and had been going on for a long time,

and although they were very anxious for us to get this

into peer review, and get it out in the literature,

there was an understanding that we weren't going to

include non-peer-reviewed publications in that report

just as a way to be fair across the table, and so that's

why it was not included. It was not excluded since I

was a member of that work group, I did not exclude my

own work because I didn't feel as if it was a good

enough quality. It was just that we agreed to these

rules, and I lived with them.

F: "Do you agree with the statement

on page 58 that the results of the Wisconsin Utility

Case Study performed by -- I'm sorry I'm not going to

butcher your name -- spell it,

V-I-J-A-Y-A-R-A-G-H-A-V-E-N, et al., indicated that, on

an annual basis, coal-fired utility boilers in Wisconsin

Page161

contributed, approximately, one to four percent of the

mercury being deposited via precipitation near Wisconsin

MDN stations. Do I agree with that statement? I have

no basis to agree or disagree with the statement. I'm

aware of the work and the report that was done for the

Wisconsin utilities. And presented to the mercury work

group. I have no reason to doubt the validity of their

work, and I would suggest that in the case of Wisconsin

utilities seeing that most of their utilities their

large utilities are located on the east side of the

lake, most of their emissions would tend to belong to

the east, and would not be deposited in the

precipitation, so it's not inconsistent with my thought,

in terms of where precipitation comes from and goes to,

in terms of delivering pollutants to the state of

Wisconsin, so that's my answer.

MADAM HEARING OFFICER: Point of

clarification, "MDN station" is that monitoring

stations?

DR. KEELER: MDN refers to the mercury

deposition network.

MR. RIESER CONTINUES:

Q. There are power plants other than those

along the eastern coast of coal-fired burning power

Page162

plants in Wisconsin, other than those along the eastern

shore of Lake Michigan, are they not, western shore?

A. Yes. I think there are a couple that are

also on the southeast side I guess or southwest side,

excuse me.

DR. KEELER: G: "Do you agree with

the statement on pages 60 to 61 that, `the local impact,

or potential hot spot, is likely overestimated by

Regional 3-D Eulerian models"? I would say, no, I don't

agree with that. It really is largely a function of the

model, in the parameterizations of that model, so I

would tend not to agree with that generalization.

Q. So when you say it depends on the model,

are there specific models that you think are likely to

overestimate local impact?

A. Again, I would have to be intimately

familiar with the model in order for me to give you an

example of one. It would depend on the picture of

emission sources that you were looking at and how the

model parameters some of its scale processes, and so

forth, in order for me to generalize that question.

Q. For example, the C-MAQ that's utilized by

U.S. EPA.

A. If one is asking me the question do I see

Page163

instances in modeling output that suggests the C-MAQ

overestimates deposition in some places, I would say

that, from my experience, that I see C-MAQ

underestimating deposition in areas where we have

measurements where there's a great deal of mercury, for

example, and other places where it overpredicts based on

the measurements, so again, I think that's a very

difficult generalization to make, and one I would not

agree with. I don't think there's a hard and fast rule

that you could say that's a correct statement.

Q. And do they also be sort of gets into H,

that -- would you agree with the statement it's likely

to be overestimated by the Team model -- T-E-A-M?

A. I am familiar with the Team model through

the publications that have been written. I am not,

intimately, with the parameterizations in the Team model

and how they handle some of those things, so I wouldn't

be able to answer, but H: I do agree that C-MAQ and

Team are both examples of 3-D Eulerian models, so the

answer to that is yes.

Q. Do you think that the C-MAQ and Team

models are useful for looking at the issue of mercury

deposition?

A. I'm, both, a modeler and a measurements

Page164

person. When one can model a physical process using

mathematics and equations, one can confirm that they

understand what controls the process. That's what the

basis for engineering is. If one can model something,

one learns and gains great insights to the physical and

chemical processes that are underlining the true

physical realities that occur in the environment. As

someone who does, both, measurement and models, I am

equally what would you say -- I look at both of them

with a sense of knowing that both are very uncertain,

and that both can be used to collaborate the other.

Models are imperfect, and models, especially for

something as complex as mercury deposition, are

something that, at this point in time, are not to the

point where I feel comfortable with using them to --

because I don't feel that they do behave and describe

the phenomenon that we see in the environment based on

our measurements, so models are useful tools, but what

you put into the model, the input parameters, how the

physical parameterizations are done in the models will

have a great impact on the quality of the output that

the model has and not all models are created equal. I

think that the models that have been published in the

peer-reviewed literature have come a long way. I think

Page165

that the state of the sciences should be applauded where

we are at at this point, but we have a long ways to go,

and it's not the modelers problem. It's the other half

of me that's trying to describe the physical and

chemical reactions that are needed in the models. If I

give them the wrong model reaction rates, then they

can't model the process properly. If I don't give them

the proper speciation in the emissions data, they cannot

predict the appropriate deposition downwind. I think

the models in terms of their description of meteorology

have gotten pretty sophisticated, but the fact remains

that it's still a struggle with actually predicting

where precipitation falls, if you look at the way that

these models predict each event that we sample, so if we

compare an event estimate from C-MAQ, or from some other

model, we do a poor job because we can't get the

precipitation rate amount or the rate correct. We don't

get the precipitation falling in the correct location.

These are problems that all models have that

parameterize meteorological processes.

Q. Would the same uncertainty be associated

with a source-receptor model?

A. All these are source-receptor models.

Source-receptor models are, basically, taking some

Page166

attribute at a receptor and providing a relationship,

how much comes from this source winds up with that

receptor, so whether you start at the source, or you

start at the receptor, they are both receptor models, so

are you asking me are these the same uncertainties in

receptor models?

Q. I'm asking you whether there are the same

uncertainties in the source-receptor model that you

identified yourself as performing.

A. So the source receptor-modeling that I

refer to and that was utilized to come to our

conclusions on the portion of coal-fired utilities at

Steubenville are receptor models. They are starting

from observations at the receptor, and they work

backwards, and they do not rely upon understanding the

chemical processes that go in the atmosphere. They

don't require that you understand the species, exact

speciation that occurs at the power plant. They only

require that you have a detailed characterization of the

measurements at that site, and that's one of the

advantages. That's why receptor modeling started

because in the early days there were measurements made

of particulates in the atmosphere, and we didn't know

where that particulates were coming from, and we were

Page167

able to identify sources that people didn't even know

were there based on using these receptor techniques and

working backwards and identifying a source of that

particular pollution, so no, they don't have -- they

don't suffer from the same uncertainties as the project

models would.

Q. My recollection is that you testified that

the source-receptor -- excuse me -- the receptor models

identify are categories of sources, correct?

A. I don't understand that question.

Q. What the receptor models identify are --

is not a specific source, a specific smokestack, for

example, but a category of sources, coal-fired power

plants?

A. Correct.

Q. So in order to locate the specific source

and identify it's proximity to the receptor, you have to

add another step, which involves meteorology, correct?

A. Meteorological modeling, correct.

Q. Is the meteorological modeling subject to

the same uncertainties that you have just described?

A. This is the beauty in this analysis that

we depend largely on observations in that type of

analysis. We do use a model, like the hy-split model

Page168

that gives you where the pollution came from. You can

start at a receptor and it will tell you where the air

came from that arrived at that point in space, and there

are uncertainties in that calculation, but we include

that in our modeling. We take into account that there's

an uncertainty in the upwind pattern. That's something,

again, that many of the other researchers that look at

this do not include that. They assume that the

trajectory is a perfect calculation, but we look at it

as a problemistic function, that the trajectory

represents the highest probability that pollution would

follow that path and propagate a certainty about that

analysis.

Q. So the uncertainty associated with the

meteorologic models that you use in your receptor

studies is less than the uncertainty inherent in the

C-MAQ or Team modeling?

A. It is because it's not part of our results

or our conclusion. When we get a result, and this is

why, when you keep asking me about the distance scales,

what's local versus regional and so forth, is that we

don't overinterpret what we're doing, and the scale that

we ascribe to that the sources could have been in is a

little bit larger than what we would like. We would

Page169

like to be able to pinpoint it to a spot, but because

there's uncertainty in these calculations, it becomes a

little bit blurry, but what's nice is, at the end of the

result, we can pinpoint an area like Southwestern Ohio

so something the size of an area of a state is the kind

of the smallest geographical region that we can

identify.

Q. When you use the meteorologic models that

you described in your receptor modeling, how far back in

time or back in distance do they go?

A. That varies from study to study and what

we're looking at. As one goes farther and father

backwards in time, the calculation has greater and

greater uncertainty, and one of the things I did as a

Greenhorn graduate student over 20 years ago was to

actually look at trajectories that went back five and

seven days compared to those that went back three days,

and look at our ability to predict where the air came

from using five-day trajectories, versus three days,

versus 24 hours, and what we found was -- and this is

actually work we did under the utility acid

precipitation project filed by my advisor, Perry Samson,

at the University of Michigan who was the principal

investigator on that. We actually used the three-day

Page170

back trajectories, and found they gave us very, very

good information describing the synoptic meteorological

conditions pertaining to air mass transport, so through

a set of analysis and meteorological tests and so forth,

we found that a three-day was the most reliable,

although understanding that, of course, air is

transported much farther than that, and again, if you

were looking at carbon monoxide, or if you are looking

at a stable interpollutant that doesn't have any

transformations or deposition, then one might select a

different approach, but since we are looking at

precipitating systems, the choice of a three-day

trajectory or five-day trajectory doesn't strongly

influence our analysis.

Q. What are your assumptions with respect to

the amount of mercury in the air, furthest back extent

of that meteorologic model?

A. We don't make any assumptions about that.

There is no inherent assumption about the amount of

pollutants going back along the trajectory.

Q. Then going back -- what are you measuring

back along the trajectory? The air movement?

A. Yeah. That's only telling us about the

meteorological conditions that occurred upwind.

Page171

Q. There's no factor for looking at where --

we made these discussions about local and regional

sources, which have a radius of about one thousand

kilometers, about?

A. Yeah.

Q. How do you know that the mercury that you

assume is from a local-regional source within this

thousand kilometer radius circle is from within that

circle and not going outside that circle?

A. What we will do is we will actually look

at the rainfall patterns that occurred -- we will take

the meteorological data and look at it. It's kind of a

the best way to describe it is you take the 30-minute

meteorological data and plot it out on the piece of

paper so that the maps that you see and then put it in

motion, so you watching it as it's moving through time,

and you can actually look and see where the air mass

came from, and where it precipitated along the path

where the air mass came, and you can look and see where

the air was flowing into the storms, and where that

storm or that cloud cell actually moved, and then

precipitated, and so we do this for each storm that we

have data for, and we can work backwards along those to

a point where we are confident that we have captured

Page172

where that storm started, where the air mass came from

and eventually precipitated and aided by NEX-RAD data,

together with all the surface meteorological

measurements, and this is a very tedious analysis, and

it takes a lot of time, as you can well imagine. If

you've ever done a meteorological analysis, we still do

things the old way where we take out maps and we plot

out things by hand and actually put these things

together, together with the data that we are collecting,

so we can look at the variability, really, as a

snapshot. Sometimes we will look at the NEX-RAD data in

five-minute intervals and play it like a movie and play

it back and look at these snapshots, and in doing that

because reactive mercury is a form that's most likely to

go into those precipitating systems, together with

particulate mercury, we can then know that those

emissions had to have come from a point in time in space

upwind where they could have been removed at our site,

so elemental mercury that was coming from China, for

example, takes a very long time to oxidize in the

atmosphere, and it can float around and float around and

float around, but it takes a long time and converts at a

very low rate. If we have a lot of mercury in our

sample, it's almost impossible for a mercury that was

Page173

emitted, say, use the example of California in the

elemental form to have it get to enough where it's going

to explain the amount of mercury we found in our

deposition sample in Ohio, and so we have an

understanding of the chemical conversion, and an

understanding of the types of mercury that are emitted,

and you can only explain that by looking at sources

within a certain vicinity that emit the types of mercury

that would go into the solution, and go into clouded

water. Of course, I'm oversimplifying this to make sure

everyone understands what I'm saying, which I know is

fairly difficult, but still, that's the basic

taking-apart and dissecting, doing a CSI case on a

precipitation event. We are taking all the information

that we can and dissecting it, knowing what we have by

looking at it over 20 years worth of precipitation data

for other species, like sulfate, nitrate, precipitation,

and using that to then determine the radius from which

these sources could come, and then having over 200

precipitation events, you put this ensemble of events,

this weight of evidence together, and then paint a

picture on the region or area that that mercury could be

coming from.

Q. Would some portion of the mercury that you

Page174

observed at the receptor site come from these more

distant sources, or is there an assumption that none of

it comes from those distant sources?

A. No. We have always stated, although

people like to pin you into a box, and say that you

don't believe there's any global transport. I've never

said that. It's, in fact, we see some 20 percent could

be coming from very distant sources beyond the regional

scale, so there is going to be global transport. If we

are able to knock down emissions to a point where let's

just say this rule went into effect, and we were able to

take 90 percent of emissions out of coal-fired

utilities, that 20 percent signal would become an

important one to be worrying about, but that signal is

varying over time, but you can see that 20 percent

signal if you look at the remote sites to the west where

there are no coal-fired utilities. You can look west of

the Mississippi, and look to see that the deposition

that they incur out there is typically less than 4

micrograms per square meter, where we are getting 20 in

Steubenville, so you do see that global signal, and it's

real, and it's something we are looking on, in terms of

getting other nations -- China, India, and other places

-- to be concerned about their mercury emissions, as

Page175

well.

Q. When you use the term "value 20 percent,"

that's 20 percent of --

A. The observed mercury deposition that we

see.

Q. The observed mercury deposition. Does

your model differentiate between coal-fired utility

mercury at different distances within that one thousand

kilometer radius circle, say, 1,500?

A. Are you asking me about, again, individual

plants?

Q. Yes.

A. No. Because of the uncertainty in the

meteorological analysis, we can't differentiate between

a plant that's located 15 kilometers from another one in

that sphere, so the answer is no.

MR. AYRES CONTINUES:

Q. Could I jump in with one question at this

point? Mr. Rieser has used this thousand kilometer

radius several times, and I just wanted to ask the

witness if that is the outer limit of what you call

"region"?

A. Yes. I think that's a good way to think

about it. I think that's really, when we are looking at

Page176

regional, that thousand kilometers is the outer region.

Q. Would it be likely to be smaller

throughout?

A. In most of the large precipitation events

that we see at Steubenville, it's actually much smaller.

It's less than 50 kilometers. One of the questions that

you asked me earlier and I thought you were going to ask

me the same question because I did answer it before was

that you asked me would we -- if, in fact, the mercury

that was the mercury that we see in the wet deposition

could we explain it by emissions of all elemental

mercury. If so, if the local power plants were only

emitting elemental mercury, could we explain our

deposition pattern? The answer is no. So, if all the

power plants only put out elemental mercury in the

entire region, then we could not explain our data, so

that's fairly important, so it really confirms the fact

that coal combustion sources are putting out a

significant amount of mercury in that reactive,

particulate form, and that's consistent with our

understanding and thought processes in the atmospheric

chemistry.

MR. RIESER: Would you read the last

part of that answer back, please?

Page177

(At which point, the previous answer

was read by the court reporter.)

MR. RIESER CONTINUES:

Q. That it's some portion of RGM and some

portion is elemental.

A. It's not an according to the model at all.

It's a confirming a point, in terms of our interpreting

our data.

Q. In what way?

A. Because our understanding of the chemistry

of elemental mercury is such that it would not see the

large amounts of mercury in the deposition that we see

based on just elemental mercury being the form that

would be in the precipitating systems that deposit the

mercury to Steubenville.

Q. In answer to Mr. Ayres' question I think

you changed something that you said before with respect

to your testimony, so I just need to confirm it. Right

at the end of your testimony, you talked about the

deposition of mercury having influence by large

precipitation and these events are associated with

emissions from local and regional sources, and we had a

long discussion about how those are defined, and you

define them in the meteorologic sense, and it was my

Page178

understanding that the combination of local and regional

implied the distance that had been using because it was

your phrase a thousand kilometers away, and since it can

come from anywhere, I'm thinking of a circle of a

thousand kilometer radius.

A. That's correct.

Q. So are you changing that definition of how

you're using "local" and "regional" in that sentence of

your testimony?

A. No. I gave a sub-answer. The answer to

that in the testimony is still correct. What I said was

that, for a number of the largest events, we determined

that they were coming from less than 50 kilometers away.

I said -- I did not make the general statement that all

of them were coming from less than 50 kilometers away,

so my general statement in my testimony is still

correct.

Q. And how are you able to determine that

they were less than 50 kilometers away?

A. Again, based on meteorological conditions

in those situations, meteorological conditions was such

that the transport was very slow, and storms did not

move very quickly, and so the spacial extent was very,

very isolated.

Page179

MR. HARLEY: Madam Hearing Officer,

out of deference to my peers who are actually being paid

to be here, we've elected to sit at the second table,

but I do want to point out that we have appeared at

these proceedings, and would be very grateful, even

though we are sitting at the second table, if we could

get copies of exhibits when they are passed around.

MADAM HEARING OFFICER: And I

apologize, Mr. Harley. I have not been looking up when

that's happening, and we'll see to it that you get

exhibits in the future. Keel.

MADAM HEARING OFFICER: Also, if we do

have problems getting copies of things, the EPA I know

has a copier here, as do we. We can get copies.

MR. KIM: We will make sure we have

copies for, at the very least, Mr. Harley.

MADAM HEARING OFFICER: With that, I

think we are ready to begin with the questioning again.

Are we ready for 7-H?

DR. KEELER: I answered 7-H.

MR. RIESER CONTINUES:

Q. I just have a couple. Is it your

understanding or belief from the data that you got that

this Steubenville the utilities that, say, within 50

Page180

kilometers of Steubenville are putting out -- are

emitting large amounts of RGM?

A. So you combined a number of statements

that I made into a new statement, which doesn't really

express what I said in the several independent

statements. A, the 50 kilometers was pertaining to a

different point, referring to a couple of the largest

precipitation events, and the second part of your

combined sentence was do I feel like there's a large

amount of RGM coming out of the power plants. That was

another statement in which I said we couldn't -- we

couldn't explain the large amount of mercury deposition

we see in our precipitation events based on local and

regional power facilities putting out all elements of

mercury that they had to have been putting out reactive

mercury for us to see that. I can't quantify the amount

of reactive mercury, but I did say that it was a

substantial amount of their emissions had to have been

in the reactive mercury form, so just to clarify, so I

didn't put those things together.

Q. Following up on that, you talked about a

20 percent global mercury I think was the phrase. Is

that correct?

A. 20 percent that came from --

Page181

Q. Global sources?

A. From distances greater than the regional

scale, let's say.

Q. How did you arrive at that number?

A. Well, A, that was, approximately, the

difference in my -- it really is a difference, if you

begin looking at the amount of deposition at sites that

are west of the Mississippi, and at periods where we get

what I would call -- I call them clean air sectors or

transports from directions where there aren't a lot of

sources. We get fairly low deposition amounts and

because we don't see strong tracer signals from those

precipitation samples and we get transport from, say,

the northwest out of Canada, but we still see some

mercury. It appears to me that that's something that

could be from oxidation of mercury that was transported

over long distances, so the absence of strong tracers in

this small amount of mercury almost always is visible in

this very small mercury events, and that's where I kind

of have a sense for 20.

Again, I didn't mean to be

quantitative that the 20 percent is our estimate at

Steubenville. That's not what I said. I said there

could be something on the order of 20 percent

Page182

contributing to a background over the Great Lakes

region. But again, the importance of that background is

going to vary over a time period. Clearly, that's a

small signal in Steubenville compared to what we see

from the local coal-fired utilities compared to the

local and regional utilities.

MADAM HEARING OFFICER: You need to

identify yourself.

MR. RIESER: Let's go on to eight,

please.

MADAM HEARING OFFICER: Mr. Zabel and

then Mr. Harley.

MR. ZABEL CONTINUES:

Q. Before we go on to eight, Dr. Keeler, I

can follow this at all. You have a monitoring station

at which you take a sample. You analyze that sample for

mercury. I'm talking about Steubenville now.

A. The precipitation source assessment?

Q. Yes.

A. Yes. We collect precipitation samples,

and we analyze those samples for mercury, a suite

(phonetic) of trace elements and major ions.

Q. Just somewhat of an aside, is that total

mercury?

Page183

A. It is total mercury.

Q. And you analyze it, as you said, for the

trace elements that, I assume, have been fingerprinted

to specific source types. Is that correct?

A. I'm sorry. I don't understand your

question.

Q. Earlier you talked about fingerprinting

specific source types.

MADAM HEARING OFFICER: I think he

used the word "signature."

DR. KEELER: That's fine, source

signature, same thing.

Q. Specific source types by trace elements.

Is that correct?

A. That's correct.

Q. So you now analyze the sample for the

mercury content and the trace element content. Is it

through that trace element and your model that you

allocate the quantity of mercury among those source

types?

A. It's all of the chemical composition, the

major ions, the elemental composition, and the mercury,

and the PMF, or unmixed models that deconvolute this

sample's contributions back to the sources that

Page184

contributed, yes.

Q. So whatever algorithm you've got built

into that model will say X percentage of that came from

coal-fired power plants. Is that correct?

A. It identifies the source which then we use

our understanding of the chemical signatures to identify

that source, yes.

Q. What do you do with that global portion

that you mentioned a moment ago?

A. Well, the global portion is usually in the

unexplained category.

Q. How big is the unexplained category?

A. In this case, that's where I got the 20

percent factor.

Q. I guess what -- and maybe this is built

into the trace element analysis is you've got an

analysis of all those other elements to a apportion the

mercury. Why don't you apportion 100 percent of it?

A. You can only apportion the amount of

mercury that has a relationship with these other

elements, so if there are -- so if the case -- if, in

fact, it's a mercury form that was chemically

transformed in the atmosphere, just been floating around

and other trace elements which were there were removed,

Page185

so it's just this gaseous elemental mercury that's

floating around the earth, and then it chemically gets

transformed because it gets removed, it would not have

the same tracer to go with it, so therefore, it would

not be able to -- the model would not statistically be

able to separate it out as a new factor, so basically,

it gets in the unexplained category.

Q. I guess I'm having trouble getting around

the unexplained category, Doctor.

A. If you think about in terms of a

regression analysis, when you do a regression, and you

can explain your R squared is .86. it tells you that 14

percent of the variance in your data is unexplained.

That's what I'm trying to say.

Q. Now I understand. Thank you.

MR. HARLEY CONTINUES:

Q. Based on your testimony, I feel very bad

for Steubenville, Ohio. I'm not sure. That's my

question. In your opinion, is there any reason to

believe there is a disproportionate or preferential

mercury deposition on Steubenville, Ohio, by comparison

to other similarly situated towns?

A. No. There is nothing unique about

Steubenville, Ohio, that would make it stand out or be

Page186

unique, in terms of it receiving some unusual amount of

mercury deposition. The quantity of deposition that's

received there is really a function of its geographical

location in the United States. The fact that the

precipitation and the storms that bring the

precipitation have an orientation that go from the

south, southwest, up to the Steubenville location, and

then right now, it's all of those factors taken

together. The high density of emissions, together with

the way that the storm tracks -- storm tracks follow

that makes Steubenville a place that receives high

mercury deposition. There's nothing in particular or

special about it.

MADAM HEARING OFFICER: Ready to move

on to question eight? Mr. Bonebrake.

MR. BONEBRAKE CONTINUES:

Q. One follow-up. The 20 percent unexplained

portion that you were just discussing with Mr. Zabel, is

that all the positive is RGM?

A. I'm sorry. I must not be making myself

clear, then. The 20 percent we were talking about was

in reference to precipitation deposition. We were

talking about an unexplained amount of mercury that was

in the deposition that we can't account, in terms of

Page187

saying, "It came from this source or that source."

Partly, that's because of the uncertainty in the

calculations and everything else, but it doesn't

really -- I can't say that it's related to RGM or that

it's in an RGM form, or anything like that, so I think I

confused you somehow.

Q. So but that 20 percent portion you don't

know of what type of mercury is comprised?

A. That's total mercury assessment, so the

total amount of mercury that's in the precipitation

sample 80 percent of it we can explain and 20 percent of

it we can't.

Q. I understand that what I'm --

A. I mean, most of it is going to be in the

oxidized form. When we do that type of analysis where

we speciate the mercury from just a total mercury to

whether it's particulate or dissolved or reactive we

wind up getting a predominantly larger amount of

oxidized mercury in that sample. It will be 70 to 80

percent oxidized mercury. Some of it is particulate,

but we don't put much credence in that because, in order

to really get at the specific numbers -- and that's not

in the paper, and has no bearing on the calculations.

It doesn't come into account at all. It's just that

Page188

sample is not taken in the way to be able to look at

that number, specifically, and it's not important to us,

but it does wind up being in an oxidized mercury form in

the rain, so question 8 --

MR. RIESER CONTINUES:

Q. Just as a follow-up on Mr. Harley's

question, this is question is asked later, but I think

it makes sense to ask it here.

MADAM HEARING OFFICER: Could you

identify what question it is later, so we can mark it

off the list, please.

MR. RIESER: This would be 13 A and B,

"How many coal-fired power generating units are located

within 50 miles of Steubenville and what is the combined

capacity in megawatts of these units?" These are the

questions.

DR. KEELER: Again, the information I

have available to me, or whatever, is the information I

would have to go to the website and pull it off from EPA

from the 1999 ICR. I mean, I have got that information

back in my lab, but -- it wasn't pertinent, in terms of

defining my conclusions, or whatever. We used that

information, and we have -- I can provide you with a

map, if you would like, that shows the sources as

Page189

provided by EPA, but the megawatt capacity and so forth,

I don't have that.

MR. RIESER CONTINUES:

Q. It wasn't important to your study?

A. It wasn't important in this -- obviously,

it is important, in terms of those plants are emitting

mercury, but it's not important, in terms of the

conclusions that I drew because I don't look at

individual plant information at all. In the modeling

studies that I do, I don't need that information for

receptor modeling, you need that for deterministic

models.

DR. KEELER: Question 8: "On page 81,

the TSD states that you suggest `the lifetime of

elemental mercury in the atmosphere is likely much

stronger than previously believed.'"

MR. RIESER CONTINUES:

Q. That should be "shorter."

A. Because my answer was no, that's not an

accurate statement of my testimony.

Q. Fair enough.

A. So the question should be it's likely much

shorter than previously believed?

Q. Correct. Is that an accurate statement?

Page190

A. Yes. I believe it is an accurate

statement. B: "What is the basis for that statement?"

One of the biggest uncertainties in all of the mercury

modeling and to understanding mercury chemistry at this

point is the rates of reaction for mercury in the

atmosphere. We learned through work that our colleagues

have done in the Arctic that elemental mercury can very

rapidly on a time frame of almost instantaneously react

with atmospheric halogens, bromine and chlorine in the

atmosphere to transform the elemental mercury to

reactive mercury and deposit to the surface of the snow

pack in such a way that all the elemental mercury is

depleted from the air over the course of hours. It's a

very interesting phenomenon that occurs at Arctic

sunrise. This really got the mercury world shook up

because it was something that we hadn't anticipated.

Most of the work that had been done prior to that

suggested that mercury chemistry was really slow, that

it would take days and days for you to oxidize mercury

and convert it from one form to another. Over the past

several years, we have learned that we have a poor

understanding of the mercury chemistry, that we don't

understand what chemicals and what compounds are

oxidizing mercury, and through the work of a large

Page191

number of scientists, we now know that mercury can

transform to the atmosphere from one form to the other.

It can be both oxidized and can be reduced and that

chemistry suggests that, in certain environments, such

as downwind of urban areas, that you can actually get

rapid oxidation of the mercury. By "rapid" there I

don't mean like in the Arctic, instantaneous. I mean

over the course of hours. In the same type of time

frame that is introduced in regional transport, you can

get transformation of elemental mercury to reactive

mercury, and so that would then shorten up the lifetime

of mercury in the atmosphere and that's the basis for my

statement.

Q. Has that phenomenon been observed anywhere

other than in the Arctic or marine-boundary

environments?

A. The phenomenon with the halogens is

something that has not been documented anywhere, but in

the Arctic, and in some cases, in the marine boundary

layer. We have not seen it in the marine boundary

layer, by the way. We have made our own measurements

and don't see that phenomenon, but others claim that

they do, but that's -- the reason for that is it's

associated with the Arctic sunrise. The whole winter

Page192

it's dark. The chemicals build up in the snow and ice.

The sun comes out, and all those chemicals get released.

It's a rapid explosion of all these chemical reactions

together with the ice that's there, so that's a very

different phenomenon. We do see in Michigan at our

sites and in Steubenville evidence of atmospheric

chemistry taking place where elemental mercury is

changing into reactive mercury, so we have observational

evidence of this happening, and we're in the process of

trying to define that better by improving our models, so

we can describe that, but it's something that is fairly

new, and not everyone has included those processes in

their models.

Q. Are reductions happening, as well, from

that?

A. Yes, they do. Reduction does occur. It

occurs in cloud water, and it's thought to occur in

other situations, as well, although we have not seen

those -- or those reduction reactions occur in our

observational studies, but we do see reduction occurring

in cloud water, and it's fairly important reaction in

terms of our cloud chemistry.

Q. Thank you.

DR. KEELER: Question 9 I think. "In

Page193

your testimony you discuss the U.S. EPA's there's a --

C-MAQ model to determine the impact of domestic mercury

sources on atmospheric mercury deposition. A: Are you

familiar with the Team model used to perform some of the

model Michigan report?" Yes. B: "Do your comments

regarding C-MAQ also apply to Team?" Could you be

specific about what comments you are referring to?

MR. RIESER CONTINUES:

Q. The comments with respect to its

uncertainties.

A. Yes. My comments are the same. I would

say that all Eulerian deterministic source oriented

models suffer from similar large uncertainties due to

the fact that they, A, don't adequately describe the

physical and chemical processes that control the

behavior of mercury in the atmosphere or the

interactions of mercury with the surface of the earth,

so I would say that they are all equally uncertain -- I

shouldn't say they are equally uncertain. They all

share in that large uncertainty an inability to model

mercury transport and fate.

Q. Is that a larger uncertainty than other

deterministic models, for example, the model used in the

Everglades?

Page194

A. No. I would say that the uncertainties in

all of these models are quite large, and again, models

are a tool, and you use the tool that you have at your

disposal, and when we did the work in the Everglades,

the model that we used was, at that time, the best

developed tool that we had. If we were to go back and

redo that analysis, we would have used a more

sophisticated tool, and redo that in a different, so

they are all very uncertain. What we know now that we

didn't know then is that we were really off in our

understanding of atmospheric chemistry.

Q. You didn't know that at the time of the

Everglades?

A. At the Everglades, yes.

Q. What ways were you off, in terms of the --

A. We really didn't understand the complexity

in the mercury chemistry at that time. We felt that the

reaction ratings were slow enough where chemistry wasn't

important. We didn't consider vertical redistribution

of mercury in ways where I think we should have. So

there's lot of uncertainties we have learned a lot over

the last 10 to 12 years that made me feel even more

uncertain about where we are at with the modeling.

Q. The C-MAQ model has been through several

Page195

versions and several episodes of peer review, has it

not?

A. I don't know that that's correct. To my

knowledge, unlike the receptor models, which have been

through exhaustive peer review, and so forth, C-MAQ, to

my knowledge, has been part of several inner

comparisons, but I do not believe that it has undergone

the same Agency verification and validation that, let's

say, that the RATA (phonetic) model did back in the acid

rain days.

MR. BONEBRAKE CONTINUES:

Q. You mentioned that you had some concerns

about uncertainties in the model used in the Florida

Study. Also, that you had acquired some additional

knowledge pertaining to some of the factors related to

that model, Dr. Keeler?

A. I'm sorry. I apologize. Could you re-ask

the question.

(At which point, the prior question

was read by the court reporter.)

DR. KEELER: I don't have concerns

over the use of the model, or in fact, if I went back, I

don't think I would change any of my conclusions based

on what I know now. I would just use a different model,

Page196

knowing the uncertainties of the chemistry are as large

as they are.

Q. Those uncertainties in chemistry, if you

were to re-run if same -- use the same model and re-run

the model, but would you change some inputs to the model

based on the additional information you have available

to you today?

A. Actually, I made that statement trying to

make the point that we have a -- we have increased our

understanding of the chemistry tremendously to the point

that, to apply a model such as that one anymore, would

not be the best model choice. It turns out, because

almost all the mercury that was emitted from the

incinerators came out in the reactive form, and the

reduction reactions that we understand that are

important in cloud, really don't -- wouldn't cause us to

change our answers at all, that, in fact, the modeling

we did in Florida probably would stand up pretty well.

If we were to re-do it, we wouldn't have to re-do much.

The things that we know now are much more important, in

terms of looking at the super long range, so global

redistribution, and really looking downwind of major

urban areas, so regional oxygen chemistry, which again,

Florida does not have very high oxygen levels. The

Page197

levels of ozone are extremely low, for the most part, in

South Florida.

Q. If I understand that correctly, you are

saying the uncertainties you talked about are related

primarily to global transport issues?

A. Global transport and urban chemistry type

reaction.

Q. Doesn't global transport have an impact on

local deposition?

A. Doesn't global transport have an impact on

local deposition? I'm not sure why you would call it --

does global transport have an impact on deposition

everywhere? Is that what you're asking me?

Q. At any particular location, some portion

of what's being deposited at that location may be

derived from a source outside of the region, as you have

described it, right?

A. Every place that you have mercury there is

a probability that some of that mercury came from an

area that you would say is not in the local or regional

area surrounding that site, and that mercury could have

come from a source that was located on the other side of

the globe, or it could have come from a molecule of

mercury that was emitted from a plant down the road that

Page198

went around the globe once, and got reemitted and came

back, and we can't differentiate that mercury, so the

global pool is a combination of mercury that was emitted

by our sources, our coal-fired utilities, our

incinerators, as well as those that were emitted from

Chinese power plants and incinerators, and no, I cannot

rule out any one location. Some mercury is part of that

reemission or long range transport of emissions from

some place else.

DR. KEELER: Question C: "Are you

aware of whether the Team results have been validated

against MDN data." It's not typical for one researcher

to look into the validation of someone else's use of

their own model. When one uses the model, or one

develops their own modeling application, we try to

validate our own work, and so I'm not intimately aware

of what validation has been done by the Team model. I

have read the peer-reviewed literature papers that have

been published on the Team model. The word "validate"

means different things to different people, and showing

a comparison of predicted wet deposition versus MDN wet

deposition, to me, is not validation. It's a

comparison, so but that doesn't mean that that model

hasn't been validated. I'm not aware of any validation

Page199

of that model, but it may or may not have been.

Q. Why would that the constitute a

validation?

A. "Validation" has a very specific meaning

when it comes to models, and in comparison of one part

of the model, output is not validation. Validation

suggests that you have compared the basic mechanisms and

processes in the model in a way that you can point to

real observations, and say that the model

subparameterizations are replicating reality. I can run

a model that will just use precipitation amount, and

predict with the same R squared the amount of mercury

deposition that you could get at almost any spot and you

could say, "Gee, I got the supermodel, and there's my

comparison, so I validated it," but to me, I'm not

describing anything. I'm just doing a statistical

explanation, so if you look in the model in literature,

and look at where models are actually validated, which

means you take the same input data and same emissions

data and same meteorological data, and you run models,

and you have various checks that you can make on the

model output at various steps to verify and validate

that the model is actually doing what you think it is

doing. That's a validation process, and again, the

Page200

model might have been validated. I have read the

peer-reviewed literature papers, and I have no basis of

knowing whether a validation was done or not.

DR. KEELER: D: "Are the

uncertainties" -- I already identified that -- answered

that one. D is no.

MR. RIESER CONTINUES:

Q. Are there difference levels of uncertainty

between the two?

A. Between C-MAQ and Team? Is that --

Q. No, between the receptor study and the

C-MAQ model?

A. Are there different levels of uncertainty?

I would say yes.

Q. In what way?

A. Again, understanding that I'm wearing two

hats. I'm the modeler and a measurement person, so I

don't want to criticize myself. I'm going to criticize

the other side, so I'm not trying to pick on one, versus

the other. In order to be able to model from a source

perspective, the fate and transport and deposition of a

pollutant, one needs to understand all of the processes

one has to understand the emissions, in terms of the

speciation; one has to understand the chemistry; one has

Page201

to understand the processing of the mercury and the

clouds; and one has to understand how the gas molecules

are interacting with the surface, all of this. That's a

lot of science. There's a lot there, and based on my

measurements, I would say that the uncertainties and our

ability to define the rates of chemical reaction of

mercury in the atmosphere and the rates of deposition of

mercury in the atmosphere, those uncertainties are much

greater than the uncertainties in our measurement

capabilities, much greater. I mean I would say that

borders a magnitude greater in source modeling than we

are in receptor modeling, so the two are completely

different.

Q. When you talk about the source modeling,

that includes the component that involves the

meteorologic model. Is that what you are doing?

A. No. Source modeling is C-MAQ and Team --

Q. I'm sorry, receptor modeling. I misspoke

A. No. When you are talking about

meteorologic modeling, again, we are using meteorologic

observations to do that. Trajectory modeling has

inherent it an uncertainty based on the time resolution

and the spacial resolution of the data that's collected

from the National Weather Service, and that uncertainty

Page202

is fairly well documented. People have been attacking

trajectory modeling for 25 years, so there's pretty good

literature describing biases and uncertainties in

trajectory calculations and how they would affect

prediction models.

Q. Does your receptor model account for the

non-linear transformations between the different forms

of mercury in the atmosphere between the source and

receptor?

A. The PMF and unmixed models do not have a

chemical transformation term in them. So they don't

take into account transformations, so that if there was

some phenomenon that one could come up with, which,

basically, what we do is we try to understand is there

some non-linear relationship that could occur and how

would that effect it, and in doing that analysis, if you

could come up with something that you could say would be

non-linear relationship you would have to figure out how

that would impact the relationship of the data at your

site, and that's inherent in the uncertainty analysis

that we do, but no, the models do not take into account

uncertainty -- or I mean, non-linear transformations.

DR. KEELER: Question E: "Can the

source receptor approach be used to predict the impact

Page203

of different regulatory approaches on mercury

deposition?" Not directly. Unlike a source model where

you can vary the emissions in the inventory, and say,

let's cut mercury emissions from some sector, and then

re-run the model, and then see what the difference is.

Receptor modeling only takes the observed results and

works backwards for telling you how much came from each

source sector, but it is -- it's a linear process, so

that you could say, if you were to cut mercury emissions

from that source, that it would a commensurate impact,

in terms of the contributions to the sample in that --

at that site, so it doesn't allow you to prognosticate.

You can't look into the future. I can't model 20/20,

for example, using a receptor modeling approach.

Question F: "In using the source

receptor approach, do you typically attempt to correlate

your results with findings from available atmospheric

deposition modeling?" Because we're using observations,

and working backwards to come up with a source, we do

try to go out and look at whether anyone else has

predicted source-receptor relationships to see how well

we are doing. In the case of all of our studies, we,

generally, try to do that. With mercury now, there are

more predictions coming out, but in the past, there

Page204

haven't been all that many, but yes, we do.

MR. RIESER CONTINUES:

Q. Have you done that with respect to

Steubenville?

A. Only in the respect that we did make a

comparison with what -- the results that were put

together by U.S. EPA using their CMAQ model.

Q. What was the result of that comparison?

A. The CMAQ model, which went into the CAMR

rule modeling, found that, for Steubenville, coal-fired

utility boilers contributed 43 percent of the mercury

deposition in the grid square where Steubenville is, and

the grid square was a 36-kilometer grid square, so 36 by

36 kilometers, so that number is an average of that grid

square. The only thing, and the biggest caveat here is

that those results were for the calendar year 2001. The

results that I have quoted here today were for our

modeling for the years 2003-2004, so it's really

comparing apples and arranges, but to get a sense for

how well CMAQ was doing, we made that comparison, so you

can't directly compare the 43 percent from the CMAQ

model to what we did because it's different years, and

the meteorology is very, very different, and since we

saw 13.2, or whatever it was in 2003, and 19.8 in 2004,

Page205

that there's a significant big uncertainty just from one

year to another, so comparing apples to oranges isn't a

good idea, but that's the only data that was available.

Q. Is there an uncertainty factor associated

with the 70 percent number that you have talked about

with respect to Steubenville?

A. What's the uncertainty number?

Q. We'll leave that for later.

DR. KEELER: G: "What steps have you

taken to evaluate the accuracy of the Steubenville

results?" As I mentioned earlier, one of the strengths

of our study really has been the extensive peer review

that was completed by EPA in its external reviewers. As

I mentioned, we provided the raw data and the models to

the external reviewers to re-run the models, and to use

whatever other tools they had at their disposal to

evaluate the conclusions and the raw quantitative

answers that we gave, and when we were given back those

results, and when they gave us our report, they,

basically, confirmed our numbers and said that these

were very solid, robust answers. The conclusions that

we got were very strong, and I would say that's a very

strong indication of the quality and accuracy of the

work that we are doing. Also, I should say, in terms of

Page206

evaluating the accuracy of the predictions, one of the

things we did is we went and looked at, again, the EPA

emissions inventory for, basically, for the whole

Midwest going from east of the Mississippi, and if you

look at that emissions database, and look at what is

burning coal and where in the greater vicinity, almost

all of the coal burned in the Midwest -- I don't

remember the exact figure, but a large majority of it is

burned by utility boilers, which again, kind of goes

along with the answers that we found from our receptor

modeling.

Q. I'm sorry. When you use the term

"vicinity" that --

A. Like I said, east of the Mississippi.

Q. The peer review, that's external to the

EPA. Is that a specific body, like the NAS, or somebody

like that?

A. EPA hired a contractor to ask for

independent review from three independent scientists who

felt they had the qualifications. I wasn't involved in

that. I have no idea how they went through the

selection process. None of the EPA people that were

involved in that study were involved in that peer review

section. I had no control over that, so these were

Page207

independent scientists who had modeling measurement and

source-receptor relationship qualifications.

DR. KEELER: Question H: "Are there

any studies that evaluate source-receptor studies?" I

think I mentioned before the EPA has been involved in a

number of, basically, model innercomparisons and

evaluations for the receptor models over the last 15

years, and those are documented in the peer review

literature, and yes, there are plenty of those, and

there's also been a lot of work being done as far as the

receptor modeling developed by EPA.

MR. RIESER CONTINUES:

Q. But it's accurate that those prior

receptor models didn't address mercury, correct?

A. Some of them did.

Q. Which ones?

A. I'm trying to think now. I think the

Hopke Group applied PMF, I think it was, looking at -- I

think it was mercury, and he also applied a

trajectory-based approach -- PSCF I think it's called --

applied to mercury measurements that they made in New

York.

Q. Do you know when those were performed?

A. I, honestly, don't recall.

Page208

MADAM HEARING OFFICER: I think we'll

-- five o'clock has arrived. We have finished -- we're

at Question No. 10.

Page209

STATE OF ILLINOIS)

COUNTY OF ST. CLAIR)SS

I, Holly A. Schmid, a Notary Public in

and for the County of Williamson, DO HEREBY CERTIFY that

pursuant to agreement between counsel there appeared

before me on June 16 and 17, 2006, at the office of the

Illinois Pollution Control Board, Springfield, Illinois,

Dr. Gerald Keeler, who was first duly sworn by me to

testify the whole truth of his knowledge touching upon

the matter in controversy aforesaid so far as he should

be examined and his examination was taken by me in

shorthand and afterwards transcribed upon the typewriter

(but not signed by the deponent, and said testimony is

herewith returned.

IN WITNESS WHEREOF I have hereunto set

my hand and affixed my Notarial Seal this 25th day of

June, 2006.

__________________________

HOLLY A. SCHMID

Notary Public -- CSR

084-98-254587

Page210