1

1 DIVISION OF ADMINISTRATIVE HEARINGS

DEPARTMENT OF ADMINISTRATION, STATE OF FLORIDA

2

3 SUGAR CANE GROWERS COOPERATIVE )

OF FLORIDA; ROTH FARMS, INC., and )

4 WEDGWORTH FARMS, INC., )

Petitioners, ) DOAH Case No. 92-3038

5 v. )

SOUTH FLORIDA WATER MANAGEMENT )

6 DISTRICT, an agency of the State )

of Florida; et al., )

7 Respondents. )

- - - - - - - - - - - - - - - - - - x

8 FLORIDA SUGAR CANE LEAGUE, INC.; )

UNITED STATES SUGAR CORPORATION; )

9 and NEW HOPE SOUTH, INC., )

Petitioners, )

10 v. ) DOAH Case No. 92-3039

SOUTH FLORIDA WATER MANAGEMENT )

11 DISTRICT, an agency of the State )

of Florida; et al., )

12 Respondents. )

- - - - - - - - - - - - - - - - - - x

13 FLORIDA FRUIT AND VEGETABLE )

ASSOCIATION; LEWIS POPE FARMS; )

14 W.E. SCHLECHTER & SONS, INC., )

and HUNDLEY FARMS, INC., )

15 Petitioners, )

v. ) DOAH Case No. 92-3040

16 SOUTH FLORIDA WATER MANAGEMENT )

DISTRICT, an agency of the State )

17 of Florida; et al., )

Respondents. )

18 - - - - - - - - - - - - - - - - - - x

100 S.E. 2nd Street

19 Miami, Florida

April 8, 1994

20 8:40 a.m. - 4:10 p.m.

21 DEPOSITION OF ED DOWNING

22 Taken before THOMAS R. NEUMANN, Registered

Professional Reporter and Notary Public in and for

23 the State of Florida at Large, pursuant to Notice of

Taking Deposition filed in the above cause.

24 - - - - - - -

2

1 APPEARANCES

2

ON BEHALF OF THE PETITIONERS SUGAR CANE GROWERS

3 COOPERATIVE OF FLORIDA, ROTH FARMS, INC., AND

WEDGWORTH FARMS, INC.

4

HOPPING, BOYD, GREEN & SAMS

5 123 South Calhoun Street

P.O. Box 6526,

6 Tallahassee, Florida 32314

BY: CAROLYN S. RAEPPLE, ESQ.

7

SOUTH FLORIDA WATER MANAGEMENT DISTRICT

8

POPHAM, HAIK, SCHNOBRICH & KAUFMAN, LTD.

9 4000 International Place

100 S.E. 2nd Street

10 Miami, Florida

BY: GREGORY CESARANO, ESQ.

11

ON BEHALF OF THE RESPONDENT-INTERVENOR

12 UNITED STATES OF AMERICA

13 TOM WATTS-FITZGERALD, ESQ.

ASSISTANT U.S. ATTORNEY

14 99 N.E. 4th Street

Miami, Florida 33132

15

16

17

INDEX

18 Witness Direct Cross Redirect Recross

ED DOWNING

19 By Mr. Cesarano: 4

By Mr. Watts-Fitzgerald: 33

20

3

1 EXHIBITS

2

3 NUMBER PAGE

4

5 Exhibit 10 through 12..............................5

6 Exhibit 13........................................20

7 Exhibit 14........................................20

8 Exhibits 15, 16..................................121

9 Exhibit 17.......................................142

10 Exhibit 18.......................................143

11 Exhibit 19.......................................143

12 Exhibit 20.......................................143

13 Exhibit 21.......................................144

14 Exhibits 22 and 23...............................144

15 Exhibit 24.......................................145

16 Exhibit 25.......................................145

17 Exhibit 26.......................................145

18 Exhibit 27.......................................145

19 Exhibit 28.......................................146

20 Exhibit 29.......................................146

21 Exhibit 30.......................................146

22

23

24

25

4

1 Thereupon --

2 ED DOWNING

3 was called as a witness and, having been first duly

4 sworn, was examined and testified as follows:

5 DIRECT EXAMINATION

6 BY MR. CESARANO:

7 Q. Good morning, Mr. Downing.

8 A. Good morning.

9 Q. We are continuing the deposition that we

10 left off last week. The rules are the same. You

11 tell me if you don't understand my questions, all

12 right?

13 A. Yes, sir.

14 Q. Since we last met, what have you been

15 doing?

16 A. Working on the classification for the 1985

17 image.

18 Q. What have you been doing? Tell me

19 specifically how you have been doing on the

20 classification.

21 A. By using the techniques as described in the

22 previous deposition, 1985 and '84 that was used as

23 the source for training sample selection in the 2A

24 study area and identified those sites by the training

25 samples set up, the classification schemes produced,

5

1 the classifications and continued, of course, the

2 analysis that I talked about also.

3 Q. Did you do anything on the other images,

4 the '87 or '89 or '91 images?

5 A. Unsupervised classification into 36 and 99

6 categories, no. We did nothing additional with

7 those.

8 MS. RAEPPLE: Off the record.

9 (Discussion off the record.)

10 MR. CESARANO: Mark these Exhibit 10.

11 (The documents referred to were thereupon

12 marked Exhibit 10 through 12 for

13 Identification.)

14 BY MR. CESARANO:

15 Q. We have marked these three documents

16 Exhibits 10, 11, 12 to your deposition. Counsel

17 produced these to me yesterday.

18 Could you explain what those are, please?

19 A. Give me one second. Document No. 12 is the

20 table indicating the area per class in the

21 interpreted TM image for 1993 document.

22 11 is a similar category map for the --

23 this is the entire Everglades study area. This is an

24 area that includes 2A and an area from Loxahatchee

25 down to the small portions of the national Everglades

6

1 park, the hectares -- class categories in hectares.

2 And No. 10 is the similar area calculation

3 table for the 1985 Landsat thematic paper image.

4 Q. When we were last taking your deposition we

5 discussed the differences in part between the

6 vegetative classes you were using on your map and the

7 classes that Mr. Rutchey used on his map. Have you

8 done any further work or analysis or attempted to

9 make those classes comparable to each other?

10 A. No. We weren't able to. We just decided

11 to use our classes with the definition that we used

12 in terms of percent cover and had no comparison. No

13 maps comparing Rutchey's classes to our classes.

14 Q. I'm going to try to make at least an

15 initial attempt to see if there is any way we might

16 be able to determine which classes of Mr. Rutchey's

17 would fall into at least your cattail classes. Did

18 you attempt to do anything like that?

19 A. Preliminarily we tried to compare roughly

20 our classes to his in some of the tables we produced

21 early on in the study.

22 Q. In looking at comparing Exhibits 10 and 12

23 which you have just identified, I note that on both

24 of those exhibits you have a class predominant

25 cattail, and we have already discussed the definition

7

1 of that, I believe 95% or greater cattail in an area,

2 correct?

3 A. We estimated roughly in the area of 95% or

4 greater to be a cattail category.

5 Q. I see in both exhibits in Exhibit 10 class

6 No. 2, predominant mix cattail/other. According to

7 class No. 3 in Exhibit 12, predominant mix

8 cattail/other?

9 A. No. The classes in the 1985 image, several

10 of the classes in this category in 1993 are combined

11 to form these classes in the '85 image. It's not a

12 one-to-one comparison by class name. If you notice,

13 there are fewer classes in '85 than there are in '93.

14 The total acreage would be the same, but the classes

15 themselves are inclusive of other classes.

16 Q. Could you explain to me which ones have

17 been combined, please?

18 A. The predominant cattail in the 1985 image

19 is predominant cattail, same definition for

20 predominant cattail in the '93 image. You go down to

21 the predominant mix cattail/other, and that includes

22 all the mixes available in the '93 category,

23 predominant mix cattail open water. All the mix

24 categories there of cattail and other would include

25 other in the predominant mix.

8

1 Q. Let me understand that. Category No. 2 in

2 the 1985 table is equivalent to categories No. 2, 3

3 and 4 in the Exhibit 12 table?

4 A. You are talking about between 10 and 11 or

5 10 and 12?

6 Q. Those are both 2A. Why don't you just tell

7 me which ones are included in category No. 2 in

8 Exhibit 12.

9 Q. Category 1 predominant cattail equals -- in

10 the -- let's start with the -- in the reference in

11 this case would be the '85 image?

12 A. In the '85 image predominant cattail and

13 predominant cattail class 1 are the same definition

14 of class. All the mix categories 2, 3 and 4 are the

15 classed or combined into the predominant mix cattail

16 other category No. 2 in the '85 image.

17 There is no equal mix category in the '85

18 image.

19 Q. Why is that?

20 A. To determine an equal mix from the air

21 photos we used we thought was not practical, so we

22 did away with that category and decided either it was

23 a mix with predominance toward either category. We

24 found no equal mix, equivalent equal mix category in

25 the '85 image. So equal mix 5, there is no

9

1 equivalent from '93 to '85.

2 Q. Category 3 in the '85 image predominant mix

3 sawgrass/other would include class 6 in the '93 image

4 which is predominant mix sawgrass/cattail, class 7

5 predominant mix sawgrass/other and 9?

6 A. Yes, and 9. Predominant sawgrass/open

7 water -- no. No. Not 9. It wouldn't be 9.

8 Predominant mix sawgrass cattail class 6 in the '93

9 image matches the predominant mix sawgrass class 3 in

10 '85.

11 Class 7 predominant mix sawgrass/other is

12 classed into class 3 in the '85 image predominant mix

13 sawgrass/other. Sparse sawgrass is equal to sparse

14 sawgrass. In other words, class 8 in the '93 image,

15 sparse sawgrass may contain periphyton. And class 5

16 in the '85 spars sawgrass may contain periphyton or

17 comparable.

18 The predominant sawgrass categories,

19 predominant sawgrass class 9, '93 image, predominant

20 sawgrass open water would fall into the predominant

21 sawgrass in the '85 image, which is class 4.

22 Predominant sawgrass in the '93 image falls into

23 predominant sawgrass in the '85 image. In other

24 words, that class 10 in the '93 image is included in

25 class 4 in the '85 image.

10

1 The slough is equivalent to slough, class

2 11 is equivalent to class 6. Class 12 in the '93

3 image is equivalent to class 7, which is broad leaf

4 vegetation/tree island. Class 13 in the '93 image

5 open water is equivalent to class 8 open water in the

6 '85 image.

7 And clouds and shadows in both the class 14

8 and clouds and shadows in the '93 image is equivalent

9 to class 9 clouds and shadows in the '85 image.

10 Q. Now, I want to ask about why you were

11 unable to determine whether or not there was an equal

12 mix of sawgrass and cattail -- or put differently,

13 why you were unable to identify that class from

14 aerial photography that you were using.

15 Why was that?

16 A. That class requires a very detailed view of

17 the site that we were unable to obtain in the 1985

18 photography. It was too close to call.

19 Q. Why are the total hectares approximately

20 100 different between the two images?

21 A. Apparently in digitizing the 2A study area

22 we were slightly off in delineation of the study area

23 in the '85 image. It was about 100 hectares

24 different.

25 Q. You told me about the accuracy evaluation

11

1 that you are presently engaged in at the time that I

2 last took your deposition. My notes reflect that at

3 that point you had gone through some 26 points. How

4 many total points were evaluated for the accuracy

5 assessment for the 1993 image?

6 A. I believe that number was 84.

7 Q. When was that accuracy assessment

8 completed?

9 A. It was completed during the deposition the

10 last time.

11 Q. What was the percentage of accuracy?

12 A. I believe that was 70% or 72%.

13 Q. Do you have the error matrix on that?

14 A. Yes.

15 Q. I would like to see it, please.

16 A. I'm sorry. I don't have that error matrix

17 with me today. It's being faxed this morning.

18 Yesterday when we finished this project -- the day

19 before when we finished this project and compiled all

20 of that NASA through -- we are members of the VIP

21 program. We were involved in the program using some

22 of the scanners that NASA has available.

23 NASA sent a film crew to our office

24 Thursday morning at 8:00. They came in and cleaned

25 out our work area, set up cameras and did interviews

12

1 for two and a half hours. And in the process, they

2 scattered stuff -- at least the records we had neatly

3 piled were moved to a different area by the film

4 crew.

5 And Pam Green is reassembling that and

6 sending me the current error matrixes. I apologize

7 for that. It was unexpected. They were supposed to

8 have been out in the field filming field work and

9 then would film in my work area Friday. But it

10 rained, so they came into my work area. But my staff

11 right now is finding that and they are faxing that to

12 us. We will have it in a few minutes.

13 MR. WATTS-FITZGERALD: Before the

14 accusations begin to fly, or I read it in the

15 paper, the Department of Justice had nothing to

16 do with the timing of this. It was actually

17 supposed to have occurred much sooner. Please

18 convey that.

19 THE WITNESS: If you know the documentary

20 thing they are doing on CNN, just remembered

21 they disrupted my work schedule by doing that.

22 BY MR. CESARANO:

23 Q. Where is it being faxed to?

24 A. Here in this office.

25 Q. To my office?

13

1 A. Yes.

2 MS. RAEPPLE: To whose attention?

3 THE WITNESS: Ed Dowling's office.

4 BY MR. CESARANO:

5 Q. Were you expecting that Pamela Green would

6 fax that first thing this morning?

7 A. Yes.

8 Q. Well, we will move off that point until it

9 gets here.

10 We were discussing with Dr. Lodge yesterday

11 some of the aerial photography, and I believe the

12 grid indicated that the scale on the national high

13 altitude program aerial photography is close to 1 to

14 65,000?

15 A. 1 to -- the scale in the photography wound

16 up being a surprise. That photography ordering to

17 NASA specifications when they sent it to us was

18 supposed to have been 1 to 58,200. But when we photo

19 reduced the quad sheets to match the photo, it didn't

20 come out to that ratio.

21 So in conversations with NASA about that

22 photography, well, obviously it was not at that

23 scale. So we matched the reduced quad sheets and

24 matched the photography and it turned out to be 1 to

25 64,900.

14

1 Q. Tell me how you went about reducing the

2 quad sheets and turning them into transparencies.

3 A. We took the in-house photography and used

4 that as a control for the quad sheet. So we matched

5 the quad sheet reduction to the NHAP, photography.

6 Q. And to your knowledge was it an exact

7 match?

8 A. It was a good map. I've never seen an

9 exact map in photo reduction.

10 Q. When you say a good map, can you quantify

11 how close?

12 A. The features were within acceptable

13 distances that we decided were necessary to do this

14 photo interpretation.

15 Q. And what were those tolerances that you

16 decided?

17 A. We attempted to match reasonably the scale

18 of the photography so when we over laid a 500 meter

19 grid we could identify the patterns within that 500

20 meter grid.

21 Q. Based solely on the reduction of the quad

22 sheets, can you tell me -- can you quantify the

23 locational error in the map?

24 A. Can I quantify the locational area in the

25 reduced quad sheets?

15

1 Q. Exactly.

2 A. I did not quantify the area in the reduced

3 quad sheets.

4 Q. But by reducing it like that, it does

5 introduce the possibility of error, does it not?

6 A. Yes.

7 Q. And the color infrared photography itself

8 has a margin of error, does it not?

9 A. Yes.

10 Q. Can you quantify that for me?

11 A. No.

12 Q. The process that Dr. Lodge used in manually

13 lining up the photography with the quad sheet also

14 has the potential to introduce error, does it not?

15 A. Yes, it does.

16 Q. Are you able to quantify that amount of

17 error?

18 A. We have not yet quantified that amount of

19 error.

20 Q. Do you believe that you will?

21 A. I don't believe it will be necessary.

22 Q. Why not?

23 A. Because we depended on photo recognition

24 within a 500 by 500 meter quad. We did not -- we

25 used a photo recognition pattern recognition within

16

1 the map and the quad sheet.

2 Q. Am I correct in stating that the total

3 locational error is a function of the margin of error

4 in the photo added to the margin of error in the grid

5 added to the margin of error introduced in the

6 process?

7 A. That would be correct if you assume you

8 were dealing in total positional error. We were

9 dealing in relational error. Those are different

10 types of errors.

11 Q. Would you explain that, please?

12 A. Positional is accuracy of each map in

13 relation to the point on the surface of the map and

14 coordinated space.

15 We were dealing more with relational

16 accuracy, which is once we identified an identifiable

17 pattern in the photo, we align those various overlays

18 and transparencies to that point. So relationally

19 within the 500 meter grid we identified areas and

20 patterns.

21 Q. In your opinion how does the total error as

22 we have just discussed relate to or compare to the

23 pixel size in the thematic mapper image of 30 meters?

24 A. As I stated already, we did not use

25 positional accuracy as the control in this photo. We

17

1 used relative accuracy. Relative positional

2 accuracy.

3 Q. Isn't it important to consider positional

4 accuracy when you are talking about extent of

5 vegetative cover?

6 A. It's important to consider positional

7 accuracy if you are basing any data you might add or

8 subtract to that map on simply a statement of

9 positional accuracy, which is not what we do.

10 Q. Talking about relational accuracy, is that

11 what you said?

12 A. Yes.

13 Q. Talking about relational accuracy and the

14 use of the color infrared photography. The dates of

15 the infrared photography were, I believe, March,

16 early March of 1984, mid February of 1980 and mid

17 February of 1985 --

18 A. There were two NHAP methods that we used in

19 the study area.

20 Q. You were using those to verify or to define

21 the vegetative classes in the November 1985 satellite

22 imagery, correct?

23 A. Yes.

24 Q. Would you not expect that in the period of

25 time between March 3, 1984 and November 2, 1985 that

18

1 there would be some significant changes in the

2 relationships of the various vegetative classes as

3 shown in the infrared photography between the

4 satellite imagery?

5 A. The relationships between the species --

6 I'm sorry.

7 Q. Between the grouping of the classes. We

8 are talking about relational accuracy. That's what

9 I'm talking about, the relationship of one species

10 location to another. In other words, it would

11 change, would it not, over that period of time?

12 A. What now? You are talking about relational

13 accuracy in terms of species communities. That's

14 really not the context I was describing relational

15 accuracy in.

16 If you could explain a little more about

17 what you mean. I'm not certain --

18 Q. As I understand it, Dr. Lodge would attempt

19 to locate various vegetative or various features in

20 the infrared photography and compare that position or

21 that spot, identify it and then locate it in the

22 satellite imagery.

23 Is that more or less correct, or you tell

24 me what your understanding of it was.

25 A. He identified the feature in the

19

1 photography and then we identified the feature

2 sometimes with his assistance in the raw image.

3 What was your question?

4 Q. How did you employ this relational or this

5 relationship, as you have described the relational

6 accuracy in that process?

7 A. The NHAP photography is color IR

8 photography. Color IR produces very good

9 differentiation in patterns in vegetation community

10 patterns. Those same patterns were apparent in the

11 raw satellite image.

12 By positioning Dr. Lodge's focus into a 500

13 by 500 meter grid as best we could determine through

14 the method I described, he would sketch out the

15 patterns that he recognized the various plant

16 communities to exhibit within that area.

17 We would take that sketch that he had

18 prepared for the patterns he saw in the infrared

19 photography, compare those patterns in the 500 by 500

20 meter grid on image which we produced in rectifying

21 the image.

22 If we could see a pattern that he had

23 described in the photo, then that became the pattern

24 we recognized in the image. From that, we selected

25 the training sample.

20

1 Is that what your question is about?

2 Q. Yes. That's what I was getting at.

3 Wouldn't you expect those patterns to change between

4 the period of time of IR photography or satellite?

5 A. Dr. Lodge has extensive knowledge of the

6 Everglades, so he would recognize any obvious shifts

7 in vegetative patterns from, say, one period to the

8 other. We had to assume, and I'm sure Dr. Lodge

9 discussed it, that he would have some recognition of

10 what was going on in those patterns and those

11 classes, those species locations.

12 MR. CESARANO: Take a short recess.

13 (Thereupon, a brief recess was taken,

14 after which the following proceedings

15 were had:)

16 MR. CESARANO: Mark that Exhibit 13.

17 (The document referred to was thereupon

18 marked Exhibit 13 for Identification.)

19 MR. CESARANO: Mark this as Exhibit 14.

20 (The document referred to was thereupon

21 marked Exhibit 14 for Identification.)

22 BY MR. CESARANO:

23 Q. We were talking about the relational

24 accuracy vis-a-vis the positional accuracy?

25 A. Did I say relational -- relative accuracy.

21

1 I'm sorry. Relative accuracy.

2 Q. And that is based on the relationship of

3 patterns, vegetative patterns to each other?

4 A. No. Relative accuracy really doesn't

5 relate specifically to vegetative mapping. Relative

6 accuracy is a mapping term that relates to the

7 accuracy of one feature's relationship to another.

8 Q. In this case, it's the vegetative patterns

9 or vegetative classes?

10 A. It's the vegetative classes, the air boat

11 trails, the tree island heads, the canal center

12 lines. It's any feature within the area you happen

13 to be attempting to bring those features together

14 into.

15 Q. Isn't it important to know the location or

16 the positional accuracy of the features at the outset

17 in order to have any confidence in the relative

18 accuracy?

19 A. It's absolutely necessary to at some point

20 to establish some type of positional accuracy when we

21 transferred the points onto the image which was

22 geographically correct.

23 Q. And the places where you established

24 positional accuracy were where?

25 A. In the image when we rectified the image.

22

1 Q. And that is based on known locations of

2 particular features?

3 A. The rectification of images -- I think we

4 went through this process already.

5 Q. We don't need details.

6 A. When you would select your ground control

7 points and then compare those known points to points

8 in the image and then geographically stretch or

9 rectify the image to that set of known points, that's

10 called rectification, and that's what we did with

11 that image.

12 Q. You did that with the satellite image?

13 A. With the satellite data, yes.

14 Q. Was it necessary to do that with the

15 quadrangle sheets?

16 A. No. Our methods did not require that we

17 rectify the quadrangle sheets.

18 Q. As you move farther away from the

19 rectification points in the image by using the

20 relative accuracy process, is there a greater margin

21 for error the farther away you get from a known

22 point?

23 A. You wouldn't do that. The points you are

24 referring to, I guess, are the ground control points,

25 the known points?

23

1 Q. Yes.

2 A. Those points were used to rectify the

3 image. The methods we used to establish the

4 geographic locations of features in the photograph

5 were all based on the relationship of those features

6 to each other. So moving from a point, you would

7 continually readjust as you move. In other words, --

8 Q. What is the margin of error that -- can you

9 quantify the margin of error in the rectification

10 process that you used?

11 A. Yes.

12 Q. What was that?

13 A. Let's see. What was it? That's part of

14 the process you would always go through when you

15 rectify an image as part of our digital files we were

16 provided. The error was very low. And frankly, I

17 don't remember -- it's in the data we have given you,

18 it's part of the record. It was well within the

19 acceptable limits for rectification of this image.

20 Q. What are acceptable limits?

21 A. The errors down to .003 you are definitely

22 down to within -- the accuracy requirements you would

23 need for us is the raw data, 25 meter by 25 meter

24 sample data set. I don't remember what ours were.

25 They were very low.

24

1 Q. Was it within .003?

2 A. I don't remember.

3 Q. But .003 is acceptable?

4 A. The software we used allowed us to quantify

5 that in terms of materials, I can't remember how many

6 meters that was.

7 Q. Plus or minus?

8 A. So many meters, yes. It's on our digital

9 tape.

10 Q. My recollection, Mr. Rutchey's report

11 indicated his was within 3 to 7 meters?

12 A. What, for his rectification?

13 Q. I think you are right. Not for

14 rectification.

15 A. You are thinking of positional accuracy.

16 He used the same process. RMS error generation and

17 converted that to meters and wound up with "X" number

18 of meters.

19 Q. Mr. Rutchey's rectification was plus or

20 minus eight meters. Do you recall if yours was

21 better or worse than that?

22 A. I don't recall.

23 Q. We spoke briefly about the tasseled cap

24 analysis that you ran initially in the 1985 image, I

25 believe?

25

1 A. Yes.

2 Q. Have you run a normalized different

3 vegetation index on the '85 image or the '93 image?

4 A. In terms of within the tasseled cap did we

5 use the coefficients that a tasseled cap normally

6 would use? Yes, we have. "Normalize" is really not

7 the right term to use there.

8 Q. What's was the right term?

9 A. Applying coefficients that are applied.

10 Q. What is the difference between those

11 results and the greeness components on the tasseled

12 cap?

13 A. What's the difference between what, now?

14 Q. Those results. You said it's not proper to

15 use "normalized," but however you phrased it.

16 A. I wouldn't use the term "normalized" to

17 apply the coefficient to a tasseled cap image mainly

18 because the normalization has taken on in this

19 proceeding.

20 We applied coefficients -- a standard set

21 of coefficients that are common in tasseled cap.

22 Those coefficients were multiplied by the brightness

23 value.

24 Q. That's a different process than the

25 tasseled cap process?

26

1 A. No. Tasseled cap is a process of applying

2 coefficients to images to reduce the amount of data

3 being increased or maintain the majority of

4 information in that data. It does that by applying a

5 set of coefficients.

6 Q. Well, in doing that process is it effective

7 to distinguish between cattail and sawgrass?

8 A. It's effective to distinguish between

9 spectral response of various species of plants and

10 ground and water.

11 Q. And in doing the map, we are concerned with

12 cattail and sawgrass, are we not?

13 A. Yes, we are.

14 Q. In your opinion, is it effective to

15 distinguish between cattail and sawgrass?

16 A. I think that it could be, yes. It could be

17 one of the two that can assist you in distinguishing

18 between the cattail and sawgrass.

19 Q. Is it similar to tasseled capping that it

20 shows vigorousness of growth?

21 A. Is what? Is it -- is what similar?

22 Q. This other process that we have just been

23 discussing using the ratios.

24 A. The coefficients are part of the tasseled

25 cap analysis. That's what tasseled cap is, it's a

27

1 set of ratios that you apply to bands in order to

2 compress data so there is no difference in what I

3 discussed and a tasseled cap. That is a tasseled

4 cap.

5 Q. Just so I'm sure I understand, that is the

6 analysis that measures total greenness of plants,

7 chlorophyl or vigorousness of growth?

8 A. I never heard it described as total

9 greeness. One component of tasseled cap analysis is

10 greeness, yes.

11 Q. Greeness values from low to high?

12 A. Low to high, yes.

13 Q. You mentioned that the difference in

14 hectares as we saw on the tape resulted from

15 digitizing the data.

16 Were you talking about the satellite

17 imagery or did you also digitize the color infrared

18 photography?

19 A. No. When I say the differences in

20 selecting out the study area, cutting it out of the

21 image.

22 Q. Did you digitize the color infrared

23 photography?

24 A. No.

25 Q. I want to understand where I went wrong. I

28

1 asked you about the normalized difference in

2 vegetation index. You said that was not the proper

3 term or description?

4 A. Based on the fact that "normalized" has

5 been used in all proceedings so far to refer to some

6 kind of adjustment to radio metric response, things

7 of that sort, I want to make sure the record

8 indicates that I'm not talking about performing a

9 normalization of historical image.

10 Tasseled cap by definition does not require

11 they do anything except apply tasseled cap images,

12 and that's what we did.

13 Q. Now I understand that. Did you understand

14 what my question was by using that term, however?

15 A. Maybe I didn't. Maybe you ought to ask it

16 again.

17 Q. As I understand it. This process is one in

18 which you determine a ratio between two bands in the

19 TM image. I think band 3 and 4, the distance and the

20 greater the ratio or the higher the number, the

21 greater the distance between the trough and the peak.

22 Are you understanding me?

23 A. Yes. I understand what you are saying, I

24 think. Continue.

25 Q. And by developing and normalizing that

29

1 ratio you arrive at a scale of brightness values

2 from -- that indicate very little vegetation or

3 greeness to the top, which is quite a large amount?

4 A. Okay. Yes. What you do is, you apply a

5 set of -- not just simply the difference between two

6 bands. You neutralize some bands by adding and

7 subtracting some bands from others and things of that

8 sort and come up with -- this has been going on since

9 the late '70's, but yes.

10 You do wind up -- I guess if you want to

11 use the term "normalized," you wind up with a new set

12 of values that contain a majority of the information

13 that was in that original set of readings but

14 compressed into a smaller space. And it is

15 reflectance values if you want to refer to that as

16 that.

17 Q. Did you do that?

18 A. Yes.

19 Q. And is that what you were describing as

20 part of the tasseled cap analysis?

21 A. Yes. That process is within the tasseled

22 cap analysis, applying those coefficients.

23 Q. Now I think I understand.

24 A. Okay.

25 Q. Counsel brought with her yesterday this

30

1 document here with the date at the bottom of April 5,

2 1994. I presume this is the latest?

3 A. This is the final.

4 Q. This is the final?

5 A. Uh-huh.

6 Q. The answer is yes?

7 A. Yes.

8 Q. When we were last together, we had some

9 discussion about some banding that appeared in the

10 image. Did you do any further -- anything further

11 with the data to try to correct or account for that

12 banding?

13 A. As I stated before, banding in my opinion

14 had no significant influence on this image. We've

15 done nothing to correct banding in this image. In

16 fact, there is no noticeable banding except in the

17 plot. I'm not convinced whether that's a plot or a

18 banding.

19 Q. What did you do with this image from the

20 last time that we were together?

21 A. Simply refined the categories, just came up

22 with the best final product that we could arrive at.

23 Q. You mentioned you are not sure whether the

24 banding is in the image or from the plotter?

25 A. I noticed no banding that had any effect on

31

1 this image. The banding, in my opinion, is not an

2 issue. The banding did not affect the classification

3 of my opinion.

4 Q. This next other document is I believe what

5 you called the entire study area?

6 A. Yes.

7 Q. You are saying, in your opinion it's a

8 non-issue, but I think it's rather clear in this

9 image. My question is, why in your opinion would the

10 banding be more distinct in the image of the entire

11 study area than the image of just 2A?

12 A. We made no adjustments for banding.

13 Q. So you don't have an opinion or don't know

14 why it would be more distinctive in the entire study

15 area than in the 2A map?

16 A. I don't think it's any more distinctive in

17 the entire study area than it is in 2A. Defective

18 banding, in my opinion, did not affect the

19 classification to any significant amount.

20 Q. Let's talk about Exhibits 13 and 14 for a

21 minute, please.

22 A. Okay.

23 Q. Could you just tell me what the

24 relationship between Exhibits 13 and 14 are, please?

25 A. Exhibit 14 is a resume, my resume.

32

1 Exhibit 13 is also an older version of my resume.

2 Q. What's the difference between the two?

3 A. This older resume, Exhibit 13, is a resume

4 that ends with my work with the state of Mississippi.

5 And this resume is a generalized, more current resume

6 of my work since then.

7 Q. And Exhibit 14 doesn't list any reports or

8 publications as Exhibit 13 does. Have you published

9 anything since the date of Exhibit 13?

10 A. Yes. Well, a few papers. Nothing that

11 applies directly to this project.

12 Q. Other than working on the 1985 image, and I

13 believe you said you refined the 1993 image somewhat

14 in the last week or so since we were last together,

15 have you done anything else in connection with this

16 project?

17 A. We have done the '85 interpretation and the

18 tasseled cap. That has been our work this week.

19 Q. I do want to get into the accuracy

20 assessment.

21 A. I apologize for that not being here. It

22 should be here. I don't know why it's not here.

23 MR. CESARANO: Other than that, I do want

24 to wait for that to come in, do you have

25 anything to go into at this point?

33

1 MR. WATTS-FITZGERALD: As long as you

2 understand he is reserving having to come back.

3 MS. RAEPPLE: Yes.

4 THE WITNESS: Would you like me to call to

5 the office?

6 MR. WATTS-FITZGERALD: Yes. We will -- we

7 are doing that. Is it my understanding we are

8 getting two accuracies, one for '93 and one for

9 '85.

10 THE WITNESS: Yes.

11 MS. RAEPPLE: Go ahead and call your

12 office.

13 (Thereupon, a brief recess was taken,

14 after which the following proceedings

15 were had.)

16 CROSS EXAMINATION

17 BY MR. WATTS-FITZGERALD:

18 Q. Good morning. I'm Tom Watts-Fitzgerald

19 representing the United States in this matter.

20 Same ground rules. If you don't understand

21 the question or need to take a break, just speak up.

22 Addressing for a moment and clarifying in

23 my mind -- the documentation may be en route -- I

24 understand the two matrixes reflecting the error

25 assessment or accuracy assessment for the 1983

34

1 imagery -- 1985 imagery and 1993 imagery that you

2 have produced will be here shortly and we will look

3 at that then.

4 There were worksheets, as that term is

5 used, by Dr. Lodge yesterday which I think you

6 referred to today that may have included traces of

7 vegetative patterns that were utilized in connection

8 with the 1985 image development.

9 Are those en route, to your knowledge?

10 A. Yes. Those are en route, to my knowledge.

11 Q. Am I correct in understanding that there

12 would be one of those for every accuracy assessment

13 point to the tune of roughly 85 of these things?

14 A. One of which one?

15 Q. One of those worksheets.

16 A. There are two worksheets. There are a set

17 of worksheets that were used in support of the

18 classification of the 1985 image, and there are a set

19 of worksheets, different type of worksheet for the

20 accuracy assessment for the '85 image. Both of those

21 are either in en route in Fed Ex or being faxed here.

22 Q. For the accuracy assessment worksheets

23 there are 84 of those or something of that order, one

24 for each point, the tables?

25 A. The total number would be listed in error

35

1 matrixes.

2 Q. Why are there less than half as many for

3 '85?

4 A. There were fewer classes and the classes

5 were broader in definition.

6 Q. By going to broader classes, that appears

7 to be reducing the apparent variability within the

8 imagery when you look at the color depictions; is

9 that true?

10 A. Reduce the variability in the image? You

11 mean the number of classes depicted in the image?

12 Q. Yes.

13 A. Obviously, yes.

14 Q. It reduces your ability to make vegetative

15 distinctions from that image to some degree?

16 A. In that the vegetative classes are

17 different since we classed, as I described earlier,

18 some classes into smaller classes, yes.

19 Q. And since you did that for the reasons you

20 stated, some of which related to the resolution

21 capabilities of the color IR from 1985, are you able

22 within a reasonable degree of certainty to compare

23 your set, the '85 imagery and '94 imagery as you

24 produced it, and make an assessment of acreage

25 changes and vegetative alterations over that time

36

1 period?

2 A. Within the limits of the data, yes.

3 Q. What could you see as the limits of the

4 data?

5 A. The fact that we have classes and the

6 interpretation that we did in '93.

7 Q. Can you make a direct comparison?

8 A. You can. Those that are directly

9 comparable that I described earlier. The others you

10 have to assume the new class is a combination of the

11 classes that were combined in the '93 image to give

12 you that same basic vegetative mix.

13 Q. Can you compare your two products to

14 Dr. Jensen's normalized time series to the various

15 years and assess vegetative alterations in comparison

16 to his assessment?

17 Can we put them side by side and reasonably

18 be able to analyze the results?

19 A. Well, since we have questions that have

20 been stated in the previous deposition about the

21 methods of Mr. Rutchey and Dr. Jensen's one-to-one

22 comparison as we stated would be inappropriate

23 because the classes are different.

24 Q. Is it my understanding that you will offer

25 no testimony at the time of hearing attempting to

37

1 make a one-to-one comparison between the work you

2 have performed and the work performed by Dr. Jensen

3 and Mr. Rutchey?

4 A. Would you define "one-to-one".

5 Q. It's the term you just used. You tell me.

6 A. You are the one that said would I do a

7 one-to-one comparison. I told you.

8 Q. No, we can read back the question. What I

9 said was, can you do a comparison. You said you

10 can't do a one-to-one.

11 A. Okay.

12 Q. What did you mean by that?

13 A. That the classes are -- as I stated earlier

14 in my deposition, the classes that Mr. Rutchey and

15 Mr. Jensen used, in our opinion, are inconsistent.

16 Therefore, there is no way to do a one-to-one

17 comparison. The classes are not the same. Our

18 classes are defined as a percent of the vegetative

19 class that we put is predominant in that class.

20 Q. I understood from your earlier deposition

21 that some of your concerns about the methodology

22 employed by Mr. Rutchey and then indirectly -- at

23 least for the '91 imagery -- by Dr. Jensen concerned

24 some of the hardware that was employed in the field;

25 is that correct?

38

1 A. That's correct.

2 Q. Have you resolved those hardware issues in

3 your mind?

4 A. Yes.

5 Q. So, in fact, you now know that the

6 equipment had the capacity to store the data during

7 the field operations?

8 A. I'm satisfied that I understand it fully.

9 My concerns about how they used the Trimble

10 equipment, if that's what you mean -- I'm not sure

11 what you mean.

12 Q. I had understood your earlier testimony to

13 be you were not confident that the equipment that was

14 used in the field in fact had the capability to store

15 the data points.

16 A. Storage was one of the concerns I had.

17 Q. Do you still have that concern?

18 A. My major concern does not involve storage.

19 Q. What are all of your concerns regarding the

20 equipment used in the field?

21 A. As stated earlier, those concerns -- as I

22 stated earlier, those were my concerns. The

23 procedures used by Mr. Rutchey are still of concern

24 to me.

25 Q. What procedures?

39

1 A. Procedures used to locate his accuracy

2 assessment once in the field.

3 Q. What about his procedures cause you

4 concern?

5 A. He had no communication to the base

6 station. He had no way to do post processing in the

7 field. Therefore, how could he have determined where

8 he was in the field when he went out with a set of

9 coordinates looking for an accuracy assessment point.

10 Q. Is it your expert opinion that one cannot

11 record the data in the field and do the differential

12 corrections and post processing afterwards?

13 A. No, it is not my opinion. That is not my

14 opinion.

15 Q. So that is possible?

16 A. Is it possible to post process data?

17 Q. Yes.

18 A. Certainly.

19 Q. I don't understand. I guess I would have

20 to ask you to elaborate what your concern was about

21 his inability to process the data in the field as one

22 of your major concerns.

23 A. According to his testimony, his paper, he

24 did no post processing in the field.

25 Q. Why is that a problem?

40

1 A. Because if you don't post process data with

2 a single hand-held free channel unit, if you don't

3 have some communication with the base station or some

4 way to adjust your position real time, then he could

5 not be exactly on the point he thought he was at

6 until he did post processing.

7 Q. And then once one had done the post

8 processing later back in the lab, for example, one

9 would know where you were while you were in the

10 field?

11 A. That's true.

12 Q. Even if that point was not what you were

13 looking for when you went out?

14 A. That is right.

15 Q. It's not a question of not knowing

16 necessarily where he really was, your concern is he

17 wasn't at the time in the field where he thought he

18 wanted to be?

19 A. He more than likely was not at the point he

20 wanted to be in while he was in the field.

21 Q. Your team used the Magellan receiver for

22 their field work for the training site selection

23 initially?

24 A. Yes, we used the Magellan.

25 Q. That did not have the capacity to store

41

1 points?

2 A. The Magellan unit doesn't store enough

3 points by itself.

4 Q. That was what Magellan model?

5 A. NAV Pro 4000.

6 Q. Which has limited storage capacity but not

7 sufficient for the work you were doing in a given

8 day?

9 A. Correct.

10 Q. Within the remote sensing community, what

11 is the percent accuracy considered acceptable -- I

12 should probably qualify that and say for this

13 purpose, user accuracy?

14 A. Within the user community, what is the --

15 Q. Within the remote sensing community?

16 A. What is the --

17 Q. Level percentage of accuracy that would be

18 considered acceptable for a map product like these

19 classified from Landsat TM data.

20 A. I think that depends on the project and the

21 person doing the work. I don't know that there is a

22 standard map accuracy. You are talking about map

23 accuracy?

24 Q. Yes.

25 A. Yes. I don't know there is a standard

42

1 acceptable level of accuracy.

2 Q. In your view, then, is there a certain

3 level of professional judgment and project specific

4 characteristics?

5 A. It has to be up to the person doing the

6 classification, yes.

7 Q. Do you, in fact, intend at the time of

8 hearing or between now and then to make an effort to

9 compare your classification products for '85 and '93

10 with Mr. Rutchey and Dr. Jensen's work?

11 A. As I stated before, we have done some

12 preliminary viewing of their products with ours. We

13 have done no extensive analysis. If I were requested

14 to do that I would do it.

15 Q. How would you go about it, given the fact

16 that you within your two products don't have

17 uniformity of classifications and do not have cross

18 uniformity with their work?

19 A. My classes are -- '85 classes are a sub set

20 of the '93 classes. There is some uniformity. It's

21 just a sub set. We would have to come up with an

22 acceptable class comparison between Mr. Jensen's and

23 Mr. Rutchey's work in order to do further comparison

24 of the two products, three products.

25 Q. You don't plan to do that unless requested

43

1 by counsel?

2 A. If I'm requested, I will. Right now we

3 don't plan to do that.

4 MR. WATTS-FITZGERALD: Counsel, if the

5 witness does that, I think we are entitled to

6 some notice because that would be a major new

7 opinion.

8 BY MR. WATTS-FITZGERALD:

9 Q. You have no final opinion on that at this

10 point?

11 A. On what?

12 Q. The comparison of your work with Dr. Jensen

13 and Mr. Rutchey's.

14 A. We have general impressions of our work

15 versus theirs, but we have no digital overlay

16 comparison yet. I haven't been asked to do that.

17 Q. Have you discussed with anyone production

18 of a digital overlay of any sort either on the

19 product that Dr. Jensen generated or Mr. Rutchey and

20 your work?

21 A. Yes.

22 Q. Have you taken any steps to produce such --

23 I assume by "overlay," you mean something like mylar

24 acetate that would go over --

25 A. I'm referring to a digital overlay, doing a

44

1 digital overlay and platting the result. You could

2 produce a hard copy overlay, but I would do it in a

3 digital environment.

4 Q. Have you begun that process?

5 A. We did some preliminary work several weeks

6 ago where we took some of Jensen's products and

7 brought them up, comparing them to ours, looking at

8 them in different windows. We did not overlay the

9 two images but looked at the differences between his

10 classes and ours.

11 Q. That was two separate images?

12 A. Yes, two separate windows.

13 Q. Was any preliminary work done through

14 software programs isolated into a single image,

15 various categories of classes, to see where there was

16 non-conformity or conformity?

17 We did some preliminary work on that. And

18 then because of the volume of data, we took it off

19 the system. We have not reinstalled it.

20 In a visual environment we set up a way to

21 look at two images on the screen, but we have done no

22 overlay.

23 Q. Do we have that on the electronic data

24 provided, I guess, yesterday?

25 A. That was a set of -- it wasn't dated. It

45

1 was a set of algorisms. You should have it on the

2 back-up tape. If you don't, we will supply one.

3 Q. What's the file name or directory for that?

4 A. I don't remember.

5 Q. You testified in response to counsel this

6 morning that with respect to area 2A the acreage

7 differential that appears in the Exhibits, I think,

8 11 and 13 was as a result of digitizing error in one

9 of the images. But I'm not sure that you said

10 specifically which of the images you determined was

11 incorrect.

12 A. Well, incorrect, I mean the digitizing, the

13 area digitized were slightly different. Obviously,

14 the acreage showed up differently. The area was

15 digitized to a slightly larger extent than '93. '85

16 had slightly more acreage than '93.

17 In eight years we lost a hundred hectares.

18 If it was my acreage, I would probably care about it.

19 Q. Did you analyze to determine where it went

20 or what portion of the image that was attributable

21 to?

22 A. No. Acreage was lost in the digitizing of

23 the study area. We included 130 hectares in one

24 image that was not included in the other.

25 Q. I have no reason to believe within the

46

1 bounds of 2A over that eight year period it grew. So

2 where did the 100 acres come from?

3 I mean, is it strictly an artifact of the

4 digitizing process? Is it an imagery problem? How

5 do you account for the differential?

6 A. Since we didn't use coordinated geometry to

7 digitize the area of 2A. To start with, digitizing

8 has in it always a percent of error, if you want to

9 call it that. You wouldn't take much difference in

10 digitizing an area of interest in one year's image to

11 another to find out less than one-tenth of one

12 percent or less than that error in the total acreage

13 of the study area.

14 Q. I take it within your professional opinion,

15 that's within the acceptable range of errors for that

16 type of process?

17 A. I would certainly think so, yes.

18 Q. We may have covered this last time around,

19 but I didn't see it in my review of the available

20 information from the earlier deposition.

21 But prior to this particular case, what

22 work had you done in the vegetative mapping field in

23 the Everglades?

24 A. None.

25 Q. Had you done any vegetative matching for

47

1 the wetlands prior to this case?

2 A. In those various industrial site studies, I

3 did using MSS data, I think a little TM data. In my

4 career I have done wetlands identification and

5 wetlands mapping mainly in terms of industrial site

6 development.

7 Q. Did you use TM data for those wetland site

8 evaluations?

9 A. I have to look at the reports and see --

10 no, I would not have used that. I used MSS on those

11 studies.

12 Q. The accuracy assessment points for '93 you

13 said you have 84 of those for the '93 image. I

14 assume -- I shouldn't assume. Are those 84 all

15 within area 2A?

16 A. Yes.

17 Q. There are no accuracy assessment points for

18 the balance of the image that shows the entire study

19 area as you have referred to it?

20 A. That's correct.

21 Q. And I noted yesterday, and it's in front of

22 us today, that that December '93 production dated

23 April 4 for the entire area has a warning or at least

24 a caveat, if you will, on it that there is no

25 accuracy associated with the areas other than WCA-2A?

48

1 A. I believe it says no accuracy assessment.

2 Accuracy is verified within Water Conservation Area

3 2A only.

4 Q. I translate that -- tell me if it's correct

5 or not -- that that means no one is making any

6 representations as to the accuracy of the image

7 beyond the boundaries of 2A?

8 A. It means the accuracy has not been

9 assessed.

10 Q. You are drawing a distinction. Tell me

11 what the accuracy is beyond 2A.

12 A. I expect the accuracy to be close to 2A.

13 Q. Quantify that.

14 A. I can't do that until we do an assessment.

15 Q. Are you going to do that?

16 A. If I'm asked to do that, I will.

17 Q. Do you have any present plans to do an

18 accuracy assessment?

19 A. No.

20 Q. How long would it take if you were asked to

21 do an assessment for the balance of the area depicted

22 in that image?

23 A. That would depend on the availability of

24 aerial photography for the entire study area.

25 Q. Where did you acquire the NHAP photography

49

1 for the '83 image?

2 A. For the '83?

3 Q. I'm sorry, '85. I keep saying that.

4 A. The '85 NHAP photography was acquired from

5 the NASA spec photography. We ordered it through the

6 base, St. Louis facility. I'm not sure where it was

7 actually shipped from, from the U.S. government.

8 Q. Now, for the '93 image you used BDA's

9 photography for 2A?

10 A. Yes.

11 Q. Was any effort made at the time you were

12 seeking to acquire CIR photography in the time frame

13 of the December '93 satellite pass to determine what

14 other CIR was available?

15 A. I'm sorry, did I make that investigation?

16 Q. Yes.

17 A. Yes.

18 Q. Was there other CIR available?

19 A. I couldn't find any, no.

20 Q. So when you said a few minutes ago it would

21 depend on the availability of CIR, I take it you are

22 really saying you can't because there is no CIR known

23 to exist?

24 A. No. Right now we are unable to find some.

25 But the way I understand it, the photography

50

1 available through Breedlove, Dennis, goes beyond 2A,

2 so I could go beyond that.

3 Q. What's your understanding of the rest of

4 the geographic coverage?

5 A. It goes outside of 2A.

6 Q. Do you know where?

7 A. Well, I have seen some indices that goes up

8 to Loxahatchee a little bit and over into 3A or 3.

9 Q. If that's the extent of the CIR imagery

10 available, would you be able to assess only those

11 portions of the areas beyond 2A that fell within the

12 scope of the vegetative coverage?

13 A. We would have to have some type -- to use

14 the same method we have to have some type of

15 photography available for the study area.

16 Q. Are any other methods available to you?

17 A. You could hold out a percent of the

18 original ground truth sites from the classifications,

19 classify using that set of ground truth points, and

20 then check your classification against your own

21 ground control points to see if, in fact, you

22 classified the image as to what you held out.

23 Q. You would go back and regenerate your image

24 with less ground truth points?

25 A. That's one option you would have, yes.

51

1 That could be done.

2 Q. Aren't you kind of robbing Peter to pay

3 Paul there? I understand your reasoning, I think, in

4 not using a ground truth point employed for

5 generation of the image or training in the image for

6 error assessment because of the bias that induces and

7 the statistical problems that would be incurred in

8 doing something like that.

9 But you in a sense end up under training

10 your algorism to generate the image as you cut back

11 at the front end of the process to reserve them for

12 later use as assessment points. You don't have that

13 many to begin with for an area that size?

14 A. Is there a question there?

15 Q. Is that true or not?

16 A. Is all of that true?

17 Q. Yes.

18 A. Yes. Of course it's true.

19 Q. Speaking of numbers and points, how many

20 accuracy points do you expect to have total for the

21 '93? Are there any more that you will be assessing?

22 A. At this point we generated 200 or 240

23 points. We possibly could assess --

24 Q. For accuracy?

25 A. For accuracy.

52

1 Q. And you used 84. Are you going to use any

2 more?

3 A. If instructed to do so, we certainly could.

4 We maintain that set of points for that reason.

5 Q. The 84 that you used, how were they

6 selected out of the 240?

7 A. In sequential order.

8 Q. And the 200 that were generated, that was

9 done with using what selection program?

10 A. Stratified random sampling.

11 Q. What measures, if any, did you take when

12 you determined to use only the first 84 to insure

13 that each of the classes was adequately represented

14 for statistical purposes in the reduced sample that

15 you chose to employ?

16 A. Since the sample was generated through

17 random numbers, since we went through the random

18 numbers sequentially we wound up with a random set of

19 the entire set.

20 Q. Who would instruct you to use additional

21 points out of the 200? Where would that decision

22 come from?

23 A. It would come from my people that hired me

24 to do this project.

25 Q. At this point you have intentions to use

53

1 additional points. That's something you are capable

2 of doing given time, but no current plans, per se?

3 A. Yes, correct.

4 Q. With regard to the accuracy assessment for

5 the 1985 image, I think you said there were 34 points

6 for that?

7 A. 34 or 37, I believe.

8 Q. How were those points selected?

9 A. Exactly the same method, random stratified

10 samples. There were 104 over 100 points. I believe

11 104 in total.

12 Q. On what basis did you determine that the 37

13 or whatever the number is, in the 30's was, that were

14 ultimately used and were identified by Dr. Lodge was

15 sufficient for statistical purposes in analyzing the

16 accuracy of the area encompassed by WCA-2?

17 A. It was just my judgment.

18 Q. Are you familiar with the body of

19 literature in the remote sensing field which

20 describes how one should go about the process of

21 assessing the appropriate number of accuracy

22 assessment points?

23 A. Yes.

24 Q. What factors did you take into account in

25 making your professional judgment that 30 was

54

1 sufficient?

2 A. Just begin the complexity of the map and

3 the major classes that we were dealing with. Major

4 classes were cattail and sawgrass.

5 Q. You mentioned in your original deposition a

6 concern because you felt that Mr. Rutchey -- and

7 perhaps Dr. Jensen, as well -- had an initial bias in

8 their assessment because they set out focused on

9 cattail, Mr. Rutchey, and sawgrass, Dr. Jensen.

10 Do you recall saying something along those

11 lines?

12 A. Stated some concerns about their method and

13 the fact about comparing the two.

14 Q. My understanding from Dr. Lodge's testimony

15 is that basically you set out to do the same thing,

16 focusing essentially on sawgrass and cattail; is that

17 not true?

18 A. That we did the same thing that Mr. Rutchey

19 or Dr. Jensen did? They did different things.

20 Q. I'm not talking methodology. I'm talking

21 mind set. You set out along with Dr. Lodge with the

22 purpose of specifically analyzing vegetative changes

23 with respect to cattail and/or sawgrass just as they

24 did, although you chose a different methodology to do

25 it?

55

1 A. Well, your question, if I understand your

2 question, yes. We set out to determine the

3 relationship between sawgrass and cattail, but we

4 feel that Dr. Jensen's methods and Mr. Rutchey's

5 methods were different. Therefore to say it was like

6 them, I guess you mean in terms of did they have a

7 goal of determining -- are you saying that both had

8 the goal of determining sawgrass and cattail?

9 Q. In both your effort and their efforts

10 looking at them collectively, yes.

11 A. Yes. The emphasis of the study was on the

12 relationship of sawgrass and cattail within 2A, yes.

13 Q. How many training samples were there for

14 the '85 image?

15 A. The acreage -- no, that wouldn't be it.

16 Yes. Let me see the tables with the acreage,

17 Exhibit 12, which is the '93 image.

18 If you are looking down at the bottom, it

19 says based on original 53 classes for December 10,

20 you see at the bottom 53 classes down at the bottom.

21 Each one, it says "based on original of each number

22 of classes."

23 Q. How do the classes tell me how many

24 training samples?

25 A. Because we produce the number of classes in

56

1 the original classification equal the number of

2 training samples. There were 35.

3 Q. You did map productions here of imagery for

4 36 training sites and 99 training sites -- I'm sorry,

5 classes. You just said classes equal training sites?

6 A. No. You are misunderstanding.

7 Q. Straighten it out for me.

8 A. What are you confused by?

9 Q. I asked you how many training samples you

10 had or training sites, if you will, for the '85

11 image. You just talked about the '93 image.

12 A. You said '85.

13 Q. I said '85 to start with. I'm not asking

14 how many classes. I understood you to say in the '93

15 image classes equals training sites and the numbers

16 don't match. I understood there were 116 roughly

17 training sites for the '93 image from your testimony

18 last time and doing all of the R, H and E's --

19 whatever that meant. Is that right, 116?

20 A. 116 observations were made.

21 Q. Training sites?

22 A. Okay. Wait. Let me understand what you

23 are asking again. Ask me one more time. I'm not

24 understanding.

25 Q. Back up. I'm looking at '93 now. You had

57

1 about 116 training sites for '93?

2 A. For the entire study area.

3 Q. When you generated the class map --

4 classified map for '93 for the entire study area, how

5 many classes did you have when you went through the

6 classification process initially?

7 A. You mean before recoding the classes into

8 general categories?

9 Q. Yes.

10 A. The map is down here and that show 36 and

11 99 classes.

12 Now I understand what you are asking.

13 There are two processes here for each year. If I'm

14 understanding correctly. A little history, I guess

15 for each year we produced an unsupervised

16 classification for 2A. 99 classes unsupervised and

17 36 classes unsupervised. That was a separate process

18 from the supervised classification.

19 Q. On the unsupervised, why did you do that?

20 A. On the unsupervised -- I thought you were

21 talking about supervised.

22 Q. We are going to go back to the supervised

23 on the unsupervised. Why did you do that?

24 A. On the unsupervised classifications we

25 produced 99 classes and 36 classes of unsupervised

58

1 classification of TM data for all of those years

2 because we wanted to see what type of spectral

3 significant classes there were in those images.

4 In other words, if you break it into 99 how

5 many classes and where are they spatially in the

6 image. If you break it into 36, where are those 36

7 clustered images. So it was to gain a feel for the

8 study area.

9 Q. But according to the printout dates, that

10 was done the end of last month. At least the

11 platting was done. When did you actually conduct

12 that exercise?

13 A. Unsupervised classification?

14 Q. Yes.

15 A. Most of that was completed in the week of

16 my deposition -- the deposition in Tallahassee,

17 Florida a week ago.

18 Q. How will you, if at all, be relying on

19 those unsupervised classification efforts in

20 formulating your opinions of testimony in this case?

21 A. They will provide an indication of spectral

22 diversity in the study areas to support analysis and

23 conclusions in the area.

24 Q. In an unsupervised classification you as

25 the operator get to set arbitrarily the number of

59

1 classifications the program should look for?

2 A. Correct.

3 Q. So if I told it I want 200 classes or 230,

4 all I have to do is set into the algorism program the

5 range of spectral signature that I want and it will

6 attempt to classify every pixel in the digitized

7 database into that number of classes; correct?

8 A. Correct.

9 Q. So the fact it's 99 or 36, per se, has no

10 significance. Those are just the numbers you chose?

11 A. No significance?

12 Q. It tells you nothing about ground truth, it

13 tells you nothing about what's really there, it just

14 assigns pixel by pixel into your set number of

15 categories subject to black spots if something falls

16 outside of the range you specified in the program?

17 A. Okay. First you make a statement that it

18 gives you no information about the image. In my

19 opinion, it does give you information about the

20 image, but it does not tell you -- it will not tell

21 you a certain class is cattail. It will tell you

22 that there are, based on what your input is, subsets

23 to this image, it will tell you what spectral classes

24 fall in that set of bands that you created.

25 Q. So if you tell it, give me 99 and here is

60

1 the specs for each of the 99, and you have an

2 incredibly homogenous area, the program could end up

3 coming back saying, sorry, there are only four

4 classes out there?

5 A. Repeat that, please.

6 Q. If I tell it I want the program to classify

7 an image in the 99 classes, unsupervised

8 classification, it will go pixel by pixel through the

9 entire image and will try to fit them into the 99 if

10 it can?

11 A. Correct.

12 Q. Some may not fit in the 99 because you are

13 specifying the range for each of those

14 classifications, are you not?

15 A. In an unsupervised classification in the

16 software -- Imagine software package, you tell it to

17 do 99 classes out of the image it will be produced.

18 Q. It will divide up your entire spectral

19 range in that image in the 99?

20 A. Yes.

21 Q. Same thing if you do it with 36?

22 A. Yes.

23 Q. After producing that work, what opinions or

24 conclusions did you derive from reviewing the

25 unsupervised class maps?

61

1 A. It gave us a feel for the effect of fire,

2 which was obvious in the unsupervised

3 classifications. It gave us a feel for where the

4 greater diversity existed in the image.

5 Q. How did you use or employ that information

6 or that sense impression that you developed in

7 producing the final product for the 1993 and 1985

8 imagery?

9 A. It served as reference material.

10 Q. So it doesn't directly affect the

11 classified products for '93 and '85?

12 A. That's correct.

13 Q. General background?

14 A. Yes. It was not directly affecting the

15 classifications, correct.

16 Q. Did you do any accuracy assessment based on

17 the imagery that was produced through the

18 unsupervised classification classes?

19 A. No.

20 Q. Did you make any effort, you or those

21 working with you at your firm, to take the

22 unsupervised classes the, 36 and 99 and identify them

23 as to vegetative type?

24 A. Yes, we did.

25 Q. There is no legend on the map that appears

62

1 to do that. Where does that appear?

2 A. It doesn't. We began -- as I stated in my

3 deposition last week, the original plan was to take

4 the 36 class or the 99, whichever we decided was most

5 appropriate, and use that in a hybrid classification

6 process. Hybrid classification process normally

7 combines an unsupervised and supervised

8 classification. But we did not use that. We did not

9 use that process on the '85 image.

10 Q. Did you use it on the '93 image?

11 A. No, we did not. It was supervised.

12 Q. So there are no hybrid products available?

13 A. That's correct. We did not do a hybrid

14 classification.

15 Q. Did you start that work and stop it or you

16 didn't begin it?

17 A. We actually never did a hybrid

18 classification. Wait. Let's see.

19 At one point when we collected about eight

20 training samples from the NHAP photography, we did

21 combine those signatures to the 36 class unsupervised

22 signature table and run the classification. But we

23 did not use that, did not ever produce it as a

24 product. Just that it was a test.

25 Q. Does that appear on the back-up tape you

63

1 brought us?

2 A. We didn't save it. It was not a

3 work-product.

4 Q. Moving now to the 1985 product, how many

5 training samples or sites were there for the 1985

6 image?

7 A. I believe there were 37.

8 Q. How many accuracy assessment points for the

9 1985 image?

10 A. Wait, let's see.

11 Q. Did you reverse the numbers? 37 roughly

12 is, in fact, the accuracy assessment number?

13 A. Just a second.

14 Q. It's not always me.

15 A. You are right. It's not. '85 -- I cannot

16 remember right now how many training samples were

17 used in the '85 image. My mind is blank. I can't

18 remember right now. Those documents are being faxed

19 here. We will have them shortly. I apologize for

20 that.

21 Q. Without the actual number, can you tell me

22 how the training sites were selected for the '85

23 imagery?

24 A. Yes. Exactly. The process was very, very

25 similar to the selection process used in the '93

64

1 image. You want me to elaborate some more?

2 Q. It was my understanding in '93 they went

3 out in the field. First, you couldn't very well go

4 out in the field in '85?

5 A. Okay. The field operations became

6 photogrammetric operations. The process, in other

7 words, the process of identifying samples were

8 similar in that we weren't in a helicopter, we were

9 perusing the NHAP photography. We found areas that

10 had recognizable vegetative patterns, zeroed in on

11 those and used those as training samples.

12 In other words, Tom would make his notes,

13 as he told you yesterday. From that, we would

14 identify the same area in the image and then select a

15 training sample from the image that met the photo

16 mission, -- if you would, observations.

17 Q. What criteria were used to decide how many

18 training sites you needed for each class that you

19 were attempting to generate?

20 A. We attempted to find some number of

21 training sites in each class and attempted to

22 stratify those per our knowledge of the image in

23 terms of most of it is sawgrass, or at the center

24 there might be cattails, something of that sort.

25 You select according to expected patterns

65

1 to some degree.

2 Q. So that was done more or less in a

3 supervised as opposed to unsupervised fashion?

4 A. Absolutely supervised.

5 Q. Because you were working from CIR

6 photography with a relatively small scale in

7 comparison to what you used for '93 with the loss of

8 resolution, et cetera., how do you account for the

9 inevitable variability within an area in terms of its

10 spectral signature?

11 A. Variability in an area was more difficult

12 to interpret. Therefore, that's why we reduced the

13 number of classes we could identify in the available

14 photography.

15 Q. Was there any class or is there any class

16 in your '85 imagery for which only one training site

17 was available or used?

18 A. I don't think so. I think we had at least

19 more than one in all classes, all field observations.

20 I can't remember one that only had one training site.

21 Q. You would be able to tell that --

22 A. When we see the worksheets, those are

23 clearly listed.

24 Q. You told Mr. Cesarano this morning that the

25 satellite image was rectified?

66

1 A. Yes. How did you determine from your

2 satellite image -- I think you were taking about '93

3 at that point, maybe this applies to '85.

4 Were both '85 and '93 geo rectified?

5 A. The '93 image was rectified and the '85 was

6 rectified to the '93 image.

7 Q. How many ground cross points did you use

8 for '93 to do that?

9 A. That is in the digital data. I'm trying to

10 remember. That is on the tape we gave you. I

11 believe there were about 34 or 30 something ground

12 control points.

13 Q. Is that throughout the entire study area?

14 A. Throughout the entire study area.

15 Q. How many within 2A, if you can recall?

16 A. I would have to go back and look at the

17 records to see.

18 Q. How did you determine the geographical

19 position of the control points?

20 A. Determine -- you mean where they were?

21 Q. Yes.

22 A. We used USGS quad sheets.

23 Q. What's the positional accuracy of quad

24 sheets?

25 A. The positional accuracy is stated by U. S.

67

1 geological survey to be plus or minus 40 feet.

2 Q. Were you working off the full sized quads

3 at this point or had you had to reduce them already?

4 A. Which image are you talking about?

5 Q. '93.

6 A. We used the full size quads.

7 Q. You said you geo rectified '85 to '93?

8 A. Yes.

9 Q. So you did not use the quad sheets for that

10 purposes for the '85 imagery?

11 A. No, we did not go to the paper quads to

12 rectify the '85 image to the '93 image. We used

13 image to image rectification.

14 Q. Because you used the quads for '93, the

15 geophysical position accuracy was the limiting factor

16 for the '93 image, or are there other factors that

17 affect the geo rectification accuracy of the image?

18 A. Our geographic rectification was based upon

19 USGS 24,000 quads. They were the control points from

20 which we selected our ground control points.

21 Q. I understand that. I'm not sure that quite

22 answered the question I had in mind.

23 A. What was the question?

24 Q. The question I had in mind is that the only

25 error input, if you will, or were there -- could

68

1 there have been other sources of positional error in

2 your process not solely attributable to the error

3 inherent in the quads?

4 A. Let me make sure I understand what you are

5 asking me. Are you asking what error might be

6 introduced in the rectification process other than

7 the error in the quad?

8 Q. Yes.

9 A. The ability of the operator to identify a

10 feature in the quad and to identify the same feature

11 in the image. Obviously, it has a potential for

12 error.

13 Q. I assume if you selected a ground control

14 point, it's because you could find it in both. Is

15 that a fair assumption?

16 A. Yes.

17 Q. The ability of the operator to find it is

18 not a factor. You used it and you found it?

19 A. No. I mean, the ability of an operator

20 always has an influence on the potential to introduce

21 error. Ground control points are usually the

22 intersection of two vectors. You have to estimate

23 that intersection so that includes the accuracy of

24 the digitizing tables, the eye of the operator. So

25 there is potential error always for operator

69

1 interface to the process.

2 Q. In your expert opinion what is the error of

3 the '93 image -- the total error, excuse me?

4 You were fitting the '93 image to the quads

5 in that entire process, what's the assessment of the

6 overall proportional error?

7 A. That's in the digital data. That's

8 normally stated. It was very small. I don't

9 remember the number. It's in the digital data

10 supplied to you. You are welcome to look at the

11 files.

12 Q. With regard to the '85 satellite image, am

13 I correct in believing there is an error inherent in

14 fitting the '85 to the '93?

15 A. There is potential error, but we used the

16 same observed points in the '93 to find that same

17 intersection, if you would, in the '85 and then

18 rectified those two together.

19 Q. What role, if any, did the reduced quads

20 play in the rectification process?

21 A. None.

22 Q. So the error inherent in fitting the '93

23 image to the full sized quads would be carried

24 forward in the effort to fit the '85 image to the '93

25 image. You can't be more accurate?

70

1 A. No, you can't. You couldn't increase

2 accuracy.

3 Q. Theoretically you could increase the

4 accuracy but you wouldn't want to say that in looking

5 at your error you would look at the total range?

6 A. Yes, that is correct.

7 Q. What is your assessment of that error in

8 '85, the '85 imagery?

9 A. We didn't do a quantitative assessment. We

10 viewed the overlay of the two images and the overlay

11 of the control points and saw no error, no increased

12 error that affected our interpretation.

13 Q. So you have no specific estimate of the

14 total location error on the '85 image as the

15 methodology for '93?

16 A. We could certainly run an RMS error on the

17 '85 image. I don't know we did that. We probably

18 did do that. I just haven't looked at it. I could

19 probably produce that if you wanted it.

20 Q. What's your pixel size for '93?

21 A. 25 meters re-examined to 25 meters.

22 Q. That means that's not what you got in the

23 original digitized data?

24 A. We got it at 25 meters. We opened it,

25 re-examined it.

71

1 Q. The scanners in the satellite don't produce

2 its data in that?

3 A. That's correct.

4 Q. So you received reprocessed data to begin

5 with?

6 A. Yes.

7 Q. Any error induced by that reprocessing

8 effort?

9 A. There could be potential for some error.

10 Q. Can you quantify it for me?

11 A. No.

12 Q. Can anybody?

13 A. I would imagine EOSAT could probably do it.

14 Q. You never seen in the literature that talks

15 about that?

16 A. I can't recall any papers. I'm sure there

17 have been people that attempted to do that.

18 Q. Didn't I see in your earlier testimony a

19 criticism that Dr. Jensen had done re-examining and

20 was using a different pixel size than the original

21 data?

22 A. I don't think you saw that. I don't

23 believe I said that.

24 Q. Well, I guess my question is, since you did

25 it you must think it's an acceptable procedure?

72

1 A. I didn't. You misunderstood what I said.

2 Q. Forget Dr. Rutchey for the moment. You

3 used the re-examined knowledge or a process set of

4 data, preliminary process, to get it into the

5 exercise you wanted to employ for some reason?

6 A. We used the standard pixel size from EOSAT

7 Corporation. We did not resample the pixel size. We

8 did not change the resolution of the data.

9 Q. It had already been done for you?

10 A. A scanner does not produce a grid. A

11 satellite produces a single reference based on the

12 central point of a scanning mechanism.

13 Every device, every mechanism for

14 establishing the satellite's recognition of a pixel

15 location on the ground is modified through some

16 re-examining method. You have to do that.

17 A pixel -- in fact, the instant field of

18 view of a scanner is usually beyond the pixel size

19 produced in the data delivered from EOSAT. So every

20 data set, every scanned data set has to be

21 re-examined in some fashion.

22 So, yes, everything has to be re-examined.

23 It's not unusual, in fact it's required that the data

24 be re-examined in some fashion in order to be usable.

25 Q. Routine operation in the field?

73

1 A. Required operation by the provider of the

2 digital data, EOSAT, SPOT, CASEY, anybody else.

3 Q. Would you agree that the overall error for

4 the '93 image is probably in the range of plus or

5 minus 20 to 30 meters?

6 A. What now? Restate that. I'm sorry.

7 Q. Would you agree that the overall error in

8 the '93 image for WCA-2A is plus or minus 20 to 30

9 meters, it's on that order?

10 A. The overall error in the '93 image -- you

11 mean the image itself, the raw data?

12 Q. Yes.

13 A. I'm sorry. Now I'm trying to understand

14 exactly what you are asking. Are you asking if I

15 think that the data from EOSAT --

16 Q. The rectification error overall for the '93

17 image.

18 A. Was within how much?

19 Q. Plus or minus 20 to 30 meters.

20 A. Our rectification?

21 Q. Yes.

22 A. I would say my recollection is better than

23 20 meters. The RMS error that we generated will tell

24 you what the estimated error is and it's in the

25 digital tape. RMS error is standard product. And to

74

1 my recollection, I can't remember exactly what it

2 was, but I would say it's much better than 30 and

3 possibly better than 20. Is that what you are

4 asking?

5 Q. Yes.

6 A. Okay.

7 Q. You described for Mr. Cesarano in some

8 degree the last time and a little more today doing

9 the KT transformation, but I don't understand how

10 that's reflected in your work.

11 Can you explain for me how the KT formation

12 has assisted you in producing either the '93 or '85

13 imagery you will be utilizing in the case since you

14 went to great trouble to produce it?

15 A. You said the KT?

16 Q. What you called tasseled cap.

17 A. You called it --

18 Q. KT.

19 A. Which is what? What's KT?

20 Q. You never heard it referred to that way,

21 after the two guys invented --

22 A. You are talking about the researchers of

23 tasseled cap?

24 Q. You said tasseled cap was one narrow

25 application of a technique.

75

1 A. You are talking about their techniques or

2 the tasseled cap?

3 Q. Where did tasseled cap come from?

4 A. That has been around a long time.

5 Q. The question is, where did it come from?

6 A. Arum did a lot of research.

7 Q. I am sorry, who?

8 A. Arum. I seen it used at NASA's Earth

9 Research Center.

10 Q. You are not answering the question. Who

11 developed the technique that you are calling tasseled

12 cap?

13 A. I don't know who developed it. My

14 knowledge of tasseled cap is based on numerous papers

15 and mention of tasseled cap over the years.

16 Q. Which of those papers sanctioned or in any

17 way suggested the validity of employing tasseled cap

18 in a sub tropical wetland environment?

19 A. I wouldn't know which paper recommended

20 that or didn't recommend that.

21 Q. Is it fair to say, then, you are aware of

22 no paper that, in fact, identifies such use in the

23 past?

24 A. I can't think of a paper that I read that

25 used tasseled cap in a sub tropical environment.

76

1 Q. Have you ever used tasseled cap in a sub

2 tropical environment to assist in the analysis of

3 spectral imagery?

4 A. No.

5 Q. Whose idea was it to use it in this case?

6 A. Tasseled cap was suggested by Dr. Roy

7 Stein.

8 Q. Who is he?

9 A. He is a professor at Sanford University.

10 Q. What's his involvement in the case?

11 A. He has been employed by some of the parties

12 in the case to just look into it and see what's going

13 on. He has been asked to form an opinion about

14 certain pieces of this case.

15 Q. How do you know that?

16 A. He has told me he was employed to

17 investigate this process.

18 Q. Have you consulted with Dr. Stein?

19 A. Yes, I have.

20 Q. Has he reviewed your work?

21 A. He has reviewed our procedures. We

22 discussed our procedures with him and spent one day

23 looking at that, and he has reviewed it.

24 Q. When was the one day looking and reviewing?

25 A. About three weeks ago.

77

1 Q. Did he provide you any comments on your

2 work?

3 A. Yes.

4 Q. In writing?

5 A. No, just comments.

6 Q. Has he produced or written products, to

7 your knowledge, resulting from his review of your

8 work?

9 A. Nothing that I have seen.

10 Q. As a result of your communications with him

11 either by person or phone or otherwise did you alter

12 your work plan in any fashion?

13 A. We introduced the tasseled cap, yes.

14 Q. Did Dr. Stein -- I presume it's Dr. Stein?

15 A. Yes.

16 Q. Did he explain to you the environment in

17 which the process you referred to as tasseled cap was

18 developed?

19 A. No, we did not.

20 Q. As I understood your testimony the last

21 time, you had concerns over the use of the Eckert

22 technique by Dr. Jensen to enable him to do his time

23 normalized series?

24 A. Yes.

25 Q. And one of your concerns, as I understood

78

1 it, was that Dr. Eckert developed his technique in a

2 semi arid ecosystem; is that correct?

3 A. It was the area, yes.

4 Q. And that you were concerned -- deeply

5 concerned, I might say, if that's a fair

6 characterization -- over the transference of that

7 technique from a semi arid environment to the sub

8 tropical environment of South Florida?

9 A. Because it required targets, as I clearly

10 stated. It required clear targets over the study

11 period.

12 Q. Are you aware of any literature that

13 suggested that the same target must be used

14 throughout a time normalized series under the Eckert

15 technique?

16 A. My understanding of the technique was -- my

17 recommendation was to use a consistent target.

18 Q. From whence do you derive this

19 understanding? Have you ever yourself employed the

20 Eckert technique?

21 A. No.

22 Q. Who told you that you got to use a

23 consistent target throughout?

24 A. It seems like I remember in the paper that

25 the target should be consistent.

79

1 Q. Are you aware of anything suggesting that

2 you can't use different targets?

3 A. I'm not aware of a paper saying that you

4 can -- what? I'm sorry.

5 Q. Are you aware of anything saying that you

6 cannot use targets that are not consistent throughout

7 the entire time normalized series?

8 A. No.

9 Q. Were you aware that the technique you

10 referred to as tasseled cap was invented or developed

11 in the study of agricultural lands in the midwest for

12 large homogenous tracks of crop?

13 A. They were used in South Carolina and three

14 other states, from what I was aware of.

15 Q. Let me ask you the question again so you

16 can answer the one I asked. Were you aware that it

17 was developed in the midwest based on assessments and

18 studies of large area homogenous crops in

19 agricultural lands?

20 A. You say "developed." I'm not aware that

21 tasseled cap is a single development. Tasseled cap

22 is an ongoing development, from what I read.

23 Q. What are you relying on for that?

24 A. I have read a few papers on tasseled cap

25 and discussed it with Dr. Roy Stein. It's an ongoing

80

1 discussion.

2 Q. Did you read these papers in connection

3 with preparing to employ the technique in this case?

4 A. Yes.

5 Q. Please identify the papers you read since I

6 don't see them in the materials.

7 A. Just papers and literature. The titles of

8 the papers, is that what you want?

9 Q. Yes.

10 A. I would have to find them. I got two

11 papers I read on tasseled cap.

12 Q. Do you have copies of those in your

13 materials at your office?

14 A. Yes, I have copies of the reports, if you

15 would like those.

16 Q. If you relied on those and employed them in

17 assisting you in applying this technique in this

18 case, I certainly do.

19 MR. WATTS-FITZGERALD: Counsel, I request

20 they be provided.

21 MS. RAEPPLE: Have they not been provided?

22 THE WITNESS: I thought they were. I'm

23 surprised if they aren't. A copy of the entire

24 file was produced.

25 Research is an ongoing proposition, but I

81

1 would be surprised they are not in the file you

2 already got. If they are not, I'll provide

3 them.

4 MR. WATTS-FITZGERALD: I indexed the

5 material I got. I don't recall --

6 THE WITNESS: I'll provide them, two

7 studies out of ASPRS one. They are discussions

8 of tasseled cap.

9 BY MR. WATTS-FITZGERALD:

10 Q. What's ASPRS?

11 A. American Society of Remote Sensing

12 Photogrammetry and --

13 Q. Are you a member?

14 A. No, I'm not.

15 Q. I take it you are not certified in

16 photogrammetric interpretations by any professional

17 organization?

18 A. No, I'm not.

19 Q. The publication listed in Exhibits 13 and

20 14, your CV and update, I didn't notice if the update

21 had anything which -- if any of those are peer review

22 journal publications.

23 A. I have done no peer review journal

24 publication except for the review through URISA. No,

25 I'm not in peer review publication.

82

1 Q. What is URISA exactly?

2 A. Urban and Regional Information Systems

3 Association.

4 Q. What is it?

5 A. It's a large group of people interested in

6 information systems, have been around for quite a

7 while.

8 Q. They have a publication?

9 A. Yes.

10 Q. You have contributed to that?

11 A. I presented many papers, as the resume

12 indicates. They have an annual conference similar to

13 ASPRS.

14 Q. Do they produce written results?

15 A. Usually they do.

16 Q. There are outlines of presentations?

17 A. Every year conference proceedings are

18 published.

19 Q. But those are not peer review, they reflect

20 whatever was presented?

21 A. It's reviewed by a group at URISA but not a

22 peer review.

23 Q. You said the tasseled cap could be a tool

24 to distinguish between sawgrass and cattail. That's

25 what you said this morning?

83

1 A. Yes.

2 Q. In what fashion in your estimation could it

3 be such a tool?

4 A. Several ways. One is to actually perform a

5 classification on a tasseled cap image and then you

6 see what those classes are.

7 Another method would be to compare

8 brightness values in a tasseled cap image to a

9 classified image brightness green. You compare

10 brightness, greenness or wetness values to a

11 classified image.

12 Q. So that's how it could be a tool?

13 A. Yes.

14 Q. Have you used it as a tool in either of the

15 two faxes you have described?

16 A. We produce the tasseled capped image

17 combining '85, '87, '89 and '93. I have used various

18 combinations of wetness, brightness and greenness

19 among those images and compared those images to our

20 classification for '85 and '93.

21 Q. I need to break that down. I'm not sure I

22 understand what you did. You merged together the

23 scatter plots, if you will, of brightness and

24 greenness and wetness for all of those years into a

25 single product?

84

1 A. We produced one 12 band image that included

2 '85, '87, '89 and '93.

3 Q. Using what satellite data?

4 A. The '85 the '87 PM image, '89 and '93 PM

5 image.

6 Q. Did you use all bands?

7 A. No. We used the tasseled cap modeling

8 capabilities with the software to produce three

9 bands. In each of those three images, the bands

10 being where -- the new bands being brightness,

11 greeness and wetness.

12 Q. When you say you produced the 12 band

13 image, what form did that take?

14 A. It's a digital image.

15 Q. Did you print that out?

16 A. It's on that tape right there. There are

17 photographs of it.

18 Q. The photographs are taken off the screen?

19 A. Yes.

20 Q. The question was, did you hard copy it?

21 A. No, we produced no hard copy of tasseled

22 cap at this point.

23 Q. Do you plan to?

24 A. If requested to do so, I will.

25 Q. And then as I understood what you said, you

85

1 compared that 12 band image to your '85?

2 A. Various components of the 12 bands image to

3 our classified '85 image and classified '93 image.

4 Q. What components of the 12 band image did

5 you compare to the '85 imagery?

6 A. Greeness and brightness and wetness.

7 Q. What were your conclusions as a result of

8 that comparison?

9 A. Our conclusion is that there is no

10 substantial change in greeness between '85 and '93.

11 Q. You have added in '93. You jumped ahead a

12 little bit. As I understand it, you did the same

13 comparison to your '93 classified image?

14 A. The tasseled cap image is a combination of

15 all of those years. So the comparisons are intrinsic

16 in the image.

17 Q. If you are comparing the 12 band image that

18 includes the two years in which -- and the same

19 imagery data that you have for '85 and '93, aren't

20 you already biased towards limiting any variation or

21 change in the greeness values?

22 A. Could you repeat that question? I do not

23 understand it.

24 Q. You took the three bands from your '85 TM

25 satellite range and put that in the 12 band?

86

1 A. I took the 36 bands excluding thermal and

2 combined that to a three band tasseled capped image.

3 Q. For '85?

4 A. Used that as part of the 12 band composite

5 image.

6 Q. The three bands are -- used tasseled caps

7 out of '85 are derived from the '85 data which is

8 used to generate your classified '85?

9 A. From the raw '85, the unclassified '85.

10 Q. Doesn't that inevitably suggest that you

11 are minimizing any variance? You are taking '85 and

12 '93 and using the same tasseled cap converted as part

13 of the comparison?

14 You don't understand what I mean?

15 A. No, I don't understand.

16 Q. I have reviewed my notes on the materials

17 provided. I can't find any indication of the fact of

18 the publications you referred to.

19 A. At the first break I'll find my copies.

20 Q. I don't know if you need to make the copies

21 if you can get me titles and reference.

22 A. You want that? Okay.

23 Q. In the efforts you described to determine

24 the 12 band image, what original TM data channels

25 were used to produce the 12 band image?

87

1 A. We used all bands in each image except the

2 thermal band.

3 Q. You said earlier that this could be a tool

4 in distinguishing between cattail and sawgrass based

5 on what you just said regarding how you employed it.

6 I don't understand you to say that you, in fact, did

7 that or attempted to do that of your analysis of the

8 '93 image, but you only attempted to determine if the

9 greeness values showed the significant shift, and you

10 concluded the answer to that was no.

11 Am I correct in my understanding?

12 A. You said a lot in that statement. Are you

13 saying -- I'm not sure what you are asking. Please

14 tell me what you are asking about the greeness.

15 Q. As you applied the analysis or the

16 comparison of your TC image to the '85 and '93

17 classified images, did you form an opinion

18 specifically about any alterations in sawgrass and

19 cattail over that time span, those species?

20 A. From the tasseled cap, singularly?

21 Q. Yes.

22 A. No.

23 Q. You told us three times now that you could

24 use the tasseled cap analysis to make such a

25 distinguishing judgment?

88

1 A. That it could be a tool in making the

2 judgment, yes.

3 Q. Have you done that?

4 A. Yes.

5 Q. How?

6 A. Visually we have compared -- we have

7 produced numerous combinations of brightness,

8 greeness and wetness from the 12 band image, and we

9 have produced those and observed greeness in

10 particular and then compared that greeness to our '85

11 classification and the '93 classification.

12 Q. And did I understand you to say the

13 greeness values from your tasseled cap '85 to '93

14 don't show any significant change?

15 A. There is not a major significant change in

16 the areas that -- the significant change -- there is

17 change I guess is significant up to the user. It

18 shows change in areas that we did the classification

19 on to see, in fact, what the greeness was at that

20 point. It showed changes in greeness. It indicates

21 greeness values in those two years.

22 Q. Can you quantify the changes you observed?

23 A. We did not attempt to quantify the changes

24 in the tasseled cap image.

25 Q. Do you plan to?

89

1 A. We could if asked.

2 Q. I take it that's a no. The question is, do

3 you plan to. Not could you do it. If somebody asked

4 you, do you plan to do it?

5 A. I'm doing what my client asked me to do. I

6 do what I'm asked to do. I can do it or not do it.

7 I'm not sure if I understand.

8 Q. Let me ask the question. Has anybody asked

9 you to do that?

10 A. No, not yet.

11 Q. Do you of your own sua sponte plan to do

12 that?

13 A. No.

14 MR. WATTS-FITZGERALD: Counsel, if you ask

15 him to do that, obviously we feel entitled to

16 certain notice on that further deposition.

17 MS. RAEPPLE: Certainly.

18 BY MR. WATTS-FITZGERALD:

19 Q. In your analysis where you found some

20 unquantified as yet shift in greeness values for 1985

21 imagery to the 1983 imagery --

22 A. '93.

23 Q. -- how much of that greeness shift can you

24 attribute within a reasonable degree of scientific

25 certainty to expansion of cattails over that time

90

1 frame?

2 A. How much of the changes in greeness can I

3 attribute to expansion of cattail?

4 Q. Yes.

5 A. The acreage figures that we produced would

6 be that combination. Those are digital files that

7 can easily be overlaid to each other. If you want to

8 quantify it pixel by pixel, we can do that.

9 Q. Let me make sure I'm clear on this. The

10 original calculations you are talking about, that's

11 based on the tasseled cap?

12 A. No, it's based on the classification.

13 Q. You are answering a question I didn't ask.

14 I'm talking about tasseled cap, based on tasseled cap

15 and the shifting greeness values, which has nothing

16 to do with your accuracy assessment.

17 In your acreage assessment in Exhibits 10,

18 11, 12, you have said you have observed a greeness

19 value shift but you can't quantify it, but there is a

20 shift over those years '85 to '93.

21 You told me the tasseled cap is a tool to

22 distinguish between cattail and sawgrass in the

23 greeness for area shifts.

24 How much of it is attributable to the

25 increase in cattail?

91

1 A. I did not tell you that tasseled cap would

2 quantify species. I said it was a tool used to help

3 assist us in quantifying the species attributable to

4 a particular species. That's what I said. Tasseled

5 cap is not -- we did not assign vegetative classes to

6 the tasseled cap images.

7 Q. Do you plan or did you attempt that in any

8 fashion thus far?

9 A. No, we have not.

10 Q. Has anyone directed you to do that?

11 A. No.

12 Q. Do you have any plans to do that at this

13 juncture?

14 A. At this point, no.

15 Q. Would you have to go through the same

16 process with color IR or would you conduct that

17 effort by reference to your classified '85 and '93

18 images?

19 A. You could do it either way.

20 Q. Which way would you do it?

21 A. I would compare it to the classified

22 images.

23 Q. It's easier. You have already got that

24 digitized, right?

25 A. We do have that.

92

1 Q. Cut Dr. Lodge out of the loop, he doesn't

2 have to sit there on the light table.

3 A. I would compare it to the classified

4 images, yes.

5 Q. So as of right now, as I understand it --

6 correct me if I'm wrong -- you have no opinion based

7 on tasseled cap as to the relative change or

8 distribution within those greeness values for '85 and

9 '93 of cattail and sawgrass?

10 A. That's correct. We did not classify the

11 tasseled cap based on species type.

12 Q. So what opinions have you formed and

13 developed based on your tasseled cap analysis that

14 are final at this time?

15 A. Tasseled cap similar to the unsupervised

16 classifications we did increase our general knowledge

17 of the study area.

18 Q. And as I understand it, it has been

19 suggested to you that there has been some change in

20 the coverage of certain species. The extent of that

21 is not quantified in terms of, say, acreage?

22 A. Not quantified in the tasseled cap but in

23 the '85 or '93 image, yes.

24 Q. We were speaking in terms of cattail and

25 sawgrass. But there are, according to Dr. Lodge's

93

1 book, something on the order of 190 other vegetative

2 species in the ecosystem at issue here.

3 How would you distinguish out from the

4 tasseled cap analysis as you have done it thus far,

5 unclassified other species of flora that have their

6 own associated greeness and brightness values?

7 A. If we did that, we would do classification

8 of the tasseled cap image.

9 Q. That's the only way to distinguish that

10 out?

11 A. We could overlay it to our existing

12 classification and see if there was correspondence.

13 Q. Have you done that, even on the screen?

14 A. No.

15 Q. Other than the photos, we have all of your

16 tasseled cap analyses on this back-up data?

17 A. That 150 megabyte tape contains the

18 complete tasseled cap model that we used.

19 Q. What system was the 150 meg produced on?

20 A. A SUN micro station work station. SPARC,

21 2, 150 megabyte, quarter inch tape unit.

22 Q. What is the tape format?

23 A. Probably TAR format. TAR back up. If it's

24 not, call us and we will tell you. I think it's TAR.

25 Q. Assuming for the sake of argument for this

94

1 question that the quantification of the greeness

2 change is relatively minor. Do you have any way

3 currently on the work you have already completed to

4 assess the greeness values you had developed from the

5 TC effort?

6 A. William Thomas or Bill Thomas, that's the T

7 you are talking about, TC tasseled cap. KT is the

8 other one.

9 Q. Do you have any way of assessing within a

10 reasonable scientific certainty whether or not the

11 mix of cattail and sawgrass species within the

12 greeness value has altered the proportion of those

13 species, if you will, from '85 to '93 based on the

14 work already completed?

15 A. Based on what we already completed today,

16 is that what you are asking?

17 Q. Yes.

18 A. Yes, I believe it did.

19 Q. On what do you base that belief?

20 A. As previously stated, we could overlay the

21 tasseled cap brightness categories and assign species

22 typed to the tasseled cap.

23 Q. In your answer just now you said "could."

24 My question was "done." You have not done that then?

25 A. I thought you said "could we."

95

1 Q. No. I am assuming the difference is very

2 small. Can you right now based on the work already

3 done tell me what the ratio shifts are or what shift,

4 if any, in the relative composition within the

5 greeness values is accounted for from '85 to '93 by

6 cattail and sawgrass?

7 A. By "can," you mean have I produced that

8 product already? My interpretation of "can" was

9 could I take the data I got and come up with that

10 answer.

11 Q. It's a semantic thing. Put the question

12 this way.

13 Tell me right now what the ratio shifts

14 were based on the greeness evaluation from '85 to

15 '93.

16 A. I understand your question now. We have

17 not done that as of right now.

18 Q. It was the way I asked the question. We

19 just look at the words a little differently.

20 A. Okay.

21 Q. The software used to produce the tasseled

22 cap analysis was still ERDAS?

23 A. ERDAS, yes.

24 Q. What version?

25 A. This model was run in the current release,

96

1 which was 8.1.

2 Q. The demonstration tape that was produced as

3 part of your discovery, it showed Mr. Erickson

4 playing with the monitors doing his golly, gee whiz

5 stuff. That was a later version not yet released?

6 A. That was one, the beta version, probably

7 8.02.

8 Q. Earlier for Mr. Cesarano you were talking

9 about the 2A image and said that you refined the

10 categories as part of your very recent work?

11 A. Yes.

12 Q. Were you talking about '85 or '93?

13 A. '93.

14 Q. How did you refine the categories?

15 A. We observed that we could break out

16 predominant -- which one was that?

17 Q. The next one down.

18 A. We determined in reviewing these classes

19 that we could break out predominant cattail open

20 water and predominant sawgrass open water. We added

21 those to the classes that we exhibited in the legend.

22 Q. Did that require you to go back to the

23 field notes?

24 A. Let's see. We reviewed field notes

25 continuously through this. We probably did, yes.

97

1 Q. And who actually conducted that exercise to

2 class the categories?

3 A. I did.

4 Q. Assisted by anyone?

5 A. Pamela Green and Terry Jackson.

6 Q. Dr. Lodge was not one of those?

7 A. Dr. Lodge has been involved in quite a bit

8 of discussion. Yes. Dr. Lodge would have had some

9 input into this, yes.

10 Q. When I look at the image, is it fair to say

11 that everything I see on the image is as a result of

12 platting out of the digitize data that has been

13 classified?

14 A. I am not sure I understand. As a result of

15 what?

16 Q. As I understand the image, it is intended

17 to convey through the color schemes and

18 classification process information to the reviewer?

19 A. Yes.

20 Q. So everything depicted on there has

21 significance to the viewer, even if it's down to the

22 cloud category which tells you what was under it?

23 A. Yes.

24 Q. If I see banding on any of these images

25 that implies to me it's in the digitized data or it's

98

1 an artifact of the platting; is that correct?

2 A. I would say that's right.

3 Q. What is the inclination to its orbital path

4 for the satellite of the sensors employed to produce

5 these two images, the '93 images?

6 A. What?

7 Q. Inclination of the sensors to the orbital

8 path of the satellite?

9 A. Inclination of the sensors to the orbital

10 path --

11 Q. What satellite are we talking about?

12 A. Landsat 5.

13 Q. Landsat 5's sensors, are they normal to the

14 footprint of the satellite?

15 A. Are they off NADER, are they pointed other

16 than straight down? Is that what you are asking?

17 Q. Yes.

18 A. They are pointed straight down. Landsat 5

19 does not have a pointing capability. Therefore, the

20 sensors are always aligned in the same direction from

21 the satellite, if I understand your question.

22 Q. Then if there were -- if banding were

23 present you would expect it to be consistent

24 throughout the areas that you mapped if there were

25 banding as a result of the sensor?

99

1 A. Expected to be consistent within that

2 image, yes.

3 Q. If you look at the image for the entire

4 study area, do you see banding in what would be the

5 Loxahatchee National Wildlife Refuge?

6 A. I do see what could be interpreted as

7 banding in the upper portion of Loxahatchee. If you

8 are speaking of this area here, is that what you are

9 talking about, here?

10 Q. The entire area of Loxahatchee.

11 A. I do not see indications of banding in the

12 southern area or 2A.

13 Q. When you hold the paper chart and look at

14 it end on, very close to end on, you still don't see

15 it?

16 A. We printed these on a -- these are color

17 electrostatic plats, and the color electrostatic

18 platting that we are using prints in rows and our

19 observation of the image on the screen was that there

20 was no significant banding.

21 I can see what you are talking about there

22 and that does have the appearance in the upper

23 Loxahatchee of possible banding in the image, but we

24 never observed it in the image on the screen and I

25 cannot tell you where that came from.

100

1 Q. I guess the problem I have when I look down

2 at WCA-3A on that image I see bands from whatever

3 source running more or less parallel east/west. And

4 when I look at Loxahatchee on the same image,

5 certainly in the northern half of it above the cloud

6 cut out, I see what appear to be bands running

7 northwest, slightly southeast.

8 If there is an artifact of your platter, I

9 would expect them to be consistent throughout the

10 entire range of the imagery. It was platted at one

11 time?

12 A. This is what you are talking about, right

13 here.

14 Q. Show me what you are talking about in the

15 image.

16 A. I see the banding effect right here. Since

17 I think you are wrong on the satellite you are not

18 interpreting what I'm talking about, the platter

19 effect which is straight horizontally.

20 Q. Down in 3A we are talking east, west or --

21 A. All right. Let me see this down here. In

22 this part of the image, what are you talking about,

23 right here.

24 Q. East/west, there are triangles, if you

25 will, in the imagery?

101

1 A. You mean this little light thing here?

2 Q. No. I see it continuously.

3 A. Where do you see it?

4 Q. I see it like every 16 lines.

5 A. Every 16 lines?

6 Q. How many sensors are there on a Landsat?

7 A. There are 7, 7 bands. I can see something

8 in this image. I do not know -- I cannot attribute

9 why that's there.

10 Q. And we can both see something in the

11 Loxahatchee part of the image?

12 A. I see something up there, yes. There is

13 some.

14 Q. Do those striations which I use, because it

15 doesn't have a normative value what the cause is,

16 they don't all run in the same direction, which is

17 roughly parallel to those in 3A, and it seems to me

18 this is an artifact of the printer. You just

19 suggested your printer prints in rows, which I assume

20 is roughly east/west normal to the paper?

21 A. Yes.

22 Q. That would not seem to account for

23 striations for Loxahatchee. I want to know what they

24 are. Are they an artifact, are they in the data or

25 is there some problem in the classification we have

102

1 not considered?

2 A. Is your question, do I know what that is?

3 Q. Yes, or can you explain that for me.

4 A. The artifact you are observing is something

5 either in the data of the platter, it looks like from

6 the angle of it. It's in the data. At least in

7 northern Loxahatchee during our classification

8 process we observed no banding that caused us

9 difficulty with the classification.

10 Frankly, I don't see -- I don't observe

11 that banding at that angle referred to, I don't

12 observe that in the southern study area. I don't see

13 it. I see perfectly horizontal lines and I do see

14 that slightly off horizontal artifact you referred to

15 here. But I do not see that here.

16 Q. Did you process the portion of the data

17 involving Loxahatchee in some way different from the

18 balance of the imagery as a result of the cutting

19 into the middle of Loxahatchee?

20 A. That was not separated out as an area of

21 interest or digitized out in any fashion from the

22 classification that occurred in this portion of the

23 study area around 2A. It was not treated separately.

24 Q. Other than the tasseled cap effort, has any

25 classification been done on images other than '85 and

103

1 '93?

2 A. Yes. We did an unsupervised classification

3 on '87 and '89.

4 Q. As I understand what you said, while you

5 classified them to the 36 and 99 class, no effort was

6 made or expended to identify vegetative species with

7 those classes?

8 A. Correct.

9 MR. WATTS-FITZGERALD: He has a call to

10 make and I'm at a good stopping point. It's a

11 quarter to 12. I would like to review these

12 materials before we continue. So if we can take

13 an early lunch and get back early --

14 MS. RAEPPLE: Do you want to take a short

15 lunch?

16 MR. WATTS-FITZGERALD: Let's come back at

17 12:45.

18 (Thereupon, a lunch recess was taken,

19 after which the following proceedings

20 were had.)

21 BY MR. WATTS-FITZGERALD:

22 Q. Your imagery for 1985, that is complete and

23 final as I understand it; is that correct?

24 A. Yes. Yes, it is.

25 Q. Did you yourself do any of the photo

104

1 interpretation in the process of developing the

2 classified images?

3 A. No, not directly. I looked at the photos

4 and gained as much familiarity as I could. But that

5 was left for Dr. Lodge.

6 Q. You stated you were going to compare Ken

7 Rutchey's procedures and results to yours or to your

8 effort.

9 Other than the limited discussion we had

10 about comparing acreages and classes and whether you

11 could overlay them or not, how you would do that?

12 Have you done anything else to analyze and compare

13 your methodologies of the final product?

14 A. To Ken Rutchey?

15 Q. Yes.

16 A. You know the GPS procedures. I stated some

17 differences in those, the way he produced classes.

18 We looked at those things, yes, if that's what you

19 are asking, yes.

20 Q. What is your opinion on the level of

21 positional error that might exist in Ken Rutchey's

22 work based on your understanding of his GPS

23 procedures?

24 A. What is my estimation of his accuracy?

25 Q. Yes.

105

1 A. I think in those -- it's really hard to say

2 post processed data. I would say he may be close to

3 what he said he was when he post processed the data.

4 My question would be how he obtained

5 information about the site when he was doing the

6 accuracy assessment. If he did have to come back and

7 post process the data and had to move that location

8 to another location, how he correlated that field

9 research with the point in terms of defining, you

10 know, the vegetative mix around the helicopter.

11 Q. You think he did all of his accuracy

12 assessment field work from the helicopter?

13 A. He stated he did a mixture of boats and

14 helicopters. He wasn't sure of the percent.

15 Q. The field points, the 230 points that were

16 generated by his random stratified sampling program

17 for accuracy assessment, are you aware of any size

18 constraints that would act to limit the possibility

19 of being in an erroneous field position in the way

20 they could affect the result prior to post

21 processing?

22 A. Am I aware of the positional error that

23 might have occurred -- one more time.

24 Q. How the point was defined, the accuracy

25 assessment point, to minimize any potential problem.

106

1 A. You mean the size of the matrix he chose

2 around the area?

3 Q. Yes.

4 A. I'm aware that he chose a matrix around the

5 area.

6 Q. Do you know how large?

7 A. I believe he chose -- let's see. I believe

8 that was 3 by 3 he chose in the field work. I'm not

9 certain.

10 But the non-post processed GPS data could

11 easily be 100 feet. I think he mentioned in the

12 paper that SA was sound, so he had to adjust those

13 sites. I assumed SA was sound.

14 Q. You have the downloaded satellite almanac.

15 Was it on or not?

16 A. The almanac doesn't tell you whether it's

17 on or not.

18 Q. Where would you look to find that?

19 A. The government, the defense mapping agency

20 will not tell you when SA is on and when it's off, to

21 my knowledge. The reason they use selective

22 accessibility, SA, is to intentionally confuse the

23 signals from the satellite. If they issued a -- the

24 purpose of SA is to confuse the signal. If you

25 generate the almanac in advance of the mix so there

107

1 would be no way to know when you generate the almanac

2 if SA is not on. The user is not notified when the

3 SA is turned on or not.

4 Q. In your accuracy assessment effort did you

5 exercise any techniques to attempt to account for

6 variability and results induced by the phrenology of

7 the flora as a result of the time interval between

8 the satellite passes and the color IR photography?

9 A. Other than Tom Lodge's knowledge of the

10 plants themselves, we took no mechanical means in the

11 image to adjust for that. Tom would do that in the

12 photo interpretation.

13 Q. How do you understand him to have done

14 that?

15 A. His knowledge of the plant material. He

16 can recognize cattail in sawgrass communities because

17 of the difference in the IR response.

18 Q. Even in the 1 to approximately 65,000 CIR

19 from 1985?

20 A. Yes.

21 Q. Well, I thought you told me that the reason

22 you classed classes was that fine distinctions became

23 very difficult to make in mixed communities in that

24 photography.

25 A. Yes.

108

1 Q. You testified you had 116 total observation

2 sites. This is your testimony last week. Would that

3 have been training sites, the total training sites,

4 116 --

5 A. Yes.

6 Q. -- we couldn't remember before lunch?

7 A. Yes.

8 Q. That extended throughout the area of study?

9 A. Yes. Some of those were removed. If, in

10 fact, the -- if the GPS was not sufficient or we

11 could not locate where it was with certainty, some of

12 those we did not use. So 116 original observations

13 to be used in this process. And through the process

14 of producing classified image, some number dropped

15 out.

16 Q. Do you know how many?

17 A. It's on the additional tape I sent you. I

18 don't know exactly how many.

19 Q. More or less than 10%? Do you have any

20 feel for it?

21 A. I guess it would be 10%.

22 Q. The approximate 12 to 18 observations you

23 testified last week were identified in the

24 Loxahatchee refuge were made from a helicopter in the

25 air; is that correct?

109

1 A. Yes.

2 Q. Were you on Board the helicopter at the

3 time?

4 A. Yes, I believe so.

5 Q. How high was the helicopter off the ground?

6 A. We were limited to 500 or 1,000 feet. The

7 pilot set the altitude. Whatever the limit that the

8 Park said was the limit. I think it was 500 or 1,000

9 feet.

10 Q. What's your assessment of the inaccuracies

11 induced in the GPS coordinates as a result of

12 attempting to figure the position of those ground

13 truth points or air hover points, whatever you want

14 to call them?

15 A. It could be substantial.

16 Q. Substantial is an adverb or adjective. Can

17 you qualify that for me?

18 A. It could be a matter of 20 meters or 100

19 meters.

20 Q. That's a differential correction?

21 A. You are talking about the accuracy of

22 hovering over where you think you are with GPS.

23 Q. I'm assuming so. Maybe I'm wrong here.

24 When you testified those were observation points, you

25 used those to train your algorism for classification?

110

1 A. Yes.

2 Q. So you needed a GPS position to go with

3 that?

4 A. Yes.

5 Q. You took that while hovering for some

6 period of time in the air?

7 A. Yes.

8 Q. They were manifestly going into a number of

9 sources of error in that.

10 Were you able to hover in a fixed spot for

11 the same period of time that was employed at the

12 ground stations to allow averaging of the

13 constellation mix?

14 A. How long? Enough to allow averaging. I

15 would not call the quality of the hovering point

16 equal to the quality of the ground.

17 Q. That's what I understand. I'm asking, can

18 you quantify that either pre or post -- post

19 processing?

20 A. We didn't quantify them. You probably

21 could, yes. We did not quantify the error in the

22 hovering points.

23 Q. With so relatively few air hover truthing

24 points in Loxahatchee in comparison to WCA-2A, is the

25 classification for Loxahatchee largely a result of

111

1 the extrapolation of spectral signatures from other

2 areas within the study area?

3 A. Other signatures outside of the study area

4 did have an influence on the Loxahatchee

5 classification, yes.

6 Q. And in the way the algorism functions and

7 the way you trained the algorism, would the fact that

8 the bulk of -- I think the majority, almost all of

9 the tracking sites were in 2A, meaning that 2A would

10 have the biggest impact in the training regime on 1?

11 A. That's possible, yes.

12 Q. If the vegetative communities in those two

13 adjacent areas are, in fact, significantly different

14 based on whatever natural factors and unnatural

15 factors are operating, does that have the potential

16 for substantially reducing the reliability of the

17 classified image for Water Conservation Area 1?

18 A. If the vegetative communities were

19 different would it reduce the reliability of the

20 classification in Loxahatchee?

21 Yes, it would.

22 Q. In the two matrixes you have provided for

23 accuracy assessment, I see no footnotes or any

24 commentary that addresses that or attempts to account

25 for it.

112

1 Was any effort made to account for that or

2 consider that facet?

3 A. Notes in the accuracy assessment table.

4 Q. There is nothing in the matrix. You have a

5 warning on one of the charts unverified, outside of

6 2A, or something to that effect.

7 I have seen nowhere in the materials any

8 effort. Of course I haven't been able to go through

9 the tape, but any effort to take into account the

10 effects of signature extrapolation, was that done?

11 A. No.

12 Q. Do you have any current plans to do that?

13 A. I have not been directed to do that.

14 Q. As I understood from your earlier testimony

15 there were 58 ground sites, training sites in 2A?

16 A. Okay.

17 Q. If you had used all 116, assuming for the

18 moment, we don't know exactly how many fell out that

19 would have been half of the total sites initially and

20 then some fell out, do you know if any of the 58 in

21 2A fell out of the training sites?

22 A. A few in 2A did fall out.

23 Q. It was across the area?

24 A. Yes.

25 Q. You stated with respect to Dr. Jensen's

113

1 report to the Water Management District, Exhibit 8

2 from the last session, that in your view it was

3 impossible to do an accuracy assessment on his work

4 and certainly none accompanied it.

5 Are you still of that view?

6 A. I'm sorry, which document.

7 Q. Dr. Jensen's report to the Water Management

8 District.

9 A. Yes.

10 Q. Are you still of the view that it is

11 impossible to do an accuracy assessment for such an

12 effort?

13 A. On the historic MSS data that he

14 classified?

15 Q. Yes.

16 A. As I stated before, if he had photography

17 for that time, then he could do an accuracy

18 assessment in the same fashion we did. I believe I

19 stated he did not do an accuracy assessment.

20 Q. You expressed a concern with respect to

21 Dr. Jensen's work that he use the two different

22 platforms with very little overlap in the band of

23 those platforms.

24 You were at Dr. Jensen's initial

25 deposition, were you not?

114

1 A. One day of it.

2 Q. And have you had an opportunity to review

3 the transcript of that deposition since then?

4 A. No, I have not.

5 Q. Do you recall his testimony regarding the

6 extent of band overlap?

7 A. Yes.

8 Q. Do you remember the reference to his

9 textbook -- actually it may have been the draft of

10 his revised textbook, the second edition, regarding

11 the actual band waive length for those respective

12 platforms quantifying on delineating the overlap?

13 A. I remember a discussion about band overlap,

14 yes.

15 Q. I guess my question, then, would be, my

16 notes indicate you said there was very little overlap

17 in the bands. Is it not a fact that in some of the

18 bands there was perfect overlap?

19 A. My statement, if I remember correctly, was

20 that the band 3 and 4 in MSS do not match band 3 in

21 SPOT, and they don't.

22 Q. That's taking into account the renumbering

23 of the bands?

24 A. The bands in SPOT do not only have a

25 partial overlap with bands 3 and 4 in MSS.

115

1 Q. Do you disagree with Dr. Jensen's

2 assessment, expressed during his deposition at which

3 you were present that the overlap was sufficient for

4 his purposes and sufficient to allow use of it as a

5 tool in satellite imagery processing?

6 A. He said it was sufficient in his work to

7 adjust signatures between the two satellites, not

8 just in general image processing.

9 Yes, I disagree with that.

10 Q. And on what do you base your assessment

11 that the overlap was insufficient?

12 A. The overlap, I didn't say, was

13 insufficient. I said there was an overlap.

14 Q. You just told me you don't think it was

15 appropriate for him to do it that way. On what basis

16 are you asserting that the overlap that did exist

17 which you have acknowledged existed to some degree

18 was not sufficient to allow that to appropriately be

19 used as it was by Dr. Jensen?

20 A. Because it is only a partial overlap

21 between one band and two bands. It's confused also

22 by the pixel size of the two satellites.

23 Q. Is it ever appropriate to use less than

24 perfect band overlap? Is there a point at which you

25 would be comfortable in using data such as that that

116

1 doesn't have a perfect identity of band width?

2 A. Are you speaking of image interpretation?

3 Are you speaking of trying to transfer a set of

4 spectral signatures from one satellite to another?

5 There is a difference.

6 Q. Start with image interpretation.

7 A. Image interpretation, you could probably

8 account for some degree of band overlap or lack of

9 total band overlap.

10 Q. How much?

11 A. I don't know.

12 Q. You never researched it?

13 A. No, I have not researched it.

14 Q. Are you aware of any literature addressing

15 that?

16 A. Addressing bands? Most of the tasseled cap

17 mentioned something about that, talks about the fact

18 between MSS and TEM band overlap or the combination

19 of those two satellites is usable. They make no

20 mention of SPOT.

21 Q. Anything to suggest that SPOT is not

22 usable? I mean, if you can overlap the other two

23 methodologies?

24 A. The difference in my statements, as I tried

25 to make clear in the deposition before, was that in

117

1 order to produce a set of signatures that are going

2 to be transferred from one satellite to the other,

3 the number of variables should be decreased.

4 When you have two bands in the MSS

5 satellite platform that have marginal overlap with

6 one band in the SPOT platform you have to make an

7 assumption of what something would look like within a

8 band that is different at different portion of the

9 spectral -- of the reflectance arena than the band

10 from which you produce the original signature.

11 Q. You just expressed a value judgment on the

12 overlap by characterizing it as marginal overlap. I

13 wanted you to tell me on what basis do you make the

14 judgment about the differences in overlap.

15 When is it only marginal, when is it

16 adequate? That's the area I'm trying to illuminate

17 and I'm not getting that through to you -- or at

18 least the answer doesn't seem to address what I'm

19 trying to find out.

20 A. I'm not sure I understand your question.

21 Q. Is there a number in your mind at which you

22 have only marginal overlap between two bands, whether

23 it be TM and MSS or TM and SPOT or MSS and SPOT?

24 Is there a point where you are saying

25 that's marginal in my mind, I wouldn't use it or it

118

1 can't be used?

2 A. I have never quantified what percent band

3 overlap between SPOT and MSS is appropriate because,

4 as I stated earlier, I did not attempt to do any

5 signature transformation between two satellite

6 platforms.

7 Q. You never attempted to do that?

8 A. No.

9 Q. Are you aware of anybody in the literature

10 doing that?

11 A. The only person as I stated before that I'm

12 aware of doing that is Dr. Jensen.

13 Q. You mentioned tasseled cap and some

14 literature on that discussing TMM and MSS.

15 Dr. Jensen didn't use tasseled cap, that's not

16 directly applicable?

17 A. Is what directly applicable?

18 Q. The literature you were mentioning or the

19 discussion of tasseled cap and the use of TMM and

20 MSS.

21 A. I'm sorry, I don't understand what you are

22 asking. Does it have anything to do with the Eckerd

23 method of using creating historic signatures?

24 Q. No. We are talking about your comment

25 about the literature in tasseled cap discussing the

119

1 use of the two different methodologies. Let's leave

2 that for a second.

3 When you were preparing for your

4 deposition, did you review all of the documents that

5 were provided in your production of materials?

6 A. Is that what he gave me?

7 Q. That's what you gave me. Who provided you

8 the article on the expert witness and litigation?

9 A. That's a corporate piece of literature.

10 Q. Have you read it?

11 A. Frankly, no, I haven't.

12 Q. What are the wave ranges -- wave length

13 ranges in each band of the Landsat satellite?

14 A. You have to provide that -- which Landsat,

15 which one, MSS or TM?

16 Q. I'm going to ask both. Start with TM.

17 A. I don't know those offhand.

18 Q. Same for MSS?

19 A. What?

20 Q. Same answer for MSS, you don't know the

21 band?

22 A. I know the band. I mean, I know red,

23 green, blue bands are the ones I use the most. I

24 don't know the frequency response.

25 Q. Is the same thing true with respect to

120

1 SPOT?

2 A. Can I quote you the point to point in each

3 band, is that what you are asking?

4 Q. Yes.

5 A. No, I can't quote that offhand.

6 Q. What did you mean when you said -- I

7 started to ask this earlier today and we got side

8 tracked. Dr. Rutchey's map was biased towards

9 cattail and Jensen's is biased towards sawgrass?

10 A. Based upon my interpretation of their

11 deposition.

12 Q. Were you aware Dr. Jensen refined his work

13 and substituted or altered his report to reflect

14 additional classification efforts?

15 A. No, I wasn't aware of that. He changed his

16 classification?

17 Q. Well, what was it about his work that held

18 you to believe there was a bias towards sawgrass?

19 A. His own statements about his methods used.

20 I described those to my understanding in the last

21 deposition. Has his work been revised?

22 Q. You have to wait until you depose me for me

23 to answer questions.

24 Did you try and find any map photography

25 for use with your '85 image from different seasons?

121

1 Were there any other available high altitude

2 photography?

3 A. In the NAP photography?

4 Q. Yes, other than the two that you used.

5 A. It seems like the NAP started after '85.

6 Q. One of the imageries was before the data?

7 A. NHAP and there is NAP. There are different

8 ones.

9 Q. You used NAP or NHAP?

10 A. National Altitude -- NAP, was a later

11 program that started after '85. We looked at both

12 indices and NHAP was the only one available that I

13 remember for the year '85.

14 MR. WATTS-FITZGERALD: Can we mark this as

15 Exhibit 15 and 16. 15 will be the accuracy

16 assessment matrix for 1985, and 16 the accuracy

17 assessment for 1993, which is also a matrix but

18 that's not the formal title on the papers that

19 were provided to us at lunch break.

20 (The documents referred to were thereupon

21 marked Exhibits 15, 16 for Identification.)

22 BY MR. WATTS-FITZGERALD:

23 Q. I read this to suggest that for 1985

24 imagery the assessment was based on a total of 39

25 assessment points; is that correct?

122

1 A. Yes.

2 Q. And for '93 the total number was 84

3 accuracy assessment points?

4 A. Yes.

5 Q. How did you statistically arrive at those

6 numbers as being adequate to determine the accuracy

7 of the two maps?

8 A. My professional judgment.

9 Q. No statistical test whatsoever?

10 A. On the image, you mean?

11 Q. No statistical test to determine such as

12 that -- are you familiar with Mr. Rutchey's

13 description how he developed his value for the

14 minimum number of points necessary to demonstrate map

15 accuracy?

16 A. As described in his report?

17 Q. Yes.

18 A. I remember reading that.

19 Q. He referenced Cochran's paper from 1978 on

20 binomial probability formulas and suggested that the

21 number of points needed to -- just for 2A to have an

22 85% map accuracy level with an error of plus or minus

23 35% was a minimum of 204. Do you agree with that?

24 A. Do I agree with that recommendation?

25 Q. With that assessment, yes.

123

1 A. That would be a nice number.

2 Q. And he, in fact, used 241 assessment

3 points, which would exceed the number, would it not?

4 A. Rutchey used a number exceeding that?

5 Q. In an accuracy assessment, is it fair to

6 say the greater number of assessment points employed

7 and verified, the greater confidence can be placed on

8 the imagery?

9 A. If the procedure supports the precision of

10 location of those points.

11 Q. Procedures being equal, then, the more the

12 merrier in terms of accuracy?

13 A. Procedures being equal to a point, the more

14 the merrier, as you say.

15 Q. I understand if you check 100% of pixels,

16 you have perfect accuracy.

17 You are expressing on Exhibits 15 and 16 a

18 confidence level in the accuracy of these two maps

19 that range from 70 to 72; the 70 being for 1993 and

20 the 72 being for 1985?

21 A. Yes.

22 Q. In your professional expertise, do you

23 believe that that level of accuracy meets the

24 standard in your field or in the field of remote

25 sensing vegetative map classifications?

124

1 A. I don't know there is a standard -- that

2 there is a standard accepted accuracy level for the a

3 map.

4 Q. I understood from your earlier testimony

5 and also understood you to say it's based on a

6 professional judgment assessment and somewhat project

7 specific.

8 What I'm asking you to do is to apply those

9 two bases for evaluation and tell me if someone in

10 your field objectively evaluating those two

11 parameters which you have described would find these

12 two accuracy assessments such that they would propose

13 a high degree of reliability on these two maps.

14 A. I think that any person in my field

15 reviewing these accuracy tables will assess the

16 accuracy at what we said it is, and that is the

17 accuracy of the map. They would say the accuracy of

18 the map is what we said it is.

19 Q. As I understood your testimony, and I want

20 to verify now -- I got the matrix in front of me.

21 Both accuracy assessment point selections of 39 and

22 84 were part of a larger field of accuracy assessment

23 points generated by a stratified random sampling

24 program?

25 A. Yes.

125

1 Q. What effect does it have by taking only 39

2 and 84, respectively, you have a significant number

3 of classes within each of those two assessments with

4 two or less sampling points?

5 A. What is the significance of the fact there

6 were two or less?

7 Q. How does that affect the result?

8 A. How do two or less samples being tested

9 affect the map accuracy result?

10 Q. I looked through these. I see a total of

11 four sites out of -- four classes out of a total of

12 20 that had no sampling points in it at all and the

13 significant additional number more than that that had

14 one or two -- which in the abstract seems like a very

15 small number.

16 A. It's a small number.

17 Q. That makes me think statistically it's a

18 very low power analysis.

19 A. Statistically since it was a stratified

20 random sample the areas that were very low acreage in

21 the classification could have very few possible

22 potential accuracy assessment signs associated with

23 them. So that would be lower.

24 Q. Is there any way for me to tell -- looking

25 at Exhibits 15 for 1985, does that suggest that on

126

1 the wire access, which is the photo taken as true

2 here for class No. 7 showing 20 correctly identified

3 interpreted image, that's what the shadow means,

4 right?

5 A. Yes, that's correct.

6 Q. And then six were incorrectly interpreted

7 for a total field of 26 true No. 7's, which were

8 predominant sawgrass. That's how the matrix works?

9 A. Yes.

10 Q. You have two error columns next to the

11 total of 26, one of 23.08, one 9.09. Is the first

12 one an omission error?

13 A. There is an error on the label. Those

14 columns, the first column is an omission. The next

15 column is commission, and the final column is

16 accuracy.

17 Q. I see.

18 A. The '93 has the correct column titles for

19 some reason, '85 does not.

20 Q. So what this implies to me is for class 7,

21 which you say because this is a stratified random,

22 must be the class with the greatest acreage?

23 A. Stratification is based upon pre-dominance

24 of that classification.

25 Q. So in the classification with the greater

127

1 acreage within the image, this is for just 2A, you

2 have a 76.92% accuracy?

3 A. In the sawgrass category.

4 Q. In the predominant sawgrass category?

5 A. Yes.

6 Q. You have 100% accuracy in the predominant

7 cattail, of which there were two sample points,

8 accuracy points?

9 A. Yes.

10 Q. Now, exactly what does the 33.33 commission

11 error mean for the predominant cattail accuracy

12 assessment?

13 A. For predominant cattail in '85?

14 Q. Yes.

15 A. Which column?

16 Q. The first true photo, predominant cattail

17 No. 1. You had a total of 2.0 omission error,

18 commission error 33.33%.

19 A. You see that column under 1? What that 33

20 means is that one point was interpreted in the

21 photography as a different category. 33%. Those are

22 our percentages.

23 Q. If you had extended this effort and

24 continued beyond the 39 to the field of assessment

25 points, would you have expected to have assessment

128

1 points in all classes?

2 A. Yes, we would. And they would be based

3 upon the classes in the classified image.

4 Q. By its ratio?

5 A. By the stratified random sampling, yes.

6 Q. Did you do a KAPA analysis for this?

7 A. No, we have not done one yet.

8 Q. Does that mean you are going to?

9 A. Well, the situation is this. We generated

10 a large number of potential stratified random

11 samplings that could be used. We used those points

12 in these accuracy assessments.

13 Once we closed the table in the imaging

14 software, it closes the table. You can't add to it,

15 you can't modify to it. KAPA is one of the functions

16 in that table.

17 So until I'm directed to close the table,

18 close the accuracy assessment, then I haven't run a

19 KAPA.

20 MR. WATTS-FITZGERALD: Counsel, do you plan

21 to direct the firm or the witness to conduct

22 that analysis because --

23 MS. RAEPPLE: Unless prohibited by the

24 hearing officer, yes, I would like him to do

25 further accuracy assessment.

129

1 If that occurs, we will provide it to you

2 and provide opportunities for further inquiry.

3 MR. WATTS-FITZGERALD: We will deal with

4 that when it happens. But in good faith, as you

5 did yesterday or the day before with Dr. Jensen,

6 we are obviously going to oppose that this has

7 to stop at some point.

8 BY MR. WATTS-FITZGERALD:

9 Q. Was the methodology for the production of

10 the accuracy assessment for 1993, Exhibit 16, the

11 same as that which you have described for 1985?

12 A. Yes.

13 Q. On the Y axis, Exhibit 16, the numbers

14 range from 1 to -- I can't read the last one, 18, but

15 they are not obviously sequential. The missing

16 classes are as a result of the class collapse --

17 A. Yes.

18 Q. -- over the time of the analysis?

19 A. Yes.

20 Q. On 16 I see a commission error for

21 predominant cattail in excess of 50%. What does that

22 signify to you?

23 A. Which figure are you looking at now?

24 Q. Exhibit 16, the figure for 1993, the

25 predominant cattail, which is the first line in the

130

1 numerology to the right of the graph. You said you

2 have total of 7?

3 A. Commission error column?

4 Q. Yes. Commission error is 53.33?

5 A. The photo interpretation, the truth found

6 and error.

7 Q. I think what was throwing us off, the total

8 column on the right on both pages, should that be

9 labeled total correct? That's really what that is,

10 isn't it, and then the total number of samples per

11 class is at the bottom of the box?

12 A. No, I don't think so.

13 Q. Let's look at position 1-1 in the box.

14 It's shaded. It says 7. The 7 is carried over to

15 the right to the total line.

16 If I go down the vertical column of 1, I

17 see in class 3, 1 was identified. It was 4. 2 were

18 identified. Class 7, 5 were identified. Those have

19 to be commission errors to stand up with 53.33?

20 A. Yes.

21 Q. So that implies that one category 3 which

22 is equal mix, one category 4 predominant mix, one

23 category 7 predominant sawgrass were identified into

24 the predominant cattail area which were not?

25 A. In the photo, yes. That is what that

131

1 indicates.

2 Q. And the photo is the true?

3 A. Photo is the true.

4 Q. Which means the image was wrong?

5 A. Well, yes. If you use the photo as ground

6 truth.

7 Q. You did, did you not?

8 A. Yes.

9 Q. You took as gospel whatever is in those

10 photos?

11 A. In order to do an accuracy assessment you

12 have to come up with something to compare your data

13 to.

14 Q. Since you weren't comparing to ground

15 truthing efforts directed at the accuracy assessment

16 points, you had to do this boot strap operation from

17 the photos?

18 MS. RAEPPLE: Object to the form.

19 THE WITNESS: Yes.

20 MR. WATTS-FITZGERALD: I shouldn't use the

21 term boot strap, that's a term of art, if that

22 was the objection.

23 MS. RAEPPLE: That was the objection.

24 THE WITNESS: Restate it again.

25 BY MR. WATTS-FITZGERALD:

132

1 Q. So this implies for '93 that your system

2 did a pretty good job placing in the predominant

3 cattail category on the image everything that is

4 predominant cattail. You are calling it 100%

5 accurate?

6 A. That's a normal statistical value, yes.

7 Q. The rest of it, the commission error says

8 that's true, but watch out because a lot of areas are

9 going to be listed as cattail that aren't?

10 A. It's saying they aren't in that class, yes.

11 There may be other mixes of cattail.

12 Q. But they are not predominant cattail as you

13 defined them?

14 A. Yes.

15 Q. In fact, 5 out of the 8 are going to be

16 predominant sawgrass?

17 A. Five out of 15.

18 Q. 5 out of 8 are incorrectly identified. I'm

19 giving you the 7.

20 A. Okay.

21 Q. Now, in '85, this methodology did not do so

22 well for predominant sawgrass and broad leaf

23 vegetation, tree islands because you got a fairly

24 significant percentage omission errors for both of

25 those classes.

133

1 You have 23.08 for predominant sawgrass and

2 50% -- you have a 100 omission error for predominant

3 sawgrass and other. So that model or that image does

4 not seem to do too well in the sawgrass range. It

5 looks like you misclassified it with a fairly high

6 degree of regularity based on the accuracy

7 assessment?

8 A. Is that a question?

9 Q. Yes. Am I interpreting that correctly?

10 A. It has a 76% of accuracy, which is higher

11 than the overall map accuracy. There are several

12 ways to look at data. You have to look at all of

13 these calculations in context.

14 Q. I was looking at all of the -- I was

15 looking for broad leaf vegetation upward. I see 50%

16 accuracy for broad vegetation/tree island. I find

17 that surprising. I find according to Dr. Lodge

18 that's one of most prominent features identifiable

19 from the imagery and CIR. I see only 76% for

20 sawgrass and zero for predominant sawgrass/other.

21 Zero for predominant cattail/other, which would

22 include, I assume, predominant cattail with some

23 sawgrass because of the way that class was classed.

24 So for the class number 3, 6, 7, 8, -- 3,

25 6, 7 and 9 the accuracy is very poor only for

134

1 predominant cattail. And as far as sawgrass with

2 periphyton, it may contain periphyton, is that right?

3 Do you recall Ken Rutchey's discussion as

4 sort of an author's observation in his report of the

5 influence of the brightness value of periphyton on

6 interpretation?

7 A. What kind of observation?

8 Q. About the influence of the brightness value

9 of periphyton communities?

10 A. You said he made some type --

11 Q. Author's observation.

12 A. Okay, go ahead.

13 Q. Do you recall that discussion in his

14 report?

15 A. Yes.

16 Q. Did you or Dr. Lodge make any effort to

17 consider the possible implications for your work of

18 the phenomenon he reports?

19 A. The reflectance of periphyton?

20 Q. Yes.

21 A. Yes, we did.

22 Q. What did you do?

23 A. Most cases where periphyton was present, it

24 was sparse sawgrass community.

25 Q. That was in your field visits for purposes

135

1 of identifying training sites that occurred during

2 what time of the year?

3 A. Our field samples were taken at the same

4 period the satellite passed over, roughly the winter.

5 Q. But for the 1985 effort you were working

6 from CIR imagery. That was roughly 10 months

7 distant. And February, March working back to

8 November when the communities -- periphyton

9 communities, at least based on what Mr. Rutchey says

10 in his report could be expected to be somewhat

11 different, was any effort made with regard to the '85

12 to take that factor into account in interpreting the

13 photo imagery?

14 A. We attempted to identify periphyton with

15 the sparseness of vegetation within it. We tried to

16 take into account the effect of periphyton.

17 Q. What does the KAPA value tell you with the

18 accuracy assessment calculations?

19 A. It would tell you how much randomness

20 influenced that classification.

21 Q. At what point does the randomness exceed

22 the acceptable limits in your profession when

23 conducting a statistical analysis and what would you

24 do if it exceeded those acceptable limits?

25 A. If it exceeded acceptable limits? There

136

1 are two divisions.

2 Q. I'm not allowed to do that. Let me start

3 with the second one first.

4 If the KAPA value you are ultimately

5 calculating, assuming that that occurs, exceeded the

6 acceptable norm or limits within the field of

7 expertise that you are going to proffer opinions

8 from, what do you do -- how do you cure the problem?

9 Can you cure the problem?

10 A. Once you complete --

11 MS. RAEPPLE: Could you just ask one

12 question at a time? Objection to the form.

13 MR. WATTS-FITZGERALD: You can still

14 answer.

15 MS. RAEPPLE: You asked three questions.

16 Which one do you want answered.

17 MR. WATTS-FITZGERALD: I have had

18 difficulty throughout getting the witness to

19 answer the question I do ask. I try to ask it

20 several ways so that we eventually get around.

21 It just takes a lot of time.

22 BY MR. WATTS-FITZGERALD:

23 Q. Can you adjust after the fact when the KAPA

24 value exceeds acceptable limits?

25 A. I don't believe so, no.

137

1 Q. Can you go back, include additional

2 assessment points in an effort to improve the

3 accuracy of your work?

4 A. The way we are set up on this project which

5 is using imaging software, once you close the table,

6 once you run the accuracy assessment in the imagining

7 software you cannot modify in any fashion that table

8 ever again.

9 If the KAPA coefficient is run on the

10 table, it's set in stone. You have to generate a new

11 set of stratified random samples and go through the

12 whole process again if you wanted to redo it at that

13 point.

14 Q. Would merely redoing it but including

15 additional accuracy assessment points -- instead of

16 stopping at 39 you go to 80 for 1985, for example,

17 will that affect the KAPA value?

18 A. Well, it could, I suppose, yes. I never

19 tried that, I don't know.

20 Q. If the accuracy assessment is valid in

21 expressing the accuracy of the image, that's what

22 it's trying to do, don't you have to go back and

23 improve the image itself to get better accuracy?

24 A. I don't know.

25 Q. If the accuracy assessment is valid and you

138

1 have a 70% accuracy in your image -- I'll give you a

2 hypothetical.

3 If you end up doing an accuracy assessment

4 that says you have a 40% accurate image, assuming 40%

5 also falls below the acceptable norm, is there any

6 alternative other than to go back and improve your

7 image?

8 A. You are talking about once you close the

9 table and -- you know, you said you have difficulty

10 having me answer questions. I'm not intending to do

11 that.

12 Q. I wasn't suggesting you were.

13 A. All right. The methods we are using, we

14 generate a table. Each of those tables, we generate

15 more random samples that we used. That is because I

16 went and had the option if the powers that be want

17 more points, we can do that. But once we closed the

18 table, then that table can never be reopened because

19 that's a quality control check.

20 So if I understand your question, if you

21 did a map, if we did this map and closed the tables

22 and it was 40% accuracy, the level of the map is 40%,

23 then the only way you could redo that accuracy

24 assessment would be to regenerate a set of points and

25 redo the whole process. Am I answering your

139

1 question?

2 Q. You actually answered the earlier one, but

3 that's good, too.

4 If your accuracy assessment is right, no

5 matter how many points you used and your accuracy is

6 only 40% based on the accuracy assessment and it's

7 valid, that's a given that implies that your image is

8 no good, that the image as platted, the result of all

9 of the work is substandard, it's deficient, it should

10 not be used?

11 A. If 40% were an acceptable accuracy level?

12 You discussed it as being acceptable.

13 Q. It's not acceptable.

14 A. If it was not acceptable and you agreed it

15 was not acceptable and you ran your assessment it was

16 less than 40% or 40%?

17 Q. And your accuracy assessment is right, you

18 have to assume it's right?

19 A. Yes. In that case you could add a thousand

20 points and assess every pixel, but it would not

21 improve the accuracy of the classification, if I

22 understand the question correctly. I think I do.

23 Q. I'm trying to figure out if they tell you

24 to go back and do more points, you said you closed

25 the tables in order to do this?

140

1 A. No. We have not closed the tables yet.

2 That's what I said.

3 Q. These are drafts?

4 A. No, these aren't drafts. These are real

5 valid tables based on number of points we have now.

6 The table ERDAS is still open. If I closed it right

7 now, it would produce exactly this result. This is

8 the same data, same set of formulas.

9 But the table in the ERDAS software is

10 still open. If I was directed to continue selecting

11 samples, I could add more samples to it. There would

12 be no closes classes where there were no samples. So

13 your questions about that could be answered and we

14 could continue.

15 I apologize for the confusion of not

16 understanding what you were talking about.

17 Q. I mentioned on Exhibits 15 and 16 certain

18 classes that have no samples or no accuracy

19 assessment points. What can you tell, if anything,

20 based on this assessment methodology about those

21 classes?

22 A. Those classes from this table, you can't

23 say.

24 Q. So if you express any opinion about those

25 categories in the two respective images, they would

141

1 be based on what?

2 A. My opinion the image.

3 Q. In your professional judgment?

4 A. Yes, it would be a professional judgment

5 call, obviously.

6 Q. And you would feel likewise constrained,

7 then, not to attempt to extrapolate these values

8 assuming they prove up after any further work or stay

9 the way they are to areas beyond 2A in a quantitative

10 sense?

11 A. You mean, would I speculate on the accuracy

12 in all of the study area, is this accuracy based on

13 the assessment 2A? I would speculate it would be

14 very similar. I wouldn't speculate that it would be

15 the same.

16 Q. As we discussed earlier, to the extent any

17 sub area within the study area has substantial

18 variation in its foliage from that found in 2A you

19 would have to diminish the accuracy values or your

20 assessment of the accuracy values, would you not?

21 A. If the foliage were different?

22 Q. The vegetation -- hypothetically, if

23 Dr. Lodge would tell you that the flora of Water

24 Conservation Area 1 is markedly different in its mix

25 and composition from the foliage in Water

142

1 Conservation Area 2 for whatever reason, would you be

2 confident in taking these quantitative values for 2A

3 and assuming or rendering a professional judgment

4 that they pretty closely, if not exactly, reflect the

5 conditions and accuracy of the image in Water

6 Conservation Area 1?

7 A. I would think there would be a high degree

8 of correlation between the accuracy in Loxahatchee

9 and the entire duty area to the 2A figures.

10 Q. Even the Loxahatchee had such a minimal

11 number of ground truthing points?

12 A. Yes.

13 MR. WATTS-FITZGERALD: I have no further

14 questions, but I want to mark exhibits.

15 I would like to mark Exhibit 17 the

16 supervised classification Landsat Thematic

17 Mapper Multispectral image for December 10, 1993

18 of the entire study area. It's got a prepared

19 date of April 4, 1994, depicting essentially the

20 northern most point of Water Conservation Area 1

21 to approximately ten kilometers of Everglades

22 National Park in the south.

23 (The document referred to was thereupon

24 marked Exhibit 17 for Identification.)

25 MR. WATTS-FITZGERALD: Exhibit 18, the map

143

1 which indicates it was produced April 5, 1994

2 scaled 1 to 50,000, Water Conservation Area 2A

3 from the supervised classification of Landsat

4 Mutispectral Thematic Mapper scene from

5 December 10, 1993.

6 (The document referred to was thereupon

7 marked Exhibit 18 for Identification.)

8 MR. WATTS-FITZGERALD: Exhibit 19 will be a

9 grid map produced and discussed by Dr. Lodge

10 which underlies the work described by the

11 witness today depicting construction lines and

12 other features utilized in the preparation of

13 their classified imagery for WCA-2A.

14 Exhibit 19 not being labeled other than

15 with the number of the exhibit.

16 (The document referred to was thereupon

17 marked Exhibit 19 for Identification.)

18 MR. WATTS-FITZGERALD: Exhibit 20 is photo

19 interpretation registration grid for NHAP 9 by 9

20 at scale 1 to 64,923.

21 (The document referred to was thereupon

22 marked Exhibit 20 for Identification.)

23 MR. WATTS-FITZGERALD: Both 19 and 20 would

24 relate to the 1985 imagery because of the scale.

25 MS. RAEPPLE: For Exhibit 19 and 18 you

144

1 will be making copies and substituting copies

2 and returning the originals to me; is that

3 correct?

4 MR. WATTS-FITZGERALD: Yes. Exhibit 21

5 will be Water Conservation Area 2A classified

6 image supervised classification of Landsat

7 Multispectral acquired November 2, 1985, the map

8 production date April 5, 1994, scale 1 to

9 50,000.

10 (The document referred to was thereupon

11 marked Exhibit 21 for Identification.)

12 MR. WATTS-FITZGERALD: Exhibits 22 and 23

13 are large grids about 4 feet by 5 feet with

14 construction lines and some features drawn in in

15 colored pencil depicting portions of WCA 2A that

16 have previously been described in Dr. Lodge's

17 initial deposition.

18 (The documents referred to were thereupon

19 marked Exhibits 22 and 23 for Identification.)

20 MR. WATTS-FITZGERALD: We are marking

21 Exhibit 24 a portion of a USGS quadrangle from

22 Fort Lauderdale to northwest 7.5 series showing

23 the southeast portion of the Loxahatchee Refuge

24 and the north central area of Water Conservation

25 Area 2 A. No date.

145

1 (The document referred to was thereupon

2 marked Exhibit 24 for Identification.)

3 MR. WATTS-FITZGERALD: Exhibit 25 is

4 Fort Lauderdale to the southeast, photo revised

5 1983 7.5 minute series.

6 (The document referred to was thereupon

7 marked Exhibit 25 for Identification.)

8 MR. WATTS-FITZGERALD: 26 is

9 Fort Lauderdale to the southwest dated 1973, no

10 revision date on it. 7.5 showing the central

11 area of WCA-2A and a portion of levee 38 east.

12 (The document referred to was thereupon

13 marked Exhibit 26 for Identification.)

14 MR. WATTS-FITZGERALD: 27 is Cooper City

15 northeast, Florida photo revised 1983, 7.5

16 minute series showing the eastern portion of

17 Water Conservation Area 2.

18 (The document referred to was thereupon

19 marked Exhibit 27 for Identification.)

20 MR. WATTS-FITZGERALD: 28, Andytown,

21 Florida. 7.5 minute series photo revised 1983

22 showing the 20 mile bend and the southeastern

23 corner of Water Conservation Area 2A where it

24 intersects 2B and WCA 3A.

146

1 (The document referred to was thereupon

2 marked Exhibit 28 for Identification.)

3 MR. WATTS-FITZGERALD: 29 is Everglades 1

4 southeast portion there of 7.5 minute series.

5 Date is 1974. No revision date.

6 (The document referred to was thereupon

7 marked Exhibit 29 for Identification.)

8 MR. WATTS-FITZGERALD: These are all

9 photographs, Exhibits 29 and 24. 26 is, as

10 well.

11 Exhibit 30 is Fort Lauderdale to the

12 northwest 1974, 7.5 minute photo showing the

13 northern portion of 2A and Loxahatchee.

14 (The document referred to was thereupon

15 marked Exhibit 30 for Identification.)

16 MR. WATTS-FITZGERALD: Finally, Exhibit 31

17 depicts the northeastern corner of Water

18 Conservation Area 2A, southern most portion of

19 Water Conservation Area 1, and adjacent portions

20 of Coral Springs. And because of the cutting of

21 the top of the map, none of the legends that

22 otherwise appear directly there on other

23 versions of Exhibit 25, Fort Lauderdale to the

24 southeast but it has various sites labeled

25 numerically and alphabetically in 2A.

147

1 MS. RAEPPLE: Mr. Watts-Fitzgerald, as for

2 Exhibits 19, 20, 22 through 31 we will need the

3 originals returned to us and we would appreciate

4 you having copies made, substituting copies as

5 exhibits to the deposition and returning them at

6 your earliest convenience.

7 MR. WATTS-FITZGERALD: I'm not sure that

8 attempts to replicate the work can be done off

9 the copies of some of these. I may not be able

10 to return then all immediately, if we are going

11 to use the photos. This is something we have to

12 talk about, getting the photos from you later.

13 But I understood that you were producing

14 the 15 meg tape and the tasseled cap as part of

15 the production. We are not marking them as

16 exhibits.

17 MS. RAEPPLE: You can keep those. As for

18 the color infrared, we will retain the

19 originals. If you need access to them for

20 whatever purpose, let us know and we will

21 provide that -- including if you want to take

22 them long enough to make copies.

23 In addition, Mr. Downing advised me at

24 lunch in particular that he overlooked providing

25 today in digital form the '85 image which you

148

1 have in hard copy. If you want it in digital

2 form we can send that to you next week.

3 MR. WATTS-FITZGERALD: We would request

4 that.

5 THE WITNESS: I'll make you one.

6 MR. WATTS-FITZGERALD: We got over the

7 lunch break all of the worksheets by Dr. Lodge

8 that were used for '85 and '93?

9 MR. CESARANO: I believe it is. That's all

10 that came in.

11 MR. WATTS-FITZGERALD: I have nothing

12 further.

13 MR. CESARANO: I have nothing further.

14 MS. RAEPPLE: No questions. We do not

15 waive.

16 (Thereupon the taking of the

17 deposition was concluded.)

18 - - - - - - - - -

149

1

2 I, ED DOWNING, do hereby certify that I

3 have read the foregoing deposition and that the same

4 is a true and accurate transcript of my testimony,

5 except for attached amendments, if any.

6

7

8

9 ----------------------------------

10

11

12

13

14

15 The signature above of ED DOWNING was

16 subscribed and sworn to before me this day of 1994.

17

18

19

20

21 -----------------------------------

22 Notary Public

23 My commission expires

150

1

2 CERTIFICATE OF OATH

3

4

5 THE STATE OF FLORIDA )

6 COUNTY OF DADE. )

7

8

9 I, the undersigned authority, certify that

10 ED DOWNING personally appeared before me and was duly

11 sworn. WITNESS my hand and official seal this 15th

12 day of April, 1994.

13

14

15 ___________________________________

16 Thomas R. Neumann, RPR

17 Notary Public - State of Florida

18 My Commission Expires: June 19, 1994

151

1 CERTIFICATE

2

3 THE STATE OF FLORIDA )

COUNTY OF DADE )

4

5

I, Thomas R. Neumann, Registered

6 Professional Reporter, do hereby certify that I was

authorized to and did report said deposition in

7 stenotype; and that the foregoing pages, numbered

from 1 to 148, inclusive are a true and correct

8 transcription of my shorthand notes of said

deposition.

9

I further certify that I am not an attorney

10 or counsel of any of the parties, nor am I a relative

or employee of any attorney or counsel or party

11 connected with the action, nor am I financially

interested in the action.

12

The foregoing certification of this

13 transcript does not apply to any reproduction of the

same by any means unless under the direct control

14 and/or direction of the certifying reporter.

15 Dated this 15th day of April, 1994.

16

______________________

17 Thomas R. Neumann, RPR

18

THE STATE OF FLORIDA )

19 COUNTY OF DADE. )

20

The foregoing certificate was acknowledged

21 before me this 15th day of April, 1994 by

Thomas R. Neumann, who is personally known to me.

22

23

________________________________

24 Notary Public - State of Florida

My Commission expires:

25

152

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15