264
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 7, 1994
20 2:05 p.m. - 5:15 p.m.
21 DEPOSITION OF THOMAS E. LODGE, Ph.D., C.E.P.
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 - - - - - - -
265
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
INDEX
17 Witness Direct Cross Redirect Recross
THOMAS E. LODGE
18 By Mr. Cesarano: 266
By Mr. Watts-Fitzgerald 309
19
266
1 Thereupon --
2 THOMAS E. LODGE
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 afternoon, Dr. Lodge. We are here to
8 continue the taking of your deposition which was
9 commenced last week.
10 Let's start by having you tell us what you
11 did since -- what you had done since the last time we
12 saw you?
13 A. The major participation I have had is in
14 the evaluation of 1984 and 1985 color infrared aerial
15 photography as an accuracy check for satellite
16 mapping that has been produced based on data from
17 that approximate time frame.
18 The color infrared photography that I used
19 is of a different scale. It's approximately 1 to
20 55,000 or so, I don't know the exact number, whereas
21 the previous color infrared photography we were using
22 was at a scale of 1 to 24,000 which matched the USGS
23 quadrangle sheets.
24 So for the assessment that I just completed
25 we had the USGS quadrangle sheets reduced to that
267
1 approximate scale so that the aerial photographs
2 could be superimposed over the quadrangle sheets.
3 And we also produced a grid, UTM coordinates in that
4 same scale so that check points for the accuracy
5 check could be superimposed on the USGS quadrangle
6 sheets and then also located on the color infrared
7 aerial photography.
8 So our accuracy assessment for the earlier
9 period then is totally based on aerial photography,
10 not based on ground truth data.
11 Q. Let me ask you some specific questions
12 about what you just told us.
13 What are the dates of the color infrared
14 photography from 1984 to '85?
15 A. We have the photographs here, and the
16 actual dates are specified on the sides, if you would
17 like me to pull one out.
18 Q. Yes, I would.
19 A. There are two different dates involved.
20 One series is March 3rd, 1984 and another set is
21 February 14th, 1985.
22 Q. How were those dates chosen?
23 A. These were what was available. I don't
24 know how that selection was made.
25 Q. Where did you obtain these?
268
1 A. These were provided to me at our Kennesaw,
2 Georgia office by Mr. Ed Downing.
3 Q. You don't know where he got them?
4 A. No, I don't know the direct source.
5 Q. The date of the satellite imagery that you
6 used these color infrared for was what, please?
7 A. That date is also on the satellite imagery
8 for the earlier time period. If I could just look at
9 it to get it right, it is November 2nd, 1985.
10 Q. In your opinion what effect would using a
11 color infrared image from March of 1984 to verify
12 satellite image in November of '85, that lapse of
13 time, what effect would that have?
14 A. Well, during the time period you could have --
15 of course, a major influence would be fire. The
16 areas where it was evidence that fire had not
17 occurred there would be whatever time period the
18 growing season had. And any time that the data
19 photograph doesn't exactly match the date of the
20 imagery, you have other kind of conditions -- not
21 just date but the actual time of reflectance,
22 direction and wetness conditions, other weather
23 conditions.
24 Q. And all of those variables can introduce
25 error into your evaluation, can it not?
269
1 A. That's correct, it can.
2 Q. And this 1985 image, the title on this
3 document that you brought, this map for 1985 states
4 that it's a supervised classification of Landsat
5 multispectral thematic mapper scene. Is that the
6 type of satellite image it is?
7 A. The type of satellite image it is really is
8 Mr. Downing's area.
9 Q. But so far as you know, that's what it is?
10 A. That is correct.
11 Q. Tell me -- well, first of all, what did you
12 bring here today that you worked with in the last
13 week or so?
14 A. I brought the color infrared photographs
15 that I worked with, the grid -- there are two grids,
16 actually, that were used to -- these are UTM
17 coordinate guides. These were used to match the
18 scale of the photography and to provide randomly
19 chosen points for an accuracy assessment.
20 Q. That's these two grids here?
21 A. Yes. One grid has purely the accuracy
22 assessment points. That grid was produced a number
23 of hours later than an earlier grid that was used
24 strictly as an overlay for the aerial photographs and
25 the USGS quadrangle sheets which had been reduced to
270
1 the size of the aerial photography.
2 Q. So that we know what we are talking about
3 here, the grid with the conservation area and what
4 appears to be the air boat trails is the one that was
5 used to locate for the proper location?
6 A. Right.
7 By superimposing these two grids, you can
8 see I went through a process of tracing on to the
9 grid showing air boat trails and it also shows some
10 tree island features.
11 I traced the randomly selected points that
12 were used for the accuracy determination, the reason
13 being that I had to have all of the information on
14 one sheet and the accuracy assessment points had not
15 been chosen as early as this grid had been made.
16 Q. I'll get into that whole process a little
17 more specifically in a few minutes.
18 I next want to ask you, was the only map
19 that you -- the only satellite image that you worked
20 with the November 2, 1985 thematic mapper image that
21 is down at the end of the table?
22 A. That's correct.
23 Q. You have not done anything with respect to
24 the map that shows the entire study area from the
25 refuge to the northern part of the Park?
271
1 A. None other than what was described in
2 the -- my earlier deposition where we talked about
3 ground truth points that we had done in Everglades
4 National Park and conservation area 2A. Mr. Downing
5 can tell you how those were used.
6 I will comment, however, that in learning
7 how to recognize details on the '84 and 85 color
8 infrared photography I needed to back check with
9 photography that I did understand from having worked
10 with in the previous time period. And in doing that,
11 I recognize that at one point apparently a grid had
12 slipped slightly, so I annotated the sheet and
13 re-evaluated previous points that had been used in
14 our accuracy check.
15 Q. That is for which map?
16 A. That is for the date of 1993. I retraced
17 some air boat trails and rechecked and handed that
18 accuracy check information in. I don't know what
19 effect it had.
20 Q. Now, you said that the scale of this
21 infrared photography that you used was different from
22 before in that it was approximately 1 to 55,000,
23 correct?
24 A. Yes.
25 Q. You didn't verify that scale?
272
1 A. The scale was verified. I don't happen to
2 have it in my head and I don't have it written down
3 on anything. It may be written on the grid.
4 Q. Up here at the top of this is some writing
5 that says printing, it says photo interpretation
6 registration grid for NHAP 9 by 9 CIR. Dispositive
7 at map scale 1 to 64,000923. Is that the scale?
8 A. That's the scale.
9 Q. Was that scale verified independently in
10 Kennesaw?
11 A. Yes.
12 Q. The USGS quad sheets are at a scale of 1 to
13 24,000?
14 A. The original are, that is right.
15 Q. You say that you reduced them so that they
16 would correspond to the scale of the infrared
17 photography that you were using?
18 A. That's correct.
19 Q. How did you reduce those quad sheets?
20 A. I didn't do that, that was done by
21 Mr. Downing.
22 Q. Do you know how it was done?
23 A. No, I don't, but I have an example
24 transparency of a USGS quadrangle sheet.
25 Q. Do you know whether the quadrangle sheet
273
1 was reduced to exactly 1 to 64,923?
2 A. I'll explain why it was not important for
3 me to remember what the actual scale is. The
4 important thing for me to verify as I work that I was
5 able to match the UTM coordinate grid lines with the
6 UTM indication on the sides of the reduced scale USGS
7 quadrangle sheets and that I was able to obtain a
8 satisfactory match of the aerial photography features
9 over the USGS quadrangle sheets. That is, there are
10 a number of features indicating canals and so on that
11 would superimpose.
12 Q. Okay. My question is not whether or not
13 you knew if the quad sheet was reduced to a
14 particular scale. What I meant to ask you was
15 whether or not you knew whether the quad sheet was
16 reduced to the exact same scale as the color infrared
17 photography.
18 A. As near as I was able to determine, that
19 was a satisfactory match. But I'll say that it was
20 not exact, that the color infrared photography had to
21 be adjusted very slightly. But there were
22 adjustments similar to the ones that I described in
23 doing the accuracy check for 1993, that the center of
24 a photograph is generally the best portion. So you
25 try to adjust so that you are sure that that center
274
1 is in the correct place. The edges sometimes are
2 slightly off, but the match requires very, very
3 little adjustment once the photograph had been set on
4 a sheet.
5 Q. Do you have any idea what type of
6 locational error was present by using those overlays
7 in the fashion you have just described?
8 A. I don't have that quantified, but in most
9 places on these aerial photographs there were visible
10 air boat trails and a few other features that were
11 also indicated on the USGS quadrangle sheets. So by
12 superimposing those I was able to obtain some degree
13 of accuracy.
14 I really don't have that quantified, but
15 the air boat trails may typically be 15 feet wide. I
16 think that that sort of degree of slip, say 15 feet,
17 is probably the best. But there may be places where
18 we are 30 or 40 feet off. But 15 feet from a
19 particular location would be a possibility.
20 Q. Now, talking about the air boat trails, did
21 those appear on the quad sheets?
22 A. A number of air boat trails are indicated
23 on USGS quad sheets, yes.
24 Q. And what were the dates of the quad sheets
25 relative to the dates of the color infrared?
275
1 A. I don't have all of the dates of the quad
2 sheets, but they are based on aerial photography that
3 was somewhat earlier.
4 And that I believe we discussed this
5 phenomenon the last time, that air boat trails change
6 from time to time but some of the major trails are
7 identifiable as being in the same place and having
8 the same curves. We were able to eliminate certain
9 trails as being new or substantially relocated.
10 A lot of that depends on features that are
11 on the edge of water conservation area 2A where, for
12 example, there are identifiable structures or roads
13 that can be seen in the aerial photography. So when
14 you get to the edge of a quad sheet there are a
15 number of match points that can be used between the
16 photography and the USGS quad sheet.
17 Q. Did you verify the location of the features
18 that you saw that were depicted on the quad sheets?
19 A. I don't understand how you mean "verify."
20 Q. Let me ask you this. What is the margin of
21 locational error on quad sheets?
22 A. I know there is some guaranteed level of
23 error. I don't have that figure in my head.
24 Mr. Downing does.
25 Q. You described in certain instances when the
276
1 overlays didn't exactly match up, didn't quite match
2 up. By how much wouldn't they match up and how does
3 that translate to distance on the ground?
4 A. I can't give you an overall number that
5 represents what my perception is even of how close it
6 would be. But error as large as a couple of hundred
7 feet is easily seen. Even a hundred feet is easily
8 seen even at this scale on the aerial photographs.
9 Because of the certain features, roadway width and so
10 on, roads can be seen.
11 There are some cases where there are stilt
12 camps, structures constructed in water conservation
13 area 2A, and those have size in the relative size of
14 a conventional house, not a large house.
15 Q. When you put these overlays down and they
16 didn't quite match up, what was the greatest distance
17 in millimeters that you can remember that it didn't
18 match up with?
19 A. Well, when looking from one side of a sheet
20 to another side in millimeters, you could be off two
21 millimeters. And obviously then when you work one
22 side of a quad sheet with features matched up as best
23 you can and you start working to the other side you
24 have to readjust to line up the features that are on
25 that size. So you are constantly correcting.
277
1 You try to keep features that are as close
2 to your points of question as possible.
3 Q. At this scale, if your map didn't line up a
4 single millimeter, what would that translate to on
5 the ground?
6 A. That would translate to 64,000 millimeters,
7 approximately, on the ground.
8 Q. How many meters would that be?
9 A. 64, meters which would be a couple of
10 hundred feet.
11 Q. And what were the sizes of the pixels in
12 the satellite imagery?
13 A. The pixels, to my understanding, is on the
14 order of 75 feet on the side. 25 meters, 80 feet on
15 a side.
16 Q. What is the smallest size feature that --
17 or I suppose you call it a mapping unit -- that you
18 can determine on the 1 to 64,000 plus color infrared
19 photo?
20 A. Well, as I said, roadways, even small dirt
21 roads are easily visible, but I don't have an actual
22 number.
23 Q. You don't think that's important to know?
24 A. Yes. I think it's of relative importance.
25 I'm confident that something as large as 30 feet
278
1 across would be visible on these photographs.
2 Q. Explain to me exactly how you did this
3 accuracy check, please, if you could come over and
4 point it out to me and kind of walk me through it.
5 A. Let me pick out an example photograph.
6 All right. First step is to superimpose
7 the USGS quad sheet on the grid of UTM coordinates.
8 The particular points of the UTM grids are printed on
9 the side of the photograph, the numbers are there,
10 and there are small blue lines that can be used to
11 match up. I did notice --
12 Q. Let me look at this.
13 A. We can put this on the window and it's
14 focusable for my eyes. Let me go to the side and
15 find a particular point for you. Right there you
16 should see 2913.
17 Q. Right.
18 A. The grid points are visible with the naked
19 eye, but it's much easier with magnification.
20 Q. You line those up?
21 A. Using numbers as well as lines to make sure
22 you are on the correct line.
23 There are some places where this particular
24 redox of the USGS quad doesn't exactly match the
25 grid. We used a correction to make sure that the
279
1 area of interest was matching. Sometimes distant
2 from the area of interest there would be a slight
3 mismatch. But when you work from point to point you
4 make sure that your lineup is as close as you can get
5 it for that area.
6 Q. Explain to me again why there would be a
7 mismatch.
8 A. Well, it perhaps happens to do with how the
9 quad sheet was reduced.
10 In my opinion I would say it was a very
11 good job because I had very little problem with that.
12 Q. Okay. So you got the quad sheet on the
13 grid. What's the next step?
14 A. Right now we are talking about an accuracy
15 check where I have areas where I'm supposed to judge
16 what the vegetation is in the photograph.
17 That means we have jumped ahead of my
18 having studied these photographs to determine what
19 kind of vegetation are there, realizing we had no
20 ground truth for that particular point in time. That
21 may have been available, but we didn't have it
22 available to us.
23 Q. Let's talk about that. I don't want to
24 jump ahead too quickly. You have this color infrared
25 image here. Before you overlaid it on the quad
280
1 transparency, you studied it, correct --
2 A. Yes.
3 Q. -- to try to determine the different types
4 of vegetation depicted in the photo, right; is that
5 correct?
6 A. That's correct.
7 Q. How did you make that determination?
8 A. We were first of all using the premise that
9 cattail at that time, if present, would be in
10 somewhat similar areas to where it is today. That is
11 particularly where it is very abundant today, it's
12 likely that it would have been present in the past.
13 So I used that as a training point.
14 We also looked at tree islands for known
15 characteristics. The heads of tree islands, the
16 actual tree island itself frequently has a pink
17 vegetation character. We cross referenced with the
18 older aerial photographs to see if we could pick out
19 those features and other items in the shape of tree
20 islands.
21 We looked at the variability -- I looked at
22 the variability in this photograph and compared it to
23 the variability in the other and was convinced that
24 the same spectrum of kinds of classes of vegetation
25 were generally visible on these photographs compared
281
1 to the others with somewhat of a qualification that
2 these photographs are accentuated towards the visible
3 blue, whereas the more recent color infrared
4 photographs have considerably more yellow and red at
5 the end of the spectrum.
6 We are looking for -- I looked for
7 patterns. Therefore, I assumed that the vegetation
8 would be similar.
9 One of the patterns that was easily
10 noticeable on the edge of areas where cattail exist
11 today, there are circular patterns that we verified
12 from the ground recently that were cattail. That's a
13 characteristic signature of cattail in a number of
14 places where it forms a circular colony, I should
15 say.
16 Those kind of features were recognizable in
17 similar areas on these older photographs.
18 Q. Is that like this?
19 A. That's correct. Plus the fact that those
20 features that you just pointed out, circular
21 photographs -- circular images are somewhat more
22 yellow than the background, blue/grey around it.
23 That is very similar to the characteristic color
24 change or color difference that we noted between
25 cattail and sawgrass in the more recent photographs.
282
1 Colors are not the same, but the general pattern of
2 change and the colors are somewhat similar.
3 So I used that as a training tool so that I
4 could identify cattail and sawgrass and tree island
5 features -- open water, of course, which is very dark
6 on both types of photography.
7 There are other subtle differences, but I
8 realize that a lot of that is too difficult to
9 determine what exactly it was without being able to
10 ground truth that at that particular time.
11 Q. What did you determine that these streaky
12 type images were?
13 A. Those streaky images were where water is
14 apparently showing through and would represent slough
15 habitat. There are times when water lilies, these
16 cover those so that a slough might look like open
17 water. It takes on a pink flush from water lilies
18 being present.
19 Q. So after you determined in the fashion that
20 you just described the types of vegetation without
21 doing some ground truthing, just by analyzing the
22 photographs --
23 A. Cross referencing.
24 Q. -- what was the next thing you did?
25 A. Then we get to the point where we broke off
283
1 here of superimposing the redox, the USGS quadrangle
2 sheets over our grid, then superimposing on a light
3 table the aerial photographs.
4 And we would start out -- I started out
5 with aerial photographs that were on the edge of the
6 conservation area where there were identifiable
7 structures so that I could make a alignment that I
8 was very confident was correct.
9 I started tracing some features, then by
10 having to reverse the sandwich here to put, for
11 example, air boat trails and some tree island
12 features onto the grid. I'll say that we were also
13 aided by some location features that we had done
14 previously with Ed Downing, under his guidance, the
15 heads of tree islands that had a very small actual
16 center of trees, those were located for the previous
17 study.
18 We made the assumption that those heads, if
19 they were recognizable of being a similar
20 configuration, were the same as the tree island
21 locations in this earlier photography.
22 That grid that was present, diagonal lines
23 that intersect on tree island heads, were put on as
24 an aid of helping to locate where these aerial
25 photographs lie.
284
1 I'll say that we didn't purport to say that
2 the tree islands haven't changed shape or shifted
3 just slightly. So we used that as an aid, not as an
4 absolute location.
5 But in general that was very helpful. It
6 correlated, is the important thing, with other
7 features in the area of the photographs.
8 Q. Okay.
9 A. After going through the exercise of tracing
10 easily recognizable air boat trails onto the grid,
11 I'll say I selected the air boat trails based on
12 their presence on the USGS quad sheet and
13 recognizable presence in a similar location on the
14 aerial photographs.
15 Q. I'm just going to make a couple of notes
16 here. Give me a second.
17 A. Okay.
18 Q. You have done the step with the quad sheet
19 transparency, the color infrared transparency, and
20 you located tree island heads and made tracings of
21 tree islands and boat trails. Then what?
22 A. After that work had been done, a set of
23 points were given to me that represented by a small
24 box -- the points are actually the center of the box.
25 Those were -- Mr. Downing's group provided those on a
285
1 second grid.
2 Q. Do you know how those points were selected
3 or generated?
4 A. They were generated by a random locator
5 that is in the software that was provided that
6 Mr. Downing used. I believe that he needs to address
7 how that was done.
8 Further, I had to trust that that was done
9 properly. It was described to me as being a random
10 set of points where the program was also selected for
11 a uniform field of vegetation about that point so
12 that a certain number of pixels would be the same.
13 So we weren't looking for a needle in a
14 haystack, but we were looking for a feature that
15 would at least have dimensions of a few hundred feet
16 of being somewhat uniform.
17 Q. How many pixels were considered at each
18 point? I think you said the point was the center of
19 the boxes.
20 A. Right.
21 Q. Is the box the size of the pixels?
22 A. The box I believe is slightly larger than
23 the group of pixels selected.
24 Q. How many were in the group of pixels?
25 A. The earlier work, it was 3 pixels by 3
286
1 giving a 9 pixel area. I'm under the assumption it
2 was the same here, but I don't in fact know that.
3 Q. The method of selection was to within a
4 random selection to select points surrounded by
5 pixels in which a certain percentage or certain
6 number of pixels were of the same vegetative type?
7 A. Not the same vegetative type. They were
8 selected by uniformity of whatever parameters the
9 data set had which had to do with color.
10 Q. Do you know whether the program that
11 selected these random points was of a stratified
12 nature? In other words, was an attempt made to
13 locate random points in each vegetative class?
14 A. I don't know the answer to that.
15 Q. So Mr. Downing brought you how many points?
16 A. I believe that we ended up with 55. The
17 total number -- the total number that was generated
18 was 100. We had to cut back on that
19 number or I wouldn't have been able to finish for
20 this morning.
21 So we assumed that I would be able to
22 finish 35. So the first 35 in that order were
23 located on the grid that was supplied.
24 Q. This grid here?
25 A. Yes. I colored them in by number so that I
287
1 would be able to find them easily.
2 Q. It looks like just basically a smudge on
3 the first 35.
4 A. Right.
5 Q. After you did that?
6 A. Then I traced those actual points onto the
7 grid that we were using for location. You will find
8 that those two grids can be exactly superimposed.
9 There are so many lines and the fact that they both
10 have the numerical -- it's produced -- the grid is
11 produced by exactly the same technology.
12 I'm sorry, the second grid does not have
13 the actual numbers on it, but it has the same
14 features that were provided in the first that are
15 tied together with the grid.
16 Q. So you traced the points onto --
17 A. Our working sheet. We call it the working
18 grid.
19 Q. This is the working grid?
20 A. Correct.
21 Q. And after you traced those on, then what
22 did you do?
23 A. Then I proceeded to align the photographs
24 on the grid using the location points that I had
25 described, but frequently having to go back and put
288
1 the USGS quadrangle sheet back under again as a
2 check, because I'm not confident in just trusting air
3 boats and tree island locations. Wherever I could
4 use other features on a quad sheet, I would do that --
5 particularly at the edge of the quadrangle sheet
6 where roads that are identifiable on the aerial
7 photograph can be seen.
8 Q. You laid the color infrared over the
9 working grid?
10 A. Over the grid. Then I looked through to
11 find the actual accuracy check points. This
12 particular magnifier is a six power magnifier. You
13 are able to see a great deal of detail about a grid
14 point.
15 If I couldn't see through the sandwich, if
16 I couldn't recognize the color signature and perhaps
17 what the vegetation was, by leaving that in place you
18 can move the photograph while still looking at
19 features of the photograph to this part of the light
20 table where you can see through everything and it's
21 not complicated by the sandwich.
22 I would make a determination then what the
23 vegetation was based on the training that I had
24 performed earlier comparing these photographs with
25 the later photographs.
289
1 I then filled out a data sheet that was
2 provided for this and turned that over to Mr. Downing
3 for whatever accuracy check he did. I do not know
4 the results of that or how he incorporated that
5 information.
6 Q. Where is that data sheet?
7 A. That's with Mr. Downing. That's part of
8 his presentation.
9 Q. Could you set up that light box so that we
10 can take a look?
11 A. Certainly. Is there a plug in the table
12 there? I have a six foot cord.
13 Q. There should be one down here. Right
14 there. Is this yours?
15 A. No, this is not the light box I was using.
16 I was using a much larger one.
17 I didn't always sandwich these in exactly
18 the same order. Sometimes I would have the USGS on
19 top of the grid, depending on what sort of features
20 you would want to see.
21 Now, it's probably not going to be possible
22 to go through the exercise of lining up the UTM
23 coordinates. You can line them up from top to
24 bottom, but the UTM coordinates are on the edge of
25 the grid here. And because of the enlargement, it's
290
1 not there.
2 But by aligning the north/south running
3 lines with the top and bottom of the USGS sheet and
4 then using a straight edge to check for the
5 particular marks on the side, you can line that up.
6 As I said, in most cases it was perfect.
7 And where there was any discrepancy --
8 Q. Let me see how you do this.
9 A. All right. Let me do this.
10 Let me find a point. I'm finding a UTM
11 coordinate 567000, and on the edge of the USGS quad
12 sheet I have that same point. It's going to be
13 easier for you to see if I put the quad sheet on top.
14 Let's try the match again.
15 The edges of conservation area 2A are shown
16 by the red lines, so you get an approximate line up
17 there. And then I take my point, again 567000, I can
18 make an alignment right there. If you move that
19 slightly, you will see that the black line is exactly
20 superimposed over the blue tic point or the UTM
21 coordinate. You do that top and bottom. Then you
22 have to get it in proper north/south orientation on
23 the edge. It requires a fair amount of work to get
24 one lined up.
25 When you have one lined up to the point
291
1 where you are satisfied, we have little pieces of
2 tape made for that purpose that fasten it so it
3 doesn't move.
4 Q. Then you put the CIR over it?
5 A. Yes. Now, for example, on this one you see
6 a feature that is a dredged lake. That feature has
7 the same bottom line as that, so a feature such as
8 that on the edge can be used to line up. By holding
9 your finger there, you can rotate and bring in other
10 features.
11 For example, there are some roadway
12 patterns down here where you can make a good
13 alignment. So you know you have alignment, at least
14 in the developed area. That was convincing that it
15 was accurate.
16 Q. This is what I really wanted to ask you.
17 These little boxes appear to be significantly larger
18 in many cases than the circular patterns.
19 A. That's correct.
20 Q. How would you make a determination as to
21 what vegetative community was in that box?
22 A. I had to go on the premise that was given
23 me that the feature I was looking for was in the
24 center of that box. That box was small enough that
25 it's fairly easy to determine what is the center.
292
1 And you are also given a piece of information that
2 you are going to be looking for something that is
3 uniform about that center dot. So you are not going
4 to be looking for a tiny feature that is not uniform.
5 You are going to be looking for something that's
6 uniform.
7 So I would look. And if by chance I found
8 a uniform feature, after I was happy that it was
9 lined up, I would try to make a judgment what that
10 vegetation was. And then that is the answer that I
11 would give to Mr. Downing as the quality check.
12 If I was unable to determine any uniform
13 signature representing vegetation, I had to write on
14 the sheet "unable to determine." I handed several
15 sheets in that were that way.
16 Q. You started with 35?
17 A. I think I started with 35, and I had
18 promised I would have at least 35 data points. There
19 were several sheets I had to admit I could not judge
20 what piece of vegetation the satellite imagery had
21 picked out, so I did 10 more.
22 Q. You mean the aerial photography?
23 A. Well, that's correct.
24 Q. You said that the satellite imagery --
25 A. The satellite imagery you see is what
293
1 picked the random points out.
2 Q. But you couldn't tell?
3 A. I couldn't tell on the aerial photographs
4 what it was, so we ended up with a fewer than 35. So
5 I went ahead and --
6 Q. Can you recall how many? Was it more than
7 five points?
8 A. It was less than five. I believe it was
9 three out of 35.
10 So I went ahead and I circled numbers 36
11 through 45 to get ten more. And in that ten I had
12 two or three that I was not able to determine. But
13 we ended up with a number that was in excess of 35.
14 I don't remember what it was actually.
15 Q. Who decided that 35 would be a sufficient
16 number for an accuracy assessment at this point?
17 A. Mr. Downing and I had agreed that based
18 somewhat on time limitations you would like to do as
19 many as you can. But at 35 you have the number at
20 least where you can quantify what your accuracy is.
21 The statistics of 20 is quite often handed
22 out as a number of points for checking. But there
23 is, of course, the phenomenon that we discussed
24 earlier the more points you check the more accurate
25 it is until you checked all of the points in the
294
1 entire environment.
2 Q. Was the number 35 determined by any sort of
3 a statistical model or formula, do you know?
4 A. No, it was not.
5 Q. That was just yours and Mr. Downing's
6 opinion what might be a sufficient number?
7 A. Yes.
8 Q. So you filled out your data sheets and gave
9 them to Mr. Downing, is that the extent of your
10 involvement?
11 A. That's the extent of my involvement.
12 One other step. In some of the situations
13 there would be a pattern in the vicinity of the point
14 that I was to determine.
15 Q. The square?
16 A. The square. For those, I made a sketch
17 description of what I saw with some details of what I
18 thought the pieces were.
19 Mr. Downing called that point up on the
20 computer. That was the actual satellite image. He
21 asked me to tell him yes or no, was that the same
22 pattern that I saw. And in all cases I said yes,
23 that's the pattern.
24 The point that I intend to be what it is is
25 that. That was blind for what the vegetation is
295
1 because I was not able to know what his grid pattern
2 was on the computer screen. I was just telling him
3 whether or not it matched what I saw as a pattern in
4 the aerial photographs.
5 Q. How many points did you do that on?
6 A. Out of the total 45, I think I recall doing
7 that on about four.
8 Q. What was the purpose?
9 A. The purpose was to tell Mr. Downing what
10 point I meant as my interpretation.
11 MR. CESARANO: Let's take a short break.
12 We have been going for an hour.
13 (Thereupon, a brief recess was taken,
14 after which the following proceedings
15 were had:)
16 BY MR. CESARANO:
17 Q. The aerial photography has a degree of
18 locational error intrinsically within itself, does it
19 not?
20 A. If the aerial photography has not been
21 rectified, it's my understanding this has not been,
22 yes, the scale will be slightly different on one side
23 than it is on the other.
24 Q. It's inevitable, is it not, that the
25 process that you used in your interpretation will
296
1 also introduce some locational error?
2 A. That's correct.
3 Q. Can you estimate the quantity or the amount
4 of total error between those two, the process?
5 A. No, I can't. The dependence that I gave to
6 the process is one of doing as careful of an
7 alignment job of the various factors that I was
8 capable of doing, and then looking for a vegetation
9 pattern which was uniform in that particular area to
10 be what I consider probably was represented by what
11 had been selected.
12 I knew the area had to be uniform. That
13 was one of the characteristics that was plugged into
14 the selection of the points.
15 Q. All right. Now, talking about the
16 uniformity of the characteristics at a particular
17 point, that also is an assumption that may be
18 incorrect because of the differences between the
19 dates of the satellite imagery and color infrared
20 photography, correct?
21 A. There may be some difference there, as we
22 had discussed earlier, that fire may be the main
23 problem in that way.
24 But, in fact, the shapes of slough and
25 sawgrass patterns is a reasonably stable feature --
297
1 through a few years, anyway. The kinds of
2 differences that you find are definitely decades
3 would give you problems. But a year or two or three,
4 shapes are easily recognizable. It would be
5 temporarily erased by fire, but they would come back.
6 Q. What effect on reflectance in the color
7 infrared photography would taking photographs at a
8 different time of the day have?
9 A. Well, that's known to be a problem. That's
10 where the user needs to be aware of the gradations of
11 tones that may be available, may be represented in a
12 particular photograph, and needs to do some homework
13 as I have described of looking over the photograph
14 and comparing it with a later date photograph to
15 become comfortable with the ability to interpret the
16 particular colors that are on a particular
17 photograph.
18 Q. And how did you account for the possible
19 error -- not in location, but differences in
20 reflectance of characteristics or features -- that
21 were not directly on NADER?
22 A. The biggest difference on reflectance had
23 to do with water. In some cases you get direct
24 reflection off water. Generally in infrared
25 photography, water is in sky blue or black.
298
1 You need to be aware that in the side of
2 the photograph where there is a reflection, that you
3 can't use that.
4 Q. I guess I wasn't talking about reflectance
5 of water, I was talking about reflectance as
6 described in brightness values of the vegetation.
7 A. I recognize that there are a number of
8 factors that could throw my interpretation off. The
9 one thing is that all of those would tend to diminish
10 my ability to be accurate.
11 So there aren't any of those factors that
12 by chance help you out. The general trend is to
13 diminish your accuracy. So when you end up with an
14 accuracy assessment done this way, the number that
15 you get is probably low based on those things, not
16 based on your ability -- on your knowledge of the
17 vegetation. It's based on your ability to interpret
18 it.
19 Q. What do you mean when you say the numbers
20 are probably low?
21 Do you mean that the percentage, the
22 accuracy assessment, is lower than the number -- the
23 number is lower than the accuracy of the map, or
24 because of these difficulties the number is lower
25 than you would like it to be?
299
1 A. What I intended to say there and didn't say
2 it very well is that my accuracy check, it's
3 reasonable to think that that might be, in fact,
4 lower than the accuracy of the image that was
5 generated, because there are so many factors in
6 interpretation that can deteriorate your ability to
7 judge the vegetation. They don't by chance enhance.
8 Maybe occasionally one enhances. But in general when
9 you have several categories of vegetation, that those
10 things deteriorate your ability to do that.
11 What I'm intending to say is that the
12 accuracy assessment gives you the low side of the
13 accuracy of the map that you are producing.
14 Q. And those things which can tend to
15 deteriorate your interpretation, we have talked about
16 a couple of them, the time of day of the photography?
17 A. Yes.
18 Q. Whether or not the image that you are
19 interpreting is on NADER, in the center of the
20 photograph or off to the side, whether the
21 photography itself is on NADER or off NADER, the
22 differences in the time of the year between the
23 photograph and the satellite image?
24 A. That is right.
25 Q. What other variables might deteriorate your
300
1 ability?
2 A. Did you mention fire?
3 Q. Fire. I want to talk more extensively
4 about that in a minute. But what else?
5 A. I think that you have --
6 Q. How about atmospheric conditions?
7 A. Yes. Atmospheric conditions would have a
8 good bit to do with what you see. The worse
9 atmospheric conditions erases your ability to see
10 anything, and that's cloud cover.
11 In the absence of cloud cover, various
12 degrees of haze deteriorated your ability, for
13 example, if the visible range of haze reduces
14 everything to a shade of blue.
15 Q. What about the type of photographic
16 equipment that is used? You use two different dates
17 of infrared photography. Do you know whether the
18 same photographic equipment was used?
19 A. I do not know.
20 Q. Could that also have an effect?
21 A. Yes. That has an effect, but that's one of
22 the things you are compensating for by studying the
23 particular aerial photograph. That is one -- in
24 fact, all of those variables that we talked about
25 have to do with judging those variations.
301
1 Q. And generally, is it not more difficult to
2 recognize or define a particular vegetative class at
3 a 1 to 64,000 -- almost 1 to 65,000 scale than it is
4 at a 1 to 24,000 scale?
5 A. Yes. In trying to learn the details of
6 this photograph, I had to pick features that were
7 larger than I would have picked on the other scale on
8 the 1 to 24,000. So the typical feature is less than
9 half the distance across on this smaller scale
10 photograph.
11 Q. Now, the last time we were together was, I
12 believe, last Tuesday, about ten days ago, nine days
13 ago, something like that.
14 Did you do any of this work that you have
15 described for the 1985 image before we met last
16 Tuesday?
17 A. No.
18 Q. So everything you have told me, all of the
19 studying and interpreting and learning about the
20 various characteristic of the vegetation reflected in
21 the infrared photography as well as actually locating
22 the points and interpreting the photography over the
23 grid has been done in about eight or nine days?
24 A. Yes. It was all done in the last two days.
25 It began Tuesday, the 5th of April.
302
1 Q. Would that have an effect on your ability
2 to accurately identify and interpret the vegetative
3 classes, the fact that you had such a compressed time
4 frame?
5 A. The -- no. I don't believe that my ability
6 to recognize the classes of vegetation had -- was
7 hampered, but perhaps the number of accuracy points
8 that I could have chosen might have been higher had I
9 been still working today.
10 But I was comfortable that I had learned as
11 much as I was going to be able to learn about the
12 interpretation of this photograph in the time frame
13 that we had.
14 Q. You did this up in Kennesaw?
15 A. That's correct.
16 Q. What time would you begin in the morning?
17 A. On Tuesday. I arrived there just
18 afternoon. So I spent from about 1:00, and I believe
19 that we quit at around 11:00 or midnight.
20 Q. And the second day?
21 A. The second day I got there at 8:00 and I
22 left at 11:30 at night. Perhaps an hour and a half
23 out, total, during the day for meals.
24 There are enough breaks. I think it's
25 important to establish that eye strain, it becomes a
303
1 factor. And I know about that factor and I know I
2 take breaks and try to do other things for a while
3 and then get back on to looking through the loop.
4 Q. What were you doing before you went to
5 Kennesaw on Tuesday?
6 A. With respect to the vegetation
7 interpretation, I was not actively involved in that.
8 I was through -- the days between I was primarily
9 finishing the final draft of my book, the final
10 chapter to turn into the St. Lucie Press -- which I
11 did Monday afternoon.
12 Q. The accuracy assessment of the 1993 map
13 that we discussed thoroughly the last time we were
14 together, I understand that there was a problem in
15 software that generated the random points; is that
16 correct?
17 A. Yes. They had to be regenerated. The
18 first time around the 3 by 3 pixel characteristic
19 that we had chosen didn't work, and we did it again.
20 Q. Doing it again, wasn't that done after your
21 deposition, the first session?
22 A. I'm sorry. I have lost track of time. In
23 fact, I can't tell you now. I believe that was after
24 my first deposition. Yes. We regenerated -- time
25 flies when you are having fun.
304
1 We regenerated a complete set of points,
2 that's correct.
3 Q. And how many did you interpret the second
4 time?
5 A. The second time, I don't know. Mr. Downing
6 has that number. Off the top of my head, I don't
7 recall what the actual number was. It was on a grid
8 sheet that we have here.
9 Q. I think we have it. I was wondering what
10 your recollection was.
11 A. I don't have it in my head.
12 Q. Do you know what the percentage accuracy is
13 for the 1985 map?
14 A. For this, no, I don't.
15 Q. You mentioned a couple of times that fire
16 could have an effect. Did you recognize or determine
17 whether or not there had been fire in the areas that
18 you were interpreting?
19 A. Yes. I think that there had been fire in
20 the western portion of 2A, and that a number of grid
21 points fell there and --
22 MR. WATTS-FITZGERALD: Can we just clarify,
23 are we talking about the accuracy assessment or
24 the training points?
25 THE WITNESS: Accuracy assessment grid
305
1 points fill in that area. That was much more
2 difficult to determine what was there. That may
3 have contributed to the inaccuracy, also. Much
4 more so than the areas that had not been burned.
5 BY MR. CESARANO:
6 Q. Do you recall how many points fell in the
7 burn area? Do you want to go over and look?
8 A. We could take a look.
9 It's a maximum of four points that were in
10 an area where the burn was fresh enough that it was
11 recognizable as a fire scar.
12 Q. And in the western area as you described
13 was the only place where there was a recognizable
14 burn area?
15 A. There had been earlier fire elsewhere.
16 There had been some recognizable fire
17 features elsewhere, but it apparently had been
18 earlier. As it wasn't as bothersome, as fire in the
19 western portion was a problem in interpretation.
20 MS. RAEPPLE: Can we take a quick break?
21 MR. CESARANO: Yes.
22 (Thereupon, a brief recess was taken,
23 after which the following proceedings
24 were had:)
25 BY MR. CESARANO:
306
1 Q. We are looking at the map, "Supervised
2 Classification of Landsat multispectral Thematic
3 Mapper Scene," acquired November 2, 1985 that was
4 brought here today. Prepared date, April 5, 1994.
5 Looking at this map I see some of what
6 appear to me to be unusual characteristics. I'm
7 talking about this curved area on the western part of
8 the map and what appears to be a rather straight
9 line, demarcation between two types of -- two
10 classes.
11 Can you explain what the reasons for those
12 types of features are?
13 A. Yes, those are apparently fire -- what I'll
14 call fire scars and --
15 Q. And the same with this larger area in the
16 middle, as well?
17 A. Yes.
18 Q. And with all of those fire scars there were
19 only 4 points?
20 A. The area where I looked at where I had
21 trouble with my photography looking at this fire scar
22 that continues down the western side and the northern
23 apex.
24 Q. Have you yet read either Dr. John Jensen's
25 report or Ken Rutchey's report to go along with your
307
1 maps?
2 A. No, I have not.
3 Q. You told us last time we were together that
4 there was some discussion that you may be asked to
5 look at Mr. Rutchey's map and slides in order to
6 verify certain portions.
7 Have you done that?
8 A. No, I have not done that.
9 Q. Do you know whether you will do that?
10 A. I will not.
11 Q. You also discussed when we were last
12 together the effect of air boat trails on monotypic
13 cattail and how it was your opinion that the air boat
14 trails were a factor in encouraging monotypic cattail
15 growth.
16 Is that an accurate recollection of mine?
17 A. Yes. I would call that a hypothesis, but
18 one of very reasonable basis that disturbance is
19 characteristically the opportunity that affords
20 cattail the ability to take root.
21 I have known that, for example, buggy
22 trails in the Big Cypress swamp frequently grow up as
23 a line of cattails.
24 Q. Now, the area that you are referring to is
25 the area generally south of the Hillsboro canal?
308
1 A. That's correct.
2 Q. Can you tell me whether -- let me say that
3 area is an area obviously dense cattail which is an
4 area of great interest and has been recently,
5 correct?
6 A. Correct.
7 Q. Can you say whether the air boat trails
8 caused the cattail or whether the cattail caused the
9 air boat trails, because it is such an area of
10 interest?
11 A. I think that it is reasonable that the
12 disturbance caused by air boats is one of the
13 disturbance factors that allows cattail to come into
14 domination.
15 There are other places clearly where
16 cattails grow without the intervention of an air boat
17 disturbance.
18 But the particular area along the Hillsboro
19 canal, particularly from an entry point where air
20 boats apparently use a boat ramp, I haven't been
21 there on foot, but you can see it in the aerial
22 photographs, that the number of trails emanating from
23 that point is just a complete maze in this corner.
24 It has also grown up with cattail.
25 So I think there is a reasonable assumption
309
1 that air boats may have been one of the contributing
2 factors in cattail growth there.
3 MR. CESARANO: I don't have anything
4 further at this point. Thank you.
5 CROSS EXAMINATION
6 BY MR. WATTS-FITZGERALD:
7 Q. Good afternoon. While we are up and moving
8 around anyway, on the light table I put the photo you
9 looked at earlier, which was the 3/3/84 photo.
10 And based on what you were just telling
11 counsel, I have been looking at the stretch south of
12 the Hillsboro canal between two visible discharge
13 structures, S-10 structures along the ridge of the
14 canal running down to the intersection that seems to
15 coincide with the air boat launch points you have
16 been describing.
17 Can you look through the loop and tell me
18 if that's the area you are referencing to counsel
19 regarding cattail growth and air boat trails?
20 A. Yes.
21 Q. That air boat launch point at the actual
22 turning point of the north/south canal on the eastern
23 border to the northwest, there is a roadway
24 intersection there. Do you see that on the photos?
25 A. Yes, I see it.
310
1 Q. Can you tell me, based on your experience
2 with this scale on the work that you have done, the
3 approximate size of that intersection?
4 A. Are you asking how many feet it is across
5 the intersection?
6 Q. Yes.
7 A. Well, it's -- from aerial photo
8 interpretation, something as white as the roadway you
9 see there, it's improper to try to tell how wide it
10 is because of the resolution phenomenon in films.
11 That if there is a very bright reflectance, the
12 signature of the feature is substantially wider than
13 what you see.
14 So if I were to try to judge the distance
15 based on the pattern that I see, I would overstate
16 it.
17 Q. Did you use any of those features in that
18 immediate vicinity as ground orientation points for
19 the process you described for 1985?
20 A. Yes, therefore you need to center on it.
21 But the roadway even at the most -- maybe the road is
22 75 feet wide. So we are talking about one pixel.
23 Q. And as you go from that intersection of the
24 levees and follow the Hillsboro canal to the
25 northwest, the structure or the vegetation that comes
311
1 in what I would call somewhere between a pink and a
2 red -- maybe magenta, what would that be?
3 A. That's generally shrubs. Cattail is one of
4 the main features that does that.
5 Q. The yellowish material to a little bit
6 further to the south into the water conservation area
7 many times associated with open water bodies, what
8 would that be?
9 A. If it's greenish-blue, that's cattail. If
10 it's bright yellow, what we are finding is cattail
11 that probably is underlain by either a fern called
12 Salvinia or duck weed.
13 Q. How did you make that determination?
14 A. We did that by our ground truth points.
15 That's the way the color signature came out.
16 Q. That's the ground truthing that you did?
17 A. For the first time around.
18 Q. Using a different set of photos with a
19 different scale?
20 A. That's correct.
21 Q. Did I understand you to say this set of
22 photos were not geo rectified?
23 A. I don't know if they were. They are a lot
24 closer in falling perfectly from side to side than
25 the first ones were, much closer.
312
1 Q. So the first ones weren't?
2 A. No.
3 Q. You don't know if these were?
4 A. I don't know, but these were much easier to
5 work with.
6 Q. You testified last time Dr. Downing was in
7 the process of attempting to digitize the DBA
8 photography to use for the earlier process, the 1 to
9 24,000. Was that done?
10 A. Not to my knowledge, that was not done.
11 Q. Have these been digitized?
12 A. Not to my knowledge.
13 Q. What was done, if anything, to your
14 knowledge to correct the BDA photos for any effects
15 induced by flight characteristics of the aircraft or
16 the path, actually?
17 A. The corrections that we did were what I
18 have described of using as many features along the
19 edge of the water conservation area that are easily
20 identifiable on the photograph as truth points and
21 lining up on those points.
22 As you progress further into the water
23 conservation area we use air boats if the signature
24 shape of the air boat was easily recognizable as
25 being the same as one that is visible on the
313
1 photograph. Air boat trails are present on the USGS
2 quadrangle sheets. Tree island shapes and sizes
3 helped in our location. Occasional houses -- these
4 are all of the things that I had gone through in the
5 last deposition for locating features.
6 Q. Were you aware that on the quadrangle sheet
7 published by the Geological Survey they identify the
8 date of revisions and the currency date of those?
9 A. Yes.
10 Q. The USGS quads that you brought with you
11 here today, are these the ones you actually used and
12 reduced?
13 A. Yes.
14 Q. If I look at a legend and find it was
15 edited in 1983, that would imply, would it not --
16 photo revised in '83, I'm looking at Ft. Lauderdale
17 to the southeast.
18 No features on this had been verified,
19 anything later than that date?
20 A. That's correct.
21 Q. So approximately something on the order of
22 November 2 is the end of the year. You got to be
23 looking at virtually two full years --
24 A. Yes.
25 Q. -- on the photo revision?
314
1 A. Yes.
2 Q. And in the same photo counsel is perusing,
3 which I identified earlier, you talked about the air
4 boat trails, and I think I see what you are talking
5 about.
6 Starting from that same intersection of the
7 north/south levee of 2A and the Hillsboro canal
8 running to the northwest, I see what appear to be
9 several fairly prominent air boat trails with an open
10 water track.
11 Is that what those would be?
12 A. Yes, that is correct.
13 Q. There are very prominent ones due east,
14 perhaps a kilometer or less south of the
15 intersection?
16 A. Yes.
17 Q. And the area, according to your supervised
18 classification for 2A on the November 2, 1985
19 thematic paper, would seem to show relatively
20 monotypical cattail in the area north of that trail.
21 Would that be correct?
22 A. That's correct.
23 Q. Now, what is the pattern within 2A, if you
24 know?
25 A. It's from north to south.
315
1 Q. Based on your training and experience as a
2 biologist and one with more than a passing interest
3 in the Everglades, would it be fair to say if the
4 water is moving north to south and the air boat trail
5 runs east to west, no factor in the growth of cattail
6 in that area could be described to the activities in
7 that area?
8 A. The immediate vicinity of the air boat
9 trail where there is disruption underneath the air
10 boat and moderate weight that is thrown by the air
11 boat, there would be an impact.
12 If we are talking about anything else, any
13 air boat factor would have to extend a little bit
14 south or north. But generally what I'm talking about
15 is disturbance by an air boat is a very, very local
16 phenomenon having to do with actual movement of water
17 created by the surge of water created.
18 Q. We are talking 10 meters, maybe less?
19 A. Probably less.
20 Q. So you would not ascribe the monotypical
21 cattail growth in the Hillsboro canal to cattail air
22 boat?
23 A. In careful scrutiny of some of the very
24 dense areas near this roadway intersection at the
25 northeast corner of 2A, by looking at several aerial
316
1 photographs you can see the area is completely
2 permeated with air boat trails. It's a solid feature
3 of air boat trails. That extends along the Hillsboro
4 canal to the west for some distance.
5 I don't have an actual distance but enough
6 that I was convinced that there may be a relationship
7 between air boat disturbance and solid growth of
8 cattail.
9 That also extends south along the eastern
10 edge of water conservation area 2A. There is
11 frequently a signature where you can see something
12 has passed in the shape of an air boat trail that has
13 now become vegetated and is no longer only water.
14 Q. If the air boats are launching in the open
15 water of the canal, how is it any different from a
16 boat doing the same thing, traveling the same water
17 body?
18 A. I would expect that any impact would be
19 less from an air boat than it would be from a boat
20 that has a propeller under the water.
21 Q. Would you expect boat operations, then, in
22 the canals further south in the Everglades ecosystem
23 to induce cattail growth along all the canals in
24 which they operate?
25 A. I would say that boat operation would be
317
1 one disturbance factor in a canal that could help
2 establish cattails along the side of the canal. But
3 there are many other things.
4 There is wind blown water movement along
5 the edges of canals. So canals, I think, have
6 problematic cattail lines along the edge. It could
7 be ascribed to many factors. I think boats may be --
8 my personal judgment is boats may be minor in canals.
9 Q. Moving away from the northeast corner a bit
10 and the immediate vicinity of that heavy air boat
11 impact that you described, looking between the two
12 S-10 structures depicted in the photo running south,
13 let's say, eight to 10 kilometers, there is still a
14 considerable amount of cattail shown in those areas.
15 Would you expect any of that impact to have
16 been derived from canal borne air boat or surface
17 propeller driven boat activity? I'm talking down in
18 here.
19 A. I would say that away from the actual
20 immediate vicinity of the boat passage -- by
21 "immediate" I'm talking a few yards on each side --
22 then I would say that that would be the extent of the
23 effect. So my concern only has to do with numbers of
24 air boats.
25 Q. You said you looked at other photos of that
318
1 roadway canal intersection, northeast corner of 2A.
2 What other photos have you looked at?
3 A. There are two dates represented here that
4 are clustered close together within a year of one
5 another. Then there is the 1993 Breedlove, Dennis
6 color infrared. Those are the only ones I looked at.
7 Q. What other date from the mid '80's did you
8 use to do the actual accuracy assessment, or did you
9 only use this run?
10 A. I only used these two sets of photographs
11 which are in '83 and '84.
12 Q. I didn't understand. You had used both
13 sets for the accuracy assessment?
14 A. Yes.
15 Q. Have you made any effort to determine
16 through historical sources or anecdotal data the
17 duration and extent and, in fact, timing of the fire
18 that may have contributed to the fire scars in 2A
19 that you testified to earlier today?
20 A. No. I recall that I said I was going to
21 attempt to research that in the last deposition. I
22 did not.
23 Q. Do you plan to do that between now and the
24 hearing?
25 A. No, I don't.
319
1 Q. You talked about studying the NHAP photo
2 sets from '83 and 84 to get a sense of what you were
3 looking at, that you utilized the training
4 opportunity of the BDA photos to enable you to
5 interpret the signatures on spectral signatures on
6 these photos; is that correct?
7 A. That's correct.
8 Q. Did you do that process side-by-side?
9 A. Yes.
10 Q. So you still have the BDA photos?
11 A. They are here.
12 Q. I didn't see them.
13 A. Yes.
14 Q. Just keep track. We didn't get them last
15 time. We may need to get access to them at some
16 point.
17 You testified about 10 days ago that the
18 western portion of WCA-2A was an area not included
19 within the coverage of the flight paths of the
20 Breedlove, Dennis aerial photos; is that correct?
21 A. Very small area is not included.
22 Q. Were any of the accuracy assessment points
23 located within the area not covered by the BDA
24 photos?
25 A. No.
320
1 Q. Did you seek to determine if any of the
2 points selected by the software or whatever it was to
3 be used for accuracy assessment coincided with the
4 training points used in the earlier map generation
5 process for the 2A for the 1993 image?
6 A. I did not check.
7 Q. Are you aware of the bias that would be
8 induced in an accuracy assessment if that, in fact,
9 occurred?
10 I guess, to put it another way, isn't that
11 a statistical no-no?
12 A. The training points were shown on our
13 accuracy check of the 1993 photography. And the
14 accuracy assessment points at that time did not
15 include any of the training points.
16 I did not check to see if, in fact, any of
17 the accuracy points for the national high altitude
18 program photographs. The older photographs were the
19 same as some of our recent training points.
20 Q. You understand the bias that induced?
21 A. Yes, I understand.
22 Q. I am thinking it through. Tell me if I'm
23 wrong here, but if you select a point for your
24 training point and you later say, gee, let that be an
25 accuracy assessment point, and you ground visited,
321
1 you hope it's 100% because you went there and checked
2 it?
3 A. Yes. That is within the same date
4 photograph.
5 Now, if an accuracy assessment point on the
6 older photography had happened to fall on what was a
7 training point in the more recent photography, I
8 don't think that introduces a bias, because what we
9 are looking at is possibility of vegetative change
10 and that could occur anywhere.
11 Q. But I understood you earlier to say that
12 you engaged certain presumptions about the stability
13 of significant vegetative features within the area,
14 and that formed a substantial basis for your
15 assessment of vegetative community identification in
16 the earlier photos.
17 A. That's correct.
18 Q. Understand I'm not asking you to quantify
19 this yet.
20 A. Because I can't.
21 Q. That's the next question. But I guess you
22 sort of made that point, if there is a point.
23 Who picked the features to be used for the
24 assessment, the accuracy assessment? I think I
25 understood you to say there was 100 -- initially 100
322
1 feature points?
2 A. The computer picked the points.
3 Q. Randomly generated. You don't know what
4 other program was involved, just random?
5 A. I was told it was random.
6 Q. Paring it down to 35, was it simply an
7 arbitrary decision to take the first 35 to maintain
8 the randomness in some fashion?
9 MS. RAEPPLE: Object to the form.
10 THE WITNESS: May I answer the question?
11 MS. RAEPPLE: Yes, if you understand it.
12 MR. WATTS-FITZGERALD: You certainly may.
13 THE WITNESS: If we had selected points
14 throughout the 100, we could have introduced a
15 bias as to where those were. By selecting them
16 in actual sequence, we avoided any kind of bias.
17 BY MR. WATTS-FITZGERALD:
18 Q. Are you familiar with the substantial body
19 of literature in photogrammetrical interpretation
20 that deals with how one selects the number of points
21 for statistical significance in an accuracy
22 assessment?
23 A. No, I'm not.
24 Q. So you are relying on Mr. Downing for that,
25 I would imagine?
323
1 A. Yes.
2 Q. Was any effort made in constructing the
3 program or selecting the points for assessment to
4 insure that all vegetative communities representative
5 of the area of the variation in the area were
6 adequately represented to avoid false results in your
7 assessment?
8 A. You have to ask Mr. Downing. It's my
9 understanding that the randomness was purely
10 geographic in nature and not that there wasn't a
11 factor built into the selection to insure that all
12 gradations were selected.
13 If that was built into the selection
14 process of the points, then I don't know about it.
15 It could have been done and I wouldn't know.
16 Q. In the actual results, because you saw the
17 results, the 35 that you examined, what was the
18 distribution of those 35 amongst seven or 10 whatever
19 it is categories of vegetative classes actually
20 reflected?
21 A. I don't know the actual numbers. But most
22 of them were in sawgrass, a lesser number in cattail.
23 And I know that on at least one occasion it hit dead
24 center on top of a tree island.
25 Q. Which type of tree island?
324
1 A. It was a head of a tree island, what would
2 have been called the bay head. Whether or not it was
3 actually the bay making the signature --
4 Q. Were you aware that the spectral signatures
5 on some of the satellite bands of a bay head is
6 indistinguishable from a brush, sawgrass, tree
7 island?
8 A. What little I have done in trying to
9 recognize things like that, I think my opinion is
10 they are not distinguishable and weren't
11 distinguishable, based on what I have done.
12 I did not put down bay head as the answer.
13 I put down tree island.
14 Q. You think its proper to classify a mixed
15 brush sawgrass community as a tree island?
16 A. We had some check points built into the
17 first determination that showed a slight pink dotting
18 represented by -- it could have been bay, could have
19 been willow, it could have been wax myrtle -- into a
20 field that otherwise appeared with the color and
21 texture of sawgrass.
22 Based on our ground truth, we realized that
23 a small number of pink dots in a field that looks
24 like sawgrass should be called sawgrass because the
25 sawgrass is overwhelmingly dominant. The only time
325
1 in a tree island feature where we called it actually
2 a tree island is in the head of the tree island where
3 there is an entire cluster of pink.
4 Q. Is cattails in the Everglades ecosystem or
5 South Florida in general senescent in November or
6 December?
7 A. The site visits we did in November and
8 December showed cattail still with a substantial
9 amount of chlorophyl. We did not see seed heads at
10 that time of the year.
11 I'm aware from other work that I have done
12 that cattail in February and March is frequently
13 brown. We still found it green with some indication
14 that it was becoming senescent -- that is, yellow and
15 orange colors coming in. Also chlorophyl.
16 Q. I believe you testified last time that that
17 effect could be accelerated or would be associated in
18 your mind with colder weather onset?
19 A. And hypothesis based on what other plants
20 do. But that's strictly a hypothesis. I stated that
21 because of my observations, that I had seen cattails
22 turn to colors where there should be easily
23 recognizable at least in visible bands from sawgrass
24 during a so-called senescent stage when it ultimately
25 turns brown. I had seen that intensely in large
326
1 areas in February and March.
2 Q. What month of the year were the BDA runs
3 made?
4 A. We have the actual date here and they were
5 in early '93. I have to get the actual -- they were
6 the 2nd of February, '93.
7 Q. Do you have any idea what the prevailing
8 weather conditions were for that?
9 A. Other than '93 was a mild winter. In terms
10 of intense cold fronts, I don't have it.
11 Q. How is your memory for 1984 and '85? How
12 is your memory for back then, November and December,
13 whenever it was back then?
14 A. '89 I remember fairly distinctly.
15 Christmas '89 was a rather harsh freeze. But, in
16 fact, I don't know for those earlier years.
17 Q. In November 2, was the shoot in '84 -- '83.
18 November 2, '83 is the accuracy assessed?
19 A. Yes.
20 Q. That's what we were just looking at, the
21 photos were shot 10 months from that 11/85 -- 3/3/84
22 and 2/85.
23 So we have February photo imagery and we
24 have -- of 10 months earlier than the satellite, and
25 you said you had a second set of CIR from March of
327
1 '84, which is 8 out in the annual cycle and 20 months
2 out in actual time.
3 A. Yes.
4 Q. Given the vagaries that might be induced by
5 weather conditions on the state of the cattail and
6 its spectral signature, could you account in any
7 fashion for possible error that might be induced in
8 the interpretation?
9 A. Yes. What I would say about that is that
10 all of those things would tend to deteriorate my
11 ability to judge what was seen by the satellite at a
12 different date.
13 The fact that the photography is at a
14 different time of the year, slightly different time
15 of the year and a different date. So I think that
16 that will be reflected in how well I did in the
17 accuracy assessment, because all of those things
18 would tend to deteriorate the accuracy assessment.
19 It would be a rare combination of
20 circumstances that would make it improve an accuracy
21 assessment.
22 Q. What I'm not quite able to follow here is
23 how something that causes a misclassification that
24 nobody can ground truth will show up as inaccuracy
25 when you are working back in time. I'm not sure I
328
1 made that clear for you. Did I?
2 A. No, you didn't.
3 Q. You have said several times that any of the
4 problems that can crop in that would affect your
5 ability to interpret the photo -- like atmospheric
6 conditions, fire, et cetera, would ultimately show up
7 in the accuracy assessment and would be reflected in
8 the ultimate assessment.
9 That may well be. But the problem I have
10 is if the inaccuracy is that you confidently
11 misidentify something because over that time span the
12 vagaries of nature distort what you think you are
13 seeing, I don't see how that shows up in the accuracy
14 assessment.
15 A. Okay. Where that shows up is in building
16 the basis for the interpretation of the satellite
17 imagery data I fed to them. And we really haven't
18 discussed a number of points for establishing the
19 vegetation of that time frame.
20 I gave them several numbers of points with
21 UTM data coordinates, what I said was cattail in the
22 '83 or '84 photographs. So that was built in.
23 Whatever signature was read by the satellite data,
24 that information was used to train the computer to
25 recognize cattail, to recognize sawgrass, to
329
1 recognize broad leaf vegetation, as would be present
2 in a tree island.
3 Q. And you knew how to classify that for the
4 November 2 run back then in '83, because you had done
5 a similar process complete with field visits in the
6 training for '93 and you told the '93 program this is
7 what you will find at the training points. Then it
8 went through the algorism. It generated ultimately
9 the map for 1993.
10 Now you are saying I'm going to read that
11 training ability into the 1983 run using photos that
12 I will interpret based on my 1993 experience. And if
13 your 1993 experience does not translate accurately
14 because the physical condition of the foliage is not
15 the same or reasonably close, an error will be
16 induced that we have no way of calculating in your
17 accuracy assessment because the program is blind to
18 it and you, in a sense, are blind to it as well
19 because you can't ground truth it?
20 A. That's correct. And what I'm saying is
21 that in our accuracy assessment, that that should be
22 reflected in a very low success rate. The number of
23 times that I -- so to say by the process of not
24 knowing exactly, I'm calling it a guess, what the
25 vegetation is, it's always an educated guess. I will
330
1 make a number of wrong guesses against an image that
2 has been trained by wrong information.
3 So you put the two together. I don't mean
4 wrong information, I mean by guesses based on '83 and
5 '84 photography. So if you put the two inaccuracies
6 together, it should make your accuracy assessment
7 come out worse and worse and worse. So what I'm
8 saying, if we get some reasonable degree of accuracy,
9 I'm trusting and crossing my fingers what Mr. Downing
10 did with my numbers that you will get tomorrow, that
11 if that is a reasonable number, that I'll hang my hat
12 on it and say we evidently found something.
13 Q. I hear a lot of like professional judgment
14 calls being articulated there. Is that a fair way to
15 characterize the process that you went through, both
16 for the '93 and the '83 process in identifying what
17 is there?
18 A. Yes.
19 Q. The process that you have pursued with this
20 serial overlay and estimation process for accuracy
21 assessment, how often in your experience in the field
22 of color IR or interpretation have you used that
23 methodology?
24 A. Do you mean prior to the exercise that we
25 have been discussing at this table?
331
1 Q. That's not the methodology used for '93?
2 A. Correct.
3 Q. So for the '83 -- '85 assessment, November
4 2 run, have you ever done it that way before?
5 A. I based it on work -- I based it all on
6 photography, but training myself by oblique 35
7 millimeter aerial photographs where you could, in
8 fact, recognize particular plants.
9 Let me give you a time frame -- a frame of
10 reference a little bit better. This would be from an
11 airplane or helicopter flying at -- as low as about
12 250 feet, shooting diagonally out the side so you can
13 recognize point blank what a pine tree is.
14 And given my experience in estuarian areas,
15 for example, it's very easy to recognize black
16 return. As you move up it's easy to recognize
17 freshwater, grass. And with that kind of training, I
18 taught myself to recognize features in vertical color
19 infrared photography and conventional color
20 photography.
21 After the fact people have ground checked
22 that and found that I did well.
23 Q. Let's focus on the question now that I
24 asked, which was how many times in the past have you
25 used this methodology as you used in your accuracy
332
1 assessment for the November 2 run?
2 A. I have done it before. I can't tell you
3 how many times. But I have been through so many
4 versions of photo interpretation in consulting work
5 that I have done this before. But not to this
6 degree, that is correct.
7 Q. Am I fair in assuming that this was for
8 like dredge and fill assessment and that sort of
9 thing?
10 A. Yes. That sort of thing, but a more broad
11 spectrum for the information required for a
12 development. Regional impact requires more than just
13 the wetland assessment, it requires the upland as
14 well, complete vegetation mapping.
15 So I have looked at more than just wetland.
16 I looked at upland and wetland.
17 Q. What would be the largest geographic area
18 where you have done this?
19 A. The largest was a portion of a 19,000 acre
20 development on the west coast of Florida. And the
21 largest plat of that, which was done with some ground
22 truth, but almost all aerial photography was 6,000
23 acres. Two other plats added to that, which were a
24 couple of thousand.
25 Q. I may have been insufficiently precise in
333
1 my question.
2 I'm asking you instances where there was no
3 ground truthing because you didn't ground truth for
4 this. You trained yourself, I understand that
5 process. But there is no ground truthing by
6 chronological limitation, the fourth dimension
7 precludes it?
8 A. There are areas in the Big Cypress Swamp I
9 have interpreted without being anywhere near it on
10 the ground.
11 Q. Going back 10 plus years with high altitude
12 CIR photography where you went through the whole
13 process of reducing the quads, adjusting the
14 photography to get a decent scale out of it?
15 A. Not quite to that extent. But, yes, a
16 number of these steps were, in fact, used. We went
17 back to 1951, black and white aerial photography.
18 Q. And when you did that, which was the point
19 of all of this, what was the percentage accuracy
20 assessment that resulted from those various projects?
21 A. I don't recall.
22 Q. Did you do them?
23 A. We had some accuracy assessment, but only
24 for the most modern features. We picked a number of
25 points, not randomly generated, but we picked points --
334
1 let's see, if we interpreted this area and this area
2 and this area, and we went and looked at them and
3 found, in fact, we had done so, we had done it
4 correctly.
5 Q. But assessing it in that fashion in the
6 field of photogrammetry has no statistical
7 significance, does it, when you can't assign a
8 percentage?
9 A. I couldn't have assigned a percent
10 accuracy on that, no.
11 Q. So in these types of projects, you selected
12 prominent features that simply someone perhaps not as
13 skilled as you wouldn't even have the ability to
14 confuse and verified that it, in fact, was accurately
15 reflected over the time period involved?
16 A. A number of the time periods involved, yes.
17 But you must realize one of the things you
18 want to ground truth the most -- in fact, it was done
19 for the Big Cypress work -- was the confusing areas
20 where we thought it was one thing and we went and
21 made sure and found that we had done well.
22 That gives you a sense of professional
23 confidence, but it's not quantified.
24 Q. But your sense of professional confidence
25 in these earlier projects based on a similar, if not
335
1 as expansive, mechanism was sufficient for you to
2 submit it and expect the people for whom you were
3 performing this work to rely on within a reasonable
4 degree of professional certainty?
5 A. That was the whole reason for going through
6 a type of accuracy assessment that we have done, is
7 to try to quantify what I would otherwise have to
8 describe as professional opinion.
9 Somebody in Kennesaw, Georgia at our main
10 office between last night and today is grading my
11 paper.
12 Q. Who is doing that?
13 A. Mr. Downing is doing that.
14 Q. In grading any of -- he didn't grade any of
15 your earlier efforts like this, did he?
16 A. The one last week. But, no.
17 Q. We haven't seen that, so I can't ask you
18 about that.
19 You don't have the number from '93?
20 A. '93, I was told it was slightly in excess
21 of 70%.
22 Q. What type of accuracy assessment is that?
23 Is it producers or users?
24 A. I don't know the difference.
25 Q. That's overall. Do you know the accuracy
336
1 assessment for individual classes?
2 A. I have been told that that was quantified.
3 And that for sawgrass and cattail which we focused on
4 most carefully -- and we, in fact, ended up being two
5 of the classes that are reasonably easy to separate --
6 that we did better. And it was other kind of classes
7 where we did not do so well.
8 Q. If that was done and you know that, why
9 don't we have it?
10 A. That was given to me verbally. I don't
11 have a piece of paper.
12 MS. RAEPPLE: Was that not produced at
13 Mr. Downing's deposition, the accuracy
14 assessment for the '93?
15 MR. CESARANO: No.
16 MS. RAEPPLE: You have never seen that?
17 MR. CESARANO: No.
18 MS. RAEPPLE: Let me see if I have it.
19 BY MR. WATTS-FITZGERALD:
20 Q. While you are looking, perhaps a collateral
21 question. The document that was provided us today
22 that shows land cover classifications for 2A based on
23 the '85 Landsat, the '93 Landsat, have you compared
24 any of those acreages to the analogous classes
25 produced by Dr. Jensen's work for any of his time
337
1 normalized series or Ken Rutchey's work?
2 A. No, I have not.
3 Q. Do you plan to do that?
4 A. I don't plan to. I may come across those
5 numbers. But the actual comparison, I believe, is in
6 the area that Mr. Downing has to substantiate for me.
7 Q. You said that you would not be analyzing
8 Ken Rutchey's photography work to determine the
9 accuracy of his plant identification and field work.
10 To your knowledge, is anyone else doing that now?
11 A. I believe that there has been -- that
12 Mr. Downing has taken a look at the methodologies
13 Mr. Rutchey used, and that whatever aspect of that he
14 will cover will be in his deposition.
15 Q. You understand I'm asking strictly about
16 the ground truthing end of it --
17 A. Okay.
18 Q. -- which I understood you to say 9 or 10
19 days ago you would do if you had the time to get
20 around to it?
21 A. Yes.
22 Q. I'm asking, has Mr. Downing taken over that
23 facet of it?
24 A. He will not be in a capacity to judge
25 whether Mr. Rutchey correctly identified cattail in
338
1 the field or correctly identified sawgrass.
2 So we may be under the assumption that what
3 he calls sawgrass and what he calls cattail is
4 correct.
5 Q. So you, anyway at this point, will not be
6 challenging any of Mr. Rutchey's interpretations that
7 may -- identifications that may underlie his work?
8 A. No. Part of the reason -- the main reason
9 that has been put aside is that the identification of
10 those plants in the field is very easy, and I don't
11 expect that there would be any fruitful basis for
12 challenging those particular plants.
13 Q. I also understood from your testimony a
14 week or 10 days ago that doing that check would be
15 incredibly difficult. Based only on photography and
16 even videotape, it would be very difficult to do.
17 There were problems when you attempted to do that?
18 A. Yes. We established that the field notes
19 are the most important.
20 Q. You described for counsel a circular
21 pattern growth of cattail in the Everglades ecosystem
22 as being one of the features that you utilized on
23 high altitude photography to assist you in your
24 identification and classification; is that correct?
25 A. Yes. That's useful in assisting, but it
339
1 does not follow everywhere. But it appears that
2 cattail in a number of locations start as an
3 individual plant or perhaps a group of plants that
4 started in a some fortuitous circumstances and then
5 reproduced by rhizome development.
6 I described to you having looked at that in
7 aerial photography at an impoundment at the west
8 coast of a Florida development called Rotunda. I had
9 some experience in looking at that in successive
10 years from aerial photography.
11 When I saw these clusters of circles in the
12 field work that we did and saw that the cattails, in
13 fact, were often like that I suppose maybe that was
14 vegetative reproduction that accounted for it. And
15 between the color signature being right and that
16 circular feature, I felt that I could recognize
17 cattail in the older photographs.
18 Q. Doesn't sawgrass also grow in circular
19 patterns many places in the Everglades?
20 A. Yes. That's where you need to have a look
21 at color in addition to that pattern. Now, where
22 sawgrass is a expansive feature over a great deal of
23 territory and a circular pattern in the sawgrass with
24 a slightly different color signature is apparent,
25 then you suspect it's a different plant.
340
1 MR. CESARANO: Off the record.
2 (Discussion off the record.)
3 BY MR. WATTS-FITZGERALD:
4 Q. Do you know if the error analysis that has
5 been done for '93 which you have heard about is -- by
6 word of mouth, does that include a complete error
7 matrix or is it just a reported number by class
8 overall?
9 Do you understand what I mean by matrix?
10 A. Yes. There is a matrix, but it identifies
11 it also by class so that Mr. Downing knows how well I
12 did in various categories.
13 Q. You described for us last time two
14 contracts involved between your firm and your clients
15 for this particular litigation.
16 Are there still just the two contracts
17 outstanding, your personal one to provide advice and
18 one to provide vegie maps?
19 A. It's not my personal one. It is a contract
20 of Law Environmental to provide my services.
21 Q. Your personal?
22 A. Yes.
23 Q. Have there been any extensions on either of
24 those contracts?
25 A. No.
341
1 Q. So despite the work, you haven't exceeded
2 your 10,000 cap yet?
3 A. Okay. The 10,000 cap was for my initial
4 assisting of Hopping, Boyd, Green & Sams in
5 developing whatever strategies or whatever knowledge
6 they needed to work on their aspects of this case.
7 My participation as a witness here is
8 separate from that and does not have a cap on it.
9 But I did not exceed the 10,000 at all on the other
10 work. I don't have the actual dollar figure, but I
11 haven't gone far over the 10,000.
12 I think I'm in excess of it totally,
13 putting everything together.
14 Q. You indicated you had not read or reviewed
15 Jensen's or Rutchey's work but that you had at some
16 time in the past reviewed other vegetative maps.
17 Have you in the last 10 days reviewed any
18 other vegetative maps, whatever source, regarding the
19 Everglades ecosystem to assist you in your work?
20 A. Not other than the historic maps that I
21 have been using in my book, particularly the Davis
22 map of 1943.
23 Q. And in what fashion would you be relying on
24 or basing any of your opinions in this matter with
25 regard to the accuracy assessment or the production
342
1 of the 1993 map on the Davis work for 1943?
2 A. Not at all.
3 Q. Did you ever figure out the significance of
4 the assignment of the letter H, R and E to your
5 training sites?
6 A. They were essentially arbitrary.
7 Mr. Downing had some reason that he could remember
8 what those sequences were. But since there were
9 similar number sequences from different trips, they
10 were assigned a letter designators to make sure they
11 weren't confused.
12 Q. When you were doing the work with the 1 to
13 24,000 grid and IR for the 1993 effort, you testified
14 that originally you started out with 10 sites that
15 would be training sites that you visited in the
16 field. And because of time constraints and other
17 factors, that was cut back to 200, it was cut back to
18 something on the order of 70 plus?
19 A. Those were not training sites.
20 Q. What were those?
21 A. Those were the accuracy assessment sites.
22 We did identify initially 200. Maybe even 250.
23 Q. That was done by random programming?
24 A. That's correct. We cut that back and the
25 actual accuracy check you asked -- I was asked
343
1 earlier how many we had done. I couldn't come up
2 with a number. I did see that I checked off at least
3 through number 89 on the map. I don't remember how
4 many we did. I would think it's probably on the
5 order of 10.
6 Q. That was done as you did for the November 2
7 '83 photo or imagery -- that was also done using CIR;
8 is that correct, color infrared photos?
9 A. Correct.
10 Q. In that process for the '93 map that you
11 produced for 2A last month or a couple of weeks ago,
12 when you dropped from the 200 assessment sites to
13 something on the order of 70 or whatever it was, was
14 that a stratified random sample?
15 A. I don't know. I do know that we had to go
16 through the same process, though, making sure that we
17 took the numbers in order to preserve the randomness.
18 Q. Was any effort made in that process to
19 insure that variability would not be understated by
20 including within that sample sufficient
21 representative sites for each of the classifications
22 appearing in the map?
23 A. I think Mr. Downing will have to answer
24 that. My assumption is that there was not built
25 into -- it was poorly geographic and not to satisfy a
344
1 certain number of points in particular classes.
2 Q. Were you aware that a substantial body of
3 literature in the field of remote sensing identifies
4 ways in which one should geographically distribute
5 your sites for purposes of assessment in terms of
6 both physical geography and classification geography?
7 A. I'm not aware of that. It becomes more
8 difficult to use randomly selected data to quantify
9 actual areas of various plant communities. It
10 becomes more difficult statistically to come up with
11 those numbers. But it can be done.
12 Q. Does it decrease the accuracy of the
13 ultimate acreage numbers generated if you don't use
14 something like a stratified random sample?
15 A. This is getting out of my area. I don't
16 know the answer to that.
17 Q. So the extent of your knowledge is yes, it
18 can cause you problems, but it can't take it beyond
19 that?
20 A. The extent of my knowledge is that we used
21 what I thought was a geographically random process of
22 selecting points. And that I'm aware that that is a
23 good way to quantify an area of various classes of
24 vegetation coverage. And it would get out of my
25 ability to do the quantification if the points were
345
1 other than randomly selected.
2 Q. I obviously had some confusion in my mind
3 over the number of sites that were assessment sites
4 thinking those were training sites. I may repeat
5 just a little bit, but I'm almost done, some of what
6 we covered the last time because I have listened to
7 too many terms of art in this field already.
8 How were the training sites for the '93
9 imagery selected?
10 A. '93 is the recent imagery. The training
11 sites were selected in the field for identifiable
12 vegetation that was of interest in this
13 classification system. So we --
14 Q. So that was a supervised classification
15 process?
16 A. That's correct. I supervised it.
17 Mr. Downing contributed to the supervision by
18 requesting that we have reasonably good coverage
19 aerially of the subject, particularly water
20 conservation area 2A. But we looked at other places.
21 Q. Was any statistical test used for that
22 purpose, to satisfy the concern that Mr. Downing
23 expressed to you?
24 A. To my knowledge, no test was used.
25 Q. This was simply a fly out there and pick a
346
1 site that looks big and of a community that we are
2 interested in?
3 A. That's correct. I'll qualify the "big" by
4 saying that we try to judge that it would be
5 substantially larger than a pixel.
6 Q. You said 10 days ago something in the order
7 of several hundred feet across?
8 A. That would be three pixels across.
9 Q. And when you set out on this process, that
10 you and Mr. Downing agreed on the number of training
11 sites throughout the system?
12 A. I think Mr. Downing had an idea of how many
13 that would be, but it was certainly tailored by our
14 experience of how confusing it was or how simple it
15 was, that we found that the areas of -- for example,
16 of dense cattail were sufficiently easy to recognize,
17 that we quit looking for dense areas of just cattail
18 and started looking for mixes of cattail and sawgrass
19 and so on that we knew might be more confusing -- if
20 it was confusing to us, might be more confusing to
21 the process.
22 Q. Were the non-confusing dense monotypical,
23 if you will, stands of cattails more prevalent in one
24 area of the ecosystem that you analyzed than in
25 others?
347
1 A. Well, the northern -- where we discussed in
2 the northern part of 2A, there are extensive areas of
3 monotypical cattail. But we found them elsewhere,
4 too. That shows up on the aerial photography pretty
5 much the way we saw them of circular or somewhat
6 circular patches that were monotypical.
7 Q. And, in fact, how many training sites did
8 you use?
9 A. I don't have the actual number. If you add
10 up, H, R, E --
11 Q. Can you give me a ballpark figure?
12 A. 70.
13 Q. Do you know their distribution by class in
14 the earlier -- larger field of classes that was
15 originally used in '93?
16 A. No, I would be guessing on those numbers.
17 Q. What did you do to insure that you had a
18 statistically significant number of training sites
19 for each class that you were going to employ and
20 ultimately employed?
21 A. The statistics of it wasn't my
22 responsibility.
23 Q. That's Mr. Downing's?
24 A. Yes.
25 Q. Did you use all of the training sites in
348
1 developing the 1993 map that was made an exhibit last
2 week?
3 A. I assumed that we did, but I don't know.
4 Q. And I know we talked about this for the
5 1985 version, but for the 1993 were any of the
6 accuracy assessment sites co-located or coincident
7 geographically with training sites?
8 A. Some were close, but there weren't any
9 direct overlaps.
10 I do recall when the 200 number was printed
11 out that there were a couple of the higher numbers
12 that did happen to overlap our training sites, but
13 the number that we selected, that went up to -- I
14 forget. I saw a 99 -- 89 on this map. I think it
15 was 100 accuracy assessment points. None of those
16 fell on a training site. That's my recollection.
17 Q. In Everglades National Park you had 10
18 sites that you had requested to visit. And as I
19 understand it, you visited like five of them. Those
20 were assessment sites or were they training sites?
21 A. Training sites.
22 Q. Did you have any assessment sites in
23 Everglades National Park?
24 A. No. We didn't do any assessment sites.
25 Assessment sites were done totally on aerial
349
1 photography.
2 Q. I'm not asking if you visited -- did you
3 have any there?
4 A. No.
5 Q. Do you know how many you had in WCA-3A
6 assessment sites?
7 A. No. I don't recall. A lot smaller than we
8 had in 2A.
9 Q. How many did you have in 2A?
10 A. The lion's share of them. I don't know the
11 actual number.
12 Q. It's your view or your understanding that
13 the lion's share being in 2A was the result of the
14 randomness of the generation program that selected
15 them?
16 A. No, we are talking about --
17 Q. Assessment?
18 A. We are talking about training sites.
19 Q. I'm talking about assessment.
20 A. There were no assessment sites anywhere
21 except in 2A.
22 Q. That answers the next question, which is
23 why there is a legend on one of these maps that says
24 no accuracy assessment sites for 2A.
25 A. That is right.
350
1 Q. Why did you not do an accuracy assessment
2 for the rest of this work?
3 A. Mr. Downing will have to answer for that.
4 I did the vegetation identification. We focused
5 accuracy checks on 2A.
6 Q. So to the trier of fact and for the moment
7 me, how do you express your confidence in the
8 materials depicted on the April 4, 1994 150,000 scale
9 map that's "Supervised Classification of the Thematic
10 MMI" for December 10, 1993 for the areas exclusive to
11 WCA-2A?
12 A. I don't have any reason to believe that the
13 accuracy would be greatly different elsewhere than it
14 would be in 2A. But, in fact, the only place we
15 quantified it is 2A, so the percentage of accuracy
16 that Mr. Downing will bring to you -- that I reported
17 was somewhere in excess of 70% -- applies to 2A. And
18 if it's any different elsewhere, I don't know. I
19 would be surprised if it was greatly different
20 elsewhere.
21 Q. That's based on just professional judgment
22 because you can't quantify it through statistical
23 evaluation or testing, or you didn't?
24 A. I did not. That's based on the fact that
25 when I fly around in the helicopter, I can recognize
351
1 cattail as easily in Everglades National Park as I
2 can in 2A. I'm under the assumption, then, that the
3 satellite, being a be