0001 1 STATE OF FLORIDA 2 DIVISION OF ADMINISTRATIVE HEARINGS 3 ---------------------------------------------------------- 4 SUGAR CANE GROWERS COOPERATIVE OF ) 5 FLORIDA, et al., ) 6 and ) Nos. 92-3038 7 FLORIDA SUGAR CANE LEAGUE, INC; et al., ) 92-3039 8 and ) 92-3040 9 FLORIDA FRUIT AND VEGETABLE ASSOCIATION,) 10 LEWIS POPE FARMS, et al., ) 11 Petitioners, ) 12 vs. ) 13 SOUTH FLORIDA WATER MANAGEMENT DISTRICT,) 14 Respondent, ) 15 and ) 16 MICCOSUKEE TRIBE OF INDIANS OF FLORIDA, ) 17 the UNITED STATES OF AMERICA, et al., ) 18 Intervenors. ) 19 ---------------------------------------------------------- 20 Deposition Upon Oral Examination Of 21 STEVEN P. MILLARD 22 Volume 1, Pages 1 - 102 23 Taken at 800 Fifth Avenue, Suite 3600, Seattle, WA 24 DATE: March 9, 1993 25 REPORTED BY: Joanne Leatiota, RPR CSR LEATIJL477Q5 LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0002 1 APPEARANCES: 2 For the United THOMAS A.W. FITZGERALD, ESQ. 3 States: Assistant United States Attorney 4 155 South Miami Avenue 5 Miami, Florida 33130 6 For the Florida ROBERT H. BLANK, ESQ. 7 Sugar Cane League: Peeples, Earl & Blank 8 One Biscayne Tower, Suite 3636 9 Two South Biscayne Boulevard 10 Miami, Florida 33131 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0003 1 DEPOSITION OF STEVEN P. MILLARD, Vol.1; taken 3-9-93 2 E X H I B I T S 3 NO. DESCRIPTION PAGE 4 1 Notice of Taking Deposition- Duces Tecum 8 5 2 Resume 22 6 3 Designation of S. Millard as expert witness 35 7 4 Surface Water Improvement and Management Plan 57 8 for The Everglades 9 5 Computer source code 70 10 6 Evergrades SWIM Plan, Appendix E 75 11 7 Table of CA1 values 82 12 8 An Analysis of Derived Phophorus Limits for 88 13 Loxahatchee National Wildlife Refuse 14 15 E X A M I N A T I O N 16 BY PAGES 17 MR. FITZGERALD 3 - 100 18 19 20 21 22 23 24 25 LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0004 1 Seattle, Washington; Tuesday, March 9, 1993 2 2:00 P.M. 3 -------------------------- 4 STEVEN P. MILLARD, witness herein, having been 5 duly sworn by the Notary, 6 testified as follows: 7 E X A M I N A T I O N 8 BY MR. FITZGERALD: 9 Q. Doctor, I am Assistant U.S. Attorney Tom 10 Fitzgerald representing the United States in this 11 matter which is respondent intervenor in the case at 12 hand. 13 Can you for the record state your name, spell 14 your last name and give us your home address. 15 A. My name is Steven Paul Millard, 16 M-i-l-l-a-r-d, and my address is 7723 - 44th Avenue 17 Northeast, Seattle, Washington 98115. 18 Q. Doctor, have you ever been deposed before? 19 A. No, I have not. 20 Q. Have you ever testified as an expert witness 21 in any litigation, administrative, civil or criminal? 22 A. No, I have not. 23 Q. Have you ever testified? 24 A. No, I have not. 25 Q. As they say, there is a first time for LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0005 STEVEN P. MILLARD, V.1, 3-9-93 1 everything. Let me just run through a few sort of 2 common ground rules just for your benefit. If at any 3 time during the deposition you want to take a break or 4 you need to stretch your legs or whatever, just tell me 5 and we'll accommodate that. If I ask a question that 6 doesn't make any sense, which is a very high 7 probability, ask me to clarify it, or if you need to 8 assume something to answer a question, please tell me 9 what your assumptions are. Otherwise the general rule 10 is try not to speculate unless you tell me you have to 11 speculate to answer and then we'll try to sort the 12 question out in a way that makes more sense if we can. 13 A. Okay. 14 Q. Based on that, we'll move on ahead. 15 MR. FITZGERALD: Counsel, do you want to 16 reserve all objections other than to the usual, to 17 expedite matters, objections to form I'd like to hear 18 them so I can try and resolve them. 19 MR. BLANK: That will be fine. 20 MR. FITZGERALD: That's been pretty much the 21 practice since we began this. 22 Q. Doctor, have you done any statistical 23 analysis of oligotrophic wetlands prior to becoming 24 involved in this case? 25 A. No. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0006 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Have you done any work in aquatic 2 environments with the exception of Love Canal which may 3 or may not be classified as aquatic environment? 4 A. Yes, I did work on data that came from a 5 monitoring program in Lake Michigan and that work -- 6 that work on that data was incorporated into my 7 dissertation and also a publication that's listed on my 8 C.V. 9 Q. What was the nature of the work you did on 10 the Lake Michigan monitoring plan? 11 A. The data came from a monitoring program that 12 was carried out to determine whether effluent from 13 power stations were affecting nearby flora and fauna, 14 and as well as I can recall, I looked at that data 15 to -- I was working on a particular problem where I was 16 investigating how temporal and/or spatial correlation 17 could affect the results of commonly used statistical 18 tests that are used to determine whether an impact was 19 in fact being realized or not. 20 Q. Who was directing that study? 21 A. That was part of a study that was funded by 22 EPRI, Electrical Power Research Institute, and I was on 23 that study as a research assistant. The PI in that 24 study, the principal investigator, was a gentleman 25 named Brian Marr who is or was part of the civil LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0007 STEVEN P. MILLARD, V.1, 3-9-93 1 engineering department at the University of 2 Washington. I believe he is still there. He was the 3 principal investigator on the project. Another 4 professor on the project named Dennis Lettenmaier is 5 actually who I worked closely with. So I worked a 6 little bit with Brian Marr, but mostly I worked with 7 Dennis Lettenmaier on that project. 8 Q. The project itself, what parameters were 9 being examined in the monitoring plan to determine if 10 the power station effluent was affecting the local 11 flora and fauna? 12 A. As I recall, the only variable that I looked 13 at was counts of zoa plankton. 14 Q. Was the program also scoped to test for the 15 chemical constituents of the effluent? 16 A. The actual monitoring program that produced 17 the zoa plankton data may well have also looked at 18 chemical concentrations. I don't recall whether it did 19 or not. As best as I can remember, I only looked at 20 the zoa plankton data in my personal analysis. 21 Q. Did you design the monitoring plan? 22 A. No, the monitoring plan had already been 23 carried out, I believe, a couple years or several years 24 before and so I was just using that data to -- I was 25 analyzing it after the monitoring plan had already been LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0008 STEVEN P. MILLARD, V.1, 3-9-93 1 carried out. 2 MR. FITZGERALD: Can we have that marked 3 Exhibit-1 for the deposition. 4 (Exhibit-1 marked.) 5 Q. I am going to show you what's been marked 6 Exhibit-1 for the deposition. That's a copy of a 7 notice of taking deposition duces tecum, and I ask if 8 you would take a moment and look at it and tell me if 9 you have seen that one before. Not that particular one 10 but similar. 11 A. Yes. My counsel Mr. Blank faxed me a copy of 12 this last week, sometime last week. 13 Q. Have you had an opportunity to review the 14 notice and the included listings of documents to be 15 provided with counsel prior to today? 16 A. I have. 17 Q. Have you had full opportunity to discuss what 18 was required by that notice? 19 A. Yes, I discussed it briefly with my counsel. 20 Q. In locating documents that would be 21 responsive to the request for production of documents 22 which begins at page 5 and the particular documents to 23 be produced beginning on page 6 through 8, how did you 24 go about ascertaining what documents were responsive 25 and actually producing those documents? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0009 STEVEN P. MILLARD, V.1, 3-9-93 1 A. I actually left that up to my counsel. 2 Q. Did you conduct any search of your personal 3 files and records to -- 4 A. Well, my counsel requested me for certain 5 reports and data on floppy disk which I then supplied 6 to my counsel. 7 Q. What data on floppy disks did you supply to 8 your counsel? 9 A. He specifically requested the data that I 10 used to produce the report entitled -- this is a long 11 one -- "An Analysis of Derived Phosphorus Limits for 12 Loxahatchee National Wildlife Refuge." 13 Q. When was that produced? When did you provide 14 it to counsel? 15 A. The report or the floppy disks? 16 Q. The floppy disks. 17 A. The floppy disks I provided to counsel I sent 18 via Federal Express last week either I believe it was 19 Wednesday, on Wednesday which would have been what -- 20 today is the 6th -- sorry, today is the 9th, so that 21 would have been -- 22 Q. The 3rd? 23 A. The 3rd if that was last Wednesday. 24 Q. How many floppy disks were there? 25 A. One. One floppy disk. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0010 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Were there any files on that other than data 2 files? 3 A. No, no. 4 Q. In addition to the one data disk and of 5 course the report you have just referenced itself and 6 the bootstrap code document -- 7 A. Right. I also sent him -- I believe in a 8 separate shipment earlier I sent him the code, the hard 9 copy of the code. 10 Q. And then your report. Did you have any other 11 documents other than those three? And document as 12 defined includes storage in whatever form, any magnetic 13 media, computer media, et cetera would be included in 14 the term "document." Did you have any other records or 15 documents that were responsive to the 12 categories of? 16 A. I don't believe so. 17 Q. Why don't we go through and we can be certain 18 of that. Your curriculum vitae you did provide and 19 that will be the next exhibit, so paragraph 1 is taken 20 care of I believe. The second paragraph calls for a 21 list of your technical, professional, or scientific 22 publications, reports, et cetera in which you are 23 identified as author or coauthor but it is limited or 24 related to Everglades research activities, water 25 quality trends, et cetera. Do you have any such LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0011 STEVEN P. MILLARD, V.1, 3-9-93 1 documents? 2 A. This is the only one. 3 Q. Just the single report. Paragraph 3 is a 4 copy of each technical, professional, or scientific 5 publication, report, article, et cetera in which you 6 are identified as an author or coauthor related to 7 analysis of Everglades related issues. Do you have 8 such documents other than the report you have just 9 mentioned? 10 A. No. 11 MR. BLANK: Again just for the record, the 12 report he is mentioning is the one that he's previously 13 identified and was tendered to counsel. 14 A. Right, right. 15 Q. This report draft dated January 8, 1993, 16 correct? 17 A. Yes. 18 Q. Other than that draft report, is there a 19 final report on the same subject? Have you finalized 20 that as yet? 21 A. No. 22 Q. Did you submit that report for referee or 23 review by any other consultant expert or colleague, for 24 example, to provide you comments on it? 25 A. Yes. I submitted the report to Dennis LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0012 STEVEN P. MILLARD, V.1, 3-9-93 1 Lettenmaier, and he reviewed the report and made a few 2 changes to wording in the report. 3 Q. Do you have the prior version or the unedited 4 copy and the reviewer comments made by Dr. 5 Lettenmaier? 6 A. I don't. He didn't make specific comments on 7 the report. I gave him an electronic copy of my 8 version and then he incorporated that into his word 9 processing software, changed a few sentences and then 10 this is the result of that. 11 Q. Do you still have a copy of the unedited 12 version on your own word processing system? 13 A. I would have -- yes, I would have an 14 electronic copy of that I believe. 15 MR. FITZGERALD: Counsel, the unedited 16 version of the witness would be responsive to paragraph 17 3 and I would request that it be produced. 18 Q. I don't know if you will get by the office 19 between now and tomorrow if you have it at home, but if 20 not you can produce it through counsel, I'd appreciate 21 it. 22 A. Yes. 23 Q. How about all documents, paragraph 4, all 24 documents relating to analyses, statistical or 25 otherwise done in relation to the Florida Everglades? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0013 STEVEN P. MILLARD, V.1, 3-9-93 1 Again I assume there is nothing but the January draft 2 report. 3 A. That's correct. 4 Q. All data whether used or not, collected as 5 part of any analysis of issues affecting the Everglades 6 Protection Area and/or Everglades Agricultural Area, 7 including published data, mean data and raw data. Do 8 you have any such material that you are aware of? 9 A. None other than the data that I supplied to 10 counsel on the floppy disk that I assume was also 11 supplied to you. 12 MR. FITZGERALD: Counsel, the floppy disk 13 data in the electronic storage form was not provided. 14 I would ask that a copy of that be provided. 15 MR. BLANK: That's fine. We just gave you a 16 printout of the disk material itself. 17 MR. FITZGERALD: I am sure the doctor will 18 tell you that that is of marginal utility in a case 19 like this. The practice has been to provide disks on 20 all that, because first, it takes a lot less room, and 21 secondly, nobody can manipulate all the data. I know 22 Dr. Davis does not want to see everything from Dr. 23 Walker's work on hard copy. 24 Q. Does that data disk include your actual 25 bootstrap runs with your out file of a thousand runs? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0014 STEVEN P. MILLARD, V.1, 3-9-93 1 A. No, it does not. 2 Q. Where is that? 3 A. That is on my computer at home and it's in -- 4 it's stored in a format that the software that I use 5 understands, so it's in a certain binary form. 6 Q. That's the software program that the firm you 7 consult with -- 8 A. That I work for, yes. S-PLUS is the name of 9 the software package. 10 Q. We are down to paragraph 6, all abstracts and 11 materials used in scientific or public presentations 12 including drafts, submitted materials, slides, overhead 13 transparencies or presentation materials. Have you 14 prepared any such materials to your knowledge? 15 A. No. 16 Q. That would relate to this matter, to the 17 Everglades matter. 18 A. Right. 19 Q. Obviously you have told us you probably have 20 that sort of thing in other areas. Have you ever done 21 any public presentations on bootstrap? 22 A. I may have talked about it briefly when I 23 teach a course instructing people how to use S-PLUS, a 24 question may have come up during the course on how you 25 would do bootstrapping using S-PLUS. But no formally LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0015 STEVEN P. MILLARD, V.1, 3-9-93 1 prepared course on bootstrapping, no. 2 Q. Have you made any presentations on your work 3 to other consultants or experts in this case? 4 A. On the work that's in this report? 5 Q. Yes. 6 A. The only other consultant that's looked at 7 this report that I am aware of is Dr. Lettenmaier. 8 Q. Has Dr. Lettenmaier been coordinating your 9 work on behalf of the sugar cane legal counsel for this 10 case? 11 A. He was the person who contacted me 12 originally, so -- and yes, you could say that he's been 13 coordinating my work. 14 Q. Other than meetings with counsel to prepare 15 for your deposition, have you met with anyone else 16 other than Dr. Lettenmaier in connection with your work 17 on this case? 18 A. No. 19 Q. In preparing for this deposition, were you 20 provided the depositions of any other prior witnesses 21 or any other materials other than the data files from 22 Environmental Services? 23 A. Yes, I was. 24 Q. What other materials were you provided? 25 A. I was provided the deposition of Dr. McClave LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0016 STEVEN P. MILLARD, V.1, 3-9-93 1 that I believe was taken last November or December, and 2 I was provided -- so I was provided the deposition and 3 all the exhibits in that deposition. And I have also 4 been provided with a copy of the Florida -- the SWIM 5 Plan, but the only appendix that I have been provided 6 with is Appendix E. 7 Q. The SWIM Plan which is a matter of record in 8 this proceeding, its appendices to the League's second 9 amended complaint is comprised of three separate 10 volumes. Do you recall which volumes you were provided 11 or is it your testimony you were only provided the 12 Appendix E excerpt? 13 A. No, I was provided -- I don't -- I am not 14 sure of the title of that document in relation to the 15 three volumes. 16 Q. The planning document? 17 A. Yes, the planning document is what I was 18 provided with as well as Appendix E. 19 Q. What did you review in the planning document 20 to prepare for your deposition? 21 A. I looked through pages -- pages 102 through 22 something like 105 or 106 which specifically talk about 23 phosphorus limits for the Loxahatchee National Wildlife 24 Refuge, and I briefly glanced at some maps in the 25 planning document and read the introduction. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0017 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Other than that portion of the SWIM Plan 2 planning document and Appendix E which is in volume 3 3 of the SWIM Plan, did you see any other portions of the 4 SWIM Plan or review it? 5 A. No. 6 Q. I assumed in an earlier question that you had 7 gotten that and reviewed it in preparation for your 8 deposition, but it suddenly occurred to me that may not 9 be a valid assumption. When in fact did you receive 10 those materials related to the SWIM Plan? 11 A. Last week. I reviewed it in preparation for 12 the deposition. 13 Q. So you did not review and have available to 14 you any of that material prior to your conduct of your 15 bootstrap analysis reflected in your January 8th 16 report? 17 A. That's correct. The only document that I 18 have reviewed is what I mention in the report, the 19 report by Dr. Walker that's dated May of 1991. 20 Q. Were there any additional materials provided 21 to you, then, other than the SWIM Plan materials and 22 Dr. McClave's deposition and exhibits to help prepare 23 for the deposition? 24 A. Not that I can recall now. I believe that 25 covers it. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0018 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Have you ever made any presentations on 2 Everglades related issues of any sort whatsoever? 3 A. No. 4 Q. Do you have any notes, notebooks, hard copies 5 of materials on computer disks which include 6 unpublished analyses related to Everglades research? 7 A. Yes, I do because I am involved in ongoing 8 analyses. 9 Q. What materials are you currently in 10 possession of which you utilize of your ongoing work? 11 A. I have obtained from Dr. Lettenmaier data on 12 phosphorus concentration and flow for the structure 13 stations S-5A and S-6. 14 Q. What form is that data in? 15 A. I received it from Dr. Lettenmaier on a 16 floppy disk, so it's -- I have a copy on a floppy disk 17 and then also on my computer. 18 Q. Do the data files contained on the floppy 19 disk from Dr. Lettenmaier identify the source of that 20 data? 21 A. No, they do not. 22 Q. Do you know the source of the data? 23 A. I believe Dr. Lettenmaier received it from 24 Environmental Services -- I believe the name of the 25 firm is Environmental Sciences & Permitting, ES&P. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0019 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Do you have any other data in hand that you 2 are or will be using for ongoing work? 3 A. Actually probably the data that I receive 4 from Dr. Lettenmaier may not only include data for S-5A 5 and S-6 but it may include data from the other 6 structures. I can't remember their other names, I 7 believe there is an S-13 and an S-16 and a few others, 8 but then that would be the bulk -- that would be all of 9 the material that I have to this point. 10 Q. The S-5A and the S-6 discharge surface waters 11 into Loxahatchee National Wildlife Refuge, and 12 hereafter I will refer to the refuge if I can kind of 13 lay that as a ground rule. The other structures you 14 may be referring to do not. Are you analyzing surface 15 flows or inflows to any other portion of the Everglades 16 Protection Area? 17 A. I am not using the data from those stations. 18 Dennis just happened to give me the data for those 19 stations all in one packet. 20 Q. Do you know what period of record the data 21 encompasses for 5A and 6? 22 A. I don't recall the exact dates. What I was 23 interested in when I -- as I am doing this continuing 24 work is actually data that overlaps the sampling dates 25 for the data that's available for the refuge now. So LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0020 STEVEN P. MILLARD, V.1, 3-9-93 1 in other words, I don't recall the starting and ending 2 dates for the S-5A and S-6 data, but I have been in the 3 process of condensing that data so that it only covers, 4 as I recall, the years 1978 through 1983, whatever 5 years. 6 Q. So more or less to match the five-year period 7 that is rationalized to baseline period of the one year 8 by Dr. Walker? 9 A. That's correct. 10 Q. So you are going to conduct further analysis, 11 considering additional data sources for the analysis 12 done by Dr. Walker in that regard? 13 A. That's correct. 14 Q. Have you ever made any proposals for research 15 or analysis on Everglades related issues? 16 A. No. 17 Q. Do you have any documents that you relied 18 upon in preparing, formulating, developing, authoring, 19 co-authoring, reviewing and/or organizing your 20 anticipated expert testimony relating to the subject 21 matter of this action, yours or anyone else's, that is 22 not amongst the material that you have provided? 23 A. No. 24 Q. Paragraph 11 recites a portion of your 25 designation as an expert witness and asks for any LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0021 STEVEN P. MILLARD, V.1, 3-9-93 1 materials or documents that you would have that you 2 employ related to phosphorus trend models, statistics, 3 STA modeling, phosphorus concentration limits, water 4 quantity and quality trends, QA/QC database. Do you 5 have any materials related to any of those areas which 6 you have not provided? 7 A. No. 8 Q. Finally, have you done any notes, commentary, 9 critique or have any documents of that sort that would 10 relate to your review of the portions of the SWIM Plan 11 or prior drafts of the SWIM Plan you have mentioned? 12 A. Nothing other than what's in the report. 13 Q. But you have testified, did you not, that you 14 did not have the SWIM Plan at all when you prepared 15 your report? 16 A. Sorry. That's correct, I am sorry. Point 12 17 only refers to the SWIM Plan, right. 18 Q. So you have no notes or prior drafts -- well, 19 you do have the prior draft and I asked you for that. 20 Your draft report recites to a document by Dr. William 21 Walker which you have analyzed in your report, and you 22 have identified that as a May '91 document. When did 23 you receive that? 24 A. I was initially approached by Dr. Lettenmaier 25 to work on this project. He called me up at the end of LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0022 STEVEN P. MILLARD, V.1, 3-9-93 1 September and so I would have met with him sometime in 2 early October, at which point he gave that document to 3 me, Dr. Walker's report to me. 4 Q. Where did you meet with Dr. Lettenmaier? 5 A. At his office at the University of 6 Washington. 7 Q. Do you know if the document you were 8 reviewing by Dr. Walker is the finalized report 9 prepared by Dr. Walker on that subject? 10 A. I do not. 11 MR. FITZGERALD: Mark this as Exhibit-2. 12 This is a copy of your resume. 13 (Exhibit-2 marked.) 14 Q. If I can invite your attention to Exhibit-2, 15 I guess my first question is, is this up-to-date as 16 best you know? I know people -- I tend to let things 17 like this slide until I have to change it so it's 18 always good to ask. 19 A. Everything is up-to-date except for my 20 current title at Statistical Sciences. I have relieved 21 myself of the burden of administration, so my title is 22 now senior training instructor and the job of 23 administration has been passed off to someone else. 24 Q. What is Statistical Sciences, Inc.? 25 A. Statistical Sciences, Incorporated is a LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0023 STEVEN P. MILLARD, V.1, 3-9-93 1 company that distributes and markets a couple different 2 software packages. The main package that we sell is 3 called S-PLUS. We also sell a product called Data 4 Junction. S-PLUS is a statistical graphical software 5 package for analyzing data. Data Junction is a 6 software package for allowing you to convert data from 7 one format to another, from spreadsheets to database to 8 statistical package; it handles a number of different 9 formats. 10 Q. Who actually developed the two software 11 packages? 12 A. S-PLUS is built on top of a software package 13 called S which was developed and is maintained at Bell 14 Labs, and we -- Statistical Sciences is licensed by 15 Bell Labs to take their software and build on top of it 16 and market and distribute it. So S-PLUS is a superset 17 of a package called S. 18 Q. How long have you been with Statistical 19 Sciences? 20 A. Since 1990. Originally I came on in the 21 spring of 1990 as a consultant to them and I was 22 formally hired in the fall of 1990. 23 Q. What is the current nature of your duties 24 with them? 25 A. I provide training courses in the software, I LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0024 STEVEN P. MILLARD, V.1, 3-9-93 1 teach people how to use the software, I develop 2 material for new training courses, maintain the 3 material for the current training courses, sometimes I 4 am called in to write documentation that's shipped with 5 the product. 6 Q. How did you happen to become acquainted with 7 SSI and go to work for them? 8 A. The gentleman who founded the company, 9 Professor Doug Martin, was one of my professors when I 10 was in graduate school. He's a professor at the 11 University of Washington in the department of 12 statistics. 13 Q. Prior to being employed by SSI you were an 14 assistant professor at St. Martin's College? 15 A. That's correct. 16 Q. What type of courses were you teaching there? 17 A. I was in the math department so I taught a 18 range of courses in math: Calculus, statistics, 19 introductory probability, introductory statistics, 20 differential equations. 21 Q. All my favorites. All my favorites. And I 22 am sure they still are with the undergrads. Did you do 23 any consulting work as well as teaching during the 24 period of '89 and '90 when you were an assistant 25 professor? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0025 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Yes. I did consulting -- I started a 2 consulting project for a professor at the University of 3 Washington who's in the obstetrics and gynecology 4 department, and I can't recall the year that that 5 project first started up. I believe it was 1987, and I 6 did consulting for him from '87 through '88 and I 7 believe through '89 as well. 8 Q. Hence your articles in the "Obstetrics and 9 Gynecology" publication? 10 A. Correct. 11 Q. Prior to taking the position at St. Martin's 12 College you were with CH2M Hill as a statistician? 13 A. That's correct. 14 Q. Which of their offices were you employed in? 15 A. They have an office in Bellevue, Washington. 16 Q. During that period did you conduct or work on 17 any projects with their employees in their Florida 18 office? 19 A. There was one project that I was involved in 20 that was based in one of their Florida offices -- I 21 don't know if they have more than one, I don't recall 22 which office it was at this time -- that had to do 23 with -- sorry, the fundamentals of it escape me at the 24 moment. It had to do with trying to determine where 25 children who were attending school where exactly they LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0026 STEVEN P. MILLARD, V.1, 3-9-93 1 lived, the housing tracts that they lived in, and I 2 think originally they were going to send out a survey 3 and we suggested that instead they look at databases 4 that were available both on where the children lived 5 and then a database from, say, the assessor's office 6 and try and match those two databases. 7 Q. Do you recall who from the Florida office you 8 worked with if not the location of the office? 9 A. The name of the person that I worked with 10 most closely, I believe her name was Becky Pickren, and 11 I don't know how you spell the last name. I would 12 guess P-i-c-k-r-e-n, perhaps P-i-c-k-r-i-n. 13 Q. How long were you with CH2M Hill? 14 A. I started there in the fall -- let's see if I 15 can remember -- the fall of 1987 and stayed through the 16 fall of 1988. 17 Q. Prior to that you were with the 18 biostatistical consulting unit, department of 19 biostatistics at the University of Washington for about 20 two years? 21 A. Actually just one year. 22 Q. I see the dates. The reason I am asking is 23 you can't really tell what that means. 24 A. Right, right. 25 Q. That's the period during which you were LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0027 STEVEN P. MILLARD, V.1, 3-9-93 1 working on the issues that related to your publications 2 in maternal hemodynamics? 3 A. Yes. The public -- the research that I did 4 that resulted in the publications in the "Obstetrics 5 and Gynecology" journals, that happened starting at the 6 end of '87. 7 Q. When were you a consultant for NeoRx 8 Corporation? 9 A. NeoRx. I believe I started working for them 10 in the fall of '87 and worked with them for a period of 11 approximately four months, maybe six months. 12 Q. For a period of time, 1985 to '86, your 13 resume indicates that you were a statistician with the 14 U.S. Geological Survey in the water resources division 15 in Sacramento. What were you doing for USGS? 16 A. As I recall, most of my work was with a soil 17 scientist there named Steve Deverel, and I worked with 18 him analyzing data on selenium concentrations in 19 run-off water from agricultural land. As I recall, the 20 Kesterson National Wildlife Refuge was where the flora 21 and fauna were being impacted because agricultural 22 run-off water contained very large amounts of an 23 element called selenium, and when that agricultural 24 water reached the refuge, it started wreaking havoc 25 with the wildlife. I believe the waterfowl, their eggs LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0028 STEVEN P. MILLARD, V.1, 3-9-93 1 that they laid, the shells were not hard. I can't 2 remember the exact details of how it impacted the 3 wildlife, but that was an issue, and so the U.S. 4 Geological Survey, this Sacramento office, was somehow 5 involved in a study of that. 6 Q. What did you actually do in the study? 7 A. I recall analyzing one main data set for 8 Steve Deverel, and it had to do with concentrations of 9 certain nutrients in two different geographic regions. 10 Q. Do you recall what nutrients were the subject 11 of that analysis? 12 A. At the current moment I don't. 13 Q. Nitrogen, phosphorus? 14 A. I really don't recall the specifics at this 15 point. 16 Q. Did any publications result from that work? 17 A. Yes. There was -- there were two 18 publications, and in fact I believe both of them are 19 listed on my C.V. 20 Q. This is S.J. Deverel and Millard, 21 "Distribution and Mobility of Selenium and Other Trace 22 Elements in Shallow Ground Water of the Western San 23 Joaquin Valley, California"? 24 A. That's correct. That's one of the 25 publications, and the other publication that resulted LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0029 STEVEN P. MILLARD, V.1, 3-9-93 1 from that work is the one that's listed right up above 2 that. 3 Q. "Nonparametric Statistical Methods"? 4 A. That's correct. 5 Q. When you were working for your doctorate, 6 what was your chosen area for your thesis? 7 A. It grew out of the work that I was doing on 8 the EPRI project as a research assistant, and there 9 were two main areas of focus in my dissertation. One 10 was the problem of designing monitoring studies where 11 you are trading off how much money you can spend on the 12 sampling program versus how effective your monitoring 13 program will be in detecting change of a certain size. 14 Q. You were under the direction of Dr. 15 Lettenmaier for that? 16 A. That's correct. I was under the direction of 17 Dr. Lettenmaier and actually my senior advisor for my 18 dissertation was Dr. Peter Guttorp who is in the 19 department of statistics at the University of 20 Washington. 21 Q. Who was on your panel? 22 A. Dr. Guttorp was my senior advisor, Dr. 23 Lettenmaier was on my committee, Dr. Gerald Van Bell 24 who is a professor in biostatistics at the University 25 of Washington -- excuse me, he was a professor of LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0030 STEVEN P. MILLARD, V.1, 3-9-93 1 biostatistics at the time. I believe he is now head of 2 the environmental health department. Dr. Loveday 3 Conquest was on my committee, she is in the center for 4 quantitative science at the University of Washington. 5 There is actually a fifth person on the committee under 6 the program that I was studying in. You have to have a 7 representative from somebody in biological sciences, so 8 I believe the person who was on my committee in that 9 regard was a Dr. -- I can't recall his name at the 10 time. He was actually only involved in an earlier part 11 of my examination that involved knowledge of biological 12 field and I am not even sure if he was at my final 13 exam. 14 Q. What biological field had you selected as 15 your specialty? 16 A. It involved -- let me think for a minute. I 17 believe it involved estimating abundance and 18 distribution of certain aquatic organisms. I'd have to 19 go back and dig out my notes. 20 Q. This is not something you have applied every 21 day ever since, I gather. 22 A. No. 23 Q. Have you done any prior consulting work for 24 agricultural industries? 25 A. No. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0031 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Have you done any prior work for the sugar 2 industry? 3 A. No. 4 Q. Or any element thereof? 5 A. No. 6 Q. Do you have a written contract for your work 7 on this case? 8 A. No. It's verbal. I have a verbal contract 9 with the firm that my counsel works for. 10 Q. Did Dr. Lettenmaier arrange that for you? 11 A. Dr. Lettenmaier was my original contact into 12 this project. The verbal contract has been discussed 13 between counsel and myself. 14 Q. Is it a fixed fee or is it time and 15 materials? What's the nature of your contractual 16 relationship? 17 A. It's time and materials at this point. 18 Q. How much time have you expended producing 19 your report and analyzing materials for the case? 20 A. I believe I spent approximately 40 hours 21 preparing this report. I'd have to go back and look at 22 my bill. 23 Q. You said at this time it's time and 24 materials. Do you anticipate or are you in the process 25 of negotiating a different arrangement? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0032 STEVEN P. MILLARD, V.1, 3-9-93 1 A. No. 2 Q. Do you anticipate that arrangement will 3 prevail absent sudden change? 4 A. I have no idea. 5 Q. At what rate do you bill out your time? 6 A. $70 an hour. 7 Q. I am not going to insult them by asking if 8 they're current on their payments. That's your 9 problem. 10 (Discussion off the record.) 11 Q. How is the scope of the work you were going 12 to perform under your verbal arrangement determined? 13 A. How is it currently determined? 14 Q. How was it determined at the outset? 15 A. Through Dr. Lettenmaier. 16 Q. What did he tell you is the nature of the 17 task to be performed by you? 18 A. Initially he gave me a copy of Dr. Walker's 19 report, the one dated May '91, and asked me to review 20 the report and then come up with some sort of way of 21 analyzing how good the estimates are in that report. 22 Q. The estimates of limits and target 23 concentrations? 24 A. Yes, yes. 25 Q. How did you proceed with that task? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0033 STEVEN P. MILLARD, V.1, 3-9-93 1 A. I was interested in assessing the adequacy of 2 the model from the start and also whether there were 3 other kinds of data available, but Dr. Lettenmaier at 4 that point said well, for the time being let's assume 5 that this is the model that we're going to use and this 6 is the data that we have, and based on these givens is 7 there some way that we can assess the variability in 8 those limits that were determined in that report, and 9 so he actually was the one who initially suggested 10 doing a bootstrap study, and so I followed up on that 11 suggestion and implemented the approach that's reported 12 in my report. 13 Q. When Dr. Lettenmaier first suggested the 14 bootstrap method to analyze the variability of the 15 limits, did you agree that that was the most 16 appropriate mechanism? 17 A. At that point I had not reviewed the report 18 in detail, so I couldn't say. But once I reviewed the 19 report, then I agreed with Dr. Lettenmaier that that 20 would be a simple way and an expedient way to get at 21 the variability in the limits. 22 Q. What other mechanisms might you have chosen? 23 A. Well, as indicated in my report, if you're 24 willing to make certain assumptions about the 25 distribution of the data or some transformation of the LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0034 STEVEN P. MILLARD, V.1, 3-9-93 1 data, then you could analytically derive the 2 distribution of the limits. You could mathematically 3 derive that. 4 Q. You have to assume a Gaussian distribution to 5 do that? 6 A. That's correct, that's correct. 7 Q. Was there any reason why you felt it was 8 inappropriate to assume a Gaussian distribution of the 9 data sets? 10 A. I didn't necessarily feel it was 11 inappropriate, but I knew that we could get an answer 12 that was reliable without having to make the assumption 13 of Gaussian distribution. 14 Q. Bootstrap does not depend on such an 15 assumption? 16 A. That's correct. 17 Q. After the initial definition and scoping by 18 Dr. Lettenmaier, did your understanding of the task 19 alter as you got into the process such that you changed 20 the scope or altered your approach in any way? 21 A. No. 22 Q. Are you familiar with the designation of 23 witness filed in this case identifying you as an expert 24 witness for the Sugar Cane League and its allied 25 parties? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0035 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Yes, I am. 2 MR. FITZGERALD: Can we mark that Exhibit-3. 3 (Exhibit-3 marked.) 4 Q. Have you seen what's marked as Exhibit-3 5 prior to me handing you the document today? 6 A. Yes, I have. 7 Q. When did you last see it? When did you first 8 see it? 9 A. This morning. 10 Q. All the above. So you did have an 11 opportunity to review this document as well in 12 preparation for your deposition? 13 A. Yes. 14 Q. Had you ever consulted with anybody regarding 15 the nature of your designation prior to October of 16 1992? 17 A. I don't understand the question. 18 Q. Well, I'd represent to you for the purposes 19 of this proceeding that this document naming you was 20 filed in October of 1992 in the administrative action 21 in Florida. From your prior testimony I understood 22 that you were approached in late September of '92 by 23 Dr. Lettenmaier and met with him here in Washington in 24 early to mid October, and this was filed no later than 25 October 26th. I don't know the precise date, but that LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0036 STEVEN P. MILLARD, V.1, 3-9-93 1 was the deadline. I know it was timely. 2 A. Yes. 3 Q. Have you discussed with Dr. Lettenmaier the 4 subject matter of your expected testimony, the 5 substance of your testimony and the grounds for your 6 opinion at that time? 7 A. Only in very general terms. 8 Q. So not in the detail that appears here? 9 A. Probably not in quite this detail, no. 10 Q. Since your initial meeting with Dr. 11 Lettenmaier in October of '92, have you had occasion to 12 meet with him and discuss the nature of the testimony 13 you might render in this proceeding, the areas on which 14 you might venture opinions? 15 A. Whenever I have met with him we have talked 16 only about the project at hand, the analysis that we're 17 carrying out, and haven't really talked about -- we 18 haven't speculated as to if this were to go to court 19 what might we be talking about. 20 Q. Have you discussed that area with anyone 21 other than Dr. Lettenmaier? Other than counsel this 22 morning presumably. 23 A. No. 24 Q. In the subject matter of expected testimony 25 area, it identifies one potential area of your LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0037 STEVEN P. MILLARD, V.1, 3-9-93 1 testimony as phosphorus trend models. With the 2 exception of the bootstrap operation you carried out 3 and your review of Dr. Walker's model contained in his 4 May '91 document, do you have any other experience with 5 phosphorus trend models? 6 A. Not with phosphorus trend models. I do with 7 trend models. 8 Q. Trend models being a fairly common 9 statistical tool? 10 A. Right. 11 Q. An environmental tool? 12 A. Right. 13 Q. Statistics. I think we'll skip that one. 14 STA modeling. Do you know what an STA is? 15 A. I believe that stands for storm treatment 16 area or something like that. 17 Q. Close enough. Storm water treatment area. 18 Are you familiar with any modeling done on the storm 19 water treatment areas pursuant to the SWIM Plan that's 20 at issue in these proceedings? 21 A. Only in a very general sense. 22 Q. Have you conducted any review of models 23 related to STAs? 24 A. No, not at this time. 25 Q. Do you represent that you are going to LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0038 STEVEN P. MILLARD, V.1, 3-9-93 1 analyze STA modeling efforts? 2 MR. BLANK: Objection. Calls for 3 speculation. 4 You can answer if you know. 5 A. I really don't know. 6 Q. No one has discussed with you -- and I'll ask 7 the question a little more artfully. Has anyone 8 discussed with you their desire or intent that you in 9 the future as part of your work in fact review STA 10 modeling in connection with the SWIM Plan either by 11 petitioners or respondent, respondent intervenors in 12 the case? 13 A. I can't remember. I discussed the SWIM Plan 14 this morning with counsel and I can't remember if that 15 was one of the topics that he suggested may come up in 16 the future. 17 Q. Other than the phosphorus concentration 18 limits and targets that you have reviewed in connection 19 with the refuge and the work that you have identified 20 in your draft document and Dr. Walker's paper, have you 21 done any other analysis of phosphorus concentration 22 limits in connection with this case? 23 A. No. 24 Q. Has it been indicated to you that anyone 25 desires you to do further work in that area? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0039 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Yes. 2 Q. In what area phosphorus concentration limits 3 -- or what geographical area would that be? 4 A. The refuge. 5 Q. Is that in connection with the additional 6 data you have received on S-5A and 6? 7 A. Yes. 8 Q. Beyond the refuge and those two structural 9 contributions to the surface water discharged into the 10 refuge, are you aware of any other areas within the 11 Everglades protection area where you may be requested 12 or have been requested to conduct phosphorus 13 concentration limit analysis? 14 A. I would speculate that I may -- 15 MR. BLANK: Please don't. 16 A. All right. So in lieu of speculating I'd 17 have to say no. 18 Q. So as I understand that answer after the 19 hiatus, it is not your understanding at this time that 20 you will be doing it, but you may should somebody ask 21 you? 22 A. That's correct. 23 Q. What analysis, if any, have you conducted 24 into water quantity issues in this analysis? 25 A. Water quantity meaning data involving flow? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0040 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Flows and hydro period in the Everglades 2 Protection Area or the Everglades Agricultural Area or 3 flow data related to the operation of the central and 4 south Florida flood control project. 5 A. I assisted Dr. Lettenmaier on a project where 6 he was developing rating curve -- what's called in the 7 hydrological literature rating curves, and I don't 8 remember what area it's for. The way I was assisting 9 him was with my expertise of S-PLUS. He needed 10 something done in S-PLUS, and so he explained to me 11 what he needed done in S-PLUS. I didn't really need to 12 know the subject matter in order to do the 13 manipulations that he requested, but the variables did 14 involve flow and concentrations. 15 Q. These are rating curves for structures in the 16 Everglades region? 17 A. I don't recall what the stations were, if 18 they were structures or where they were. 19 Q. Other than providing the technical assistance 20 with the S-PLUS programming, have you yourself 21 conducted any water quantity analysis? 22 A. I have not, no. 23 Q. Other than the incidental stage levels that 24 you would have had to consider? 25 A. No, not at this point. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0041 STEVEN P. MILLARD, V.1, 3-9-93 1 MR. BLANK: Just for clarification purposes, 2 when you refer to incidental stage levels you are 3 referring to the stage levels on the refuge? 4 Q. Stage level statistic data related to the 5 refuge that was referenced in Dr. Walker's paper and 6 referenced in your paper. 7 Is your current understanding of the scope of 8 work which you will be called upon to perform hereafter 9 through the conclusion of the case whenever, if ever, 10 will involve water quantity analysis? 11 A. I am sorry, could you restate the question. 12 Just say it again. You don't have to restate it. 13 Q. Is it your understanding that hereafter you 14 are going to conduct water quantity analysis in this 15 case? 16 A. Yes, I will be involved in looking at flow 17 data, specifically at the structures S-5A and S-6. 18 Q. How about elsewhere in the Everglades region? 19 A. I don't know at this point. 20 Q. Other than the report you have produced 21 related to the targets and limits for the refuge, have 22 you done any other quality trend analysis in connection 23 with this case? 24 A. No. 25 Q. Are you preparing to do any? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0042 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Yes. That would be involved in the analysis 2 that I plan on doing where I am incorporating the data 3 from the S-5A and S-6 structures. 4 Q. When you incorporate that data, do you plan 5 to rerun your bootstrap program using additional data 6 points from the inflow structures? 7 A. It's possible. 8 Q. How will you calibrate that with the stage 9 data that is the variable for the refuge? 10 A. At this point I don't know. I haven't 11 thought about it enough. 12 Q. Have you done any QA/QC database analysis in 13 connection with this case? 14 A. Nothing other than what I noted in my report 15 on the discrepancy on the data I had obtained versus 16 the data that was printed in the Walker report. 17 Q. Other than Dr. Lettenmaier, have you worked 18 with or consulted with any other experts in relation to 19 any of those six areas of expected testimony? 20 A. No. 21 MR. BLANK: Just for clarification purposes, 22 I believe you did have some contact with someone at 23 ESP. 24 A. Sorry. I needed to obtain the date -- when I 25 was reviewing Dr. Walker's report and I was LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0043 STEVEN P. MILLARD, V.1, 3-9-93 1 preparing -- doing the work to prepare this report, I 2 needed to obtain data on the stage values at the three 3 stations that were used in the Walker report. Dr. 4 Lettenmaier -- I am not sure if he had that data or 5 not, so he suggested that I contact Bruce Myhre at ES&P 6 and so Dr. Myhre sent me three files that contained the 7 stage data on the three stations. 8 Q. Was the source of that data identified in the 9 data files? 10 A. I don't believe so. 11 Q. Did the data files encompass more than the 12 five-year period addressed in Dr. Walker's report? 13 A. Yes, they did. 14 Q. In the six areas we have just gone through 15 seriatim related to the subject matter of your expected 16 testimony, have you formed any final opinions? 17 A. No. 18 Q. The designation cites as a substance of 19 expected testimony analysis of district phosphorus 20 trend models. Assuming for the sake of this deposition 21 that that is the substance of expected testimony, what 22 analysis of South Florida Water Management District or 23 Army Corps of Engineers, Jacksonville District, because 24 I can't tell which one it means, what phosphorus trend 25 models have you conducted to date? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0044 STEVEN P. MILLARD, V.1, 3-9-93 1 A. The only one that I have conducted to date 2 where I have created a report is this review of the 3 Walker report. 4 Q. You understand Dr. Walker is not an employee 5 of either of those two districts. 6 A. (Witness nods.) 7 Q. You say you have produced only this report. 8 Have you conducted analysis of Corps of Engineers or 9 water management district phosphorus trend models? 10 A. No, not at this point. 11 Q. What testimony, if any, would you expect to 12 give in the issue of phosphorus concentration limits? 13 MR. BLANK: Again, Doctor, please don't 14 speculate. 15 A. I don't know. 16 Q. Rather than go through the six categories 17 there, would it be fair to say that essentially your 18 answers on those would be as with your answers on the 19 subject matter of expected testimony, that either of 20 those areas have not yet been identified to you as 21 within the scope of work someone wishes you to perform 22 or the precise nature has not been identified to you as 23 yet? 24 A. Correct. 25 Q. The only one I'd like to specifically touch LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0045 STEVEN P. MILLARD, V.1, 3-9-93 1 on is the review of other consultants' and experts' 2 analyses. Other than Dr. Walker's May '91 paper, have 3 you conducted a review of any other consultants' or 4 experts' analysis in connection with this case? 5 A. No. 6 Q. Has anyone asked you to conduct analysis of 7 other experts' or consultants' work? 8 A. Not at this time. 9 Q. In what ways do you foresee or plan to refine 10 your draft document hereafter in order to finalize it? 11 A. There are a number of proposed tasks on page 12 5 and 6 of my report, and I assume that I will be 13 talking with Dr. Lettenmaier and counsel as to which, 14 if any, of those tasks they may want me to perform or 15 other tasks. 16 Q. I take it, then -- or am I correct in taking 17 it, then, that thus far you are not authorized or 18 directed to proceed with any of those follow-on efforts 19 identified in your draft report? 20 A. Not the ones that have been identified in my 21 draft report, no. 22 Q. Who solicited you to conduct the additional 23 analysis in the refuge with the S-5A and S-6 data? 24 A. Dr. Lettenmaier. 25 Q. What was his expressed purpose for having you LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0046 STEVEN P. MILLARD, V.1, 3-9-93 1 look at that particular data? 2 A. He said the request came from counsel, from 3 Mr. Blank. 4 Q. Do you know what you are looking for or how 5 that data will fit with the draft as you have conducted 6 it thus far? 7 A. Yes. As I understand it, the question is if 8 limits are imposed in the refuse on how much phosphorus 9 is allowed in the refuge, the question is whether there 10 is in fact any relationship between the amount of 11 phosphorus at the inflow structures S-5A and S-6 and 12 the actual phosphorus that's observed in the refuge. 13 So the point of my analysis that I have been requested 14 to work on is to look at what, if any, relationship 15 there is between phosphorus concentrations at the 16 inflow structures S-5A and S-6 and phosphorus 17 concentration in the marsh. 18 Q. In conducting your work thus far, have you 19 had access or been provided any of the earlier drafts 20 of the SWIM Plan, the several iterations it went 21 through prior to -- 22 A. No, no. 23 Q. In the same vein, were you provided any of 24 the earlier iterations of Dr. Walker's work related to 25 targeted interim concentration limits of total LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0047 STEVEN P. MILLARD, V.1, 3-9-93 1 phosphorus in the refuge? 2 A. No. 3 Q. The final category in your designation of 4 witness relates to the grounds and date for opinion. 5 The initial category there or I suppose it's intended 6 to be a grounds, it says, "See SWIM process 7 procedures." Do you have any idea what that means? Or 8 put another way, can you explain to me what it means? 9 A. I don't know what that means. 10 Q. Would the same answer apply to the related 11 data and documents rubrics, with the exception of Dr. 12 Walker's document that you had prior to the production 13 of your January draft? 14 A. Well, the second part of the related data and 15 documents, I assume what that means is whatever 16 opinions I come up with would be based on the data that 17 the water district is using to come up with limits. 18 Q. Do you know what DER is? 19 A. I assume that stands for Department of 20 Environmental Resources, something like that, a state 21 agency. 22 Q. What DER policies and procedures have you had 23 access to and reviewed in connection with your work on 24 the case thus far? 25 A. The only ones I would have in my possession LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0048 STEVEN P. MILLARD, V.1, 3-9-93 1 is whatever is printed in the SWIM -- the document -- 2 MR. BLANK: The planning document? 3 A. The planning document, which as I say, I have 4 only looked at briefly. 5 Q. You understand that the Everglades SWIM Plan 6 planning document dated March 13 of 1992 is adopted by 7 the board of the South Florida Water Management 8 District which is not a subsidiary or related directly 9 to DER? 10 A. In that case, then I have not seen any DER 11 policies and procedures. 12 Q. With regard to the date of your final 13 opinion, this notice, which is, as I said, filed back 14 in October of this past year, thought you might have 15 your final opinions by February 15th of 1993. 16 Obviously that's not the case, correct? 17 A. That's correct. 18 Q. It further says that it's dependent upon 19 completion of work in progress and analysis of data to 20 result from timely entry and access to the park and 21 refuge. If additional sampling is done in the refuge, 22 would you employ data derived from further sampling 23 program in the refuge to update your bootstrap program, 24 your bootstrap run? 25 A. If requested by counsel, I could look at that LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0049 STEVEN P. MILLARD, V.1, 3-9-93 1 data and think about how I might incorporate it in 2 whatever modeling procedure's come up with. Whether I 3 would use that to update my bootstrap procedure is -- I 4 don't know at this point. 5 Q. Assume the same sampling stations employed in 6 Dr. Walker's analysis and your analysis of Dr. Walker's 7 analysis continue to be sampled. Does it have any 8 value to you in your analysis to have that data on -- 9 where the sampling is done on a monthly basis 10 hereafter? 11 A. Yes, that would help in the analysis -- well, 12 I have to -- I have to answer that with a caveat. It 13 would help with analysis assuming that nothing has 14 changed drastically in the sampling procedure and how 15 the samples are analyzed between the dates when the 16 original data was taken between '78 and '83 and the way 17 the data is being collected now and analyzed. If 18 substantially different laboratory techniques are being 19 used, then that could call into question the validity 20 of combining those two data sets. 21 Q. Isn't that equally true with respect to the 22 five years' worth of data and any data you might crank 23 into the equation from S-5A and 6? 24 A. That's certainly true. If there has been a 25 drastic change in laboratory analysis over that LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0050 STEVEN P. MILLARD, V.1, 3-9-93 1 five-year period, then that would affect any results 2 that you get from analyzing that data. The reason that 3 I bring the point up is just because from my 4 perspective there is a large gap in time between when 5 the last observation was taken in the refuge that's in 6 Walker's report and the current time, and so as a 7 statistician, that always rings a bell, an alarm. 8 Q. If and when that data became available to 9 you, if the data falls within the historic range that 10 your table and Dr. Walker's table identified for the 11 base period of five years and the 14 stations that -- 12 or 16 stations that appear there, so if the data falls 13 within that range, does that tell you anything about 14 the accuracy of the predictive value of the model 15 constructed by Dr. Walker? Can you validate the model 16 from that? 17 MR. BLANK: Object to the form of the 18 question. It's very vague and ambiguous to me. I am 19 not sure what you are asking. 20 Doctor, if you can answer it, go ahead. 21 Q. It isn't to me, but it is to the witness. 22 I can ask it another way if you prefer. 23 A. Actually could you ask it again. I was going 24 to proffer an answer, but that was going to be full of 25 caveats so why don't you -- LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0051 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. If data from a current sampling program had 2 given geometric mean stages arithmetically derived as 3 they are in your report and Dr. Walker's report falls 4 within the predictive range of Dr. Walker's model, 5 would that not tend to validate the predictive capacity 6 of the model with respect to interim targets and 7 limits? 8 MR. BLANK: Just for the record let me repeat 9 the same objection, but go ahead if you have an 10 answer. 11 A. My answer to that is that I can't say for 12 sure because you're making the assumption that the 13 model that was used by Dr. Walker is a good one in 14 predicting -- in modeling the system, and it may be 15 that if in fact -- if you use a different model to 16 model the system, then the new data that you are 17 collecting may actually show that you are not falling 18 into -- you are not -- really have something else going 19 on. 20 Q. Go ahead. I didn't mean to cut you off. 21 A. I was going to say the reason that I say that 22 is you prefaced your question by saying for a given 23 stage, you said for a given geometric mean stage,, 24 actually I think they just used the straight average 25 for the stage. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0052 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. That's why I changed that. 2 A. You said for a given stage if the new data 3 that's being collected seems to fall into the range 4 that was observed under the -- with the originally 5 collected data in the '78 to '83 period, your question 6 is wouldn't that validate the model that Walker 7 developed, and I am saying not necessarily because it 8 may be that stage is not the best variable to use to 9 try and model the system. 10 Q. What in the data sets available to you 11 suggests itself as a better variable? 12 A. I have not looked at the data closely enough 13 to see if there are other variables that might be 14 better. I haven't gone that far in my analysis. 15 Q. So at least at this point you cannot identify 16 any variable that would be superior to stage data? 17 A. No. 18 Q. In reviewing the data that was available and 19 in reviewing the SWIM Plan, you are aware, are you not, 20 that a relationship has been identified between stage 21 data and concentration levels in surface water in 22 Loxahatchee refuge? 23 A. That was pointed out in Dr. Walker's report. 24 Q. It is also in the portion of the SWIM Plan 25 that you read, was it not? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0053 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Yes, I believe there was a sentence that made 2 a statement to that effect. 3 Q. Do you have any basis right now for 4 suggesting that that's incorrect? 5 A. No, I can't say that's incorrect. 6 Q. I understand that there is a null hypothesis 7 there. 8 What ongoing work in the refuge is the 9 resolution of your draft into a final document 10 dependent upon? If it's not additional testing and 11 sampling, because you have just suggested that in fact 12 may not tell you anything. 13 MR. BLANK: Objection. That's not what the 14 witness testified to. 15 Q. Am I incorrect in understanding that 16 additional test data from the refuge, current test 17 data, may in fact have no value whatsoever in 18 validating or invalidating the model that Dr. Walker 19 has propounded? 20 A. I believe what I said, the point I was trying 21 to make if I didn't make the point was that whatever 22 model you are looking at, if you want to use the data 23 that was originally collected in the '78 to '83 time 24 period and take that data and combine it with the data 25 that's currently being collected, you need to be LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0054 STEVEN P. MILLARD, V.1, 3-9-93 1 careful about any differences in collecting procedure, 2 analyzing procedure between those two time periods. 3 Q. That's not a novel concept, though, is it? 4 That's something that you would anticipate any 5 reasonably competent sampler, statistician, 6 environmental engineer would have in mind. That's 7 always a danger in looking at a long period of record 8 for data. 9 A. That's correct, that's correct. 10 Q. And you understand that the predictive 11 capacity of the model incorporated in Appendix E to the 12 SWIM Plan utilizes that five-year period to set certain 13 limits, set limits and set targets for Loxahatchee. 14 A. Yes. 15 MR. BLANK: Let me just interject something 16 for clarification. Are we talking about Appendix E or 17 the Walker report now? 18 MR. FITZGERALD: I said Appendix E now. 19 Q. Do you understand Appendix E has a model as 20 well? 21 A. Yes. 22 Q. You understand that it's not the same model 23 as Dr. Walker's May 1991 paper? 24 A. Yes. 25 Q. Is it fair to say that it's similar in LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0055 STEVEN P. MILLARD, V.1, 3-9-93 1 concept but executed slightly differently in detail? 2 A. In layperson language, yes, you could say 3 that. 4 Q. That's for the benefit of the hearing officer 5 actually. We'll talk about the difference. 6 So what are you waiting for from Loxahatchee? 7 A. In order to -- 8 Q. Finalize this draft. That's really how we 9 got around. That's where we were going with all that. 10 A. Okay. What I would want to do before I 11 finalize this report is engage in at least some of 12 these proposed tasks. 13 Q. Am I correct in understanding that your 14 proposed tasks are not dependent on analysis of data 15 from current testing or sampling in the refuge? 16 A. I am sorry, say that -- could you repeat the 17 question. 18 Q. You have cited to the five or six things-to- 19 do ideas. Are any of those dependent on current 20 sampling in the refuge? 21 A. They're not dependent on current sampling in 22 the refuge so that I could form some opinion of the 23 model based on that data. However, obviously if there 24 is other data available and now I know of that, then I 25 will say at this point that I cannot form a final LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0056 STEVEN P. MILLARD, V.1, 3-9-93 1 opinion until I have seen that data as well. 2 Q. When did you become aware that other data was 3 available? 4 A. Oh, I have known that for at least a month. 5 Q. Has anybody provided you with that data? 6 A. No. 7 Q. Have you asked for it? 8 A. Dr. Lettenmaier told me that it's in the 9 process of being collected and he was -- he thought 10 that probably only one or two dates were actually 11 available at this point and so it wouldn't be worth 12 looking at until a few more observations were 13 available. 14 Q. Are you doing any work with respect to 15 Everglades National Park? 16 A. Not currently. 17 Q. Has anybody defined your scope of work to 18 include any issues directly related to Everglades 19 National Park inflows or phosphorus concentration 20 limits for the park as distinct from the refuge? 21 A. Not currently. 22 Q. So that portion of the grounds/date for 23 opinion is probably somebody just ran the word 24 processor with the standard language. 25 MR. FITZGERALD: Could we have this marked as LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0057 STEVEN P. MILLARD, V.1, 3-9-93 1 Exhibit-4. 2 (Exhibit-4 marked.) 3 Q. Doctor, I am providing you an excerpt of the 4 planning document of the Surface Water Improvement and 5 Management Plan, SWIM Plan if you will, for the 6 Everglades adopted March 13, 1992, by the South Florida 7 Water Management District, and the excerpts encompass 8 page 100 through page 105. If I can refer to paragraph 9 captioned number 2, Phosphorus Levels for LNWR, 10 Loxahatchee National Wildlife Refuge. This is the 11 portion of the SWIM Plan planning document, you 12 mentioned earlier you had reviewed the pages 102 to 13 105? 14 A. Yes, I have briefly read this. 15 Q. You are aware that the actual model as 16 derived and adopted by the water management district 17 employs 14 internal marsh stations to set its limits -- 18 A. Yes. 19 Q. -- and targets. And three water gauging 20 stations or stage data stations for stage data 21 variables in the equations? 22 A. Yes. 23 Q. The top of page 103 there is a sentence that 24 says, "To account for the observed correlation between 25 marsh total phosphorus concentration and stage, the LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0058 STEVEN P. MILLARD, V.1, 3-9-93 1 concentration levels vary with the average interior 2 stage on the date of sample collection." That's what 3 you were referring to earlier about the -- 4 A. The statement to the effect -- 5 Q. The potential problem with the model if that 6 turns out not to be true. Or no, I am sorry -- 7 A. I didn't really say whether that was true or 8 not. There may be other variables to look at. 9 Q. In reviewing Dr. Walker's work and in 10 reviewing the material contained in this document, did 11 you find anything to suggest that the linear 12 relationship, the simple regression analysis that was 13 conducted by Dr. Walker to establish that relationship 14 was incorrect? 15 A. I have not at this point, because I have not 16 done an analysis of the goodness of fit of that model. 17 To produce this report, as I said, Dr. Lettenmaier said 18 let's take as given that this is the model that will be 19 used. 20 Q. Do you understand what the report means or 21 this section of the planning document means when it 22 talks about a ten percent rejection level for the three 23 stage data stations? 24 MR. BLANK: Where are you reading, Counsel? 25 MR. FITZGERALD: It appears several places. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0059 STEVEN P. MILLARD, V.1, 3-9-93 1 It's on page 104 and it's also back there I think on 2 102. 3 MR. BLANK: I saw ten percent -- 4 MR. FITZGERALD: On top of 103. 5 MR. BLANK: That's relating to the 14 station 6 interior marsh mean concentration. 7 MR. FITZGERALD: Yes. 8 Q. Then on 105 -- I am sorry, 104, the middle of 9 the page, the paragraph starts "Effective July 1," it 10 talks about "the ten percent rejection level of 11 stations CA1-5, CA1-6 and CA1-16 at a given mean daily 12 stage." Do you understand what they are talking about 13 there by a ten percent rejection level? 14 A. I believe what they mean is the -- when they 15 developed the interim levels in Appendix E, they fit a 16 regression model that involved the variable stage as 17 predictor variable and also a time variable to account 18 whether you were in the baseline period or not. When 19 they fit this regression model, you can then come up 20 with a number of statistics and a number of ways of 21 quantifying statistics associated with that fit, and I 22 believe what they're referring to here when they talk 23 about a ten percent rejection level is the following: 24 For a given value of stage, if you make the assumption 25 of normally distributed errors, then you can predict LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0060 STEVEN P. MILLARD, V.1, 3-9-93 1 what the distribution of the response variable is, in 2 this case the response variable is the geometric 3 mean -- sorry, in their model the response variable is 4 actually the mean of the log of the observations. 5 So for a given value of stage, if you make 6 the assumption normally distributed errors, then a 7 standard statistical method or technique tells you that 8 the distribution at that given level of stage is going 9 to be a normal distribution with a certain mean and a 10 certain variance. Since you know the distribution, you 11 can state what the 90th percentile is of that 12 distribution. If I remember correctly, that is what 13 they were using as the interim limit or the interim 14 level is what they're calling it here is the 90th 15 percentile of that distribution for a given value of 16 mean daily stage. 17 So what that means is if your model is 18 correct, if the assumptions of normal errors is 19 correct, if your model is correct, you have got the 20 right variables, then for a given value of mean stage 21 you can say that 90 percent of the time the mean of the 22 log should fall below this value. You can actually 23 transform the results back into the original units of 24 parts per billion, so you can say 90 percent of the 25 time your observed geometric mean should fall below LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0061 STEVEN P. MILLARD, V.1, 3-9-93 1 this number. So what that means is that ten percent of 2 the time your geometric mean is going to fall above 3 that number. 4 Q. If your goal is to protect a resource and 5 that is defined as not allowing above a certain 6 concentration of, say, phosphorus, to be detected in 7 the area in your compliance test, am I correct in 8 understanding that by setting a 90 percent confidence 9 interval what you are saying is that in 90 percent of 10 the tests if your predictive model is correct, you 11 should in fact fall within the acceptable limits that 12 you have defined? 13 A. If you are defining the limit to be the 90th 14 percentile, then what that means is if the true 15 concentrations are -- if the true geometric mean 16 concentration is the same as it was in the baseline 17 period, then you would not exceed that limit except ten 18 percent of the time. 19 Q. So you are accepting a ten percent risk or 20 probability that you will exceed based on the model as 21 designed? 22 A. That's correct. 23 Q. Is it valid to assume a normal error 24 distribution in this case? 25 A. I couldn't answer that until I looked at the LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0062 STEVEN P. MILLARD, V.1, 3-9-93 1 data more closely. Actually what they assumed was that 2 the original data, they assumed a log normal 3 distribution because they took the log of the data and 4 assumed that in log space -- logarithmic space -- 5 Q. When you say "they," to whom are you 6 referring? 7 A. The people who developed the model that's 8 described in Appendix E. 9 Q. Dr. Walker did not employ that technique 10 initially, did he, in his paper that you reviewed? 11 A. No. The model that -- when I -- the 12 difference between the model in Dr. Walker's and the 13 model in Appendix E, I believe there's two main 14 differences. The first difference has to do with the 15 data that were used. In Appendix E data from stations 16 1 and 2 were not used. There were only data from 14 17 stations used, whereas in the report by Dr. Walker data 18 from 16 marsh stations were used. 19 The second difference is that Dr. Walker, 20 instead of taking the log of the observations and then 21 doing his regression and log space, he looked at 22 geometric means, which is essentially you take the log 23 of the data, take the mean and then transform it back 24 to the original units, that's one way of thinking of a 25 geometric mean, and then actually instead of using the LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0063 STEVEN P. MILLARD, V.1, 3-9-93 1 geometric mean as the response variable he made another 2 transformation. He took one over the square root of 3 the geometric mean, and that he used as his response 4 variable in the regression model. So he assumed that 5 one over the square root of the geometric mean that 6 data were normally distributed. 7 Q. It's your testimony that you can't verify 8 that that is correct or not without doing further 9 analysis? 10 A. Yes. Let me say it's my testimony that I 11 can't say whether whether it looks like taking a log 12 transformation or doing what Dr. Walker did appears to 13 get you closer to a normal distribution or not. You 14 can't really say what transformation is correct, 15 because in reality you can't ever say that anything is 16 precisely normally distributed. 17 Q. Is the choice between the two, then, a matter 18 of personal preference or is there some factor that 19 would dictate you would use one or the other? 20 A. I am not sure what led Dr. Walker to use the 21 transformation that he did. It may be that he 22 originally took the log of the observations and for 23 some reason after looking at certain diagnostic plots 24 decided that that didn't work as far as bringing it 25 close to a normal distribution. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0064 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. If one were able to bring to this work 2 experiential input from a limnologist, someone of that 3 character, could their input tilt you in one direction 4 or the other in choosing between the two techniques? 5 A. Most statisticians if they're doing analysis 6 would want to consult with an environmental scientist 7 to ask what's known in their field about how certain 8 quantities behave, and in fact I am aware from my work 9 in working with environmental data and water quality 10 data that in fact a lot of times people do assume that 11 the data are log normally distributed. 12 Q. What's your understanding of why stations 1 13 and 2 were dropped by the water management district in 14 its Appendix E analysis? 15 A. I don't think I recall all of the 16 explanation. I do recall that there were quite a 17 number of missing values at those two stations 18 something on the order of 40 percent, and I believe 19 that was at least one of the reasons given. Of course 20 that wouldn't be satisfactory to a statistician for why 21 you would drop data. 22 Q. You recall the 40 percent, so I know you have 23 an excellent short-term memory. When did you read 24 Appendix E? 25 A. I believe I read it yesterday. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0065 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. Do you recall the discussion in Appendix E 2 related to the rejection of those values because as 3 well the perceived link between stage data and 4 concentration levels was such that at extremely low 5 water stages certain of the sampling sites could dry 6 down or become very, very difficult to sample, thereby 7 introducing a bias in the sample results? 8 A. I do recall that explanation now that you 9 bring it up. 10 Q. Is that the kind of thing that you would be 11 consulting with the limnologist or the environmental 12 scientist to learn in order to determine if your 13 population of data can validly be manipulated? Not in 14 the sense that -- you can always manipulate it whether 15 it's real world valid or -- this is your reality check, 16 right? Given the statements of the district proponents 17 in Appendix E, assuming for our purposes just for this 18 question that those are correct and there is that 19 direct linkage and they would come up with relatively 20 asymptotical values for peak concentration as the water 21 stage diminishes towards drydown, would that be a valid 22 reason in your mind for rejecting the data points from 23 stations 1 and 2 and reducing your field from 16 24 stations to 14? 25 A. Before I answer that question, let me see if LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0066 STEVEN P. MILLARD, V.1, 3-9-93 1 I can summarize it. So the question is if certain 2 environmental scientists or limnologists explained that 3 the process that's going on is when the stage is so low 4 that you do have a problem with sampling for those two 5 stations, 1 and 2, if the environmental scientist 6 and/or the limnologist said that is that a valid reason 7 for dropping those stations? Is that the question? 8 Q. Yes, along with the 40 percent missing data 9 during the period of record. 10 A. I can't give you a definite opinion at this 11 point, because as a statistician I would have to worry 12 about is it really only those two stations that you 13 have that problem with or are there other stations as 14 well. 15 Q. So you would seek further information of an 16 experiential nature to answer that? 17 A. That's correct. 18 Q. How much absent data would you be willing to 19 accept in conducting this kind of analysis before you 20 would want to reject a data point? I take it from your 21 testimony 40 percent doesn't trouble you particularly. 22 A. When you say rejected data point, you really 23 mean rejected a particular station? 24 Q. Yes. I am sorry, by rejected data point I 25 mean sampling site. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0067 STEVEN P. MILLARD, V.1, 3-9-93 1 A. A statistician -- well, at least my 2 philosophy is that you want to use as much data as you 3 can. Even if they're missing values in a long series 4 you would still want to use as much data as you can, 5 and if there is not a good -- a very good reason for 6 leaving that data out, you would want to include that 7 in your model. 8 Q. If you have missing data coupled with a known 9 bias effect from the available data, would that 10 persuade you in the direction of omitting that data? 11 A. In a situation like that, I would either have 12 to investigate with the environmental scientist if 13 there is a way to somehow correct for that bias or else 14 leave those data out. 15 Q. Have you ever indicated to Dr. Lettenmaier or 16 anyone else that refinements would occur in your draft 17 report to take it to a finalized version based upon 18 water quality data generated as a result of further 19 sampling in the refuge? 20 A. Do I see -- let me again try and summarize 21 the question. Have I indicated to Dennis that my 22 report would be -- 23 Q. Or anyone else. 24 A. Or anyone else. -- would be revised after I 25 received the data that's being collected now? I am not LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0068 STEVEN P. MILLARD, V.1, 3-9-93 1 sure I understand your question. 2 Q. Yes. That's a fair summary. 3 A. At this point we haven't really discussed all 4 the details what would be involved in finalizing this 5 report, so I guess in general terms I have talked with 6 Dennis about yes, once we get new data we will probably 7 renew this study, perhaps do other bootstrap, maybe 8 with this model, maybe with a different model. 9 Q. Why would you bother doing another bootstrap 10 on Dr. Walker's model when it's not in the SWIM Plan? 11 A. Well, I guess I'd have to say at this point I 12 am not sure. 13 Q. Did anyone ever explain to you prior to your 14 conduct of the bootstrap analysis that in fact Dr. 15 Walker's May '91 model was not incorporated in the SWIM 16 Plan? 17 A. No, no one explained that to me. 18 Q. When did you first become aware that the SWIM 19 Plan Appendix E equations were not the same as Dr. 20 Walker's? 21 A. Last week. 22 Q. Did you ask Dr. Lettenmaier why he did not 23 have you do this analysis on the version that's in the 24 challenged SWIM Plan that's germane to this proceeding? 25 A. Yes, I asked him, and he said that he himself LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0069 STEVEN P. MILLARD, V.1, 3-9-93 1 wasn't sure at that point which model was actually 2 going to be adopted by the water district, whether it 3 was going to be Walker's or whether -- Walker's model 4 or whether it's going to be what's in the appendix. 5 Q. Do you see the date of adoption on Exhibit-4? 6 A. Yes. 7 Q. That's before you first met with Dr. 8 Lettenmaier, isn't it? 9 A. Yes. 10 Q. In fact it's a good six months before Dr. 11 Lettenmaier first approached you to do any work in this 12 case. 13 A. Yes. 14 Q. So the version adopted was already adopted at 15 the time he approached you. 16 A. Uh-huh. 17 Q. Which is not Dr. Walker's model -- 18 MR. BLANK: Counsel, when you use the term 19 "adopted," you are referring to the approval by the 20 board of the SWIM Plan, are you not? 21 MR. FITZGERALD: The board formally adopted 22 by a vote the SWIM Plan on March 13th. I understand 23 the legal implication of the challenge, but for the 24 purposes of the board it's adopted subject to 25 resolution of the case. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0070 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. I suspect you were not aware of that, but for 2 our purposes counsel is certainly correct it's adopted 3 subject to the outcome of the case. But at the time -- 4 I need to explore a little bit. What was your 5 understanding of what Dr. Lettenmaier meant? To your 6 understanding was he drawing a distinction between what 7 was actually in the SWIM Plan? 8 A. I don't think he was clear in his mind 9 what -- which -- what the final model was or is that 10 has been adopted. 11 Q. Is he assisting you in this work? 12 A. In which work are you talking about? 13 Q. The refinement and finalization of your draft 14 report. 15 A. Yes, I am working with him. 16 (Exhibit-5 marked.) 17 Q. I am showing you what's marked Exhibit-5. Is 18 this is the computer source code document you mentioned 19 at the outset of the deposition that you had provided 20 through counsel that document your application of 21 bootstrap? 22 A. That's correct. 23 Q. I have seen bootstrap programs that can 24 produce an out column for N equal to 1,000 as yours 25 does in a fraction of the length of your program. What LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0071 STEVEN P. MILLARD, V.1, 3-9-93 1 other capabilities are reflected in your program as 2 provided beyond conducting the strict bootstrap 3 analysis? 4 A. Well, the way the program works is it -- for 5 each bootstrap sample it samples with replacement 6 the -- what's called the regression matrix. In this 7 case that would be the rows would correspond to the -- 8 Q. Station, sampling station. Rows, do you mean 9 horizontally or vertically? 10 A. I will have to go back to look at this to 11 tell you whether I mean rows or columns, but -- 12 Q. Let me suggest to you that rows are dates and 13 columns are stations. 14 A. Okay. 15 Q. And point you at your draft document where I 16 think you show that in your -- 17 A. Here we go. 14 rows and 3 columns. The rows 18 correspond to the dates and the columns correspond to 19 the predictor variables in the model. So for each 20 bootstrap sample I randomly sampled those 14 rows with 21 replacement, did a regression fit, computed the limits 22 based on the same methodology that Dr. Walker used. 23 Now, the result of -- sorry, let me go back and ask -- 24 let me make sure I understand your original question. 25 When your original question are you -- is it your LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0072 STEVEN P. MILLARD, V.1, 3-9-93 1 understanding that all these pages were involved in 2 performing the bootstrap? 3 Q. That is -- 4 A. Okay, that is an incorrect assumption. The 5 actual function that was used to perform the bootstrap 6 is this function named bootstrap.lm which goes from the 7 bottom of that page up to about -- 8 Q. The bottom of the first page through about 9 eight lines on the second page? 10 A. Right, and then there is another function 11 bootstrap.pred.limits, which I can't recall whether I 12 used one of these or both of these. The other 13 functions here that you see listed out involve the 14 functions that I used, for example, to produce the 15 pictures in the report. 16 Q. You anticipate my question. I was going to 17 ask you whether or not the program source codes as you 18 have provided them in that document that runs, what, 19 ten, 15 pages can generate graphical displays. I take 20 it from your answer the answer is yes. 21 A. If you have S-PLUS. 22 Q. A little plug. I was going to ask you did 23 you do so in aid of your analysis? 24 A. Did I -- 25 Q. Generate graphical displays. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0073 STEVEN P. MILLARD, V.1, 3-9-93 1 A. Right. And those are the graphical displays 2 that you see in this report. 3 Q. Other than the whisker graphs and the block 4 graphs and whatnot that you have provided as part of 5 your draft report from January 8th, 1993, did you 6 produce any other graphic displays to assist you in 7 your analysis? 8 A. Not in this particular analysis. There are 9 probably some functions listed in here that I used to 10 look at some of the raw data or maybe even do some 11 diagnostics on the model, but that I never proceeded to 12 completion with that work. 13 Q. Did you save the diagnostics to memory? 14 A. To this memory or to my computer? 15 Q. Not to soft memory. Electronic. 16 A. The diagnostics of the model are probably -- 17 either the code to produce the diagnostics and/or the 18 code to produce pictures may be on my computer, but as 19 I said, it turned out that Dennis wanted me to assume 20 that this is the model that we're going to use and so 21 he didn't want me to pursue looking at diagnostics for 22 that model. 23 Q. The purpose of looking at diagnostics for 24 that model would be, and correct me if I am wrong, if 25 you wanted to try and design a better model. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0074 STEVEN P. MILLARD, V.1, 3-9-93 1 A. That's correct, or just assess how well the 2 current model's doing. 3 Q. So you didn't hard copy any of that 4 diagnostic work or the other pictures? 5 A. I may have, but it never was put into any 6 report that I produced. In fact you might consider 7 that work in progress if I ever end up doing anything 8 like that. 9 Q. Other than diagnostics, did you conduct any 10 other manipulations or analysis of the data to evaluate 11 the model proposed by Dr. Walker or the program 12 incorporated in Appendix E of the SWIM Plan? 13 A. I think I have -- I know I have fit some 14 other models where I think I used -- instead of using 15 the geometric mean as -- or actually one over the 16 square root of the geometric mean as the response 17 variable, I probably looked at different 18 transformations of the raw data, and I may have some 19 diagnostics printed out from those models, but in fact 20 I have not looked at them to see -- to interpret them. 21 Q. You believe those have been printed out? 22 A. I believe I printed out at least a couple 23 pictures of diagnostics, a couple graphs of diagnostics 24 from that. 25 MR. FITZGERALD: Counsel, I think those are LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0075 STEVEN P. MILLARD, V.1, 3-9-93 1 producible under the notice. And I would like them 2 along with the other items that we identified earlier. 3 I certainly would not delay the proceeding for them, 4 but I think because they were done in relationship to 5 the development of this report which has been produced 6 that they are subject to discovery at this point. 7 Q. There is no proprietary information in there 8 or anything other than -- I mean those diagnostics and 9 materials that you have developed, they're all related 10 to your evaluation of Dr. Walker's model and the work 11 that he was doing? There's no proprietary data as the 12 foundation of that work? 13 A. No, no. 14 (Exhibit-6 marked.) 15 Q. Showing you what's marked as Exhibit-6 for 16 the deposition, I represent to you that this is an 17 excerpt from the Everglades SWIM Plan as adopted by the 18 South Florida Water Management District on March 13, 19 1992, from the appendices volume, and it's from 20 Appendix E, pages E-16 through E-23, captioned 21 "Phosphorus Levels for Loxahatchee National Wildlife 22 Refuge." I ask you to take a minute, just look through 23 that and tell me if this in fact is the portion of 24 Appendix E you were referring to earlier that you 25 received within the last few days and read yesterday I LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0076 STEVEN P. MILLARD, V.1, 3-9-93 1 guess. 2 A. That's correct. 3 Q. The second paragraph on page E-16 states that 4 "Statistical analyses of these data," as identified in 5 the earlier paragraph, "utilize the logarithmic scale 6 of total phosphorus concentration which was found to 7 improve additivity," which is easier to spell than say, 8 refers you to figure C and then says, "and reduce 9 skewness," s-k-e-w-n-e-s-s. This is one of the 10 recommendations or comments you had in your draft 11 report on Dr. Walker's work that this is an approach 12 that should be pursued, was it not, the use of the 13 logarithmic scale? 14 A. The comment that I made in my report -- 15 Q. That's page 3, comment number 2? 16 A. Yes, page 3, comment number 2. I was curious 17 as to why he used the geometric mean instead of taking 18 the log of the data and then using a transformation of 19 it, because taking the geometric mean and then taking 20 one over the square root is not something that you see 21 commonly done all the time in analyzing environmental 22 data. 23 Q. If you conducted a similar analysis of the 24 total phosphorus concentration and found that in fact 25 what is stated here is correct, that using a LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0077 STEVEN P. MILLARD, V.1, 3-9-93 1 logarithmic scale improves additivity -- is that how 2 it's pronounced? 3 A. Yes. 4 Q. -- and reduces skewness, would that persuade 5 you of the validity of using the logarithmic scale as 6 the preferred method? 7 A. I would actually have to look at the data 8 itself, look at diagnostic plots of the data, things 9 like what are called Q-Q plots to assess whether taking 10 the log of the data really makes it close to a normal 11 distribution and whether that's, quote, the best way of 12 doing it. I doubt that you will ever get a 13 statistician to tell you there is a best way for doing 14 it. 15 Q. I can live with that. Do you agree that 16 analysis of the database which you conducted in order 17 to do your review of Dr. Walker's work exhibits a 18 permanence of the interior marsh stations, that is, 19 over time their total phosphorus values will tend to 20 stand in the same ratio, or to put it another way, the 21 difference in their log concentrations will tend to 22 remain constant? 23 A. I am sorry, can you repeat the question one 24 more time. 25 Q. Let me do it the easy way. You have reviewed LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0078 STEVEN P. MILLARD, V.1, 3-9-93 1 the data. Do you agree that the database, which is 2 described herein and which is the same database 3 essentially but for the two stations that you analyzed, 4 demonstrate permanence of the interior marsh stations? 5 MR. BLANK: I object to the question as being 6 vague and ambiguous. I don't understand what you mean 7 by "permanence of the marsh stations." 8 Q. Do you see, Doctor, on page E-16, second 9 paragraph, the term "permanence" as that appears at the 10 start of the second full sentence? 11 A. Yes. As I understand this sentence, what 12 they're talking about is in statistical jargon there is 13 an idea of a fixed effect and a random effect, and the 14 difference between the two has to do with the 15 population that you are trying to make an inference 16 towards. So what they're saying is in this analysis 17 you would treat these stations as being in statistical 18 terms a fixed effect factor rather than a random effect 19 factor, because they're making the assumption that in 20 the future sampling will only be done at these 21 particular 16 locations and the limits will be applied 22 from data only from these 16 particular locations. So 23 I agree with the sentence and the thinking that you 24 would call these stations fixed effects if in fact 25 that's how you are going to conduct the sampling LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0079 STEVEN P. MILLARD, V.1, 3-9-93 1 program in the future. 2 Q. You understand from the balance of that 3 discussion that that's exactly how they are going to 4 conduct the sampling. Once they reject stations 1 and 5 2, that is the plan as it stands? 6 A. Right, right. So that part of the sentence I 7 understand and as a statistician would agree with. 8 Q. We already talked about station CA1-1 and 9 CA1-2 on the 40 percent absent data. 10 MR. BLANK: I am sorry, Counsel, are you 11 posing a question? 12 MR. FITZGERALD: Trying to, although I think 13 maybe we already covered it and we can move on. I 14 think we did kind of cover that. 15 (Brief recess.) 16 Q. Would you agree with the characterization in 17 the Exhibit-6 that the 14 stations, excluding out CA1-1 18 and CA1-2, were almost always represented adequately in 19 the sample for statistical purposes? 20 A. As I recall, I would say yes. I don't think 21 there were too many missing values. 22 Q. Well, you analyzed the same data field in 23 reviewing Dr. Walker's work. 24 A. Right. 25 Q. Did you find there would be absent values to LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0080 STEVEN P. MILLARD, V.1, 3-9-93 1 the point that it was not usable within reasonable 2 statistical limits? 3 A. No, no. 4 Q. In the data did you observe the correlation 5 between low stage and high concentrations of 6 phosphorus? 7 A. At this point all I can say is that I 8 observed the -- I reproduced Walker's model which in 9 fact has a statistically significant coefficient for 10 stage which that means yes, there is a -- there is a 11 significant correlation between concentration and the 12 variable stage. 13 Q. What was the level of statistical 14 significance? 15 A. In the model, when both the time variable 16 that indicates whether you are in the baseline period 17 or not and stage or in the model you get a 18 coefficient -- oh, actually I don't have the P-value 19 for the coefficient listed in here, but I am pretty 20 sure it was less than a common standard of five 21 percent. 22 Q. I mentioned earlier the concern voiced in 23 Exhibit-6 about extremely low stage levels and the 24 possibility of sampling difficulties resulting 25 therefrom, biasing the result. Can that happen in your LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0081 STEVEN P. MILLARD, V.1, 3-9-93 1 estimation? 2 A. I am not qualified as an environmental 3 scientist to talk about the physical nature of the 4 system, so I can express no opinion on that. 5 Q. What, statistically speaking, is bias? 6 A. Most statisticians when they are talking 7 about bias mean how far away your estimate or the 8 number that your estimate is tending to, how far away 9 that is from the true population value. 10 Q. In your analysis or review of Dr. Walker's 11 work, did you exclude the March and June of 1979 data 12 at stages 15.21 feet and 14.88 feet respectively from 13 your analysis because of excessive low stage? 14 A. The reason that I excluded those observations 15 is because I was trying to reproduce Dr. Walker's 16 results. 17 Q. Did Dr. Walker exclude those or did the 18 district exclude them in their version in Appendix E or 19 both? 20 A. I believe both. 21 Q. What's your understanding of what one was 22 supposed to do under Dr. Walker's approach if stage 23 level exceeded the highest level in the period of 24 record? 25 A. I can't recall what he said in his report. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0082 STEVEN P. MILLARD, V.1, 3-9-93 1 He may have just said you can't apply these limits in 2 that case, but I don't remember. 3 Q. If that was what he said, would you agree 4 with that from a statistical perspective? 5 A. In general, yes, a statistician will always 6 say that if you developed a model with predictor 7 variables in a certain range and then you want to apply 8 that model where your predictor variables are outside 9 the range of your original data, that's really not 10 valid. In practice, people in the financial industry 11 do that all the time. 12 Q. They sell stock that way, I know. 13 MR. FITZGERALD: If we can mark this as 14 Exhibit-7. 15 (Exhibit-7 marked.) 16 Q. Showing you what's marked Exhibit-7 to the 17 deposition which is a tabular document captioned "wcal 18 underline tp.new" dated Wednesday, March 3, at 13:19:35 19 1993, page number 1. Can you explain what that 20 document is for me? 21 A. This is a printout of the data that I used in 22 trying to reproduce Dr. Walker's result in preparing 23 this report. The file name wca1, not L, underscore, 24 tp.new, that name was -- came from Dr. Lettenmaier. He 25 is the one who created this file. I believe the way he LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0083 STEVEN P. MILLARD, V.1, 3-9-93 1 created this file was by taking data that he had 2 received, I believe he received it from ES&P, and 3 extracted parts of some larger file to produce this 4 file. 5 Q. Did you cross-check the data in this file 6 against the data used by Dr. Walker and the data used 7 by the district in Appendix E? 8 A. I cross-checked this data against the data 9 that appeared in Dr. Walker's report. And in fact 10 that's what gave rise to the discrepancies and 11 omissions that I talk about in my report under the 12 heading "data validation." 13 Q. So you had not conducted a similar comparison 14 with the data used by the district in Appendix E? 15 A. In Appendix E, no, I have not compared these 16 data to what's in Appendix E. 17 Q. What did you understand the objective of the 18 analysis by Dr. Walker in his May '91 paper to be? 19 A. Well, according to what's stated in his 20 report, it was -- the intent was to set limits on 21 phosphorus concentration in the refuge. 22 Q. Would you agree with the district's 23 characterization in Appendix E that their purpose was 24 to track performance of a control plan and account for 25 sources of variability? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0084 STEVEN P. MILLARD, V.1, 3-9-93 1 A. I have no way of saying that that was their 2 purpose or not. If they say that's their purpose, then 3 that's their purpose. 4 Q. Does their statistical analysis support that 5 statement of purpose? 6 MR. BLANK: If you know, Doctor. 7 A. And their statement of purpose again is? 8 Q. To track performance of a control plan and 9 account for sources of variability. 10 A. I don't know the answer to that because I 11 don't know what control plan they're talking about. 12 Q. What did you understand the term "target" to 13 mean as used by Dr. Walker? 14 A. I am not sure why Dr. Walker used the 15 terminology target. I just remember that he used the 16 word "target" to designate the 50th percentile of the 17 predicted distribution of phosphorus at a given stage. 18 Q. So that would be a predicted value in the 19 base period? 20 A. That's correct, a predicted value -- 21 Q. Of concentration? 22 A. Of concentration assuming that the system was 23 acting like it was in the base period. 24 Q. What did you understand the term "limit" to 25 mean as used by Dr. Walker? LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0085 STEVEN P. MILLARD, V.1, 3-9-93 1 A. I could only guess that it may be -- end up 2 being some sort of compliance limit. 3 Q. As he defined it, wasn't it the upper 90 4 percent confidence level for the predicted value -- 5 A. Yes, that's what he defined it to be. Yes, 6 he used the word -- 7 Q. That's what I -- 8 A. Sorry, he used the term "interim limit" to 9 define the upper 90th percentile of the distribution of 10 phosphorus concentration for a given level of stage 11 based on using 16 marsh stations. 12 Q. And the base period itself, what did you 13 understand base period to be? 14 A. I believe that's the period of sampling that 15 included only observations from -- I would have to look 16 at his report again. I believe it was June of -- 17 Q. '78? 18 A. '78 through July of '79, I can't remember 19 exactly. 20 Q. In reviewing the data set -- 21 MR. BLANK: Just a minute, Counsel, I think 22 he's looking at something with regard to his prior 23 answer. 24 A. This still doesn't help me determine what the 25 baseline period is. Yes, if I recall correctly, it was LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0086 STEVEN P. MILLARD, V.1, 3-9-93 1 June of '78 through July of '79. 2 Q. In reviewing the data set of water quality 3 information for the refuge, how many stations and how 4 many samples are used in the SWIM Plan, Appendix E, for 5 the analysis conducted there? 6 A. As I stated earlier, when I looked at 7 Appendix E and was comparing it in my mind with what 8 Dr. Walker did, I noted that two stations in Appendix E 9 were left out compared to the stations that were used 10 in Dr. Walker's report, and those stations are 11 designated I believe CA1 and CA2. Sorry, CA1-1 and 12 CA1-2. 13 Q. Do you know how many total samples were 14 represented by the data field or data population 15 employed by the district? 16 A. How many dates? How many sampling dates were 17 used? 18 Q. Samples. 19 A. How many -- 20 Q. Sampling events. 21 A. On how many different dates were samples 22 taken? 23 Q. Yes. 24 A. I believe the same dates were used in 25 Appendix E as what was used in Dr. Walker's report. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0087 STEVEN P. MILLARD, V.1, 3-9-93 1 Q. So on the order of 190 thereabouts? 2 A. No, only 14 samplings dates. 3 Q. 14 by 14, 14 dates, 14 -- 4 A. So total number of observations? 5 Q. Yes. 6 A. Yes. 7 Q. Do the differences in stage data or the 8 computation of median total phosphorus values that you 9 note in pages 1 and 2 of your draft have a substantial 10 effect on your analysis or results? 11 A. No. The medians that are reported in the 12 table of Walker that are not the same value as what the 13 conventional median would be reported as, that doesn't 14 affect the result of the analysis. That's a 15 preliminary statistic. 16 Q. How about the stage data differences? 17 A. The difference between the data that I 18 received from Dr. Lettenmaier and the data -- or sorry, 19 the data on stage I actually received directly from 20 ES&P. The differences between the data that I received 21 from ES&P and the data that's reported in the Walker 22 report I don't believe would affect the results in a 23 large way, although I can't remember if I tried a 24 couple different ways. You can see from the values 25 that I ended up using my results were fairly close to LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0088 STEVEN P. MILLARD, V.1, 3-9-93 1 Dr. Walker's. But it may be that a prudent thing for 2 me to do is go back and actually use the different 3 values of stages that were reported. I don't think 4 actually that it would make a big difference because 5 you're averaging over three stations. 6 Q. So you didn't, I take it, then, conduct any 7 like formal sensitivity analysis of the implications of 8 those decimal variations or whatever the total 9 variations were? 10 A. No, I don't think so. 11 Q. Do you know what values the district used in 12 Appendix E? 13 A. No, I don't. Not unless they're reported in 14 there. But I haven't looked at the data in that. 15 Q. On page 2 of your report -- we should make 16 your report, incidentally, an exhibit. 17 The problem is mine has the title cut off so 18 I had to find out what it was and fill it in. Your fax 19 machine or however I got it -- I assume it was faxed at 20 some point. Do you have a clean copy? 21 A. This is a clean copy. 22 Q. Why don't we mark that as the next numbered 23 exhibit which will be 8. 24 (Exhibit-8 marked.) 25 Q. Using all 16 stations as employed by Dr. LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0089 STEVEN P. MILLARD, V.1, 3-9-93 1 Walker, the estimated coefficients and estimates of 2 their standard -- let me start that over again. It 3 sounds right on the record but wrong the way I intoned 4 it. Using all 16 stations, the estimated coefficients 5 and estimates of their standard errors between your 6 report, your draft report Exhibit-8, and Dr. Walker's 7 work as you note in your report agree to the third 8 decimal for the intercept and to the fourth decimal for 9 the other coefficients? 10 A. That's correct. 11 Q. What does the intercept coefficient signify? 12 A. In terms of -- 13 Q. In laymen's terms what does it mean? 14 A. Interpreting it in the model? 15 Q. Yes. 16 A. It means what would be the value of the 17 geometric mean phosphorus concentration or actually in 18 this case whatever transformation he used, one over the 19 square root of the geometric mean, what would the value 20 of that be if the stage was zero, and the time variable 21 was coded as zero which is the baseline period, so in 22 the baseline period what if the stage was zero, then 23 that's what the intercept represents. Physically that 24 doesn't make sense, but it's common -- it's common 25 practice to include an intercept term in a regression LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0090 STEVEN P. MILLARD, V.1, 3-9-93 1 model. 2 Q. And you agree to the third decimal place? 3 A. That's correct. 4 Q. What do the other coefficients mean in which 5 you agree to the fourth decimal place? 6 A. The coefficient for the baseline period is 7 telling you what the -- for a given value of stage what 8 was the change in phosphorus concentration between the 9 baseline period and the period after that, and in this 10 case you have to interpret it in terms of all this 11 transformation that's been done, it's in terms of one 12 over the square root of the geometric mean. The 13 coefficient for the average stage is simply telling you 14 for one unit increase in stage what is the increase or 15 decrease in the phosphorus concentration or one over 16 the square root of the geometric mean of phosphorus 17 concentration. 18 Q. It appears to my uneducated eye that your 19 verification of the Walker fit to the third or fourth 20 decimal place shows a very high correlation in your 21 analysis than his. Is that correct? 22 A. It means that I probably -- it means that the 23 data that I used was very close to the data that he 24 used. Ideally those numbers should agree exactly. If 25 we had exactly the same data, those numbers should LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0091 STEVEN P. MILLARD, V.1, 3-9-93 1 agree. 2 Q. The variance as we discussed earlier, the 3 discrepancies noted in your report amongst the data, 4 would that account for the variation in your mind? 5 A. That could very well account for the 6 differences. 7 Q. What do you mean by regression fit? The term 8 you use there on page 2. 9 A. I need to know better the context. 10 Q. This may be my error. If you look at page 6 11 on your report, table 1 is a comparison of regression 12 fits and it seems to be -- yes, table 1 which is 13 referred to under your verification of Walker fit on 14 page 2. When you say comparison of regression fits -- 15 regression fits, to what are you referring there? 16 A. What I mean there is I am comparing the 17 estimated coefficients that Walker came up with when he 18 did his regression fit and the estimated coefficients 19 that I came up with when I did my regression fit. 20 Q. What is a regression fit? 21 A. A regression fit refers to the -- a 22 particular statistical technique, and the technique 23 that is employed in statistical terms is your finding 24 for a number of predictor variables in the model you 25 are assuming that the response variable is linearly LARSEN & SMITH, INC., 1325 4TH AVE, SEATTLE, WA 98101 (206)623-6717 0092 STEVEN P. MILLARD, V.1, 3-