Rep: OEB Doc: 12WVW Rev: 0 ONTARIO ENERGY BOARD Volume: 2 7 OCTOBER 2003 BEFORE: P. SOMMERVILLE PRESIDING MEMBER A. BIRCHENOUGH MEMBER 1 RP-2003-0063 2 IN THE MATTER OF the Ontario Energy Board Act, 1998, S.O. 1998, c.15 (Sched. B); AND IN THE MATTER OF an Application by Union Gas Limited for an Order or Orders approving or fixing just and reasonable rates and other charges for the sale, distribution, storage, and transmission of gas for the period commencing January 1, 2004. 3 RP-2003-0063 4 7 OCTOBER 2003 5 HEARING HELD AT TORONTO, ONTARIO 6 APPEARANCES 7 PAT MORAN Board Counsel MARTIN DAVIES Board Staff JAMES WIGHTMAN Board Staff MICHAEL PENNY Union Gas Limited TOM BRETT Ontario Association of School Business Officials MICHAEL JANIGAN Vulnerable Energy Consumers' Coalition ROBERT WARREN Consumers Association of Canada ALICK RYDER City of Kitchener GEORGE VEGH CEED, OESC, Superior Energy Management, Union Energy JAY SHEPHERD Ontario Public School Boards Association MIMI SINGH CME RANDY AIKEN London Property Management Association, Wholesale Gas Service Purchasers Group SCOTT STOLL Northern Cross Energy TIBOR HAYNAL TransCanada PipeLines ROBERT ROWE Enbridge Gas Distribution Inc. PETER THOMPSON Industrial Gas Users Association BRIAN DINGWALL Energy Probe, HVAC Coalition, Distributed Energy Association DERECK FRANCIS Energy Objective VALERIE YOUNG Ontario Association of Physical Plant Administrators PETER SCULLY City of Timmins, City of Sudbury, FNOM JOHN RATTRAY Ontario Power Generation 8 TABLE OF CONTENTS 9 PRELIMINARY MATTERS: [16] UNION GAS LIMITED - PANEL 2; WEAVER, FOGWILL, ROOT [23] EXAMINATION BY MR. PENNY: [27] CROSS-EXAMINATION BY MR. BRETT: [155] CROSS-EXAMINATION BY MR. JANIGAN: [480] CROSS-EXAMINATION BY MR. SHEPHERD: [676] 10 EXHIBITS 11 EXHIBIT NO. M.2.1: DOCUMENT ENTITLED "FUTURES RESEARCH WEATHER REPORT, OCTOBER 2001, 2001-2002 WINTER OUTLOOK" [284] EXHIBIT NO. M.2.2: DOCUMENT FROM THE WORLD METEOROLOGICAL ORGANIZATION WEB SITE, FILED BY VECC [571] EXHIBIT NO. M.2.3: CROSS-EXAMINATION MATERIALS FROM THE ONTARIO PUBLIC SCHOOL BOARDS' ASSOCIATION [687] 12 UNDERTAKINGS 13 UNDERTAKING NO. N.2.1: UNION GAS UNDERTAKES TO PERFORM A CALCULATION USING THEIR 20-YEAR TREND METHOD OF WHAT THEY EXPECT THE DEGREE DAYS WOULD BE IN EACH OF THE NEXT SEVEN YEARS, STARTING WITH A BASE OF THE 2003 YEAR, USING THE 4,033, AND ASSUMING THE ACTUAL TEMPERATURES IN EACH OF THE SEVEN YEARS OUT FROM 2003 DROP BY AN AVERAGE OF FOUR DEGREE DAYS A YEAR, INCORPORATING THAT DECREASE IN HEATING DAYS IN EACH SUCCESSIVE YEAR [430] 14 --- Upon commencing at 9:40 a.m. 15 MR. SOMMERVILLE: Thank you very much. Please be seated. This is the resumption of the Union 2004 rates application. Are there any preliminary matters before we begin the next panel? 16 PRELIMINARY MATTERS: 17 MR. PENNY: Yes, Mr. Chairman, just two administrative matters. I wanted to indicate for the record that answer to transcript Undertaking N.1.1 has been provided to parties. This was the risk-management program policies and procedures which had been filed in prior proceedings, and I think it was Mr. Janigan that asked if it could be made available. And so now that's been made available. 18 And then there are a few witnesses who we were not proposing to call, people whose names who appear in interrogatories principally who we were not proposing to call, and we have -- unless it became necessary. We've made affidavits available for those witnesses, adopting their evidence. 19 MR. SOMMERVILLE: That would be Mr. Feldman, Ms. McShane and Mr. Rietdyk. 20 MR. PENNY: That's correct. It's anticipated there might be a few more, but there's a few people out of town and we don't have them yet. We propose that we move to the next step which would be the witnesses addressing the weather methodology. I just wanted to make -- I had some discussion with a few people yesterday I guess privately or offline and I thought it would be worth making the point more publicly that this witness panel is addressing the methodology itself, and the -- either the revenue and forecasting or the operational impacts are not being dealt with by this panel. The revenue/through-put forecasting impacts will be dealt with by the next panel, panel 4, and the other operational impacts, such as storage, transportation, allocation, that kind of thing, will be dealt with by the panel following them, the gas supply panel, which is panel 5. 21 MR. SOMMERVILLE: Thank you. 22 MR. PENNY: So with that, if Dr. Weaver, Mr. Fogwill, and Mr. Root could be sworn. 23 UNION GAS LIMITED - PANEL 2; WEAVER, FOGWILL, ROOT 24 A.WEAVER; Sworn. 25 A.FOGWILL; Sworn. 26 S.ROOT; Sworn. 27 EXAMINATION BY MR. PENNY: 28 MR. PENNY: Thank you, Mr. Chairman. Let me begin, just by way of background, with you, Mr. Root, since you're sitting close to me. You're the president and CEO of WeatherBank Inc. 29 MR. ROOT: That's correct. 30 MR. PENNY: And WeatherBank, I understand, is an environmental and weather forecasting company. 31 MR. ROOT: Correct. 32 MR. PENNY: You have a bachelor of science degree and a masters of science degree in meteorology from the University of Utah. 33 MR. ROOT: Yes, sir. 34 MR. PENNY: I understand you're a certified consulting meteorologist with the American Meteorological Society. 35 MR. ROOT: That's correct. 36 MR. PENNY: And you're also a member, among other organizations, of the National Weather Association of the United States. 37 MR. ROOT: Correct. 38 MR. PENNY: And before being with WeatherBank, I think you've been with WeatherBank since 1984; is that right? 39 MR. ROOT: Correct. 40 MR. PENNY: And you were first there as vice-president and then, since 1992, as the president and CEO? 41 MR. ROOT: Yes. 42 MR. PENNY: Before joining WeatherBank, I understand you were the president and CEO of Spectra Computer Systems. 43 MR. ROOT: Yes. 44 MR. PENNY: You've also held positions as senior meteorologist with both Beak Consultants and a firm called Kennecott Copper Corporation. 45 MR. ROOT: Yes, sir. 46 MR. PENNY: You were also the fire and weather forecaster for the Wasatch National Forecast Service in Salt Lake City, Utah. 47 MR. ROOT: Yes. 48 MR. PENNY: And you've been at various times a broadcast meteorologist for various radio and television markets in the Utah area. 49 MR. PENNY: That's correct. 50 MR. PENNY: And you have published many papers on empirical modelling techniques for climatological forecasting. 51 MR. ROOT: Yes. 52 MR. PENNY: You've also been a guest lecturer from time to time in university and professional development settings on matters relating to computer modelling of climatic and atmospheric conditions. 53 MR. ROOT: Yes. 54 MR. PENNY: What does WeatherBank do, at least insofar as it relates to the issues that are presented here with trends in climatological change? 55 MR. ROOT: Approximately 65 percent of WeatherBank's gross business is in the day-to-day forecasting of weather conditions for energy clients across North America. 56 MR. PENNY: All right. And you, I understand, sir, have testified before in court proceedings in the United States? 57 MR. ROOT: I have, yes. 58 MR. PENNY: And you testified as and were qualified as an expert in weather matters. 59 MR. ROOT: Correct. 60 MR. PENNY: And the report which you prepared for these proceedings, sir, which appears in the book at Exhibit C.1, tab 4, appendix B, that report was prepared by you or under your supervision? 61 MR. ROOT: It was prepared by me. 62 MR. PENNY: And for the purposes of this proceeding, do you adopt that evidence? 63 MR. ROOT: Yes. 64 MR. PENNY: And do the extent you participated in any answers to interrogatories on weather normalization methodology, do you adopt those answers? 65 MR. ROOT: Yes. 66 MR. PENNY: And Mr. Fogwill, you are currently the manager of Market Knowledge for Union Gas? 67 MR. FOGWILL: That's correct. 68 MR. PENNY: And you have held positions previously with Centra Gas B.C. as senior economist? 69 MR. FOGWILL: Yes. 70 MR. PENNY: You were with the British Columbia Energy Council as an energy policy research officer? 71 MR. FOGWILL: That's correct. 72 MR. PENNY: And I understand that you have a masters of resource management with -- from the school of resource and environmental management at Simon Fraser University. 73 MR. FOGWILL: That's correct. 74 MR. PENNY: You also have an advanced B.Sc. from the department of geography from the University of Saskatchewan. 75 MR. FOGWILL: Yes. 76 MR. PENNY: And you have testified previously before this Board in RP-2002-0130, 1999-0017, and EBRO-499. 77 MR. FOGWILL: Yes. 78 MR. PENNY: And, Mr. Fogwill, you have filed both current evidence in this case and a portion of some evidence that was previously filed in Union's earlier customer review process. 79 MR. FOGWILL: That's correct. 80 MR. PENNY: And that appears at Exhibit C.1, tab 4. 81 MR. FOGWILL: That's correct. 82 MR. PENNY: And do you adopt that evidence, sir? 83 MR. FOGWILL: Yes, I do. 84 MR. PENNY: And in the interrogatory process, to the extent that you participated in answers to interrogatories on the issues that you've addressed in relation to weather methodology, do you adopt those answers? 85 MR. FOGWILL: Yes. 86 MR. PENNY: Now, Dr. Weaver, I understand that you are currently a full professor and the Canada research chair in atmospheric science at the school of earth and ocean sciences at the University of Victoria. 87 DR. WEAVER: That's correct. 88 MR. PENNY: And you have an honours bachelors of science in mathematics and physics from the University of Victoria? 89 DR. WEAVER: Yes. 90 MR. PENNY: You have a postgraduate degree in applied math from Cambridge University? 91 DR. WEAVER: That's correct. 92 MR. PENNY: And then you earned, in 1987, your Ph.D. in applied mathematics from the University of British Columbia? 93 DR. WEAVER: That's correct. 94 MR. PENNY: Your Ph.D. was on the subject of numerical and analytical modelling of oceanic and atmospheric processes? 95 DR. WEAVER: Yes, that's correct. 96 MR. PENNY: I gather before joining the University of Victoria in 1992, you spent three years as a natural sciences and engineering research council, university research fellow in the department of atmospheric and oceanic sciences at the McGill University? 97 DR. WEAVER: Yes, that's correct. 98 MR. PENNY: You've written over 120 peer-reviewed papers in climate methodology, oceanography, earth science, policy and education journals? 99 DR. WEAVER: Yes, that's correct. 100 MR. PENNY: Now, you've also -- and you've also been awarded numerous research grants and awards. 101 DR. WEAVER: Yes, that's correct. 102 MR. PENNY: Now, -- I should have indicated earlier, Mr. Chairman, because the evidence was filed a couple of years ago, I did ask Dr. Weaver to provide an updated CV and that's been made available to parties. 103 Dr. Weaver, this document is your current CV? 104 DR. WEAVER: Yes, that's correct. 105 MR. PENNY: All right, thank you. 106 Now, I'm not going to go into the long list of your publications, and the sheer volume of your professional involvement makes it hard to highlight any one thing. But let me try and pick out a few. 107 I understand that you participated in the United Nations Intergovernmental Panel on Climate Change and were the lead author of some of the chapters of both the second and the third scientific assessments that were prepared by that body? 108 DR. WEAVER: Yes, that's correct. 109 MR. PENNY: Can you describe what the UN Intergovernmental Panel on Climate Change is and what your involvement was? 110 DR. WEAVER: Yes. The IPCC, as it's known, is a group of scientists. I participated in something which was working group 1, which conducts a scientific assessment of the state of our knowledge of climate, climate change science, and climate variability. Its first volume was in 1990. I was not involved in that. I was involved in the assessment -- the second assessment and the third assessment, which appeared in 2001, and I'm also involved in the scoping of the fourth assessment will appear in 2007. So it is a group of scientists that work under the United Nations umbrella to provide a state of assessment of climate science about once every five years. 111 MR. PENNY: And when we see in your curriculum vitae that you were the lead author of various chapters, I think it was 6, 7 and 8, if I remember correctly, but what does that mean? 112 DR. WEAVER: Well, each chapter has a number of authors associated with the writing of it. A lead author are those people who are associated with writing it, so I was one of about six or seven authors of the individual chapters that I participated in. 113 MR. PENNY: And I understand you're still a member of the UN World Climate Research Program? 114 DR. WEAVER: That's correct. I am a member of one committee, this is the working group on coupled modellings, which is a UN committee and I'm also -- 115 MR. PENNY: What is coupled modelling? 116 DR. WEAVER: These are the models that are used to make climate change projections, i.e., to understand what's happening in the future. 117 I'm also co-chair of the Clivar/Pages Intersection Panel. What that is, is a group that links the climate dynamicists, the applied mathematicians and physicists which those that have more expertise in the area of observational records to look at climate variability in past climates, that is, climates from about today through to about 135,000 years ago. 118 MR. PENNY: All right, thank you. And among other things, you're also a member of the National Academy of Sciences, National Research Council Board on Atmospheric Science and Climate? 119 DR. WEAVER: Yes. I'm not on the board of Atmospheric Science and Climate, I'm a member of the Climate Change Committee, Climate Sciences Committee. 120 MR. PENNY: And what is the work of the Committee on Climate Change? 121 DR. WEAVER: That acts as an advisory committee, it's arm's length from the government. It's considered an advisory committee. When U.S. government or other agencies in the U.S. require information on climate science, they approach this committee for recommendations. It periodically writes reports which are commissioned by, for example, the EPA in the U.S. and other organizations. 122 MR. PENNY: Thank you. I gather in 2002 you received your research fellowship as one of the top 20 scientists in Canada under the age of 40? 123 DR. WEAVER: That's correct. That's a CIAR award. 124 MR. PENNY: And you were selected by BC Business magazine as one of the 25 power thinkers in British Columbia? 125 DR. WEAVER: That's correct. 126 MR. PENNY: And you have also been selected as one of the top five Canadian scientists by Time Magazine of Canada? 127 DR. WEAVER: That's also correct. 128 MR. PENNY: Now, you prepared a document at tab C.1, tab 4, appendix A of A, entitled "Global Warming: The Intergovernmental Panel on Climate change and Surface Air Temperature Trends in Ontario." Was that report prepared by you or under your supervision? 129 DR. WEAVER: It was done by me exclusively. 130 MR. PENNY: And do you adopt that report for the purposes of your testimony today? 131 DR. WEAVER: I do indeed. 132 MR. PENNY: And to the extent you were involved in the answers to any written interrogatories, do you adopt those answers as well? 133 DR. WEAVER: Yes, I do. 134 MR. PENNY: Thank you, Dr. Weaver. 135 Now, Mr. Fogwill, this evidence, as everyone knows, was filed with the Board over two years ago so there's been lots of time to digest it and I'm sure that the Board and the parties have read these reports with interest so I'm not going to ask anyone to summarize their evidence. There are three distinct pieces here, Dr. Weaver's report, Mr. Root's report, and then there's the Union piece on the analysis and modelling that it did. 136 What I would ask you to do is simply explain what the purpose of filing each of these three pieces was and how they fit together? 137 MR. FOGWILL: The three pieces fit together in a layer. Dr. Weaver's evidence provides the basis for the premise that there are some changes in the climate conditions. Mr. Root's evidence is a piece that looks at an appropriate way to view weather normalization going forward. And the key element of Mr. Root's evidence is that he identified a time period, the 20 years, and also a trending component. Those are the two pieces that we really took and used to do an evaluation, which is the third piece, the Union Gas component. 138 MR. PENNY: In general terms, what was involved in the Union Gas component? 139 MR. FOGWILL: The Union Gas component looks at not only the recommendation from WeatherBank, but also other weather methods that are used by gas utilities in the country. 140 MR. PENNY: All right. So when it came time for Union to do its analysis, what were the major factors or measures that you looked at in evaluating the various possibilities? 141 MR. FOGWILL: Well, we had five, but there were three main ones that we looked at. One was symmetry, the second one was accuracy, and the third one was stability. 142 MR. PENNY: So in this context of developing an appropriate weather methodology, can you explain what you mean by "symmetry, accuracy, and stability," and why those are important? 143 MR. FOGWILL: Well, symmetry is just a measure of how much bias there is in the analysis. If it's a completely symmetrical method, then you would expect the same amount of observations above and below the trend line. And in that case what we're looking at is trying to get something which is not bias towards either underrepresenting what the future weather will be or overrepresenting in terms of a higher number of heating degree days or a lower number of heating degree days. 144 The second measure which is accuracy, is a measure of how much error there is between the forecast estimate and what actually happens. And in that case what we want to do is try to find something that minimize that error, because the greater the error, the greater the possibility that there is going to be complications due to inappropriate planning. 145 MR. PENNY: Just briefly, how do you measure accuracy, or for that matter symmetry, over what time period and with what data set? 146 MR. FOGWILL: Well, the analysis we did was over an 18-year period and it used a data set from the Toronto Pearson Station. 147 MR. PENNY: And what about stability? What's the significance of stability? 148 MR. FOGWILL: The significance of stability is how the estimate will change, using whatever method, over time, from one year to the next. If a method is more stable, that means that your estimate from one year to the next will vary very little; if it's less stable, that estimate will vary widely. You want to look at something that has greater stability because if you have some significant changes from one year to the next, that could have some implications related to gas supply planning or the demand forecast through the demand forecast rates. 149 MR. PENNY: All right, thank you. 150 And you mentioned the forecasting a few times in your response. Is the purpose of the weather normalization methodology in Union's rate-making process to forecast what the weather will be in a particular year? 151 MR. FOGWILL: No, it's not. It's really trying to minimize the variability of weather over a significantly longer period than one year. 152 MR. PENNY: Thank you. 153 Thank you, Mr. Chairman. Those are all my questions in examination-in-chief. 154 MR. SOMMERVILLE: Mr. Brett. 155 CROSS-EXAMINATION BY MR. BRETT: 156 MR. BRETT: Thank you, Mr. Chairman. Good morning, Mr. Birchenough. Good morning, panel. 157 My name is Tom Brett. I represent the Ontario Association of School Business Officials. My first series of questions, Dr. Weaver, are for you, and they're mainly of an informational nature and I'll slip back and forth between the three documents that you have filed here. 158 The first point I wanted to confirm with you is set out at page 11 of your report. These are C, tab 4, appendix A. It's just a historical statement. You said over the twentieth century the increase in temperature has been about .6 degrees centigrade with an error range of plus or minus .2 degrees centigrade. Am I reading that right? 159 DR. WEAVER: Yes, that's the IPCC assessment value, .6, and that's a globally averaged number. So it's not applicable to any locality, it's averaged over the entire globe. 160 MR. BRETT: All right, thanks. And the base -- what is the base for that calculation? In other words, it's warming -- the increase in temperature is .6 degrees increasing from what? 161 DR. WEAVER: So what is done, if I could refer you to page 12, over the page, the figure at the top there, that's figure 3(a), what is done is an average is defined for reference purposes only as the average temperature globally averaged between the years 1961 to 1990; the numbers are then expressed as anomalies relative to the average, that is, departures from the 1961 to 1990 average, and the best-fit trend line is put through the data and that best-fit trend line will give you a .6 plus or minus 2. 162 MR. BRETT: And there's nothing -- in the first sentence below those two diagrams on page 12, you talk about the 1961 to 1990 mean. Along the vertical axis of the two graphs, you talk about the 1961 to 1990 average, I take it there's nothing in those two -- nothing to be made of that distinction? 163 DR. WEAVER: No, it's simply used as a reference. If I could draw an analogy, it would be like when you go to the doctor to see whether your child is above normal or below normal in terms of its height, the doctor will say your child is above normal or below normal, but he'll just be referring or she'll be referring to the average defined from some number. That doesn't tell you anything about how much above normal you'll be or what rate you're growing at, et cetera. It's purely for reference purposes. 164 MR. BRETT: Okay. But if I can take that and turn it just a bit, that -- is it also correct to say, then, that the -- that over the 100-year period, the 100-year historical period, I guess it's actually 40 years, in any event, you could say, using that reference, the '61 to '90 period as a reference, that the increase in temperature has been over that period .6. 165 DR. WEAVER: The increase is not -- it doesn't -- the increase is not dependent on what you use as a reference. The increase is .6 degrees. The reference -- what you use as a reference is irrelevant. It is .6 degree warming over the twentieth century. Whether you have decided to define the average as 1961 to 1990, 1971 to 2000, whatever, the trend is the same. 166 MR. BRETT: Okay. Now, then, you make the point, going back to page 11, that most of the warming occurred during two periods, 1910 to 1945, and 1976 to 2000, and you can see this trend on -- the pattern on figure 3(a) back on page 12 and you can see it again on the set of global maps at the top of page 13 where you break out the trends in four time periods; correct? 167 DR. WEAVER: That's correct, yes. 168 MR. BRETT: And the period -- what the graph on page 12 and the maps on page 13 show, it seems to me, is that from the period in between those two periods where most of the warming occurred, that is, for the period from roughly 1945 to 1976, you had a cooling, you had a cooling trend within this overall warming trend. 169 DR. WEAVER: Yes, that's correct. 170 MR. BRETT: So that it is fair to say that within the long-term warming trend, whether you look back historically or look forward, and I'll come to the forward period in a moment, that you can have consistent with that long-term gradual warming trend, you can have, for want of a better word, intermediate periods where the weather is cooling. 171 DR. WEAVER: This is not weather, this is climate. That's an important distinction because weather and climate are very different. Climate is, by definition, statistics of weather. So you can have -- so your question, then -- could you repeat the question then to me. 172 MR. BRETT: Well, you had mentioned to me that on the graph on page 12 and on the maps on page 13, there was, for the period of time between 1945 and 1976, a cooling trend displayed. 173 DR. WEAVER: Yes. 174 MR. BRETT: You're saying that's a cooling trend in climate? 175 DR. WEAVER: That's a climate cooling, yes, not weather. Climate and weather are different. But what -- where -- I mean what I tried to discuss later towards the end of this page 29, for example, is that there are many factors that cause climate to change. There are natural factors, such as changes in volcanic emissions and changes in sun's solar intensity, and there are not natural factors such as changes in the emission of sulphate aerosols. 176 Now, that particular trend, that cooling trend is most likely explained by the existence of increased volcanic activity thereby cooling the earth, together with the fact that, back in those days, there was very little control on industrial emissions of sulphate aerosols which act to cool. And since then, there's been large caps on the emission of sulphate aerosols which has reduced the cooling effect of anthropogenic activities or human activities. 177 So we are able to reduce -- reproduce that trend together with its intervening cooling trend with the climate models that are used to make future projections, as I showed on page 29, figure 14. 178 MR. BRETT: All right. Well, I'm going to come to the future projections in just a moment. But I take it that in that area we just discussed, in that period of time that we just discussed where there was a cooling trend in climate, that does translate into, one way or another, it translates into more degree days as opposed to fewer during that period. 179 DR. WEAVER: I would say -- I mean in a globally average sense, yes. 180 MR. BRETT: I'm speaking -- no, I'm speaking globally first here. I'll go to the regional thing later. I want to focus first on global -- 181 MR. PENNY: I'm not sure what the point of comparing heating degree day to the globe is, Mr. Chairman. 182 MR. BRETT: Why don't you let me finish my cross-examination, Mr. Penny. 183 Mr. Chairman, I don't intend to telegraph to Mr. Penny where I'm going. I want to start, with your forbearance, at the global level, I then want to move to the regional level in Ontario. I have a game plan here. I'd like to follow it. 184 MR. SOMMERVILLE: Proceed, Mr. Brett. 185 MR. BRETT: Thank you. 186 All right. Then if I could refer you for a moment to page 5 of the technical summary of the IPCC report. I just want to read you a passage. I think this is just to clarify something which we already talked about. But in the second paragraph there, page 5 of the technical report, halfway down the paragraph: 187 "The recent regional patterns of temperature change have been shown to be related, in part, to various phases of atmospheric/oceanic oscillations, such as the North Atlantic/Arctic Oscillation and possibly the Pacific Decadal Oscillation. Therefore, regional temperature trends over a few decades can be strongly influenced by regional variability in the climate system and can depart appreciably from a global average. The 1910 --" well, I don't need the rest of it. 188 I think that's emphasizing what you just said, that you can have regional distinctions, quite notable, within the general global average or global trend. 189 DR. WEAVER: Absolutely. 190 MR. BRETT: Okay. Now, in that same -- in that same section, and I'll come back to this a bit later, but they say in the first sentence I read you: 191 "The regional patterns of temperature change have been shown to be related in part to various phases of ..." et cetera, on down to "and possibly the Pacific Decadal Oscillation." 192 Could you explain briefly what that -- or explain as you like what that is. 193 DR. WEAVER: Well, first of all, this report was -- this is a document that was written in 2000 which is very shortly after this issue of the Pacific Decadal Oscillation came about. I would argue that the Pacific Decadal Oscillation is not an oscillation. What it is is an index which shows -- which represents the -- basically the surface temperatures of the North Pacific ocean. It gives an indication of the surface temperatures of the North Pacific ocean and it's evolution with time. Now, where that came about, it came about by a fellow named Nate Mantua, who was at the University of Washington, who was looking at salmon migration and other fish species and trying to link it in with change patterns of sea surface temperatures. 194 It is -- I mean, I don't know very many people who would actually say it's an oscillation because you have a hundred and some odd years of data and this is supposed to have a 50-year period, you simply cannot attribute an oscillation to something where you have one period of data records. So I personally don't believe it is an oscillation, I just believe it represents periods in time where the surface temperatures in the north are warmer or colder than the longer period average. 195 MR. BRETT: Okay, I'll come back to this in a bit. But given what you've said, could it still influence regional temperatures? 196 DR. WEAVER: The phenomena in my interpretation of the Pacific decadal oscillation is it in itself does nothing. But, yes, the surface temperatures of the North Pacific can influence regional temperatures, and those temperatures are just expressed through this tool which is called the Pacific decadal oscillation. What I mean by that is PDO is not a physical phenomenon. The sea-surface temperatures are physical phenomena, and it is the sea-surface temperatures in the Pacific that can affect regional climate. 197 MR. BRETT: Thank you. I think that you've made a -- there are a number of other points that you've made about the particularity of temperature changes in your paper. You talk about nighttime temperatures, nighttime lows increasing more than maximum temperatures characteristically. You talk of warming over land increasing more than warming over ocean characteristically. You talk about the northern regions where there's snow cover incurring greater warming than other regions because of the impact of the lack of the snow cover, and so on. And I don't think I'll take you through the details of that. I'd like to go now to where you move into the period of looking ahead. 198 In the policy document, in this policy summary, and you make this statement in your paper as well, "over the next hundred years, globally average surface temperatures are projected to increase" and you give a range of 1.4 to 5.8 degrees C; right? 199 Now, that is a wide range of about 300 percent. You mention in your paper at page 11, you mention indirectly a best guess that has been made by one scientist of something in the order of 2 degrees Centigrade. You quote a paper by Mr. Wigley. I guess that's not important to me. What is important to me is there is a best guess in the scientific community about what that number is? 200 DR. WEAVER: The word "best guess" did not come from Tom Wigley alone. Back in the earlier IPCC assessments there was a value judgements, that is, best estimates as to what the most likely scenario would be and the 2 degrees was the best estimate. 201 In the third assessment report they moved away from this best estimate because, for reasons I'm not -- well, I guess because really it's a best estimate and there's uncertainty here, rather than trying to look at a range. 202 Now, the range 1.5 to 5.8 comes about from many scenarios as to future emissions of various gases, including sulphate aerosols, as well as many models around the world running many times. It's a wide range because it incorporates uncertainty through different models as well as uncertainty within the scenarios in the future. 203 However, the best estimate, if you're asking me personally, my personal feeling is that 2 degrees still stands the test of time as being the best estimate. 204 MR. BRETT: Okay, thank you, that's what I was after. 205 And, again, we probably will have a replay of our earlier conversation, but we're looking ahead and we're saying, over the next hundred years, the best estimate is that the climate is going to change -- it's going to warm by 2 degrees C. The base, again I say, from where? And I guess there are a number of base periods that are mentioned in your report. And at page 41, for example, it looks like you're saying that the scientific community is coming round to using the period 1971 to 2000 as a new sort of "normal period," if you like. 206 Now, can that be viewed as a -- can you say that if we were to take -- let me put it in sort of simple terms which help me at least. If Ontario, if the relevant part of Ontario has an average winter temperature over the past 30 years of X degrees Centigrade, and leaving aside for the moment your comments about whether or not the warming experience in Ontario would be average, higher than average or much higher than average -- I'll come back to that because I have some questions on that point -- can you say, though, that effectively the 2 degrees Centigrade can be compared with what I've said is the average Ontario temperature in degrees centigrade? 207 In other words, if the average Ontario winter temperature were, for sake of argument, 15, then we would be saying that if you applied the global pattern to Ontario, you would be saying that the temperature -- the winter temperature in Ontario is likely to warm by 2 degrees, in percentage terms, 2 over 15 times a hundred in the next 110 years. Can you say that? Is that consistent with -- 208 DR. WEAVER: Well, first of all, I will not let the 2 degrees stand because I have said quite clearly in my report that you cannot take the global number and apply it to Ontario. 209 MR. BRETT: Yes, and I acknowledge that. 210 DR. WEAVER: But it must be amplified, and I think it's important that it be there for the record, that that would be amplified in Ontario. So it would be amplified. 211 MR. BRETT: Okay. 212 DR. WEAVER: So the 2 degree warming, where these numbers would come, is you would take the 1961 to 1990 average and you would add 2 degrees to that. So that would be your -- and that would be globally, and that would include summer and winter. 213 Now, as I've mentioned, the warming is amplified in the winter and it's amplified in higher latitudes and it's amplified over land. So you can't take it into the winter in Ontario and say it's 2 degrees. It's larger because it's winter, larger, because it's Ontario, which is high latitudes, and larger because it's land -- 214 MR. BRETT: In Ontario I think they consider winter for degree-day purposes from September to May, so that you -- you know, you -- well, I'll come back to Ontario in a moment. I didn't mean to not give you the opportunity to speak about the regional differentiation and you've seized on it quite properly. But the fact of the matter is, subject to those qualifications, it's the 1961 to 1990 versus the 2 degrees. 215 DR. WEAVER: Well, I mean, that would be what you would use, yes. Some people have done 2 degrees relative to -- see, the reason -- what's happening here is, why I need some further expansions here is that what it's done is it's 2 degrees warming based on a model simulation of a present-day climate. So where the 2 degrees comes from is the models have the present-day climate and then they look -- and to a projected climate, and it's 2 degrees warmer than the model present-day climate. 216 Now, in terms of translating that into the observational record, you would then do it from 1961 to 1990, add 2 degrees. 217 MR. BRETT: I see. But you're also saying -- are you saying that the 2-degree estimate still takes into account the sort of -- as well as you can from point in view, the dynamics of the system, going forward? In other words, you're saying the base is moving constantly, in a sense? You're saying the models apply to today's weather, and if you apply the model to today's weather you come out with a number, as you say, an estimate globally of 2. You seem to be suggesting if you apply that model to -- I think all you're really saying there, if I'm not mistaken, if you were to apply that model in 2010, you'd be applying it to a different set of climate? 218 DR. WEAVER: No, what I'm saying is that the best projection as to what would happen is to draw a straight trend line from 0 to 2 and every year just add -- if you're looking at a hundred years, add 2 divided by 100 and add the warming. 219 MR. BRETT: Well, that's what I thought might be the result, and I appreciate your confirming on that. So it's just a straight arithmetic division. All right. 220 Now, if I could turn to page -- to figure 12, this would be I believe page 25 of your study. It's the one with the maps on it. Figure -- the middle of page 25, the global map. I want to talk a little bit about Ontario here. You say that figure 12 shows a region called the GRL region. 221 DR. WEAVER: Yes. 222 MR. BRETT: And that -- if I look at that on the map, the GRL region is the region in which you have said has much greater than average, at least part of it has much greater than average warming and therefore you apply an amplify occasion factor of 40 percent, the degree of warming over any surface which fits into the GRL box; right? 223 DR. WEAVER: For the winter seasons for both scenarios there is a much greater than average warming in that box there, yes. That isn't me, this is the IPCC assessment. 224 MR. BRETT: I understand. 225 DR. WEAVER: Okay. 226 MR. BRETT: But my -- but if you look at that map, most of Ontario, the part of Ontario where the people live and where the gas is used would fall into -- would not fall into the GRL, it would fall into the CNA or the ENA which is a category of greater-than-average warming but not much greater-than-average warming. 227 DR. WEAVER: That's correct. 228 MR. BRETT: The greater than average is James Bay and Northern Ontario, where a few people live. 229 DR. WEAVER: Yes, that's correct. 230 MR. BRETT: So my question to you is: Given that fact, what is the factor that would be applied? Is there a factor like a 20-percent factor or a 15, or is there any number comparable to the 40 percent? 231 DR. WEAVER: See, this is -- the issue of regional forecasts or regional projections in climate is one where it's entering its infancy in terms of the science. What we would say is that it's greater than average, but we would be loathe to give you an amplification factor simply because it's difficult to actually quantify how someone would actually obtain such an amplification factor. The uncertainty in that factor would be difficult. All I can say is would be greater than average. 232 MR. BRETT: Thank you. I appreciate that. Well, then, if -- if we were to try and extrapolate -- now that we're sort of focused in on Ontario and we have a range -- we know that in Ontario, over land, in the winter, we're going to have -- and because of the altitude, latitude, we're going to have a higher factor -- a higher warming over a hundred years than the global figure of point -- of 2. So we would have to -- what it would be, I suppose, is you would say, given the state of the science and so on, it's very difficult but -- I mean it could be 3, I suppose, theoretically, although that's probably high. It may be somewhere between 2 and 3. 233 DR. WEAVER: I would say 3 would not be high. It would probably be a good estimate. I mean, you can -- I'm looking here at figure 11 right on page 24, right beside it. You'll see why it's difficult for me to give you a number is because this is the result from one model, the Canadian Climate Centre model under one scenario, that is, under one assumption into future technology paths, future energy use, future population growth; and in that particular case you see that it's kind of -- you see the amplification there in Ontario and you see it's dark and you see it's, you know, between 3 and 5 degrees in that particular situation. 234 Now, you could find another model which has slightly different physics, slightly different feedbacks, may give you a -- a very similar pattern but slightly different magnitude. And this same model under a different scenario would give you a different number. 235 MR. BRETT: I understand. So if we took 3, for example, as the number and we did some of those -- a little bit of the arithmetic we were just sort of hinting at earlier, you would take .03 -- let's assume that -- let's assume, and these are purely my assumptions, that we have 4,000 degree days in Ontario as our average number of degree days over the base period that we spoke about. And let's assume that we're going to have a 3-degree cooling over a hundred years. You divide that out and you get -- that's effectively .03 -- let me be a little careful here. That's about, I think -- well, it's about .15 percent, .15 percent per year in percentage terms. 236 Would you take that, subject to check? 237 DR. WEAVER: I would take your calculation subject to check. However, what I would be doing is I wouldn't be calculating an estimated trend from a model, I'd be using the observed data to actually estimate that trend, because the model -- because you can't pick the three -- you're picking 3 degrees. 238 So my point is you have to be -- by doing that, you're going to end up telling me that some number is .15 percent and I suspect I'm going to be told that the -- that it's higher or lower than the number you can get out from the forecast. But that's going to presuppose that you would use a model with one scenario under one projection to predict a trend. We're not or they're not, you're not. The idea is you should use the observed data from which to get the trend and then that observed data will narrow down which of the scenarios is actually better. You're not going to get the trend out of the models, you're going to get it from the data. 239 MR. BRETT: I think you're ahead of me. All I really wanted to use -- all I wanted to use the .3 for is to get a very general idea of how long it takes over time -- I want an illustrative set of numbers here. I quite realize that it's not going to be linear, that it's going to be this sort of movement and, as you've said, over periods of time, warming tends to take place over certain periods of time and then some cooling and some more warming and so on. All I really want to do is get my -- for my own purposes, I want to say -- you can put the qualification on it. I'm not trying to get a trend here, I'm not trying to get a trend to counter Union's trend. 240 I'm trying to get a rough measure if we're going to get that pace. My questions are all about pace. They're all about pace of warming compared to what's implicit in Union's proposals about pace of warming, all right? 241 And under your analysis, you're talking about 3 degrees C in 110 years. That represents a pace of warming of approximately 6 degree days for a year for a hundred years. 242 DR. WEAVER: Well, I haven't done that calculation. 243 MR. BRETT: No, you haven't, but I've done a rough calculation, using the base of 4,000 and going out 100 years. That's all I'm trying to do. I'm not trying to get more at this stage, any more rarified than that. 244 It seems to me to be fair, you put some general propositions out there about the degree of -- about what science knows about the rate at which the world and Ontario is warming. In order for us to make anything of that and relate that to what Union's doing, we need to translate that into what that means, what that would mean over 110-year period in terms of the rate at which degree days would change. 245 DR. WEAVER: Yes, but I have not said in my report that in Ontario the number of 3 degrees is going to be the number. 246 MR. BRETT: No, you haven't. 247 DR. WEAVER: I have said above average relative to a global mean with a best estimate of 2. Now, because -- because a science would not allow me to give a better estimate than those statements. But what I would say is if you're going to estimate a trend, you're going to estimate a trend from the data that exists, because that is where you're seeing the trend. And that will allow you to suggest which of the possible scenarios is the most likely evolution of that. And so it's -- so in the questions you're -- 248 MR. BRETT: Let me put it another way. 249 MR. PENNY: Hang on, hang on. Sorry. Mr. Brett made great umbrage at my interjection, and perhaps Mr. Brett can give the witness the courtesy and allow the witness, and maybe he can allow the witness. 250 May I just interject. Having nothing to do with Mr. Brett, Mr. Chairman, but I notice the report is struggling from time to time. So can I remind Dr. Weaver to try and slow down. 251 DR. WEAVER: Slow down, thanks. I've lost my train now with double interjections now of where I was going. 252 What I was trying to say is that when calculating what trend exists, one must use the existing data. In making a projection to the future, all that one can say for this region is that it will be amplified relative to the global mean. I have not assigned a number, I cannot assign a number. I can just say it will be amplified relative to the global mean, and that what actually transpires will result from the scenario which evolves in terms of the emissions that are globally done in terms of the various things that affect climate. 253 And I can't give any more than that and I can't verify your number of degree-heating days because I haven't seen the calculation, I haven't done the calculation, and I wouldn't do the calculation because it would be an assumption that I am validating a number. 254 What I can say is that the trend exists today. That trend is discussed and borne out in the record. That trend is consistent with what models said should have happened, and what models say will happen is a continuation of that trend, although it's not quite clear, the regional ramifications, a hundred years from now. 255 MR. BRETT: Thank you. And I take it you would also agree that you could have -- I think you did agree with this earlier, that you could have within that long-term secular trend, you could have 10-year or even 20-year subtrends, if you like, in the reverse direction. 256 DR. WEAVER: I mean, climate has variability associated with it. Climate has El Ninos which affect climate. Sure, you will get periods when it's above normal, above the trend, and periods where it's below the trend. It may be above the trend for three years, it may be below the trend for three years. The bottom line is what one is trying to do when estimating a trend is make it such that it is equally likely to be above as it is to be below. The trend, by definition, is trying to make it equally likely it will be above or below on the long-term average. 257 MR. BRETT: At the beginning of your evidence, Dr. Weaver, where you outline what you were asked to provide at page 2, you say you were asked to provide an assessment as to whether there is a noticeable and sustainable change in climate condition that could indicate if future temperatures in Ontario would differ from historical data; right? 258 DR. WEAVER: That's what I've said here, yes. Correct. 259 MR. BRETT: In your conclusion, you go on to say -- sorry, let me just get the relevant page here. You were not asked at that time to pronounce on whether or not -- whether Union's particular proposal would be the best, fairest way of the various alternatives that are out there to reflect this broad trend. I take it that's fair? 260 DR. WEAVER: Yes. I had not seen until very recently the actual Union proposal as to what they were going to do. 261 MR. BRETT: And then you conclude at the very end of your report, "I firmly believe..." And this is at page -- 262 DR. WEAVER: 41. 263 MR. BRETT: -- 41. 264 "...that any natural gas rate setting which does not account for the observed and projected trends in Ontario's temperature, or equivalently, degree days, would not accurately reflect the known changes that have occurred and will continue to occur in Ontario's climate." 265 Now, I'd take it that you would agree with me that given that there is a long-term warming trend, or in the event there's a gradual long-term warming trend, which you and your colleagues believe there is, any number of these methods that are being discussed in the Union proposal, any of the rolling averages - the 30-year rolling average, the 20-year rolling average, the 10-year rolling average, the 20-year trend, a longer term trend - any of these methods, the Enbridge method, any of these methods would reflect, to varying degrees, would reflect the secular warming trend. 266 DR. WEAVER: The key statement in there is "varying degrees," so some would reflect it very well, some would reflect it extremely poorly. A 30-year average would reflect that extremely poorly. Something that includes a trend would reflect it very well, and in fact, best. 267 MR. BRETT: If there were, however, a countertrend, a cooling trend for the next 20 years, as the authors of this report, the Smith Barney report that I left with Mr. Reghelini, if that were to occur, though, the 20-year trend would not be necessarily the best, would not reflect what actually happens the best. 268 DR. WEAVER: Could you, first of all, point to me where in the Smith Barney report it refers to a 20-year cooling period? 269 MR. BRETT: Let me get the Smith Barney report as I was going to come to that next anyway. 270 MR. SOMMERVILLE: The Board does not have -- 271 MR. BRETT: No, I was just about to get it out for you, Mr. Chairman. I only have a certain number of copies. 272 MR. PENNY: Mr. Chairman, I'll make this comment about this particular document so that I won't have to make this particular little speech again each time we get these materials from other intervenors, and my cautionary comment is as follows: 273 I think you've heard me on this before in past cases. It's, of course, open to cross-examiners to put whatever they like to witnesses, as long as it's relevant, but the mere fact of filing material, of course, doesn't make it evidence. There will be, you'll find in the course of the next few days, a good deal of articles written by people that I gather intervenors want to put to the witnesses. And I simply take the position that the filing -- while the document may be filed by identification purposes doesn't make it evidence. 274 If the cross-examiners want to put statements from these articles to the witnesses, that's fair, but unless they're adopted by or agreed with by the witnesses, these are not evidence. And of course, the authors of these -- this case on this issue is frankly notable by the fact that although we've had this evidence out for more than two years, there is no evidence from any intervenor on this issue. And of course, the authors of these articles are not coming to testify, and therefore, we're not in a position to cross-examine. 275 MR. BRETT: I understand that, sir. I'm just using this as a tool to cross-examine, to get at this question of the impact of this Pacific oscillation. 276 MR. SOMMERVILLE: Mr. Brett. 277 MR. BRETT: I'll be very brief. I was actually -- you've had a chance to read this, have you? 278 DR. WEAVER: I have read this document, yes. 279 MR. SOMMERVILLE: Dr. Weaver and Mr. Brett, I guess it would be appropriate, and not to -- this is in no way a contradiction of the point of view that you will want to argue perhaps later, Mr. Penny, but I think it is appropriate that the document be given an exhibit number. 280 MR. PENNY: Yes. I wasn't taking any exception to that. 281 MR. SOMMERVILLE: So it can be appropriately filed in the case. 282 You're aware of the position Mr. Penny takes in this document, and I see the other intervenors here and I expect that position to extend to all like instances. 283 MR. MORAN: Mr. Chair, on that basis, this will become Exhibit M.2.1, a document entitled "Futures Research Weather Report, October 2001, the 2001-2002 Winter Outlook." 284 EXHIBIT NO. M.2.1: DOCUMENT ENTITLED "FUTURES RESEARCH WEATHER REPORT, OCTOBER 2001, 2001-2002 WINTER OUTLOOK" 285 MR. BRETT: Can you tell me, Dr. Weaver, have you seen a number of these documents? Is this what you would describe as a typical sort of one-year projection by working meteorologists in the industry? 286 DR. WEAVER: No, I wouldn't. This is obviously -- I have not heard of these names. This is not a refereed report -- article. It looks like a research document done for a particular brokerage. And I would like to say off the top that I don't -- this is speculation, largely speculation, and I wouldn't call it a scientific document, I will I would call it a consulting report. 287 MR. BRETT: I don't make any particular claim to that, Doctor. This is, as I say, obviously a meteorologist working for a trading firm. All of these trading firms has these people. Anyone who is trading commodities has these kind of people who do these kind of analyses for them on an ongoing basis and they try their best to get a handle on the next six months. I think that's all they purport to do. 288 But in any event, he talks in here about the Pacific Decadal Oscillation, which is also something that the third technical summary to the IPCC spoke of, although I've talked about its influences being possible on the grounds that it was, I guess, very new, as you said. 289 Nonetheless, at page 7 of the report, on the left-hand side, the middle paragraph, he is talking about the previous winter, the '99, 2000-01 winter. 290 "In looking back at last winter, we now feel that the PDO was the key variable as to why last winter was so different than the winters of the late-1990s." 291 I take it he means it was much colder than the late '90s. 292 "In other words, we view the striking shift in winter weather last year as being more due to the PDO than to any other variable. Since the PDO theory is so new, we will first define and review it." And he goes on to discuss it. 293 Is his description of the PDO in the next couple of pages, while it might not pass muster in a scientific journal, is it roughly accurate? 294 DR. WEAVER: I don't have a problem with his description of the way the index, the PDO index is defined. There is no theory of a PDO. The PDO doesn't have a theory, it is simply an index. 295 MR. BRETT: It's a statement of facts. 296 DR. WEAVER: It's an index. It's saying we're going to characterize the North Pacific temperatures by this index. There's no theory as to its evolution. So I took exception in my reading of this to the words, "The theory of the PDO," because such a theory doesn't exist, to my knowledge. 297 MR. BRETT: Well, they have labeled it an index. I guess it was me that -- 298 DR. WEAVER: No, they labeled it a theory. 299 MR. BRETT: But they also on page 8 -- page 9, they have a graph at the top of page 9 where they say, "Pacific Decadal Oscillation Index," and I was just going to ask, it doesn't look like it's been measured for any time prior to 1930. But anyway, you can see below, in the next diagram below that on page 9, they've tried to sort of link average winter temperatures and this is mainly I guess in the United States and particularly in the central U.S. with this index, with the movement of this index. 300 DR. WEAVER: Yes. But what they've done here -- first of all, the top figure, it's an NCEP, UKMO, United Kingdom Meteorological Office, NCEP, so that's obviously not their figure. 301 MR. BRETT: They've taken this -- 302 DR. WEAVER: From another cite, and that's where the word index comes from. I don't know where the bottom figure came from. All this is saying is that, you know, when you have cold -- there's no causality in here. It's saying there are two things. That on average, when the Pacific temperatures are such and such, when they're warm, you probably find it affects climate over North America the same way. I could argue that I could explain this solely through the changes in radiative forcing that have occurred over the twentieth century, that is, in periods when there is enhanced volcanic activities, you would expect to have periods of cooler temperatures in the North Pacific and cooler temperatures over land. 303 There's no causality in drawing these two things and say, look, they go up and down here, and this is part of the problem I have with this theory that they have discussed. 304 MR. BRETT: So you say these cooling subcycle within a warming cycle, you would explain these more in terms of either policy, that is, impact of rules and regulations of what people are allowed to emit, on the one hand, and natural causes such as volcanic -- 305 DR. WEAVER: Sorry, I'm losing the train here. 306 MR. BRETT: You made a point earlier. You were going to explain -- you were explaining about half an hour ago why the period between 1945 and 1970 was -- showed a cooling trend, and I thought you made reference to the fact that in those years, there was no regulation by government of the release of aerosols into the atmosphere. The more aerosols going into the atmosphere, I take other things being equal from this scientific literature, the cooler it gets. 307 DR. WEAVER: All things being equal, yes. Of sulphate aerosols, yes. 308 MR. BRETT: Right. That was my point. Okay. And then I guess to page 10, and I take it that, the left-hand column of page 10, the top paragraph, this really is getting back to your question of a few minutes ago to me. 309 "This chart points out the striking similarity of the winter rankings during these positive and negative PDO cycles. Actually, this study tells us more about long-term trends than it does about any particular winter. Because of that, it provides us with a possible answer to the question, Will the next ten years be like the last ten? At this point, based on the PDO theory alone, the chances of the warming trend of the 1990s continuing and strengthening during the current decade," and he's referring there to the decade there of 2000 to 2010, "are remote." 310 DR. WEAVER: Well, at this stage, I'd like to introduce the validation of this forecast. It's very important because this is a 2001-2002 winter outlook. Any time you do such a thing, the first thing you should do is validate the response. So they're basically saying, "Based solely on the PDO," and then they further say in here: 311 "Based on the variable, we have analyzed the similar years that we observed and the projections we have assumed, we feel that the winter of 2001-2002 will rank near or in the coldest winters historically. The odds do not point to an exceptionally cold winter such as the one of the top ten coldest, but that is not totally out of the question. One outcome that we believe is out of the question is the possibility of an extremely warm winter." 312 MR. BRETT: Which winter are we talking about now? 313 DR. WEAVER: That's the forecast. This is actually a forecast for the winter 2001-2002. That was the purpose of this report. 314 MR. BRETT: Right, I understand. 315 DR. WEAVER: So I have the -- you can actually look at the NAAO results for 2001 and 2002. It was the ninth warmest year on record, and, in fact, it was the warmest year ever in the northeast United States which is the primary forecast year. In fact, this forecast was a hundred percent wrong. It forecast very cold conditions. It turned out it was extremely warm. So this forecast turned out to be completely wrong. Why, I would argue, because the theory on which it was built is flawed. 316 In addition, the Pacific Decadal Oscillation that they make a big deal about switched sign a few years later. So therefore it is no longer in this so-called negative phase, it's switched back. And the reason is because there's no predictive skill associated with this so-called Pacific Decadal Oscillation, which is really an index. 317 MR. BRETT: Well, fine, I won't get into an argument with you. I'll look at what you've said when I see the transcript. I'm looking at, in front of me Exhibit J.26-35 which says that in the winter of 2002, in Ontario, there were 3,631 degree days; in the winter of 2003, there were 4,033 degree days predicted. I don't know what numbers you're looking at. 318 But the second point that I would like to make is that in their analysis, in their analysis, those authors did point out that the -- their focus of attention was, for the most part, the central part of the United States. They admitted all along the northeast temperatures were somewhat different. They are people that are in the commodity and the cattle forecasting business, and their interest is particularly in the heartland temperatures. Let me just leave it at that because I think we're going to get into the realm of argument here in a big way, and I think that's not going to be too helpful. 319 Let me just go on to you, Mr. Fogwill, for a moment. 320 Mr. Chairman, I was going to shift gears now. I have some questions for Mr. Fogwill. Do you want me to keep going? Is this time to take a break? 321 MR. SOMMERVILLE: If this is a convenient time, we can adjourn now until 11:10. 322 MR. BRETT: All right. 323 MR. WARREN: Mr. Chairman, just before you go, I hasn't appreciated that this panel was going to be dealing with the methodology issues exclusively. My questions with respect to weather are more mundane dealing with the issues of rate impact and how it's applied. And with your leave, I will excuse myself and return, I presume, tomorrow, based on the scheduled planning. 324 Thank you, sir. 325 MR. SOMMERVILLE: Mr. Warren. 326 We'll stand adjourned until 11:10. 327 --- Recess taken at 10:54 a.m. 328 --- On resuming at 11:20 a.m. 329 MR. SOMMERVILLE: Thank you. 330 Mr. Brett. 331 MR. BRETT: Thank you, Mr. Chairman, Mr. Birchenough. 332 I have a few questions for you, Mr. Fogwill. Just briefly, as I understand it, the existing method that Union uses to set rates, weather normalization and rate setting purposes, is the 30-year moving average? 333 MR. ROOT: That's correct. 334 MR. BRETT: And to do that, as I understand it, you make a different calculation each year, but you drop off the first year and add another year to the end of the 30-year period and take your average. It's a very simple, straightforward -- 335 MR. FOGWILL: That's correct. 336 MR. BRETT: And what you're moving toward, what you will move to, if you get the Board's approval here, is a 20-year trend where you will -- you've described how you would do this at page 2 of your evidence. And I don't need to get into the detail of it at all. I just really want to flag what you're doing. You're taking the last 20 years and you're effectively creating an equation by regression analysis techniques. You're building an equation that displays a flat line and that line trends and predicts next year's weather? 337 MR. FOGWILL: It doesn't predict next year's weather. It is a regression analysis that is used to provide an estimate for a reference, if you will, of what we use for planning purposes. 338 MR. BRETT: So, in other words, a number of degree days is likely to occur. 339 MR. FOGWILL: No, it's an estimate of the number of degree days that is the reference point that we use for planning purposes. So it doesn't necessarily have a linkage to what is actually going to occur in that year. 340 MR. BRETT: Well, I'm a little confused. I thought one of the purposes of the change was to effectively -- well, let me read you something. Perhaps this is the best way to do it. 341 If we look at the evidence that you filed two years ago, at page 4 is the main evidence, if you like, at paragraph 14, first sentence. 342 "The weather normalization method that has been used to set rates to date consistently --" wait a minute, that's not what I want. That's... 343 I'm sorry, I go back to page 1 of C.1, tab 4, page 1 of 7, the evidence of yourself and Mr. Gardiner. I apologize for that. The second paragraph on that page: 344 "The primary objective of an acceptable weather normalization method is to set a weather normal level that will best reflect what future weather is typically expected to be." 345 Now, I think you're using this to -- as I understand it, you're using it for rate-making purposes. You're using it for some other purposes already, but let's leave that to one side. What you're seeking to do is to get approval to use it to set rates, as part of the rate-making process. 346 MR. FOGWILL: It's a planning assumption and we're using it to -- we use the weather normal method to determine what the demand forecast is, and through that it affects gas supply planning and it will affect rates. 347 MR. BRETT: Right. And you apply it to that part of the gas supply -- to the gas demand that is the general use category; correct? 348 MR. FOGWILL: That's correct. 349 MR. BRETT: You don't apply it to the contract rate classes. 350 MR. FOGWILL: That's correct. 351 MR. BRETT: Okay. So the small rate classes. 352 Getting back to where I started, you then -- but isn't what you -- in order to do that, if that's what you're seeking to do, you're seeking to try and get the most accurate possible indicator of what next year's weather, next year's degree days will be, are you not, in order to be able to estimate as best you can the impact of any kind of a trend or a change, a likely change in degree days from the current year? That's what you're trying to do; is that not the case? I mean, you're trying to assess what the likely demand for gas is going to be in the general rate class in the subsequent year, the test year. And one of the factors that affects the demand for gas in the test year, as I understand it, is the degree days in the test year. I thought I was asking a very straightforward question. 353 What you're seeking to do here is to get the best method to predict what the degree days in the test year will be. That's all I'm asking, really. 354 MR. FOGWILL: Well, no, it's not quite accurate. What we're trying to do is use a method that, over a period of time significantly larger than one year, we will have a method that's symmetrical, accurate, and stable. So it doesn't necessarily mean that, in any given year, we're going to get an estimate that's going to be close to what Mr. Root, for example, might forecast for that year. 355 MR. BRETT: So in other words, that sentence I just read to you on the first page of your evidence, really when it says "that will best reflect what future weather is typically expected to be," it doesn't mean future weather in the next year, it means future weather over some undefined future time; is that correct? The next 10 years, the next 20? 356 MR. FOGWILL: If you look at the sentence following that, on line 15 it says, "Union and customers will then be kept neutral with respect to weather in the long term," and the long term here, we're looking at probably in the five- to seven-year period. 357 MR. BRETT: Five to seven years. So in other words, your hierarchy of objectives for doing this is to ensure that over a five- to seven-year period, you're as likely to overforecast as underforecast, and it doesn't really matter whether you're wildly inaccurate in either direction so long as you balance out over that five-year period, your objectives are satisfied; is that what you're saying? 358 MR. FOGWILL: No. If you look at how -- 359 MR. BRETT: Sorry, go ahead. 360 MR. FOGWILL: If we look at how we've assessed the methods, that balancing is the symmetry component, so how many estimates are above or below the observation. But there's also the accuracy component, that is, how close they are between the estimate and the actual heating degree days that show up, as well as the stability of the measure, or the estimate year over year. So it's not that if it's a symmetrical method everything else is not considered. 361 MR. BRETT: But accuracy -- you're saying, then, that accuracy is not your foremost consideration. Your foremost consideration is that you essentially remove the weather risk from Union, that you get a balanced -- that you get a balanced -- that you're as likely to be overforecasting as underforecasting. Accuracy is a secondary consideration. 362 MR. FOGWILL: Accuracy is a secondary consideration to the symmetry for the examination, yes. 363 MR. BRETT: I'll ask you to turn up J.26-35, please. It's an IR response that just cites current weather data going back to 1971. That's J.26-35. You've found that all right, Mr. Fogwill? 364 MR. FOGWILL: Yes. 365 MR. BRETT: Just before I ask you about that, if you just can also have handy the -- going back to your evidence, your exhibit -- the evidence you filed two years ago, the main evidence, what I started to ask you about a minute ago, page 4, page 4 of 31. If you could have your finger on that at the same time. 366 If you look at paragraph 14 of that page, you start off by saying: 367 "The weather normalization method that has been used to set rates to day consistently overestimates the heating demand by customers by 7.6 percent in a typical year." 368 Now, I'd take it you'd agree with me, Mr. Fogwill, that that statement, as it reads, you need to qualify that statement to make that accurate. What you're really talking about is weather since 1985, correct, which is the piece of time that you're using for your analysis. 369 MR. FOGWILL: That's correct, the 7.6 relates from the period of '85 to 2000. 370 MR. BRETT: Now, looking over at 26-35, if you would. If you go back and look at the years 1971 through 1984, the previous decade or so, would you agree with me or would you take, subject to check, that in that period of time, the 30-year average forecast has forecast -- has actually underforecast in each and every year except for 1973 and 1975. In other words, there was a pattern of systematic underforecasting in that period -- that earlier period; is that fair? 371 MR. FOGWILL: Which period are you talking about? 372 MR. BRETT: The period 1971 to 1983 -- 1984, sorry. 373 MR. FOGWILL: If you look at -- 374 MR. BRETT: Sorry, what -- yeah, that's right. What you're getting is -- you're getting forecast. Let me make sure I've got this right. You're getting forecasts that are under the actual experience, except for those two years. In other words, it was consistently or, for the most part, colder than forecast as opposed to -- my point being, to be compared with this recent period that you've used where it's been mainly warmer. 375 MR. FOGWILL: If I look at that interrogatory response, there's actually a couple of other years where it was warmer. 376 MR. BRETT: All right. And those were? 377 MR. FOGWILL: Looks like 1983, 1984. 378 MR. BRETT: All right. Now, if you take the -- will you take also, subject to check, that if you took the cumulative amount of underforecasting over that period, you've done an exercise that you just -- we just spoke about where you've taken a cumulative amount of overcasting for the period '85 to 2000, divided it out and come up with 7.6 percent. 379 If you take it, subject to check, if you did a similar exercise -- you know, you don't need to rush and do this. If I'm wrong, you can tell me that tomorrow. My quick calculation was there's an underforecast in the average of about 3.5 percent over that period. 380 MR. FOGWILL: I'd have to check that before I could actually comment on it. 381 MR. BRETT: Well, you check that and get back to me if it's either way. 382 Now, then, maybe you could give me an undertaking -- could I have an undertaking please, to get that verified, to see what the -- what I'd be looking for, Mr. Chairman, would be -- Mr. Fogwill, would be the average underforecasting that results from applying the 30-year average to the 30-year period preceding each of the years 1971 through 1983 inclusive. 383 MR. SOMMERVILLE: This is taken from page 2 of 2 of Exhibit J.26-35? 384 MR. BRETT: Correct. 385 MR. SOMMERVILLE: Mr. Brett, insofar as this is a simple arithmetic endeavour, I'm not going to ask the witness to produce that. 386 MR. BRETT: All right. 387 MR. SOMMERVILLE: That's something that's available to you to produce. 388 MR. BRETT: Okay, fair enough. Fair enough, Mr. Chair. I can do that. 389 MR. SOMMERVILLE: Thank you. 390 MR. BRETT: I'll speed things up. 391 All right. Mr. Fogwill, I'd like to ask you this, and I don't -- I'd like you to -- this isn't quite as straightforward. But you're proposing to use a 20-year method and you've described what your objectives are in doing that. I don't know how this regression analysis is going to work, of course, and I think there will be others that will ask you more questions about the nature of the regression analysis and the array of choices and whether it's the appropriate method and so on to use. I'm not going to get into that area. 392 But I do -- I'm interested very much in this scenario: Let's suppose that you do employ -- you get the change you want and you then use that method for the next 10 years in forecasting the following year's weather. And let's assume that the actual degree days in each of the next 10 years are -- starting with 2003, estimated degree days of 4,033, which I gather is still a best estimate. That's a 6 and 6 estimate that you gave in an IR response. There may be a slightly better estimate now. Is there, or is that still your best estimate for 2003 degree days? That's contained in the bottom corner there of J.26-35. 393 MR. FOGWILL: We have the benefit of a few extra months so it would change from that number. 394 MR. BRETT: It will. Have you had any idea -- have you seen how much it has changed yet? Have you the 9 and 3? 395 MR. FOGWILL: No. 396 MR. BRETT: All right, well, no matter. What I want to get at, hearkening back to my earlier conversation this morning, but if you had a degree day decrease each year from this number of 4,033, if it decreased four degree days over each and every one of the next 10 years, I would like to know to what -- I would like to know how that compared with what your 20-year trend forecast would throw up for each of those 10 years. Am I clear? 397 Unless I'm missing something, it's a very clear request. 398 MR. FOGWILL: The trend -- the 20-year trend, the magnitude of the trend will change each year that you add a new year. 399 MR. BRETT: I understand that. 400 MR. FOGWILL: So that the actual change or decline in the number of heating degree days will vary from one year to the next. 401 MR. BRETT: That's the forecast, the decline. 402 MR. FOGWILL: Yes. 403 MR. BRETT: In other words, you're saying there's a different line developed each year with a different slope. 404 MR. FOGWILL: Yes. 405 MR. BRETT: I understand that. But I'm saying -- I'm saying at the same time the actuals are going to be what the actuals are going to be. 406 MR. FOGWILL: Right. 407 MR. BRETT: And I'm saying that I'd like to see that exercise done, the new line created each year for the next 10 years, on the assumption that the actual doesn't respond to what the line says it's going to do, it actually drops by four degree days a year. Do you see what I'm getting at? 408 MR. FOGWILL: Well, each year that you recalculate the trend, you're going to get -- you're going to incorporate that information. So if you've got a consistent decline of, let's say, four heating degree days, if you have it over a long enough period of time, the trend will exactly match that trend in the actual data. 409 MR. BRETT: Well, maybe if I could pick up on your point, "over a long enough period of time." You were talking five to seven years. Why don't we just do it for five -- for seven years. If you do it for 40 years, I understand what you're saying. But I think that it's not the same if you do it for five to seven years, is it? If you do it for five to seven years, that line is going to be driven, in part, off the pattern of temperatures in the last 10 to 12 years here. 410 MR. FOGWILL: Yes. 411 MR. BRETT: That's what primarily will drive the slope of that line. 412 MR. FOGWILL: No, not primarily because all the points have to be included in the analysis. So -- 413 MR. BRETT: But in the first year there was only one new point and there's 20 existing points. 414 MR. FOGWILL: 19 existing points. 415 MR. BRETT: All right. Would you do that? 416 MR. FOGWILL: I'm not sure of the benefit of that. I'm struggling here with looking at this in a hypothetical situation, assuming a small decline year over year, and how that would actually benefit the review of whether it's a more reasonable method to use than the 30-year average. 417 MR. BRETT: Well, I think it would be helpful. It would certainly tell you something about symmetry, perhaps. 418 MR. FOGWILL: But if we are just guessing at what the future numbers are going to be, then any result of the symmetry would also be a guess. 419 MR. BRETT: We're putting out scenarios of what future numbers to be, and to do it properly I should ask you to do it for each of two, three, four, five, six, seven, eight, nine. But I'm not doing that, I'm asking you to pick four. Why? Because four is a nice small number, consistent in a broad sense with what scientists expect to happen to climate over a hundred years. I could pick three or two. And I'd like to see the result of that trend compared with what one would expect would be the normal sort of warming progression. I'm not saying there isn't one. But I think that the operative words are "gradual warming progression." 420 MR. SOMMERVILLE: Mr. Brett, could you state -- restate specifically the undertaking that you're seeking? 421 MR. BRETT: Yes, sir. It's really that Union perform a calculation using their 20-year trend method of what they expect the degree days would be in each of the next seven years, on the assumption that the actual temperature -- starting with a base of the 2003 year, so starting with 20 years going back from 2003 -- 422 MR. SOMMERVILLE: Using the 4,033 -- 423 MR. BRETT: Using the 4,033. 424 -- and assuming the actual temperatures in each of the seven years out from 2003 drop by an average of four degree days a year. 425 MR. SOMMERVILLE: The only exception being, Mr. Fogwill, I think this is your qualification, is that the trend, normally applied, would incorporate that decrease in heating days in each successive year, and that you will want to make that qualification when you make the undertaking. 426 MR. FOGWILL: Yes, Mr. Chairman. I guess the other point, though, is if we're looking at starting with a number in order to make that adjustment, we're making an assumption that the 2003 is a representative year that we start that analysis with. And if you look at the information of all the actual heating degree days, there's significant variability, so... 427 MR. SOMMERVILLE: If there's one thing we know about weather, it's that it is variable. I'm going to ask you to do this undertaking. It will be the subject of argument, what actually flows from it. But if you would -- if you could conduct that, I think, Mr. Brett, that's what you're seeking? 428 MR. BRETT: Yes, sir, thank you. 429 MR. MORAN: That will become Undertaking N.2.1, Mr. Chair. 430 UNDERTAKING NO. N.2.1: UNION GAS UNDERTAKES TO PERFORM A CALCULATION USING THEIR 20-YEAR TREND METHOD OF WHAT THEY EXPECT THE DEGREE DAYS WOULD BE IN EACH OF THE NEXT SEVEN YEARS, STARTING WITH A BASE OF THE 2003 YEAR, USING THE 4,033, AND ASSUMING THE ACTUAL TEMPERATURES IN EACH OF THE SEVEN YEARS OUT FROM 2003 DROP BY AN AVERAGE OF FOUR DEGREE DAYS A YEAR, INCORPORATING THAT DECREASE IN HEATING DAYS IN EACH SUCCESSIVE YEAR 431 MR. BRETT: Just a few questions on the table, your methodology, Mr. Fogwill, your matrix, if you like. 432 MR. FOGWILL: Will you give me the reference you're referring to? 433 MR. BRETT: Well, I'm about to. For the moment, let me just work my way into this. 434 Can you confirm to me that -- I think you've done this already, but in constructing your grid to judge these various proposed methods, to judge the existing method and the various other methods that all the other utilities in Canada use and the method that you're proposing, you have five characteristics here. And I think you touched on this with the Board just a moment ago. You have symmetry, accuracy, simplicity, sustainability, and the fifth one here -- I'm on page 23 of your evidence, Mr. Fogwill -- and you say symmetry has a weighting of 3, accuracy a weight of 2, and the objectives of stability, simplicity, and sustainability a weighting of 1. So you weight these criteria. We've talked about how you define accuracy and how you define symmetry. 435 Now, I just note in passing that -- well, never mind that. 436 If we could look at your table on page 22 where you actually set out the results, the analysis of the alternative weather normalization methods. Let's start that way. And we're now working on your assumptions, Mr. Fogwill; that is to say, we're working with your method, which is to take the last 15 years, starting in 1985, which contains, I think by everyone's concurrence, these rather spectacularly warm years in the latter part of the 1990s in historical terms. But nonetheless, we're working with your assumptions. 437 If you look at the error, I want to look at the column under accuracy, this is table 2, under the column of accuracy, you have two measures of accuracy, route mean squared error and mean absolute percentage error. And I wanted to ask you about the difference between those numbers and then how that gets translated into the sort of shorthand method you've developed to guide you in your evaluation of these various proposals. 438 And if we look at the 30-year average, we see a mean absolute percentage error of 7.9 percent; do you see that? 439 MR. FOGWILL: Yes, I do. 440 MR. BRETT: And looking down at the 20-year trend, the mean absolute percentage error of 7.0; right? 441 MR. FOGWILL: Yes. 442 MR. BRETT: And that's a difference of something in the order of, well, roughly 13, 14 percent; right? 443 MR. FOGWILL: Percentage difference, you mean like the .9 -- 444 MR. BRETT: Yeah, 7.9 minus 7.0 over 7.9, very roughly. 445 MR. FOGWILL: That sounds approximately correct. 446 MR. BRETT: Okay. Now, then, if you -- in terms of the simplicity -- well, let me then go over -- what you do then, as I understand it, is you, having done these statistical calculations for each of symmetry and accuracy, you then assign a weighting of 1 to 7, right, for each of the seven methods, 1 being the lowest, 7 being the highest; is that correct? 447 MR. FOGWILL: That's correct. 448 MR. BRETT: And so you -- when you do that multiplication, if we go over to table 4, which is a little further on in your evidence at page 24, you show the -- and these are the points that you're putting together to get this overall point ranking. You see the 20-year trend with 10 points and the 30-year average with 2 points under the column MAPE; do you see that? 449 MR. FOGWILL: Yes, I do. 450 MR. BRETT: So what you've done, what you've done by using this point method of 1 to 7 is you've taken essentially a difference of -- a relatively modest difference, 13 percent, in the error magnitude of the two methods and you've transformed that into a rubric here, a kind of accounting method, where the 20-year trend for that category gets 10 times the value of the 30-year average. 451 So isn't there a bit of -- isn't there a distortion there? I mean the difference in the error in those two methods is not that great, but at the end of the day, the way you were doing this little table, you end up making one look much more -- much worse than the other. Does it not amplify the differences? 452 MR. FOGWILL: No, I don't think so, because it's not really intended to give specific numbers that have any significance outside of the simple ranking that we've got in place there. All we're trying to do is list the seven methods and say, Based on all the objectives that we've got, the accuracy, symmetry, what have you, which ones work better and which ones don't work as well. And so that's all we're intending to do there. So the numbers are really unitless and just represent how they rank on an overall basis. 453 MR. BRETT: Mr. Fogwill, when you look at simplicity as a criteria, you've given it a ranking of 1, and if you look at table 4, you have shown the 30-year average of the various averages as -- the various moving average techniques as simpler than the trend. To your credit, you have shown that. But it's a different order of magnitude. We agree, I take it, that the method -- that the moving average method is a much simpler method than a method that involves regression analysis in terms of ability to understand what's happening, visibility, transparency, and the like. 454 MR. FOGWILL: It's a simpler mathematical function, yes. 455 MR. BRETT: All right. Now, I think your -- the evidence is -- maybe you can just confirm this with me. I don't have the number in front of me, but I believe the evidence is that the impact of moving from a 30-year moving average to a 20-year trend on the revenue requirement for the test year is $20.6 million; is that correct? 456 MR. FOGWILL: I think that might be high. The number I have in my head is 20.4. 457 MR. BRETT: These are going to be some very simple questions. I want to finish here. They're sort of just to put the very simplest parameters around this thing. So they will only be half a page instead of a full page. 458 Do you say any particular reason why -- the second question: Do you see any particular reason why there ought to be two different methods in Ontario for doing weather normalization? Are the climatic conditions under which Union operates so vastly different than those under which Enbridge operates and justifies two separate methods of weather normalization, or should there just be one? 459 MR. FOGWILL: I don't think it matters. I mean, we've been operating under two different methods for over 10 years now so I don't think there's any legitimacy to having one method or two methods or however many you want. 460 MR. BRETT: Wouldn't one be -- well, let's just pause a moment there. I mean, wouldn't one be a lot simpler for people? 461 MR. FOGWILL: Well, the method we have is fairly straightforward, and we think it's something that people can understand. I'm not sure about the Enbridge method. I really don't know, you know, if one method is easier to understand for everyone. I'm sure everyone has all the -- all the different utilities have got a different way of managing their business. 462 MR. BRETT: Do you -- if you were to be successful with this proposal, would you -- would you cease to use weather hedges? 463 MR. FOGWILL: I don't know. 464 MR. BRETT: You haven't thought about that at all? 465 MR. FOGWILL: It's not my area of -- 466 MR. BRETT: I guess I should ask Ms. Elliot about that. 467 MR. FOGWILL: That would probably be better. 468 MR. BRETT: Mr. Fogwill, you're the -- I believe from looking at your evidence, and appendix E lists the -- I don't know that you need to turn this up, but just for reference for the Board, appendix E is was the weather normalization methods of other Canadian gas utilities, and it lists the methods used by all the Canadian gas utilities. Your proposed method is unique in the sense that no other gas utility has, at the moment, a trend method; correct? 469 MR. FOGWILL: Well, no, that's not quite true. The Enbridge method does include a trending component to it. 470 MR. BRETT: All right. That's the one that you -- the one that you folks consider to be complicated. But apart from the Enbridge one, am I right, apart from Enbridge, that no other Canadian entity -- utility has that, has a trend method? 471 MR. FOGWILL: That's correct right now. 472 MR. BRETT: Have you discovered any U.S. utility, gas utility which uses a 20-year trend method? 473 MR. FOGWILL: I haven't looked. 474 MR. BRETT: You haven't looked, okay. So you're not aware of whether anywhere in the U.S. or Europe there's another gas utility that uses the 20-year trend. 475 MR. FOGWILL: I'm not aware of that. 476 MR. BRETT: Those are my questions. Thank you, Mr. Chairman, Mr. Birchenough. 477 MR. SOMMERVILLE: Mr. Brett. 478 Mr. Janigan, geographically you're next. 479 MR. JANIGAN: Thank you, Mr. Chair. 480 CROSS-EXAMINATION BY MR. JANIGAN: 481 MR. JANIGAN: I'm taking peril with cross-examining from my laptop with fat fingers to scroll, so it's always a challenge. 482 Now, according to -- panel, my name is Michael Janigan and I'm counsel for the Vulnerable Energy Consumers' Coalition, who represent generally low-income consumers who are system sales customers of Union, by and large. I have some questions concerning the methodology and its implications. 483 Now, Mr. Fogwill, according to Union's evidence, the weather normalization methodology is significant because it affects the customer consumption forecast of volumes; is that correct? 484 MR. FOGWILL: That's correct. 485 MR. JANIGAN: And I believe your evidence states at C.1, tab 4, page 6, that the effects of using the new methodology are significant. The weather normalization assumption affects forward forecasted customer consumption, storage and transportation allocations, gas supply planning, and operations. 486 MR. FOGWILL: What was the reference again, C.1 -- 487 MR. JANIGAN: Tab 4, page 6. 488 MR. FOGWILL: Page 6. 489 MR. JANIGAN: Under the heading "Using the Impacts of the 20-Tear Trend." 490 MR. FOGWILL: Yes. 491 MR. JANIGAN: And when we forecast volumes in the context of a rates proceeding, we obviously want a methodology that is going to lend itself to producing an accurate forecast of volumes; correct? 492 MR. FOGWILL: Accuracy is one of the components, yes. 493 MR. JANIGAN: For volumes, not -- 494 MR. FOGWILL: Yes. For volumes, yes, correct. 495 MR. JANIGAN: And weather normalization is only one of the methodologies that affects the weather forecast; correct? 496 MR. FOGWILL: Weather methodology is one of the input assumptions, yes. 497 MR. JANIGAN: And there are other input methodologies that affect it. It's a package, really; isn't it? 498 MR. FOGWILL: The term methodology is not correct here. It's an input assumption. 499 MR. JANIGAN: Okay. There are other input assumptions -- 500 MR. FOGWILL: There are other input assumption that affect that forecast. 501 MR. JANIGAN: Okay. And the outcome of the volumes forecast is key in total as a driving force as to whether the company is over- or underearning. 502 MR. FOGWILL: The linkage between the demand forecast and earnings is not clear. 503 MR. JANIGAN: Well, the outcome of the total volume forecast, in terms of the rate proceeding, is a key factor for the company as to whether they will earn a reasonable rate of return; would you not agree? 504 MR. FOGWILL: Well, a forecast is not -- 505 MR. JANIGAN: The approved forecast. 506 MR. FOGWILL: Well, the approved forecast is not necessarily going to be what's going to actually transpire, so it really has -- you have to look at the what the actuals are. 507 MR. JANIGAN: Yes. 508 MR. FOGWILL: But in terms of dealing with how the revenues or returns would come out of something like this, you're probably better to ask Pat Elliott. 509 MR. JANIGAN: Well, I'm keeping this on pretty simple terms, as you may appreciate. But let's take an example. 510 If the test year forecast is above the actual volumes, there's a probability that you're going to be operating under a cost of service rate-making process that will end up in you not recovering a fair rate of return in rates. 511 MR. FOGWILL: Again, you're probably best to pose that question to Pat Elliott. 512 MR. JANIGAN: Well, it's Union's view that this new weather normalization methodology will lend itself to producing a more accurate volume forecast? 513 MR. FOGWILL: Yes. 514 MR. JANIGAN: Now, according to the Union evidence, you note that the weather normalization of the 30-year average has resulted in an overestimate of heating demand from customers, and this, in effect, translates into increasing the estimate of volume and revenues from customers. Am I correct on that? 515 MR. FOGWILL: Yes. 516 MR. JANIGAN: And in your evidence from the previous case, RP-2002-0130, at Exhibit B, tab 2, page 1 -- sorry, page 4, para 4, para 14, you indicate that the -- well, let me turn that up. 517 MR. PENNY: Sorry, could we have the reference again, please? 518 MR. JANIGAN: It is Exhibit B, tab 2, page 4. That the flaws in this method, about three-quarters of the way down the page: 519 "Over the flaws in this method --" 520 MR. PENNY: Sorry, Mr. Janigan. 521 MR. JANIGAN: Of the old evidence, EB-2002 -- sorry, RP-2002-0130. 522 MR. SOMMERVILLE: It's under appendix A of the evidence, Mr. Fogwill's evidence in this case. 523 MR. JANIGAN: I'm sorry. 524 MR. SOMMERVILLE: Page 4 of 31. 525 MR. JANIGAN: That's correct. Thank you, Mr. Chair. 526 Do you have that reference now, Mr. Fogwill? 527 MR. FOGWILL: Yes, I do. 528 MR. JANIGAN: Thank you. And about three-quarters of the way down the page it says: 529 "However, the flaws in this method increase the volume and revenues from customers to the point where it significantly affects the ability of Union to manage under a PBR, even considering its adjustment factors." 530 Now, I assume that means that this impacts the rate of return. 531 MR. FOGWILL: Again, I think the question in terms of the rate of return should be addressed to Pat Elliott. 532 MR. JANIGAN: Okay. But when you put this in -- when you put this in your evidence, "it significantly affects the ability of Union to manage under a PBR," what were you thinking about? What was the significant effect that you were thinking about then? Was it on money, was it on -- 533 MR. FOGWILL: It was looking at the general service market volumes and distribution revenues. 534 MR. JANIGAN: Okay. It was money you were thinking about? 535 MR. FOGWILL: In terms of the revenue component it was the distribution revenues. 536 MR. JANIGAN: The significant effect? 537 MR. FOGWILL: Yes. 538 MR. JANIGAN: All right. And the 30-year weather normalization method was coming up with a volume forecast that was higher than possible to deal with as it was overestimating the number of heating degree days, in your estimate. 539 MR. FOGWILL: Could you just rephrase that? 540 MR. JANIGAN: Well, in effect, the 30-year forecast was coming up with a higher number for heating degree days than you believed was warranted by the evidence. 541 MR. FOGWILL: That's correct. 542 MR. JANIGAN: Okay. Now, I wonder if I could take you to IR response J.165 to Board Staff. And in table 1, this sets out through-put volumes and notes them case by case in three columns, one being actual volumes, the second being normalized actual volumes, and the third being Board-approved forecast volumes; is that correct? 543 MR. FOGWILL: That's correct. 544 MR. JANIGAN: And the volumes that we're talking about in each of these cases, for each of these years, were derived in part by use of the 30-year normalization method for weather; is that correct? 545 MR. FOGWILL: Yes. And if we look at this, would you agree with me that the actual volumes versus Board-approved forecast volumes shows that, of the 23 forecasts shown, 15 times out of 23 the actual volumes were in excess of the Board-approved forecast volumes? 546 MR. FOGWILL: I think it's important to distinguish the fact -- 547 MR. JANIGAN: Can we establish that that's correct, first of all, and then could I have you answer after that? 548 MR. FOGWILL: How many did you say? 549 MR. JANIGAN: That if we look at the actual volumes versus Board-approved forecast volumes, it shows of the 23 forecast volumes shown, 15 times out of 23 times the actual volumes were in excess of the Board-approved forecast volumes. Would you take that, subject to check? 550 MR. FOGWILL: Yes, I will. However, I would like to point out that the impacts associated with the weather normalization method for either forecasting gas supply or revenues are being dealt with by other panels. So we're really here to talk about how the method actually is reflective of the actual weather conditions and what the potential impacts are in the future. So I'm not sure if these questions are better directed at other panels. 551 MR. JANIGAN: But as we agreed earlier, the weather is an input into the volumes forecast. 552 MR. FOGWILL: Yes, it is. But, again, these impacts should be directed to other panels, the forecasting panel, for example, the gas supply panel, and the director of finance, Pat Elliot, for any revenue issues. 553 MR. JANIGAN: Just before we leave that, if we compare the normalized actual volumes to the Board-approved forecast volumes generated with a 30-year average method, 18 times out of 23 times, the normalized average volumes were in excess of the Board-approved forecast volumes. Would you take that, subject to check? 554 MR. FOGWILL: Yes, subject to check. 555 MR. JANIGAN: And if we look at that same interrogatory, on page 3, the number of times in which Union Gas's actual rate of return was in excess of the rate of return approved by the Board was 11 out of 19 times; am I correct on that? Would you take that, subject to check? 556 MR. FOGWILL: Yes. 557 MR. JANIGAN: Okay. And according to the same table, the number of times the Centra Gas actual rate of return was in excess of the approved forecast was 7 out of 13 times. Would you take that, subject to check? 558 MR. FOGWILL: Yes. 559 MR. JANIGAN: Okay. Now, I wonder if I could have you turn up Exhibit J.34-48. It's an interrogatory to VECC. And I'm looking at question C of this interrogatory, which indicates: "Please confirm that weather stations from Environment Canada measures normal to be a 30-year average." And your answer is: "The current Environment Canada weather normal standard is the 30-year average of the heating degree days from the years 1971 to 2000." 560 That continues to be the case, as far as you're aware? Probably Mr. Fogwill can answer that. 561 MR. FOGWILL: As far as I'm aware, that is the current reference point that Environment Canada is using. 562 MR. JANIGAN: And I believe in appendix B, your evidence, Mr. Root, that you note that at the -- page 1, the second paragraph of your evidence: 563 "At the International Meteorological Conference in Warsaw in 1935, the years 1901 to 1930 were selected as the international standard period for temperature normals. 564 The temperature normals determined by the United States Weather Service, however, were computed from the period 1921 to 1950. Thereafter, standard procedures have been modified to recalculate the temperature normals at the end of every decade using the preceding 30 years. This practice was established to take account of slow changes in climate and to add more recently established weather stations to the network of observed normals." 565 Does that continue to be the practice of the U.S. Weather Service as far as you know? 566 MR. ROOT: As far as I know, yes, it is. 567 MR. JANIGAN: And I believe that I have sent to your counsel a document that is an extract from the web site of the World Meteorological Organization. I wonder if I could have that turned up and marked. 568 MR. MORAN: Mr. Chair, this would become Exhibit M.2.2. 569 MR. SOMMERVILLE: Thank you, Mr. Moran. 570 MR. MORAN: A document from the World Meteorological Organization web site, filed by VECC. 571 EXHIBIT NO. M.2.2: DOCUMENT FROM THE WORLD METEOROLOGICAL ORGANIZATION WEB SITE, FILED BY VECC 572 MR. JANIGAN: I missed the reference. It was M.2.2? 573 MR. MORAN: That's right. 574 MR. JANIGAN: And as I understand it, the World Meteorological Organization is a 185-member organization operating within the United Nations, that, according to the web site, that provides the authoritative scientific voice on the state and behaviour of the earth's atmosphere and climate. 575 Dr. Weaver, you're undoubtedly familiar with this organization? 576 DR. WEAVER: Yeah. That's -- I haven't got the exact wording, but that sounds pretty accurate, yes. 577 MR. JANIGAN: And I note on the last page of this exhibit, it says that -- the development of the client database project at the top: 578 "That this activity, although not under the specific name of the current project, has been ongoing since the inception of the WMO in 1950, notably through the publication of climatological standard normals which are defined as averages of climatological data for consecutive periods of 30 years." 579 As far as you're aware, Dr. Weaver, do they continue to use 30 years to calculate their climatological standard normals? 580 DR. WEAVER: Yes, they do. 581 MR. JANIGAN: And has any international meteorological group that you're aware of, or any meteorological service operating under the aegis of a sovereign nation change the methods that calculates normals from a 30-year average to a 20-year trend methodology recently? 582 DR. WEAVER: The answer is in terms of a reference, I am not aware of anyone doing that. However, again, the issue here is what is that reference normally used for, and again I go back it my analogy of bringing your child into the doctor's office, whereby when you seek information from the doctor as to whether or not your child is above or below normal height, the doctor needs a reference with which to define above or below normal. That is not telling you anything about how fast your son is growing or what the son will be like next year. It is simply the matter of defining a normal. 583 And the reason why they shift the normal, as they have done here, is, for example, when doing seasonal prediction in a national facility like NCEP, it got really easy to forecast above-normal conditions, because every year the seasonal forecast would say above normal. Why? Because clearly the reference was a bad reference. The reference had warmed, i.e., it was always warmer than the reference so they redefined what normal is basically so that they could have a more realistic explanation as to what is the reference, a more up-to-date reference. 584 MR. JANIGAN: I turn to your example which appeals to me when dealing with the meteorological evidence. The growth normal that you cited when you go into your doctor's office with your child, presumably that growth normal is indicative of trends for the growth of children across the sample -- 585 DR. WEAVER: Not at all. You will find, actually, and this is something I do know, if you look at the European handbook - I have young children - look at the European handbook for what a normal child is, it's quite different from the American -- North American handbook. The reason is because children are smaller at the same age in Europe than here. So for them, the appropriate reference is the average of conditions in the European children, whereas for us it's North American children. It doesn't make sense for us to compare to a different group. 586 MR. JANIGAN: But let's deal with -- let's say if you were in Europe, you would have a European normal. Is that what you're saying? 587 DR. WEAVER: You would have a reference with which you would define above or below normal. It doesn't say anything about rates. It says that typically, at the age of X, the average of all the people with such age as X is height Y. 588 MR. JANIGAN: And if you plotted them chronologically, presumably you'd get a trend. 589 DR. WEAVER: If you -- what do you mean? 590 MR. JANIGAN: Well, if you have the chronological age of children and you're looking at the growth rate, and given your -- I understand your caveat that within that range of the growth rate there are measures both above and below, but you would get a trend for growth for children -- 591 DR. WEAVER: No, you would have to redefine a new normal for each age, that is, for the age of 6 you would have to define a normal what is average for children of age 6. 592 MR. JANIGAN: Yes. 593 DR. WEAVER: For the age of 7, you would have to define what is average. You cannot get any predictive scale out of a reference definition. 594 MR. JANIGAN: But if you plotted a line presumably between those particular observations, between the observation for a child at 6 and the observation of a child at 7 and the observation of a child at 8, you would get a trend line. 595 DR. WEAVER: You would, but this is where the analogy begins to break down. I don't think you could do that and draw an analogy to the meteorological age in that aspect. 596 MR. JANIGAN: Let's return to the meteorological aspect while maintaining the normals. I'm beginning to question the growth charts myself. 597 Presumably, the value of this data, of having the 30-year norm, is that it has some predictive value in some respects associated with the -- 598 DR. WEAVER: It has no predictive value. It is simply a reference from which you can define what is above or below normal. 599 MR. JANIGAN: Okay. And those normals are based upon 30 years of observations? 600 DR. WEAVER: Yes. 601 MR. JANIGAN: And in the event that, for example, they felt that climate change was casting such a pronounced effect upon those observations that it would be best to shorten the horizon of those observations, would they not adopt a 20-year window rather than a 30-year window? 602 DR. WEAVER: Knowing how these international organizations work, I would think that there's an awful lot of corporate memory in these institutions and 30 years would be around for a long time. When you're defining normal, 30 years is -- I mean, I don't actually know historically why they chose 30 as opposed to 20 or 25. Maybe one of the panel members knows this. But I would expect it would continue, normal being 30, because 30 years is a sufficiently long time in which to define it. No other reason which I can think of. I don't know why they would use 20 or 30 or 40. 603 MR. JANIGAN: I've been told in statistics that in order to develop a normally distributed population on a bell curve, one needs at least 30. Am I correct on that? 604 DR. WEAVER: Yes, you are. 605 MR. JANIGAN: Holy smokes. Okay. 606 Now, Mr. Root, am I correct in understanding from the responses I believe Mr. Fogwill gave to Mr. Brett concerning the 20-year trend, am I correct in understanding that the 20-year trend that you recommend hasn't been adopted by any utility in the United States? 607 MR. ROOT: I can't answer that correctly because many times I'm not told. WeatherBank has many energy clients that we have done studies for, provided data for, that has been used to generate 30-year averages, 30-year rolling averages, 20-year averages, 20-year rolling averages, 10-year averages, and 10-year rolling averages. To some extent, clients have used that database to generate trends as well on all three of those scenarios. Unfortunately, I'm not always given the exact details of the final answers of what they've decided upon. We're just the weather guys and as a result we're not always brought into the loop. 608 I know that with discussions from my clients that, by the way, represent a large number of the regulated and deregulated energy providers across North America, that many times they have deviated from the 30-year average. What they ended up with, I'm not always sure. 609 MR. JANIGAN: Let me rephrase that. Do you know of any regulator that's adopted a 20-year trend for their purposes? 610 MR. ROOT: I don't have that data here. I can't answer that. I don't know. I can't quote one entity or two or three. I don't know. 611 MR. JANIGAN: Okay. Now, with some trepidation I come back to your evidence again, Dr. Weaver, and I'd like to just revisit those tables that Mr. Brett was looking at associated with appendix A, page 13. 612 DR. WEAVER: These are the figure 4? 613 MR. JANIGAN: That's correct. 614 DR. WEAVER: All right, thank you. 615 MR. JANIGAN: And am I correct to understand that you're in agreement that these tables seem to show that the weather, over time, has cycles? 616 DR. WEAVER: First of all, these are not showing anything about weather. Weather and climate, again, it's an important distinction. 617 MR. JANIGAN: For temperature. 618 DR. WEAVER: What these figures show is that the annual mean temperature trend -- it doesn't say anything about cycles. It says that the trend is not constant at all times, that is, there are periods where it's actually cooled a little bit and periods where it's warmed and periods where it's warmed further. What this is saying is that it is not a perfect trend, that there is climate variability about the mean trend. So at any given time you'll find that there's warmer periods or cooler periods, but taken together, the D panel is showing a warmer trend throughout this century. 619 MR. JANIGAN: Given the overall warming trend, would it be fair to suggest the possibility of a cooler period following D, for example, based on the experience in the earlier graphs? 620 DR. WEAVER: I would say a resounding no. 621 MR. JANIGAN: Okay. 622 DR. WEAVER: That is, it is possible that, relative to the 1990s, at some point we may be down for a year or two, it is -- given what we know about the radiative forcing associated with the various greenhouse gases, it would take an awful lot of volcanoes, for example, things like that -- of course, you can have individual years where radiative forcing would change -- if a lot of volcanoes go off in three or four years, it could cool for a couple of years. However, these are very short-lived phenomena and they will affect year by year but not the longer term trend. 623 We believe as a scientific community that we understand some of the reasons behind panel C there, which is the cooling between 1946 and '75. As a scientific community, we believe that's associated with increased volcanic activity as well as a significant production of what are known as sulphate aerosols. These were associated with acid rain way back then and they've been severely curtailed in the emissions which has suppressed a lot of that cooling there. 624 We believe we understand with an extraordinarily high degree of confidence what is causing the warming, and that is why we believe that warming will continue in the future. 625 MR. JANIGAN: I guess my question is not so much from the standpoint of the overall trend but whether or not there are cycles within that trend where you may see a 25- or 30-year period where the temperature may decline for a period of time after it's gone up. 626 DR. WEAVER: There will always be variability -- I like the word variability than cycle, and I could explain that briefly. Cycling implies a periodicity. Variability means it's not at any particular period, it's not every 20 years. It's over, you know, 20, 40, 50, some broad range. So, yes, there will be variability; however, that variability is superimposed upon a warming trend. So if you envision a straight line, we don't have a board here, a straight line going up, as this goes up, there will be wiggles on that straight line so that at any given time, you know, there will be dips relative to previous years but that would be superimposed upon a warming trend. And what you're seeing in '46 to '75 is one example, but this will continue on with such wiggles in the future. 627 MR. JANIGAN: Let me go back to your terminology which is variability and periods of variability. Is it possible to give an estimate of the periods of years that these periods seem to manifest? 628 DR. WEAVER: I can say that the single biggest source of climate variability is what is known as the El Nino/La Nina oscillation, and the periodicity associated with that is three to five years. That is the single biggest source of climate variability. Then there is periodicity on a slightly longer time scale, it's called decadal variability, and this is in the order of 10 years, and that's largely thought to be in and around the North Atlantic region and perhaps in and around the Antarctic circumpolar region. 629 Then you start to get to longer time scale variability on the century time scale and even longer, and I think they're irrelevant relative to this kind of issue. This issue of 50-year time scale PDO variability is something that really -- I mean, you can see some variability in the system on these time periods; however, there's no theory, so to speak, as to explain where that's coming from, whether or not it's just natural fluctuations in the ocean or whether it's something else that we don't know about. I can't answer. But, yes, there is variability. But almost -- if you look at where the energy or where the maximum variability is, it's in that El Nino band, that is the big one. 630 MR. JANIGAN: Let's say a period of variability occurred similar to what occurred from '46 to '75, notwithstanding the fact that the overall temperature measured as against the '46 to '75 is not as low, would not be as low as against that period. If this variability lasted longer than a 20-year time horizon, wouldn't it be difficult to capture the downward -- the downward trend in any variability circumstance by use of the 20-year method rather than a 30-year method? 631 DR. WEAVER: This is a hypothetical situation that I don't, you know, personally see as in the cards. So you're suggesting a hypothetical situation whereby suddenly you have 50 years or something of cooling? 632 MR. JANIGAN: No, I meant -- I'm suggesting exactly the same what happened in example C, if that occurred again. 633 DR. WEAVER: From 1946 to 1975, so 30 years of cooling. What you would find, of course, in any trend, that you would systematically be trying to catch up, so you would systemically make a error in it, that's correct. But is this a possibility? The answer is no, in my opinion. 634 MR. JANIGAN: So you don't believe there would be such a period of variability that would exhibit a cooling pattern even at a higher temperature than existed -- 635 DR. WEAVER: I believe that there is potential for such -- for climate variability. If -- as I believe this is explained because of things like tropospheric aerosols, which we believe we understand quite well, and if this cooling trend is associated with that together with volcanos, you know, I can't see it happening in the foreseeable future, unless there's a whole pile of volcanos going off year after year. And anybody -- these would last about -- they would have a climate effect for about 18 months. 636 Predicting volcanic -- I mean you can't -- I always say this in public lectures on climate: You can't make policy based on volcanic eruption predictions because you can't predict them. 637 MR. JANIGAN: And if we went back further to the period of -- as you've listed here in A, would we see the same sort of, almost neat periods of variability between warm and cold lasting into the 19th century? 638 DR. WEAVER: Sorry, I don't ... 639 MR. JANIGAN: Well, we've got periods -- 640 DR. WEAVER: You mean the blue? 641 MR. JANIGAN: Yes. 642 DR. WEAVER: What the red means, the red means over the period 1901 to 1999, that's a positive warming trend. The blue means, over that period, it's a cooling trend. So there's a number of caveats that one has to start right off the bat with this. There's very little data at the turn of the century over large regions, that's caveat number 1. And caveat number 2 is we know that most of the warming, or a significant part of the warming that we attribute to climate change is happening in the latter part of the century. Why? Because there's a delay in the climate system to a perturbated -- radiated force in perturbation, it catches up, in some sense. 643 So you would likely not see a continuation of this pattern, you would probably see, in fact, almost clearly see it would be all red but not as red in some places as others. There may be a few small blue patches, you know, in the North Atlantic and there may be some blue patches in developing parts where the sulphate emissions are huge, associated with older technologies, but by and large this would be pretty red. 644 MR. JANIGAN: So the trends that are illustrated by C, in your view, aren't likely to occur again for that period of time. 645 DR. WEAVER: I will never say "unlikely to occur again." In the next 50 years? Very unlikely. In the next million years? Perhaps sometime during that time. But in the next 50 years, very unlikely. 646 MR. JANIGAN: Well, Mr. Penny will be here, but I don't think I will. 647 MR. PENNY: Not here. It only feels like a million. 648 MR. JANIGAN: Will it be warmer or colder is what I want to ask. 649 Finally, Dr. Weaver, is there anything inconsistent in accepting all of your evidence on climate change but letting it be reflected by a system that uses 30 years of observations rather than 20? 650 DR. WEAVER: Say that again. Is there anything -- my report did not at any point rely on a definition of normal being 30 years. So my report here is not dependent on a 30-year reference. It's based on the science behind this issue. 651 MR. JANIGAN: In other words, it's not -- likewise, it's not dependent on a 20-year reference. 652 DR. WEAVER: Nor is it dependent on a 20-year reference. 653 MR. JANIGAN: Thank you. 654 Now, Mr. Fogwill, maybe you can help me with this question: How long has the 30-year rolling average been used? 655 MR. FOGWILL: I know it's been used from at least the mid-1980s, but I couldn't say how far back before that. 656 MR. JANIGAN: And this change to the proposed methodology will have the impact that's set out on J.163, I believe. 657 MR. FOGWILL: Are you referring to the rate impact? 658 MR. JANIGAN: Yes. That sets it out there? 659 MR. FOGWILL: Yes. 660 MR. JANIGAN: Okay, thank you. 661 Thank you, Mr. Chairman, those are all my questions for this panel. 662 MR. SOMMERVILLE: Thank you, Mr. Janigan. 663 We'll adjourn until 2:00. 664 MR. PENNY: I just observe, Mr. Chairman, as a thought before lunch, that based on the estimate - Mr. Brett estimated an hour and was two hours; Mr. Janigan estimated a half an hour and was 45 or 50 minutes - at that rate, our four weeks is going to turn into eight weeks. I don't know whether there's -- I'd understood from your remarks at the opening of the hearing that the estimates were to be gathered not only for the purposes of trying to get a guess on how long we were going to be but also for the purposes of imposing some discipline on the scope of cross-examination. Perhaps I misunderstood you on that. 665 MR. SOMMERVILLE: Thank you, Mr. Penny. I do not currently have any estimates, and it is my intent -- I expect that tomorrow we will have a package of estimates, going forward. Within reasonable boundaries, the Board will be looking to those estimates to guide our progress through this case. And where an intervenor has good reason to stray and to go well beyond that estimate, that's one thing; but where it seems as though we're just kind of exceeding the estimate, the Board may well intervene to indicate that we do expect those estimates to be observed, once again, within reasonable boundaries. 666 I hear you, Mr. Penny, and I think if -- does anyone have any submissions with respect to that? I'm not trying to put anybody into a straightjacket. But without some measure of ongoing diligence, we will drift to an unacceptable length in this proceeding. 667 MR. JANIGAN: I take it, Mr. Chairman, it would be useful as well to -- if we are aware of the fact that our earlier estimate has turned out to be incorrect, and once again the degree of accuracy is -- I remember yesterday the agree of accuracy of forecasts, the further out it goes, well, the further the panel goes, to some extent, affects the accuracy. But if there are any significant changes, I take it that we should advise both Mr. Moran and the Board? 668 MR. SOMMERVILLE: We anticipated the degree of variability. The other aspect of this, and this is also where the cooperation and coordination of activities is so important, is that -- and not a criticism certainly not of this panel in any way shape or form. Sometimes witnesses want to repeat answers, and that's another aspect of time usage that needs to be observed and managed to some extent. So we'll be diligent on that side as well, if that's of any comfort to anyone. 669 Having said all that, we'll rise until 2:00. Thank you. 670 --- Luncheon recess taken at 12:40 p.m. 671 --- On resuming at 2:04 p.m. 672 MR. SOMMERVILLE: Thank you. Please be seated. 673 I'm advised that the intervenors have worked out an order for this afternoon. 674 Mr. Shepherd. 675 MR. SHEPHERD: Thank you, Mr. Chairman. 676 CROSS-EXAMINATION BY MR. SHEPHERD: 677 MR. SHEPHERD: Dr. Weaver -- 678 MR. PENNY: Just before we begin, Mr. Chairman, may I say for the record that there are -- my apologies to Mr. Shepherd. Before the lunch break we had passed out but I had neglected to put on the record that we had filed Exhibit N.1.2, which was Mr. Dent's undertaking to Ms. Singh with respect to that chart and the numbers, and then Mr. Dent, on reading the transcript, realized that he had incorrectly reported a certain reporting relationship and so he's prepared a transcript correction with respect to that and you have that as well. 679 MR. SOMMERVILLE: We have both of those. 680 MR. PENNY: Thank you. 681 Sorry, Mr. Shepherd. 682 MR. SHEPHERD: Mr. Chairman, the School Boards' will be making reference to a package of cross-exam materials that you have. 683 MR. SOMMERVILLE: Yes, we do. 684 MR. SHEPHERD: I wonder if we can give that an exhibit number. 685 MR. MORAN: Mr. Chair, this would become Exhibit M.2.3, a document entitled "Cross-examination materials from the Ontario Public School Boards' Association." 686 MR. SOMMERVILLE: Thank you, Mr. Moran. 687 EXHIBIT NO. M.2.3: CROSS-EXAMINATION MATERIALS FROM THE ONTARIO PUBLIC SCHOOL BOARDS' ASSOCIATION 688 MR. SHEPHERD: Mr. Chairman, I think all intervenors have been given copies, but there are extra copies at the back of the room if anybody is missing anything. 689 MR. SOMMERVILLE: Thank you, Mr. Shepherd. 690 MR. SHEPHERD: Dr. Weaver, I have a couple of questions for you to understand what your role is in this. 691 If I were to ask you whether next year is going to be warmer than this year, you can't tell me that, can you? 692 DR. WEAVER: Off the top of my head, no. I could probably assess you a likelihood as to whether it would be warmer than normal, if you were to define what is normal. 693 MR. SHEPHERD: But your evidence in this case isn't telling us that next year is going to be warmer? 694 DR. WEAVER: No, that's not my evidence in this case. 695 MR. SHEPHERD: And similarly, your evidence in this case is not expressing a probability that next year will be warmer than this year? 696 DR. WEAVER: That was not my task. My task in this case was to discuss the underlying justification, if any, of the existence of trends in temperatures in the Ontario region. 697 MR. SHEPHERD: I understand that. So similarly, you couldn't tell us, based on the evidence you filed in this case, not expressing a statement on your limitations academically, Dr. Weaver, but relating to this case, whether the next three years are going to be warmer than the last three years or the next ten years are going to be warmer than the last ten years. 698 DR. WEAVER: I can say with a great deal of confidence that, on average, the next ten years are going to be warmer than the last ten years, yes. 699 MR. SHEPHERD: What probability would you say that is? 700 DR. WEAVER: I would be loathe to give a real number. I'd say it is -- I mean, a good chance, so I would say greater than 66 percent -- I mean 66 percent. I don't like to give an actual number because that implies I've done a calculation, and I'm doing it off the top of my head without doing a calculation. I was not asked to prepare such a calculation before coming here. 701 MR. SHEPHERD: Given the evidence you've filed, I take it that the extent to which it's warmer in the next ten years than the last ten years, you're not talking about 10 degrees, you're talking about a small amount but a measurable amount. 702 DR. WEAVER: Small but measurable amount, amplified in the winter seasons, yes. 703 MR. SHEPHERD: Okay. 704 The second area that I want to ask you a couple questions about is, I wasn't sure when you were answering Mr. Brett this morning, whether you were expressing an opinion, a professional opinion on the Union Gas proposed methodology. Were you? 705 DR. WEAVER: I was not asked to provide a professional opinion on the actual methodology itself. I was asked to provide professional opinion as to the existence of trends in temperature data. I'm not a financial forecaster so that's not my area of expertise. 706 MR. SHEPHERD: So this morning you said that a 30-year average is not going to predict the future weather very well and a 20-year trend is likely to. That's what I thought you -- 707 DR. WEAVER: I would have used the word climate and not weather. I would have said that, given the existence of a trend in the temperature record that we can explain and we know the causality of why it exists, then if you continually use a running 30-year average to predict the future, you will systematically underestimate the temperatures because you're always playing catch-up. If you use a trend, you actually are there right off the bat. So the change from year to year is much -- the difference is much less. That's what I was trying to say. And that's not in terms of costs or supply or volume of gas or whatever. It's strictly in terms of temperature. I don't wish to stray into the gas aspects because that's not my area of expertise. 708 MR. SHEPHERD: Well, we're only talking about heating degree days here -- 709 DR. WEAVER: Sure. 710 MR. SHEPHERD: Which is temperature. 711 DR. WEAVER: Yeah. 712 MR. SHEPHERD: So I understand you to be saying that a 20-year trend is going to be better than a 30-year average -- 713 DR. WEAVER: Yes. 714 MR. SHEPHERD: -- for heating degree days? 715 DR. WEAVER: If you have a trend, as I have argued that there is and will continue to be, a 30-year average will always underestimate the temperatures; that is, if you have a warming trend and you base an average on 30 years running, it's always lower because you're always playing catch-up. If you have a trend, you're likely going to capture right off the bat and modify it, ever so slightly, year to year as you go on, which I understand, since I've been shown, is the essence of the Union Gas argument. 716 MR. SHEPHERD: So your expert opinion, then, is their proposed methodology is better than their existing methodology? 717 DR. WEAVER: I mean, I would say my opinion is yes, but my expert opinion would be if this is applied exclusively to heating degree days, my expert opinion would be yes. If it's in terms of volume of gas, I can't give an expert opinion on that. In terms of heating degree days, my expert opinion is that the use of a trend will better capture what will happen than the continued use of a climatological norm. 718 MR. SHEPHERD: Okay. I assume, then, you've -- I assume, then, you've -- are you finished? 719 DR. WEAVER: Yes. 720 MR. SHEPHERD: I assume you've looked at the two methodologies, the old one and the new one, and have done a comparison. 721 DR. WEAVER: I looked at the results put to me by Union Gas, yes, and looked at the different trends in its comparison with what exists. 722 MR. SHEPHERD: And so do you have that analysis? 723 DR. WEAVER: I didn't put it in my submission to the Board. It should be in our package here. I have it right here. 724 MR. FOGWILL: Can you clarify the question you're asking? 725 MR. SHEPHERD: Well, Dr. Weaver has given his expert opinion that one method is better than the other. I want to see his comparison to show it's correct. 726 DR. WEAVER: In terms of the temperature? This is figure 3 on page 29 of 31, of Exhibit B, tab 2. 727 MR. FOGWILL: That's wrong. 728 MR. SHEPHERD: Is this your evidence, Dr. Weaver? 729 DR. WEAVER: It's not my evidence, no. 730 MR. SHEPHERD: I'm not asking for Mr. Fogwill's expert opinion, I'm asking for your expert opinion. 731 MR. PENNY: Sorry. Mr. Chairman, the question was asked whether Dr. Weaver had an opinion, he indicated that he did and he was asked what it was and he has indicated what it was. Now, I think we need to be clear about what it is that Mr. Shepherd is asking for. If we can just slow it down a bit, can I ask, through you, for Mr. Shepherd to be clear and specific about what it is that he's asking Dr. Weaver for. 732 MR. SHEPHERD: Mr. Chairman, in my experience, when experts express an opinion, they normally do some sort of an analysis, some sort of a careful analysis, if it's a comparison, to compare one to the other. That's what I'm asking are, that analysis. 733 MR. SOMMERVILLE: I didn't hear Mr. Penny objecting to what you're asking for. I think it was a matter of clarifying it fully for the witness what it is that you're seeking. 734 DR. WEAVER: Yes. 735 MR. SOMMERVILLE: Perhaps you could just restate the question. 736 MR. SHEPHERD: I'm looking for the expert's analysis comparing the two -- 737 DR. WEAVER: What was asked of me originally is had I looked at the Union Gas analysis and I responded yes, in the affirmative. I did not redo their calculation. It's a simple Excel spreadsheet. I stated my opinion on my reading of their analysis, which I did prior to today, and also on the fact that, in doing normal statistical analysis of climatological records, you don't make opinions based on climatological norms. 738 I'm in the business of looking at climate change forecast. We incorporate trend analysis in this. It's fundamental. It's fundamental to seasonal prediction, and it's fundamental to any discussion of climate change. It is a discussion of trend. That's where my expert opinion is coming from. I was not involved in the development of their tool. I was involved in the validation after the fact, by being asked whether or not it seemed reasonable to me based on my understanding of the climate record. 739 MR. SOMMERVILLE: Thank you. 740 MR. SHEPHERD: So you read their evidence and you agree with it, is what you're saying. 741 DR. WEAVER: I said I have read the evidence and I found no reason to disagree with it. I mean I -- 742 MR. SHEPHERD: Okay. Let me turn to you, Mr. Fogwill. 743 If I understand the company's evidence correctly, there's two primary reasons why you want to change your weather method. First, you believe there's a long-term trend towards warmer weather, and that's why we're talking about global warming, for example; and second, you believe that based on the past data, the existing method is not sufficiently accurate and you want to move to a method of greater accuracy. Do I have those -- are those the two reasons you want to change? 744 MR. FOGWILL: Well, those are two of the main reasons, yes. And if I can just sort of rephrase your initial statement, it's the symmetry around the heating degree days that we're actually looking for. So the trend just seems to be the method in this case, that gets a more symmetrical performance. 745 MR. SHEPHERD: So if the 30-year average was more accurate but 20-year trend was more symmetrical, you would prefer 20-year trend. 746 MR. FOGWILL: Yes. 747 MR. SHEPHERD: I'm sorry, I'm -- okay. So let's start with how these methods reflect trends. It's true, isn't it, that both of these methods reflect trends. It's just that regression trend analysis reflects trends more quickly. 748 MR. FOGWILL: Yeah, that's true. The average methods will always pick up on the underlying behaviour of the data, but the trend methods get there right away. 749 MR. SHEPHERD: I wonder if you could look at page 2 of our cross-examination materials. This is a copy of Exhibit J.26-35. And on the second page, and you've referred to this already today, is a list of -- and this is your numbers, right, actual heating degree days, 30-year average and 20-year trend projections for each of those years? 750 MR. FOGWILL: That's correct. 751 MR. SHEPHERD: These are your numbers, not somebody else's. 752 MR. FOGWILL: That's correct. 753 MR. SHEPHERD: Let me clarify one thing. 2003, that number that is 4,033, that's a six and six number; right? 754 MR. SHEPHERD: That's correct, yeah. 755 MR. SHEPHERD: And the first six is actual. 756 MR. FOGWILL: Yes. 757 MR. SHEPHERD: And the second six is based on 20-year trend. 758 MR. FOGWILL: That's correct. 759 MR. SHEPHERD: Now, 20-year trend currently projects substantially lower than 4,000 heating degree days per year, doesn't it? 760 MR. FOGWILL: That's correct. 761 MR. SHEPHERD: So the forecast component of that is already pushing that number downward relative to the first half of the year; correct? 762 MR. FOGWILL: That's right. And that's why we don't really like to use that year in terms of any analysis for the methods, because it's -- we just want to stay with the actual observed information. That has a weakness in it because of that forecasting component, and it wasn't part of our analysis at all. This was requested in the interrogatories and so we put it in there as asked. 763 MR. SHEPHERD: Well, it's also a year in which the 30-year average is a lot closer to reality than the 20-year trend, isn't it? 764 MR. FOGWILL: The -- no. The 30-year average and the 20-year trend are always going to be a significant gap between those two. 765 MR. SHEPHERD: Sorry, I misstated my question, Mr. Fogwill, my apologies. The 30-year average projects 4,046 heating degree days in 2003; the 20-year trend projects 3,631 heating degree days in that year. And it appears fairly clear that this year, the 30-year average is going to be closer to reality than 20-year trend, isn't it? 766 MR. FOGWILL: At this point it's hard to say because we've got several heating months ahead of us, October, November, and December, and in this case we've only got six months of actuals. So we only have -- we can't draw the conclusion that it's going to be closer to the 30-year average or the 20-year trend. But I guess the other point I want to emphasize is, just looking at one year is not an appropriate way of evaluating which method it better. You have to look at it over a longer period of time, and I already mentioned the five to seven years. So it's -- the performance of the method over time that's important in each individual year, we're not necessarily going to be able to predict what the actual weather is going to be, or heating degree days, I should say. 767 MR. SHEPHERD: In fact, it's true, isn't it, that the longer the period of time you look at, the more likely the comparison is to be reliable; is that true? If you look at one year, you don't know which method is going to be better; if you look at two, it will be a little more valuable; if you look at a hundred, it will be a lot more valuable. True? 768 MR. FOGWILL: No, that's not necessarily the case. One of the things we have to keep in mind is that over a period of time, if we look at some changes in the climactic conditions that we're facing, looking at it in the past, say from the 1930s to the 1960s or even before that, we're in a different climate condition today, and that's the reason why we asked Dr. Weaver to provide his expert opinion. We're in a different climate; we're moving towards warmer weather. We could read about it in the papers and come up here and say that's the case. But instead of doing that, we brought in an expert who says that is the case. So going back into the past and saying that a 30-year average is a better method does not necessarily look at what the future is. 769 MR. SHEPHERD: Would you take a look at page 4 of our cross-examination materials, Exhibit M.2.3. This is a plot of actuals, which are the blue line, 30-year average predictions, which is the red line, and 20-year trend, which is the yellow line. Does this look right to you? 770 MR. FOGWILL: Yes, it does. 771 MR. SHEPHERD: And so you're going to get some years like 2003, and I understand that it's six and six information, but you're going to get some years where one method is better than the other, and you can pick examples along the way, like 1984, for example. But your feeling is that on balance, that sloped yellow line is more reflective of the long-term weather trend; is that right? 772 MR. FOGWILL: Yes, based on all the information we have right now, the trend will provide a more symmetrical and accurate reference. 773 MR. SHEPHERD: Now, that sloped yellow line appears to have dropped something like 600 heating degree days in 33 years. Is it your evidence that you expect that trend to continue? 774 MR. FOGWILL: I don't know. What we're trying to do is just look at the -- how the trend moves over time, and it's definitely going to change. The slope of that line is going to change as you add one year and drop another year. Next year, it could be, you know, a flatter slope so it's more -- more horizontal; the following year it could be steeper. It really just depends on how the data is -- how the observations are used in the linear trend. 775 MR. SHEPHERD: Mr. Fogwill, if I understand the concept of trends correctly, and I'm not a statistician, believe me, as you'll find out soon enough, a trend line has two characteristics; it has a direction and a slope. Is that right? 776 MR. FOGWILL: That's correct. 777 MR. SHEPHERD: And the direction, it's just whether it goes up or down. If it goes up, it means increasing heating degree day, the weather is getting colder, and if it goes down, it means decreasing heating degree days and the weather is getting warmer; is that right? 778 DR. WEAVER: Just a small clarification. It's only got one. It's just a slope. The sign of that number determines whether it's up or down. It's not a slope in direction. It's slope with a plus or minus sign in front of it which will give you direction. 779 MR. SHEPHERD: I understand from a mathematician's point of view that's true, but I'm just a layman here and it looks to me like there's two things there and so bear with me if I try to deal with them separately. We can call the first one the sign rather than the direction, if you'd prefer. 780 That's correct, right? The direction is either warmer or colder -- sorry, warmer or colder. 781 MR. FOGWILL: Yes. Yes. 782 MR. SHEPHERD: All right. And the slope is -- that's just the steepness of the line and that just reflects how quickly it's moving in a particular direction; right? The speed at which the weather is getting either warmer or colder. 783 MR. FOGWILL: It's an awkward characterization, but yes. 784 MR. SHEPHERD: Is it correct that a past data set of 10 years is not going to be able to capture a trend that's longer than 10 years, is it? 785 MR. FOGWILL: Well, a 10-year trend would be limited to the time period you're using, so yes. 786 MR. SHEPHERD: And so, for example, Dr. Weaver, you were concerned about the Pacific Decadal Oscillation and said that a hundred years of data wasn't enough to effectively determine whether there's a 50-year oscillation; is that right? 787 DR. WEAVER: That's absolutely correct. 788 MR. SHEPHERD: And that would be true of this sort of trend as well, wouldn't it? In terms of heating degree days? 789 DR. WEAVER: Hang on. What do you mean by this? What do you mean by "would be true." You have to clarify. 790 MR. SHEPHERD: Well, if you have three years of data, you can't determine from that whether there's a trend in heating degree days, can you? 791 DR. WEAVER: You need two points to draw a straight line, so you can get a straight line with two points. Three points will give you a straight line and you start to get an estimate of how well it fits, which is basically a correlation, the slope. 792 MR. SHEPHERD: Yes. 793 DR. WEAVER: As you have more points, you have a correlation coefficient which you hope remains high, and you can say it correlates well with some degree of difference. You can draw a straight line between any two points. The question is how significant is that straight line, and the significance of the straight line is measured by looking at the spread of other points around that straight line. 794 MR. SHEPHERD: I understand that. So in the same way as you didn't think the Pacific Decadal Oscillation is useful because you only have a hundred years of data, so if we only had three years of data on heating degree days, we wouldn't be able to -- 795 DR. WEAVER: With three years of data, I would not want to go into the future, unless you have an underlying physical understanding of what caused those three years of data to give the results they've given. Then you have a predictive tool. In this case, the scientific community understands why the curve is trending up, and the understanding is because of radiative forcing associated with greenhouse gases. Therefore, we are able to make predictions using dynamical models and simpler models as to the warming of this trend. 796 That's the difference between the PDO. There's no predictive skill in that, there's no theory, so to speak. It's really an index. It's just a measurement of how warm or cold the North Pacific temperatures are. So it's a different kettle of fish. We have an underlying physical understanding of the cause of this trend, and Allan here just calculated it. We know why it's there. This isn't applicable to the PDO. 797 MR. SHEPHERD: Now let's get back to direction versus slope. What you told us is that we know that the line is going to slope downwards, the heating degree days line. The temperature line is going to slope upward. We know that; we have lots of data about that. You haven't told us anything about how quickly it's going to slope, have you? 798 DR. WEAVER: Actually, in my evidence I believe I did say it's goings at .13 degrees per decade globally, and that all I can say off the global scale, on the regional scale applicable to Ontario, is it's going to be amplified, especially in the winter, but all year round in Ontario. So all I know, it will be greater than .13 degrees per decade in Ontario. That's all I can give you as a best estimate. 799 MR. SHEPHERD: It hasn't gone up 2.3 degrees in the last 33 years, has it? 800 DR. WEAVER: No, it hasn't. Well, I don't know. I mean, I haven't got the Ontario -- well, let me see if I can find the Ontario data here. I mean globally -- well, let me get the exact numbers from my evidence that I put forward. 801 That trend of .13 per decade, I'm trying to get exactly over the years with which it was done globally. In 10 years it would be 1 degree. I have to look at the numbers here just for a second. 802 That number is coming from 1976 or 19 -- the age of the satellite era, I believe the year 1979, to the year 1999, 2000, somewhere in there, and it's based on the global temperature record, which I showed in figure 3(a) on page 12 of my report, which showed an increase of about .6 degrees over 20 years. So that's -- let me just calculate this. The number .1 per decade is a number that's in the report and sits in my head. 803 MR. SHEPHERD: Dr. Weaver, if it's easier, we could leave this until after the break. 804 DR. WEAVER: Yeah. I would like to check -- I would like to come back to that because I'd like to go through the technical summary to see where I got the numbers from and how it was obtained. It's difficult to calculate this on the spot and I want to do it right. 805 MR. SHEPHERD: I wouldn't ask you to. If it's possible, then, Mr. Chairman, I don't think we need an undertaking if he's just going to calculate it on the break, if it's possible, I wonder -- 806 DR. WEAVER: It would take five minutes of peace, without the pressure. 807 MR. SHEPHERD: Just tell us, as closely as possible, how much the temperature is warmer this year than in 1971. 808 DR. WEAVER: 2003? The year is not over. 809 MR. SHEPHERD: Last year. I don't care. 810 DR. WEAVER: 2002. Oh, I can tell you -- 811 MR. SHEPHERD: Not actually, but according to the projections of climate change overall. I'm not looking for heating degree days each year, I'm looking for the trends that you're seeing. 812 DR. WEAVER: We project future, we use historic data and then project future. So how do you mean... 813 MR. SHEPHERD: Well, you have a trend line for the last 33 years. 814 DR. WEAVER: That's based on the observed data. 815 MR. SHEPHERD: Okay. But if 33 years ago was a particularly cold year and last year was a particularly warm year, that doesn't tell you much, does it? 816 DR. WEAVER: No, you're creating a line between two points. 817 MR. SHEPHERD: That's right. So that's not useful. I'm asking for the useful line, not the unuseful line. 818 DR. WEAVER: Okay. I'll look that up on the break. 819 MR. SHEPHERD: Thank you. I guess, Dr. Weaver, let me ask you one more question with respect to this sort of area of trends generally. When you were being asked -- when you were being questioned by Mr. Janigan, you said that there was no reasonable possibility that, in the next 50 years ago, there would be 30 years of cooling similar to 1946 to 1975. Why is that? 820 DR. WEAVER: Because what drives the climate system -- there's a number -- there's inherent natural climate variability and there's things -- which are things like El Nino and things like that, and then there is radiative force, that is things that change the balance of energy between the amount of energy received by the earth from space and the amount of energy the earth gives out to space. 821 MR. SHEPHERD: Yes. 822 DR. WEAVER: Okay. Unless -- given that we're not going to stop consuming energy tomorrow, there will be continued increase of greenhouse gases into the atmosphere which will continually provide a radiative forcing that will continue to warm. So globally, every possible scenario, using every model in the world, indicates that it will continue to warm over the next century; that is, every model in every country, using every scenario of economic growth, technological growth, population growth, et cetera, every one of them says it will warm this century. 823 MR. SHEPHERD: And that projection is between 1.4 and 5.8? 824 DR. WEAVER: That's the 1.4 to 5.8. But again, the spread in that, because there are some models that are -- you know, it a United Nations process and if you are a UN member state, your model will be included. So I would tend to throw out, based on my expert opinion on what the particular model is doing, some of the extremes. 825 But do I believe the 5.8? No, I don't. As I said, my personal best estimate is around 2 to 3 degrees of global warming relative to 1900. That's my feeling. That's my estimate. So that would give 2 degrees in a hundred years is .02 degrees per decade; is that right? 826 MR. SHEPHERD: That's .2 degrees. 827 DR. WEAVER: I got a degree in mathematics. 828 DR. WEAVER: So that's 2 degrees in a hundred years is the best estimate, in my opinion, as to what the warming will be. And that's on the lower end, and I think some of the people -- some of the more skeptical people in the community would tend to be on that 1.4 end, or the lower end, and some of the more extreme people would tend to argue for the positive feedbacks through things like permafrost melt and stuff that we don't really understand could amplify this in some crazy way and give you the higher end but I personally don't believe that that's the case. 829 MR. SHEPHERD: I'm not disagreeing with the long-term trend, Dr. Weaver. I'm asking a more specific question. You've given evidence that, in any given period of time there's variability. 830 DR. WEAVER: Yes, I have. 831 MR. SHEPHERD: And you might have a trend, but around the trend you have variability. So I don't understand why the variability, which we already saw in the last century, in '46 to '75, why that couldn't arise again in the next 50 years. 832 DR. WEAVER: I'll tell you why. Okay. In the last 10,000 years, we have very good records from ice cores, from pollen records, from ocean sediment records, on and on and on, tree rings, bore hole temperatures. These records have allowed us to assess what the magnitude of climate variability has been over the last 10,000 years. Globally averaged, the peak amplitude of that variability is about half a degree. That's globally averaged. 833 The warming in this century is already above that. It's already -- I mean you can see that in the hockey stick figure which I put -- figure 3(b) in my test, my piece, which is on the -- my report. That's an early reconstruction. That was the figures one done. It received some criticism because it was an early one, and it's a normal processing in science, people question. That's on page 12. 834 Since that time there have been many more reconstructions, all of which come with very similar, slightly different wiggles, similar conclusions, similar statements. The maximum amplitude of climate variability on a global scale is about half a degree, maybe .6 max, globally in the last 10,000 years. It is not appropriate to make any comparisons earlier than 10,000 years, things like massive abrupt climate change and so on and so forth, because these happened in a climate which was fundamentally different from today where there would be 3 kilometres of ice on top of us which doesn't exist as of today. 835 Climate variability fundamentally depends on the mean climate state itself; that is, cold climates are much more unstable than are climates like the Holocene. 836 So I can say that there is no known mechanism that can create anything which will give global cooling on the order of half a degree, sustained for 30 years, when we're talking about a 2-degree -- and that can't dominate the projected temperature increase of 2 degrees over this century. 837 So the maximum we can get is half a degree cooling through any mechanism of climate variability we've known that it's existed in the last 10,000 years. So you add half a degree to a 2-degree warming, it's still going to be a blip such that it will still continue in that direction, and maybe a maximum even at the extreme case of half a degree below, in my opinion. 838 MR. SHEPHERD: But then remind us how did it happened in 1946. We were already into global warming in 1946, weren't we? The effect was already known then. 839 DR. WEAVER: There are two things. The science of climate change goes back to Jean Baptiste Leguerrier, 200 years ago. It's trivial physics, right. It's -- you put in the atmosphere a gas that is radiatively active, that is, it absorbs outgoing radiation and reradiates it back, but it's transparent to incoming solars; hence the so-called greenhouse effect. 840 In 1946, we did have some -- we had warming going on there. Industrial emissions had gone up, but so had the emissions of things like sulphate aerosols, and they went up quite dramatically with industrialization. These are the aerosols that are predominantly produced, the things like coal burning, and they act to cool the planet. In addition, there are periods in the twentieth century record when there's more volcanic activity than other periods. When there's more volcanic activity, it tends to put in what are known stratospheric aerosols into the atmosphere. These last a little bit of time, maybe a year and a half max, and then they fall out and they are rained out to the ground. 841 Pinatubo was the last such big event in 1991, which is the reason why solely that 1991 does not make it into the top ten years of warmest years on record. Every one of the other '90s years is in the top years on record. 1991 is just off. It's because of Pinatubo. 842 Similarly, there's fluctuations in the intensity of the sun. They are very small, but they can have a small effect. And what we're seeing in this record is the additive effect of all the different radiative forcing. These are things like greenhouse gases, aerosols, changes in the sun, changes in the volcanic emissions, and change in landcover, agricultural use. These combine together to provide a time series of radiative forcing which, when you drive the climate models are which are used to make the future projections, they give a very reasonable reproduction of the actual trend that we've observed. 843 So we have a good understanding of why this twentieth century record looks the way it does. 844 MR. SHEPHERD: I'm just trying to take that in. It sounds like, and tell me whether I'm misunderstanding this, it sounds like what you're saying is that over the last, I don't know, 50 years, things have happened which prevent us from having, in the future, any significant long-term cooling period. 845 DR. WEAVER: You will not get another ice age for another 50,000 years without any of this greenhouse effect -- 846 MR. SHEPHERD: Dr. Weaver, I'm sure that Union doesn't really want an ice age. We're talking about much smaller cooling amounts. 847 DR. WEAVER: I mean, unless there were a large number of volcanos going off systematically within a few years, I cannot envision any mechanism by which we could go through cooling, none at all. We look at -- I mean the evidence is bountiful. Mountain glaciers are a lovely example of integrating the daily or interannual, the long time scale variability. 848 The reason why they're good examples is they don't come and go in a year, they take a long time. Every single mountain glacier in the world, with the exception of a few minor ones in the North Atlantic, are melting. Places like Kilimanjaro, there won't be ice in 20 years. You just have to go to the Columbia Icefields and look at the stakes to see how the melting is occurring. This is happening rapidly. This is happening much more rapidly than it had happened in the past, and we know why. This is going to continue into the future. 849 Again, I need to say I like to read it out, but it's the David Letterman top 10 list. It's my top 10 list. Top 10 warmest years in a thousand-year record are: Number 1, 1998; number 2, 2002, despite what the prediction came from this earlier submission for one of the coldest years on record from the Salomon Barney Report, 2002 was the second warmest year globally on record. 2001 was the third globally warmest. 1997, the fourth; 1995, the fifth; 1990, the sixth, 1999, the seventh; 2000, the eighth; 1991, the ninth; 1987, the tenth. 850 It just reads like that. You can see it in every figure. We know why. There's a sense of denial out there as to why, but the scientific community knows why and we know why it's not going to stop. It's as simple as that. Simple science, simple physics going back to Arhennius a hundred years, and Flores 200 years before that. I don't know what more I can say. 851 MR. SHEPHERD: Well, as I said, I wasn't disagreeing with global warming. However, I am trying to understand the impact of this on the variability of climate year to year, and I understand you to be saying the variability is going to be reduced. 852 DR. WEAVER: No. The scientific community is split in terms of how it will affect the most important mechanism of climate variability, that is, El Nino. Some have argued, and the arguments make sense, that El Ninos will become more frequent; others have argued that, you know, they might become permanent, and some will say no change. So there's not a consensus, although it's leaning towards, you know, more El Ninos, but there's not a consensus there. 853 What I can say is the likelihood of very, very strong El Ninos would be increased and the likelihood of very strong La Ninas, the opposite, would decrease. That is, the likelihood of the warm episodes will become greater and the likelihood of the cold episodes will become less, and that's simply because you're putting the variability on a trend and the cold episode is down here and the warm episode is up here, and it's higher than the reference. 854 And even if there is no change to the mechanism of El Nino itself, the likelihood of greater and stronger El Ninos is increasing and the likelihood of greater and stronger La Ninas is decreasing. And you can see that because the two strongest El Ninos on record, and that's in a long, long time series, that goes back into the proxy record, is 1982/'83, 1997/'98. These two El Ninos were the two strongest on record. 1992, '94, I forget the exact years, was basically a very long-lived El Nino. Again, these are all anomalous in the recent record. 855 MR. SHEPHERD: If the long-term variability, like 1946 to 1975 sort of period, is no longer possible, does that also apply to shorter-term variability like you talked about the North Atlantic Decadal Oscillation. Is that going to stop having an impact on cooler versus warmer trends? 856 DR. WEAVER: That's an interesting one. That's actually a research area. The variability on the decadal time scale in the North Atlantic, it's something I was involved in back in the '90s. There's some thought that sea ice plays a crucial role in that, and what we do know is that aerial extent of sea ice is significantly decreasing, well observed with satellite data. 857 So it's not clear how that will evolve. It's not clear to me, it's not clear -- some will give strong opinions one way or the other. I tend to resist -- I mean, I've written papers arguing that decadal variability in the North Atlantic will reduce because sea ice extent is reducing. 858 And it's in my CV, the Holland et al paper in GRL, Geophysical Research Letters, a couple of years ago, if you want the reference. We argued that decadal variability in the North Atlantic will reduce as a consequence of the retreating sea ice because we argued that the source of that is random fluctuations in the amount of fresh water being emitted to the North Atlantic from melting sea ice. When you have less of the sea ice there, you have less variability in the amount of fresh water that's going into the Atlantic on a year-to-year basis. 859 But again, I would just argue that there's no consensus on that. I know maybe I'm confusing you, but I'm just trying to state -- 860 MR. SHEPHERD: You are definitely. Much more detailed than my mind can handle. I was asking a much simpler question. You've given evidence that there's no reasonable likelihood that there will be a 30-year period of cooling in the foreseeable future. 861 DR. WEAVER: Absolutely. 862 MR. SHEPHERD: The long term. Is it also true that there is no likelihood that there will be a 10-year period of cooling, or are there still effects that could cause that? 863 DR. WEAVER: I mean, this is where you come back to what do you call cooling, relative to what? Now, are we talking about relative to the 1961 to 1990 average? 864 MR. SHEPHERD: Cooling relative to current trend. 865 DR. WEAVER: To the current trend. Ten years, I would say not. Five years, perhaps, but very unlikely. 866 MR. SHEPHERD: All right. 867 DR. WEAVER: One year, yes. 868 MR. SHEPHERD: Okay. Now, you've said that in the twentieth century there was a 0.6 degree warming trend over the globe; correct? 869 DR. WEAVER: Yes. 870 MR. SHEPHERD: Mr. Fogwill, what does that translate to in terms of heating degree days? 871 MR. FOGWILL: 0.6 over -- 872 MR. SHEPHERD: Over a hundred years. 873 MR. FOGWILL: I'd have to do the calculation and get back to you on that. 874 MR. SHEPHERD: It's something like 1.6 heating degree days per year, isn't it, in that range? 875 MR. FOGWILL: For a hundred years? 876 MR. SHEPHERD: 0.6 times 365 heating days divided by 100. I'm in the right range, aren't I? 877 MR. FOGWILL: Yeah. 878 MR. SHEPHERD: And you had a discussion earlier with Mr. Brett about the next decade -- the next century, sorry, and I understand, Dr. Weaver, that you're saying that the range is 2 degrees to 3 degrees warming over the next hundred years. 879 DR. WEAVER: In my opinion, not according to the IPCC assessment. They would argue 1.4 to 5.8. I would argue 2 to 3. 880 MR. SHEPHERD: You're the expert we have in front of us. 881 DR. WEAVER: Okay, so I'll say 2 to 3. I mean, but there's -- 882 MR. SHEPHERD: And will you confirm or accept, subject to check, Mr. Fogwill, that 2 degrees over a hundred years is 4.8 heating degree days per year? Again, we're in the ballpark; right? 883 MR. FOGWILL: Yes, I'll accept that, subject to check. 884 MR. SHEPHERD: Now, as I understand what you're saying, Dr. Weaver, you're saying that the impact of these various, what did you call them, amplifications for southern Ontario, the fact that global warming is stronger in the winter, stronger over land, and stronger in high latitudes is going to be something less than 40 percent of the projection; right? 885 DR. WEAVER: Let me check. That's seasonally dependent, too, that might be with reference to the annual mean. But then you have to amplify it again in the winter, because typically the warming is amplified in the winter and it's amplified where you are in the hemisphere latitudinally. Let me just pull this figure out here. 886 MR. SHEPHERD: I'm on page 25 of, I guess this is your report. Is this your report? On page 25 of your report you say that: 887 "In areas that are much greater than average warming, it's more than 40 percent above the global average." 888 DR. WEAVER: Right. 889 MR. SHEPHERD: And you've said in testimony today that southern Ontario is not going to be in that category, it's going to be in the next category down. So it's not going to be more than 40 percent, is it? 890 DR. WEAVER: Right. That's correct, yeah. 891 MR. SHEPHERD: So the maximum then, if your 2 degrees is right, then -- 892 DR. WEAVER: Well, it's more in the winter and less in the summer. I don't have the exact numbers in front of me, but 2 degrees would be something like 4 in the winter and 1 in the summer, something like that, and maybe transitions in the spring. So it's amplified in the winter. Then it would be 40 percent on the 4 degrees, not 40 percent on the 2. 2 degrees is an annual mean average and it's averaged everywhere in the globe. You have to amplify that in the winter relative to the summer, and you have to amplify that in Ontario region relative to non-Ontario region. 893 The warming is maximum where the feedback is strongest, the ice/snow albedo feedback is strongest in the winter and transitions and so it amplifies warmer as you get less snow cover you warm further. But snow isn't an issue in summer temperatures, but it is an issue in winter temperatures and early spring. If you have snow, the sun is being reflected and so it's cool. If you don't have the snow, it's absorbed and it warms further. That's the reason it's amplified in the winter/spring. 894 So what I'm getting at is the 2 degrees globally averaged is globally averaged and annually averaged, so that has to be amplified in the winter and amplified in Ontario. So it's much -- if you would like to pick a particular model, you can pick the winter months on figure 24 here from the Canadian Climate Centre model, this is the Canadian projection, for the years 2041 to 2060, showing between -- this is about three to five -- I would say this is in the five degrees area for this -- 895 MR. SHEPHERD: Dr. Weaver, where are you? 896 DR. WEAVER: Page 24 on my report, right on the other side of figure 11. 897 MR. SHEPHERD: Yes. 898 DR. WEAVER: If you look at that picture, and you want to look as an example of a number, this is -- this would say this particular model under this scenario would suggest 5 degree winter warming in the years 2040 to 2060 relative to the 1971 to 1990 average. So they would suggest it would be 5 degrees warmer in winter there. If you look in the summer it's slightly less. It's probably closer to 3 degrees there. 899 MR. SHEPHERD: This is 2040 to 2060, so this is probably a later rate case. I'm wondering whether you have more current information. I think it would be very useful to the Board to know what your projection is of warming over the next 10 years in Southern Ontario, both winter and summer. 900 DR. WEAVER: Well, again, you're asking me to do something that I've said already in evidence that is not able to be done; that is, quantify with a real number, a raw number, the amount of warming over 10 years. To be honest, if I could do this, I'd be a very wealthy person. You can't, because this is asking for a level of detail and a specific number that is not possible to give. I can only tell you above normal versus below normal. That's all I can tell you 901 MR. SHEPHERD: You can only tell us direction, not slope. 902 DR. WEAVER: No, and that's the reason why when you -- when I testified that I thought it was -- what they did, I could see no reason to disagree with it, was that the trend is being modified as you move forward in time to either increase or decrease as you add more data into the system. My understanding of the method Allan Fogwill developed with Union Gas is that changing trends, as you get more data, is incorporated into their analysis. 903 MR. SHEPHERD: Mr. Fogwill, one last thing on global warming. If what you want to do is capture the impact of global warming, why don't you just have an express adjustment to your heating degree days each year to adjust for the fact that global warming is pushing it downwards? Why do you have to go to a completely new methodology? 904 MR. FOGWILL: We looked for an approach that we were -- that would help us incorporate the trends in the observed data that we saw. Global warming was one component of the perspective that we were looking at. And one thing that I was able to see, based on my assessment, is that there is actually no annual or longer term specific number to use. Everything I've seen has always been a range, and every scientist I've ever heard talk about it has always expressed numbers with caveats around it. Trying to establish a number to offset a 30-year average would mean we would have to have some sort of proof or support for that number, and there's just no possible way we could ever find that. 905 So we wanted to look for an approach that took the observed data and actually recognized that there was a change in the weather, in the actual climate conditions, and based on the information we've got from Dr. Weaver and also the work done by Mr. Root to identify trending in a 20-year period as reasonable parameters to use, that's where we came up with the 20-year trend. 906 I guess the other point I'd just like to identify here is that, based on hearing Dr. Weaver talk about the fact of the amplification in Ontario of the global change, and going back to the point that was raised about it being around a 4.5 percent per year adjustment using the global estimate, and if you amplify that in Ontario, that would, I'm guessing, move you in the 10 to 20 heating degree day range per year change, which is, in fact, the slope of the line that we have in our 20-year trend, which is about 15 right now. 907 So there's a consistency between the change in the global temperature projections over time, the amplification in Ontario, and the slope of the line that we are actually using. They're not exact but they're -- they don't seem to be at odds with each other. 908 DR. WEAVER: May I add as well. I would like to add, and I could also address the earlier question you asked me to do during the break because I was able to -- I was on the gun for a significant portion of time there. 909 The trend, as observed in the last -- latter period, which is the period 1976 to today, is .2 degrees per decade. That's the trend that's observed between 1976 and today in the global surface air temperature record, which was .13 in the tropics. In my report I referred to the tropical differences of .13 degrees per decade. .2 degrees per decade times ten decades in the next century is 2 degrees, which is my estimate. I just needed some time to reflect on that. 910 MR. SHEPHERD: That's wonderful. 911 DR. WEAVER: Thanks. 912 MR. SHEPHERD: Mr. Chairman, I'm going on to another area. Did you wish to take a break or is this a little early yet? 913 MR. SOMMERVILLE: I think it's a little early. Let's carry on, Mr. Shepherd. 914 MR. SHEPHERD: I'm happy to plow forward. 915 Mr. Fogwill, the way the 20-year trend method works is that you hit a straight line to the last 20 years of data so that it, in effect, goes down the middle of the data, right, and minimizes the variations in the data around that straight line? 916 MR. FOGWILL: That's correct. 917 MR. SHEPHERD: And then you project that line out into the next year, and wherever it ends up, that's your projection for next year; right? 918 MR. FOGWILL: That's correct. 919 MR. SHEPHERD: So that analysis gives you a direction and a slope that reflects what is the underlying trend in the data. 920 MR. FOGWILL: For that snapshot of the 20-year period, yes. 921 MR. SHEPHERD: Well, you're saying it has predictive value; correct? 922 MR. FOGWILL: No, I have never said that. What we're trying to do is use a method that gives us a reference from which to do planning. It doesn't necessarily predict about what's going to happen in the next year. There's no predictive value in there at all. If we could forecast the weather, we'd be in different positions. 923 MR. SHEPHERD: Well, in fact, I understood you to say this morning that it's not next year that this trend gets right, it's the overall trend over time. Over the next five or six years, it should be fairly close; right? 924 MR. FOGWILL: No, that's incorrect. It's not that the current trend is fixed over the next five or six years. It's that the methodology of readjusting the trend every year is going to get you a more symmetrical and accurate estimate over that period of time. So you're not comparing a fixed equation over many years, you're comparing the results of this equation and then the results of the next equation and then the results of the third equation and seeing how those results fit, because the equation is always changing because you're recalculating the equation. 925 MR. SHEPHERD: So you're saying that the trend in the last 20 years of data has no value in and of itself. It's not expressing an actual trend, is it? It's just, as Dr. Weaver said earlier, it's just some numbers. 926 MR. FOGWILL: It's looking at the last 20 years of observed information and projecting the trend of that 20-year period. The next 20-year period will be a different trend, different slope. 927 MR. SHEPHERD: Yeah, sorry, that wasn't the question. What I'm trying to understand is what is the intrinsic value of this 20-year trend? What is it intrinsically telling you? What is the information it's giving you? Or is it just numbers? 928 MR. FOGWILL: The only observation you can make from the 20-year trend is that it's a warming trend. 929 MR. SHEPHERD: So it doesn't tell you anything about how quickly it's warming, does it? 930 MR. FOGWILL: No, not over a longer period of time. Not one snapshot over a 20-year period. 931 MR. SHEPHERD: But what you're proposing to this Board is that we use it to predict what the weather is going to be like next year, isn't it? 932 MR. FOGWILL: No. Again, we're not predicting the weather next year. We're trying to find a planning estimate that minimizes or maximizes the symmetry of the estimate to the actual over a period of time, as well as reducing the inaccuracy of the estimate over a longer period of time. We're not trying to predict what's going to happen next year. 933 MR. SHEPHERD: Dr. Weaver, when you were talking about the Pacific Decadal Oscillation, I understood you to say that one of the reasons why you didn't like that whole analysis was because, in effect, you don't know why -- there's no particular reason to think that it's happening for a particular cause, and therefore, the fact that the numbers show a pattern doesn't tell you anything; is that right? 934 DR. WEAVER: With the Pacific Decadal -- I wouldn't even call it an oscillation. With the low frequency variability of the Pacific Decadal Index, yes, I believe what you said is correct; that is, it's just changing going up and down. For some years there's more El Ninos, some years there's not, combine that with changes in the radiative forcing that have occurred this century. 935 MR. SHEPHERD: So, Mr. Fogwill, if the last 20 years of data doesn't have any underlying cause why the line is a particular slope -- 936 DR. WEAVER: I have testified that there is a cause, and that cause is increased radiative forcing. We know the reason why there is a warming trend in the record. I have testified many times already on that. We know why there is a warming trend in the record. 937 MR. SHEPHERD: Sorry, Dr. Weaver. First of all, my question is to Mr. Fogwill. 938 DR. WEAVER: My understanding is we can all answer. 939 MR. SHEPHERD: Secondly, my question was not the direction, it was the slope. 940 Mr. Fogwill, my question again is: Does the last 20 years of data show you any underlying cause why a particular slope is the correct trend? 941 MR. FOGWILL: No, it doesn't. All it's saying is that there's a warming trend. It doesn't say what the magnitude of that warming trend is because it's a 20-year period and that slope of the line, which is actually the trend of that 20-year period, will change when you add a year and drop a year. 942 MR. SHEPHERD: I wonder if, Mr. Fogwill, you can turn to page 5 of our cross-examination materials. This is Exhibit M.2.3. All this does, and this is just math, it just extends your current trend line that's predicting 2004 out another 20 years. The trend that you've identified in the last 20 years implies that 20 years from now, we'll be at around 3,200 heating degree days. Is that what you think is going to be the case? 943 MR. FOGWILL: No, no, this is inaccurate in terms of representing our method. Our method is recalculated each year. So the slope of this line is only good for one year, our test year, and then next year we will recalculate it again and we'll get a new slope and it will be good for the following year, so on and so forth. 944 MR. SHEPHERD: Didn't you just tell us it had no predictive value for next year? 945 MR. FOGWILL: We're not identifying it as being used for predictive value, we're identifying it as used for a benchmark for planning purposes. And what that benchmark will say is that when you recalculate the 20-year trend, you're going to get a new slope of the line, and therefore, a new estimate. 946 So we don't ever extend out a 20-year trend equation beyond the time period we need the information to do our planning. 947 MR. SHEPHERD: I wonder if you could turn to Exhibit 26.43. I'm sorry, I didn't include it in the materials. 948 MR. PENNY: That's J.26-43. 949 MR. SHEPHERD: Do you have that, Mr. Fogwill? 950 MR. FOGWILL: J.26-43? 951 MR. SHEPHERD: Yes. 952 MR. FOGWILL: Yes, I do. 953 MR. SHEPHERD: In that you say on your current method it would predict that in 2009 you would have 3,229 degree days. That's what it says; right? 954 MR. FOGWILL: It does say that, but we would never use the current method to project out that far. And this analysis is -- although we produced it in response to the interrogatory, it includes a lot of forecasted information and our method is not based on using forecasted information. 955 MR. SHEPHERD: Sorry, 26.43, you say, includes forecasted information? 956 MR. FOGWILL: Well, it must, because we're looking at a 20-year period, we're going out to 2009, and that's the future. 957 MR. SHEPHERD: Sorry, I'm just getting it back again. This says "heating degree days based on 20-year trend," and it compares it to heating degree days based on the WeatherBank method. So you're saying this isn't what your method would actually project? 958 MR. FOGWILL: No, because our method gets reassessed each year with the 20 years of actual information, and in this case we do not have that. It's an erroneous comparison. 959 MR. SHEPHERD: So you're saying that you, in fact, can't predict, with the 20-year trend, what 2009 is going to be. 960 MR. FOGWILL: That's what I've said all along. We don't even predict what's going to happen next year. All we use the 20-year trend to do is provide a reference point for planning. 961 MR. SHEPHERD: You keep saying you're not trying to predict, but this is a forecast upon which you base rates; isn't it? 962 MR. FOGWILL: Yes. 963 MR. SHEPHERD: So ratepayers actually have to pay money based on this non-prediction, don't they? 964 MR. FOGWILL: The rates are set by this -- using this and turning it into a demand forecast, yes. 965 MR. SHEPHERD: I guess I'm trying to understand what the value of the 20-year trend is. If it's not telling you what's most likely to happen next year, and you don't care whether it's most accurate anyway, and it's not telling you what you expect the trend to be in the long term, then what is it telling you? What's its explanatory value in statistical terms? 966 MR. FOGWILL: Well, first of all, I would never characterize the 20-year trend in the terms that you've used. What I would say is that what we're trying to do here is to provide a reference that reduces the variability for our planning process year after year, that variability that's inherent in trying to assess what the weather is going to be year after year. If we used a forecasting method that was a pure forecast, which is some of the services that WeatherBank provides to different clients, that number could vary significantly year over year. That's not the approach we take to our planning process. We're trying to reduce the variability associated with weather over the longer term, and the 20-year trend provides us with that estimate. 967 MR. SHEPHERD: Well, the reduction of variability, that's the test of stability; correct? The method is less variable, it's more stable. 968 MR. FOGWILL: Well, no, the whole premise of weather normalization is to minimize the impact of the weather year by year in your planning process, to help improve your ability to plan in terms of your demand forecast, your gas supply, and other aspects of the company. 969 MR. SHEPHERD: So when you say "variability," then, you mean error level. 970 MR. FOGWILL: I'm talking about weather variability here. 971 MR. SHEPHERD: And variability means the difference between your forecast and actual; correct? 972 MR. FOGWILL: No. It means the difference year to year of what actually happens in terms of weather. 973 MR. SHEPHERD: And how does a forecasting method -- sorry, a planning method affect how the actual weather varies from year to year? 974 MR. FOGWILL: It doesn't affect how the actual weather varies from year to year. What it does do, it gives you a benchmark to use in your planning instead of trying to guess or forecast what the next year's weather is going to be. So if the actual weather from one year to the next varied by, I don't know, 400 heating degree days, you'd have a significant change in all your plans and need to readjust a lot of your activities because of this large swing in heating degree days. And what we're trying to do with the weather methodology is define a condition under which people use as a planning assumption. That's it. 975 MR. SHEPHERD: I'm still confused, I'm sorry. If variability is not how accurate your method is, then I don't understand the example you just gave. It's not actual weather, you're not trying to control the actual weather, you're trying to get your forecast as close to actual as right, is that correct, so that you can plan correctly. 976 MR. FOGWILL: Over a longer period of time, yes. Not over a one-year period. 977 MR. SHEPHERD: Just as sort of a mathematical truth, tell me whether this is true: If you use a regression-based trend method, then if there are short-term trends in the data, a shorter data set will exaggerate those short-term trends; isn't that right? 978 MR. FOGWILL: Say that once more. 979 MR. SHEPHERD: If there are short-term trends in the data, then a shorter data set will exaggerate those short-term trends; is that correct mathematically? 980 MR. FOGWILL: No, not necessarily. If you have a short-term trend in the data and the period of your analysis is longer than that trend, it's going to be affected by those observations that are outside that trend. It's very difficult to say how those additional observations will affect that short-term trend. And similarly, the case is if you have a short period of time with a trend in that data and you take a small subset of that, you actually -- you don't know how that is going to represent that trend line. 981 MR. SHEPHERD: Mr. Fogwill, please take a look at page 4 of our cross-examination materials. If you used five-year trend analysis, you'd have a completely different planning assumption for next year than if you used 10 years or 15 years or 20 years; correct? 982 MR. FOGWILL: Yes. 983 MR. SHEPHERD: It would be in a different direction; right? 984 MR. FOGWILL: It definitely could be. 985 MR. SHEPHERD: If there is a short-term trend, in that case five years, there's a trend upwards in the data, the shorter data set will exaggerate that trend and will miss the longer trend; correct? 986 MR. FOGWILL: Can you repeat the question again for me, please. 987 MR. SHEPHERD: If you use a shorter-term trend analysis, then that will exaggerate the short-term trend and will miss any longer-term trends; correct? 988 MR. FOGWILL: The fewer data points that you use, the more uncertainty there is going to be in the resulting trend from the equation matching the trend in the longer data set. 989 MR. SHEPHERD: Sorry, I don't understand how that was responsive to the question. Could you try again? 990 MR. FOGWILL: Okay. If you have a -- if you have a period of data and you're taking a smaller period in order to do the trend analysis, the correlation between the trend that you get from that calculation and the inherent trend in the data will be low, meaning that they won't match that well. The smaller the number of points you use, the more mismatched the two trends will be, the calculated one versus the one that's actually in the data. 991 MR. SHEPHERD: Is it the smaller number of points or the shorter the period? 992 MR. FOGWILL: Essentially it's the same thing. The fewer points. For example, you take the five years you were talking about, you take the five years and say, I'm going develop a trend analysis on that five-year period, and you get a certain result for that trend, that trend will not match a trend that may be inherent in a 20-year period, for example. And if you had the 20-year period and you did an analysis on 10 years, there's a stronger likelihood that the 10-year trend calculated will match the 20-year trend actual. 993 MR. SHEPHERD: Sorry, I hate to disagree with you since it's your evidence, but am I correct in saying that it's the time period that you're covering, not the number of data points, that is relevant, because, for example, if you picked years 1, 4, 7, 13, 18, and 20, that would be more likely to capture the long-term trend than picking years 15 to 20 in a 20-year set; right? 994 MR. FOGWILL: If you're using the same number of points, then the accuracy or the match between the trend that you calculate and the inherent trend that's in the observed data will not differ. So if you use 1, 12, 14, 18, if you have five data points, you still will have the same uncertainty associated with the trend that you calculated from those five data points as you will if you take the 15 to 20. 995 MR. SHEPHERD: It won't go in the wrong direction, will it? 996 MR. FOGWILL: It could, yes. 997 MR. SHEPHERD: Only if it's a random walk; isn't that right? If there's an underlying substantive trend in the data, that can't be the case, can it? Just look at this example and tell us how that could possibly be the case. 998 MR. PENNY: Well, there are a number of questions that were just asked. I'm not sure which one Mr. Shepherd wants answered. 999 MR. SHEPHERD: Mr. Chairman, I withdraw the question. I'm not getting anywhere on it. 1000 I'm going into a completely new area, so if this is convenient, Mr. Chairman. 1001 MR. SOMMERVILLE: I think it is convenient. We will adjourn until 25 minutes to the hour. 1002 MR. SHEPHERD: Mr. Chairman, you may be interested, I have about 30 to 35 minutes left. 1003 MR. SOMMERVILLE: Is there -- are there other intervenors who wish to cross-examine this panel? 1004 MR. AIKEN: Yes. 1005 MR. SOMMERVILLE: How long do you expect to be, Mr. Aiken? 1006 MR. AIKEN: Probably in the neighbourhood of 30 to 45 minutes. 1007 MR. SOMMERVILLE: And others? 1008 MR. RYDER: Yes, I'll be about 20 minutes. 1009 MR. THOMPSON: Yes, Mr. Chairman, about 30 minutes for IGUA. 1010 MR. SOMMERVILLE: If we add that up, we're over time. I haven't -- I'll ask one of the panel, perhaps, to add it up for me. I'm being facetious, Doctor. 1011 DR. WEAVER: My math isn't very good. 1012 MR. SOMMERVILLE: It looks to me that we're probably going to go over to tomorrow morning. If that's not an inconvenience to the -- you have two witnesses from out of town here, Mr. Penny. 1013 MR. PENNY: Yes, but they're available for both days. But based on the assessment that Mr. Moran did yesterday, we actually fully expected that we would not get done today. 1014 MR. SOMMERVILLE: In that respect, what I think we may look to is to conclude Mr. Shepherd's cross-examination today and leave the cross-examination of Mr. Aiken, Mr. Thompson, and Mr. Ryder for tomorrow. 1015 MR. PENNY: Does that mean your intention was to break at roughly 4:00 then today? 1016 MR. SOMMERVILLE: That would be the -- 1017 MR. PENNY: Because we could get probably another half an hour in there anyway. 1018 MR. SOMMERVILLE: We'll go until 4:30, and that's holding you, Mr. Shepherd, to your estimate now and, Mr. Aiken, you to yours. 1019 MR. AIKEN: Sure. 1020 MR. SOMMERVILLE: And we'll proceed tomorrow morning with Mr. Ryder and Mr. Thompson. Is that convenient to each of you? 1021 MR. THOMPSON: Yes, that's satisfactory. Thank you. 1022 MR. SOMMERVILLE: Thank you. We'll break until, say, 20 minutes to the hour. Thank you. 1023 --- Recess taken at 3:25 p.m. 1024 --- On resuming at 3:45 p.m. 1025 MR. SOMMERVILLE: Thank you. Please be seated. I see that the word has got out that the gentlemen may remove the jackets, and ladies too. 1026 Mr. Shepherd. 1027 MR. SHEPHERD: Mr. Chairman, just to clarify the timing, perhaps, Mr. Chair, it would have been wiser to go to the panel for the math, because I think if I'm 30 to 35 minutes, I think Mr. Aiken will probably not have enough time to finish by 4:30. 1028 MR. SOMMERVILLE: We'll see where we get. If we all tighten our belts, we might just make it. 1029 MR. SHEPHERD: Mr. Fogwill, can you turn to page 1 of our cross-examination materials. This is a copy of an interrogatory response, J.26-36. You had a discussion with Mr. Brett this morning with respect to the weighting of the various tests that you used to compare these methodologies. And I take it that it's correct that if you take out all of the weightings and the double counting, that the 30-year average and the 20-year trend are pretty close, 18 as opposed to 20 points on your scale. 1030 MR. FOGWILL: The 20-year trend is at 22 points; right? 1031 MR. SHEPHERD: I think that's only if you add another competitor to the mix, being 40-year trend. 1032 MR. FOGWILL: I'm sorry, you're looking at the numbers above there. Yes, that's correct. 1033 MR. SHEPHERD: And it's true, isn't it, that if you design a marking system to compare anything, you can design it -- and I'm not saying that you did -- but you can design it so that anyone wins really, just by weighting the criteria, couldn't you? 1034 I'm not saying you did, I'm saying you could. 1035 MR. FOGWILL: It's possible. But I think what's important to note is that what we try to do is be very transparent in how we selected the method. We didn't go into the selection of the method saying that we want to prove that the 20-year trend is the one we want to go with. We went in with an open mind saying which is the best method to predict or, I should say, give as a longer-term reference for planning over a five- to seven-year period. 1036 MR. SHEPHERD: I wonder if you could turn to page 7 of the cross-examination materials. Mr. Fogwill, this is a set of calculations of your numbers from J.26-35, and I'm going to ask you a series of questions about them. You had this in advance. I wonder if you've had a chance to check these calculations or not? 1037 MR. FOGWILL: Yes, I've checked them. 1038 MR. SHEPHERD: And are there any errors? 1039 MR. FOGWILL: Yeah, there's a few. 1040 MR. SHEPHERD: Can you point them out, then. 1041 MR. FOGWILL: I have different numbers for the mean percent error, for the 20-year trend, for the raw comparison, for the '71 to '84, and '85 to 2003. 1042 MR. SHEPHERD: Sorry, slow down a bit. For mean percent error? 1043 MR. FOGWILL: Mean absolute percent error. 1044 MR. SHEPHERD: MAPE, okay. For which year? 1045 MR. FOGWILL: '71 to'84, and '85 to 2003. 1046 MR. SHEPHERD: You have different numbers for each? 1047 MR. FOGWILL: For the 20-year trend. 1048 MR. SHEPHERD: Okay. And what are the numbers you have? 1049 MR. FOGWILL: I have 4.4 for the '71 to '84, and 6.9 for the '85 to 2003. 1050 MR. SHEPHERD: And so that also increases your '71 to 2003 number. Is that also -- 1051 MR. FOGWILL: No, that was correct. 1052 MR. SHEPHERD: Are there any other corrections? 1053 MR. FOGWILL: The root mean squared error for the raw comparison, 20-year trend, 1971 to 1984, I have 215. 1054 MR. SHEPHERD: Okay. 1055 MR. FOGWILL: And the mean percent error, 20-year trend, raw comparison, 1985 to 2003, I have 1.6 -- minus 1.6. 1056 MR. SHEPHERD: Minus 1.6. 1057 Subject to those corrections, are those all the corrections, sir? 1058 MR. FOGWILL: No. 1059 MR. SHEPHERD: There's more. 1060 MR. FOGWILL: If you look under comparison excluding the extremes, 20-year trend proposed, all the results are different than the ones that you've submitted to us. 1061 MR. SHEPHERD: All right. Can you just run down them, then, and we'll take them down, starting from the top. 1062 MR. FOGWILL: Under "average heating degree day variation from actual, '71 to '84, it's 101. 1063 MR. SHEPHERD: 101. 1064 MR. FOGWILL: 101. '85 to '03 is 58. 1065 MR. SHEPHERD: Plus or minus? 1066 MR. FOGWILL: Plus. '71 to '03 is 79. For the mean absolute percent error, or MAPE, '71 to '84, it's 3.9. And these are all related to the 20-year trend. 1067 MR. SHEPHERD: Yes. 1068 MR. FOGWILL: '85 to '03, it's 5.2. '71 to '03 is 4.6. For the root mean squared error, '71 to '84 is 184. '85 to '03, it's 228. And '71 to '03 is 208. 1069 For the mean percent error, '71 to '84 is minus 2.5. '85 to '03 is minus 1.8. And '71 to '03 is minus 2.1. 1070 MR. SHEPHERD: Minus 2.1? 1071 MR. FOGWILL: Yes. 1072 MR. SHEPHERD: Thank you. Subject to those corrections, do you accept that these numbers are correct calculations of the data set that you provided in J.26-35? 1073 MR. FOGWILL: Yes. 1074 MR. SHEPHERD: Thank you. So as you've already proven so conclusively, I'm not a statistician, so in looking at a comparison of accuracy, I start with the sort of common sense, which I guess is which is -- which method is closer to projecting heating degree days each year. And correct me if I'm wrong, but it's true, isn't it, that from 1971 to '84, they were virtually identical, very close, the two methods. 1075 MR. FOGWILL: I'm sorry, you're looking at? 1076 MR. SHEPHERD: Heating degree day variation from actual. 1077 MR. FOGWILL: Well, one is minus 120 and the other one is 127. 1078 MR. SHEPHERD: So, in fact, the 30-year average underforecast heating degree days in that period, whereas the 20-year trend overforecast; right? 1079 MR. FOGWILL: Yes. 1080 MR. SHEPHERD: But wasn't your concern with 30-year average is that it would consistently overforecast heating degree days? 1081 MR. FOGWILL: Don't forget we're looking at a different period of time. You've got the '71 to '84 here, and we've done the analysis going forward. 1082 MR. SHEPHERD: Okay. But it's correct, isn't it, that one of your big concerns was that 30-year average overforecasts heating degree days on a regular basis. 1083 MR. FOGWILL: Yeah, and it does right now. 1084 MR. SHEPHERD: But in fact from '71 to '84, it was 20-year trend that was underforecasting and 30-year trend was overforecasting; right? 1085 MR. FOGWILL: Yes, correct. 1086 MR. SHEPHERD: And in '85 to 2003 in the period you looked at, both methods overforecast, but 30-year average overforecast a lot more; correct? 1087 MR. FOGWILL: Sorry, you said '85 to 2003 -- 1088 MR. SHEPHERD: Yes, the period that you used. You used -- 1089 MR. FOGWILL: The heating degree day variation from actual for the 30-year average was 264, and for the 20-year trend, it's 46. 1090 MR. SHEPHERD: So they both overforecast, but 30-year average overforecast much more. 1091 MR. FOGWILL: Yes. 1092 MR. SHEPHERD: Okay. So then if you just look at the variation from actual, it looks to me, and tell me if this is correct, like on accuracy, the 20-year trend is a bit better, and certainly in the last period it's a lot better; but on symmetry, in fact, 30-year average is better because, in at least one period, it was underforecasting while the other was overforecasting; correct? 1093 MR. FOGWILL: You're comparing the symmetry of two blocks of time; is that the understanding of your question? 1094 MR. SHEPHERD: It's a common sense comparison. We're going to get more technical comparisons in a second. If you look at those two periods, the period you looked at and the one before, the full set of data that you had available to you, you see that the first half, they're on either side of the -- of actual, and in the second half, they're both on the same side. Correct? 1095 MR. FOGWILL: Yes. Yes. 1096 MR. SHEPHERD: And the average difference in forecasting accuracy in terms of heating degree days over that entire 33-year period is only 33 degrees; right? 1097 MR. FOGWILL: The '71 to -- 1098 MR. SHEPHERD: The whole 33 -- 1099 MR. FOGWILL: Yes. Yes. I think it's important to note, though, that in terms of accuracy, the more precise way to measure accuracy is by using the mean absolute percent error and the root mean squared error tests. 1100 MR. SHEPHERD: We're going to come to those. 1101 MR. FOGWILL: Okay. 1102 MR. SHEPHERD: Forgive me if I try the common sense approach first, and then we'll go to the mathematical approach. 1103 So that 33 degree days, that's -- that 33 degree days, average heating degree day variation each year, is about eight-tenths of 1 percent or so of your heating degree days in a typical year; isn't that right? 1104 MR. FOGWILL: We're talking about ... 1105 MR. SHEPHERD: The accuracy differential between the two. 1106 MR. FOGWILL: The accuracy differential between the two is -- you really have to look at the statistical test in terms of defining accuracy, and that's why we got -- 1107 MR. SHEPHERD: Sorry, you don't think that this Board should be concerned with how close these methods are to getting the right number in heating degree days? 1108 MR. FOGWILL: Well, but if you look at the root mean squared error test, that is what will give you a relative assessment of the accuracy of the two methods. And it shows that the results from that test -- so that the 20-year trend is a more accurate method. And that gives you -- 1109 MR. SHEPHERD: We'll come to that in a minute, okay? 1110 MR. FOGWILL: Okay. I'm just having a difficult time saying that, you know, trying to identify it in terms of "common sense." Looking at the average heating degree day variance from actual sort of ignores the fact that there's some rigors assigned to these statistical tests that have really been established to evaluate the accuracy of the different forecasting methods. I'm trying to substitute that. 1111 MR. SHEPHERD: We're going to come to the question of which tests are most accurate and why. But I guess I'd like to follow up on your assertion that the actual variation -- the variation from actual in heating degree days isn't a relevant test. Why do you think it's not a relevant test? 1112 MR. FOGWILL: Well, because I think it's more relevant to look at the assessment using the mean absolute percent error and the root mean squared error. 1113 MR. SHEPHERD: So you're not saying there's anything wrong with the heating degree day variation, you're saying the other methods are better. 1114 MR. FOGWILL: I haven't thought about using that average heating degree day variation to assign a level of accuracy to the test. So I couldn't say if it's appropriate or -- at this point, I don't think it is appropriate. I'd rather use the other tests because those are the ones that we've reviewed all the different methods with, as opposed to just these two. 1115 MR. SHEPHERD: Okay. So let's, then, go -- by the way, just before I leave this, one of the things that happens in a data set is that extreme numbers within the data set can skew the whole data; right? Can skew your averages, your trend, et cetera; isn't that right? 1116 MR. FOGWILL: Not if they're valid data points, then they're representing what's actually happening and the circumstances of the -- of whatever you're evaluating, even if they are extreme. 1117 MR. SHEPHERD: I'm not saying they aren't valid, I'm saying they have a significant impact on your result; right? 1118 MR. FOGWILL: Yes. 1119 MR. SHEPHERD: So if one year one of the methods is out 900 heating degree days, let's say, which I think one of them was at one point, that's going to have an impact on 33 years of data of 27 heating degree days error rate; right? Just divide 900 by 33. 1120 MR. FOGWILL: Yes. 1121 MR. SHEPHERD: So being out a lot once is going to have a more significant impact than being out a little bit a lot of times, generally speaking; isn't that right? 1122 MR. FOGWILL: I actually don't know if that's the case. I'd have to think about that some more, because it depends on the magnitude of all those other ones. 1123 MR. PENNY: I think we need clarification of what you mean by "out" in this context, because as I understood Mr. Fogwill's evidence, what we're talking about is actual weather on actual days or heating degree days, not the product of the methods. 1124 MR. SHEPHERD: Well, Mr. Chairman, Mr. Penny has not understood the question. I think Mr. Fogwill did. We are talking, in fact, about the two methods and how they predict actual weather. So it would probably be useful, unless Mr. Penny has a substantive comment relating to the question I did ask, if I could continue asking Mr. Fogwill the questions. 1125 MR. SOMMERVILLE: If you could repeat the question, Mr. Shepherd. 1126 MR. SHEPHERD: I'm trying to remember it. 1127 Mr. Fogwill, if you have a very extreme error from the one methodology to actual, compared to actual, that will have a bigger impact on the averages or the trend, generally speaking, than a large number of small errors; isn't that true? I guess it depends on how big the small errors are; right? 1128 MR. FOGWILL: That's right. 1129 MR. SHEPHERD: It's just arithmetic, isn't it? 1130 MR. FOGWILL: Yes. 1131 MR. SHEPHERD: All right. 1132 Let me ask you this: If the existing method was bang on to actuals 32 years out of 33 -- let's say it was; we should be so lucky, right -- but one year it's out 900 heating degree days, it's just way off, and the new method is sometimes over, sometimes short, always 60, 70, 80, heating degree days every year, but on average it's only out 10 heating degree days on average every year, which is the better method for you to use? 1133 MR. FOGWILL: Well, keeping in mind that one of the important objectives that we were looking at was symmetry, if one method is more symmetrical and close to the set of data, then we would probably go with that one. 1134 MR. SHEPHERD: So in that example you would still prefer the new method? 1135 MR. FOGWILL: It would actually depend on how they performed relative to the data set itself. So if they were symmetrical, then you'd look at the more symmetrical method and then look -- and accuracy would be another element that you'd have to consider. 1136 MR. SHEPHERD: Let's go to mean absolute percent error and root mean squared error. 1137 First, can you describe the difference between the two methods? 1138 MR. FOGWILL: Well, I guess the major difference is the mean absolute percent error is used to evaluate data sets that have different units, and we had to do that because heating degree days are a unit that we use and for most of the methods it's used that way, but we also evaluated one of the methods which was the average of minimum and maximum temperatures and it's not effectively the same unit, so we had to use some sort of percentage accuracy measurement. 1139 MR. SHEPHERD: So that's the main difference between the two tests? 1140 MR. FOGWILL: Yes. 1141 MR. SHEPHERD: I thought the main difference was that root mean squared error gave more weight to the extreme numbers and less weight to the closer variations; whereas, mean absolute percent error is arithmetic; isn't that correct? 1142 MR. FOGWILL: The mean absolute percent error -- sorry, the root mean squared error allows you to look at large errors in a data set, so you're correct there. The other point I was trying to raise was that the mean absolute percent error is needed in order to evaluate different types of data. 1143 MR. SHEPHERD: I understand that. I'm thinking more about the math here than about the functional usage. 1144 MR. FOGWILL: Okay. 1145 MR. SHEPHERD: So from a mathematical point of view, if you have a data set that has some small variations and some extreme variations, the root mean squared error will tend to give more weight to those extremes and the mean absolute percent average will give less weight to those extremes? 1146 MR. FOGWILL: Yes. 1147 MR. SHEPHERD: It will still give more weight than the smaller amounts, but it's arithmetic rather than geometric? 1148 MR. FOGWILL: Yes. 1149 MR. SHEPHERD: So in terms of the mean absolute percent error, your evidence is that the 20-year trend performs slightly better, 6.9 percent to 7.7 percent, relative to the 30-year average; correct? 1150 MR. FOGWILL: Yes. 1151 MR. SHEPHERD: But in fact, if you take a longer data set, you find that in the '70s and early '80s, 30-year average performed better; isn't that correct? 1152 MR. FOGWILL: That's true, but we're not in the '70s and early '80s. We're in 2003, looking at 2004, so we really do have to look at the more recent history. That's why we've focused on the last 18 to 20 years. 1153 MR. SHEPHERD: Over the entire 33 years of available data, the two tests are about a dead heat; correct? 1154 MR. FOGWILL: No, the 20-year trend is still the better method. 1155 MR. SHEPHERD: By 0.2 percent. 1156 MR. FOGWILL: That's not the only element that's involved in it. The mean percent error, the number is much better there in terms of symmetry. In terms of the root mean squared error, it wins as well. I mean the 30-year average, every time that I've had to look at reviewing the 30-year average, either on a five-year basis, a 10-year basis, the 18 years that we've got in the current evidence, the 17 years that we had in the previous version of the evidence, the 31 years here, the 30-year average loses to the 20-year trend. The 20-year trend either comes in first or second, however you cut and slice those numbers. 1157 MR. SHEPHERD: That's what we're trying to do is go through each of the ways you cut and slice it. If you'll bear with me, it takes me a little bit longer to get through it perhaps than you. On the mean absolute percent error, the difference is not a significant difference. 0.2 percent is not a significant difference, is it? 1158 MR. FOGWILL: I would consider it significant. 1159 MR. SHEPHERD: How many heating degree days is that? 1160 MR. FOGWILL: 0.2 percent on -- 1161 MR. SHEPHERD: It's about eight, isn't it? 1162 MR. FOGWILL: I'd have to check that one. Give me a minute. About eight. 1163 MR. SHEPHERD: Now, let's go to root mean squared error. In that case you have a test where the extreme errors have more weight, and you see that the 30-year average fairs more poorly; correct? 1164 MR. FOGWILL: That's correct. 1165 MR. SHEPHERD: In fact, if you go back to page 4 of the cross-examination materials, you can see why that is, right? You have some years like 1998 and 2001 and 1990 that have very big influence on the overall numbers; isn't that right? They're going to have much more influence on the 30-year average method than they are on the trend method. 1166 MR. FOGWILL: They will have more influence on the 20-year trend method than they will on the 30-year average. 1167 MR. SHEPHERD: They will have, sorry? 1168 MR. FOGWILL: More influence on the 20-year trend method than on the 30-year average. 1169 MR. SHEPHERD: Why is that? 1170 MR. FOGWILL: You've only got 20 years of observation as opposed to 30. 1171 MR. SHEPHERD: But the variation from test to actual -- for example, take 1998. The difference between what average projects and what the actual is is much greater than what the trend projects and what the actual is; right? 1172 MR. FOGWILL: Yes. 1173 MR. SHEPHERD: An RMSE test exacerbates that difference, gives more weight to that difference than to a difference like in 1989, where both of them are relatively small difference; correct? 1174 MR. FOGWILL: Yeah. 1175 MR. SHEPHERD: Now, it's correct, isn't it, that mean absolute percent error and root mean squared error are both tests that don't care what the direction of the error is? They don't consider symmetry, they only consider raw accuracy. 1176 MR. FOGWILL: Right. That's correct, yeah. 1177 MR. SHEPHERD: All right. Whereas mean percent error also considers symmetry; correct? 1178 MR. FOGWILL: That's correct. 1179 MR. SHEPHERD: In fact, in your analysis, your ranking, you didn't include mean percent error as an accuracy test, you include it as a symmetry test; correct? 1180 MR. FOGWILL: That's correct. 1181 MR. SHEPHERD: All right. And the results, it appears, is that for the period you studied, there's a substantial difference. Both underforecast but the 20-year trend underforecast by a much lower percent than 30-year average did. And we saw that up in the heating degree day comparison. It's the same comparison; right? 1182 MR. FOGWILL: Yeah. And I think it's important just to go back to your graph that you've reproduced on page number 4, and just visually you can see how well the 20-year trend does in terms of the symmetry around the data points as compared to the 30-year average. 1183 MR. SHEPHERD: Well, that's bound to be the case, though. Mathematically, a trend line using regression always goes down the middle of the data; correct? 1184 MR. FOGWILL: Correct. 1185 MR. SHEPHERD: So we don't need to discuss which is the better method if symmetry is the only question because a regression trend line is going to be the better method. It has to be; right? 1186 MR. FOGWILL: Well thank you. Yes. 1187 MR. SHEPHERD: Okay. It's true that over the entire 33 years, the difference between the 30-year average and the 20-year trend in terms of mean percent error is not a substantial one, is it? You focused on 1985 to 2003, but 1971 to 2003 is not in the same order of magnitude, is it? 1188 MR. FOGWILL: That's correct. But again I'd have to go back and repeat that we're not looking at what happened in the '70s, we're looking at what's happening in the next five to seven years, and the 30-year average is just not the method to consider. 1189 MR. SHEPHERD: And then this is because you think that global warming started in 1985? 1190 MR. FOGWILL: No, it's looking at the fact that from the information we've got from Dr. Weaver, and also the information we've got from Mr. Root, that a more reasonable approach to take, if there's changing conditions, is to look at a shorter term period and trend it. So that's -- and if we're looking at a shorter term period, we have to evaluate over that shorter term period. 1191 MR. SHEPHERD: Dr. Weaver, is it your evidence that using a shorter data set is a more reliable than a longer data set? 1192 DR. WEAVER: I didn't say that. 1193 MR. SHEPHERD: I'm sorry, that's what Mr. Fogwill said you told him. 1194 DR. WEAVER: Sorry -- 1195 MR. PENNY: I think what Mr. Fogwill was talking about was Mr. Root's recommendation, not Dr. Weaver's. 1196 MR. SHEPHERD: Oh, fine. Mr. Root, are you telling us that a shorter term data set is better to calculate a trend than a longer term data set? 1197 MR. ROOT: My studies have shown that, yes. 1198 MR. SHEPHERD: Now, finally, there's three tests that you didn't do that -- two tests, I guess, one of which has two variations, that -- again, going back to common sense, I would have thought were logical ones. One is which method is closest to actual, which method is closest to being right each year. And it's true that if you count them up, the 30-year average is more likely to be right each year than the 20-year trend; isn't that right? 1199 MR. FOGWILL: Can you refer to something in particular on that sheet? 1200 MR. SHEPHERD: Number of times closest to actual, 1971 to 2003, 18 to 15. 30-year average wins. Did you look at that? 1201 MR. FOGWILL: Yes. And what I found, or what I can tell you about this is that it's another form of accuracy measure, and it ignores how far away from the observed data, the estimate is. So that variation is not considered in terms of this just listing how close they are, one to the other. 1202 MR. SHEPHERD: The other test that I guess intuitively I would have thought you looked at is a sort of -- I would call it the more-likely-to-be-right test, and you define right as being within a particular level of accuracy. And so if you test that saying right is within 1 percent, 30-year average still wins, doesn't it? It's more likely to be within 1 percent than the 20-year trend is. 1203 MR. FOGWILL: And you're referring to? 1204 MR. SHEPHERD: Number of times within 1 percent, five to three. 1205 MR. FOGWILL: For the line 1971 to 2003? 1206 MR. SHEPHERD: Okay. 1207 MR. FOGWILL: Well, that is -- again, I come back to the point that these -- this listing of how close they are within 1 percent, within 5 percent, ignores the fact of how much error there is in all the other data points, or even in those data points themselves. So from an accuracy point of view, I don't think it's an appropriate way of determining how accurate the method is. 1208 MR. SHEPHERD: If the 30-year average was within 1 percent 32 years out of 33, and the 20-year trend three years out of 33, would you still say, Well, you shouldn't consider that fact? 1209 MR. FOGWILL: Well, it's not there. And if it was there, then it would show up in the root mean squared error or the mean absolute percent error test. 1210 MR. SHEPHERD: But we've seen that the extremes push those tests off and, in fact, root mean squared error pushes it off quite a lot because it's geometric; right? 1211 MR. FOGWILL: Right. 1212 MR. SHEPHERD: And in fact, isn't there some value to you in knowing that your method is more likely to be within a range, a planning range? For example, if you took, let's say, 5 percent, 5 percent is, what, 200 heating degree days. So beyond 200 heating degree days, you've got problems anyway, right, in terms of planning. But if you're within 200, you're probably planning pretty accurately, aren't you? 1213 MR. FOGWILL: I'd say if you're within -- I really couldn't say if 200 is the magic number in terms of where we'd have some concern about the actual activities that we're involved in. But I guess I still come back to the point about the root mean squared error test does take that into account. I mean, if there is a number of observations that are significantly away from -- sorry, if there are a number of estimates that are significantly away from the observation or what actually happens, then that's going to be compounded and you want to stay away from methods that do that. 1214 MR. SHEPHERD: The root mean error squared test says that you have to move to a 20-year trend, but if you do that, then you're less likely to be within 5 percent; is that correct? 1215 MR. FOGWILL: Well, in terms of the 1971 to 2003 data set -- 1216 MR. SHEPHERD: It doesn't matter which one you use, does it? 1217 MR. FOGWILL: Right. 1218 MR. SHEPHERD: Pretty consistent. What we're trying to do, as I understand what we're trying to do here, whether successfully or not is not for me to say, is to help the Board understand how they make this assessment. 1219 MR. FOGWILL: Right. 1220 MR. SHEPHERD: Do you believe that the Board should take into account that fact, that one test is more likely to be within 5 percent or 1 percent than the other test? Do you think that that's something the Board should take into account; yes or no? 1221 MR. FOGWILL: No. When we put this analysis together, we looked at how we would do the analysis for the symmetry, the accuracy, and the stability, and we came up with these very straightforward statistical tests that are in a lot of manuals related to operational management. So we just chose those and it never shows up in terms of the number of times it's within 5 percent of the observed data. That is a means of evaluating whether one method is better than the another. So I think the method we've chosen is reasonable and gives an accurate representation of the process we went through to decide on the 20-year trend. And it just seemed very straightforward to me that the 20-year trend seems to win out no matter how you slice it. 1222 MR. ROOT: If I might add some discussion to that. In all of the forecasting that we do for our energy clients, in every case, in every discussion on forecast accuracy, our clients have mentioned to us, not only the energy clients but also the ISOs, that they would rather have us employ techniques that present the lowest root mean squared error because they would rather live with a variability that's small versus the likelihood of having a number of days more accurate but have a huge error or a number of huge errors, because it's those huge extremes in forecast accuracy that they cannot deal with. And in their mind, they would rather have a lower oscillation around the actual. And in our forecasting techniques, we've had to force our team members to go away from the likelihood and the desire to hit that one out of ten forecasts right on in favour of hitting all ten within a short distance. And that's what we've found. 1223 MR. SHEPHERD: Mr. Chairman, until that last answer, I was 33 minutes, but I'll still claim victory on my timing. Those are all my questions. 1224 MR. SOMMERVILLE: Mr. Aiken, you're going to be 30 minutes or so? 1225 MR. AIKEN: Yes, 30 to 45 minutes, probably. 1226 MR. SOMMERVILLE: I'm wondering, it's very hot in this room right now, whether the balance of convenience isn't to try to do this tomorrow morning rather than carry on tonight. 1227 MR. PENNY: We're, of course, in your hands, Mr. Chairman. Our preference would be to forge ahead, because there's still -- this is only panel number 2 and we've got a long way to go. 1228 MR. THOMPSON: We've got three to go. 1229 MR. SOMMERVILLE: Mr. Thompson is tomorrow, Mr. Ryder is tomorrow and Mr. Moran is tomorrow. 1230 MR. AIKEN: I can advise that if I were to go tomorrow morning, I might be able to cut down on the amount of time. 1231 MR. SOMMERVILLE: Let me indicate too that I think the understanding is that we'll start at 10:00 tomorrow and, in fact, we can start at 9:30 again. We've adjusted the meeting that was intruding on that. We can start tomorrow morning at 9:30, and I think that's what we'll do. I thank the panel for your efforts today and ask you to come back tomorrow. 1232 We'll adjourn until tomorrow morning at 9:30. 1233 --- Whereupon the hearing adjourned at 4:25 p.m.