The MacNeil/Lehrer Report; 7202; Weather Forecasting

[Tease]

ROBERT MacNEIL [voice-over]: With the nation reeling from a week of unexpectedly severe weather, has all the new technology made weather forecasting any better?

[Titles]

MacNEIL: Good evening. Here in New York and across the Northeast today, the first daffodils and the tender buds of spring got buried under one of the heaviest April snowstorms on record. Winds of up to 60 miles an hour piled up huge drifts, crippling traffic and other services.The National Weather Service station in New York called it a "life-threatening blizzard." On top of everything else, the snowstorm wiped out the first day of the major league baseball season in many cities. But today's storm was just the tail end of a long assault recently of freakish weather -- killing floods, avalanches and tornadoes to end a winter that was longer and more severe than most meteorologists predicted. As every television viewer knows, the weathermen have a lot of new scientific backup these days from satellite pictures to radar. Tonight, has all the new technology made weather forecasting any more accurate? Jim?

JIM LEHRER: Robin, there's a tendency to think of weather forecasting as fun and games, probably because we receive most forecasts from people on television who wear loud ties and blazers, draw cute cartoons on blackboards, or simply giggle and joke around a lot. But there are some very serious sides to forecasting -- the death, injury and destruction that can come from a faulty forecast being the most obvious and deadly. But in the last 10 years another no-nonsense side has emerged -- the business side, the business of forecasting the weather to make money, both for the forecasters and their customers. There are now hundreds of private commercial weather forecasting firms around the country; most are run by local TV weather personalities or other meteorologists who have opted to do their predicting in the commercial world rather than for the government or at university research institutions. The private firms' clients range from the largest corporations to farmers seeking guidance on when to plant. Most take basic information supplied by the National Weather Service and then refine it to meet the specific needs of a private client -- say a major trade association wanting to know a year ago if it would be wise to schedule a convention in New York City today, April 6th. One of the newest wrinkles of forecasting in the business world has been the hiring of staff meteorologists by New York brokerage houses. The first to do so was E. F. Hutton, and the one they hired was Gail Martell. She now heads a staff of three Hutton meteorologists based in Milwaukee where she is with us tonight from public station WMVS. Ms. Martell, why does a brokerage house need a staff meteorologist, much less three of them?

GAIL MARTELL: Well, maybe first I should clarify that I work for E. F. Hutton's commodities department, so I have nothing to do with stocks, bonds and other types of investments. I believe that my being hired was prompted by a very significant weather event back in 1972. The Soviet Union had a very severe drought in that year. Subsequently, through a series of secretive purchases the Soviets bought up a massive quantity of United States grain. And some people called this "the great grain robbery." Subsequently, prices skyrocketed, and I think E. F. Hutton decided they couldn't afford to be surprised by such an event.

LEHRER: Well, what kind of information do you supply?

Ms. MARTELL: We supply a daily forecast for the United States and for important foreign growing areas, and along with it an interpretation of how we think the crops are progressing, what we think the prospects are.

LEHRER: Do you take -- what do you base that on -- basic National Weather Service information, then recycle it, or what?

Ms. MARTELL: For the United States, yes, it's the same basic tools that your TV forecasters would have. We also get some information from a private consulting firm on all of the foreign weather.

LEHRER: Well, to borrow a phrase, what is special about your forecast? In other words, why should somebody listen to E. F. Hutton when when you speak on weather that they couldn't pick up somewhere else?

Ms. MARTELL: Well, I think there are a couple of reasons that I could answer. One is the interpretation aspect. It's in other words notonly enough to know that it's raining two to three inches in Iowa in the middle of July; you also have to know what stage of development the corn or the soybean crop is in. Is that rainfall in July beneficial or is it detrimental, and how does it look for crop production for that particular growing season? The other aspect I think is timeliness. If you're a commodities broker and market prices are fluctuating up and down during the day, you don't want to have to be bothered to call Des Moines or to call some other area in the Midwest; you want a forecast and you want an updated forecast now. And we are able to provide that kind of service.

LEHRER: You are probably not the best one to ask this, but I'm going to ask it anyhow. Has the decision by E. F. Hutton to hire you and your two colleagues proved to be a good one? I mean, can you give me some instances where you were able to make either E. F. Hutton some money or some of your clients some money as a result of your weather forecasting?

Ms. MARTELL: Well, I think the fact that there are three of us -- and I started it out five years ago -- signifies that they do think it's worthwhile. As far as exact dollar amounts, that would be pretty difficult for me to say. You see, we give the information out; we're not sure how the brokers are reacting to it. Sometimes they may believe us and act on the forecast; sometimes not. But to get a little bit more specific, I think that if a very dramatic weather event takes place -- say, a freeze or flooding or drought -- and we accurately forecast it, I think it could mean perhaps 1,500 to 2,000 dollars on one contract of, say, corn. And a lot of these brokers buy more than one contract.

LEHRER: So you think you've made some folks some money.

Ms. MARTELL: I believe so. More often than not. I think our track record is pretty good.

LEHRER: How was your track record on the crazy weather of the last few days? Did you predict this?

Ms. MARTELL: This was not one of my banner forecasts. I think that I was a little taken aback by the very severe storm that we had Saturday in the Midwest. We had hurricane force winds; we had record low pressure, blinding snow, and I think that I just wasn't paying real close attention to this current storm system, thinking that there was no way it could intensify, that we could have two very severe storms back to back. So I wasn't real sharp on this one, no. After it produced seven to 10 inches in the Midwest, it was pretty easy to make the forecast for a foot of snow in New York, but I have to admit I was a little -- not real sharp on it.

LEHRER: I see. Thank you. Robin?

MacNEIL: Now we hear from one of the many private forecasting companies, Fleet Weather, Incorporated, of New York. Its co-founder is James Witt, and he differs from most others in claiming that he's able to make very long-range forecasts. His clients include IBM, radio and television stations, shipping companies and highway departments. First of all, did you predict this storm?

JAMES WITT: Well, we predicted it short range, but not on long range.

MacNEIL: What does that mean?

Mr. WITT: For example, two or three days ago we saw the storm in its developmental stages and were able to say this was going to be a severe storm for the New York area for this particular day. But if last year you had asked me, "Is there going to be a major snowstorm on April 6th?" -- I would have said no.

MacNEIL: Why would you have said no?

Mr. WITT: Well, first of all, climatology, which say the chances of getting a major snowstorm are almost nonexistent -- I think the last time we had a very major snowstorm was back in 1915 in April, and the time before that was 1875, so you figure two times in a hundred and some-odd years? I don't think I would go and call for a major snowstorm.

MacNEIL: So you would have advised somebody to have a convention in New York today?

Mr. WITT: Yeah, sure, have one.

MacNEIL: I see. Did you predict how severe this winter just over, or nearly over, was going to be?

Mr. WITT: Yes. I thought we did a very fine job on this particular winter. We thought that it would be characterized by numerous storm systems along the East Coast, especially the northeastern portion of the coast; and we also said that the drought situation that the Northeast has been experiencing would be just a memory by the end of this particular winter, and that it would be below-normal temperatures, as cold air would rush behind --

MacNEIL: The drought going would mean a lot of snow would fall, and --

Mr. WITT: Not necessarily. We did not say it would be a lot of snow; we said precipitation would be above normal, but that much of it would be in the form of rain, because temperatures would be too warm.

MacNEIL: I see.

Mr. WITT: Even though -- and I have to clarify -- even though a temperature is too warm for snow, behind each storm northwest winds would pull very cold air into the region and put temperatures for the entire winter below normal -- which was the case.

MacNEIL: When did you say and what did you say? When exactly did you say it was going to be a bad winter?

Mr. WITT: Well, actually, the time frame makes very little difference. For example, if I make a long-range forecast, it can be made three years in advance, four years, five years -- that really has no bearing on it. It is just as easy to make a forecast four years from now as it is next month using the particular method that I use.

MacNEIL: And what is the particular method that you use?

Mr. WITT: Well, what I do is I go back through history, about 80 to 100 years of weather maps, and select maps that I will use for a future forecast. Give you an example: next December 26th -- that would be 1982 --

MacNEIL: Day after Christmas.

Mr. WITT: Day after Christmas. In selecting maps for forecasting that particular day and putting them in a folder, as I review them I find out that many show storminess over the northeastern portion of the United States, especially in New England. In making the forecast, if someone was to ask me, "What do you think weather conditions will be like in New England next December 26th?" -- I would say I would think there could be a snowstorm right after Christmas.

MacNEIL: Are you just operating on a hunch, or are you operating on the probability of the number of December 26ths on which there have been snowstorms?

Mr. WITT: Oh, no, it has nothing to do with the number of 26ths that there have been snowstorms; it has nothing to do with that whatsoever. It is a matter of selecting past maps to make future projections. And the method that we use to select the past maps has not been published and I am not ready to publish that as of yet.

MacNEIL: I see. But is it just a hunch?

Mr. WITT: No. No, it's a lot of research.

MacNEIL: I see. How much has modern technology -- the use of satellites, the use of radar, the use of computers to calculate all this -- really improved weather forecasting?

Mr. WITT: I think that it has done wonders for the meteorological profession. For example, take today's snowstorm. I think most meteorologists in New York City were predicting eight to 12 inches of accumulation before it all ended late in the afternoon or evening. That was a perfect forecast. I think years ago, 10 years ago, it would not have been possible. We were receiving satellite photos today every 15 minutes, and you could see exactly where the center of the storm was located.

MacNEIL: Well, how did that help? How actually would it help somebody living in New York that this new technology enabled them to do that?

Mr. WITT: Well, I think people were very prepared this morning; when they woke up they heard there would be eight to 12 inches of snow. Schools were closed even though snow was just beginning early this morning. A lot of businesses closed early, saying "Why open? It's going to be a really bad day." That helps people.

MacNEIL: Could you make larger claims than that? Could you say it saved lives, or saved money, or saved property?

Mr. WITT: Yes, I think I could say that.

MacNEIL: Well, thank you. Jim?

LEHRER: You can't help but wonder what the professionals at the National Weather Service think of Mr. Witt's methods and private forecasting generally. One of the service's top professionals is Donald Gilman, chief of its long-range forecasting department, who has been a meteorologist with the Weather Service for 23 years. Mr. Gilman, what do you think of Mr. Witt's long-range forecasting method?

DONALD GILMAN: Well, we just heard him say that he hasn't published it, and therefore it's really very difficult to make a judgment. It sounds as if it may be a method based on what we call analogues, and uses them in an attempt to make daily forecasts a long ways ahead. Now, analogues are an old idea in meteorology, and in some ways they could be helpful.

LEHRER: Analogues meaning looking at past -- using maps, as he says, past weather maps of certain days and trying to --

Mr. GILMAN: Yes, ordinarily you try to pick past situations that in some important ways have a very strong resemblance to the recent and current weather patterns. The trick partly is finding one like that.

LEHRER: Is that a hard trick?

Mr. GILMAN: I think it is. I think that the evidence that we see in examining and comparing past cases with the present and with each other is that the weather does not repeat itself, and --

LEHRER: Never?

Mr. WITT: Never. At least, never within the records.

LEHRER: So you could spend hours and days looking at Mr. Witt's maps and you're not going to find anything, right?

Mr. GILMAN: Well, it depends on your -- what you might call the tolerance limits. If you widen the limits, then you could find the resemblances -- there certainly are family resemblances. Now, it's not clear to me, however, that that's really the method we're talking about.

LEHRER: We'll give him a chance to confirm or deny that in a moment. But what method do you use? You are the acknowledged expert on long-range forecasting in this country; what method do you use?

Mr. GILMAN: Well, our methods are really very simple, because it's not possible to make scientifically sound complicated ones, as far as we know. They are fundamentally statistical in the character, and they also go to past data; they attempt to find some relationship between fall and winter, upper-level circulation patterns in summer and winter, of spring and winter; some dependencies from one season to the next, or as much as one year to the next. What you find this way is always weak clues scattered around the Northern Hemisphere, and we sort of piece them together and make a consistent picture out of them, and then interpret the weather at the ground from that picture.

LEHRER: How has the so-called new technology improved your ability to long-range forecast?

Mr. GILMAN: I think as far as long-range forecasting is concerned it hasn't had very much impact. It helps, of course, in doing data processing and preparing maps and things like that. The long-range forecast problem is still recognized as largely unsolved and very recalcitrant, a very obdurate sort of problem; and the modeling techniques, the really strong computer techniques that have now become the backbone of short-range forecasting will not reach into that time range. They run out of accuracy pretty much around five, six or seven days in the future, and beyond that you have to use much simpler, more empirical and more, sort of cobbler's methods, if you will.

LEHRER: So if anybody, whether it's a private forecaster or anybody else, gives you a forecast that is definite after five or six days, you would say, "This guy's going with his gut" or "This guy's going with some guesswork"?

Mr. GILMAN: One would have to say, what evidence is there for it? What track record, public track record, stands behind something which does not seem to be justified by what theory we have for weather variations and what experience we have.

LEHRER: What was your service's record on predicting this crazy winter we have just gone through and in fact are still going through, at least in some parts of the country?

Mr. GILMAN: Well, in the terms in which we make the forecast, which are quite simple terms, it was a fairly successful one. We were, I think, a little better than three-quarters right across the country in calling it as above or below normal in various areas. So we feel that we got a good general idea of the winter, and that's really as far as we think the state of the art goes.

LEHRER: Did your long-range forecast in any way show any signs that it was going to go this late in this part of the country, that we were going to have severe weather in the first part of April?

Mr. GILMAN: I think that the update on the forecast that we made later than the first of December held the pattern of cold in the East and mild in the West -- very similar to the winter forecast. But I haven't looked at the one that followed that, which is really the one that reaches into April. And in any case, an individual storm in any individual month can have nothing whatever to do -- or may -- have nothing whatever to do with the character of the month or the season as a whole. So whether what we're having right now in the Northeast tells us very much, even about the rest of April, is still open to question.It could be a [unintelligible] -- just one of those terrible things that comes along.

LEHRER: Mr. Gilman, finally, let me ask you this. Does it bother you as a professional, the National Weather Service, when private forecasters take your information, recycle it or refine it and then sell it?

Mr. GILMAN: Well, what we forecast goes right into the public domain, it's available to anybody; and as private consultants, people can often tailor it to the needs of their clients in a way that we can't possibly do because we have to deal with everybody simultaneously.

LEHRER: So you don't feel like you're in competition with these folks?

Mr. GILMAN: No.

LEHRER: All right, thank you. Robin?

MacNEIL: Mr. Witt, Mr. Gilman sounds a bit skeptical about the way you do things, says specific forecasts beyond about a week or so don't seem to be justified by the theory; that the long-range problem is largely unsolved from his point of view.

Mr. WITT: Yes, I realize that is his feeling, and he did bring up the analogue method. And I fully agree with him that I do not think the analogue method works either -- I do not use an analogue method per se. I use specific maps for a specific day; I do not continue to carry this system on for several days.

MacNEIL: What percentage of success would you claim for your forecasts, some of your long-range forecasts?

Mr. WITT: Well, as I mentioned earlier, I have not really done a statistical study per se; although one year, we did make 20 forecasts for the upcoming winter -- specific days. At the end of the winter we summarized and found out how we did; we had hit 17 of the 20 -- statistically 85%, and statistically significant. We have also on occasion, on a major New York television station, predicted the first snowfall of the season. And it hit right to the day -- and this was done months and months in advance. In New England this year, we predicted the first major snowstorm right to the day.

MacNEIL: Ms. Martell in Milwaukee, how does this long-range forecasting business sound to you?

Ms. MARTELL: I have not seen much skill demonstrated in my line of business. People have brought me occasionally some long-range forecasts about what the weather is going to do in the coming summer, and maybe it's right, but then I follow that particular forecaster into the fall and into the winter, and he's wrong. And I've not seen it demonstrated that there is significant skill in the field of long-range forecasting.

MacNEIL: Mr. Gilman, from a scientific point of view, if Mr. Witt is able, through whatever method, to satisfy his clients with the accuracy of his forecasts, does that disturb you at all? Is there anything wrong with that?

Mr. GILMAN: I suppose that's really --

MacNEIL: Don't mean to put you on the spot, but I would --

Mr. GILMAN: Well, it sounds like it's really up to the clients, then, to decide. There's really very little way for me to comment without seeing the evidence in trying to judge it.

MacNEIL: I see. What is, in your view, why can't you, what is the limitation -- with all the equipment at one's command nowadays -- what is the limitation on being able to predict something with any accuracy beyond five to seven days? Why can't you do it, in your view?

Mr. GILMAN: I think it's because when you're attempting to make a daily forecast, one after another like that, that you're starting, with our forecasting system, with imperfect information. There are errors in there to begin with -- partly just because you can't sample the whole atmosphere everywhere simultaneously and not have any holes in the observations. Those errors, we find, grow and interact with each other as you calculate forward in rather small time steps with the big machines. There's a lot of physics in the models that are used in these machines -- not all of the atmosphere's physics, but a good deal of it -- and it's also known that this kind of equation has certain instabilities, so that if you make any kind of error, even errors of approximation in the calculations, let alone errors of the starting conditions, these too will grow. And if you're a little bit wrong about how the physics is put in, or you can't deal with very small-scale information in your calculations, all of those errors grow and they essentially infect the larger aspects of the forecast a little more each day, and the rate of infection spreads, and after, say, five, six, seven days -- depending on what you're trying to forecast and perhaps also on where you started from -- you'd have a forecast that perhaps still looks like a weather map, but it has no more resemblance to what's going to occur than it does to any other weather map at the same time of year.

MacNEIL: How do you answer that, Mr. Witt?

Mr. WITT: Well, I agree with that if you use computers, etcetera, to make the long-range forecast -- but that is not my method. In other words, I do not take one system and keep on projecting it into the future.

MacNEIL: In other words, you're not starting from living data right now -- temperatures sampled in various places, pressures, thickness of cloud, all that -- you're not -- highs and lows -- you're not taking that kind of stuff?

Mr. WITT: No, I am not.

MacNEIL: You're leaping way ahead.

Mr. WITT: I am doing that on short-range forecasts for my clients but not long range. It's a totally different method.

MacNEIL: I see. And does that -- if he is overleaping all that current information which may be in error the more you forecast through it, what does that make you think, Mr. Gilman?

Mr. GILMAN: I'm left in the dark, because I know of no way by which you could put your finger on past cases that would allow them to be used as a forecast.

MacNEIL: All right. Well, thank you. Jim?

LEHRER: Mr. Witt, I've got to ask you: why won't you tell us your method?

Mr. WITT: Well, right now I am part of a private organization, a private corporation. And the -- we are not ready at this time to release the information, partly because the time involved to publish and release would be one factor; the second is, economically it would not be wise for the corporation to let the "secret" out into the meteorological community.

LEHRER: Why? Why not? I mean, just commercially, somebody would steal your idea and steal your customers, is that it? I mean, it's a business decision, in other words, right?

Mr. WITT: Exactly, yes.

LEHRER: But someday Don Gilman is going to be told what your method is, right?

Mr. WITT: He certainly will.

LEHRER: Okay, well, we'll have you back then. In the meantime, let me ask each one of you, beginning with you, Ms. Martell. Is the crazy weather over now? Or are we about to have some dramatic occurrences in the spring and summer?

Ms. MARTELL: Well, since none of us believe in long-range forecasting, that's a tough one to answer. I will say that I believe that we're in a period of very erratic weather. In the '60s, for example, it appeared that we didn't get many weather extremes -- wet or dry, very hot or very cold -- whereas in the decade of the '70s we saw a lot more of that. And I believe we're still in that extremely variable pattern. So based on that I would say probably we will have some more extremes -- perhaps an early frost, perhaps extreme wetness, maybe drought this summer. But I can't say which way the extremes will be.

LEHRER: Can you, Mr. Gilman?

Mr. GILMAN: No. First of all, I --

LEHRER: Oh, come on, try.

Mr. GILMAN: I think even the question of whether we are now in an unusually extreme period is an open question. And Ms. Martell has I think forgotten a few things about the 1960s. There was a really spectacular winter in the winter of 1962-63. It was pretty bad in this country, it was terrible in Europe. And one can point to instances like that in almost every decade. Perhaps this decade does stand out a little, but it would be very dangerous, I think, to say, "Oh, this is the kind of decade we are in, and this is where we are going."

LEHRER: Mr. Witt, finally, how about you? What's the spring and summer going to be?

Mr. WITT: Well, I wasn't looking so much at the spring and summer as I was looking at the decade of the '80s, and would like to put my thoughts about the '80s. I see it in the northeastern and the eastern half of the United States as being a decade of cold temperatures and dry weather for the eastern half of the United States, and in the Rocky Mountains probably milder than normal for the winters.

LEHRER: We'll have you back in 1990 to see if you were right.

Mr. WITT: Fine.

LEHRER: Robin?

MacNEIL: Thank you. Ms. Martell in Milwaukee, Mr. Gilman in Washington, and Mr. Witt in New York, thank you all. Good night, Jim.

LEHRER: Good night, Robin.

MacNEIL: That's all for tonight. We will be back tomorrow night -- weather permitting. I'm Robert MacNeil. Good night.