Behind the scenes of long-range forecasting
Exploring what goes into a long-range forecast with AccuWeather's lead long-range forecaster, Paul Pastelok.
Transcript
00:00Welcome to AccuWeather's Ask the Experts. I'm your host, Jeff Cornish. On this show,
00:16we go beyond the everyday forecasts to give you the how and why on all the cool and interesting
00:21things that you may be wondering about and may have wanted to ask about in the fields
00:26of weather, space, and science. And in this episode, we're going to be exploring another
00:30way in which AccuWeather provides superior accuracy deep into the future. While the short-term
00:36forecast is often the utmost importance, people, business, farmers, and also a whole variety
00:42of people benefit from knowing the weather trends coming into the next few months. We're
00:46talking about AccuWeather's long-range team, and they're ready to deliver this crucial
00:51information. So joining us now to discuss this is lead U.S. AccuWeather long-range
00:55forecaster, Paul Pastelok. Paul, great to be with you once again. Thanks for being here.
00:59Hey, thanks, Jeff.
01:00Well, I always love talking to you because you do some fascinating things. And for a lot of us
01:05who spend our time in the next three to seven days in the forecast world, there are a lot of
01:10questions about long-range forecasting.
01:11Well, hey, you do help out once in a while with the wildfire forecast. I do grab you and pull
01:17you out of the short range and get you into the long range at times. Remember that.
01:20I know. I appreciate that. A little bit of a knack for fire weather, the volunteer fire
01:25service. But you've been with AccuWeather for over three decades, and you've become our lead
01:30U.S. long-range forecaster about 10 years ago. So how did that happen? And how did you get
01:36interested in weather just kind of at ground zero?
01:38Well, to be honest with you, it wasn't really my choice. It was funny. I was asked, you
01:46know, when you start a job, you start down low and you have a lot of seniority people ahead of
01:51you, and they have their permanent positions, their set times and ways. I was kind of more of
01:56a filler. I kind of came in whenever someone was off, I filled in for their schedule. And then
02:01one day I got asked to kind of, why don't you include into the long-range department? We have
02:07this one guy that needs a little help. And I said, sure. For about six months, I did that. And
02:12then he ended up leaving. And then all of a sudden, I'm sitting in my office and they come down and
02:18they say, you're in charge. And so I just had to pick it up from there. And it was fast pace from
02:25there because I had to research a lot more. I had to look into getting a team built. And so it was
02:31basically by accident that it just fell in my lap.
02:34And we want to talk more about long-range forecasting. This is a fascinating thing. And I
02:38know that it can be leveraged by big businesses to improve their bottom line. But fundamentally,
02:44how do you produce a long-range forecast? I know there's no real easy, straightforward answer to
02:48this.
02:48I took an approach. I thought about this for a long time. And what I felt, I had to teach myself.
02:57It's something I didn't really quite learn in school. I learned the physics of weather. But long
03:02range and the approach came from me and my team around me that I did develop. But a lot of the
03:09research is done before you do the actual forecast. I mean, you got to look at past data, the
03:13climatology, all those kind of things. And it takes time before you actually start indulging into the
03:18forecast. And then I would look at past years and I developed this what we call an analog chart. And
03:24it looked at past years and some of the what we call teleconnections and sea surface temperature
03:30anomalies and El Nino, La Nina, things like that, we would put in this chart and try to match up to
03:37what happened in the past to what's happening now. And this chart took me a while. You weigh things
03:43and then all of a sudden you get higher years that stick out and it kind of paints a picture of what
03:48you expect coming forward. Then I take that and then I match it up with forecast models. And so when
03:54you put that approach to it, you're getting a lot more involved in just looking at a few models that
03:59are not always that accurate when you look out three, four, five months out ahead. So that's the
04:04kind of approach our team does. And then, of course, everyone's we have a meeting just before the
04:09seasonal forecast gets put out. We get together for about two hours. Maybe I bring doughnuts or
04:14something like that for the meeting and then we put it together. And that's how we come on our
04:17forecast. And a lot of the things that you're looking at on the baseline tend to be things in the
04:22atmosphere that don't change day to day, like ocean water temperatures, as you were saying. Yes,
04:26things that if they are such a way now, they're likely to be on a similar state, maybe five,
04:31seven, ten days down the road. Yeah, exactly. And so those are tough things that you have to adjust
04:36and adapt to. They they could they could change within a couple of weeks and throw off a forecast
04:43because you talk about what we call marine heat waves. OK, and they're becoming more and more
04:48popular to the public now. And they can change positioning of where high pressure areas are and
04:55low pressure areas are. And so this is something we look at that maybe the forecast models don't
05:01pick up right away. And so these are the kind of things that help with our long range forecasting
05:06is picking up those little things that turn out to be big things in the results because of warm
05:11water temperatures that marine heat waves put out. So and there's a lot of conversation about heat
05:16and changes in climate when it comes to analogs, weather patterns that resemble times in the past
05:23that could be kind of tipping the hat to tipping the scales as to as to the long range forecast.
05:28Is it becoming increasingly difficult to to rely on analogs or not so much? It's changing because
05:34our climate's changing and things like the water temperatures are warming more in the mid latitudes.
05:41The polar regions are changing different from what they were 15, 20, 30 years ago. So what we learn
05:48in our research from these papers, okay, from the past may not be the quite the result in the present
05:54time. And that's why with long range forecasting, it's not so much operational. It's looking and
06:00indulging into these papers, the updated ones, especially from new people that show these
06:06changes that are going on. And we have to adapt as well. What we may have saw 20 years ago from
06:11one teleconnection or something could be different from what's going to happen nowadays. And so
06:18it's it's challenging to adjust and make those changes because it could change your entire
06:25forecast. When I started studying meteorology at Penn State, John Porter was in my class. Okay. He
06:31was one of two in my class. We graduated with 34 students and he was one of two who had ever seen
06:36a computer model before day one of college. Now I just interact with enough students on Twitter or
06:42X that it's clear that almost anybody interested in meteorology now is already looking at the
06:47models coming into college. It's so accessible now compared to 25 years ago. So if you're a
06:53future forecaster, you want to do meteorology and you want to do long range meteorology.
06:57What would you give as advice for a future long range forecaster to work on their skills at this
07:03point? I'm a pattern recognizer. I like to look at patterns and kind of try to see what that looks
07:10like from year to year or from past years. There are so many forecast models compared to when I
07:15first came into the business. So there are many out there that will strictly go with the models.
07:20I see it on Twitter. I see it on other things that say, hey, this is showing this. But you still have
07:25to look at the physics and research involved with these every pattern from season to season,
07:32month to month, whatever. And I say, make sure you do the work, do the research, look at this stuff
07:38ahead of time. Know the all the teleconnections, know how the water temperatures react from
07:44different seasons. You got to have all that in your library. It's not just looking at the
07:51forecast models. I know they can do much more than they did in the past, but they can be off. And
07:55you're talking about a forecast that can get derailed very easily in the first month and it
08:00could affect the following month. So you need to, I always like to look at what can go wrong in this
08:06forecast better than what's right with this forecast. And I think taking that approach
08:10sometimes is another advice to people out there doing long range forecasting. And we use the
08:15consensus-based approach to forecasting here at AccuWeather. So it's not just one person's opinion.
08:20You lead the team, but you also bounce things off of your team members. I have members that are
08:25experts in wildfire, severe weather, drought, and having another pair of eyes and then bringing all
08:32that knowledge into the forecast is great. And sometimes I can't remember it all in my head.
08:37So I have a series of notebooks. In fact, I have about 10 of them still where I write in for
08:43different types of things and it keeps me organized. And if I forget something, I can look back.
08:47That's great.
08:48Something from three, four years ago. So I still do it kind of the old-fashioned way,
08:52but it helps. It keeps me organized. And that's another thing. Stay organized when you're doing
08:57these long range forecasts. Once you start getting on, it'll throw you off.
09:01Okay. Good advice. Well, we do have a viewer question here and we want to get to this
09:06question from Chris in Minnesota who writes, what goes into generating a 45-day or a 90-day
09:11forecast? These exist on AccuWeather.com and AccuWeather Pro site and so forth.
09:15Absolutely. We do a 90-day forecast every quarterly. And what we do is, again, the same
09:21approach. We start looking at past years that match up to what's going on or what has happened
09:26in the past. We put all that information. Is it a La Nina year? Is it an El Nino year? Is it going
09:33to be this, this? The water temperatures globally, too. We look on the other side of the globe.
09:37We put it all together and then we come up with the modeling side by side. I put it all in a
09:43PowerPoint, present it to my team, and then we sit there and we adjust it and make changes and
09:49come up with a consensus forecast that we can present out. And that's how we generally do
09:54our approach. Okay. We use Slack to do a lot of our internal communication here at AccuWeather.
09:59I lurk on the long-range channel sometimes. There's a lot that goes on there. I learn a lot
10:03just by reading your stuff. And I throw that out there so I can get comments, push some of my team
10:10to say, hey, is this looking right to you? Because this is what it looks like to me. And
10:15I get responses. And I think that's how you get the best forecast out of everybody. Okay. Very
10:21good. Well, coming up next, Paul and I are going to discuss some different seasonal long-range
10:26forecasts that he and his team provide here at AccuWeather. And we're going to talk about
10:31differences in forecasting for each when we look that far into the future. And coming up later,
10:36we're going to take you back in time with the incredible story behind what's been called the
10:41most important weather forecast in history. Stay with us. We have plenty more coming your way.
11:05Welcome back to AccuWeather's Ask the Experts. I'm your host, Jeff Cornish.
11:09And we are talking about AccuWeather's long-range forecasting team. And our expert today is
11:13AccuWeather senior meteorologist and lead U.S. AccuWeather long-range forecaster,
11:18Paul Pastelak. And, Paul, again, we appreciate your insight. Glad you're making time for us.
11:22We do have some more questions for you today. All right. Sounds good. Far away.
11:26Well, we do long-range forecasts for different seasons. We have spring, summer, fall, and winter.
11:30So which season, even hurricane season, another big one that is very important,
11:35but obviously which season is the most challenging? Some are easier than others, I'm sure.
11:41Spring. I have to say, I say that all the time when I get asked that because there is so much
11:46that we put into our spring forecast. It's not just about temperatures. It's not about precip as
11:52much. It's what it's people ask me, is it is it going to be a really rough, severe weather season?
11:58Is it going to be is it going to be tough as far as drought or flooding? Which one? Those specifics
12:04is what we put into our spring forecast. You can see some of the graphics on the screen.
12:09We pride ourselves in hitting, you know, kind of more the more interesting, the more specific
12:15things. And we put it out there for a three month forecast. Not many other outlets do that.
12:20OK, so they they kind of break it up a little bit and put special forecasts from here to here. But
12:25we put it all together into one forecast. And an example, one of them, the late frost forecast.
12:31I mean, we do a lot of research into what's the normal frost time period,
12:35when we think it's going to be faster for some places and later for others. So that is something
12:42that we pride ourselves in. And it takes a lot of time to put these specific. And there's a lot of
12:48it in the springtime. And I know that when it comes to wintertime, there's always a big appetite
12:52for that snowfall forecast, especially in the northeast. A lot of other forecast operations,
12:58they put some generic words on a map. You put actual numbers and forecast ranges for some
13:03specific cities and we put it out. We start a preview one like in late August, early September,
13:07and then we update throughout the season. And of course, everyone wants to know when the last snow
13:13is going to be. And so that carries over into the spring season. So we do a lot of that stuff
13:17as well in spring. And once we put out a seasonal forecast ahead of time, does it just sit on the
13:22shelf or do you continually update that at certain benchmarks as we move toward that season?
13:27We'll put out a three or four times the snow forecast, whether it's changing or not. A little
13:32adjustment here and there because things can change very drastically depending on who gets
13:38the snow and who doesn't and storm tracks as well. So we will update what we need to update,
13:43make the adjustments and go from there. So we don't let it sit on the shelf at all.
13:48And before the break, we kind of hit it at this. I think this is one of the most fascinating things
13:52at AccuWeather. Who are the recipients of some of this information? There are some big businesses
13:58that can leverage this intel in some really unique ways. Some of our bigger clients are
14:02insurance companies. They need to know where they need to properly put people if there's going to be
14:06a severe weather break or a frequency of severe weather in a certain area of the country. So
14:11they'll make sure they'll highlight that area. Another thing too is retail. A lot of your big
14:16stores and chains, they want to know ahead of time where they have to move their product, have it in
14:21distribution centers and not to overload or not have enough. You need to have that,
14:26know that supply and weather does affect how these businesses adjust. And so we have some
14:32of those bigger clients asking us all the time, not just one season, sometimes two seasons ahead.
14:37I believe that. That's interesting. And some of these clients are huge. So if we can help
14:40their bottom line, they do something on a huge scale by even 1%. That's a big difference.
14:45And even the emergency management teams in certain cities look to us ahead of time. I've
14:52done some presentations in the past to let them know this is how far out we can. They always ask
14:57that question, how far out can you predict this? And we let them know that. We let them know this
15:02is where we're at right now. We can give you some preview stuff and then update as we go along. And
15:08that helps them very much in preparing in case of a tropical system or severe weather outbreaks.
15:13They probably appreciate the confidence level and the transparency that you sometimes express
15:17as well. Well, it's time to answer some more viewer questions. We want to hear from Greg
15:21in Atlanta. And Greg, what do you want to ask the expert? Yeah, how accurate are long-range forecasts?
15:31How accurate are forecasts? Depends on what you're talking about. Are you talking about
15:35temperatures? Everything has got a different way of measuring as far as accuracy goes. And
15:42we feel that we are pretty darn good. We're up there. As far as getting things done, we have
15:47looked back in the past and compared our numbers. And we're pretty steady, especially with
15:53temperatures. Severe weather trends is really one of our highlights. Forecasting the number
16:00of tornadoes can be difficult, but forecasting the trends, the areas of the country, we do very good
16:06at that. And also getting people out on tropical, tropical season of forecast, another area that
16:11we're very accurate as well on ahead of time. So I think we do pretty well. And we really take
16:18our chances and push the limit on some of these other detailed type of things like drought and
16:23flooding and those things. And if we always stayed back where the historical comfort level was, then
16:28we'd never move the ball downfield in terms of advancing the science and improving the forecast.
16:32Absolutely. People would just go with, you know, climatology all the time. And
16:36that's not where we're at at this point, right?
16:40No.
16:41Well, our next question comes from Taylor in New Jersey. And Taylor writes,
16:44do you consult the old farmer's almanac? Sometimes even, you know, I'll get questions
16:48like these just as a meteorologist. Taylor, it's in my backpack all the time.
16:54It's always there. I always have one. It never goes without me. I mean,
16:58there's some really good, interesting stuff in there.
17:00The sun and moon, right?
17:01Absolutely. I can't remember all that stuff.
17:04Right. How about the forecast intel, though?
17:06But the forecast, the pretty pictures, I get it. It's entertaining. It looks nice.
17:13You know, I do look at it. I look at everything. You know, I mean, I see what other people are
17:17doing. You know, where are they from my forecast? You know, I don't copy their forecast. I look at
17:23to see maybe they're seeing something that I don't see at this point. And then I indulge in looking
17:28in that area and say, hey, wait a minute, maybe this is, you know, on the right track,
17:32make some adjustments here. I will do that from time to time, but to be honest,
17:36but most of the time I just do it just to see, you know, the comparison to ours.
17:40All right. It's good to be aware of what other people are talking about
17:43in general, even just in the public. Absolutely.
17:45Well, then we'll wrap up our question and answer segment. We do want to thank AccuWeather's lead
17:49long range forecaster, Paul Pastelak, for joining us on this edition of Ask the Experts.
17:54After the break, we're going to have the incredible story
17:56behind the weather forecast that changed history.
18:04Welcome back to AccuWeather's Ask the Experts. It is time for weather wise. And in this episode,
18:09we talk about what's been called the most important weather forecast in history.
18:13World War II, 1944, American, British and Canadian troops are preparing
18:18to invade the coast of Normandy. And weather played an enormous role in what we now know
18:23as D-Day. So before radar, computers and satellites, weather prediction was based on
18:29past records or surface observations, meaning predicting more than a day or two in advance
18:34was nearly impossible. Plans for the D-Day invasion required a triple assault, and the
18:39weather was crucial for each part of the mission. Pilots needed clear skies and a full moon for
18:44visibility. Naval operations needed calm seas with low winds to get troops ashore,
18:51and then ground troops needed a low tide to see German obstacles waiting them on the beach.
18:56The original plan was to storm the beaches on June 5th, but General Eisenhower's chief
19:01meteorologist, Group Captain James Martin Stagg, made a pivotal forecast. He advised waiting one
19:08day until June 6th. Going in as planned, June 5th, with potentially rough and stormy seas in
19:13the English Channel, would have jeopardized the operation. Waiting an extra day could allow
19:18Germans to see the invasion fleet assembling off the southern English coast, though.
19:22So with favorable weather on June 6th, the successful Normandy invasion took the Germans
19:26by surprise, and it was the beginning of the path to victory in Europe and eventually the end of
19:31World War II. Thanks so much for joining us, and remember, when you have a question, you can email
19:36us at asktheexpertsatactiveweather.com. You can also call us at 888-566-6606. Thanks for joining us,
19:44have a great one.