AccuWeather founder Joel Myers explains how his State College, Pa.-based company and the business of TV forecasting have changed over the decades, the qualifications needed for calling the weather on TV and the role of government in the now heavily commercialized business of forecasting. This is the first of a three-part special report on weather. Part 2 will appear Wednesday morning and part 3 will run Thursday.
Internet, Mobile Extend AccuWeather’s Range
Joel Myers and AccuWeather have been part of the local TV news scene since 1972 when the weather service began providing TV and radio stations with market-exclusive forecasts.
Since that time, AccuWeather has jumped the bounds of broadcasting into other media. It sells weather info to newspapers and digital media clients. In addition, it operates its own cable networks, a well-trafficked website and mobile apps — all with the popular brand or some variation of it.
Myers founded the company in 1962 while still a graduate meteorology student at Penn State. In the early years, his clients for were non-media businesses and government agencies in need of accurate forecasts.
Although it has offices in New York (sales) and Wichita, Kan. (non-media weather consulting), AccuWeather remains based in State College, Pa., the home of Penn State, with which Myers has strong ties. He received his B.S., M.A. and Ph.D. there, served on the faculty until the demands of his business became too great and then served on the board of trustees for many years. He was named trustee emeritus in July 2014.
In 2007, Myers handed off the job of CEO to his brother and longtime business partner, Barry, but he remains the chairman and president, the heart of the company and an international authority on meteorology.
In this interview with TVNewsCheck Editor Harry A. Jessell, Myers talks about how his business and the business of TV forecasting have changed over the decades, the qualifications needed for calling the weather on TV and the role of government in the now heavily commercialized business of forecasting.
An edited transcript:
You’ve been at this for more than 50 years. What changes have you seen in the science of meteorology over that span?
Tremendous changes. When I started, we used paper maps. The amount of data was much smaller than it is today, the value of the computer models was much smaller, the forecast visibility varied widely. Accuracy depended mostly on the forecasters — their pattern recognition, their intelligence, their intuition, their experience and their ability to just stick with a weather pattern virtually 24/7 and get a feel for it.
What was the biggest change in technology?
Computer modeling that uses the most advanced mathematics and differential equations to explain the motion of the atmosphere and more powerful computers with higher and higher speeds. That progress really drove increased accuracy of forecasts further out.
Now, clearly: garbage in, garbage out. More and better observations have improved the accuracy and the benefit of the computer modeling. Satellite imagery has helped because it has provided details from over the oceans, which were data-sparse areas.
Is there a way to objectively measure the accuracy of forecasts over the years?
Oh, absolutely. There are many measurements that have shown that the accuracy has increased tremendously. When I started, the average error for a one-day forecast of the temperature was four to five degrees. Today, it’s down under two degrees. We have valuable statistics that show AccuWeather is the most accurate. I have dedicated my life to making sure that we are the most accurate, and improving the accuracy of our forecast continuously.
We have run a comparison against the National Weather Service forecasts in Washington over 20-some years, day by day, month by month. We beat them by an average of 11% on temperature forecasts. There have only been a handful of months we haven’t won.
We have beaten the Weather Channel, out biggest competitor, over the last three months on high and low temperatures by 7% to 9%. There are similar results on precipitation, but temperatures are easiest to compare.
How about weather reporting on TV? How has that evolved?
It has improved along with the science. AccuWeather was the first to produce a five-day forecast. When we started in the 1960s and into the ’70s, the forecast available from the government was today, tonight and tomorrow. That was it. That’s what most TV stations presented.
When did the five-day forecast come in?
We started with ch. 6 [WPVI] in Philadelphia in 1972. That was revolutionary. Then we went first with a seven-day and then the 10-day. We pioneered this right along — then the 15 and the 20 and the 25 and the 30. Now on our site we have, day by day, 45-day temperature forecasts for three million locations around the globe. We are the only people that have that.
What’s the margin of error on the 45th day?
It’s about the same margin as the fifth day of the five-day forecast when we started that in 1972. It’s not nearly as accurate as the forecast for today, tomorrow and the day after that and so on. The accuracy deteriorates with time.
Again, measuring skill has to be measured against climatology and persistence. In other words, if you have all the statistics for where you are, of what the past weather has been on that date, the normal high and low, you know that this summer is running a little cooler than normal, you factor that in, you set up an independent view or forecast of that. If we can beat that statistically, then there is skill. That is the meteorologist definition of skill. We have skill up to 45 days. That’s why we do it. We are a scientific organization.
TV stations don’t generally, at least not around here [New York], go beyond seven days.
Yes, that’s correct because on TV you have limited time. Most stations only go five or seven, a few go 10 because that’s the main focus clearly for the people watching. You want to keep it moving along — what’s going to happen in the next few hours, what’s going to happen this evening, what’s going to happen tomorrow and then for the weekend.
So that makes sense for a two-and-a-half minute weather presentation. You want to focus more on the locality, the variations, the details of the forecast over the upcoming hours or days rather than the long range. But we do have some stations that have that capability to go out 45 days on our StoryTeller system and display it in a graph.
What kind of education and expertise does a TV weather person need these days or ideally have?
It varies all over the place. You have a few Ph.D.s in meteorology. The majority have bachelor’s degrees, often in meteorology, but even there the skills vary. Some went to schools with full-fledged meteorological programs with a great pedigree, others may not have. Some may have a degree in something else or something related. Some may have gotten degrees online from Mississippi State, which has a good solid online program, but it’s not the same as being in a classroom for four years.
Some are very good communicators. They do an excellent job. They may not even have a degree in meteorology and that’s fine. They have some of the highest ratings because they can relate and they have got a lot of experience.
I don’t want to criticize or be negative about any because in the end a number of the meteorologists do rely on data from us or our competitors. Others take great pride in doing their own forecasting and that’s fine. We work with all of them no matter how they want to work with us.
But our argument is that even if you are a great forecaster, having experts here to talk to and being able to banter back and forth will make you better because the weather is never clear-cut. There are always questions.
From the start, AccuWeather has embraced digital media. You are now big into mobile. What’s worked best for you?
They have all worked. When the Internet came along, some of our in-house people said, you don’t want to put AccuWeather on the Internet for free because you have your business customers that are paying good money for this information. I said, listen, we don’t control the world. The Internet is there. We’re going to participate.
At a convention in Atlanta, a speaker said you have to decide whether you are going to have a subscription site or a free site. You should not do more than one. And I got up and said, I don’t agree; you can do both. I said we’re not making any money yet on the free site, but we know once we reach critical mass we will sell advertising.
We now have more and more people subscribing to our subscription sites, AccuWeather Premium and AccuWeather Pro, and we are selling content to other sites more and more. In the beginning, the most money came from selling to other sites.
What about now?
Now there are still two very vibrant parts of our business. Now we get the most money from advertising because we have four billion page views a month on the free website (accuweather.com). We are up to 12 billion requests for data a day from around the world. That’s a 160,000 a second. We respond to each one of those within three tenths of a second wherever it is from — South Korea, North Korea, Buenos Aires, London.
So how is your business split between media and non-media revenue?
We are privately held so we don’t give those figures out, but I can tell you that all of them are growing.
Which business is bigger?
We have more media than we have business. Media are a bigger share in terms of overall revenue.
What kind of non-media customers do you serve?
We serve all the major railroads in North America. A lot of retail. You would recognize all of them. Everybody from oil drilling to construction, insurance companies for planning and making decisions, just about every type of industry.
You have been critical of the National Weather Service for competing with companies like yours. Do you still think that the NWS is out of bounds?
When I started this business, 98% of the forecasts reaching the public came from the National Weather Service. The private sector provided less than 2%. Today, 98% of the forecasts comes from the private sector and 2% comes from the government.
Now why is that? The reason is that the private sector — AccuWeather, the Weather Channel and so on — provide weather forecasts that are more useful or are more accurate or more relevant. It didn’t happen for any other reason, OK?
When they take resources and don’t have the best observations and don’t have the best computer models and don’t have the best warnings and they have people that are out there taking the products that we have developed and trying to copy them and offer them with taxpayer dollars that most people aren’t seeing anyway, it is a waste of resources. It hurts the entire enterprise and it hurts the entire U.S. economy.
Still, the government has a key role.
Three things: they ought to have the best weather observations, they ought to have the best computer modeling and they ought to issue warnings to save lives. That’s what they should be focused on because everything else has already been commercialized.
Are you happy with what they are doing now?
By and large. They are moving in the right direction, yes. There is a lot of progress that has been made, a lot of the things that we suggested that were considered radical 20 or 25 years ago have been adopted to the credit and the benefit of the American economy and America’s scientific leadership. I have acknowledged that in my keynote speech at the American Meteorological Society in January. I talked about that in the summer meeting of the AMS last year.
One of great public policy debates is over climate change and how much of it is caused by human activity. As a practical matter, does climate change affect your ability to forecast the weather?
No, because with climate you are talking about changes over decades and centuries. It has made people more aware and more sensitive and of course some people are claiming that the dramatic weather changes are more common due to the warming. I question that. There have always been extremes.
So, it’s an academic discussion for you?
In terms of weather forecasting, for the most part it is. Now when you set high temperature records, it’s a little more than an academic discussion, but it doesn’t, beyond that, affect the predictions on a day-to-day basis.
What’s coming in terms of weather forecasting and your ability to make them more accurate?
The accuracy will continue to improve. We will extend the temperature forecast beyond 45 days. You know, we now have MinuteCasts that predict the start and end of precipitation within two hours. That will be extended.
Plus, we just released the AccUcast app that was featured on the Apple Store. You can go and look at it. It has crowd-sourcing reports so we are going to be getting millions of reports from users around the world that will feed into our models.
The more observations, the better the forecasts.
This is the first of a three-part special report on weather. Part 2 will appear Wednesday morning and part 3 will run Thursday.