The weather is feeling less predictable than ever. Which is why, when AccuWeather announced a brand new 90-day forecast tool last week, it sounded a little good to be true. Is it? Yes. It’s highly misleading, and meteorologists are not having it.
“AccuWeather is putting out a product that has no demonstrated value, and they’ve never proven otherwise,” Jason Samenow, a meteorologist with the Washington Post Capital Weather Gang, told Gizmodo.
Weather forecasts are notorious for being wrong, and their accuracy diminishes sharply with time. Most meteorologists agree that daily forecasts beyond 7 to 10 days are borderline useless.
But not AccuWeather. The private forecasting company has spent years pushing the envelope (a euphemism for “sparking the ire of the meteorological community”) by issuing longer-term forecasts than anyone else is in the game. In 2013, AccuWeather began releasing 45-day forecasts, prompting Samenow to conduct an independent analysis of the tool. His conclusion? It’s a joke, offering no more accurate information than historical average conditions.
It’s no surprise that AccuWeather’s 90-day forecast tool is already raising eyebrows. I didn’t want to write it off immediately, so I reached out to AccuWeather to learn how accurate (get it?) the company itself believes a 90-day forecast really can be. I asked AccuWeather if it can estimate a probability that its 90-day forecast will be correct, and how the reliability of the 90-day forecast compares with that of shorter term forecasts.
“There’s a lot of ways to look at that,” Jon Porter, vice president of Innovation and Development at AccuWeather told me, before launching into a canned response on topics ranging from information availability to proprietary weather prediction technology.
AccuWeather offered no estimate of the accuracy or reliability of its 90-day prediction tool, although Porter did stress that users have found the tool to be “valuable and interesting”. Like the 45-day prediction tool, the 90-day forecast appears to collate information from a range of weather models, as well as an extensive historical archive.
I asked whether I could use AccuWeather’s 90-day prediction tool to plan a wedding in July. Porter emphasised that the 90-day prediction tool is more about long-term trends.
“With regard to planning a specific event, we’re always telling people not to focus on a specific day, but to look at the time period around it to get the trends,” he said. “Is it going to be a dry, cool time, or a warmer, wetter time?”
Sounds reasonable enough. At this point, I had not actually seen the 90-day forecast tool. So when Porter and I got off the phone, I went over to AccuWeather’s website to take a peek at the long-term outlook for Philadelphia. Based on our conversation, I expected to find a tool that presented weekly or monthly temperature and precipitation estimates out to 90 days, maybe even some margins of error. Instead, this is what I found:
Image: AccuWeather screenshot
There’s no nice way to say it: this forecast is nonsense. And I’m pretty sure AccuWeather knows it — the company itself told me that the 90-day forecast tool is “more about long-term trends than individual days”. How on Earth the average layperson is supposed to glean that from a tool that literally presents individual, day-by-day forecasts out to 90 days, with no obvious disclaimers, is beyond me.
A long-term forecast for Ocean City, NJ Image: AccuWeather
Many people won’t. Instead, confronted with today’s 90-day forecast for Ocean City, NJ, a New Yorker may well decide to book her favourite beachfront property for the 7th through the 9th, rather than risking that thunderstorm (60 per cent chance!) on Sunday, July 10th.
I’m not blaming AccuWeather for wanting to provide folks with long-term data, and it’s possible that these 90-day forecasts are somewhat better than what I’d come up with drawing random forecasts out of a hat. (How much better? I really can’t say.) But the company needs to be real with folks about what it’s offering, and display the information in a manner that emphasises the uncertainty. During our conversation, Porter stressed that AccuWeather believes the public has a right to as much forecast information as possible.
“Meteorologists have this type of information available,” he said. “It’s a question of making the information available to the public and taking a leadership role.”
To meteorologists, however, the idea that you can’t trust a 90-day forecast is second nature. As Samenow puts it, “I’m not going to look at models that provide data out to 90 days because I don’t think they’re any good.”
But to the average human who wants a tool that can answer questions like “What will it be like when I go on holiday?” AccuWeather’s new forecasts will surely disappoint.