AccuWeather Shocker Forecast: December To Be Colder Than July

When my editor asked me to write about AccuWeather’s recent launch of a service that predicts weather conditions nearly three months into the future, I jumped right on it.

Weather forecasts for 90 days out?

Based on everything I knew, such a feat was impossible. And I’ll confess, I enjoy writing about dubious scientific claims and the people and companies who push them.

Would you trust weather forecasts made three months in advance? Let us know in the comments.

Here’s the basic problem with a 90-day forecast: Daily weather patterns are just too complicated to make any kind of precise predictions for more than about two weeks in advance.

My initial story was published Monday along with my joyously unscientific headline: “90-Day Forecast? 95% Chance of Bunk.”

This is when the AccuWeather, whose spokesman had initially left me hanging when I asked tough questions, finally got back to me.

He saw my original story – and apparently wasn't especially tickled by the headline – but he still wanted me to hear from a real AccuWeather expert.

Just talk to Vice President of Innovations Jon Porter and maybe I'd see the light.

Intent on finding even one aspect of AccuWeather's claim that I could accept, I listened to Porter. Maybe I had been overly skeptical. Maybe I just didn't understand any of the science I had reviewed and that had demonstrated, convincingly, that a 90-day forecast is about as reliable as one predicting the weather nine years from now.

So I listened some more. I was trying. I swear.

And then I needed to ask only one question to prove to me — and I like to think the spokesman and his expert — that AccuWeather had about as much chance predicting the weather 90 days out as it did controlling the weather next week.

Here was the question to Porter: See where you have thunderstorms forecast on Saturday, July 6, 2016? I want to have a party that day but I'd rather my guests not get soaked or struck by lightning.

Should I cancel?

He told me — and I admittedly am paraphrasing here just as I paraphrased the question — to party on.

Despite a forecast of thunderstorms, Porter admitted that there was no actual reason to believe that thunderstorms are particularly likely that day in this new, 90-day forecast.

It still might rain! But it might not. Anyone could have told me that.

In sum, it took AccuWeather more than a full day to acknowledge that its 90-day forecast does not include an actual forecast for any particular day.

Porter explained: The 90-day predictions are “a useful trend tool,” which users appreciate, and which is “relevant to the way they’re used to looking at weather forecasts.”

He continued, “What we’re talking about is not necessarily the specific high temperature on a particular day.” Rather, Porter argued, the 90-day forecast reflects broader trends about the type of weather than can be expected within a certain time frame.

Except if you look at AccuWeather’s actual forecasts, this is not at all the most obvious reading of the graphic. It looks just like a typical 5-day forecast, only stretched out over several calendar months.

Here’s what it seems AccuWeather is doing. It's taking some empirical data that suggests that, say, July 2016 will probably have warmer than average temperatures in a particular area. Then they create a set of probable weather conditions that would fit this trend. There’s good reason to think there will be some thunderstorms in July, too, so they add a few thunderstorms to their "predictions," essentially randomly, across the calendar.

As those future days get closer, they adjust the predictions until they enter the 14-day window of being actually somewhat predictable.

So here is what we have: AccuWeather touting forecasts to the company's scores of users who might use them to plan family activities, school events, or even work schedules.

And why not? In an AccuWeather release, it explicitly says that the 90-day forecast is “a valuable tool for planning the best time of the season for road trips, vacations and outdoor activities.”

The most impressive part of the 90-day forecast is, in fact, not very impressive at all. It is this: It can tell you with some certainty that snow is more likely in the winter than in the summer.

And one more thing the forecast shows, this with absolute certainty: Some companies have the snowballs to say just about anything.

As for my party on July 6? My screenshot of the prediction shows the thunderstorm clearly. But a few hours after I asked Porter about my party, somehow the prediction – around 75 days out – suddenly got much sunnier.

Photo Credit: Kiyonobu Ito via Flickr