AccuWeather, Man, What Are You Doing?

AccuWeather published a story today predicting that tornado activity this year will focus on springtime twisters in Tornado Alley and that the United States will see 1,075 tornadoes in the year 2019, coming in six percent below average. Oh, word?

The statement offers no qualification. They don’t show their work. There’s some text about what they expect this year and what’s normal and how tornadoes get ratings, but there’s nothing more about that extremely specific number that says “this is an estimate” or “here’s the margin of error” or “we arrived at this number by taking Phil the Groundhog hostage and pumping him full of M&Ms until he was wired enough to shake a crystal ball and see through the time-space continuum.”

Nope. Just a matter-of-fact statement: “AccuWeather predicts there will be 1,075 tornadoes in 2019, which is nine percent more than the 987 tornadoes in 2018.”

Okie dokie.

The company started releasing 45-day weather forecasts earlier this decade, stretching them out to 90 days after a couple of years. The company says that these forecasts are really trends to help you get a general idea of the weather in a few months, but they’re…uh, not accurate…and, even as they’re justified as “just trends,” they’re certainly presented as a detailed weather forecast to their target audience.

For example, AccuWeather’s forecast for my town for Sunday, May 26— 90 days from today—shows a high of 74°F and a low of 54°F with a 40% chance of thunderstorms. Their forecast for May 26 issued on February 26 predicts an east wind of 7 MPH with gusts to 15 MPH. They say that the Sunday three months from today will see rain for three hours that amounts to two-tenths of an inch.

Trend? No. And the tornado forecast they issued today isn’t a trend, either. They know that people are going to read this PR stunt literally. Whether or not it’s accurate doesn’t even begin to approach the fact that it’s not scientifically justifiable to predict an exact number of tornadoes for the year or tell us what the weather’s going to be like with mind-bending specificity 2,160 hours from now.

A specific number of tornadoes doesn’t really help people much, anyway. What’s 1,075 tornadoes in the grand scheme of things? How many tornadoes constitutes an outbreak? How many tornadoes does it take to destroy your house? You could make basically the same argument when it comes to hurricane season predictions, but at least those involve ranges and some general vagueness.

I’m sure that some news organization or weather blog is going to reach out to AccuWeather for comment. They’re going to try to get an answer for how AccuWeather arrived at this extremely specific number. Whoever speaks with them will likely talk in circles and avoid answering the question. It’s proprietary. It’s a trend. Take it seriously, not literally. But clarification after the fact doesn’t matter. The statement is out there. The number is out there. That’s what they chose to present to the world.

AccuWeather predicts 1,075 tornadoes this year. And yet, somehow, they couldn’t predict the reaction to last month’s marketing stunt putting down National Weather Service meteorologists to boost their own products during the lengthy government shutdown that caused many of those hard-working government scientists to drain their bank accounts to zero to stay afloat.

I should point out that there are a lot of fantastic scientists who work for that organization and I consider many of them good friends. Stuff like this, though? This is coming from the top. And as long as the top-level culture gets a rush whenever the weather community condemns their latest marketing stunt, this is the kind of stuff we’re going to have to deal with.