The Washington PostDemocracy Dies in Darkness

Meteorologists say they are sick of clickbait and misleading social media weather forecasts

On Wednesday, the American (GFS) model was predicting snow from Washington to Boston next week in its forecast eight days into the future, which is not reliable and has no credibility. Some Facebook accounts shared this forecast.
Placeholder while article actions load

Like so many others, it was a dire forecast.

“The HUGE DEADLY OCTOBER BLIZZARD & SNOWSTORM is still on [its way] this upcoming Thursday [and] & Friday October 10th-11th 2019!!!!!” read the viral Facebook post. It’s a post which amassed more than 4,000 shares after being published Oct. 6.

Its parent page, titled “METEOROLOGISTS,” has a following of more than 30,000 accounts. Some anxiously braced for what was being advertised as a “Dangerous Early OCTOBER MONSTER BLIZZARD!!!!!” The post called for snowfall amounts of up to 25 inches, with a map indicating that the heaviest would fall in Minnesota.

In the end, no place in Minnesota saw more than 5 inches in three days.

Five inches was also the maximum in Nebraska, while South Dakota got a jackpot of 14.5 inches — in the part of the map that had shown less. Yet METEOROLOGISTS’ social media posts often receive an order of magnitude more exposure than those of the local National Weather Service office in Sioux Falls, S.D.

“Verify your source, avoid clickbait, and be wary of exaggerated forecasts,” wrote Peter Rogers in an email, the warning coordination meteorologist at the Sioux Falls office. “We hope keep people safe and protected from hazardous weather.”

It turns out the popular Facebook page “METEOROLOGISTS” is not operated by a meteorologist. Page manager Richard Cabney says he doesn’t hold a degree in the field.

“I believe I do better than most TV Meteorologists and the National Weather Service on most occasions,” Cabney wrote in an email. “I have been Reading NOAA Weather Maps & Forecasting the Weather since 1983 when I was 15. I got my Weather Maps at the TV Stations …”

This Facebook post was just one of many that professionals in the weather industry have found overselling the severity of weather events, generating great concern.

Beware of extreme long-range forecasts

Long-range forecasts are a particular problem.

On Wednesday, for example, a number of Facebook pages shared a forecast from the American GFS model showing heavy snow from Washington to Boston next Thursday, eight days into the future, when such predictions are not reliable. Other, more accurate models showed no such snow threat.

But with that shaky forecast propagating through social media, several meteorologists felt compelled to confront it.

“I will not buy into the hype. I will not post a phantom map,” posted meteorologist Justin Berk, who provides forecasts for Maryland on Facebook.

“I am saying that this event is NOT likely to happen even though you may be seeing this image being shared,” wrote meteorologist Dave Tolleris, who provides weather commentary for the Mid-Atlantic on his Facebook page WxRisk.

By Thursday, the American model no longer showed the snow threat.

Such instances of hype-filled and suspect long-range forecasts routinely appear on social media and catch fire.

One Dec. 23 post from “First Coast Tropics Watch” picked up 2,200 shares on Facebook after showing an apocryphal raw model output blasting northeastern Florida with snow on Jan. 4. The high temperature in Jacksonville ended up being 74 degrees that day and there was no snow to be found.

“You can look at some climate information about things like that happening,” said Andrew Shashy, a meteorologist at the National Weather Service in Jacksonville. “Given the location, not realistic.”

On Jan. 13, the same page posted a GFS model forecast depicting snow in Alabama and the coastal Carolinas. That system never materialized.

“We put it out there and we let the free people think for themselves,” wrote the page manager of First Coast Tropics Watch, who declined to give their name, in an email to The Washington Post. “It captivates an audience and keeps them interested in the weather.”

But, the page manager wrote, “Safety was and is our first concern."

The page’s manager noted that they “always write a disclaimer” about long-range model outputs and forecasts.

“Working for a news outlet or publication and getting paid for it, you are limited to 5 day sometimes bland forecasts,” the page manager wrote. “Sometimes, you all get it right and sometimes you don’t. I have the freedom and ability to go out in time. Sometimes the models are right and sometimes they’re not.”

Peer-reviewed research has shown that it is not possible to produce accurate forecasts of specific conditions beyond 7 to 10 days — and frequently the limit is less than that.

Study says ‘specific’ weather forecasts can’t be made more than 10 days in advance

The practice of hoisting long-range model forecasts is also at odds with the American Meteorological Society’s statement on best practices for social media use.

“Only a skilled forecaster has the knowledge to recognize these limitations and their effects on the computer model forecast,” writes the AMS.

Sensational forecasts sell

Often, the more extreme the social media forecast, the more exposure it gets.

“We’ve all seen those pages that call for the worst-case scenario in every severe weather [situation] or winter storm, and those that share ‘fantasy land’ data to stir the pot,” said Beth Carpenter, a forensic meteorologist and owner of Thermodynamic Solutions LLC.

A Facebook bulletin from the page Fulltilt weather which hyped a January severe weather situation in the South was shared more than 22,000 times. It warned of a “severe risk for long track tornadoes” and “hurricane force winds between 60 to 90 mph” for “the whole eastern portion of Texas.”

This forecast was much more strongly worded and dire than the National Weather Service forecast for scattered damaging winds and isolated tornado potential.

The page manager of Fulltilt, who declined to provide their name, did not respond to an inquiry about the accuracy of the page’s forecasts, instead writing that “people like you get paid for a story … and the end of the day people['s] lives are all that matter."

Meteorologists express frustration

Many professional meteorologists fear that such unrealistic and sensational forecasts undermine public trust in actual meteorologists — trust that such professionals have spent years or decades earning. And it’s that trust that can be critical during impactful and at times life-threatening weather.

“False information is a big problem for us,” said Nick Lilja, chief meteorologist in Hattiesburg, Miss., and an adjunct professor at Southern Mississippi University. “I spend about one-third of my day telling people that what they saw from [some] Facebook or Twitter account isn’t accurate.”

“Literally anyone can post a picture of a weather model showing a ‘bread and milk run’ triggering snowstorm,” wrote Brad Carl, a former Oklahoma broadcast meteorologist, and now a climate specialist at the Nature Conservancy of Oklahoma.

“Unfortunately, it’s common on social media to run across, say, a cherry-picked single image of a weather model showing a potentially extreme snow event or severe weather setup more than a week out.”

Cherry-picking is a major theme in many of the weather posts that go viral.

“News flash: not everything you read on the internet is true,” wrote Howard Bernstein, a meteorologist at WUSA-TV in Washington, in an email.

“Thanks to the internet, every one has access to weather models, but a lot of people have no understanding what they are looking at. So, anybody who wants to can post a model image and write their ‘interpretation’ as a so-called expert.”

Bernstein agrees that it makes the lives of meteorologists even more difficult, since they’re the ones who have to mop up the mess.

“Then we hear that so and so [said] there was going to be a foot of snow, then we have to deal with posts, calls, etc.”

“In these situations, professional meteorologists receive hordes of questions asking things like, ‘why are you not showing these big snow totals?’ ” Carpenter wrote. “Meteorologists then have to try to justify why they aren’t calling for the storm of the century as Random Joe is calling for. It takes time out of our day and confuses the public."

Many of the meteorologists with whom we spoke echoed a similar sentiment: Find a source, compare it with others, and stick with it.

“Verify!” wrote Bernstein. “Follow someone for a bit and get to know them. Trust takes time.”

Read more:

How to cut through the nonsense in winter storm forecasts

Flake news: Please disregard snow total forecasts on your weather apps

Here’s why computer models 10 days into the future cannot be trusted

The great Facebook flake-out: A little, not a lot of snow for East Coast this weekend

Beware of faulty, flaky Facebook weather forecasts