31 January 2014

The Great Facebook Blizzard- Storms and Rumors of Storms

Posted by Dan Satterfield

This image posted on a facebook page went viral this week and caused thousands to think a blizzard was coming.

This image posted on a Facebook page went viral this week and caused thousands to think a blizzard was coming.

Another piece of model output was posted online by an Alabama newspaper reporter.

Another piece of model output was posted online by an Alabama newspaper reporter.

It apparently got started with two images. One from the Facebook page of “Weatherboy weather” and another that got published in an online newspaper article in Russellville, Arkansas. It was also posted on AL.COM (the online portal of several major Alabama newspapers). The result was a crush of emails and messages to meteorologists at TV stations and NWS offices across the eastern U.S.! The actual origination may be different on the Alabama, case because the image also showed up on the blogs of meteorologists in North Alabama as well.

In some cases there was a graphic warning that this was not a forecast, and was extremely unlikely to happen, but by the time it got shared a thousand times online, that was lost (and I suspect it was never there to begin with in some cases). It resulted in lengthy special weather statements from an NWS office trying to bring some facts to the wild rumors, but it will probably take some days before this begins to fade away. Most meteorologists I know (including me) have posted notes on social media about this not being real (and have given up answering anymore questions on it).

Is Posting Raw Model Data A Mistake?

This episode highlights the issue of posting raw model data online. It’s an issue that has been under discussion for a while now among meteorologists at broadcast outlets. The concern is that this can be shared without context, and become a forecast when it’s not. I’ve been very careful about it for a while now, and only for short-range situations when I think the model is very likely correct (and it matches my forecast). After this episode, I suspect a lot more forecasters will think twice in the future before posting raw data as well. We also have quite a few on air forecasters who do not have a significant background in atmospheric physics and tend to be “model forecasters” rather than meteorologists.

One forecaster expressed his frustration at the whole episode this way...

One forecaster expressed his frustration at the whole episode this way…

Unfortunately, this will not solve the problem, because some of these viral predictions were posted by younger students and hobbyists (who look at the model data online), and have learned to read it to some extent. I hesitate to criticize this because, I think it’s great that we have students as young as high school who are learning the value of what one can do when you combine supercomputers with math and the laws of physics. The problem here is that a few of them have learned that wild forecasts will get you thousands of followers. Add in some confirmation bias, where folks remember only the few times they were correct and forget the big errors, and you are all set. This has worked very well for the Farmers Almanac (and astrologers) for over 200 years. (The Almanac forecast of a blizzard at this Sunday’s Super Bowl may be their undoing though!)

My friend and fellow forecaster John Trout had this to say, and I agree with it completely:

“With the proliferation of the World Wide Web and one’s access to data previously reserved for knowledgeable sources, suddenly everyone is releasing anything they deem worthy on all sorts of topics. Imagine today’s modern Doctor and how many times a day they must hear, “That’s not what I read on the internet”. What the lay person is left with are the shocking details without the “background knowledge” and an inability to decipher the data without an informed opinion.

The truth is no one knows if this graphic will present the weather it suggests. Modern day forecast accuracy beyond 7 Days falls off dramatically.

Science Literacy- Or The Lack Thereof

The poor science literacy of most Americans is a significant contributor to this online blizzard. The public should know that weather forecasts beyond 5-7 days are not at all reliable, but most do not. Private firms like Accu-Weather have made it even worse by putting out forecasts for a month in advance, and continue to do so after a study showed that the forecasts were WORSE than using the climatological averages. Some TV stations regularly air a ten-day forecast as well, (you should write to the management and suggest they stop it) when the best forecasters can usually do is forecast warmer or colder than normal and even then with only limited skill that varies day-to-day and season to season. Try explaining this to someone who believes there is something to those astrological horoscopes they read each day in the newspaper.

A Basic Guide To Any Forecast You See Online.

1. Is it from the NWS or a local broadcast meteorologist you trust? If not, disregard.

2. Is the forecast calling for snow or a specific storm to hit beyond 3-5 days from now. If so, disregard and if it is a broadcast meteorologist who is doing it, change the channel. No professional forecaster with a real science background would likely do this.

3. Did you get the forecast from a friend or did you see it on the NWS or a TV Station website/ smart phone App.?? If it was from a friend, and you cannot verify where it came from, ignore it and for my sake, please do not share it!

4. Disregard ANY forecast that is making a determinate forecast of  snow/tropical storms or specific temperatures beyond 7 days. No exceptions.

(Correction- an earlier version had Russellville, AL instead of Arkansas. Sorry for the error.)