Tag Archives: numerical weather prediction

Believe it or not, some snowstorm forecasts on Facebook are bogus!

Snowfall forecast map from the European numerical weather prediction model run on Dec. 16, 2013 for the forecast period Dec. 22-23. What actually happened Dec. 22 and 23 wasn't even close to this!
Widely shared snowfall forecast map from the European numerical weather prediction model run on Dec. 16, 2013 for the forecast period Dec. 22-23. What actually happened Dec. 22 and 23 wasn’t even close to this!

Winter weather is just around the corner in Indiana, which means so are authentic-looking but bogus long-range snowstorm forecasts on social media.

It won’t be long before we see claims that a storm a week or more away will bring huge snow accumulations. Many will have official-looking forecast maps, like the one above (which turned out to be wrong, by the way).

But these posts won’t be the work of professional meteorologists. Many will be the creations of school kids, passing themselves off as weather experts.

This is Winter Weather Preparedness Week in Indiana, so it seems like a good time to prepare readers for the ominous-looking but unreliable snow forecasts they’ll soon see.

To understand what amateur weather enthusiasts put on social media, it helps to know something about the computer programs that professional meteorologists use to guide their forecasts. These programs are called numerical weather prediction models. They simulate Earth’s atmosphere by describing it in a complex series of very complicated mathematical formulas.

The programs built on these formulas run several times a day on supercomputers around the world. Much of the output of these programs is available on the Web, in both numeric and graphical form.

The output of computerized atmosphere models is inherently inaccurate for several reasons, including:

  1. It’s not yet possible to completely describe our chaotic atmosphere in mathematical equations and
  2. The programs don’t have access to enough data about what our atmosphere is doing at the time they run (e.g. what the temperature, wind speed and wind direction are 10,000 feet over any given part of the planet).

Nonetheless, these programs kick out predictions of what the weather might be at any location at any time, as far in the future as 16 days, despite that fact that no computer or human can reliably forecast the weather that far in advance.

Now, imagine a young weather enthusiast who craves attention and loves snowstorms (because they get him out of school). When he sees an indication of heavy accumulations in the output of a single computer model, he might paste that model’s map into a Facebook post in which he writes a dire forecast of impending doom. Such an amateur forecaster might not be aware of (or care about) the model limitations described above. But she’ll love all the “likes” and shares her post receives!

So how do I know what to believe? First, I’m automatically suspicious of any social media post that forecasts specific snowfall amounts more than a couple days in advance. Second, I ignore any forecast that doesn’t come directly from professional sources I trust, such as:

  • The National Weather Service.
  • Local, degreed broadcast meteorologists.
  • Certain commercial weather forecasting companies.

If that ominous snowstorm forecast didn’t come from one of the above, I won’t share it on social media. I hope you’ll join me in that practice.