The Washington PostDemocracy Dies in Darkness

Don’t trust an internet snowstorm forecast more than a week into the future

Lee Anderson adds to the pile of snow beside the sidewalk in front of his house in Somerville, Mass., Tuesday, Feb. 10, 2015. (AP Photo/Josh Reynolds)

It’s the time of year again when rumors about snowstorms 10 or even 15 to 20 days into the future are spreading like wildfire. Here’s a simple message: Don’t trust them.

I’m going to try to make it easy for everyone.  If you see a forecast on Facebook (or Twitter or any online venue) for a specific storm targeting a specific area more than seven days into the future, disregard it and don’t share it.  Renowned Birmingham, Ala. meteorologist James Spann advises “Think before you share.”

I’ll be less delicate:

If it’s beyond a week, it reeks.

Repeat that.

If it’s beyond a week, it reeks.

[Beware of faulty, flaky Facebook weather forecasts]

Forecast models, which predict specific snow amounts, project out to 16 days in the future. But, in most cases, can only provide a realistic sense of the storm timing and areas at risk 5-7 days into future, and pin down the details 24-72 hours ahead of time.

Snowfall forecasts more than a week into the future (and sometimes less) are simply not reliable.  No person and no model has proven otherwise.

The best forecasters can do in the period one to two weeks into the future is identify whether the general pattern is more or less conducive to cold weather and storminess than normal.  But they cannot credibly predict where a particular storm will develop, track, and the amount and type of precipitation it will produce.  Weather models try to do this, but they’re more often wrong than right.

Yet weather websites exist that make these unreliable forecasts freely available.  And so anyone, irrespective of their meteorological credentials or experience, can post them.  The Capital Weather Gang and others have seen and documented cases in which novices, amateurs and pros have posted maps showing the “threat” of a monster snowstorm 10 days or more ahead of time.  These threats have not materialized.

[The great Facebook flake-out]

The posts incite excitement and panic.  Often, they are shared hundreds if not thousands of times, providing instant gratification for the individuals posting them.

But when these forecasts don’t pan out – usually the case – it harms the credibility of the community of more responsible forecasters.  And, as James Spann writes, can result in bad decisions.

Every winter, responsible meteorologists – who don’t issue these kinds of faulty long-range forecasts – struggle with how to deal with this problem.  Do they ignore these forecasts even when getting peppered with questions about them?  Do they shoot them down? Do they shame their originators?

My take is that it’s unproductive to call out the bad actors.  When this same problem came up with hype-filled hurricane forecasts in 2014, I wrote: “It’s a never-ending and unwinnable game of whack-a-mole.”

[Hurricane hype is here to stay; Forecasters must adapt]

Some wonder if more prominently displaying professional “seals” earned by meteorologists schooled in responsible forecasting will solve the problem. But seals are ultimately just labels which many people disregard.

The best way for forecasters to confront this issue is to continually educate the public on the limitations of weather forecasts and what information to trust.  And to keep in mind that trust has to be earned by establishing a track record of accurate forecasting and sustained engagement with their audience.

Meanwhile, it will help if consumers evaluate the credibility of sources they review.  Ask questions like: ‘Have I heard of this source before? Is the information on this page regularly updated? Is the information filled with hype and click-bait? Is it professionally presented?’

And remember: “Think before you share” and, of course, “If it’s beyond a week, it reeks.”