Weather model fantasy-land: How to handle far off “threats” of unlikely snow?

Here we go. We can’t even get through the first week of November, and there is social media chatter about possible snow 9-10 days into the future.

I’ve said it many times, but to repeat: forecasts for storms 5-7 days or more into the future have very little reliability. Most model simulations for snow in this range will not materialize. Often, even forecasting winter storms within 60 hours is a major challenge.

Nevertheless, you will inevitability see Facebook and Twitter postings about possible snow events in this range from both amateur weather enthusiasts and professional meteorologists – especially if advertised by the revered European model.

Related: Beware of faulty, flaky Facebook weather forecasts

In today’s case, meteorologist Henry Margusity at AccuWeather and Justin Berk, a Baltimore meteorologist with an active Facebook following, discuss snow simulated 9 days from now by the European model.

European model showing accumulated snowfall on the evening of November 14 (WeatherBell.com)

European model showing accumulated snowfall on the evening of November 14 (WeatherBell.com)

It’s debatable whether model simulations at this range are even worth mentioning, but Margusity and Berk don’t hold back. In Berk’s case, if a goal is generating discussion and drawing eyeballs, he succeeds. His discussion attracts loads of “likes” and “shares” tempered by some critical comments about overhype. In defense of both Margusity and Berk, their discussions are appropriately qualified.

Margusity writes: “so there [are] a lot of indications that the ECMWF [European model] run is just wrong. However, sometimes the ECMWF hits a home run and it’s possible the model is actually on to something. The implications of the ECMWF run would be snow across the mid-Atlantic for that time period.”

Berk writes: “The often reliable European Computer Model is trying to generate a cold storm by the end of next week. It might be a little overdone. First bet is that areas near the water and big cities tend to stay warmer… So for you this would be an even longer ‘long shot’.”

With other meteorologists discussing such far-off threats with very low likelihood, inevitably we end up getting questions about them. In the past, we’ve usually held off from discussing “virtual storms” more than a week or so into the future. But, if readers really want our take on them, I’m open to hearing that argument.

To be honest, I’m not in love with the idea of discussing models in 7+ day time frame, because this is a slippery slope. If we discuss too many “model storms” at long ranges, it may lead to too many such posts, a perception that we overhype potential snow (even if we generally downplay) and people then tuning out legitimate threats. What I’d rather do, and what we’ve done in the past, is discuss the pattern very generally for the period beyond one week, and indicate whether odds for wintry weather are above or below normal.

Let us know what you think about communicating far-off, very low likelihood threats. Do you want to hear about them? Do you have ideas about how we should frame such discussions? Thanks in advance…

PS The latest European model run, after showing some snow last night, now shows a mild rain 9 days from now – which makes a lot more sense given the pattern and time of year.

Also on Capital Weather Gang

Fall foliage: top 20 pictures in Washington, D.C. region from Weather Gang readers