Monday, October 01, 2012

Why Posting Computer Model Snowfall Maps Is Not A Good Idea

This is how bad rumors get started.

Yes, the GFS is showing a cold snap for the weekend and the first half of next week.  The Euro is as well.  The GFS shows a weak impulse riding along the frontal zone that pushes through this weekend, bringing a shot of showers to the region on Monday night of next week.  The Euro does not.

Because the shot of showers falls in a cool airmass -- the GFS is suggesting in some of its algorithms that it may snow along I-95 despite raw data that shows temperatures in the 40's in the city...and raw GFS data is typically colder than reality.  Some GFS maps from various providers are showing snow -- others are not.  It is the same model and if it's varying that much within different map providers, especially with an unfavorable lower atmosphere, it's going to be tough to snow unless precipitation is going bonkers.

Raw data from the GFS as of earlier today says "no snow" for Philly or outlying areas next Monday.

While one algorithm suggests possible snow, other algorithms on precipitation show a chilled rain...if the GFS' forecast is right.

It's fair for forecasters to state that it will be a good bit colder at this time next week.  Measurable snow at this time of the year is rather rare though -- we've been in a bit of a run of luck as it has snowed locally in two of the last four Octobers (2008 and 2011).  Before then, October snow is a rather rare treat around these parts.  It requires a fresh, strong source of cool air to pump in and a strong enough storm system tracking nearby to help enforce the cool regime.  I'm not sure a wave of energy travelling along a cold front is enough strength to help that happen...especially since the Euro's depiction at this point shows little support for the storm as any "wave" is a few hundred miles offshore.

Maps like the one linked at the top of the article often lead to viral rumors suggesting "it will" snow next week when nobody is really suggesting it...and the computer model based for those forecasts can't event agree if it will snow or not.  It's important to note that there's more to a forecast than just the pretty maps -- if the lower atmosphere is too mild and if temperatures aloft are not supportive of snow, it's not likely to snow.

Of course, we first would need to even see the GFS and Euro to come to agreement on it even precipitating overhead next Monday.  But what's 300 plus miles amongst rival computer models?