Thursday, August 29, 2013

Why Predicting Snow Chances Around The Super Bowl Is Not Rocket Science

God bless the Farmer's Almanac and its attempts for thinking ahead with yet another "cold and snow" outlook for this winter.

The Almanac, which issues forecasts every August for the coming winter, is predicting the second straight cold and snowy winter around these parts.

Last winter's outlook (with our 7.9" of snow in Philadelphia) did not pan out all too well.  Even with that obstacle in the way, the outlook doesn't look dramatically different from what we saw last year (see the graphic on the left).

Putting the similarities aside, what has the headline chasers all jumping up and down with a free headline and content to click is the prediction of a snowfall near the first ten days of February, especially with the Super Bowl being played in North Jersey this year.

Quoting them:  "Over the Northeast and Mid-Atlantic states, we are “red-flagging” the first ten days of February for possible heavy winter weather. More importantly, on February 2, Super Bowl XLVIII will be played at MetLife Stadium in New Jersey’s Meadowlands—the very first time a Super Bowl will be played outdoors in a typically cold weather environment. We are forecasting stormy weather for this, the biggest of sporting venues. But even if we are off by a day or two with the timing of copious wind, rain, and snow, we wish to stress that this particular part of the winter season will be particularly volatile and especially turbulent."

Using climatology alone, such a prediction is not overly unrealistic.  The first seven days of February, historically, are the most prone to snowfall in Philadelphia. February 1st-7th have seen 168 days of measurable snowfall since 1885. Playing straight odds, 18.75 percent of all days in this seven day period see at least a tenth of an inch of snow.

Kreskin's foresight and Nostradamus' quatrains aren't needed, nor are a secret blend of forecasting herbs and spices.

Snowfall frequency in Philadelphia peaks in early February, as it does in New York. As the season progresses, the odds of snowfall on a given day diminish.  By the last week of February, those odds drop to 10.78 percent, nearly half of what they were three weeks prior.

So, yeah, predicting a higher risk of snow in early February in the Mid Atlantic is not any more surprising than July heat. Predicting the specifics of "how much" is not different.

Using three inches as the criteria of "bigger" snow, the first seven days of February provides the best odds of a three inch snowfall in a given day. Three inches of snow have fallen on 31 days over this timeframe (many snowfalls work through a two day period but for the sake of this post, we're only looking at snows that fall in a given day so some snowfalls may not count as a three inch that are or larger snows may double count), odds of 3.5 percent. Not as great a chance, obviously, butt out of any seven day window it's your best chance. Coincidentally, the odds of any particular snow being a larger snow increase through February.  The first seven days of February provides an 18.5 percent chance (31/168) of any snowfall producing three inches, increasing to just over 20 percent (20.2) for February 8-14, then hanging at 20 percent over the last two weeks of the month.

Again, the odds of it actually snowing are low...but the odds of any one of those snows being a particularly decent snow is not terrible over the last three weeks of February.

If it does indeed snow in early February, the Almanac may try to pat themselves on the back for getting it right...but climatology and past performance show such a prediction is not incredibly daring.