How accurately can anyone predict snowfall in December — or worse, five months out, in March?
AccuWeather just issued its forecast for the 2014-2015 winter season. AccuWeather is calling for my hometown, New York, to “close this season with snow totals just above normal.”
Now just to be clear, I have no ill will against the folks at AccuWeather. They were very friendly when I visited as a member of Penn State Weather Camp a decade ago. But I’m skeptical a snowfall prediction made so far ahead will be consistently accurate.
I have seen snowfall forecasts in New York City miss by 20 inches — and that was when they were predicting just two and a half days out.
Snowfall forecasting is hard. To get snow, you need both cold air and precipitation at the same time. You can have a colder and wetter than normal winter and still have below average snowfall. You can also have an extremely warm winter and get cold air to meet precipitation at exactly the right time to produce snow. For example, the 2005-2006 winter in New York was much warmer than normal, but the mix of the elements in early February produced the biggest snowfall on record in New York.
To figure out how accurate Accuweather has been in its long-range forecasts, I looked back at its winter forecasts over the past 10 years. This isn’t a full rundown of every city or region in the nation, though the links in this article will allow you to do your own analysis.
AccuWeather’s daily forecasts are near the top of the accuracy rankings for New York City. Yet, its long-range snowfall forecasts for New York City left much to be desired.
- 2004-2005: It went with average or 25 percent above average snowfall. New York received 41 inches — 87 percent above the 30-year average (from 1971 to 2000) of 22 inches. That’s a 62 percentage point error.
- 2005-2006: AccuWeather predicted snowfall 25 percent or more above average. New York received 40 inches — 83 percent above average. Right on.
- 2006-2007: Again AccuWeather aimed high, with a forecast of 25 to 50 percent above average snowfall. New York got only 12.4 inches of snow, 57 percent of the 30-year average. That’s not even close.
- 2007-2008: This was a good call. AccuWeather predicted below average snowfall, and New York got below average snowfall (11.9 inches).
- 2008-2009: Two in a row. They went with above the prior year’s 11.9 inches of snow, but not “gangbusters” along I-95 (including New York City). New York got 27.6 inches.
- 2009-2010: AccuWeather is on a roll now, calling for a “cold and snowy” winter. New York City got 51.4 inches of snow.
- 2010-2011: AccuWeather’s streak came to an end here. It predicted an average snowfall season, but 61.9 inches fell. That’s not even in the same ballpark of the 1981-2010 average of 25.3 inches. It’s 145 percent above average.
- 2011-2012: So how did AccuWeather react to being so far off on the previous year’s forecast? It decided to get really specific, predicting 33 inches of snow. Only 7.4 inches accumulated in New York City. Oops.
- 2012-2013: It went broad again and called for a “snow joke” this season with above average snowfall. The joke, however, was on us. The seasonal snowfall (26.1 inches) was within 1 inch of average.
- 2013-2014: Once again, AccuWeather got specific and said “New York City… should not expect to beat last year’s totals.” They argued that New York City was right on the border of average and below average snowfall. Almost 5 feet of snow fell (57.4 inches), beating the prior year by 31.3 inches.
So how did AccuWeather do in our accuracy tally? Four of 10 forecasts for seasonal snowfall were clearly accurate, six were not. One could give AccuWeather credit for calling for above snowfall in 2004-2005, which would make it five for 10. But the past four years been have particularly bad for the folks from State College.
So, AccuWeather doesn’t have some secret recipe to forecast snowfall, but neither does anyone else. The International Research Institute for Climate and Society at Columbia University has been providing accuracy measures for its long-range forecasts since 1999. Its forecasts for January, February and March for New York City this far in advance for both precipitation and temperature have indicated “no skill” (i.e. it’s random).
Weather is complicated, and we’re not going to know how much snow we’ll get this winter until we get it.