Wednesday, October 09, 2013

Do October Temperatures Matter For Winter?

October temperatures have often been a link that many look to for a determination of what winter weather may be like. Warm Octobers typically yield warm winters, with cool Octobers yielding cold winters. In some years, such as 2007, our warmest October on record yielded a rather snow and snowless winter.

How strong is that correlation?

I took each of the 73 years that data has been measured at Philadelphia International and broke them down by coolest third, middle third, and warmest third for the entire month to get a sense of the numbers.

For inclusion in the warmest third, October temperatures needed to be at or above 58.4 degrees. For the coolest, at or below 55.8. The raw average of the last 73 years is 57.3 degrees but the distribution is roughly equal between the three camps.

If you're curious, the 1981-2010 average is 57.5 -- there is very little difference between it and the whole of the sample as October's climate has been less impacted by general warming overall.

In years where the average temperature for October ended up 58.4 or above, winters tended to be warmer than average and slightly less snowier than average and consequently, the cooler Octobers tended to correlate to cooler winters overall with a bit more snowfall than average.  This correlation works at about a 70 to 75 percent clip, meaning that more often than not October temperatures will foretell the winter to come.  In October is warm, 70 percent of the following winters were warmer than the long term average. If October is cool, 74 percent of those winters were cooler than the long term average.

Now, there are some notable exceptions to this.  For instance, October 2010 was warmer than average (59.0 degrees) but ended up colder and snowier than average. You can thank massive arctic and Atlantic blocking for the cold and snow.  October 1995 was also in this camp (warm October, cold winter).

Likewise, a cold October like 1972 yielded a warm, snowless winter thanks to El Nino. October 1952 is another exception as well (cold October, mild winter).  That said, a 70 percent correlation is a pretty solid batting average to have on one's side when going into winter forecasting...and it's a big reason why we hold out on issuing a winter forecast until late in the month.

Now, what about those mushy middle October -- the 25 or so that fall in between? Compared to the long term temperature average of 34.3 and snowfall of 21.8", temperatures and snowfall for the following winter after a neutral October are pretty close to average. There are some wide arrays in snow and temperature in this data set -- some of the warmest...and snowiest...winters on record fall in this camp. Among them, 1997-98 (warm), 2009-10 (snowy), 2011-12 (warm) were all in between 55.81 and 58.39 for October temperatures. The middle is the most volatile arguably in data sample and, not surprisingly, the closest in distribution as 13 of the 25 Octobers were below the long-term average for winter and 12 were above.

The coming three weeks will be critical in shaping the final direction of October's temperatures, although the very warm start we had will very likely bump us towards a warm finish for the month.  It isn't the only factor that shapes winter around here but the correlation is strong enough that one does need to take a look at it when shaping a seasonal outlook.