Why Cold Now?

To the great surprise of many observers and prognosticators (and I fall into both camps), Alaska is having something of a cold winter, at least since mid-December.  There's still plenty of winter left, and it may yet turn around, but so far it's a very distinct and striking change from recent winters, which were of course much warmer than normal.

The chart below shows the daily temperature anomaly from Fairbanks since September 1.  Nothing "different" happened until mid-December, but since then the cold has proven to have considerable staying power.  For example, today is the 25th consecutive day with a daily high temperature below +5°F, and with more cold air moving in, another week seems likely to be added.  We have to look all the way back to 1975 to find a stretch of 30 days or more below +5°F; the record is 48 days ending January 27, 1943.

An obvious question that has been much on my mind in recent weeks is, "Why?"  Why is this winter turning out so differently from recent years?  To illustrate the change in another way, consider the 500mb height anomaly since December 1 (first map below) and the striking contrast with the average Dec-Feb anomaly in the previous six winters (second map below).  What could drive such a significant pattern change?  Is there an "external" influence elsewhere in the global climate system that could explain the difference?

One aspect to notice is that this winter's circulation pattern bears more than a slight resemblance to a typical La Niña winter pattern.  The map below show the Dec-Feb anomaly in winters with strong La Niña episodes, based on CPC's ONI index.  There's less ridging over the Bering Sea this winter than in the La Niña pattern, and less of western North America has been affected by a trough, but there's obviously some similarity with the La Niña signal.

What's odd, however, is that we do not have a La Niña SST pattern in the equatorial Pacific this winter; in fact it has been considerably warmer than normal in the central tropical Pacific, and the pattern has been strongly aligned with a typical "central Pacific El Niño", or "El Niño Modoki", event.  Until recent weeks there has also been an intensely positive Indian Ocean Dipole, which is usually found in association with El Niño; so we wouldn't expect to see a La Niña influence in the mid- to high latitudes.

Are there any other factors that could be driving the pattern in a cold direction for Alaska?  Frankly, it's difficult to find any, and even with the benefit of hindsight it is difficult to see any compelling reasons why the pattern would flip this winter.  As part of my regular work responsibilities I look at a lot of seasonal forecast guidance every month, and nearly all statistical and "pattern matching" tools suggested that another warm winter was on the cards for Alaska; for instance, the maps below show objective statistical forecasts based on October-November SST patterns in the tropics and Northern Hemisphere.  (These statistical forecasts are only marginally skillful, but they're better than random chance.)

The experts at CPC also thought a warm winter was more likely than not - see below.

Perhaps the only real hint that something might turn out different this time came from some of the dynamical models in the NMME guidance.  In particular, a few of the models showed a cold signal from southern Alaska to northwestern Canada; but others showed pronounced warmth in the same area, and it was easy to dismiss the cold solutions.  In hindsight, if the winter continues the way it's going, the best forecast may have been produced by the new GEM-NEMO model from Environment Canada: not only did it show pronounced cold in much of Alaska, but it captured the unusual warmth that has prevailed in the lower 48 and Europe.

In light of this, it seems that the GEM-NEMO model "caught on" to a driving influence that turned out to be significant, leading to a good forecast.  There's a chance that further investigation might reveal what that influence was and lend some insight for the future; or it might just be that the model was "lucky" this time.  But in any case, hopefully I'll be able to do a more productive post-mortem of the winter pattern when it's all over.