Cooling Off

The last two nights have brought hard freezes to many interior and northern locations, and temperatures below 20°F were widespread this morning across the northwestern interior and Brooks Range.  Arctic Village reached 15°F with a bit of snow on the ground; the Norutak Lake RAWS and Huslia CRN sites saw 16°F; and yesterday Tanana dropped to 18°F.  It is slightly early for cold like this, but not dramatically so.  With a substantial trough entrenched over Alaska, we would expect nothing less.

A bigger surprise is that the Fairbanks airport hasn't reached the freezing point yet.  Last year set a new record for latest first freeze (September 28), and only one other year (1974) went past September 25 without a freeze.  According to the NWS it may happen on Saturday or Sunday, but if not the record will be broken again.

Visible satellite imagery from today shows extensive snow cover in the central and eastern Brooks Range, but apparently not across most of the North Slope - and notice the complete lack of sea ice to the north (click to enlarge).  Of course we're only a couple of weeks past the annual minimum for Arctic sea ice extent.

The next few weeks have the most rapid rate of seasonal temperature decline for interior Alaska, and one way of thinking about this is to consider the probability that any one week will be colder than the previous one.  In Fairbanks this probability gets as high as 80% in the first half of October: there's no going back at this point.

If we compare this probability curve to the annual seasonal cycle of "normal" temperature, there's one curious feature (at least to data nerds like me): the aforementioned 80% probability occurs around October 10-15, but the decline in normal temperature is greatest about 10-15 days later.  The reverse occurs in the spring: the probability of one week being warmer than the last peaks in late April, but the normal temperature rises most quickly around April 10-15.  (Note that I'm using the entire history of Fairbanks data here, not just recent decades.)

I think what's happening here is that the variance of temperature is much higher in winter, and that tends to reduce the probability of "predictable" week-to-week changes as the normal changes in late autumn or early spring.  Imagine if the temperature variance were extremely low from day to day, so that temperatures simply followed the normal curve; then the sign of the weekly changes would be highly predictable.  But if variance were extremely high, the changes in the normal would be rather insignificant compared to the variance, and the probability would be much closer to 50%.

So in terms of the predictable, relentless grind lower in temperature, the next few weeks are as bad as it gets.  There's still a long way to go after that, of course, but with more ups and downs along the way.