Getting Chilly

As we move into the last week of September, snow levels are dropping on the mountains, and autumn colors are fading along with the daylight.  But freezing temperatures have been unusually absent and/or mild so far for much of the interior: Fairbanks airport hasn't even dropped below 35°F yet, and Bettles just had its first sub-32°F reading yesterday morning (along with a trace of snow).  This lack of cold is almost unprecedented for the time of year.

The most rapid pace of temperature declines (based on the climatological normal) occurs in October for most of the valley-level interior, as snow cover usually becomes established during the month; this greatly aids the development of persistent temperature inversions, and shuts off the warming effect of the waning sun.  Early-mid October is also the time of most rapid cooling for coastal southern Alaska, where snow cover isn't yet common at low elevation.

Here's a map of the date of "peak cooling" based on ERA5 1991-2020 data:

Note that this is based on smoothing the progression of daily normal values with a harmonic function; the choice of method will affect the result to some extent.

The sharp gradient over the Bering Sea is interesting.  It seems that where ice cover is common by mid-winter, the maximum cooling is delayed, because (I suppose) greater cooling tends to develop after at least some ice is in place.  But in more southerly regions, where ice cover is unusual, peak cooling occurs in mid-October, similar to the Gulf of Alaska.

Here's an equivalent map for the peak rate of spring warming:

The west-east contrast across the interior is more striking in spring, as warming picks up rapidly in the east while it's held back by the ice-covered waters near the West Coast.