Friday, May 13, 2016

Possible End to Warmth

While making a routine seasonal forecast the other day I stumbled on something surprising and, I think, quite interesting.  Here's what I came across: the map below shows the temperature pattern during summers (June through August) following a strongly positive PDO phase in late winter (February through April).

This year, as we all know, the PDO phase has been extremely positive so far; the February-April average PDO index was second only to 1941.  With above-normal temperatures continuing in apparently unending fashion across Alaska this spring, it is interesting to see that summer tends to be cooler than normal when the PDO is strongly positive in late winter.  Note that this is not because the PDO suddenly turns negative in these years; out of the 10 years shown above, the June-August PDO remained strongly positive in 4 of 10 years.  Two more of the years had a significantly positive summer PDO, and it was close to neutral on average in 3 years and became negative in only 1 year.

The map above is derived from the NCEP/NCAR reanalysis, which as we've seen before is not very good at all for Fairbanks temperatures in summer, so I checked the result with station data from Fairbanks.  Here's what the analog years looked like in Fairbanks; I've plotted the monthly temperature anomaly normalized by the standard deviation.  The black line shows the median of the 10 years for each month.

As expected the late winter and spring tend to be warmer than normal when the PDO is strongly positive, but the warm anomaly goes away by June, and each month from June through September is more likely to be cooler than normal.  Remarkably, August temperatures were nearly 1 standard deviation below normal on average in the 10 analog years.  It's fascinating also to see the warmth return in October and November.

Here's the corresponding chart for precipitation, expressed as a percentage of the 1981-2010 median for each month.  Dry conditions are typical in January through March in these positive PDO years, but the normally dry month of April is often relatively moist - as happened this year.  There is not much signal for precipitation in the rest of the year; May, June, and September show a tendency to be dry, but July and August show no marked departure from normal.

We can get a sense of the progression of the temperature signal by looking at month-by-month patterns from the reanalysis for the same analog years:

The expansion of the cool anomaly in August and September is pretty remarkable.  The maps below show the sequence of events from the perspective of 500mb height, with yellows showing a tendency for above-normal height (ridging or high pressure) and blues showing below-normal heights.  It appears that the interior's cool weather in June and July tends to be associated with high pressure to the north and cool northeasterly flow.  However, in August there is a strong tendency for a trough over the Beaufort Sea, which would bring a cool airmass to interior Alaska from the west or northwest; and in September there is often troughing over southeast Alaska, which is certainly a cool pattern.

Despite the lack of an obvious signal in the Fairbanks precipitation data, a broader look at precipitation in the analog years suggests that July is more likely to be wet than dry.

The other summer months don't show a notable precipitation signal, but the reanalysis solar radiation data suggests that relatively sunny conditions are favored in August and especially September (see below).  So if we look all the way ahead to early autumn, the PDO analog suggests that relatively chilly and clear conditions may be likely.  It will certainly be interesting to see if 2016 matches the historical pattern.

Update: for comparison, here's NOAA's forecast for June-August, issued last month.  NOAA will update their forecast on Thursday, but I doubt it will change much.


  1. The dip in temps during summer is totally unexpected. It will be interesting in how it plays out. Perhaps I should get some outside work done now before it gets cold.

    Are the differences for each PDO modulated by the sign of ENSO? Perhaps you can get a better prediction if we assume that a strong La Nina is approaching.

    1. Eric, I like the idea of looking at the influence of ENSO too - the flip to La Nina might well have something to do with this. I'll take a look.