Arctic Temperature Update

Back in June 2021, I wrote about a set of 32 longstanding climate sites that should allow for a reasonably comprehensive ongoing assessment of "ground truth" Arctic air temperatures over land; so it's time for an update.  One reason I think this is worthwhile is because global reanalysis models - a common and generally excellent source of information about climate variability -  tend to struggle with Arctic temperatures because of deficient representations of sea ice, snow, cloud, and radiation physics.  For example:

On the warm bias in atmospheric reanalysis induce by the missing snow over Arctic sea-ice

Evaluation of Six Atmospheric Reanalyses over Arctic Sea Ice from Winter to Early Summer

Of course there are also significant difficulties with long-term temperature trends at surface observing sites - there's no really "pristine" long-term surface data - but it is interesting to take a look at the month-to-month and year-to-year variations for sites with relatively long and stable histories.

With that in mind, here's what the GHCN data has to say about summer 2022 at 30 of my 32 sites (the other 2 had quality control flags this summer):

Kotzebue was the coolest of the lot, with the coolest summer since 2006.  This is actually the third consecutive summer with an average temperature below the 1991-2020 normal in Kotzebue, but this summer's temperature would have been normal about 60 years ago.

More broadly for these Arctic land sites, summer 2022 was a bit warmer than 2021 but cooler than 2020.

 

Below are the monthly temperature anomalies from May-September this year, expressed in terms of standard deviations.

Two of the monthly site anomalies stand out as being particularly unusual.  Barentsburg on Svalbard had its warmest June on record by some margin, and it was also - easily - the most anomalously warm month on record in terms of standard deviations (compared to the 1991-2020 climate).

The same was true of Hall Beach (locally Sanirajak) in Nunavut in July: warmest July on record, and most anomalously warm month on record.  But we need to put an asterisk by this one, because an alternative site at the same location, which does not report as reliably, was not quite as warm: only 2.4 standard deviations above normal.  The previously more reliable site that I used for the map above seems to have been running warm since May - an example of a problem with surface station data.

Looking at a longer-term view of monthly anomalies since 2010, the 32-station mean seems to have come down slightly in 2021 and 2022 compared to the post-super-El Niño warmth in 2016-2020, and that may be because of the persistent La Niña in the last two years.  There's been an absence of very large positive temperature anomalies since winter 2020-21, but that's mostly because the Arctic was relatively cool last winter, at the time of year when anomalies are largest in amplitude.

When the anomalies are standardized by monthly variance, the 32-station maximum jumps up dramatically because of the Barentsburg and Hall Beach anomalies this summer.  Also, notice that while the 32-site average may have avoided jumping to new highs, there has also been an absence of significant cold this year: none of the sites has been below -1.5 standard deviations so far this year.

And check out last March: warmth was remarkably widespread, with the coldest site being only 0.2 standard deviations below normal.  This itself is a record for "lack of cold" at any of the sites.  The history of this data does not yet have a month with all 32 sites above the 1991-2020 normal, but it will probably happen sooner or later.