Tuesday, July 4, 2023

Daily Temperature Records

Interior, south-central, and southwestern mainland Alaska have had a relatively cool year so far overall; by "relatively" I mean in comparison to the previous decade.

One way to look at this is to consider the number of daily temperature records that have been set in recent months: for example, Anchorage hasn't set or tied any daily high temperature records so far this year, compared to 12 instances in 2022 and at least 4 instances in each year from 2013-2022.  Bethel set one high temperature record on March 6, compared to 6 records last year.

Of course, low temperature records are a rare breed these days, so there haven't been many of those either (Nome's record cold in April notwithstanding).

I thought it would be interesting to use ERA5 gridded data to look at daily records in a continuous spatial framework.  I don't have as much confidence in the ERA5 temperature data prior to 1979 (the start of the modern satellite era), so I calculated calendar-day records for the 1979-2022 period of record.  Here are maps of the number of days this year that have set new records compared to that 44-year historical period.

The absence of new high temperature records is quite notable for central and southern mainland Alaska, but there have been several record-warm days along the Arctic coast and in the Aleutians.  As for (post-1978) record cold, the epicenter has been in northwestern Alaska to the south of the Brooks Range.  Ground-truth data from Kotzebue fully confirms this, with 4 all-time daily cold records broken in April (April 7 and 10-12), and 10 days with post-1978 cold records.

We can create an objective measure of the "number" of records each day by calculating the grid area that sets a record, and then we can take the average for any window of time.  On an annual basis, this is the result:


On the warm side, 2019 was the "winner" for daily warm records, which makes sense as it was the warmest year of record statewide for Alaska (1925-present) - the only year on record with an annual mean temperature above freezing.

It's no real surprise to see that 1992 had the most daily cold records, although it wasn't the coldest year since 1979 - both 1999 and 2012 were colder on the basis of the annual statewide mean.  But 1992 saw astonishing cold relative to normal in September, and May was also very cold.

The increase in coverage of warm records is quite dramatic, as we'd expect, but the decrease in cold records is even more significant.  Notice that the last couple of years have been close to "normal" for coverage of warm records, but cold records haven't reached anywhere near normal.  This is consistent with the well-known fact that daily minimum temperatures are generally warming faster than daily maxima around the world.

How about the summer season?  Here we see that 2004 really stands out, and indeed it was the warmest summer on record statewide (1925-present), and also the worst fire season on record.


Here's a map showing the locations (blue shading) where 2004 was the warmest summer on record, based on ERA5 data since 1979.  But 2019 was the warmest summer in south-central and southwestern Alaska, and some other years show up for Arctic Alaska.

If we look at the area where 2004 had the most daily record highs, there's a good correspondence with the mean temperature map overall, as we'd expect.

But the fickle nature of daily records means that there's also more noise - for example, Fairbanks shows up as having the most daily record highs in 1994 (see the tiny yellow patch in the central interior above).  Somewhat remarkably, this is correct: Fairbanks did have its most (broken or tied) daily record highs in summer 1994, whether you look at 1979-2022 or the entire period of record.

Here's the number of daily record highs for five major sites based on the 1979-2022 period, using ground-truth station data (click to enlarge):


For Anchorage, 2019 stands out by far (it was by far the warmest summer on record), but 2004 comes out on top for Juneau, Nome, and Utqiaġvik.  The latter is interesting, because summer 1989 was much warmer than 2004 in Utqiaġvik, but it produced hardly any record highs.

The overall consistency between ERA5 and ground-truth data is quite encouraging.  Going forward, it will be nice to have this new tool for analyzing the "daily records" aspect of climate variability on a complete spatial basis.

No comments:

Post a Comment