Brink of Winter

Autumn is quickly drawing to a close for interior Alaska, with a severe freeze in the usual cold spots this morning, and significant snow predicted for later this weekend in and around the eastern Alaska Range.

This morning's chill was notable in the hills above Fairbanks, with widespread lows below 20°F.  Down at valley level, the Goldstream Creek observer measured 18°F, and the Salcha RAWS reached 10°F, the third morning at 11°F or below in recent days.  It's not particularly unusual, but the rate of the temperature decline never ceases to amaze (me) at this time of year.  [Update Saturday morning: down to 7°F this morning at Salcha.)

Looking at statewide data, it's typical to see a temperature below 10°F somewhere in the state before September ends, and sub-zero is not rare.  Two years ago the early cold was more severe.  Usually the coldest reports at this time of year come from locations in and around the Brooks Range, such as Ivotuk, Toolik, or Chandalar Lake.

Despite the chill, there's less cold air around in the Northern Hemisphere than in previous years.  Here's a chart I made today to illustrate the daily evolution of the areal extent of air below -5°C at the 850mb level (roughly 4500' above sea level):

A threshold of -5°C at 850 mb has been suggested as a good way to delineate the lower tropospheric cold pool that expands each winter and virtually disappears in summer.  The figure above shows that the cold pool, so defined, typically extends across 25-30% of the hemisphere in January.  This year the cold pool's growth is lagging the 1991-2020 normal significantly, and in fact this month's cold pool area is the lowest on record so far (for the month to date).

Here's a map to illustrate the cold pool's extent in the most recent day of data from ERA5: 

Based on data since 1950, the year with the most expansive cold pool in September was - surprise, surprise - 1992, when Alaska saw its record-setting early cold snap.  Here's what the 850 mb temperature map looked like on the same date in 1992:

Clearly it wasn't just Alaska that was affected by unusual early cold in the high latitudes that year.

Finally, it's worth noting that despite the sharp freezes, there is still a lot of color on the trees in and around Fairbanks.  I find this quite remarkable at the end of September.  Nenana and Cleary Summit webcam views from the last day or so:

Here's an animation of the foliage changes at Nenana:

Getting Frosty

Widespread overnight freezing temperatures arrived across Fairbanks and much of the interior late last week.  This is a "top 10 latest" first freeze in the official Fairbanks climate record; the 1991-2020 median date for first 32°F or lower is September 11.

Fairbanks climate data since 1930 shows the expansion of the growing season at both ends:

The trend is somewhat greater for the first freeze of autumn (1.8 days later per decade) than for the last freeze of spring (1.2 days earlier per decade).  Combined, the growing season (based on this 32°F metric) has lengthened by 3.0 days per decade, or nearly a full month since 1930.  The last time the growing season was 100 days or less was in 1996, but such a short growing season used to be fairly common.

Localized urban warming does not explain the trend at the official climate site (the airport since 1951, Fort Wainwright in the 1940s, downtown previously), as the data from the university farm paint a similar picture - see below.  The farm is at the far western end of campus, in a relatively open and undeveloped location, and has always been far more prone to early and late freezes.

Based on the 32°F threshold, the trends from 1930-present are actually greater at the farm, with the growing season lengthening by 4.2 days per decade.  The official climate site has never reported a freeze in July, but it occurred in some of the earlier years at the farm.

For the full history since 1905 at the farm, the trends are 1.5 days per decade later for first freeze, and 1.6 days per decade earlier for last freeze.  Note that I used July 15 as the cut-off between "late" versus "early" freeze.

The annual total of growing degree days (accumulated "heat units" above 50°F) has risen tremendously in Fairbanks, with the linear trend amounting to 5% per decade, or about a 60% increase from the 1930 baseline (change from ~750 to ~1200).  The chart below shows that a very noticeable jump occurred in the late 1960s, and it's also striking to see that sub-1000 GDD seasons have all but disappeared in recent years.

Subsea Cable Fixed

Good news this week for many residents of northern and western Alaska: the Arctic undersea fiber optic cable that was damaged back in June has been repaired, allowing internet and phone service to get back to normal.  This is a story I've been following, because the cable was apparently severed by ice scouring on the sea bed about 35 miles north of Oliktok Point, just up the coast from Prudhoe Bay.  Read more here:

https://alaskapublic.org/2023/09/19/northern-alaska-cable-break-repaired-after-14-weeks-of-internet-outages/

https://alaskapublic.org/2023/06/12/cut-cable-causes-weeks-long-north-slope-northwest-alaska-internet-and-cellphone-outages/

The aspect I find interesting here is that the break occurred in water "about 90 feet deep", and it seems remarkable to me that Arctic sea ice could ever reach the ocean floor at that depth.  Frankly I was skeptical to begin with, thinking that either the depth was misquoted, or that the cable was damaged for some other reason.  But Quintillion's webpage confirms "ROV inspection shows 20 meters of accessible cable on the seabed floor. The target is 50-60 meters of accessible cable in order to reach the surface with workable slack for splicing operations. The water depth in the area is ~25 Meters."

The second article linked above cites Quintillion's president with the following explanation:

“It’s in an area where the land-fast ice and the sea ice kind of meet,” he said. “And so ice scour – which is essentially moving ice hitting the land ice and then (sinking) kind of like a tectonic plate – I suppose that it scarred the seabed floor, and then went down below the depth where we were buried.

(end quote)

It seems that a large wind stress event would be needed to drive sea ice down to such a great depth, so it's worth looking at the weather data.  As it happens, there is some evidence of high winds on June 11, the date the damage occurred.  The following chart shows hourly 10m wind speed from the ERA5 reanalysis at 71°N, 150°W, which is very close to the stated location of the event:

According to the ERA5 data, the wind speed peaked on the afternoon and early evening of June 10, from a direction just slightly north of easterly.  The peak sustained wind speed is not particularly high - less than 30mph - but the ERA5 wind speeds can't necessarily be taken at face value.  The Deadhorse ASOS reported sustained winds of 30-35mph, with higher gusts, at the same time, so the winds may have been somewhat higher over the ice (and there was still 100% ice cover at that time).

To put the peak ERA5 wind in context, I pulled out ERA5's maximum hourly wind speed for the month of June in past years at the same location.  Since 2000, nearly half of all years have seen a June wind speed higher than this year's event, so it really doesn't stand out as an unusual event.

Here's the surface analysis chart for the same time, 4pm AKDT on June 10, courtesy of Environment Canada; click to enlarge.  There's a modest pressure gradient across the near-coastal waters north of Alaska, consistent with brisk easterly winds, but again nothing especially unusual.

It seems that further analysis of the event would require information about ice thickness and motion in the area, and perhaps satellite imagery would reveal the ice subduction process described by Quintillion.

Any sea ice experts care to weigh in?

August Climate Data

Looking back at climate data for August, it was a very warm month indeed - the 3rd warmest on record for the state overall.  Only in 1977 and 2004 was the month of August warmer for Alaska as a whole.  It was also the warmest month of the 2023 summer relative to normal; there were only a few days in August that were marginally cooler than normal overall.

Regionally, the warmth was most unusual across the eastern interior, and southwestern Alaska was relatively cooler.  A major trough-ridge contrast explains the pattern; both the trough over the Aleutians and the ridge over western Canada (and SE Alaska) were quite strong for the time of year, and the circulation anomaly pumped a great deal of warm air northward into Alaska.

South-central Alaska saw yet another wet and cloudy month, and the Southeast had more rain than in the very dry July, but it was still drier than normal.  ERA5 and NCEI disagree on whether the Southeast Interior saw moisture relief, but Rick Thoman notes that rains were highly variable in that area; some places remained very dry.

The trend towards warmer conditions as summer progressed probably had at least some connection to the developing El Niño, because a ridge over southern and eastern Alaska, and northwestern Canada, is a typical feature when El Niño is occurring in August.  The map below shows the typical August 500mb height patterns during El Niño.

However, other aspects of the circulation pattern from eastern Asia to Canada were highly atypical for El Niño.  Notice the tendency for lower heights/pressure near the Sea of Japan: that's the opposite of what happened in August this year (see the first map in this post).  Also, most of Canada tends to be cooler than normal in August during El Niño (first map below), but in fact it was very warm across Canada's northern tier, and the record fire season just kept on going.  The heat over Japan and northern Canada was relentless this summer, and that certainly can't be pinned on El Niño.

A better explanation for the overall North Pacific - North America pattern is the ongoing negative PDO phase.  It's very unusual indeed for a negative PDO to persist as El Niño emerges, but this year the PDO has remained entrenched in the negative phase; and the warmth to the east of Japan has a lot to do with that (i.e. the warm SSTs there correspond to a negative PDO).

Here's the spatial (air) temperature pattern for August when the PDO is negative.  It doesn't line up with the warmth in the eastern Pacific this year, but otherwise it's not a bad match from Japan to northern Canada.  Texas tends to be hot too - and they had a really awful summer.

Rounding out the Alaska climate maps for August, it was cloudier than normal once again for western Alaska and the northern Gulf coast.  It was also relatively windy from the Bering Sea to the western and northern interior.

It's also interesting to note the very high humidity (dewpoint) for much of the state.  This goes a long way to explaining the extremely high overnight temperatures and therefore also the remarkably delayed autumn foliage in the Fairbanks area.

For future reference, I'll include maps below for climatological summer, June through August.  It was the second warmest summer on record for the North Slope, and third warmest for the eastern interior climate divisions, according to the NCEI data.  2004 still holds the summer temperature record by some margin for the state as a whole and for the west coast, interior, and North Slope, as well as the northeast Gulf and central and southern panhandle regions.  2019 was the warmest summer for south-central and southwestern Alaska.

Rainfall, wind, and cloudiness were all exceptionally high for the northern Gulf Coast and southwestern mainland, but Southeast had really quite a dry and sunny summer.

And finally, it's just interesting to note again the dramatic shift that took place in mid-July as the weather suddenly and belatedly turned very warm.

Slowly Cooling Off

Overnight temperatures are now finally dipping below freezing in colder spots around the interior - and it's about time.  The coldest reports this morning came from the usual places:

Salcha RAWS  22°F

Circle Hot Springs  25°F

Chicken  25°F

Bettles saw its first freeze of the season, 28°F, which is very late: the record latest is September 16 in 1978.

The coldest temperature so far at Fairbanks airport is only 38°F, which ties with 1989 and 2007 for record highest up to this date.

Astonishingly, it appears the autumn colors are *still* not fully developed around the Fairbanks area, although a change is starting to show up.  Webcam views from Nenana and Cleary Summit today:

It's hard to believe that on this date in 1992, the permanent winter snow cover was established in Fairbanks.  See the following write-up from Rick Thoman, with photos from this date 31 years ago!

https://ak-wx.blogspot.com/2013/09/septmeber-1992the-first-snows.html

Main blog entry here:

https://ak-wx.blogspot.com/2013/09/semptember-1992setting-table.html

Arctic Sea Ice Loss Revisited

An interesting study was published last week on the topic of Arctic sea ice loss, led by Igor Polyakov of UAF.  Here's a write-up from the Geophysical Institute:

https://www.gi.alaska.edu/news/new-research-points-mechanisms-atlantification-arctic-ocean

The thrust of the paper is that a change in the Arctic atmospheric circulation has contributed to the lack of sea ice loss since 2007, when the seasonal minimum reached its first dramatic record low.  Since then, Arctic sea ice has been relatively stable, and the new work suggests this is because the new wind patterns have been transporting more fresh water into the Pacific side of the basin.  Fresh water is less dense than salt water and resides immediately under the sea ice, and it serves to insulate the ice from warmer ocean water that would otherwise tend to melt the ice from below.

The following figure from the paper shows the rather surprising hiatus in sea ice loss (based on September ice extent), and the inset map shows the change in upper ocean "available potential energy", which signifies the quantity of fresh water at the surface.  The change to more fresh water and more stable conditions is largely found in the Beaufort and northern Chukchi Seas.

The paper presents a variety of oceanographic analyses to support the hypothesis about redistribution of fresh water into the Pacific sector, and the basic idea seems plausible to me.  However, to throw in my own two cents, I also think it's likely that a portion of the fresh water volume increase has arisen because of greater precipitation and greater river inflows in the past 15 years.

Let's illustrate these two additional factors.  First, it's well-known that the Arctic hydrological cycle is intensifying in response to rapid warming, and the 2021 NOAA Arctic Report Card highlighted the increase in Arctic river discharge, particularly for Eurasian rivers (providing the bulk of Arctic Ocean inflows):

Here's the link:

https://arctic.noaa.gov/report-card/report-card-2021/river-discharge/

The total discharge of all rivers to the Arctic Ocean is estimated to be approximately 5000 km³/year, which is equivalent to about 35 cm of fresh water depth over the entire Arctic Ocean area - but much of this water will reside close to the continents or be transported into the Pacific sector and the Beaufort Gyre, where salinity is low:

It seems plain that an increase in discharge will tend to fortify the fresh water layer in the Pacific/Canada sector of the ocean, especially with the more favorable wind circulation of recent years.  The percentage change in river flows evidently isn't large, but the recent effect may have been magnified by the change in transport patterns.  Here's a diagram of ocean and ice transport around the basin, from Polyakov et al:

The other factor may be increased precipitation.  Ground-truth precipitation measurements are very scarce and uncertain, of course, so here's the ERA5 reanalysis estimate of annual precipitation change between the two 15-year periods that are compared in Polyakov et al.

An increase of 5-15% is widespread across the Asian side of the Arctic Ocean, according to the model.  How much does this amount to?  The annual mean precipitation is very low, only about 250-300 mm (10-12 inches) across much of the area where the increase has occurred:

But in the northern Barents Sea, where annual precipitation is higher and the percentage increase has been large, the absolute increase amounts to 50-100 mm/year.

Interestingly, the total Arctic Ocean fresh water input appears to be roughly evenly split between the average ~35cm depth of river discharge and the average precipitation of about 20-50 cm.  (Some of the snow will evaporate, of course, but some of it melts in summer and feeds the surface fresh water layer.)  With both fresh water sources producing higher volume in recent years, it seems this has probably contributed to the build-up of fresh water and thereby to the longevity of Arctic sea ice.

The question remaining in my mind is how long can sea ice melt be postponed by these mechanisms in the face of rapid atmospheric (and presumably ocean subsurface) warming?  Polyakov et al suggest that the atmospheric circulation regime of the last 15 years is likely to reverse back to the opposite phase very soon, which would be unfavorable for sea ice and perhaps bring about another rapid decline.  It does seem unlikely that sea ice will hold on much longer, unless something very unexpected happens.  Courtesy of Zack Labe's website, Arctic temperatures are on a steep uptrend, including since 2007:

But time will tell, and I imagine that a few more surprises lie ahead.

Record August Warmth

Just a quick note to re-emphasize how extraordinarily unusual the warmth of the last 6 weeks has been for the central and eastern interior, and particularly in terms of daily minimum temperatures.

Here's the leaderboard for average daily low temperatures in Fairbanks from July 16 through the end of August:

2023   55.2°F

2016   52.7°F

1990   52.5°F

2004   52.3°F

1974   52.3°F

2017   52.2°F

2021   52.0°F

This year stands out like a sore thumb.  The departure from normal represents a 3.3 standard deviation anomaly compared to the 1991-2020 distribution.  This is an extreme climate event, with a "return period" of many hundreds of years assuming a stable climate.

Accordingly, it's perhaps no surprise to see that there is *still* very little color in the foliage, based on Nenana and Cleary Summit webcam views from yesterday afternoon:

Here's a chart of Fairbanks airport daily minimum temperatures since June 1:

Looking at monthly statistics for August, Eagle had its warmest August on record (1905 is warmer in the books, but the minimum temperatures look much too high).  Northway was also the warmest on record, while Fairbanks was 3rd warmest overall, but warmest for daily low temperatures.  Gulkana also had its warmest August for daily low temperatures.

But change is in the air: snow was falling in the hills above Fairbanks yesterday morning.  Here's a view from Wickersham Dome, before the sky cleared out later in the day.

Snow at 2600 feet north of Fairbanks this morning! @NWSFairbanks @Climatologist49 pic.twitter.com/pNCsUXk0YO

— Adam Gill (@adam_gillwx) September 2, 2023