September Climate Data

With September climate data now available, let's look back at the predominant weather patterns for the month.  The monthly-mean circulation anomaly was quite similar to August, with a high-pressure ridge retaining a strong influence across southern Alaska.  In contrast, low pressure was dominant over the Arctic Ocean, and the Arctic Oscillation was virtually tied with 1975 as the most positive on record in the month of September.  The mid-atmosphere height (pressure) anomaly shows the strong north-south gradient:

The corollary of a strong pressure gradient is, of course, strong wind, and the enhanced westerly flow brought abundant moisture and frequent rainfall to central and northern Alaska.  Again the pattern showed similarities to August, and much drier than normal conditions continued from the Kenai Peninsula to Southeast Alaska.

The monthly precipitation total in Fairbanks was the highest for September since 1925, and as noted in Rick's comments, some rivers saw their highest measured flow on record for so late in the season.

Check out the contrast in sunshine fortunes across the state: western Alaska and the northern interior were extremely cloudy, but southeastern regions were much sunnier than usual.

Despite the broad similarity in the pattern, temperatures were not as overwhelmingly warm in September as in August; the flow direction across central and northern Alaska had more of a southerly component in August.  The figures below highlight a strong north-south gradient that fits with the pressure gradient and the enhanced westerly flow.

Warmth in southern Alaska was amplified by ongoing high sea surface temperatures across the North Pacific to the south of Alaska, and especially to the south of the Aleutians.  Although the core region of unusual warmth is quite a long way from mainland Alaska, southwesterly winds commonly track from the Aleutian region up into southern Alaska, making the oceanic warm anomaly a "source region" for unseasonably warm air.

As for Arctic sea ice, the minimum ice extent on September 12 was the 10th lowest in the satellite-era period of record.  Seasonal meltout was less than might have been anticipated at the start of summer, owing to a relatively cool summer with persistent low pressure across the high Arctic Ocean.  For this reason, the NSIDC describes the melt season as "unusual in the extreme".  My impression, however, is that much of the ice that qualified for "ice extent" (based on the 15% area threshold) on the eastern side of the pole was very tenuous, and a different weather pattern could easily have produced a large drop in extent.  Here's a map of estimated ice concentration just a few days before the date of the seasonal minimum:

Compare the eastern hemisphere coverage to the NSIDC "ice extent" map from the same date:

Weather Data from Central

I recently noticed that weather data from the town of Central started showing up on the surface observations map, and it seems that 24-hour data is indeed now flowing from instruments there for the first time since 1989.  Central's location tucked up against the east side of the White Mountains makes it an interesting spot for weather.  Here's the location (click to enlarge):

So far this month, Central has seen persistent snowfall on four days, and judging from webcam photos today, there is a modest snow cover:

Interestingly, Central saw snow for much of Sunday and Monday earlier this week, while Fairbanks saw none, as cold air flowing from the northeast was lifted over the north side of the terrain but subsided on the Fairbanks side.  The temperature in Central was as much as 18°F colder on Sunday afternoon: 27°F and snowing in Central, 45°F and dry in Fairbanks.  We normally think of chinook warming reaching Fairbanks under southerly flow across the Alaska Range, but the White Mountains can also produce a similar effect.

The sheltered interior location of Central creates a surprisingly cold winter climate compared to some other northern and eastern interior valley sites.  To illustrate this, I calculated the monthly average temperature difference of several locations compared to Fairbanks, using all available overlapping dates in the joint periods of record back to 1930.  (The climate history is a lot more complete in some locations than in others, but the temperature differences between two locations tend to be quite stable over time, so I'm not worried about issues of trend in this analysis.)  Click to enlarge:

Central, shown with a green line, is about 8°F colder than Fairbanks in late winter, and is also significantly colder than Eagle to the southeast.  Even though Central is above the Yukon River valley (its elevation of 940' is higher than Eagle's elevation), Central's winter cold is more akin to that of Fort Yukon (430'), although it's not quite as severe as the lower elevations of the Yukon Flats.

Central is also quite a bit colder than Bettles in mid-winter, even though Central is well to the south; but it's certainly not as cold as Allakaket (not far from Bettles).  As an aside, Allakaket is a real contender with Chicken for coldest winter climate in inhabited interior Alaska, based on the respective comparisons with Fairbanks (but Allakaket data ended in 1982).

Central's ICAO code is PACE, and the latest observations can be viewed here:

https://www.weather.gov/wrh/timeseries?site=pace

[Update October 9]

A case in point this morning: low temperatures around 20°F in the Fairbanks area, but 6°F in Central.  Eagle (not shown on the map) got to 13°F.

Much Colder

The first deep freeze of autumn has enveloped northern Alaska in recent days, and some notably chilly conditions have been registered for this early in October.  Here are some of the lowest readings in the past few days:

October 1:  12°F Ambler

October 2:  8°F Bettles

October 3:  5°F Wiseman, 2°F Kanuti NWR RAWS

Temperatures of 14°F (-10°C) or lower were particularly widespread yesterday morning, and a few sub-zero readings showed up on the North Slope:

Snow cover is widespread on the hills and - judging from webcam photos - much of the southeastern interior, but the western interior is mostly snow-free to the south of the Yukon River.  Here's a satellite image from yesterday afternoon, with blue color enhancement showing where the algorithm picked up snow cover:

How about this for late autumn views in the northeastern interior settlement of Arctic Village this morning:

Light snow fell in Fairbanks for about 10 hours on Thursday, but the measured accumulation was only 0.1", and that melted out rather quickly with a high temperature of 36°F later in the day.  The following day - October 2 - was chilly, with a high temperature of only 33°F; but it's not unheard-of to get daily high temperatures below freezing by this time of year.

It's interesting to look at how the calendar affects the probability that snow on the ground will melt out before further snow arrives in Fairbanks - consider the chart below (click to enlarge).

At the beginning of October there is still about an 80% chance that a 1-inch snow cover will disappear at some point in the subsequent days or weeks, but this drops to 50% on October 10 and only 20% on October 20.

The probability of a 3-inch snow cover melting out in Fairbanks drops even more rapidly in early October: from over 60% in the first few days of the month to 25% by October 10.  So even a relatively heavy snowfall at the very beginning of October is not safe from subsequent melting; but you only have to wait a week or so for much better odds of the snowpack remaining through winter.

As for a 6-inch snow depth, there have only been 3 years in which such a substantial snow depth melted out in late autumn: 1936, 1972, and 2015.  In 1936, 6 inches fell on the 17th of October but disappeared by the 26th, and in 1972, temperatures above 50°F in mid-October melted an apparently robust early October snowpack.  The 2015 meltout occurred following record snow in late September, and snow didn't return until October 29th.  The blog archives make for interesting reading, e.g.

https://ak-wx.blogspot.com/2015/09/historic-snow.html

https://ak-wx.blogspot.com/2015/10/fairbanks-melt-out.html

Of course no mention of early winter snowpack loss would be complete without mentioning the "brown Christmas" of 1934, following the great early-December chinook - unquestionably one of the most extreme weather events in the recorded climate history of Fairbanks.

https://ak-wx.blogspot.com/2011/12/its-white-christmas.html

Snow Showing Up

Snow fell in the hills above Fairbanks today, and this evening's scene has a wintry look.  Here are a couple of webcam photos from Cleary Summit (2250' on the Steese Highway) today:

Snow is also now on the ground in Bettles:

Fairbanks airport reported just a few minutes of snow falling this afternoon, but no accumulation; valley-level temperatures are still well above freezing in Fairbanks-land.

Late September has become pretty normal for the season's first observation of falling snow in recent years in Fairbanks, while the first accumulating snow tends to be in the first half of October - but with considerable year-to-year variation, of course.  The chart below shows the long-term variation.

It's remarkable how often the first snowflakes used to be observed around the beginning of September; but on the other hand, the first measurable snow and snowpack onset dates were comparable to today's climate back in the 1930s-1940s.

A Few Notes

A change of weather pattern in the past week or so has brought an end (for now) to extremely wet weather across interior and western Alaska, and clear skies have produced widespread overnight freezes for the past several days - as befits the time of year.  Contrast the mid-atmosphere height (pressure) pattern in the past week with the pattern that produced excessive interior rains earlier in the month:

It's worth pointing out that Google's WeatherNext2 AI forecast anticipated the change from about 10 days in advance, with a stronger signal than other models: see the end of my post here:

https://ak-wx.blogspot.com/2026/09/windy-and-wet.html

The WeatherNext model is easily the most skillful of the current generation of medium-range forecast models, and has just been upgraded to version 3.

Rick Thoman's latest post discusses the very unusual late season peak in water flow for several interior rivers: in some instances, the highest flows on record this late in the season.

https://alaskaclimate.substack.com/p/rivers-running-high-in-interior-alaska

The cloudy, wet weather earlier in the month seems to have significantly delayed the progress of autumn foliage colors across interior and southern Alaska.  Colors were peaking in Fairbanks last weekend, but remarkably there was very little color in the Anchorage area earlier this week.  Here's a photo I took on Wednesday morning (September 23) looking over a very green southern portion of Anchorage (Turnagain Arm to the left):

On another note, western Alaska "dodged a bullet" this week in terms of a strong Bering Sea storm involving ex-typhoon Dujuan, which previously caused a lot of problems in Japan.  High winds affected southwestern Alaska, bringing memories of last year's devastating ex-Halong, but the latest event was more of a glancing blow than a direct hit, and flooding was not severe.  The sequence of maps below shows the 6-hourly progression of low pressure from the central Aleutians to the Bering Sea waters west of Nunivak Island.

Peak wind gusts on Thursday exceeded 84 mph in Cold Bay, with sustained winds as high as 62 mph - the highest sustained wind measurement for over 6 years in Cold Bay.

78 mph gusts were observed on Akun Island - near Dutch Harbor - and Cape Newenham on the mainland briefly measured 86 mph in a squall.  Here's a map of reported peak wind gusts in mph:

Finally, the unseasonable daytime warmth that is now prevailing in the interior (courtesy of chinook winds downstream of the Bering low) will soon end, and the inevitable temperature decline toward freeze-up will resume.  Judging only from the temperature forecast, Google's WeatherNext2 model suggests snow may be in the cards by the end of the week; the following graphic shows the forecast ensemble 25th, 50th, and 75th percentiles for Fairbanks:

Still Raining

I'm traveling this week and most of next, so posts will be light, but it's worth noting the exceptional persistence of wet weather in Fairbanks.  With well over an inch of rain in the last two days (and it's still falling), this month has brought 3.77" through the 16th, making this already the wettest September since 1925.

With incessant cloud cover, the average daily temperature range so far this month in Fairbanks is only 12.4°F, far lower than the normal range of 20°F, and easily the lowest on record.  No crisp frosty mornings and dry sunny days this year.

Here's a chart (click to enlarge) of the number of hours with rain, snow, or drizzle reported in Fairbanks in the first half of September since 1950.  The 2026 total would rise quite a bit if we included the 16th and 17th, as it's been raining almost non-stop in the last two days.

Notice the consecutive high number of hours since 2022 - this is the fifth straight year with unusually frequent precipitation in the first half of September.  It seems quite likely that this is related to the persistence of unusual warmth in the northwestern Pacific Ocean since 2022.  Some may recall that the emergence of that sea surface temperature anomaly was implicated in the intensity of ex-Typhoon Merbok in September of that year:

https://ak-wx.blogspot.com/2022/09/more-on-ex-merbok.html

Last month's SST anomaly map shows the similarity of the ocean temperature pattern this autumn:

August and Summer Review

Now that August climate data is complete, it's time to pen a few comments on the unusual summer of 2026.

First of all, it's interesting to observe that August was the warmest month of summer statewide, and that's rare: it has only happened twice in the last century, in 1941 and 1959 (according to NOAA/NCEI data).  Here's this summer's daily statewide temperature index from UAF (note that this is an index based on departures from normal, not absolute temperatures):

In 102 years of data, June has been the warmest month of the year in Alaska five times, including a tie with July in 2024.  July is the warmest month in over 90% of years.  Of course, this breakdown varies with region.  Fairbanks more often sees June as the warmest month, and has never had August as the warmest month, but southern maritime locations more often see peak warmth in August (e.g. over 75% of years in Cold Bay).

August was also the first warmer-than-normal month statewide since last November, breaking a streak of 8 consecutive below-normal months (although April was essentially normal) - based on the 1991-2020 normal.  The last time 8 months in a row were cooler than normal was in 2012, not long before the sudden reversal (which proved enduring) to persistent warmth in early summer 2013.

Here's the mid-atmosphere pressure map from August, showing that high-pressure ridging became a prominent feature across all of southern Alaska.

The ridge was fostered by, and also acted to reinforce, much above normal sea surface temperatures from the Sea of Japan to the waters south of the Aleutians.  This is basically a negative PDO signal, although the warm pool is shifted slightly north compared to its classical position.

August warmth was most anomalous in northern and eastern Alaska: the North Slope was very warm indeed, and NCEI ranks the Southeast Interior division as the 5th warmest on record (but not as warm as 2023).

August rainfall was feast or famine, so to speak, with a sharp contrast between very damp conditions in the western interior and much drier weather to the south and east of Fairbanks.  It was also very dry in the Aleutians, courtesy of the North Pacific ridge.

Sadly, the very warm and dry conditions in the southeastern interior allowed a damaging fire to occur in Tok.  To quote Rick Thoman, "More than 40 homes in Tok were damaged or destroyed by the fire, making it one of the most destructive fires in terms of societal impacts in Alaska history".

The Tok fire illustrates the adage more often used in hurricane preparedness: "it only takes one".  It's no comfort to those affected, but the statewide seasonal wildfire activity was far below normal this summer - only 89,000 acres burned, the lowest total since 1995.

The very inactive fire season is in some ways a surprise, because despite the overall cool weather in June and July, there was a lot of lightning for a 3-week period in late June and early July.

Here are some maps to illustrate the spatial distribution of this season's lightning and its departure from normal.  The southeastern interior saw less lightning than normal, but a much larger area from the Y-K Delta to the western and northern interior was significantly more active than normal.

The Lower Yukon fire monitoring zone was the most unusually active region in Alaska, with a new seasonal record for total lightning strikes (data since 2012).

On the inactive side, the upper Tanana River region saw the most significant lightning deficit compared to normal:

The lack of significant fire growth in the wake of the June-July lightning onslaught can be mainly attributed to the wet spring, along with a notably cool May, preventing fuels from becoming very dry before the onset of lightning season.  Of course the absence of really warm weather in June and July helped a lot too.

As one example of the "missing warmth" in Alaska's interior, Fairbanks only reached 81°F for the highest temperature of the summer; this has only happened a few other times in the period of record.  The normal summer peak temperature in Fairbanks in recent decades is 88°F.  In contrast, southeast Alaska reached unusually high temperatures in mid-August, and the period of record does not show any other year with such a contrast (i.e. peak summer temperatures around 5°F warmer in Southeast than in Fairbanks).

Wrapping up the discussion with some maps of the overall summer climate anomalies, we see that southern Alaska and the central interior were generally cooler than the 30-year normal, but not dramatically so in most locations.

A more significant anomaly is seen in the summer's total rainfall, which was well above normal in the southwestern mainland and far below normal in southeast Alaska.

NCEI's data shows the central and northern panhandle regions with the second lowest summer rainfall in the last 30 years (3rd percentile on my map) - 2004 was considerably drier.  However, summers drier than this year occurred with some regularity prior to 1995, so the drought isn't quite as remarkable on a long-term basis.  Here's a chart for the central panhandle division:

Summer sunshine was most abundant (relative to normal) in the extremities of Alaska - west, north, southwest, and southeast.

It's interesting to see the unusual combination of sunnier and wetter than normal in western Alaska; this happened because June (long daylight hours) was much sunnier than normal, and August (typically wetter) was much wetter than normal - and these two months dominated the respective seasonal anomalies.

Finally, for completeness - July was windy in southern Alaska, and August was windy in the southwestern mainland, but all three months were notably less windy than normal on the North Slope.  With less breeze than normal, the mosquito season may have been an especially bad one for the long-suffering residents of the far noth.