Winter at O'Brien Creek

Back in 2012, before I took over as lead author on this blog, Rick Thoman penned a very nice essay about candidate locations for the coldest place in Alaska.  I've long been interested in his comment there that a short-lived co-operative observing site at O'Brien Creek in the Fortymile Country tended to be colder in winter than the famously cold settlement of Chicken.  After all this time I've taken a close look at the data, and I can confirm that indeed O'Brien Creek was (and presumably still is) slightly colder than Chicken in winter.  This would give it the coldest observed winter climate in interior Alaska.

The O'Brien Creek co-operative observers (Larry and June Taylor, to whom we owe a debt of gratitude!) recorded the weather at their place of residence from May 2001 through August 2010, giving us 9 winters of mostly complete climate data.  There are some missing days or periods here and there, but overall the November-March data is 94% complete for this period; this is about the same as for Chicken.

Looking at all dates for which data is available from both O'Brien Creek and Chicken, we find that the November-March mean temperature was 0.5°F colder at O'Brien Creek; so overall it is a very small difference, and not statistically significant.  However, the differences are larger in late winter and are strongly significant for both February and March.  The chart below shows the mean daily maximum and minimum temperatures by month for the overlapping period of record; we see that differences are very small in November through January, but O'Brien Creek is colder in February and has lower minimum temperatures in March.

The November-March mean temperatures for each winter with mostly complete data are shown below, with the top pair of lines showing mean maximum temperatures and the middle lines showing mean minimum temperatures.  The bottom pair of lines shows the differences between the two locations, with the scale on the right axis.  The winter of 2005-2006 produced relatively warmer daily maximum temperatures at O'Brien Creek, but in most of the other winters O'Brien Creek was slightly colder.  February mean temperatures were colder at O'Brien Creek in all but one winter, and in February 2007 the difference was a substantial 6.7°F.  The 2006-2007 November-March mean daily minimum temperature of -30.6°F is remarkably cold and ranks up (down) there with the likes of Umiat, Arctic Village, and Allakaket for historic cold.

It's interesting to look at the absolute minimum and maximum temperatures observed at the two locations during the period of overlapping data; see below.  As before, the green and black lines show the comparison for maximum and minimum temperatures, and the red and blue lines show the differences.  For November through January there are no obvious systematic differences, but from mid-February through March the extremes become relatively colder at O'Brien Creek for both maximum and minimum temperature.  So for example O'Brien Creek recorded -61°F on March 3, 2007, and -51°F on March 15, 2009, whereas the latest dates for -60°F and -50°F at Chicken are February 24 and March 4, respectively - and that's using the entire period of record at Chicken, which is now almost 20 years.

The tendency for cold late winter conditions at O'Brien Creek seems very likely to be a direct result of the steep nearby topography, which maintains long hours of shadow over the valley until well into spring.  The topographic map below shows the situation, with terrain rising up steeply in most directions, and the photo below (courtesy of Rick Thoman) further illustrates the nature of the surroundings.  The deep valley may also help shelter the valley floor from prevailing wind patterns, creating more frequent calm conditions.

The topographic map for Chicken (below, on a somewhat expanded scale) shows much less terrain variation in the vicinity of the station, and therefore it seems likely that both solar and wind-induced warming would have more of an effect on Chicken in late winter.  As cold as Chicken is, then, it is not ideally situated for cold, as the deeper valleys of the Fortymile Country are capable of producing even colder winters on average.

50 Below Reached

The negative half-century mark (-50°F) was measured for the first time this winter in Alaska today (to my knowledge); the "lucky" spot was the Kanuti Lake RAWS (elevation 524') in the Kanuti National Wildlife Refuge.  Kanuti Lake is about 40 miles southeast of Allakaket.  The hourly observations, shown below, did not actually hit -50°F, but after rounding to the nearest degree the minimum temperature would be reported as -50°F (and a slightly lower reading presumably occurred at an intermediate time).  Note the extreme rise in temperature between 10am and 11am AKST this morning, when the first stirring of a breeze apparently mixed out the (no doubt very shallow) surface-based inversion.

As far as I'm aware, the last time -50°F or lower was observed in Alaska was early last February.  The Sag River/Pump #3 site came close on December 4, with -49°F, and Kanuti Lake was almost as cold (-48°F) on December 10.  -50°F or colder is observed every winter somewhere in Alaska, although in some winters that is about as cold as it gets.

Here's a nice webcam photo from Arctic Village yesterday at solar noon, with the sun about 1° below the horizontal and some light ice fog evident near the ground.  The airport reported -45°F at the time.

Strong Polar Vortex

It's been almost two years now since media discussion in the U.S. latched onto the idea of the "polar vortex" as a means of explaining the unusual cold that affected parts of the lower 48, while Alaska was - as usual - experiencing the opposite temperature anomalies.  Google's search history shows the spike in interest in the term "polar vortex" in early January 2014.

From a scientific perspective, the discussion in 2014 was misplaced, as the real polar vortex is a phenomenon in the stratosphere that most certainly did not migrate southward into the upper Midwest; it seems the term was used to refer loosely to the southward displacement of cold air that often resides over northern Canada in winter in association with a cyclonic circulation in the troposphere.

The behavior of the real stratospheric polar vortex is actually much more interesting this winter, as the vortex is currently very intense - more so than ever previously observed at this time of year in the era of balloon observations back to 1948.  Using an index of 10mb westerly wind speed from 50°N to 80°N, the average strength of the circulation since November 1 is more than 10% higher than in any previous year during this period (see the chart below).  The polar vortex is a wintertime phenomenon, and according to the long-term normal, the 10mb wind speed in this latitude band usually peaks at just over 30 m/s in early January.  However, this year the average since November 1 is over 39 m/s, and the wind speed reached 45 m/s on December 5, which was the earliest on record for such a strong circulation.  (As an aside, the apparent long-term upward trend in the chart is more than a little interesting, although the data from the early years might be questionable, as 10mb observations were pretty thin back then.)

What does this mean for weather patterns down here at the surface?  The strength of the stratospheric polar vortex is correlated with the Arctic Oscillation, which is a measure of the pressure gradient between middle and high latitudes, and of course this makes sense: we would expect a strong cyclonic circulation aloft to occur in tandem with a strong westerly flow at low levels.  The magnitude of the correlation is not particularly strong, but nevertheless this year's tropospheric conditions bear out the relationship, as the AO has been significantly positive on average since the beginning of November.  The 500mb height anomaly since November 1 (see below) shows lower than normal heights (pressure) from far eastern Russia to Greenland and the northern North Atlantic area.  Unusual ridging over eastern North America and Europe has brought extremely unusual warmth, with many locations undoubtedly seeing their warmest December on record.

The map below shows the characteristic 500mb height pattern during 10 previous years in which the polar vortex was unusually strong in November and December.  The locations of the ridges over eastern North America and Europe are quite similar to what has occurred this year, which suggests that the recent weather anomalies are probably closely connected to the polar vortex strength.  In this sense, then, the media could be justified in blaming the excessive warmth on the polar vortex this winter; but I doubt they'll pick up that story.

The relationship between Alaska's weather and the polar vortex strength is not as clear, and more investigation might be worthwhile.  However, I'll point out that low pressure in the Arctic basin, and a positive AO phase, are somewhat favorable for enhanced snowfall in Fairbanks, because lower pressure to the north favors a more westerly flow regime.  The chart below shows that November-January snowfall is rarely low when the AO is significantly positive, and conversely a strongly negative AO greatly increases the odds of below-normal snowfall.  With this in mind, I think we can place part of the blame for interior Alaska's snowy November on the strongly positive AO phase, although the details of the snowy weather were more closely tied to the pressure pattern in the North Pacific-Bering Sea-Alaska sector, and this is only partly related to the larger-scale AO phenomenon.  With December so far being almost snowless in Fairbanks and Bettles, the AO phase is obviously only a small part of the story.

Winter Temperature Variance

I'll present this more or less without comment, as I'm feeling a bit under the weather this weekend (no pun intended); but I thought it would be interesting to examine the long-term changes in variance of temperature during winter in Fairbanks.  As we've noted before, there were some remarkable extremes in the 1930s; so has there been a long-term decrease in temperature variance?

The answer is yes.  The chart below shows the November-March standard deviation of daily, weekly, and 30-day mean temperature anomalies, with the anomalies calculated relative to contemporary normals (which have warmed substantially over time).  The variance has decreased at a similar percentage rate for each time scale from daily through 30-day mean temperatures.  We can conclude that the modern winter climate of Fairbanks has somewhat less variability of temperature than in earlier decades prior to about 1980.

What does the decreasing variance look like in practice?  The charts below show daily temperature anomalies for 1950-51, which had daily and 30-day variance very close to the 1930-1959 averages, and for 1993-94, which was very similar to the 1985-2014 average.  The decrease in variance is about 10% on both time scales, which is small but arguably just about perceptible on the charts.

The winter with the highest combined daily and 30-day variance was 1980, and that with the lowest was 1987 (interestingly the latter coming at the end of a strong El Niño episode).

December 1934 Chinook

A few weeks ago reader Mike suggested that we take a look at the great chinook event of December 1934, which brought record-breaking warmth to interior Alaska and led to the only "brown" Christmas in Fairbanks history.  The magnitude of the warm anomaly was so extreme that nothing really comparable has happened in the 8 decades since.

To set the stage, Fairbanks had experienced a dry early summer (4th driest June on record) but a rather wet August, followed by a mild and very dry autumn (2nd driest October-November on record).  By November the Pacific Decadal Oscillation (PDO) phase was significantly positive, as cool ocean temperatures dominated the central North Pacific and waters were relatively warmer along the North American coast (see figures below).  However, temperatures in the tropical Pacific were near normal, with neither El Niño nor La Niña conditions in play.  The Atlantic Multidecadal Oscillation (AMO) was in a long-term positive phase, but Atlantic waters were not particularly warm overall in late 1934, and the 1934 Atlantic hurricane season had been inactive compared to surrounding years.  In the lower 48 the summer of 1934 was extremely hot and dry as the Dust Bowl drought intensified.

The synoptic conditions that led to the December chinook and heat wave are illustrated in the figures below, which are taken from the 20th Century Reanalysis.  The left column shows maps of daily mean 500mb height from December 1-10, and the right column shows the corresponding maps for sea-level pressure.  As we would expect, the event involved an intense trough over the Bering Sea and an enormous ridge over southeast Alaska and western Canada, with an extremely strong and persistent pressure gradient between the two features leading to southerly flow from the sub-tropics into far northern latitudes. While the magnitude of the anomalies was huge, the persistence of the setup was perhaps even more astonishing.

The highest temperatures observed during the period December 1-10 are shown in the map below.  The interior Alaska "winner" was Nenana, with 61°F on December 8.  Fairbanks reached 58°F on the 5th, but even more remarkably exceeded 50°F for 5 consecutive days from the 4th through the 8th.  These are the only December days in Fairbanks history to have reached 50°F, and only a couple of other days have exceeded 50°F from November through February (November 25, 1936 and January 16, 2009).  The highest temperature observed statewide was 64°F at Sitka Magnetic Observatory on the 8th.

Overnight minimum temperatures were also extremely high for the time of year, with Fairbanks recording its only instance of a 40°F daily minimum temperature between October 21 and April 18.  The ability of temperatures to stay so high throughout the long hours of December night was aided by the elimination of snow cover with the first 50°F day on the 4th, and the snow cover was then reported at zero from the 5th through the 11th.  The rapid loss of snow cover was in turn helped by the lack of earlier snow pack resulting from the dry autumn; there was only 1 inch on the ground on November 28, and a few inches that fell just prior to the chinook quickly melted out.  Snow cover was not regained permanently until December 26, so Fairbanks missed out on a white Christmas for the only time in its history.

The magnitude of the warm anomaly can also be appreciated by comparing the observed temperatures to the 1981-2010 normals.  The December 5, 1934 daily mean temperature of 49°F was 51.6°F above the 1981-2010 normal, which is the largest warm anomaly of any day in Fairbanks history.  Of course the modern climate is warmer than that of the early 20th century - the difference is about 5°F in December - so actually the anomaly was more like 55°F in comparison to the earlier normal.  I would guess that the list of places in the world that have ever been 55°F above normal is fairly short.

For the week ending December 9, 1934, the weekly mean temperature was 41.4°F above the 1981-2010 normal; the next warmest week was in January 1981 (37.8°F above normal).

Reader Mike speculated that such an anomalous event might occur about once every 200 years.  If we assume the true peak daily anomaly was +55°F, and if we assume the variance is unchanged over time, then the December 5 anomaly was a little less than +4 standard deviations.  In a Gaussian distribution, a +4 SD anomaly would occur once every 31574 days on average, or once every 86 years; so the chance of this happening in any one winter (November-March) might be about 1 in 200.  Good guess, Mike!  The only additional comment I would make is that this is just a rough estimate, because the temperature variance is unlikely to be stable over time; it is subject to long-term trends and also probably depends on multi-decadal or longer timescale behavior of major climate phenomena like the PDO or ENSO.  As I suggested here, it seems that the climate regime of the 1930s may have favored high-amplitude circulation anomalies near Alaska, and of course it may not be a coincidence that the Dust Bowl occurred in the same decade.

Late Freeze-Up

For some weeks now I've been checking the Nenana Ice Classic webcam every now and then to see if the Tanana River is finally frozen over.  Remarkably, it still isn't, as there remains a substantial width of open water that has been producing copious amounts of steam fog in the recent colder weather.  I'm not sure just how unusual it is to have open water this late - and of course the open patch may be less significant than it appears on the webcam - but nevertheless it seems notable.  Here's a shot from earlier today, with a temperature of -20°F at the nearby airport.

The late freeze-up is at least partially explained by the slow accumulation of freezing degree days so far this winter.  As seen in the chart below, the freezing degree days so far this season are barely above the level of the past two winters, which were notoriously mild.  There have been other years with lower FDDs through this date, but the only other time 3 such mild freeze-up seasons occurred back to back was in 1979-1981.

Slow freeze-up this year was also noted in a recent article here about ice fishing conditions near Kotzebue.  The article states that the timing of freeze-up was "once relatively consistent" but has "become sporadic over the past decade or so".  I thought it would be interesting to see if there's any evidence of this in the history of early winter freezing degree days in Kotzebue; the chart below shows the annual number of FDDs in September, October, and November.  Note that I show a data point only if 2 or less days are missing in the month, and for any missing days I fill in the values with normal temperatures to avoid a bias.

It's plain to see that the October FDDs have decreased rather substantially since 2002, and the November FDDs were also very low last year and this year.  We would therefore expect that freeze-up has become later in Kotzebue since 2002, but I'm not sure we can say anything about changing variability of freeze-up dates.  If anything I would guess that freeze-up has been more consistently late recently, whereas in earlier decades it appears that it was sometimes late and sometimes very early.  A more sophisticated freeze-up model using daily temperatures might reveal more about the nature of the changes.

One other point of interest on the chart is the warmer conditions that are evident in October in the early decades, particularly the 1940s and early 1950s; a very substantial cooling shift occurred in the mid-1950s, and the next 45 years were generally colder than the decades on either end.

Out of curiosity I also created the charts for Barrow and Fairbanks - see below.  The recent warming trend is very striking in all 3 months in Barrow, with the 10-year means dropping well below anything observed before; and here we can certainly say that interannual variability has decreased.  This year and last year brought slightly higher FDDs in October in Barrow (as we noted here), but there's still some way to go before Barrow might see a "normal" October.

The chart for Fairbanks shows relatively little change over the long-term, although the period since 2002 has been relatively mild on average, and the last three freeze-up seasons have certainly been delayed in comparison to normal.

West-East Temperature Gradient

In recent weeks, the western half of interior Alaska has been colder relative to normal than the eastern half, and this is characteristic of El Niño winters.  Here is the average temperature anomaly since November 15 for five stations lying approximately west-east across the state:

Nome   -0.9°F
Kaltag   +0.3°F
Tanana   +0.8°F
Fairbanks   +5.8°F
Eagle AP   +8.0°F

According to Papineau's study in 2001, El Niño winter temperatures are typically "near normal in western Alaska but significantly warmer than normal for the eastern two-thirds of the state", so the recent pattern matches this nicely.  As El Niño won't die out any time soon, we might expect the general pattern to persist on average during the heart of winter.

The idea of west-east temperature gradients led me to wonder where the climate of Fairbanks falls in terms of normal temperatures along the 65th parallel.  The figure below shows one answer, taken from the CFS reanalysis; according to this data, Fairbanks winter temperatures are close to (but slightly above) average for the latitude.  The eastern North Atlantic is the epicenter of warmth at high northern latitudes, and of course eastern Russia is the pole of cold.  It's interesting to see the similar east-west gradient across North America and across Russia; with westerly mean flow, increased distance from the western oceans translates directly into lower temperatures.

ERA Reanalysis Precipitation

Returning to the subject of normal winter snowfall patterns across the interior, Peter Bieniek of UAF has very graciously provided the results of his downscaled ERA reanalysis for the period 1979-2013.  The ERA reanalysis is a product of the highly-regarded European Centre for Medium-Range Weather Forecasts (ECMWF) and is independent of the CFS reanalysis that I looked at earlier.  What the UAF group has done is to run a higher-resolution numerical model over an Alaska-centric domain while constraining the model by the (global) ERA reanalysis; the result is a higher resolution ("downscaled") historical analysis for Alaska.  It's a great idea and should be a very useful product.

The map below shows the results of the UAF/ERA reanalysis for mean November-March precipitation over the interior, just as in the earlier map for the CFSR data (also included below).  Note that the shaded boxes represent the native resolution of the model in the case of the CFSR, but I've interpolated to a finer grid in the case of the UAF/ERA data.  The actual resolution of the UAF model runs was 20km, compared to about 30km for CFSR.

It's interesting to note the dramatic differences between the two reanalysis products, which highlights the tremendous uncertainties in the output from these models.  The UAF/ERA precipitation shows very much less sensitivity to the relatively modest terrain features of interior Alaska, but it does show a pronounced maximum on the south slopes of the western Brooks Range.  The high precipitation from Kaltag and the southern Nulato Hills south along the Yukon River is one of the best points of agreement between the models, which supports the original idea that Kaltag is one of the snowiest lowland locations in the interior.

The upper Tanana River valley is strangely moist in the UAF/ERA data, as there is little sign of the precipitation minimum that we know is observed in that area.  The chart below shows the same point-by-point comparison that I showed earlier, but including the UAF/ERA results.  Northway and Gulkana are both much too wet in the UAF/ERA output, and consequently the overall correlation to the observed values for these 8 locations is lower than for the CFSR data.  My next step will be to assess the performance of the two models in terms of the interannual variability of winter precipitation at these locations.