Winter Storm Warning

The highly unusual September weather continues in Fairbanks, with a Winter Storm Warning now in effect for another round of snow, and potentially even heavier than Friday's dump.  The latest forecast calls for 6-12", which is really amazing for the time of year.

It probably goes without saying that the upper-level flow pattern associated with this storm is very unusual, but it's worth pointing out one aspect that seems particularly strange: heavy snow is expected in Fairbanks with a strong southerly flow aloft.  Assuming the snowfall pans out as expected, this is indeed rare.  Looking at historical data from 1948, I found 58 distinct heavy snowfall events that brought 8" or more of snowfall in association with 500 mb wind speeds of at least 30 knots on any of the surrounding days.  Only 3 of these events saw 30+ knot winds from a direction between 90° and 200° (i.e. easterly to just west of southerly): November 19-20, 1964, December 18-19, 1968, and October 16-17, 1974.  Southerly winds aloft typically bring strong down-sloping to the central interior of Alaska, which is very unfavorable for precipitation.  It will certainly be interesting to see if the computer models are correct in suggesting that it's different this time.

The series of maps below shows the predicted 500 mb height and wind fields at 12-hour increments beginning at 3pm today and ending at 3am on Wednesday.  The potent trough is already swinging down across western Alaska and will cut off south of the Kenai Peninsula, then pull northwards.

The 850 mb forecast shows colder air working its way in aloft tonight and becoming easily cold enough for snow in Fairbanks.  A striking frontal zone sets up across the eastern interior by tomorrow afternoon.

The sequence of 12-hourly precipitation accumulation maps clearly shows the zone of substantial snowfall associated with the frontal zone stretching northeast from the Kenai Peninsula to northeast Alaska.  It will certainly be interesting to see how this one plays out.

Update September 29: here's the 3am AKST 250 mb analysis from today.  The densely stippled areas show the highest jet speeds.  The right-entrance region and the left-exit region of the jet maxima are both favorable for deep tropospheric ascent; this zone extends from about Valdez to Fairbanks in this analysis.

Heavy September Snow - Will It Survive?

Fairbanks received an extraordinary 6.7" of snow yesterday, which puts this year firmly in second place for September snowfall through the 25th of the month.  Of course the epic September of 1992 still holds the record, with an amazing 24.4" of snow by the 20th.  If no more snow falls this month, the month will end up in 5th place for total September snowfall (1930-present).

This year's early snowfall is consistent with a modest long-term trend towards earlier autumn snow in Fairbanks.  From 1930-1970, the median date of first measurable snowfall was October 5, but since 1971 the median date has been September 29.  Furthermore, from 1930 to 1990, only 7 years out of 61 saw at least 1" of snow by September 27, but this has occurred 7 more times in the 25 years from 1991 to present.  The chart below shows annual snowfall totals for September 1-27.

At least part of the trend appears to be attributable to increasing precipitation and "storminess" in September.  From 1930 to 1990, the median September precipitation was 0.80", but this increased to 1.19" in 1991-2014.  In the earlier period, only 59 September days (3.2%) saw 0.25" of precipitation or more, compared to 41 days (5.5%) since 1991.  Given that the atmosphere is often cold enough for snow in Fairbanks by late September, an increased frequency of substantial precipitation events may be enough to account for the increased snowfall.

The maps below show MSLP and 500mb height anomaly maps for the two periods, according to reanalysis data.  Note the more unsettled pattern in the recent period.

	1948-1990	1991-2014
MSLP
500mb height

An interesting question now is, will the snow cover melt out in Fairbanks before winter arrives?  From one perspective, it seems very likely: measurable snow cover (1" or more) in September has ALWAYS melted out to a trace or less, except in 1992 (i.e. 13 of 14 times).  This includes 1972, when 8" on the 30th melted out by October 17.  We are, after all, a full 3 weeks ahead of the average date of establishment of the permanent winter snow cover (October 17) and the normal high temperature is still 48 °F.

On the other hand, when September snow cover occurs, it is usually only 1-2", but yesterday brought a snow depth of 5" - the second earliest on record.  The first occurrence of 5" snow depth has stuck around all winter in 82 of 85 years (including 1992), with the only exceptions being 1934, 1936, and 1972.  The chart below shows the frequency of melt-out after the initial occurrences of various snow depths.   So from this perspective the odds look better; but of course most years don't see 5" on the ground until much later (median date October 30).

All in all, I suspect there is at least a two-thirds chance that the snow cover melts out in Fairbanks itself, even if more snow occurs on Tuesday and Wednesday, as seems possible.  However, with 9" of snow reported just a few miles north of Fairbanks, it seems quite likely that elevated and/or sheltered locations not far from Fairbanks may just have initiated their winter snowpack.

Here's a post from last year on a similar topic.  Last year's October 5 snowpack of 4" did manage to survive, so 2014 actually tied for 3rd earliest arrival date of the permanent winter snowpack.

http://ak-wx.blogspot.com/2014/10/early-snowpack.html

[Update Sep 28: That was quick.  The latest snow depth observation from Fairbanks airport shows the snow cover has melted back to only a trace in just 2 days.  That's what steady rain and temperatures in the upper 30s will do.  But more snow appears to be on the way.]

September Temperatures

With valley-level snow in the forecast for tomorrow, it appears that a more wintry scene is about to descend on Fairbanks.  If the forecast verifies, this first measurable snow of the season will be about a week early compared to normal.  Two years ago the first snow fell on the 18th, but all other years in the past decade have seen snow arrive at a later date.

So far this month, Fairbanks is slightly more than half a standard deviation below normal in terms of average daily mean temperature (through the 23rd).  September 2013 was a little cooler through this date, but ended only 0.4 SD below normal, and every other September since 2005 has been warmer than normal.  This makes September the most persistently warm month of the year in the last decade (in Fairbanks, and compared to normal of course).

I thought it would be interesting to look at the global sea surface temperature (SST) patterns that most often precede cool Septembers in Fairbanks.  To do this, I took the 10 coolest Septembers since 1950 and calculated the frequency of above-normal SSTs in the preceding August at each grid point, using the ERSSTv4 data set.  See the results below.

Cool Septembers tend to be preceded by warmer than average waters over the eastern North Pacific.  The reverse analysis for the 10 warmest Septembers, shown below, indicates that - counterintuitively - a cool eastern Pacific is more favorable for a warm September in Fairbanks.  Other regions where the SST patterns are generally opposite between cool and warm Septembers are in the western subtropical Pacific (southwest of Hawaii) and in the eastern North Atlantic, where August SSTs seem to be positively correlated with September temperatures in Fairbanks.

The SST analysis from this August (below) shows tremendously widespread warmth in the eastern tropical and North Pacific.  The warm conditions from the Gulf of Alaska to Baja California are certainly more closely aligned with the pattern that favors a cool September, which matches what we are seeing this month.

NOAA's Climate Prediction Center, taking into account many different predictors, expected a higher likelihood of a warm September (see below, top left panel).  In view of the warm waters surrounding Alaska, it would have been difficult to anticipate a cool month, but the SST analysis above seems to provide a hint that perhaps unusual warmth was not as likely as we might have thought.

Freeze-Up Begins

Some smaller fresh water bodies are freezing up at relatively low elevation across the interior and north.  Here are a few webcam shots from today of small lakes near the much larger Teshekpuk Lake, near the Arctic coast:

Teshekpuk Lake itself only shows a small margin of ice along the shore:

Some shoreline ice is evident at Inigok:

Toolik Lake is not yet frozen over, despite air temperatures hovering around 15-20°F for the past couple of days; the lake temperature remains above +3°C.  The Colville River at Umiat is still open:

Finally, there was ice on ponds at about 900' elevation in the upper Chena River valley late Sunday afternoon.  With the forecast remaining on the cold side, it seems quite likely this may not melt until spring.

[Update Sep 24: Toolik Field Station and Galbraith Lake both reached 0°F yesterday morning, the 23rd.  Ivotuk dropped to 1°F.  Additional information from NWS Fairbanks indicates that on Monday and Tuesday "north of Finger Mtn (97 Mile Dalton) the lakes are mostly frozen over. At Wiseman, Wiseman Creek was mostly ice covered and there was pan ice on the Koyukuk. The coop observer at Wiseman, who has lived there her whole life, thought freeze-up was early, but not extremely so."]

Snow Remaining at Elevation

Higher elevations in the interior now have a modest but unequivocal snow cover as a result of the chilly, wet conditions of the past week.  I was able to hike most of the way up Wickersham Dome this afternoon, but found myself wishing for snowshoes and eventually gave up on post-holing through a good 12" of snow, and more in places.  There were some patches of snow at 1900' elevation, a more-or-less continuous cover above about 2500', and then the depth increased rapidly above 2700'.  Interestingly very light snow was falling over the hill, although the cloud base was fairly high and visibility was good.

Including today, Fairbanks has remained below 50°F for 7 consecutive days, which is rather unusual for this early in the season; it happened before in 1935, 1987, and of course 1992.  Is it possible that Fairbanks won't reach 50°F again this year?  Yes, it's possible, though unlikely.  It probably won't get there in the next week, and the CPC 8-14 day forecast is decidedly cool.  However, there's still about a 65% chance of reaching 50°F in October or later.

Here is CPC's 6-10 day forecast:

and the 8-14 day forecast is below.  It seems like a long time since decidedly cool weather has persisted over Alaska.

Snow Cover, Chilly Forecast

Incessant rain has melted off yesterday's 4" of snow from Keystone Ridge, but other more northerly locations are looking wintry today, including Arctic Village and the central Brooks Range above about 2000'.  Based on the forecast, it seems this may be the permanent winter snow cover at middle to upper elevations of the Brooks Range; but presumably that is not particularly unusual.  Here are a few webcam shots from earlier today.

Arctic Village

Coldfoot

Chandalar Shelf

The medium-range forecast is showing a fresh dose of cold air moving into northern and interior Alaska over the weekend, so it appears the cool spell will persist.  The models have steadily trended colder in the past several days, so it will be interesting to see just how chilly it gets.  Here's an example of how the models have trended colder: the chart below shows the forecast 850mb temperature for a grid point near Fairbanks, from the last 4 runs of the ECMWF deterministic model.  The earliest run (Saturday afternoon's run) is in blue and the most recent (this morning's) is in purple.  To put this in perspective, the normal 850mb  temperature at Fairbanks drops below 0°C on the 20th.  In 1992, the 850mb temperature reached -11°C on the 10th, -12°C on the 15th, and -14°C on the 22nd.

Early Wet Snow

Snow mixed with rain was reported at Fairbanks airport today for a period of slightly less than one hour.  This is the earliest snowfall reported in the hourly (or sub-hourly) observations since 1992, the year of the great September snowfall and cold snap (when the permanent winter snowpack was established on the 13th, and 10" was on the ground by the 15th).

I should mention, though, that the official climate history contains a trace of snow on earlier dates on a number of occasions in more recent years, presumably due to a passing flurry or a few stray flakes in a rain shower.  We also have the 0.1" of snow that was reported on August 25, 1995, although the daily minimum temperature was 44°F - I'm not sure what happened there.

Accumulating snow was observed today at the higher elevations near Fairbanks, such as on Ester Dome (2400') and Cleary Summit on the Steese Highway (2233') - thanks to Brian for pointing out the latter.  Here are some webcam shots.

Ester Dome:

Cleary Summit:

The chilly conditions are a change from what has been a relatively mild month so far (after the cold spell of late August).  Soon, of course, the term "relatively mild" will take on a different meaning, as the peak rate of seasonal cooling is only a month away.

Update 10am Sep 13: There's a report of 3" of new snow this morning on Keystone Ridge by 8:45am, with heavy snow continuing.  Here's a photo, courtesy of Heather Hrawiec:

ENSO Height Patterns

I've been very busy lately with several research projects having varying degrees of relevance to Alaska, but came across some results that may be of interest to some here.  We recently discussed the influence of El Niño on Alaska winter climate, and the maps below give additional insight on this topic.  The 4-panel figure below (click to enlarge) shows the following:

- top left: mean 500mb height during the top 10 El Niño (positive ENSO) winters since 1950
- bottom left: mean 500mb height anomaly, i.e. difference from normal, for the same winters
- top right: variance of the daily 500mb height anomalies
- bottom right: variance anomaly, i.e. difference from normal

The bottom left panel shows that El Niño winters tend to bring below-normal heights (troughing) over the Bering Sea, Aleutians, and northeast Pacific, and above-normal heights over Canada.  This is a relatively stable pattern as far as Alaska is concerned, as the variance is lower than normal from the Bering Sea across Alaska to western Canada.

The La Niña patterns are opposite in some respects, as we would expect - see below.  The climatologically favored trough over the Bering Sea is greatly weakened and heights tend to be above normal, especially over and south of the Aleutians.  Also, the variance is enhanced over the north Pacific and Alaska, and especially over the southern Bering Sea and Aleutian chain.  This means that the flow pattern is unstable and flips around more than normal.

The difference between El Niño and La Niña winters is greatest near the southernmost Aleutian islands around the date line.  Consequently I pulled out the daily height anomalies at 50°N 180°W and plotted the frequency distribution during the top 10 El Niño and La Niña winters - see below.  It's very interesting to see that even though the mean 500mb height is above normal in La Niña winters, the most common occurrence is for the height to be below normal.  In other words, the distribution is highly skewed; it's also positively skewed for El Niño.  This is something I did not expect.

In terms of impacts on Alaska, a key feature is the elevated frequency of very high heights during La Niña winters; this means that strong upper-level ridging (blocking) is much more common over the Bering Sea during La Niña, and it is these events that bring unusual cold to much of Alaska.

Below are corresponding maps for the top and bottom 10 PDO winters since 1950.  There's much that could be said here, but I'll just point out that the positive PDO phase amplifies the ridge over western Canada much more than El Niño does, and that is why the positive PDO is more reliably connected to warm winter conditions in interior Alaska.

Winter Weather Dates

Here's a chart showing the temporal evolution of onset and ending dates for a few metrics of cold and snow conditions in Fairbanks.  Reader Eric suggested doing this for various seasonal metrics throughout the entire year, but I thought I'd start with just cold and snow conditions; there's enough to focus on here for now.

A couple of interesting features stand out:

- The first freeze has become considerably later over time (no August freeze since 1996), but the first measurable snowfall has become earlier, and the establishment of the permanent winter snow cover has also become slightly earlier.  Prior to about 1970 there was usually a lengthy period between the arrival of frost and the arrival of snow, but these days the transition is relatively compressed.  The slightly earlier arrival of snow cover is consistent with a long-term cooling trend in Fairbanks in autumn: most of October and early November have become colder since 1930 (in contrast to the rest of the year).

- Late winter and spring have become warmer over time, with fewer instances of late freezes and late 0°F occurrences, but it does not appear that the melt-out date has become earlier.  Consequently the continuously snow-covered season has become slightly longer.

Announcement and Invitation

Regular readers who reside in Fairbanks may be interested to know that I will be giving a talk at UAF on Monday September 21 (11 am); all are very welcome to attend.  I'll be discussing the relationship between the North Pacific "blob" of very warm SSTs and Alaska winter climate.  Here's the announcement page, including a rather outdated photo of yours truly:

https://accap.uaf.edu/?q=The_BLOB

I'll be staying in Fairbanks for several days and would also like to set up an informal gathering for blog readers - if there's interest from others.  I'd be grateful for any suggestions of a suitable establishment where we might enjoy some conversation and perhaps a bite to eat.  I imagine early evening would work best; I'll be available from Sep 18-21 inclusive.  Significant others and/or children would be welcome.  Feel free to leave a comment below, or email me at richard [dot] james [at] prescientweather.com.

First Freeze

The Fairbanks airport observing site saw its first freeze this morning, with a minimum temperature of 30°F.  Temperatures were at or below freezing for at least 4 hours.  This is about a week ahead of the long-term normal for first freeze, and about 2 weeks ahead of the normal of the past 20 years.  The chart below shows the long-term history.  It's nice to see that the era of late freezes, noted by Rick in the early days of this blog, appears to be over for now.

Here's a lovely shot from one of the Fairbanks air quality webcams this morning (click to enlarge).

August Chill

Meteorological summer is ending with a nip in the air across interior Alaska, with hard freezes and some snowfall in certain locations.  The temperature dipped to 20°F this morning at the Norutak Lake RAWS (elev 800'), between Bettles and Ambler; it was 25°F at the Bettles airport.  Closer to Fairbanks, it was 23°F at the Livengood RAWS, and many locations on the north side of Fairbanks dipped just below freezing.  The airport made it to 33°F.  Here is what was reported at the UAF ACRC weather station this morning:

As for snow, light amounts were observed even at some valley locations, such as Delta Junction.  The images below were taken yesterday morning at the Delta Junction FAA webcam; the airport reported light snow for more than 2 hours, with visibility down to 3/4 mile for a time, and a temperature of 32°F.  It's an interesting coincidence that snow fell in both June and August this year in Delta Junction.  However, any accumulation yesterday was probably no more than a trace; the earliest recorded measurable snowfall occurred on September 2, 1983.