Monday, August 22, 2016

No Freeze Yet

None of the usually colder reporting sites around the Fairbanks area has seen a freeze yet this month, and this is unusual compared to recent years.  The normally chilly Goldstream Creek and Ester 5NE (Goldstream Valley Bottom) COOP sites came close a few days ago, with both sites reaching 33°F, but the closest site to Fairbanks to hit 32°F so far is a COOP site 20 miles southeast of Delta Junction.

Looking back at the past 10 years, here are the coldest temperatures reported by August 22 in the general vicinity of Fairbanks:

Aug 9, 2006   29°F   Ester 5NE COOP
Aug 10, 2007   30°F   Fairbanks RAWS
Aug 12, 2008   28°F   North Pole COOP and Ester 5NE COOP
Aug 20, 2009   23°F   Fairbanks RAWS
Aug 22, 2010   26°F   Fairbanks RAWS
Aug 6, 2011   30°F   Mile 42 Steese Highway COOP
Aug 22, 2012   25°F   Ester 5NE COOP
Aug 22, 2013   30°F   Mile 42 Steese Highway COOP
Aug 22, 2014   27°F   Ester 5NE COOP
Aug 15, 2015   31°F   Goldstream Creek COOP

The last time the first freeze report came in later than August 22 was in 2004, when it occurred on the 25th.  That year also saw Fairbanks airport stay at or above 45°F until the 28th, the latest on record; but if current forecasts are to be believed, that record may be in danger this year.

The chart below shows daily low temperatures at Ester 5NE this summer.  In the 18 years that this COOP site has been operational, freezes have been recorded even in the height of summer - for example:

31°F   June 25, 2009
31°F   July 4, 2002
31°F   July 9, 2012
31°F   July 20, 2011
30°F   July 21, 2014

Consider this as a metric of how persistently mild it's been this summer: the low temperature at Ester 5NE stayed at or above 40°F for 54 consecutive days this summer; this is more than double the previous record of 25 days, set in 2005.

Thursday, August 18, 2016

Drier and Warmer

Fairbanks has been enjoying a spell of delightful late summer weather recently, which I'm sure is most welcome after the very wet weather of June and July.  The past week included the longest dry spell since mid-May, and 14 of the past 15 days (including today) have reached 70°F; but it hasn't been excessively warm, as the temperature hasn't reached 80°F since the mid-July heat wave.

It's interesting to note that the average daily high temperature so far this month is 71.9°F, which is higher than last month's average: so far, August days have been warmer than July days, which is unusual.  If the anomaly continues and August ends warmer than July for high temperatures, it will be only the 4th time this has happened in Fairbanks history (1930-present) - and the previous occasions all followed very cool July's, unlike this year.  But note that August has never been warmer than July in terms of overall average (daily mean) temperature, and it's very unlikely to happen this year.

The warmth in Fairbanks this month is consistent with, and merely a small part of, the ongoing exceptional warm anomaly that has affected the state of Alaska all year.  The chart below shows the daily average of temperature anomalies during the last year for 25 stations scattered around the state, from Annette Island to St Paul Island and across the interior up to the Arctic coast (credit to Rick Thoman for devising this kind of analysis).  According to my calculations, the statewide index has been below normal only 2 days so far this year, and that by only the smallest of margins; the persistence of the warm anomaly is quite astonishing.

Monday, August 15, 2016

Strong Arctic Storm

A very strong low pressure system developed over the Arctic Ocean north of Siberia yesterday and intensified last night as it migrated eastward to the date line.  This morning the Canadian surface analysis estimated the minimum central pressure at 969mb, which is very low for the time of year.  The sequence of images below shows some analyses from the past 36 hours.

3am AKST yesterday:

3pm AKST yesterday:

9am AKST today:

The current position and intensity of the storm is quite reminiscent of the 2012 "Arctic hurricane" (documented here on this blog), although today's storm tracked from the west rather than the south, and it's not quite as strong as the 2012 storm.  The higher overall latitude of this storm may mean that the associated high winds do less damage to Arctic ice than the 2012 event; the 2012 storm is believed to have been one of the factors that led to that year's record low ice extent.  Today's chart of sea ice extent from the NSIDC conveniently shows the comparison of recent conditions to 2012 - see below, and note the dip in early August 2012.  This was probably at least partly caused by the storm, and so we may well see a similar sudden dip in sea ice extent in the next few days.

The current storm and that of August 2012 appear to be consistent with a long-term trend towards stronger storms during summer over the portion of the Arctic north of Alaska; see the chart below, created from the NCEP/NCAR global reanalysis.  One caveat here is that the reanalysis system ingests oceanic weather reports, so it's possible that the paucity of observations over the Arctic in earlier years prevented the reanalysis from fully capturing all the storms that actually occurred; however, the model is very capable of transporting information from data-rich to data-sparse areas, so this may not be a significant issue.  Here's an NSIDC article giving some additional perspective:

[Update August 16] Here's a chart of observed sea-level pressure at a buoy that happened to be located very close indeed to the track of the storm center.  If the measurement is accurate, then the storm was a little stronger than estimated by Environment Canada.

Saturday, August 13, 2016

Colville Delta Dryness

Reader Tracy recently asked about conditions around the Colville River delta of the North Slope this summer in light of very dry ground conditions that she observed there during July.  How is this best explained - dry weather, warm weather, lack of winter snowfall, or something else?

The first thing to note is that precipitation has not been unusually low this summer; the Colville Village COOP reported 1.81" of precipitation in June and July, compared to a 19-year median of 1.44".  Similarly, the Kuparuk COOP observer measured 1.91" of precipitation compared to a 31-year median of 1.20" for the June-July period.  Nuiqsut airport reported 1.84", their second highest June-July total on record (17-year median of 0.88").

Looking at temperatures, the June-July period was slightly but not dramatically warmer than normal, except for the heat wave in mid-July.  Overall Colville Village was 1.5°F above normal for the two months, and Kuparuk was 1.3°F above normal; three of the last four years were warmer than 2016 over this period at Colville Village.  However, if we look back to May we find very unusual warmth, which contributed to a very early melt-off of the winter snow cover.  Colville Village reported zero snow depth (i.e. no snow over more than 50% of the ground area) on May 14, the earliest date on record (data from 1997-present).  Kuparuk reported melt-off on May 12, also very much earlier than normal (but not a record).  For the rest of May after melt-off, temperatures were persistently above normal in both locations.

The chart below shows that melt-off typically occurred in early June prior to last year at Colville Village (at least since 1997), so this year's early melt added 2-3 weeks of warming and drying of the ground surface in early summer.

At first blush the early melt-off seems to be significant relative to a short summer season: by July 15 this year the area had been snow-free for 9 weeks, compared to around 6 weeks in a normal year.  With evaporation taking place for 50% longer, and with the potential for 24-hour sunshine at this time of year, ground moisture levels might be expected to take a hit relative to normal.  However, temperatures were still low in absolute terms in the second half of May, and it is a cloudy time of year, so total evaporation may have been only slightly enhanced compared to a normal summer.

I was hoping to dig into the detailed evaporation calculation using data from Barrow's CRN site, but unfortunately the NCEI website is not working at present.  In lieu of a full calculation for Barrow, I computed the daily average vapor pressure deficit from Nuiqsut airport, which in any case is much closer to the Colville delta.  The chart below shows the sum of daily average vapor pressure deficit values from the date of melt-off to July 31 in each year since 1999.  Recall that the vapor pressure deficit is directly proportional to the evaporation rate, all else being equal (e.g. wind and solar radiation).  The results indicate that evaporation this year was probably higher than the long-term average but may have been no higher than 2015, and was perhaps lower than 2015.  It's clear that the evaporation is more closely tied to summer temperatures and humidity (and no doubt cloudiness and wind) than to melt-off date, because it's just too cool in May to accomplish much evaporation regardless of whether the ground is snow covered or not.

Looking again at the melt-off chart, another point of interest is that the peak snow depth in winter 2015-2016 was only 9" at Colville Village, which is the lowest in the 20-year history.  It's not a large difference in terms of absolute water content of the snowpack, but the small snow deficit may have contributed to reduced ground moisture levels this summer.  In conclusion, I'd say the dry ground conditions observed by Tracy were caused by a combination of factors, including at least: a slight lack of snowpack last winter, an early melt-off allowing for early warming and drying of the ground, and slightly unusual warmth (and one heat wave) this summer.  It would be interesting to hear others' perspectives; and the faithful climate observers at Colville Village would probably have some useful insight as well.

Here is a parallel melt-off chart for Kuparuk with its longer observing history.  Last winter's snow depth was low there too compared to recent years, although there is a pronounced long-term trend towards rising winter snow depth; this may or may not be real, as it could be related to changing measurement practices or variations in measurement location.

Wednesday, August 10, 2016

Statewide Thaw

The exceptionally persistent and unusual warmth across Alaska in recent months and years (since mid-2013) has been well-publicized on many fronts, and deservedly so.  Here's another angle on the amazing warmth that I haven't seen discussed elsewhere (although perhaps I missed it): it is now very likely that Alaska's statewide 12-month running average temperature will rise above freezing for the first time on record within the next 2 months.  This is according to NOAA's climate division data for Alaska.

We're already very close to the 32°F threshold, as the 12-month average was 31.9°F for periods ending in both April and July of this year.  Last year both August and September were cooler than normal statewide, so if this year continues warm or even normal, it will be easy to bump the 12-month average above freezing.  For example, an August-September mean temperature of 44.65°F would be needed to push the 12-month average up to 32°F, and only 7 out of the past 35 years were cooler than this (most recently last year).

The charts below illustrate the likely range of 12-month running mean temperatures through the end of this year by appending the 1981-2015 distribution of multi-month means to the preceding averages.  Note that anything except continued very unusual warmth will bring the 12-month running average back down below freezing by early winter; it's rather unlikely that the annual mean temperatures will remain above freezing for more than a month or two.  After all, the 1981-2010 normal is 27.2°F, and presumably more normal temperatures will return, at least for a spell, sooner or later.

Monday, August 8, 2016

Radar-Estimated Rainfall

In view of the extraordinary rainfall in the Fairbanks area this summer, I thought it would be interesting to look at radar data to get a better idea of the spatial distribution of the rainfall amounts in June and July.  Using a standard relationship between radar reflectivity and rain rate, I calculated daily rainfall estimates based on data from the Pedro Dome radar site just north of Fairbanks; the figure below shows the June-July estimated totals (click to enlarge).  Observed amounts at surface measuring sites are marked in black.

Here's a comparison of estimated and observed total precipitation by location:

Big Delta AP: 5.97" observed vs 6.22" estimated
Delta 6N COOP: 6.79" vs 6.78"
Nenana AP: 7.39" vs 9.53"
Fairbanks AP: 8.26" vs 6.95"
Eagle Summit SNOTEL: 8.50" vs 6.38"
Clear Sky COOP: 9.30" vs 9.28"
North Pole COOP: 9.77" vs  7.71"
Teuchet Creek SNOTEL: 10.10" vs 11.26"
Ft Knox Mine COOP: 10.94" vs 8.22"
Keystone Ridge COOP: 12.24" vs 8.08"
Mt Ryan SNOTEL: 12.50" vs 10.35"
Monument Creek SNOTEL: 13.30" vs 9.99"
Little Chena Ridge SNOTEL: 14.30" vs 11.92"
Upper Nome Creek SNOTEL: 16.60" vs 11.70"
Munson Ridge SNOTEL: 19.00" vs 18.56"

On average for the 15 stations, the radar algorithm underestimated the total precipitation by 11%, which is not too bad.  In a few spots the radar estimates were excellent, and in a few spots they were notably bad; for example, Keystone Ridge apparently observed 52% more rainfall than the radar indicated.  Localized differences like this might be related to local topographic enhancement or diminution of rainfall compared to what is estimated from radar reflectivity at the height of the radar beam (which increases with distance from the radar site).  It's also possible that the surface measurements are incorrect at some of the observing sites.  The charts below show comparisons of the daily rainfall amounts at Fairbanks airport and Munson Ridge SNOTEL; both of these sites report precipitation for the midnight-to-midnight period, which matches the period that I used for the radar calculations.

Regardless of local discrepancies and potential errors, the main point of the exercise is to get a rough look at the spatial distribution of rainfall during June and July, and the radar estimate serves this purpose quite well.  Based on the SNOTEL data, we already knew there was a broad area of enhanced rainfall in the hills east and northeast of Fairbanks, but the increased rainfall between Fairbanks and Nenana is interesting.  The radar estimate suggests that it was also very wet on the south side of the Tanana valley but north of the higher terrain of the Alaska Range.

Thursday, August 4, 2016

Warm Nights

The temperature dipped below 50°F this morning at Fairbanks airport, which is the first time since June 23.  The 41 consecutive days at or above 50°F is easily a record for Fairbanks; the previous record was 32 days (ending July 26, 1975; July 1975 was the warmest month in Fairbanks history).  Given that the normal low temperature peaks at only 53°F (on about July 6), the absence of any chill in the air during the past 6 weeks is remarkable.

The chart below shows the daily maximum and minimum temperatures since May 1 compared to the 1981-2010 normals.  While daily high temperatures have more often than not been below normal since mid-May, mostly because of all the cloud and rain, low temperatures have been unusually warm.  The average diurnal range so far this summer (since June 1) is only 17.4°F, which is the second lowest on record; only 2014 - which was similarly wet - had a smaller mean diurnal range through this date.

The reduced diurnal range this summer is consistent with the long-term trend, as summer daily low temperatures have experienced 3 times as much warming as summer daily high temperatures in the past 85 years in Fairbanks.  However, the chart below shows that most of the warming in overnight minima seems to have occurred between about 1955 and 1975; this preceded the 1976 PDO shift, so we can't attribute the change to the PDO.  Furthermore, precipitation did not increase concurrently with nighttime temperatures, and in fact the long-term trend in summer precipitation is slightly negative, so increased cloudiness may not be a contributing factor.  A more likely explanation may be population growth and urbanization, which was rapid from the 1950s onward.

It would be interesting to look into these trends a bit more with the help of data from other stations and also with upper-air data from Fairbanks; I'll plan to take this up in a subsequent post.  For now, here's a related post from 3 years ago that I had all but forgotten about until now.