Warming Trend

The recent two-week spell of unseasonable chill has ended in Fairbanks, with temperatures finally getting back above 70 °F the past three days.  As reader Gary pointed out, the Climate Prediction Center is expecting a warmer than normal August, which I'm sure would be most welcome for residents of the Interior.  I thought it would be interesting to look at the statistical chance of a warm August based on the observed wetter than normal conditions in June and July.  We went through a similar exercise last month based on having a wet June, and suggested then that "unusual warmth appears to be a more likely outcome for the late summer".

The scatterplot below is the same as the first figure in last month's post, but now I've colored the points based on the August mean temperature; the colors correspond to the four quartiles of the 1949-2013 distribution for August temperature.  The slight correlation between June and July precipitation was noted last month.  The most notable feature of the color distribution seems to be that a cool August is favored when a dry June gives way to a wet July (upper left quadrant).  This year we fall in the upper right quadrant, and here we see a slight tendency for warmer than normal conditions in August; 10 of 16 years in the quadrant had a warmer than median August, and 5 of 16 were in the top quartile.  The two exceptions with high June rainfall were 1949 and 1955, but these were negative PDO summers.  This fact suggests that we should divide out the negative and positive PDO years - see the next two scatterplots for this breakdown.

The scatterplot for the positive PDO phase (as we are seeing this year) shows a possibly more robust indication of warmth in August, with 6 of 8 years seeing a warm August after a wet June and July.  It's also interesting that the aforementioned flip from dry June to wet July seems to be more common in negative PDO years.

In summary, wet June and July conditions are, perhaps counter-intuitively, associated with a higher than normal chance of a warm August in Fairbanks, especially when a positive PDO phase is observed.  This analysis agrees with the CPC outlook.

Fairbanks Precipitation Ranking

The forecast for the remaining two days of July in Fairbanks does not include any mention of rain. Therefore, I feel (reasonably) comfortable adding a preliminary rainfall ranking table.

Through July 29th, Fairbanks is approaching their annual normal precipitation of 10.81". When comparing similar monthly intervals from different years, only one January to July period was wetter (1929). In fact, every monthly grouping (ending in July) is ranked in the top 3. The May to July and June to July are far and away the wettest such periods on record in Fairbanks.

Should these numbers change in the next two days, this post will be updated.

Arctic Chill

Cool weather has continued lately in Barrow (see chart below), and it now looks likely that the June-July period there will end up as the coolest in more than 30 years.  The unusual atmospheric circulation is responsible, with persistent troughing over the Chukchi Sea.

Based on the NWS forecast for the remainder of July, the mean June-July temperature in Barrow should end up near 36.0 °F, which would be the coolest since 1982 - and about equal to the 1951-1980 normal.

 Here's the mean 500 mb height from June 1 to July 24:

And the 500 mb height anomaly; with this pattern, a cool, damp summer in interior and northern Alaska is inevitable.

Precipitable Water

As most readers of this blog know, precipitable water (PW) is a measure of the amount of moisture that is contained in a column of air extending from the surface to about 300 millibars. It also represents a measure of the volume of water that can contribute to a precipitation event. The PW amount is calculated from upper air balloon soundings released from each of fourteen sites around the state. It stands to reason that more moisture in the atmosphere will lead to more precipitation. In fact, anomalously high PW values are frequently noted in Area Forecast Discussions as indicators of heavy rain potential. So, let's put this to the test.

Since this June-July has been the wettest on record for Fairbanks (by far) I wanted to see if a correlation existed between the average daily PW value and the precipitation totals during this two month period. Using a June-July time period also has the advantage of eliminating snow events and it is also the season of peak thunderstorm activity.

When looking at it on an annual scale (June-July constituting a single year) there is no correlation between average PW over the 61-day period and the total rainfall over the same period (see Figure 1). For Fairbanks, this June-July is already the wettest on record but the average daily PW has been unremarkable.

Figure 1. Average June-July PW (orange) and total June-July precipitation (green) in Fairbanks between 1971 and 2014.

When we group days by the amount of precipitation that fell and calculate an average of those days, a strong correlation exists (see Figure 2). Days with no precipitation or only a Trace have substantially lower PW than those days with over 0.50" of precipitation. The three days this month with over 1.00" of rain in Fairbanks had a PW average of 1.16".

The apparent discrepancy between the two charts is primarily a result of the number of days with precipitation not taken into account in the first graph. If most of the rain falls in 3 or 4 days, the PW from those days is averaged with the PW from the other 57 or 58 days and thus blends in with the other values.

Figure 2. Average June-July PW value grouped by observed precipitation amounts in Fairbanks between 1971 and 2014.

Note: Here is a site ( http://www.crh.noaa.gov/unr/?n=pw ) that has PW climatology for all upper air stations in the U.S.

Still Wet in Bettles

On Monday I noted that Bettles had seen measurable rainfall on 13 of the past 14 days, and including today's heavy rain showers this is now up to 17 of the past 18 days.  Rain has fallen on the past 14 days in a row, which is close to record territory; the warm-season record for consecutive days of measurable precipitation is 16 days (ending September 8, 2012).  The most remarkable aspect of the pattern may be the frequency with which the rain has been relatively heavy; 17 days since June 1st have seen rainfall of at least 0.25 inches, compared to a record of 11 days in the period June 1 - July 23, 1984.

The chart below shows the daily precipitation and high and low temperatures since May 1 in Bettles; it hasn't been much of a summer.  The lack of warm nights is quite notable: only 5 days have had daily minima above 50 °F, which is just about a record low pace through this date.

It goes without saying that all of this is an extraordinary turn-around from last summer's heat and dryness.

Brooks Range Snow Forecast

A significant snow event is about to get under way in the Brooks Range; the following statements were issued by the NWS in Fairbanks this morning:

AKZ206-240000-
NORTHEASTERN BROOKS RANGE-
INCLUDING...ANAKTUVUK PASS...ATIGUN PASS...GALBRAITH LAKE... SAGWON...FRANKLIN BLUFFS
428 AM AKDT WED JUL 23 2014

...SIGNIFICANT SNOWFALL THURSDAY THROUGH FRIDAY...

SNOW WILL DEVELOP THURSDAY ABOVE 3000 FT AND LOWER TO 1000 FT THURSDAY NIGHT. SNOW WILL CONTINUE THROUGH FRIDAY. STORM TOTAL SNOW AMOUNTS OF 3 TO 6 INCHES ARE POSSIBLE WITH THE HIGHEST AMOUNTS ABOVE 3000 FT.



AKZ218-240000-
SOUTHEASTERN BROOKS RANGE-
INCLUDING...ARCTIC VILLAGE...WISEMAN...COLDFOOT... CHANDALAR D.O.T. CAMP...INIAKUK LAKE
432 AM AKDT WED JUL 23 2014

...SIGNIFICANT SNOWFALL THURSDAY EVENING THROUGH THROUGH FRIDAY FROM THE PIPELINE WEST...

FROM THE PIPELINE WEST...SNOW WILL DEVELOP THURSDAY AFTERNOON ABOVE 5000 FT AND LOWER TO 2000 FT THURSDAY NIGHT. SNOW WILL CONTINUE THROUGH FRIDAY. STORM TOTAL SNOW AMOUNTS OF 2 TO 5 INCHES ARE POSSIBLE WITH THE HIGHEST AMOUNTS ABOVE 3000 FT. THIS COULD MAKE TRAVEL DIFFICULT OVER ATIGUN PASS THURSDAY NIGHT.


Snow has already been falling in the western Brooks Range, as evidenced by the Red Dog FAA webcam photo from earlier today (elevation 962' MSL):

I took a quick look at the latest high-resolution (6km) NAM model forecast for the area, and indeed the model indicates substantial snowfall in the higher elevations over the next two days.  Below is a map of the estimated total snow accumulation through Friday evening (note: the model predicts snow accumulation in liquid equivalent, and I used a 10-1 ratio to estimate total depth).  I added markers for some of the FAA webcam locations, including Anaktuvuk Pass (red), Chandalar Shelf DOT (blue), Coldfoot (green), and Arctic Village (turquoise).

For historical context, here's a map that Brian put together a few days ago, showing all Alaska observing locations that have reported measurable snow accumulations in July (per GHCN data).  The Brooks Range location with 2.5" is Chandalar Lake, and that event occurred on July 17, 2003.

Precipitation Percentiles

A few weeks ago Rick started a discussion (via e-mail) regarding precipitation percentiles. Since temperatures are pretty normally distributed, the terms 'above normal' and 'below normal' are universal when the time scale is a single day. However, precipitation is an entirely different matter. If a station receives precipitation on 20% of days throughout the year, then 80% of days will experience below normal precipitation. From a statistical perspective this is counter intuitive. Using a probabilistic measure provides a meaningful baseline from which to compare a single day's precipitation to determine whether or not it is a high or low probability event.

Figure 1 shows the daily precipitation values for the 75th, 80th, 85th, 90th, 95th, and 99th percentiles for Fairbanks between 1915 and 2013. Measurable precipitation values for 2014 are overlaid as blue diamonds. The daily values were interpolated from a spline of monthly values. Importantly, all values were included when developing the percentile lines – including days with no precipitation.

For example, on March 1st, the 99th precipitation percentile is 0.25". In other words, there is a 1% chance of observing a precipitation value greater than 0.25" on March 1st. Since there are 365 days in a year, we expect to exceed the 99th percentile 3.65 times annually. Figure 2 shows the number of exceedences of the 99th percentile for every year since 1915. So far in 2014 the 99th percentile has been exceeded 6 times. In fact, those 6 exceedences occurred in a span of 20 days. The only time that occurred in a shorter time period was back in March of 1918 when 6 exceedences occurred in a span of 11 days.

Figure 1. Precipitation percentiles for Fairbanks (1915-2013) with 2014 precipitation shown as diamonds.

Figure 2. Number of days that the 99th percentile was exceeded for Fairbanks (1915-2014). 2014 data is through July 20th. 

Cool Summer Days

Summer 2014 is turning out to be cooler than normal in interior Alaska, as highlighted by reader Gary who mentioned yesterday that for the first time in many years it's been necessary to run the heating stoves in July in Fairbanks.  The temperature anomaly is mostly seen in the daily high temperatures, which have been more than 3 °F below normal on average since June 1.  In contrast, the low temperatures have actually been above normal on average; the compressed diurnal range is directly related to the abundant cloud cover and precipitation.

Below are charts showing the daily high temperatures compared to normal at Fairbanks and Keystone Ridge, and also at Bettles where they have been having a very wet time of it.  Not only is it the wettest summer-to-date on record at Bettles, they have had measurable precipitation on 34 out of 50 days since June 1, which is also a record.  The last two weeks have been particularly bad, with measurable rainfall on 13 of 14 days and with daytime temperatures dropping to levels more typical of early September.

Here's another way of looking at the distribution of high temperatures since June 1: the charts below show the counts of days exceeding a range of threshold temperatures.  We can see that the frequency has been especially lacking for days on the warm side, with (for example) only 6 days reaching 77 °F at Fairbanks, compared to a normal of 15 days.  Up on Keystone Ridge, only 2 days have reached 75 °F, compared to a normal of 11 days.  The anomalies appear particularly large on Keystone Ridge, but part of this could be due to the station history beginning in August 1996; the "normal" period only includes the (presumably) warmer half of the 1981-2010 period, so current anomalies will look cooler relative to normal than for stations with longer histories.

Update July 22: reader Eric asked about corresponding charts for last summer and last winter.  Here they are for the standard three-month seasons.  Summer 2013 was really excessive in terms of the number of days above 80 and 85 °F.

Barrow Upper-Air Climatology

As a companion to yesterday's post, here are the 1981-2010 normals for the temperature aloft at Barrow, as observed by the twice-daily balloon soundings.  The color scales are the same as in the Fairbanks plots.

Looking first at the cold half of the year, we see that a deep low-level temperature inversion is characteristic of the atmospheric profile at Barrow, and in this respect the climatology is similar to Fairbanks.  The warmest temperature in the "normal" sounding is found above 3000 feet from late November all the way into May.  However, the inversion strength at the surface is nowhere near as strong as at Fairbanks, because the near-constant wind mixes the low-level air and usually prevents radiative cooling from getting out of hand.

In the afternoon climatology we see that the surface inversion ceases to be present in early April as solar heating picks up, and the afternoon surface inversion doesn't return until late October.  However, the inversion aloft persists through the entire summer and disappears only from late August through September and October.

It is very interesting to observe that there are two distinct maxima in low-level lapse rate, in which temperatures are relatively high at the surface compared to aloft.  The first, and lesser, maximum is found in May when surface temperatures are warming rapidly, and solar heating is relatively strong, but the air aloft is still cold.  The second, more significant, maximum is in late August through early October, when the open ocean nearby keeps surface temperatures relatively high while the air aloft is cooling.  Between these two periods, particularly in early July, lapse rates are notably lower (more prone to inversion) in both morning and afternoon; this period is close to the time of peak warmth aloft.

Lastly, I'll point out that the average morning sounding shows no surface-based inversion throughout September and the first few days of October, because the ocean warming is sufficient to offset nighttime radiative cooling during this period.  Contrast this situation with that in Fairbanks, where even in the height of summer the morning sounding still shows a healthy inversion.

Temperatures Aloft (Climatology)

In my last post I looked at recent temperatures observed by the balloon soundings from Barrow and Fairbanks, as compared to the 1981-2010 climatology (normal).  I thought it would be interesting also to look at the temperature climatology itself to see if there is anything to learn.  Starting with Fairbanks, the images below show, in the top panel, the normal temperature throughout the year in the lowest ~8000 feet of the atmosphere,
and in the lower panel, the lapse rate (multiplied by -1 for easier interpretation).  The first image shows the normals for 12 UTC (3am AKST), and the second image shows 00 UTC (3pm AKST).  Click on the images for a better view; discussion is below.

There is not too much surprising in the Fairbanks climatology, for those familiar with the march of the seasons in the Alaskan interior.  From late October through early March the "normal" sounding shows a low-level temperature inversion even in the afternoon, and the inversion remains strong throughout the day in December and January.  For three months of deep winter (mid-November to mid-February), it is usual for the temperature lapse rate to be inverted to above 4000 feet elevation, i.e. the warmest temperature in the sounding typically occurs above 4000 feet.

The most interesting feature of the warm-season climatology may be the "super-adiabatic" lapse rate near the surface that is seen at 3pm from late March all the way through the beginning of October.  A super-adiabatic layer has temperature decreasing with height at more than 9.8 °C/km, which means that the layer is absolutely unstable to dry convective overturning.  The phenomenon is ubiquitous over land surfaces around the world when there is adequate solar radiation, because strong solar heating of the ground causes the lowest layers of air to gain heat more quickly than any heat removal mechanism (radiation, conduction, convection) can transport it elsewhere.  The implications of such a layer are that thermally-driven turbulence will be active, a fact that aircraft pilots are well aware of.

The 3am lapse rate climatology shows that low-level inversions are typically found in the early morning throughout the year, but are shallow from March through most of October.   The morning inversion is weakest in early August, and also shows a slight secondary minimum in early October; both of these times correspond to increased cloud cover and precipitation frequency, but usually without snow on the ground.  As snow cover typically becomes established later in October, the morning inversion strength rapidly picks up owing to efficient radiative cooling.

Given the length of this post, I'll discuss the Barrow temperature climatology in a separate post soon.

Temperatures Aloft

For several months I've been meaning to create a climatology of upper-air temperatures for Alaska observing stations at various heights above the ground.  The goal here is to visualize the evolution of temperature anomalies in a time-height cross-section, similar to the plot I showed here, but for departure from normal as well as just temperature.  Well, I finally got around to it, and the results are posted below for Fairbanks and Barrow since June 1.

There was a good deal of cool air around in June at both locations, with temperatures struggling to get much above freezing over Barrow until mid-June; but it's been quite warm lately, with freezing levels recently poking above 10,000 feet at Barrow.  Chilly Arctic air has just returned to the North Slope, however: Barrow reported freezing drizzle and 0.05" of ice accumulation overnight last night.

Sea Ice Update

The Barrow sea ice webcam has shown a nearly ice-free view for a week or so now (see below).  A nice satellite shot from the Suomi-NPP satellite two days ago (below, image from 22:43 UTC July 10) confirms that ice has retreated far from the coast on the west side of Point Barrow, but is hanging on near the coast on the east side.

Comparing the basin-wide ice concentration map to those from the past few years, we can see that the higher concentrations are more extensive than last year at this date, and much more than in 2012 (the record melt year).  However, the areal extent of ice cover is comparable to recent years and still well below the 1981-2010 normal.

Just for the record I'll also note that the latest CFSv2 forecast, which I wrote about earlier, is still showing above-normal ice extent in September, but the forecast is now much lower for August and has pulled back a little for September.  I would wager that the forecast will continue to retreat and we'll end up below-normal for September - though perhaps not as far below as in recent years.

Southeast Thunderstorm Environment

Reader Mike posed a question the other day about the kind of weather pattern that supports thunderstorms in southeast Alaska.  Of course, thunder is quite rare in that part of the world, as low-level temperatures are rarely warm and humid enough to create sufficient instability for deep convection.  However, thunderstorms do occasionally develop and it's interesting to look at the large-scale weather patterns that accompany these events.

To address this, I examined the hourly weather observations from both Haines and Juneau airports since 1998, when the ASOS platform was introduced.  Remarkably, I was unable to find a single instance of a thunderstorm being reported at Haines; it seems unlikely that this could be correct, but perhaps thunder really is that rare, as the town is surrounded by high mountains on nearly all sides.  In Juneau the situation is better (for this study), with 13 distinct thunderstorm episodes reported since 1998.  Several of these occurred in the cold season and were undoubtedly related to strong lifting associated with vigorous frontal passages.  I'm more interested in the 8 summer events, which are as follows:

June 28, 2000
June 27-28, 2004
July 24, 2004
August 10, 2004
August 18, 2004
June 17, 2013
June 24-25, 2013
July 13, 2013

Interestingly, all but one of these events occurred in 2004 and 2013, which were of course very warm summers in Alaska.  Looking at the sea surface temperature maps from these months, every one shows above-normal temperatures in the northeast Pacific close to the Alaska panhandle; so this appears to be an important factor.

Looking more closely at the individual events, I created maps of the 500 mb height (pressure) anomaly and 850 mb temperature anomaly at 12-hour intervals surrounding the onset time of each thunderstorm event.  These maps are shown below.  Note that two of the episodes included two distinct thunderstorms separated by many hours, so for these I created two sets of maps.  We can (perhaps simplistically) summarize the weather situation in Juneau for each event as follows:

June 28, 2000  -- upper-level low arriving from the Pacific, low-level cold front
June 27-28, 2004  -- very warm low-level air retreating northwestward, upper-level ridging to the north
July 24, 2004  -- upper-level trough arriving from the Pacific, low-level cold front
August 10, 2004  -- cold front arriving from the Pacific
August 18, 2004  -- very warm low-level air, ridging aloft, no obvious trigger
June 17, 2013  -- upper-level low diving down from the Yukon (described by Mike in his comment)
June 24-25, 2013  -- very warm low-level air migrating to the northwest, weak ridging to the north
July 13, 2013  -- upper-level low moving slowly southward over British Columbia.

Overall, we have three events with an obvious cold front arriving from the Pacific; two events with an upper-level low moving southward over Canada; and three events with no obvious upper-level trigger but with very warm air residing over or migrating away from the area.  So we see that there is considerable variety in the specifics of the weather patterns that bring thunderstorms to Juneau; but a common factor to all of the events is the unusually warm ocean water along the coast.  In years with near-normal or below-normal ocean temperatures, it appears that summer thunderstorms are much rarer in Juneau and presumably also in surrounding locations near the coast.

 

Statewide Temperature Record Frequency

As a follow up to Richard's excellent post regarding the frequency of daily records in Fairbanks, I thought it would be interesting to see if the experience of Fairbanks was representative of the rest of the state. Right away there exists an apples-to-oranges problem; namely, how do you compare a station with 100 years of records to a station with 50 years of records? The station with a 50-year record should be twice as likely to break a daily record as the station with a 100-year record on any given day.

To eliminate this problem, I took the daily climate data for 14 stations across the state and only used data from 1955 to the present. That way, all 14 stations contain the exact same period of record. This makes for an apples-to-apples comparison. The stations utilized are: Anchorage, Annette Island, Barrow, Bethel, Cold Bay, Fairbanks, Juneau, King Salmon, Kodiak, Kotzebue, McGrath, Nome, St. Paul, and Yakutat (see Figure 1).

For example, if Station 'A' has a record low on July 10th of 35°F set in 1935 and their second lowest temperature for July 10th was 37°F in 1965, the 1965 value is assigned as the record for that date and the 1935 value is ignored. Also in this example, 1965 is also given a tally of 1 for that daily record. If other days in 1965 achieved record temperatures, additional tallies are given to 1965. Ties are prorated to the each year that a tie was observed. For each record category (high max, low max, etc.), every station has a total tally of 365 – a value of 1 for each day of the year. In the series of figures below, each of the four temperature metrics (high max, low max, high min, and low min) are charted for each year followed by a combined chart. On the y-axis, the value is the number of records per year per station. Using an example of 3 stations in 2013, Annette Island had 6 record high maximums, Cold Bay had 19, and Juneau had 3. The average for those 3 stations is 9.3 record high maximums in 2013.

Note that 2014 is treated equally as compared to other years. Therefore, you may want to double the values assigned to 2014 to extrapolate how the end-of-year values might look.

Interpretation:

High maximums:

I expected the number of record high maximums to show more of an upward trend but that was not the case. There is a slight upward trend but the it is unlikely to pass a statistical significance test

Low maximums:

The rate of low maximum records was far more interesting. Many of the low maximum records were set prior to the PDO shift of the mid-70s. Were they cloudy, rainy summer records or cold, clear winter records? I'll have to do some more digging around to answer that question. The variability and magnitude of the values through the mid-70s is remarkable.

High minimums:

Minimum (low) temperatures are probably more sensitive to long-term climate change, urbanization, and vegetative changes. More than any other category, the number of high minimums closely mirrors the mid 70-PDO shift (see Figure 6).

Low minimums:

A gentle, steady downward trend in low minimum records is easily apparent. As stated in the previous paragraph, minimum temperatures are especially sensitive to changing conditions. Interestingly, the PDO shift only briefly interrupts (steeper slope for a few years) the trend toward reduced low minimum occurrences.

Overall:

A gradual decrease in temperature records during this 59-year period was observed. Does that mean the climate is more homogeneous? My initial reaction is that warmer temperatures generally have less variability. For example, daily standard deviations are smaller for stations with warmer temperatures (e.g., Anchorage and Fairbanks in January).

Figure 1. Stations used in the 1955-2014 records analysis.

Figure 2. Average number of record high maximum temperatures per year per station (1955-2014).

 Figure 3. Average number of record low maximum temperatures per year per station (1955-2014).

Figure 4. Average number of record high minimum temperatures per year per station (1955-2014).

Figure 5. Average number of record low minimum temperatures per year per station (1955-2014).

Figure 6. Average number of all daily record temperatures per year per station (1955-2014).

Figure 7. Average annual high and low temperature of the 14 stations used in this analysis (1955-present)