Weather warnings have been hoisted across interior and northern Alaska today, as a powerful cold front makes its way eastward across the state. Heavy snow was reported at Fairbanks airport between 3 and 4 pm, and this is a rarity in the Fairbanks climate. The last time heavy snow was reported in Fairbanks was in the epic snowstorm of February 2011. The current weather situation bears resemblance to the active weather pattern in the last 10 days of that month - read more about it here:
Looking farther back in the history of hourly observations at Fairbanks, heavy snow has been reported on only 17 previous occasions since 1950 (I threw out a couple of bogus reports). Only 12 of these were during the winter months of November through March, so that's an average rate of about once every 5 years during winter. Interestingly February has seen more occurrences of heavy snow than any other month, with 4 previous events that produced snow totals of 16.8" (1977), 5.9" (1981), 14.5" (1996), and 18.6" (2011).
Here's the scene at the Golden Heart Plaza at 4pm, with Iron Dog festivities hurriedly wrapping up under the heavy snowfall.
The following maps illustrate the meteorological situation at 3pm AKST, with a pronounced 500mb shortwave and vorticity maximum just upstream of Fairbanks and a strong 925mb temperature gradient across northwestern Alaska.
The preliminary 3pm surface analysis from Environment Canada shows (in the far upper left, click to enlarge) the strong surface low just to the northeast of Barrow.
As the low moves east and colder air rushes in, a severe blizzard is expected to develop over the eastern North Slope. The NWS forecast discussion notes that "wind gusts are expected to increase to at least 80 mph this evening". The intense temperature gradient and wind associated with the front are evident in the graphics below, which show a time-height cross-section of forecast conditions at Barter Island. As of this hour, winds have gusted to 51 mph at Deadhorse and at least 58 mph at Nuiqsut (although the anemometer has now stopped reporting). Interestingly the Umiat RAWS is reporting wind gusts to 54 mph, so the high winds are extending well inland from the ocean.
Finally, for reference, here is a recent plot of METAR observations showing the heavy snow in Fairbanks and the building blizzard on the eastern North Slope.
As a follow-up to Monday's post on freezing rain trends, I thought it would be interesting to divide the historical winter rain events according to temperature so that we can see the relative changes in frequency of "plain rain" versus "freezing rain". I define "plain rain" as rain at temperatures of 32 °F or higher, and freezing rain as rain at sub-freezing temperatures. Of course, plain rain in winter generally still freezes on road surfaces in interior and northern Alaska, even if the air temperature is well above freezing, so the distinction may not be practically all that useful - but the results are interesting nonetheless.
The charts below show the number of days each winter on which rain was reported, with red columns indicating reports at which the temperature was simultaneously at or above 32 °F, and blue columns indicating "freezing rain" reports. The 10-year trailing mean lines are colored correspondingly to highlight the long-term variations.
In Fairbanks, McGrath, and Kotzebue, there is a fairly clear trend towards plain rain taking up a larger proportion of all winter rain events in recent years. The changes in Nome are not quite as clear, as freezing rain also appears to have become more common in the past 20 years. Rain of any kind is rare in Barrow in winter, of course, although it does happen, and even at temperatures above freezing on three occasions since 1950.
As in the previous post, the 6-station mean consolidates the overall trend quite nicely (see below). There has been quite a pronounced increase in the frequency of plain rain, beginning (it would seem) with the anomalous winter of 2002-3. It is interesting to bear in mind that the PDO was predominantly negative from 2007-2012, so the warmer rain is presumably not attributable to warmer nearby ocean temperatures. Based on a Mann-Whitney U-test for non-normally distributed data, the increase in plain rain frequency is very statistically significant (p<0.02) and is not sensitive to the choice of years for the two sub-samples.
To confirm the changes at Fairbanks, I examined the temperature observations that accompanied all rain observations from 1950 to 2014, and compared the distributions from 1950-1982 and 1983-2014 (see below). The shift to warmer temperatures is notable, as temperatures above 34 °F have become much more common during winter rain events, while significantly sub-freezing temperatures have become much less common. This matches what we saw last week: the rain occurred at temperatures of 32, 33, and 34 °F. In other words, Fairbanks now tends to see plain rain relatively often in winter, as opposed to true freezing rain which was the favored outcome (when it rained) in the past.
Due to poor snow conditions in the southern mainland, the Iditarod race start was moved from Willow to Fairbanks. This is the second time in history (1973-2014) that the race start was moved due to low snow in the southern portion of the route. The other year that the race start was moved to Fairbanks was in 2003. The official restart will occur on Monday, March 9th. Approximately half way through the race, the 2015 route will join with the traditional northern route at the Koyukon village of Ruby. A map of the 2015 route is shown in Figure 1.
Figure 1. Iditarod route in 2015. Populated places along the route are shown as yellow circles.
Analysis Period
What does climatology tell us about the expected conditions during this year's Iditarod? To answer that question, we need to select an analysis period. In the last 20 years, the winning time for the Iditarod is 9 days, 7 hours, and 28 minutes. The top tier of mushers usually finish in about 10 days. Therefore, I chose a time period of 10 days for my analysis. Since the start date this year is on March 9th, the time period for climatological analysis is March 9th to March 18th. This brings up an important point. The analysis in this post is not a climate analysis of past Iditarods. Previous races usually began on a Sunday and the calendar date differs from year to year. This analysis is looking at the same time period over a number of years to assess what the range of temperature and snow conditions have been like along the route used in 2015.
Stations
There are eleven stations along the 2015 route in the Global Historical Climatology Network (GHCN) database. In this analysis I have restricted the date range to the 1973-2014 time period. This corresponds exactly to the years in which the Iditarod race has occurred. That gives us 42 possible years of data. Some stations, like Fairbanks and Nome, have outstanding data for all 42 years. Other stations have incomplete or interrupted data. A map of the climate stations used in this analysis is shown in Figure 2.
Figure 2. Climate stations along the 2015 Iditarod route.
Temperatures
When assessing the temperatures on a yearly basis, I used an average of all stations for all 10 days of the analysis period. This yields a single temperature value for each year. Since the stations are fairly well distributed geographically, arithmetic averaging is sufficient.
As you can see in Figure 3, a wide range of temperatures have been observed since 1973. Keep in mind that these are average daily temperatures (high plus low divided by two). Four years have seen an average temperature of 20°F or warmer and seven years have seen an average temperature below 0°F. Interestingly, the years since 2000 have been, on average, a little cooler than the years in the 1970's and 1980's. Based on climatology, it is reasonable to expect temperatures to average between +10°F and -10°F.
A note of caution, do not over interpret the trend in Figure 3. Remember that this year's Yukon Quest happened to take place during a brutally cold week during one of the warmest winters on record.
Figure 3. Average temperature for the March 9 to March 18 time period for the 11 stations along the 2015 route shown in Figure 2.
Now that we have seen the annual temperature chart, let's look and see how the warmest and coldest years varied on a day-to-day basis during the March 9 to March 18 time period. Figure 4 shows the three warmest and three coldest years.
Figure 3. Average temperature for the March 9 to March 18 time period for the three warmest and the three coldest years along the 2015 route shown in Figure 2.
Even in the very warmest years, only the last day or two was above freezing – even then only by a couple of degrees. The coldest years were quite cold. While not even close to this year's Yukon Quest, temperature have average as low as -25°F (1995) with average low temperatures near -40°C/F. Keep in mind that temperatures on the river ice can be 10°F colder than at the climate stations.
On a broader scale, we can look at temperatures on a statewide basis using the ESRL Reanalysis tool. I put together a video of temperatures from March 9 to March 18 for each year between 1948 and 2014. Video 1 shows the time lapse of the temperature departures from normal from year to year.
Video 1. Temperature departure from normal from 1948 to 2014 during the March 9 to March 18 time period. The 1981-2010 climate normals are used as the temperature baseline.
Snow
What does a musher really care about? Snow, of course. How has the snow been during the last 42 years along the 2015 route? The short answer is that it is consistently good. Figure 3 shows the average snow depth and average new snow for all stations along the 2015 route during the 1973-2014 time period.
Figure 4. Average new snow and snow depth for the March 9 to March 18 time period between 1973 and 2014 along the 2015 route shown in Figure 2.
The average snow depth was 20.8" and the lowest was still 9" (1986). Six years have seen over 30" of snow on average with a peak of 42" in 2009. As for new snow, most years have recorded 2"-5" of new snow during the 10-day period – average is 2.4".
Conclusion
While there are no guarantees in life, history tells us that the route chosen for the 2015 Iditarod is consistently cold and consistently snowy. Mush on!
This morning the National Weather Service noted the occurrence two lightning strikes concurrent with snow in Nome – thunder snow. This is quite the event. Figure 1 shows the text of the Special Weather Statement that was issued by the NWS office in Nome. Since the event was observed by NWS personnel, we assume that the report is valid. Unfortunately the two closest lightning detectors to Nome are both non-operational at the moment so we cannot tell how widespread the event was.
Figure 1. Special Weather Statement issued this morning by the Nome, Alaska, NWS office.
The METARs (see Figure 2) indicate the beginning and ending of the event. Several "special" METARs were issued that note the presence moderate snow and blowing snow concurrent with the lightning. The temperature was noted to be -2°C (28°F) during the thunderstorm.
Figure 2. Screenshot of METARs for Nome, Alaska, between 17:15 Z and 17:31 Z.
Thunder snow requires an unusual set of atmospheric occurrence to come together at just the right time. I am not a forecaster and am not qualified to provide context on the specific atmospheric dynamics at the time. However, for those out there more qualified than I am, here is the 12 Z (3 a.m.) upper air sounding from Nome this morning.
Figure 3. Upper air Skew T/Log P plot from Nome, Alaska, for February 24, 2015 (12 Z).
Just how rare of an occurrence is this? Surprisingly, it is not unprecedented. In fact there are 11 instances of thunderstorms in Nome with a temperature below freezing prior to today's occurrence. Figure 4 shows all of the hourly observations (1960-present) with snow and/or sub-freezing temperatures and a thunderstorm. Only 69 days in Nome have recorded a thunderstorm since 1960; the fact that 11 of those days were winter occurrences is remarkable. Even more amazing is that three instances occurred with a temperature of 2°F or colder!
Figure 4. Hourly observations from Nome, Alaska, that contain a thunderstorm and sub-freezing temperatures.
Fairbanks and many other interior communities were afflicted with freezing rain over the weekend in what seems to have become an annual winter ritual of late. Last winter I looked at the history of significant (0.1" or more) winter rain and freezing rain events in Fairbanks; we can now add another event to the list, which means that four of the last five winters have seen a significant event. This is quite a remarkable sequence, but as we noted before there have also been periods in the past when winter rain occurred with uncommon frequency.
Another way of looking at the historical trends is to extract reports of rain or freezing rain from the hourly observations. This method includes all winter rain events, not just the most significant ones, but it does not account for the quantity of rain that fell; some of the events will have been just a trace. The series of charts below shows the results when we do this for six locations with a reliable history of hourly observations since 1950. Note that I looked at just the deep winter (Dec-Feb) period, and I've used the same vertical scale for all the charts to highlight the differences in the typical frequency of winter rain.
It's interesting to observe that in Fairbanks, none of the recent winters comes close to 1962-63 in terms of the number of days with rain. I confirmed this by looking at the Local Climatological Data publication; remarkably, Fairbanks reported "glaze" in January 1963 on 10 separate days (some of these were freezing drizzle events). However, the persistent recurrence of rain in recent winters appears to be unprecedented since 1950, with this winter (2014-15) being the 7th in a row with at least one report of rain. The results also show that Nome and (especially) Kotzebue have seen a higher frequency of winter rain in the past decade than in earlier years.
Another interesting observation is that last winter (2013-14) was the first one with at least 2 rain reports (days) at all 6 locations. Only 4 previous winters had at least one report at each location. We can also note that only 5 winters had zero reports of rain at all of the locations - most recently in 2003-4. In other words, nearly every year brings rain in deep winter to at least one of these locations.
The chart below shows an attempt to combine the 6 station series into an overall mean. For each winter, I took the percent of normal at each station and then calculated the 6-station mean of these numbers. This is better than just averaging the 6 series together, because that result would be dominated by the higher values at Nome and McGrath.
The mean chart illustrates the highly anomalous nature of last winter and also shows that the 10-year trailing mean is slightly higher than at any earlier point since 1950, although the winter of 1962-63 produced a similar trailing mean. The 10-year trailing median provides an alternative measure of the long-term variations, but this is no higher than the value in about 1980.
In what may be the least surprising weather development this winter, warm conditions have returned to many parts of Alaska in recent days. The PDO is still very strongly positive. Bettles has reported a high temperature of 30 °F the last four days including today; the last time four consecutive days were this warm in December through February was in 1991.
The charts below show the temperatures compared to normal since November 1 at Bettles, Fairbanks, Barrow, and Kotzebue. The late January cold snap stands out as significant at Bettles - although it wasn't anything special by historical standards. Remarkably, Barrow has not yet seen a single day with daily mean temperature more than 1 standard deviation below normal, and this is unprecedented if we compare historical temperatures to the modern normal.
The snow drought continues in Kotzebue, with a meager 4.3" of snow in 2015 so far, and still only 4 inches on the ground. The snow depth has not yet exceeded 4", which is a record low. The ranked observations of total snowfall and maximum snow depth through February 18 are shown below. The colored markers highlight the most recent 12 years, showing that this winter is the complete opposite of the very snowy conditions observed often in recent winters.
I've been giving some thought lately to possible methods of measuring the degree of persistence of unusual temperatures within a season or a year. For example, some months, seasons, or years are characterized by long stretches of significantly above-normal or below-normal temperatures, and others by short-lived warm or cold spells, with frequent alternation or variation. Two winters that contrasted strongly in this respect in Fairbanks are shown below.
In devising a metric for the degree of persistence, it's clear that the mean departure from normal is not particularly of interest; a season or year could end up near normal even though many long warm and cold spells were observed. I considered revisiting a measure of the average magnitude of the departures from normal, or the frequency of large departures from normal (as e.g. here), but this tells us primarily about the degree of "extremeness", not the degree of persistence. I'm specifically interested in a metric that can tell the difference between temperatures that tend to remain consistently and significantly on one side of normal, as opposed to highly variable temperatures, over the course of the month/season/year. The motivation for this is primarily to see if "persistence" has changed noticeably over recent decades.
The index I came up with - in experimental form - measures persistence by adding up the changes in mean temperature anomaly from day to day in terms of standard deviations. For example, if the temperature is 1 SD above normal one day, and 1 SD below normal the next, then +2.0 is added to the index. If the temperature is unchanged relative to normal, then nothing is added. Also, I limit the magnitude of the temperature anomalies to +/- 1 SD, because I wish to capture the idea that anomalies greater than 1 SD are all qualitatively similar - for example, we would call the weather "unusually warm" regardless of whether the temperature is 1, 2, or 3 SD above normal. Limiting the anomalies to +/- 1 SD means that the maximum daily index increment is +2.0.
After calculating all the daily increments over a period, the average is taken, and then for cosmetic purposes the "persistence index" is finally computed as the reciprocal of that average, minus 0.5. This conversion is desirable so that higher values reflect greater persistence, and so that we get a zero index value ("zero" persistence) in the hypothetical case when temperatures flip from +1 SD to -1 SD every single day.
If you're still with me, the result is an index that is quite easy to calculate and, I think, fairly reflects the degree of persistence within a chosen time period. The chart below shows the annual value of the index for the November-March season in Fairbanks, and the 10-year trailing mean is shown in red. It's interesting to observe that the 10-year mean remained nearly stationary from the 1930s through the early 60s, then it was somewhat higher until the late 80s as there were no low-persistence winters from 1969 through 1989. However, in the last 25 years the index has been lower again on average, but with strong variability. The winter of 1980-81 saw the highest degree of persistence, and 2008-2009 the lowest.
The chart below shows that the persistence index is only weakly connected to mean temperature anomaly, which is a good thing as I'm interested in how persistence varies independently of mean temperature. We can see that the most anomalous winters are associated with high persistence, and this is obviously unavoidable, but there is little correlation over most of the range of temperature.
The climatology of the persistence index is shown below, based on the long-term mean for each overlapping three-month "season". It's not too surprising that persistence is highest in winter and lowest in summer; temperature anomalies tend to stick around longer in winter. This is presumably because tropospheric Rossby waves have longer wavelengths in winter and therefore move more slowly.
Finally, the annual mean persistence index shows some interesting decadal-scale variations (see below), but the most recent two decades have been generally unremarkable compared to the earlier history.
I'd welcome ideas from readers on alternative ways to measure persistence. The next step is to apply the procedure to upper-air data so that we can find out if upper-air flow patterns have become more or less persistent over time.
A few days ago the Iditarod's board of directors voted to move the (re)start of the race from Anchorage to Fairbanks, as in 2003, because of a lack of snow along some rugged parts of the trail in the Alaska Range. I couldn't help noticing a comment by race director Stan Hooley, reported by the News-Miner as "I think we’re all worried about the weather [...] There’s a pattern that’s developed. And that’s a concern to all of us." In view of this, it's of interest to look at the snow depth data from several first-order observing stations to see if there is any evidence of changing snow conditions in recent years.
The charts below show the mean snow depth in February and March for several locations with a long history of reliable data. The first chart shows the data for Anchorage, McGrath, and Nome (along the traditional Iditarod route), and the second chart pertains to Kotzebue, Bethel, and King Salmon (for comparison purposes).
The first thing that jumps out is that there's no evidence of systematically lower snow depth in recent years at these locations, and in fact quite the opposite is true; on average, snow depth has been higher in the last 20-25 years in Anchorage, Nome, and Kotzebue, and slightly higher in McGrath. The averages for two consecutive 30-year periods and for the most recent decade are shown below. It's worth noting that at Nome, where the increase has been rather dramatic, the difference between the two 30-year periods is not statistically significant; however, the difference between 1955-1984 and 2005-2014 is very statistically significant, according to a Mann-Whitney U-test.
The second interesting aspect is the marked increase in variability of the snow depth at McGrath, Nome, and Kotzebue (especially the latter). See below for a summary chart. At McGrath, despite the fact that snow depth has increased on average, 4 of the lowest 9 snow depths have occurred since 2003, along with 3 of the top 9 snow depths. The situation is more extreme in Kotzebue: of the past 11 years, only 1 (2007) did not fall in the top or bottom 20% of the overall historical distribution.
Although this is only a cursory look at historical snow data, the results suggest that perhaps the problem for sled dog racing in Alaska may not be a lack of snow in the mean, but too much variability from year to year. Typically one thinks of climate changes primarily in terms of shifting the mean (or median); but this appears to be a rather clear example in which changes of variance have been very large - and possibly more significant in terms of impacts for human and animal activity.
[Update February 24: reader Andy mentioned the snow shadow on the northwest side of the Alaska Range and asked about snow depth data from Farewell Lake. The chart below shows the distribution of daily snow depth for 17 winters at that location; the median is between 5 and 10 inches for much of the winter, and very low snow is not at all uncommon in February and March.]
This is a follow-up post concerning the topic of Monday's post, i.e. that this winter has been remarkably lacking in wind in Fairbanks. Brian showed that the same is not true just a short distance up the Tanana River at Eielson AFB, but I haven't had a chance to examine other locations yet; so for now I'll continue with the focus on conditions in Fairbanks.
The first item of analysis I looked at was the mean pressure gradient this winter near Fairbanks, compared to previous winters (see below). The mean gradient has been lower than the long-term mean, so this may be part of the answer for the windless conditions. However, the pressure gradient is no lower than in some other years of the last decade or so, and therefore we can't exclusively blame this winter's extreme lack of wind on weak pressure gradients at the synoptic scale.
Secondly, I looked at the mean wind speed at several heights above ground using the balloon sounding data from Fairbanks airport (see below). We've seen these long-term trends before in a previous post. Interestingly the wind speed at 500m above ground has been near a record high this winter, while the wind speed at 50m has been near a record low; the overall trend since the early 1990's suggests an increasing degree of decoupling between the surface-level air and the air aloft. Note that I obtained the wind speed at each level by interpolating the wind vectors between reported heights on each sounding. The main point here is that the sounding data confirms that the low-level winds have been unusually low this winter (although not a record in this data).
Next I calculated the mean inversion strength in the lowest 100m of the atmosphere, again from the sounding data - see below. We would expect to see a relatively strong low-level inversion in association with light winds, and indeed we see that the near-surface inversion has been stronger than in most past years; so this is physically consistent.
Finally, I looked at the distribution of wind direction this winter, because reader Gary suggested that southeasterly flow can set up a low-level eddy circulation in the Fairbanks bowl, and this would seem to favor local stagnation of the flow. The chart below shows that surface winds (when not calm) have mostly occurred from the north through northeast quadrant, with a secondary peak in frequency from the south. However, at a level of just 50m above the ground the flow has been mostly out of the northeast through east-southeast.
Looking at the frequency of 50m winds out of the east to southeast quadrant (90-135°), we find that this winter is second only to 2002-2003 (see below). East-southeasterly winds have been considerably more common than usual just above the surface, and so this supports Gary's hypothesis that flow from this direction favors light winds and stagnation in the Fairbanks area.
In summary, the results so far suggest that this winter's lack of wind may have been partly caused by the overall benign pressure pattern, with weak synoptic-scale gradients (related to persistent ridging aloft), and also partly by a prevalence of (light) southeasterly flows. It appears to be a plausible hypothesis that light southeasterly winds interact with local topography to create near-stagnant flow in the Fairbanks area; and this happens to have been a common occurrence this winter.
