Wednesday, September 19, 2018

Historical Context for Recent Extreme

Following up on Saturday's post, I did a bit more work with the NCEP/NCAR global reanalysis data to get a better sense of just how unusual the recent circulation anomaly has been.  How many times has a comparable two-week anomaly been observed in the past, not just over Alaska but anywhere in the Northern Hemisphere?

I addressed this question with the following series of calculations:

(a) Obtain a smoothed daily climatology (normal) of 500mb heights for the 1958-2017 period.  Note that the NCEP/NCAR reanalysis starts in 1948, but the quality is lower before 1958 owing to the paucity of upper-air observations.
(b) Calculate mean height anomalies for 15-day periods ending on every day since 1958.
(c) Extract a smoothed daily climatology of the standard deviation of the 15-day anomalies.
(d) Find the daily maximum and minimum of standardized 15-day anomalies across the entire Northern Hemisphere.

Here are the results: the red line shows the daily hemispheric maxima of standardized 15-day height anomalies, and the blue line indicates the hemispheric minima.  (Click to enlarge)

Remarkably, the 15-day anomaly ending just 3 days ago at a grid point over the Bering Sea was the largest of any in the entire Northern Hemisphere history since 1958.  It was also larger in magnitude than any 15-day negative anomaly.  The closest competitor on the positive side was an extreme blocking ridge over northern Greenland in November 1965; here's a pair of maps showing a comparison of the two events:

Of course the long-term global rising trend in 500mb heights gives a slight "advantage" to the recent anomaly in comparison to a fixed climatology, so I re-did steps (b) - (d) above after removing the 1958-2017 linear trend (calculated for each day of the year and each grid point).  In this case the 1965 event moves into first place as the most extreme positive 15-day height anomaly at 500mb (in terms of standard deviations).

It would be interesting to see how other atmospheric variables like temperature and moisture reflect the nearly unprecedented nature of this month's anomaly, and it would also be worth looking at other, more modern data sets.

But based on these results, it seems safe to say that the recent high pressure ridge over the Bering Sea and western Alaska has been one of the most extreme 15-day weather anomalies in recent decades anywhere in the Northern Hemisphere.

Saturday, September 15, 2018

Record Bering Ridge

Much of Alaska has enjoyed a most remarkable extended spell of clear, dry, calm, and warm autumn weather in the past couple of weeks, thanks to an extremely strong and persistent ridge of high pressure centered over the Bering Sea.  Here's a map (click to enlarge) showing the 500mb height anomaly (departure from normal) in the first two weeks of this month, expressed in terms of standard deviations.  Near St. Lawrence Island the 500mb height has averaged more than 2.5 standard deviations above normal for a period of two weeks.  This is a remarkable anomaly.

Courtesy of Environment Canada, here's the 500mb analysis from last Saturday afternoon when the blocking ridge was in full swing.

As of yesterday afternoon the ridge was much weaker but still quite sufficient to dominate the weather over much of Alaska.

The chart below shows the 500mb height measurements from the sounding site on St. Paul Island in the southeastern Bering Sea.  The gray lines represent the one-standard deviation range either side of the 1981-2010 normal.  The persistence of the very strong positive height anomalies is notable, to say the least.

Another perspective on just how unusual this is can be gleaned from the history of September 1-14 measurements at St. Paul Island - see below.  I'm not sure that I've ever seen a multi-decadal climate chart with such an extreme anomaly in only one year that lies so far outside the envelope of observed climate.  It's quite astonishing.

Unfortunately Alaska's sounding sites farther north have not been reporting regularly this month, but I pulled out the equivalent data from the NCEP/NCAR Reanalysis at a grid point near the Bering Strait - see below.  The result is similar but actually even more extreme relative to the climate of recent decades.

Compared to the 1981-2010 climate, the mean 500mb height of the past 14 days has been 5.1 standard deviations above normal.  If the reanalysis is correct (and it should be quite accurate for 500mb height), and if the 500mb climate is Gaussian and stationary (unchanging), then we'd expect this kind of anomaly to occur less than once every million years, on average (for this particular date window).  Obviously one or more of the assumptions is false, but this illustrates the magnitude of the extreme.

What might have caused such a remarkable departure from normal?  I suggest a couple of possible causes.  First, it seems like the chain of events to set up the ridge started in late August when a burst of tropical cyclone activity occurred in the West Pacific.  Typhoon Cimaron (Category 3) hit Japan on August 23 and then the remnants headed up to the Aleutians, and on the same day Typhoon Soulik (also Category 3) hit South Korea and then moved into the Bering Sea.  Then about 10 days later Typhoon Jebi (Category 5 at one point) hit Japan (September 4) and headed for northeastern Russia.  The upper-level circulation of the latter storm, in particular, moved up the west side of the Bering Sea ridge on September 6 and seems to have provided a reinforcing boost for the ridge by transporting huge quantities of warm, moist air up to the region.

Here are the paths of the 3 storms - maps courtesy of Wikipedia.




A second contributing factor may be that sea surface temperatures have been far warmer than normal in the northern and western Bering Sea in recent weeks, and this unusual ocean warmth probably helped boost the overall warmth of the atmospheric column and build the ridge in the same area.  Here's an SST anomaly map that Rick Thoman posted on Twitter recently:

Finally, to illustrate the impacts of the weather pattern, here are a few nice shots of the lovely autumn scene across the state, courtesy of the FAA webcam views on Thursday evening.

Kivalina (far northwest Alaska)

Chandalar Shelf (in the central Brooks Range)


Teller (western Seward Peninsula)

Grayling (on the lower Yukon River)


Denali National Park



Yakutat (one of the wettest places in Alaska, especially at this time of year normally)

And here's a true-color shot from the Suomi polar orbiter on Thursday afternoon (click to enlarge).  The reddish-brown tundra colors of autumn are quite distinct in the north as well as in higher altitude areas throughout the state.

Saturday, September 8, 2018

Temperature Trends

In last week's post I mentioned that the abundant cloud cover over Fairbanks in August produced much lower than normal diurnal temperature variations; the average day/night temperature difference was almost the smallest on record.  Below is a chart (click to enlarge) showing the average diurnal temperature range in August, beginning in 1930 for the university's Experiment Farm and 1952 for the official Fairbanks climate record (when observations moved to the airport).  The downward trend is highly statistically significant at both locations; late summer now tends to see less difference between daily high and low temperatures than in the past.

It's of interest to see how the trends in daily high and low temperatures contribute to the diurnal range trend, and to see how this varies through the year.  Accordingly, the chart below shows smoothed daily values of the 1930-2017 linear trends in each quantity at the Experiment Station.  The relative positions of the blue and red lines show that daily low temperatures have warmed more than daily high temperatures at all times of the year, and therefore the diurnal range has been reduced throughout the year.  Note that both high and low temperatures have cooled slightly in mid-autumn, but the diurnal range has still been reduced.

The largest change in diurnal range has occurred in mid-summer.  Daily low temperatures have risen by more than +0.7 °F/decade from mid-May through the end of July, but high temperatures have not changed significantly in high summer, and so the diurnal range trend peaks at a rather remarkable -0.85 °F/decade (8.5 °F/century!) around July 1st.  It's clear therefore that the reduction in diurnal range in summer is attributable to nighttime warming rather than daytime cooling; as a simple example, the Experiment Farm has seen only four days with a freeze in June in the past 30 years, but a summer freeze was not uncommon in the earlier years (31 freezes in June and July of the 1930s and 1940s).

Using Fairbanks airport data since 1952, there are some differences as we would expect, but some of the same signals are evident: the winter peak in overall warming, the autumn lack of warming, and the summer trend (albeit less pronounced) towards smaller diurnal range.

If we re-do the calculations for Experiment Station over the shorter time period, there is a lot more similarity between the results for the two sites.

Finally, a chart of the June-July diurnal temperature range illustrates the changes at the two sites in the high summer months.  I'm not sure what was going on at the airport in the 1990s, and one does wonder if the transition to ASOS instruments in 1998 could have affected the numbers.  But the long history of consistent data from the Experiment Farm tells a clear story of remarkable change in this aspect of Fairbanks climate.

Friday, August 31, 2018

Seasonal Sunshine

As summer comes to an end on a wet and rather chilly note in the interior, it's interesting to observe that August has not only been very wet in Fairbanks, but cloud cover has been exceptionally high as well.  Two lines of evidence illustrate the remarkable nature of the anomaly:

- The month's average diurnal temperature range (difference between the daily high and low temperatures) is just about the smallest on record for August in Fairbanks (1930-present); this indicates an absence of the clear skies that tend to produce warmer days and cooler nights at this time of year.  If the temperature rises no higher than 53°F today, then this month's diurnal range will be exactly equal to the record low value from August 1998.  Interestingly the third-place year was 2015, and August of last year also had a very small mean diurnal range.

- The solar radiation at the CRN station near Fairbanks has been far below normal and easily the lowest for August in the 16-year history of the site; total solar energy has been only about two-thirds of normal.  Solar panel owners in Fairbanks-land are probably not happy at the moment.

The chart below (click to enlarge) shows the seasonal cycle of solar radiation at the Fairbanks CRN site.  The blue columns indicate the 2003-2017 monthly averages of measured radiation, and the additional gray area in each column shows how much greater the radiation would be if it matched the theoretical "clear sky" input (ignoring any shading effects from topography).  In March and April the Fairbanks CRN site receives about 70% of the theoretical maximum, but increasing cloud cover causes this to drop to only 50% by September.  Consequently, for example, August produces less solar energy than April, even though the sun is higher in the sky in August.  The same contrast is true of July versus May.

Below is the same chart for Utqiaġvik, formerly Barrow, where the CRN site also has a nice 15-year history.  Not only is the seasonal cycle in solar radiation even more pronounced than in Fairbanks because of the higher latitude (notice that Utqiaġvik sees more solar radiation than Fairbanks in May and June), but the seasonal change in cloud cover is also more dramatic.

For context, here's the same chart for Champaign, IL, at 40°N latitude.  The measured solar energy is not all that different from the Alaskan sites in May, but by the end of August there is a pronounced contrast.  And of course, as I discussed recently, this accelerating seasonal difference is related to the quick end of summer in the far north and also the typical seasonal change to wetter weather in August.

Sunday, August 26, 2018

How Damp has it Been?

Rick T. here. Over on my Twitter feed someone opined/asked about the recent spell of wet weather in Fairbanks-land and that it was more than we've had in recent years. So this got me to thinking again about how we might quantify "dampness" using climate data. I'm thinking of "dampness" in an Alaskan content, so not in terms of humidity or dew points but rather in terms of frequent rainy weather.

I've looked at this a couple of times before, but especially in 2016, which by almost any reckoning was a very wet summer in Fairbanks. In that work I combined the monthly total precipitation with the number of days with measurable precipitation to come up with a simple cumulative index that looks like this:
Figure 1
This produces values that seem intuitively in the ballpark: 2016 had the third highest May-August value, while 2004 and 2013 have low index values. However, this kind of index, while fine for looking back at the past, is less useful in near real-time since it relies monthly data and is really designed to look at compare complete seasons. People don't tend to experience weather at a monthly scale with a "hard reset" at the first of each month. Perhaps "frequent wet weather" would be better assessed using daily data. Which brings me back to the Twitter comment:  by August 22ⁿᵈ, for this person at least, the warm, dry weather of late July was clearly no longer on their "environmental radar".  So below is an effort at a dampness index with a shorter time horizon.

Here I've used the same combination (total precipitation times days with measurable precipitation) but applied this over a running 15-day window, so that each day gets an index value. So, to illustrate, on June 30, the index is calculated as the total precipitation from June 16-30ᵗʰ times the number of days with measurable precipitation in the same 15 day period. For July 1ˢᵗ, the window is June 17 to July 01. Do that for the whole of the warm season. For the last few years, the daily plot of this index looks like this (through Aug 25, 2018):
Figure 2
I particularly like this presentation, as the timing of "frequent wet weather" through the summer stands out clearly and make year-to-year comparison comparatively easy.

Of course, we can derived multi-month summary statistics from the daily index. Here's a plot of the average daily May-August index value:
Figure 3
This graphic does show some differences with the monthly-derived plot (Figure 1, above). For instance, 1967 now ranks as the second "dampiest" summer, and 1930 drops down a bit. Both of these seem like improvements. Even before the flooding rains of mid-August, 1967 had been a rainy summer, and 1930 had a lot of rain but much of that was concentrated in short bursts, e.g. 1.80" on July 2ⁿᵈ was from one thunderstorm: most of that rain fell in under an hour: the other 30 days in July had a total of 0.82"of rain. On the other hand, 1949 was a crappy summer no matter how you slice it. And in case you're wondering, with less than a week to go in August, the average index value for May-August 2018 is a comparatively low 3.7, though that will creep up a little with rain likely the next few days. 

Perhaps a future improvement of this kind of index would be to incorporate some temperature measure, which might improve resolving between days with convective showers and those day-long steady rains. 

Wet and Westerly August

As more rain moves across the interior today, adding to the moisture tally in this very wet August, it's interesting to note that it has been nearly a decade since Fairbanks saw an August that was both wet and notably wetter than July.  As we noted last year, July has become considerably wetter than August in recent years, mostly because of an increase in frequency of the heaviest rain events in July.

Here's another look at the big picture based on 20-year running means of the July and August precipitation totals; in the past 20 years, July has become as wet as August used to be in the 1930s and 1940s.  (Click to enlarge the image.)

But this year is bucking the trend, as July produced only 1.01" of rain in Fairbanks, but August is above 3.5" and climbing; this will be one of the wetter Augusts on record in Fairbanks.  Also of note is that Fairbanks saw a daily total of 1" earlier this month, and this hasn't happened in August since 1990; whereas July has produced 8 such days since 2003.

A couple of weeks ago I discussed the connection between August rains and strengthening westerly flow aloft at this time of year, and as if to reinforce that message, the speed of the westerly flow above Fairbanks this month has been close to a record.

Moreover, it turns out that the very heavy rains of August 5-6 were associated with the strongest westerly flow on record for so early in the autumn, as measured by Fairbanks balloon soundings.  The sounding below had a mean westerly wind component of 65 knots, or hurricane force, based on the values at 850, 700, 500, and 300 mb.

The view of the sky from the UAF webcam back on August 5th looked - to me - decidedly maritime, which might reflect the fact that the air aloft was making the trip from the Bering Sea to Fairbanks in a matter of 6 hours or less.

The previous record for strongest westerly flow before August 15 was on August 12, 1967, and Fairbanksans will know that date: it was the single wettest day in Fairbanks history, and the flooding was catastrophic.  Here's Rick Thoman's 50th anniversary blog post about it last year:

Saturday, August 18, 2018

Fire Season Ends

Alaska wildfire acreage has not increased in two weeks now, and with all the rain it's safe to assume that the fire season is over for 2018.  The total acreage burned was 399,000 acres, which is less than the last 3 years and nearly 40% below the median of the past couple of decades.  June was the busiest month for firefighters, although another 100,000 acres burned at the end of July and the very beginning of August.  Click to enlarge the figure below.

The chart below shows a parallel view of the statewide lightning data.  Despite a fast start in June, the state has seen considerably less lightning than the past 3 years, which were all quite active based on the short history of the ALDN TOA data.

The spatial distribution of fire was a little unusual this year, with the central Tanana Zone seeing over 50% of the state's fire acreage; this has only happened a couple of times before since 1990.  Indeed the Tanana Zone acreage was more than 50% above normal, and the Galena Zone also saw more fire than normal, but the rest of the state was much less active than normal.

Finally, the length of the fire season was very close to normal, based on the length of time (54 days) between the 5th and 95th percentile of the total acreage.  According to this definition, since 1995 the fire season length has typically been around 40-80 days, although it was as long as 112 days in 2007 and as short as 9 days in 2001.  The chart below shows an interesting absence of August fire in the past several years, despite wild variation in the overall level of activity.  (But in fact there is no correlation between total acreage and length of the season based on the definition here.)

Monday, August 13, 2018

Why is August so Wet?

August is known as a rainy month in western and interior Alaska, and this year has already proved the adage in spades.  Extraordinary rains fell across the southern and especially southeastern interior last week, with multi-day totals exceeding 3" in some spots - including Northway, which typically sees only about half of that for the entire month and has only twice had a wetter August in total.  (Coincidentally, Northway also had excessive rain a few months ago, with 2.1" falling on May 1st alone.)

The climatological shift to wetter weather in late summer is an interesting aspect of Alaska's climate that deserves a bit of analysis and explanation.  First, let's establish that August really is wetter than other months in much of the west and interior - as well as farther afield in northern latitudes.  The map below shows the difference between July and August precipitation, based on rain gauge data from 1981-2017 (click to enlarge).  The continental interior south of 60°N (Europe, Asia, and North America) is mostly drier in August than in July, but at higher latitudes increased precipitation is seen quite widely, including over the British Isles, most of northern Russia, much of Alaska, and northernmost Canada.  The change is very notable around the North Pacific and Bering Sea coastline.

The high-latitude trend is dramatically reversed from August to September in most sectors, with the notable exception of southern Alaska and the Pacific coast to the south of Alaska.

A map of grid boxes where August is the wettest month of the year reveals the general tendency for a seasonal peak in precipitation in late summer for many areas north of 60°N.

Looking at the climate division data for Alaska as a whole, the July-to-August increase in precipitation is the largest month-to-month change of the year (although September is wetter overall for the state because of the very large amounts that fall during autumn in the southeast).  The climate division data also confirm that August is the wettest month of the year for the North Slope, West Coast, and Central Interior divisions.  The eastern interior is slightly wetter in July, as suggested by the first map above.

If we dig into data from a few key individual climate sites from Anchorage north, we again confirm that August is the wettest month from the Y-K Delta region northward to the north-central interior and the North Slope.  July is slightly wetter than August in Fairbanks (although it didn't used to be), but the daily frequency of rainfall is higher in August, and the number of hours with rain is 25% higher in August than in July.

How can we explain the late summer peak in rainfall for western and northern Alaska, and more generally across the high-latitude regions of the Northern Hemisphere?  In a nutshell, the August maximum occurs because the north-south temperature gradient strengthens in late summer at high latitudes, leading to more vigorous westerly flow and more energetic large-scale low pressure systems that generate widespread rain.

To illustrate, consider the July and August mean 300mb height maps below (the heights are in units of decameters; 900 decameters is about 30,000 feet, so we're looking at the upper troposphere).  Heights are lower where the troposphere is colder, so the purple region represents the swirl of cold air near the pole, i.e. the tropospheric polar vortex.  Evidently the polar vortex strengthens quite a bit between July and August as the angle of the sun decreases over the Arctic and the region of 24-hour daylight shrinks; the total solar energy input over the Arctic is much less in August than in July.

However, notice that over the Pacific Ocean to the south the 300mb height increases slightly in August; this is because August's solar input is not reduced as dramatically in the south, and of course the Pacific Ocean has an enormous heat capacity and has a large lag in the seasonal temperature cycle.  Consequently, the atmosphere is still warming up as July transitions to August over the Pacific to the south of Alaska.

Here's a map to show the difference in heights between July and August.

The contrasting changes between north and south imply that the north-south gradient of height becomes tighter in August throughout the whole sub-Arctic region, and this means that the westerly wind speed picks up, because the pressure (height) gradient is the driving force for the jet stream.  Here's the resulting change in westerly 300mb wind from July to August:

The increase in wind speed is most notable from the Sea of Okhotsk across to southern Alaska, which are areas that see a pronounced increase in rainfall from July to August.  This is no coincidence: the mid-latitude jet stream is associated with large-scale weather disturbances that bring clouds and precipitation to broad areas.

Looking more closely at the Alaska sector, the figure below shows how the jet stream's position and strength evolve through the year in a longitude band just east of the date line (130-180°W).  The jet stream is strongest in winter owing to the strong north-south temperature gradient, and its average location is farthest south in January and February - far to the south of Alaska.  As summer progresses, the North Pacific jet stream weakens and moves farther north, reaching its most northerly position in August at about 50°N.  This is still well south of most of Alaska, but of course there is tremendous variability from week to week, and storm systems can occur over a wide range of latitude.

The July-August change that we noted above is evident at about 60°N in the diagram.  The strengthening of westerly flow over the Bering Sea and Alaska implies that more energy is available for the growth of large-scale low pressure systems that migrate generally eastward and bring widespread precipitation across these areas.

Balloon sounding data from Fairbanks confirm the strengthening of westerly flow aloft in August, and at the lower level of 700mb the westerly flow peaks in August before decreasing again in September.

The southward migration of the jet stream in September can partly explain the change to drier conditions in early autumn, but the September drying trend is also closely related to the rapid decrease in atmospheric moisture as seasonal cooling develops in earnest.  On the chart below I've added the seasonal variation in precipitable water in Fairbanks; as the atmosphere gets colder in autumn, it "holds" less water and so precipitation rates must decrease, all else being equal.

I'll make one last comment regarding an obvious question that arises from the first map: why does the eastern (and especially southeastern) interior turn drier, not wetter, in August?  The August drop-off in climatological normal precipitation is very notable in Northway.  I surmise that this happens because the strengthening westerly flow increases the rain shadow effect in the eastern interior; the August mean flow in this area is actually more west-southwesterly than westerly, and so the Alaska Range is often upstream.  An increased influence of downsloping may also explain the August drying trend in most of western Canada to the east of the coastal ranges (see the first map at top).