Thursday, July 18, 2019

Lightning Distribution

As a follow-up to last week's post on extreme lightning activity, I thought it would be interesting to take a look at the spatial distribution of lightning in Alaska, as documented by the Alaska Lightning Detection Network.  As many readers know, the ALDN system underwent a transition in 2012 to a new network with different sensors, so the data are not directly comparable pre- and post-2012.  However, we can look at each period separately and make some qualitative comparisons.

The map below (click to enlarge) shows the density of lightning strikes in the 1986-2012 data, and you can zoom in by viewing the map on the ArcGIS platform at the following link:

The ArcGIS link also includes a second layer that shows the 2012-2018 data; you should be able to toggle the layers by navigating to "Content" on the left side of the interface.  Here's a static version of the 2012-2018 map:

Note that the data values are essentially arbitrary; they are obtained by counting the number of lightning strikes in latitude-longitude grid boxes, so there is a small north-south discrepancy in terms of the land area within each grid box.  Also, as mentioned above the new ALDN sensors record lightning strikes at a different rate than the old network, so we can't do a comparison of total strikes between the two periods.

However, it does seem that we can make one very interesting observation, which is that recent years have seen much more lightning in the southern interior than in the rest of the interior, and this is very different from the spatial distribution in the earlier decades.  The 1986-2012 history shows the highest concentration of lightning in the Yukon-Tanana uplands, and especially in the hills just to the east and northeast of Fairbanks.  This seems reasonable and agrees with earlier results here:

In contrast, the data from the past 7 years shows the highest lightning density along the west and north side of the Alaska Range and in the upper Tanana River valley.  It's a striking difference (pardon the pun).

To confirm the result, I re-calculated the number of strikes within two similar latitude/longitude rectangles: one focused on the 2012-2018 hotspot to the south of McGrath, and the other centered on the hills near Fairbanks.  In the 1986-2012 data, the Fairbanks area saw about twice as much lightning as the McGrath region, but in 2012-2018 the ratio was approximately reversed.

Of course the past 7 years is a short period from which to draw any conclusions, especially when the sum total of lightning activity is so heavily influenced by a relatively small number of very active days.  For example, 20% of all 2012-2018 lightning in my box south of McGrath occurred in a single day: June 23, 2015 (see below).  So the apparent southward shift in lightning density since 2012 might be related to just a few very unusual events, and therefore might not be statistically significant.  But it is quite intriguing nonetheless, and I can't help wondering if the exceptional North Pacific warmth of the past several years might have something to do with it.  Some interesting research could be done here.

Friday, July 12, 2019

Lightning Outbreak

Yesterday was an extraordinary day for lightning in Alaska, with the highest number of lightning strikes recorded for any single day in the 2012-present history of the modern Alaska Lightning Detection Network.  The previous daily record from 2016 was broken by about 25%, based on strike counts for 24 hour periods ending at 6 am.

Note that the ALDN database includes plenty of lightning data from northwestern Canada, but I'm excluding anything east of 141°W (the north-south Alaska/Yukon border) so as to focus on Alaska.  (But yesterday also broke the record for the entire domain, and also for days defined as midnight-to-midnight.)

The annual totals and peak daily numbers for each year since 2012 are shown below.  It will be interesting to see if this year sets a record for the annual total; we are already past the climatological peak in lightning activity for Alaska, but a few more big days could push 2019 into first place.

Monday, July 8, 2019

Heat and Smoke

As I'm sure all readers are fully aware, an extraordinarily intense heat wave has been occurring across much of southern Alaska in the past week, and longstanding temperature records have been falling left and right.   Here's a quick summary that Rick Thoman posted yesterday:


There's far too much to comment on, but the remarkable July 4 heat in Anchorage may be the most astonishing statistic: the previous all-time record for the city's official climate site was only 85°F.

Dense smoke from wildfires has held temperatures down in some parts of the interior, including Fairbanks, but many locations in western and southern Alaska are well up into the 80s today.  Particularly striking to me is the 87°F currently being reported from Noatak in the northwest - above the Arctic Circle - and 92°F earlier today at the high-quality CRN station to the southeast of Ruby.

Here's an annotated true-color satellite image, courtesy of Rick Thoman and UAF:

Click on the image below for a higher resolution version:

 As bad as the smoke is at the moment in Fairbanks, this summer has a long way to go to equal the awful summer of 2004 in terms of duration and choking density of smoke - see the chart below, showing data from the Fairbanks airport ASOS instrument.  Here's hoping that 2004 remains an outlier.

Tuesday, July 2, 2019

Wildfire Activity

Increasingly unusual warmth in the past couple of weeks has allowed wildfire activity to pick up substantially across Alaska, and as of today the state has seen about double the normal fire acreage for the date.  The typical midpoint for statewide fire acreage is only about a week away, and so 2019's acreage is close to exceeding the median value for the entire season.  The last 3 years were relatively normal for statewide fire activity (at least based on a "modern" normal since 1995), but with another round of heat currently developing across Alaska, there's little doubt that 2019 will end up with significantly above-normal wildfire activity.  Click to enlarge the figure below.

The fire that is drawing the most attention from fire suppression experts is the Shovel Creek fire, which is very close to Fairbanks - just the other side of Murphy Dome.  The fire is not nearly as large as some other fires in the state, but it is expected to grow and there is a high risk to properties near Fairbanks.  Consequently a national-level command (Type 1 IMT) is taking over the suppression efforts tomorrow; I'm not sure how often this occurs in Alaska.

Here are a couple of maps showing how the Shovel Creek fire grew between Friday (3,424 acres) and yesterday (10,008 acres).  Ester Dome (on the west side of Fairbanks) is located in the lower right of the map domain.  Let's hope the trend does not continue.

Last year I commented on the positive correlation between North Pacific sea surface temperatures and Alaska wildfire activity, so perhaps it should be no surprise that this fire season is becoming active.  The Bering Sea and Gulf of Alaska surface temperatures have been far above normal since winter, and Rick Thoman has been documenting (via his Twitter feed) temperature anomalies of several degrees Celsius to the west of Alaska in recent weeks.  From a global perspective, the warmth in the North Pacific is one of the more striking features in the global oceans at present; here's a map of May SST departure from normal in terms of standard deviations.

To reinforce the idea of a link between Alaska fire and North Pacific temperatures, the chart below shows a 25 year history of the North Pacific Mode index in May and subsequent Alaska fire acreage.  Remarkably, all of the very active fire seasons in recent decades have occurred when the NPM index was positive.  Recall that the NPM represents a specific pattern of SST anomalies that extends west-east across the North Pacific to the south of Alaska; not surprisingly, the NPM was positive in May and has become strongly positive in recent weeks (see figures below).

The long-term history of the NPM index does not have a rising trend over time, but this is by design: the NPM definition proposed by Hartmann (2015) removes the linear trend from the SST data before doing the EOF analysis.  In contrast, if we look at area-average SSTs from 40°N to the Bering Strait, there is a very significant warming trend, and the past several years have been persistently very warm.

It stands to reason that higher temperatures would favor increased wildfire activity, all else being equal, because earlier snowmelt allows fuels to dry out earlier in the year, and increased evapotranspiration creates larger moisture deficits if rainfall does not increase.  The long-term increase in temperature has probably therefore produced a background increase in fire activity, or an increase in what is "normal" for a multi-year or decadal period.

However, the year-to-year variability in ocean and atmosphere patterns is still very large and obviously controls the level of fire activity in any given year; and the NPM appears to be closely connected to the key weather factors that influence Alaska fire activity.  A lot more research could be done on this, and it seems likely that useful seasonal fire predictions might be possible.  If anyone is interested in collaboration, leave a comment!

Monday, June 24, 2019

Utqiaġvik Temperature Records

Exceptional warmth occurred last week over much of northern Alaska, and a significant new record high temperature was set at Utqiaġvik (formerly Barrow), where the 73°F on Thursday was the highest on record for the month of June; the previous record was 72°F in 1996.  June therefore joins May, October, and January as having set or tied calendar-month high temperature records in the last few years.

There are some interesting aspects to the distribution of Utqiaġvik's calendar-month records over the decades, but first I'll state the obvious: average temperatures have increased very dramatically in the 100+ years of climate observations, and daily record high temperatures have been broken or tied with high frequency in recent years.  The second chart below shows a running 10-year total number of high temperature records, both for daily maximum temperature and for high daily minimum temperature.  Very nearly half of all of the calendar-day high temperature records have been set or tied since 1990.  Note that I'm only using the Weather Bureau/NWS era of 1930-present for the analysis of records.

The distribution of daily low temperature records is even more striking, because the frequency has dropped to just about zero in recent years.  In fact, the last time a calendar-day record was broken (not tied) for daily minimum temperature was in February 2009, and if we allow data from the 1920s, we have to go back to 2006 to find a new low temperature record.

In light of this, the much smaller sample of calendar-month records is interesting; see the chart below.  The red columns show the number of calendar-month high temperature records for non-overlapping decades since 1930; these are the monthly records (set or tied) for daily maximum temperature, of which four have been reached in the past few years.  The blue columns show the high records for daily minimum temperature, and a few of these have also been broken or tied in recent years (January, July, and November).

The obviously interesting aspect of this is how many of Utqiaġvik's calendar-month records were set in the 1930s and are still standing; in fact the largest number of warm records (daily max and daily min) from a single decade is from the 1930s.  This is really quite surprising, given the ramp-up in mean temperatures and daily warm records.  Of course the calendar-month records represent a very small and rather arbitrary sample of the most extreme events, and a more rigorous statistical analysis would be needed to make a definitive statement; but this does seem to suggest that the very extreme warm tails of the temperature distribution have not shifted as much as the less extreme parts of the distribution.  In other words, it appears that the modern warmer climate is not producing very-rare warm extremes of the same amplitude (relative to the mean) as the climate of the 1930s.  I've noted before on this blog that the 1930s was a time of wild extremes in Alaska climate, and this is another piece of evidence in that direction.

Here's the distribution of calendar-month cold records in Utqiaġvik.  The 1970s really stand out, and this is fairly consistent with the mean temperature and the daily records from that decade, although the monthly records suggest that extreme anomalies on the cold side were particularly concentrated in that decade, just as warm extremes were concentrated in the 1930s.

Saturday, June 15, 2019

Permafrost Update

Back in September I posted some brief analysis of soil temperature profiles at one of the Fairbanks monitoring sites run by UAF's Permafrost Laboratory.  Not surprisingly (in view of the remarkable warmth of recent years) the data showed a dramatic warming trend, and so it's of interest to do an update now that another year of data is available.

The charts below are the same as before, but with 2017's annual values appended with the dashed black lines.  2017's mean temperature was slightly lower than 2016 in most of the column, and the annual minimum was lower than for 2016 at each level between about 0.25m and 1.5m depth.  We know why this is: winter 2016-2017 was cooler than the other winters since 2013-14, and March 2017 in particular was unusually cold, so this provided a slight recovery in the condition of the permafrost.

Readers will note that the warming trend apparently continued unabated at the top layers just below the surface, but this appears to be at least partly related to some rather substantial ground subsidence relative to the temperature instrument in recent years.  From 2007 through 2016, the second level below ground was reported as 12cm, but in 2016-17 this changed to 8cm, and in the most recent year (2017-2018) it was reported as only 4cm.  The top-level sensor started at 2.5cm below ground but is apparently now 6cm above ground!  I haven't made the vertical adjustment on the chart, but obviously this could make a rather big difference at the top levels.

It's worth looking again at the temperature trace from the sensor at 0.525m (now 0.44m) - see below.  The effect of March 2017 is obvious, but I think what is more striking is the fact that this level almost failed to re-freeze in the subsequent winter (2017-2018); the temperature didn't drop below freezing until March 14, the minimum was -0.24°C on April 11, and it was back up to -0.09°C by the end of the data series on June 1.

Of course winter 2017-2018 was very warm in Fairbanks (13°F above normal in December), but the unusually deep snow cover probably also contributed to the lack of a subsurface cold wave in late winter; the ground was well insulated from the cold above, and there wasn't much cold to diffuse downward.

These results are all taken from just one location; what about others?  I did similar analysis with data from two sensors at the Bonanza Creek ecological research site, which is about half way between Fairbanks and Nenana.  At the first sensor, the trends since 2010 are similar to those at Smith Lake, but 2017 produced very similar temperatures to 2016 instead of seeing a small recovery towards colder conditions.  (Note that the sensor depths have trended in the opposite direction at this location, with the top sensor dropping from 7cm in 2010 to 11cm in 2017-18.)

The temperature time series at 52cm depth is quite remarkable when we look at the late winter minima; there has been a stair-step pattern of 2°C increases at two-year intervals (2014, 2016, 2018).  And just like at Smith Lake, the last winter in the series shows no hard freeze at this level about 20 inches below the surface; the winter's minimum was only -0.1°C.

The same thing is evident closer to the surface; even at the shallow level of 22cm (now 26cm), the 2017-2018 winter minimum was only -0.7°C.

And for comparison, here's a quick look at the same levels from the second permafrost site at Bonanza Creek.  It's essentially the same story, although winter 2015-2016 also saw very minimal re-freeze at the half-meter depth.

In conclusion, it's striking to see the rapid pace of change at these permafrost sites in interior Alaska.  Not only has warming occurred throughout, and the active layer deepened (i.e. seasonal thaw extends deeper), but we're seeing layers that were formally permafrost now almost fail to refreeze in winter (e.g. 0.5m level at Smith Lake).

In last September's post I said that "Smith Lake #1 may soon see seasonal ice atop permanent thaw", but I wasn't imagining that something like this would appear in the very next year of data.  What will the next annual update reveal?

Saturday, June 8, 2019

Peak Rainfall Rates

Intrigued by the extreme rainfall rate that was reported last Sunday at Fairbanks airport, I procured the history of 1-minute precipitation observations from the Fairbanks ASOS; this data goes back to 2000 and now provides a nearly two-decade history of high-resolution rainfall data in the warm season.  I had to pick out about 40 obviously erroneous data points, and it's likely I didn't find all the problems, but I checked the most extreme events of each year in the remaining data and they all look fine.

There is a lot of interesting analysis that could be done on the data, but the charts below provide the perspective for last week's rain.  The greatest 5-minute rainfall amount was 0.36" on July 4, 2005, and the top 10-minute and 15-minute totals occurred on July 21, 2010.  Both of these events saw peak 1-minute rates of 0.09", or 5.4"/hour, and that's also the highest observed for a 1-minute total.

The 1-minute data is only updated once a month, so we won't know for a few more weeks whether the recent event broke these records, but the realtime data from the ASOS does suggest so: 0.55" in 8 minutes is greater than the 10-minute record of 0.53".

What's perhaps even more striking is that the 10-minute record for June is only 0.22"; it's much more common (relatively) for extremely heavy rain to fall in July.  There have been only 6 events in total with at least 0.25" in 10 minutes, and of these 5 were in July and 1 was in August (August 17, 2008).

It's also curious to note something of a discrepancy with the NOAA precipitation atlas: the NOAA frequency estimates indicate a recurrence interval of about 20 years for 0.25" in 5 minutes, but it appears Fairbanks has now seen this happen 4 times in 20 years.