Friday, May 25, 2018

Sunshine at CRN Sites

After reviewing the status of Alaska's CRN data last week, curiosity led me to take a look at the solar radiation measurements from these sites.  For example, it's interesting to find out what these instruments have to say about how solar radiation varies across the state - at least for the last few years.

The chart below shows a summary of April-October total solar energy for the 11 sites that have been in operation since at least 2013 and have essentially complete data for these months.  The horizontal black lines show the 5-year mean for each site, and the blue bars indicate the minimum and maximum values within this brief 5-year period.

The sites near Fairbanks and Tok are the only interior sites represented here, and as we would expect they are at the top for solar energy.  Interestingly Metlakatla, on Annette Island in Alaska's far southeast, is in 3rd place and close behind Fairbanks.  It's not surprising to find that rainy Sitka is easily in last place; but note that the difference in solar energy between the cloudiest site and the sunniest site is less than 50%.

The range between minimum and maximum seasonal totals is remarkably high at the Red Dog site, but this is mostly because of a very low total from 2013; this looks suspicious and might be incorrect.  (If we exclude 2013, Red Dog actually comes in above Fairbanks for mean solar energy, and the Red Dog instrument also reported the highest seasonal total of any site in any year.)  More robust, perhaps, is the very low variance of solar radiation at the sites near Barrow (Utqiaġvik) and Tok; it's quite extraordinary, actually, how consistent the solar energy has been at these locations in the past 5 years.  But in the case of Barrow, data from earlier years showed higher values - see below.

The decrease of solar energy at the Barrow CRN site, evident in the chart below, is highly statistically significant even over such a short period.  Presumably this is related to the increase in open water area in the Arctic Ocean and therefore higher evaporation and increased cloudiness in recent years.

Friday, May 18, 2018

Improvement in CRN Data

I've mentioned the US Climate Reference Network program on previous occasions - it's a national effort to install high-quality climate monitoring instruments throughout the nation in locations that are unlikely to be affected significantly by urbanization in the coming decades.  So far 21 such sites have been installed in Alaska, with the most recent being near Toolik Lake and Cordova last summer.  Several more are planned:

Read more about the program here:

In previous years it has been very disappointing to see that several of the sites had serious and persistent problems with missing temperature data in the winter months.   This is apparently caused by the fuel systems being unable to produce sufficient electrical power to operate the instruments and other electronics during cold weather.  It's a little surprising that the system engineering wasn't up to the challenge of Alaska's climate, but one must concede that it's no small task to run a complex array of instruments without external power throughout the deep cold and dark of the high-latitude winter.

But happily there is now some good news: some modifications were made in last summer's maintenance visits, and the past winter saw a significant reduction in the amount of missing data from a number of sites.  The chart below shows (in green) the statewide percentage of all November-March days for which daily high and low temperature data are available in GHCN, and the black columns show the number of sites.  The network was in a rather sorry state in winter 2015-2016, with more than 20% of days missing from the 18 sites around the state, but in the past winter only 7% of days were missing statewide.  If the improving trend continues, we'll soon be in good shape.

Here's the percent complete over the lifetime of each site.  Notice that the really bad locations are some of the coldest and most remote locations, whereas the instruments in southeast Alaska are performing just fine.

The sites that improved dramatically this winter were Deadhorse, Nowitna (Ruby 44 ESE), and the sites near Selawik and King Salmon: whatever tweaks were made at these locations worked nicely.  On the other hand, there are still significant problems at Ivotuk and Denali 27 N.  Let's hope that one more year does the trick to get the network running as intended.

Saturday, May 12, 2018

Chena Basin Snowpack

Continuing with the theme of snowpack, here's an update on the conclusion of the snow season in the Chena Basin above Fairbanks.  Back in March I noted that the amount of snow on the ground was at near-record levels based on data from 5 SNOTEL sites in the hills, and it turns out that the snowpack did indeed reach an unprecedented level by late in April.  Historical data goes back to 1981.

As of a week ago, 3 of the 5 sites were reporting a greater snowpack (in terms of water equivalent) than in any other year on the same date.  Of course the major warm-up in the past few days has melted most of the snow at the lower elevation sites - Teuchet Creek (1640') melted out yesterday - but Mt Ryan (2800') still has 8" of water equivalent on the ground.  Even more remarkably, Munson Ridge (3100') is reporting 15.2" still on the ground (a record for the date) and a 47" snow depth.  The beginning of hiking season will be somewhat delayed in the White Mountains this year.

Here's a chart of snow water equivalent at Munson Ridge for this year and some other notable years.  The greatest snowpack of record occurred in 1991, but no other year retained so much snow at this late date.  It will be interesting to see how long it lasts; in 1982 there was still snow on the ground on June 13.

And here's an update to the multi-station chart that I showed before; the late winter snowpack has been much greater than in any recent year.  (The rapid rise in percent of normal at the end of April reflects the rapidly diminishing median value, as the median snowpack drops to zero in early May for 3 of the 5 sites.)

Tuesday, May 8, 2018

Snowpack Season

The winter snow cover melted out on Friday in Fairbanks, according to measurements by the National Weather Service at the airport.  This is somewhat later than normal, which is not surprising in view of the healthy snowpack at the end of winter and the rather cool temperatures of late.

The chart below shows the 88-year history of dates when the continuous winter snowpack began and ended in Fairbanks.  The variance of dates is greater for snowpack onset, but that's nearly all because of two major outliers in 1934 (the great December chinook) and 1992 (early arrival of winter).

The dashed lines show long-term linear trend lines for the two series of dates, but instead of least squares regression I've used quantile regression, with each trend line showing the best estimate of how the median has changed over time.  The reason to go with quantile regression is to avoid an undue influence from outliers, which seem likely to be a problem here, especially for the snowpack onset dates.

It's interesting to see that the regression estimates indicate that snowpack onset has become earlier by nearly a week, and meltout has become slightly (about 3 days) later, over the 88 years.  But it's important to note the degree of uncertainty: the 90% confidence interval includes a zero trend for both series, so neither trend is statistically significant; we can't confidently rule out the possibility that the trends are just a reflection of random chance.

One thing we can say with confidence, however, is that snowpack meltout has not become significantly earlier over time, and this is intriguing.  We know that April - when most of the melting occurs - has become significantly warmer, so how is the snow not disappearing more quickly?  There might be a variety of explanations involving changes to sunshine or precipitation during melt season, but the simplest explanation may be the best: it seems there is just more snow on the ground these days.  The chart below shows the snow depth in Fairbanks on April 1st each year, and there's an upward trend that is significant at p~0.05.

More investigation will be required to determine whether and how this trend is related to precipitation changes during winter; my impression from previous work is that Fairbanks winter precipitation has NOT increased over the long haul, so it seems this is a rather interesting question.

Thursday, May 3, 2018

Nenana Breakup

Tuesday was the big day for tripod watchers in Nenana, as the ice moved enough to stop the Ice Classic clock early in the afternoon.  May 1st is just about normal compared to recent decades, but it's interesting to note that the ice went out very "easily" when we consider how little warmth was available for thawing.  The normal (1981-present) accumulation of thawing degree days (excess of mean daily temperatures above freezing) by May 1 is about 170°F, but this year the total was barely half of that (88°F through May 1).

Here's an update of the figure I produced last week, with the observed (green) line extended to the right to show the forecast verification.  The last week or so before breakup turned out to be cooler than the previous week's forecast indicated for Fairbanks, and indeed the second half of April was significantly colder than normal in Fairbanks, but that didn't keep the ice in place for long.

The chart above shows that breakup occurred this year with cumulative TDDs not even reaching the 5th percentile of the historical range, and the scatter plot below confirms this point: only a small handful of other years since 1930 had so little thawing by the time breakup arrived.

As an aside, the relationship of TDDs and breakup date also reveals an interesting tendency for the latest breakups to occur after a relatively small quantity of thawing; in other words, the later the ice holds on, the less cumulative warmth is needed to produce breakup.  This undoubtedly reflects the increasing influence of solar insolation as the calendar advances; by mid-May, the strong direct influence of the sun on the ice accelerates breakup compared to when it would be expected from temperatures alone.

Another interesting aspect of how "easily" breakup occurred this year is that May 1 was one of the coldest days on record for the conclusion of the Nenana Ice Classic.  It's unusual for the daily high temperature in Fairbanks to be less than 50°F on breakup day, and breakup is usually late when this does happen; but this year the high temperature was only 43°F.  The chart below shows how unusual this is, especially compared to recent decades.

How can we explain the relative ease with which the ice moved out this year?  I think at least two factors were at play: first, the ice was very thin - barely over 24" thick at the end of March - owing to the combination of a thick snowpack and a warm winter.  Second, the ample snowpack across the region has produced higher than normal runoff, leading to increased pressure on the ice.  It's also possible that the rain and snow on Monday and Tuesday was the straw that broke the camel's back; Fairbanks reported nearly a quarter-inch of rain and snow (that soon melted).

Here's a webcam view of the scene at Nenana on Wednesday morning, with a dusting of fresh snow from overnight snow showers; the wreckage of the tripod sits in the broken ice on the left.

Finally, and for the record, it's fun to note that there was a trace or more of fresh snow on the ground at UAF's West Ridge for 3 consecutive mornings this week (Monday, Tuesday, and Wednesday) - see below.  Such are the travails of spring in interior Alaska.

Monday, April 30, 2018

Snow in the Hills

Spring suffered another setback in the interior today, as a round of steady precipitation brought the snow line down to valley-level in the Fairbanks area for a few hours.  Mixed rain and snow was reported from the airport during the morning hours and there was a brief threat of a fresh snow cover at UAF:

Snowfall of 2-4" was reported from the hills, according to the NWS, and temperatures have been below freezing all day above about 2000'.  The webcam views from Cleary Summit showed the Steese Highway becoming snow-covered between 5 and 6 am this morning:

However, the road surface cleared off by mid-morning (10:30 in the image below) under the influence of solar heating despite thick cloud cover and ongoing snowfall.

As Rick noted the other day, this spring is turning into something of a cool one relative to recent norms, although it's been more notable for the lack of warmth than the occurrence of unusual cold: there have been only 2 days above 50°F so far this year, compared to 12-22 such days by the end of April in the past 4 years.  This April's mean temperature has been below freezing in Fairbanks for the first time since the record cold April of 2013, but for most of the 20th century it was more common than not to have a sub-freezing monthly mean in April.

More chilly and damp weather, with snow in the hills and probably in town too, looks rather likely in the next couple of days.  It's unusual to see lengthy spells of rain and snow at this time of year - the chart below shows that the long-term frequency of rain or snow reaches its seasonal minimum in early May, based on hourly reports from Fairbanks airport since 1950.

This morning's 500mb chart shows a very typical set-up for wet weather in Fairbanks-land: vigorous flow out of the west-southwest in the middle levels of the atmosphere.

Saturday, April 28, 2018

When will it warm-up in Fairbanks? Adjusting Climatology for Trend

Hi, Rick T. here with some ideas on how we can use the historical record (i.e. climatology) in the face of significant trends. What I want to look at in this post are estimating threshold exceedance dates, e.g. first or last date in the season of the temperature exceeding some threshold. This is very commonly available for freeze dates, such as this plot of the probability of first date of freezing temperature at Fairbanks Airport in autumn (courtesy of the Western Regional Climate Center), but of course can be constructed for any threshold of interest. However, an underlying assumption for the use of this kind of information for planning or forecasting is that the past is a reliable guide to the future, i.e. there are no significant trends. 

The high temperature so far (through April 27) at Fairbanks Airport has been 54F. This is only slightly lower than average for the historical record, but is notably low for recent years: only four springs in the past 30 years have had a lower high temperatures at this point in the season. So when is going to really get warm? Or at least, when can we expect to get warm based on the usual annual cycle? We of course turn to climatology. The first thing we can do is construct a plot like the one above just based on the historical data. From this, we see that the median date of the first 70F is May 21, and any day between May 15 and May 25 would be pretty typical, while dates before May 7th or June 1st would be quite unusual.

However, when we plot the first 70F of the warm season as a time series, we see that the the first occurrence of 70F has been occurring, on average, earlier in recent years, as the the linear regression trend line slopes downward (and is statistically significant at the 95% confidence level), with the typical first 70F now about week earlier than in the 1930s.
So how could we adjust our cumulative distribution plot to account for the trend? If we assume that the variablity remains the same, and that looks like a reasonable assumtion in this case, then it's pretty easy:
  • First detrend the observed dates by, for each year, subtracting the regression value from the observed. This leaves us with the regression residuals. 
  • Next, take the regression estimate for the present (in this case May 16th) and add that back into the residuals. This in effect "converts" the historical record to the current climatology.
  • Last, construct a cumulative distribution plot on our "converted" data. 
When we do that, it makes a significant difference because of the trend:
So based on our present climate, we would estimate that there is about a 30% chance of having the first 70F temperature in Fairbanks before May 12th, whereas using only the historical record we would estimated less than a 20% chance. Similarly, the chances of getting through May without having a 70F or higher temperature is now less than 5%, half of the historical-only estimate. Is any of this meaningful? That of course depends on your needs.

Putting it all together, here is the final graphic, with the cumulation distibution plots for 60F, 70F and 80F temperatures adjusted to reflect the probablies of exceedance based on today's reality and not on a different past.

North Pacific Forecast

A couple of months ago I wrote a post on the very unusual warmth that the seasonal forecast models were expecting across the North Pacific this summer.  It was a long-range forecast at the time, but as summer draws nearer the models are not backing off.  The two maps below show a comparison of the February (top) and April (bottom) forecasts for June-August mean sea surface temperature departure from normal; the message is very similar, with the models expecting very anomalous warmth in the north-central and northwestern North Pacific.

As noted in the previous post, the forecasts are consistent with a strongly positive phase of the North Pacific Mode (NPM); here's a chart showing the NPM index values that the models are expecting.

 We've seen in the past that the NPM phase is correlated with winter precipitation patterns across interior and northern Alaska, with above-normal snowfall often occurring when the NPM is positive.  But what can we say about potential impacts for the next few months?  Looking at Fairbanks first, the charts below show the historical relationship of the NPM index with May-July mean temperature and total precipitation.

There's a slight but barely significant tilt towards cooler temperatures at this time of year when the NPM is more positive, but the precipitation chart is perhaps more interesting as it hints at a non-linear relationship.  It appears that the positive NPM phase tends to favor either very wet or very dry weather in Fairbanks, as the 6 driest years and several of the wettest years (in May-July) occurred with a positive NPM phase; but near-normal rainfall is somewhat more likely when the NPM is negative.

Recent years have borne out the wide variation in summer rainfall amounts during a positive NPM phase, as 2013 was very dry (and extremely warm), but 2014 brought record summer rainfall.  Both summers had a strongly positive NPM phase.

The map below shows the typical 500mb height pattern for positive NPM years when dry weather prevails in Fairbanks; not surprisingly the most common location for high pressure is over the eastern interior.

In contrast, years with a positive NPM phase but wet May-July weather in Fairbanks have the ridge axis located farther to the southwest, and importantly these years have an active trough over the Chukchi Sea; so this pattern favors strong westerly flow that brings wet frontal systems across the state.

From a North Pacific-wide perspective, the two patterns are rather similar, but the differences are crucial for rainfall in interior Alaska.  The maps below show the precipitation patterns associated with the two sets of positive NPM years.

So which pattern should we pick to accompany this summer's expected North Pacific warmth?  Well, the seasonal forecast models are clearly choosing the wet option, as a rather strong majority of the NMME ensemble members are showing significantly above-normal precipitation across western and northern Alaska.

The Climate Prediction Center is also going for wet, and the signal is a strong one for western Alaska; it's unusual to see 50+% probabilities of upper-tercile precipitation in the seasonal forecast.

There seems to be little reason to disagree with the strong model signal - it's usually unwise to do so - but just for fun I pulled up the May-July patterns that occurred in past years when Bering Sea ice was very low in the preceding winter (although nowhere near as low as this year): see below.  The sea-level pressure pattern (top map) supports the idea of low pressure over the Arctic waters north of Alaska, but there's also a high pressure signal over most of the state, and interestingly these years were more dry than wet in the interior.

There's obviously more work that could be done to pick apart the varying ways that North Pacific SSTs interact with Alaska's climate at this time of year, but I expect we'll learn something just from watching the pattern unfold in the next few months.  A wet summer seems more likely, but the historical NPM analysis suggests that it could go the other way - and so I think there is a bit more uncertainty about the forecast than the models and CPC would suggest.