Saturday, November 17, 2018

Alaska Observation Palooza



Hi, Rick T. here. In searching back through Deep Cold posts, it looks like we've never put up a review of a subject near and dear to my heart: an overview of point-based weather and climate observations in Alaska. So, I thought it might be worthwhile to lay out some of the details. Note: this piece is acronym heavy. I've included a decoder table at the end.

When it comes to weather observations: right now here in Alaska, we are living in the golden age of real time environmental observations: the good old days were definitely not always good. When I started with the NWS in Alaska in 1988, for most of the year there were less than four dozen reliable, real-time 24-hour/per weather observations in the entire state, and some of those (ones with the asterisk in the graphics below) only had temperature, wind and pressure for part, or all, of the day. Every single one of these were operated by either then NWS, FAA or DoD.

There were more part-time or irregular weather observations the 1980s. There were perhaps 20 contract aviation observations (paid for by NWS or FAA) that took 6 to 16 observations a day, e.g. Umiat, Slana and Ambler. By the late 1980s the DoD dew line stations were no longer regularly reporting 24 hours per day but sent observation on an occasional schedule. There were a tiny number of NWS operated remote stations (RAMOS) that reported temperature and winds and were still operating in places like Anaktuvuk Pass and Sitkanak (southern Kodiak Island). There were some, but not usually real-time observations from a small number of the Soil Conservation Service (now NRCS) SNOTel sites. In the summer there were even more, from Alaska Fire Services RAWS (which were deployed and then removed at the start and end of the fire season) and from relatively few river gauges which had temperature and tipping bucket sensors.

Fast forward to today: a quick count of observations in Alaska available on the MESO-WEST website for 3am Saturday morning revealed about 785 separate observations with at least one meteorlogical element, and to that you could add more than a dozen more home weather stations that are online only on Weather Underground. Here's a graphic I made up in 2017 that identifies the operators and types of observation in the eastern Interior and Copper River Basin.
This shows most of the classes of 24 hour per day weather observations that are currently deployed around Alaska, with the exception of some coastal specific observations. You'll see that most of the platforms belong to some Federal agency (e.g. NWS, FAA, BLM, NRCS) but there are some others, including state of Alaska DOT and private (the home weather stations). The stations marked as USARRAY are an interesting class. These are actually part of a temporary, high density seismic monitoring network on which weather sensors have been installed. Only a  fraction of the stations have weather stations included (the full map is here). These stations have provided never before seen real-time weather detail over the North Slope and Brooks Range, you can as see in this graphic (same 2017 project):


On the climate side of the house, it's a different story. The March 1988 edition of the NCDC publication Climatological Data had data from about 130 stations in Alaska reporting daily temperature extremes and precipitation data, and this represents the vast majority of climate observations that were made that month in Alaska. About 75 of these were NWS cooperative sites which made observations once a day and the rest NWS and FAA (including the contract observations) and DoD sites. In March 2018, there were just over 100 stations with data in that publication, and NWS cooperative sites was still number about 75. The losses are mostly from the drop off of the aviation contract and DoD sites: almost none of those became strictly cooperative stations.

The big difference beween 1988 and 2018 is that now there are multiple sources of readily available climate data that are not included in the Climatological Data publication. These include explicit daily climate data from unaugmented NWS ASOS (e.g. Kaltag and Seldovia) and the nearly two dozen NCEI flagship Climate Reference Network stations as well as derived climate data from the 100+ RAWS (mostly operated by the Alaska Fire Service and National Park Service) and four dozen NRCS SNOTel sites. Now one can question the climate value of some of these platforms, e.g. RAWS, which we know report temperatures that are too warm during high-sun, low-wind situations to due insufficient thermistor shielding, but clearly, Climatological Data is no longer the definitive source for Alaska climate data. Rather, it's perhaps the definitive source for NWS cooperative data, only some of which is reported in near real-time, and of course this does (eventually) make it into online climate sources, e.g. the NWS NOWData and scACIS.

But that's not the end of the story. Since 2010 the FAA has upgraded nearly all of the early 1990s era AWOS, the first generation automated operational weather observing systems that included visibility and ceiling height, with modern automated equipment which is functionally very similar to the NWS ASOS, as well as installing this modern equipment at at airports that previously lacked any weather observations. There are about 80 of these these FAA owned and operated systems that report temperatures and precipitation in the same way as ASOS, including sites with long histories of climate observations as well as sites that have never had climate observations before. While these systems don't handle frozen precipitation, the temperature data is reliable, but at the moment none of this is being used systematically in most climate analyzes. It's not true that it's been tossed out: the data is all archived at NCEI and other online locations, but we're not making much use of it. Here's a plot of these stations climate data we're not using:


As a sidenote, I do keep track of the daily data from several of these location that have historic climate data (Kaktovik, Unalakleet, Gambell, Ft. Yukon) and use it in my work. So while we have many, many more weather observations nowadays, climate data has not expanded, and in fact over much of northern and western Alaska we have less now than we did 30 years ago.

Acronyms:
  • ASOS  Automated Surface Observing System: the NWS standard, requires commercial power
  • AWOS Automated Weather Observing System: the first generation FAA station, did not report climate data, required commercial power 
  • NRCS: Natural Resources Conservation Service: Successor to Soil Conservation Service, part of the US Department of Agriculture. 
  • RAWS: Remote Automated Weather Station: the workhorse of fire weather. Run on solar or wind power, often installed at higher elevations 
  • SNOTel: Snow Telemetry: NRCS station widely in the West and Alaska primarily for to measuring snow pack but now includes in a variety of meteorological parameters. 
  • USARRAY: US Array: a 15-year program to place a dense network of permanent and portable seismographs across the continental United States. Installed in Alaska 2015-17, scheduled for removal starting in 2019. 



Thursday, November 15, 2018

Warm Winter Ahead?

Today the U.S. Climate Prediction Center came out with their seasonal forecast for winter, i.e. December through February, and to probably no-one's surprise they are showing a rather high chance of above-normal temperatures in Alaska in the coming months.


By way of reminder, the CPC forecast is probabilistic and aims to predict the chances of each of the three terciles of the 1981-2010 distribution; so a 50-60% chance of "above normal" (as for most of Alaska above) means that the probability of the upper tercile is 50-60% rather than the baseline normal probability of 33%.  This is a big shift in the probability distribution and a very strong warm signal; it's the first time that CPC has shown such a large area of 50+% warmth in Alaska for Dec-Feb.

Here's the corresponding precipitation forecast.


My blog co-author Rick will be presenting much of the rationale behind the forecast in his regular monthly webinar tomorrow, and listeners are guaranteed to learn something even if they're already familiar with the complexities of seasonal forecasting.

https://accap.uaf.edu/November2018

Not to steal Rick's thunder, but I'll suggest just a few reasons why a very warm forecast seems reasonable if not inevitable.  First, sea surface temperatures to the west and south of Alaska have been extremely warm in recent months, so regardless of how the weather pattern plays out, there is a lot of residual heat available.  The map below (click to enlarge) shows that October SSTs were more than 3 standard deviations above the 1981-2010 normal near Alaska, according to NOAA's ERSST data.  This is a pretty extraordinary warm anomaly.



Second, El Niño has developed in the tropical Pacific Ocean.  As we speculated back in July (see here), this El Niño episode is focused in the central rather than eastern tropical Pacific, so it's a so-called Modoki El Niño.  It's interesting to consider the different implications for Alaska of having a Modoki rather than East Pacific El Niño; I've found that the results are a bit sensitive to the definition that's used, but the overall message seems to be that Modoki episodes favor warmth more widely across the Bering Sea - Alaska region.  East Pacific El Niño's are more closely linked to warmth in eastern and southeast Alaska than in the rest of the state.  But in any case, an El Niño winter is most definitely a warm signal overall.

Third, the long-range computer model forecasts are showing pronounced warm signals for the upcoming winter.  Here's a sample.

The NMME multi-model ensemble mean anomaly: note the +3-4°C anomaly near the Bering Strait.
 The NMME (non-calibrated) tercile probability forecast:

The UK Met Office ensemble mean anomaly:

And the Japanese seasonal model - see below.  It's a little hard to see the coastlines, so I've circled Alaska in red.  The top left panel shows 500mb height, indicating a southerly flow over Alaska; the middle left panel shows 850mb temperature - note that the model expects the warmest conditions in the Northern Hemisphere (relative to normal) over Alaska.  The bottom left panel shows sea-level pressure - note the Bering Sea trough.



Finally, the map below shows a statistical forecast based on sea surface temperatures that I developed at work recently; this is an ensemble mean forecast from a number of models that have undergone cross-validated historical testing to determine the optimal predictors.  Note that the baseline is the 1950-2017 trend, so the forecast for Alaska would be even warmer relative to the standard 1981-2010 climatology.  The skill of these statistical forecasts is modest at best, but they're worth having because they provide independent guidance to complement the computer models.


So is there any contradictory guidance showing a cold or even a normal winter?  Having looked at a very broad collection of predictors in the past couple of weeks, the answer is "almost none".  The only hint I found was in looking at past years in which late summer and early autumn produced strong and persistent high pressure ridging over the northern North Pacific, similar to this year.  The subsequent winters had a modest tendency for unusual high pressure over the Bering Sea, which is a cold pattern for Alaska (see below); but the signal is not very striking, and in any case most of the "analog" years are taken from earlier decades that were colder to begin with.

Tune in to Rick's talk tomorrow for much more detail and considerably more expertise related to the winter forecast!



Tuesday, November 6, 2018

North Pacific Blog Post

It's been a while since I did one of these, but I've just posted an update on the North Pacific "blob tracker" blog.

https://alaskapacificblob.wordpress.com/2018/11/06/record-north-pacific-warmth/

I'll aim to add some interpretation for the Alaska climate scene on here in the coming days.

Tuesday, October 30, 2018

Cooler At Last

The air temperature is much more seasonable across interior and northern Alaska today, courtesy of a strong cold front that pushed its way across the state on Sunday.  Here's yesterday morning's 500mb map, showing the associated upper-level low over the North Slope (click to enlarge).



Fairbanks has seen a high temperature of "only" about 17°F today, but this is completely normal for the time of year.  It's quite a dramatic change, however, as this is the first day with a sub-freezing high temperature, and prior to Saturday the coldest day had a high of 37°F.

The sudden change to colder got me thinking about the statistics of sudden "one-way" temperature drops in autumn or early winter in Fairbanks.  (And let the reader beware - what follows is a rather arcane discussion.)  By "one-way" I mean an instance when daily high temperatures drop below a threshold that was not breached earlier in the season, and then the temperature fails to rise back above that threshold at any later date in the same year.  For example, as noted above the coldest day in Fairbanks until Saturday had a high of 37°F, but it's possible that we've seen a "one-way" change to colder, as 37°F may not be reached again until the spring.  This would constitute a "one-way" drop in high temperatures at the 37°F threshold.

Digging through Fairbanks climate data reveals that this kind of thing is fairly common as a result of the high rate of seasonal cooling; about 30% of all years see a "one-way" change to colder high temperatures at some point in the autumn.  The most dramatic was in 1950, when up until September 26 the coldest day had a high of 46°F, but no temperature above 38°F was observed from that date on; that's the largest one-way drop (8°F) that I found.

Here's a chart showing the high temperature thresholds and dates for which these "one-way" cooling events have occurred in Fairbanks.  Unsurprisingly, the earliest was in the absurdly cold September of 1992; the latest was in 1942, when the temperature sank below 22°F at the end of October and did not reach that level again in the same year.  (Note that I haven't included winter after December 31 in this analysis.)



Looking at a number of other sites in Alaska, the frequency of sudden one-way cooling is highest for the most continental sites and lower for more maritime sites, as we would expect.  For example, Anchorage sees such an event in about 20% of years, but Bettles and Northway manage it more often than not at some point in the autumn.

On a related note, it's worth pointing out that Fairbanks has never seen a true one-way temperature drop in which subsequent daily high temperatures never reach the lowest daily low temperatures prior to the drop.  I rather doubt if this has been observed anywhere in the world, as it would require both an extreme rate of normal seasonal cooling and an extreme shift in the weather pattern at the right time.

To finish up, here's a pleasing clear-sky scene from UAF this evening just after sunset.  Officially the snow cover is still only a "trace" - it hasn't made it up to 1 inch yet.

Wednesday, October 24, 2018

Excessive Warmth

There's a lot that could be said about the ongoing persistent and increasingly extreme warmth (relative to normal) across much of Alaska, but time doesn't permit more than a brief update at this point.

To illustrate, Fairbanks has seen 3 days in the past week with a high temperature of 50°F or higher - and to put this into context, the 1981-2010 normal high and low temperature for October 24th are 25°F and 10°F respectively.  Fairbanks doesn't normally see more than a small handful of days above freezing from this point on, so 50°F is quite extreme.

Today was actually the warmest day in Fairbanks since the first week of the month, with a high of 52°F.  This degree of warmth has not occurred so late in the season since the 1930s: it happened in 1934 (the year of the great December chinook) and also in November 1936 and late October 1938.  But similar conditions occurred in 2013, of course, with 51°F on October 28.

The map below shows the culprit for today's balmy conditions: a deep southerly flow aloft, leading to downslope warming to the north of the Alaska Range.  As we would expect, the middle Tanana River valley was warm and windy today owing to the chinook flow; Delta Junction reported wind gusts over 50mph.


Here's the snowless scene this evening on campus at UAF: a disheartening prospect for winter enthusiasts as well as, more seriously, residents who rely on frozen land and water for important activities.


Here's a comparison of this month's daily mean temperatures to the two other years that were outstandingly warm in Fairbanks in the month of October: 2013 and 1938.  No other years have come close to having the same amplitude of warm anomaly for this calendar month.  But neither of those years could hold off winter's freeze until the very end of the month; and so surely it is just a matter of a few days now until the landscape is finally transformed.


Saturday, October 20, 2018

No Rain in Yakutat

Hi, Rick T. here with an analysis of one piece of exceptional September we just came through. Specifically, the lack of precipitation on the Gulf of Alaska Coast. One of the most amazing climate statistics to come out of September was from Yakutat, where there was an astonishing 20 consecutive days with no rain at all, not even a sprinkle. Considering that Yakutat averages 24 days in September with some rain and that this streak was several days longer than any previous dry streak this time of year, the question is, just how unusual was 20 straight days with no rain at all?
Richard, Brian and I had a long email string on ways to think about analyzing this event. With their help, here's what I came up with.

First, some background to set the stage.
  • Since  I am looking at days with zero precipitation, not even a trace, I used data since 1947, since this is entirely within the Weather Bureau/Weather Service era of 24-hour per day observations and there is no missing daily precipitation data. 
  • The frequency of precipitation varies seasonally, e.g. May in Yakutat averages twice as many days without any precipitation as October, so we need to limit the analysis to this time of year. Therefore I confined the analysis to the early autumn (August through October) season. I'm also assuming there is no trend in dry days streaks (which is the case for the total number of dry days in ASO).
  • For statistical analysis, the independence of events is often an important underlying assumption. So while it's easy to generate simple counts of consecutive days without precipitation, it took a bit more work to find the independent streaks. To illustrate this, a simple count revels that there are two streaks of 19 days with zero precipitation during August through October, 1947 to 2018. However, both of these streaks are simply subsets of the 20-day streak from this past September (Sep 2-20 and Sep 3-21). So removing all the streaks that are simply subsets of longer ones, here's what we find for the counts of independent, non-overlapping streaks of specific lengths:
So we see that in the past 72 years there have been only 12 independent streaks that were more than a week long during the early autumn, and only four streaks longer than ten days. We want to get a climatological handle on such unusual events so we can answer questions like: how unusual was this? Could it happen again?

There are a number of ways to potentially answer such questions, but the one I'll provide here involves our old friend, regression. But rather than linear regression (which obviously is not appropriate), I tried mathematical forms that have the potential to represent what we see in the plot above: large and rapid changes as we move from left to right along the x (horizontal) axis. Two commonly used forms for distributions of this shape are exponential and power law. In order to facilitate this analysis I first converted the raw count values into frequencies per year and then plotted the frequency on a log scale, which results in this:
This is the same information as in first figure, just displayed in a different way. But it allows us to immediately to see that that an exponential fit is not likely to work so well. How do we know that? Well, with the y (vertical) axis plotted in log scale, an exponential fit will appear as a straight line. Just eyeballing the top of the bars, you can see that a straight line will fit pretty well for streaks of 11 days or less, but then fails to capture the handful of events longer duration. For that, a power law fit works out better. Now a well established issue with power law fits is that the often only part of a distribution (typically the right tail) is well represented by a power law. How does that work out in this case? I systematically fitted a power law using the observed frequency of all the streak lengths, i.e. 1 day to 25 days (everything about 20 is zero). Then I fitted a power law to streaks of two days or longer, then three days or longer, etc. The "winner" was the fit that had the lowest root mean squared error but still utilized most of the data (there are more sophisticated ways to do this but I have not had the time to implement them, though in this case will lead to the same answer).  It turns out that the best fit was for runs to two or more days and it looks like this:
So based on this analysis, the streak of 20 completely dry days in row has only about a 0.7% chance of occurring in any particular August through October season. I've noted the return period as calculated from the fit on the graphic for selected streaks, though I don't really like to do that because it's easy to misinterpret. Why do it? People like to see it, and in principle it is perhaps a more intuitive way to express low likelihood events. But, it is important to remember that a long return period is the inverse of a very small number, and so small changes in the fit result in big changes in the return period. So if I improve this analysis and come up with probability for a 20-day streak as 0.9% in any given year, that's a small change from 0.7%, but the return period would drop by 40 years, to 111. 

So from this analysis, the 20 dry days in a row at Yakutat this September was likely a once in a lifetime event, at least if you're of mature years. After all, 0.7% annual chance of occurrence means that, assuming no change and no year-to-year correlation, that there is about a 30% that this will happen at least once in the next 50 years. 
  

Friday, October 19, 2018

Bering Ridge Wrap-Up

As a postscript to my earlier analysis of the intense ridge over the Bering Sea and western Alaska (see here and here), the charts below indicate the magnitude of the recent 500mb anomaly at 15-day, 30-day, and 45-day time scales compared to the Northern Hemispheric extremes since 1958.  As before, we're looking at the standardized 500mb height anomaly, i.e. the departure from normal divided by the standard deviation, and I've removed the (seasonally-varying) long-term linear trend at each location.  The charts show the daily maximum and minimum values of the standardized anomaly across the entire Northern Hemisphere since 1958.  Click to enlarge the images.




Based on this analysis, the upper-level ridge that affected western Alaska in recent weeks was most anomalous on a 30-day time scale.  Remarkably, the 30-day standardized height anomaly at 500mb just to the south of Nome was the most extreme in the global reanalysis history back to 1958, for either Northern or Southern Hemisphere, and for either positive or negative anomalies.  Here's a map of the peak 30-day height anomaly:


In the earlier post I noted that the record 15-day ridge was an extreme high-pressure block over northern Greenland in November 1965.  But interestingly the record event for a 45-day time scale was also over Alaska, in the late winter of 1989.  This event appears to have been related to a major disruption of the stratospheric polar vortex (a "sudden stratospheric warming") in February 1989.


Finally, here's a chart showing Southern Hemisphere extremes on a 45-day basis.  The record for most anomalous ridge occurred near the southern tip of South America in the early austral winter (late April - early June) of 2016.