Wednesday, November 15, 2017

Frequency of Snow

The scene in Fairbanks is wintry at last, with persistent light snow events in the past week bringing the snow depth up to 9 inches.  This is a little above normal for the time of year.

Beginning last Wednesday, 6 consecutive days produced measurable snowfall in Fairbanks; this seems a little unusual, but it's typical for a 6-day snowy period like this to occur at least once in a winter.  The record for consecutive days with measurable snow is 16 days in November 1994, and the next longest periods are:

14 days ending Oct 24, 1970
13 days ending Nov 24, 1988
12 days ending Jan 10, 1987
12 days ending Nov 8, 1996

It's a bit curious that prior to 1965 (i.e. 1930-1964), the longest streak of snowy days was only 10 days; but perhaps there was a tendency to overlook very small (e.g. 0.1") snow accumulations in the early years.

The fact that 3 of the longest 5 periods ended in November is consistent with the peak in daily snowfall frequency at this time of year.  The figure below shows a smoothed daily frequency of measurable snow (blue line) and also shows the frequency of snowfall when there was snow on the previous day (purple line).  As we would expect, the chance of snow is higher if the previous day was snowy.

Taking a quick look at webcams around the area, freeze-up is still not complete in Fairbanks, although the Tanana River at Nenana is ice-covered now.

Friday, November 10, 2017

Utqiaġvik Wind Events

Back in late September, Alaska's northernmost community of Utqiaġvik (formerly Barrow) experienced a damaging coastal flood event that washed away some lengthy sections of road and caused significant erosion to the coast.  The following article provides detail on the damage:

As noted in the article, a similar event happened in 2015; here's a blog post from the time:

Some of the coastal flooding that occurred in the latest event is evident in this shot from UAF's sea ice webcam:

Later that day the berms were repaired along the beach, but obviously the city's defences are rather fragile.

In view of these events, it's worth taking a look at some historical data to see if high wind events have become more common, and if so, by how much.  To do this I used gridded reanalysis data to estimate the sea-level pressure gradient at 6 hour intervals since 1950 for a grid point close to Utqiaġvik.   Of course the pressure gradient is closely related to the wind speed, so the idea is to use the pressure data as a proxy for estimating changes in wind conditions.

But why not just use the historical wind speed data from Barrow?  The reason is that there are too many uncertainties related to changing measurement practices over time - i.e. it's likely that changes in instrumentation, measuring height, and measuring procedure (such as averaging time), as well as a changing urban environment, have created artificial changes in the long-term wind speed data.  In contrast, the sea-level pressure field in a reanalysis should (by design) provide a consistent estimate over time, although not all reanalysis techniques are equally good, and of course a reanalysis is just a model.  Nevertheless I think the reanalysis data has more potential to provide a useful answer.

The chart below shows annual series derived from the 6-hourly pressure gradient from the NCEP/NCAR reanalysis at a grid point near Barrow.  The two horizontal dashed lines show 1957-2016 mean values and help to highlight subtle changes.  Note that data prior to 1957 is a bit suspect because the number of balloon soundings around the Northern Hemisphere was far smaller in earlier years.

It's interesting to see that there seems to have been a subtle upward shift in pressure gradient since about 2000, and for the middle series (the annual mean of the 12 monthly maxima) most years since 2000 have been above the long-term mean.  The 1957-2016 linear trend in the middle series is statistically significant (p=0.03).

Here's a similar analysis from the 20th Century Reanalysis, which actually goes all the way back to 1851.  Note the suspiciously high pressure gradient values prior to 1957 - this supports the idea that the early estimates are less good.

The annual mean pressure gradient values in the two data sets are correlated at R=0.77, which isn't too bad, although obviously there can be large differences in how the reanalyses handle individual events; the "annual maximum" series are correlated at only R=0.58, and the 20th Century Reanalysis seems to have a very different assessment of the event in 2013 (a blizzard in mid-January).

The 20th Century Reanalysis shows less of an indication of higher pressure gradient in recent years; there is still an upward trend, but it's not really statistically significant (p=0.08 for the middle series).

How about individual wind events that might tend to produce storm surge flooding in Barrow?  To look at this I pulled out all 6-hourly instances when the pressure gradient was above a certain threshold AND Barrow reported a wind direction between 250° and 360°, i.e. winds from west-southwesterly through northerly; we'll call these "flooding winds".  Strong winds from this direction would tend to pile up water on the coast.  Of course there is much more to creating storm surge than just instantaneous wind speed and direction, but this is a quick look.

The chart below shows the number of "flooding wind" hours each year in which the pressure gradient exceeded 3mb per 100km, which is about the 95th percentile year-round.  There was a notable lull in these strong wind events for about 15 years starting in 1976, and this year has been windy, but overall there seems to be little trend for the annual number of hours.

Looking just at the autumn season when shore protection from sea ice is reduced, the picture is a little different - see below.  Interestingly for this time of year it does appear that there has been a long-term increase in strong pressure gradient events that could cause coastal flooding, although the last decade or so has not been notably worse than say the 1990s.

The 20th Century Reanalysis again shows a slightly different result.  Here I've used a slightly lower pressure gradient threshold, because the values are systematically lower, but other than a couple of high totals in 1993 and 2002, the past few decades have not seen an unusual number of "flooding wind" hours.

In summary, the well-known NCEP/NCAR reanalysis provides some evidence of an increasing trend in pressure gradient and therefore presumably increasing winds year-round near Barrow, and the data also suggest that somewhat more strong westerly/northwesterly wind events have occurred in autumn since around 1990.  However, the 20th Century Reanalysis provides only marginal support for the idea, so in the absence of more information it's difficult to draw a firm conclusion.  Perhaps the most useful thing we can say is that there's certainly no evidence of a slackening in pressure gradients, and given the profound reduction in sea ice extent, the rate of shore erosion is likely to be higher than in the past.

As an aside, there's one other interesting aspect of the data that I noted: a pronounced increase in pressure gradient in February, see below.  The other months with the most notable increases were June and July, but November has seen a downward trend.

Saturday, November 4, 2017

Extreme Snowfall In Fairbanks

Hi, Rick T. here with a post on the frequency of heavy snowfall in Fairbanks, specifically the return period (or, as I prefer, the annual probability) of snowfall amounts. You're perhaps familiar with this concept as applied to rainfall or river flows, and is an important factor to be considering in many engineering applications. We don't see so much of this kind of analysis with regard to snow, though we should, e.g. the maximum expected snow load during the lifetime of a building is (or ought to be) an important design criteria in snow county.

To illustrate, I'll go through the procedure with the two-day snowfall.  For each snow season since 1919-20, we find the max snowfall total on consecutive calendar days.  This gives us an "extreme values" time series. This seasonal maximum has ranged from 2.6" in 1952-53 to 26.9" in 1965-66. Now note that this maximum two-day total can occur at any point in the season, though realistically, late September to May. While the highest value typically occurs in the November to January time frame, it can be much earlier or later. For instance, in 2015-16, the highest two-day total was 13.5" September 29-30, while in 1991-92, the highest two-day total of 9.6" occurred in mid-May. So putting this in graphical form, on the left is the annual time series. It is clear just from inspection that there is no temporal trend, which  simplifies the analysis. On the right is a histogram of those annual values. For many cold seasons, the highest two-day snow total is in the range of 5-9", but with a moderately long tail toward higher values.

Now we'll fit the data to a generalized extreme value distribution, and I plot the results as a function of return period, as in the graphic below. The observed seasonal maximum in this plot are rank ordered, the green line gives the best fit and the dashed lines are confidence intervals of the fit.

So from this you can pick off the return period (better: annual probability) for two-day total snowfall. So the 5-year return period (20% chance in any individual season) is about 12", while the 20-year return period (5% chance in any given season) is just over 18". If we do this analysis for various accumulation intervals, we get this (leaving off the confidence internals for simplicity):
For reference I've plotted the highest observed value in the past 97 winters. These are used in the fit and so are not independent, but this does give an idea of the (relative) rarity. So both the 7 and 14-day totals are a bit lower than the fitted 100-year return period, while 72.4" that fell in the 31 days ending January 25, 1937 is more unusual. A word of caution: the 24 hour return periods are only for the Weather Bureau/NWS era (since winter 1929-30) and, since c. 2001 there are no six hour snowfall observations. The only 24 hour snowfall totals currently available are the midnight to midnight, calendar day total and the 3pm-3pm AKST totals (which can be back-calculated from the end-of-day and 3pm AKST climate summary products). The 3pm-3pm totals are not regularly used by NCEI, so for the high end snowfalls since 2009 I've checked them myself. The only difference from the published values (which are maximum calendar day values) is for 2016-17, when the highest calendar day total is 10.4" (Dec 28) but the highest 3pm-3pm total is 12.4" (Dec 28-29).

Thursday, November 2, 2017

Missing Soundings

An article today in the Washington Post brought attention to a somewhat unsettling trend in NOAA's upper-air observing program at sites in Alaska.  In short, there has been a reduction in the number of balloon soundings being launched from several of Alaska's upper-air observing sites, and it is likely that this will have some (unquantified) negative effects on the quality of weather forecasts, especially for downstream locations.

The chart below verifies the issue, as the percentage of missing soundings from Alaska's 13 sounding sites has increased rather substantially since late summer.

Looking back at 2016 (see below), over 90% of days had all, or all but one, of the usual 26 soundings per day across the state, although there seems to have been something of an increase in missing data in the second half of the year.  Presumably the problem has been developing slowly over time, but for reasons cited in the article it has worsened considerably of late.

My perspective on the issue is that it's fairly inexcusable, even if the argument can be made that the coverage of 13 upper-air sites over Alaska is quite good compared to other high-latitude parts of the world.  Given the economic importance of forecasts for the lower 48 and Canada, high-density sampling of the atmosphere upstream over Alaska is really critical, and balloon soundings have tremendous value for numerical weather prediction.  Surprisingly, two of the sites that have been cut back are Utqiaġvik/Barrow and Kotzebue, and these are the only U.S. upper-air sites in the Arctic.

I would argue that just purely on a financial basis, it makes sense to at least maintain the existing sounding network.  Compared to the cost of satellite observing platforms, the value of the forecasts, and the cost of the computers to crunch all the numbers, the radiosonde network provides enormous value for money.

Wednesday, November 1, 2017

Out-of-Season Rainfall

My last post made mention of the plain rain (no snow) that fell in Fairbanks last Friday, and Monday of this week brought a second and heavier round of late autumn rainfall: 0.25" of liquid fell onto nearly bare ground accompanied by only a trace of snowfall.  It is very rare - in fact unprecedented in the Weather Bureau/NWS era (1930-present) - to have this much rain falling on essentially bare ground at this late date in the autumn.

Based on 2-day rainfall totals, the previous record for latest date with this much plain rain (trace or zero snow) falling on bare ground (trace or zero snow depth) was October 23-24 of 1981, when more than a quarter-inch of rain occurred together with temperatures as high as 49°F.  More recently, a similar amount fell over 3 days ending on October 29 in 2013.  My post from the time noted that the 2013 event was the latest on record for plain rain (any amount) with zero snow on the ground, and that record still stands as there has been a small amount of snow remaining on the ground this week in Fairbanks.

The chart below shows an attempt to visualize the historical occurrences of out-of-season rainfall on bare ground in Fairbanks.  I've used 2-day rainfall totals to capture events that span the midnight boundary, and the markers show all rain events from October 1 through May 15 in which (a) no more than a trace of snow fell, and (b) there was no more than a trace of snow on the ground.  Note that I required conditions (a) and (b) for both of the 2 days; for the sake of simplicity I've excluded events for which 1 of the 2 days had snow.

It's clear that plain rain on bare ground is (or used to be) rare after mid-October, and the only instances that occurred in November were in 1954 and 1962; both of these were light events.  However, 3 of the past 5 years have now seen significant "plain rain" events after mid-October.  Looking at the other side of winter, it's interesting to note that the past several years have also seen a few significant rain events relatively early in the year; in fact the top 3 early and late events have all occurred since 2010.

The top 3 separate rain-on-bare-ground events after October 25 are:

0.26"  October 27-29, 2013
0.25"  October 30, 2017
0.10"  October 27, 2017

and for spring the top 3 events prior to May 4 are:

0.33"  April 29 - May 2, 2010
0.31"  April 25-27, 2016
0.44"  May 2-3, 2016

For those who are interested, the sequence of maps below shows the progression of the 500mb flow in association with Monday's rain event.  The maps show the analysis at 4pm AKDT Sunday (top), 4am Monday (middle), and 4pm Monday (bottom).  Remarkably, most of the 0.25" of rain fell between 8am and 11am on Monday, or half way between the 2nd and 3rd maps - i.e. just as the vigorous upper-level low was moving into the Y-K Delta region.  Note the very long fetch of warm air being drawn up from the south around the ridge over the northeastern Pacific.  Not only was this an extremely warm air mass (hence the lack of snow), but it was moisture-laden and suitable for generating heavy precipitation.

[Update Nov 2]

Courtesy of Gary's comment and San Francisco State University, I've added below the 300mb wind maps from 4am and 10am on Monday.  Note the strong jet stream flow impinging on the southern interior.  The second map shows that Fairbanks was located close to the nose of the jet and perhaps a little to the left of the jet axis, at the time of the heavy rainfall; the "left exit" region of a jet stream maximum is a favorable zone for producing heavy precipitation.

Here's a 2-hour radar animation ending at 9am on Monday, courtesy of Plymouth State University.

Saturday, October 28, 2017

Thaw and Rain

Exceptional warmth for the time of year has developed across most of Alaska, putting a halt on freeze-up and even melting away the early snow cover in places in the past couple of days.  Most of the state has seen temperatures above freezing, including parts of the North Slope today; 37°F was observed on the Sag River at 69°N this afternoon.

On a statewide basis the "heat wave" is among the most significant observed this year when compared to normal temperatures for the date.  The chart below, courtesy of Rick Thoman, shows a daily temperature index for 25 sites around the state; the values (ranging from -10 to +10) correspond to percentiles within the historical distribution.  Since April there have been far more days in the upper tercile (red dots) than the lower tercile (blue dots), and yesterday's index value was the second highest of the year after June 8.

In addition to the warmth there has been rain: one-tenth of an inch of plain rain in Fairbanks yesterday, with nary a trace of snow falling.  In Tanana it has been raining steadily since mid-morning today (over 8 hours straight now), and temperatures aren't even particularly close to freezing; the dewpoint reached 37°F this afternoon, which is the highest on record for this late in the year - in fact it's the highest dewpoint observed there between October 19 and March 21 (based on data from 1950-present).

Here's the rather miserable scene in Tanana this afternoon: rain falling on a thin snow cover, with moderate ice moving in the Yukon River.

With the temperature in Fairbanks getting up to 42°F on Thursday, there is now only a trace of snow on the ground in the Golden Heart City.  Based on the long-term history, less than 20% of years have no real snow cover in Fairbanks on or after this date, but this year makes it 4 out of the last 5 years (2013, 2015, 2016, 2017).  The scene this afternoon on the UAF webcam - with more green than white - is not normal for October 28.

On another topic, the webcam shot is of interest for another reason too - notice the bright sun dog to the east of the sun.  Sun dogs are a rather common optical phenomenon caused by refraction and scattering of light through flat hexagonal ice crystals, generally (as in this instance) in high cirrus cloud.  The image also shows a hint of a 22° halo above and below the sun dog.  These phenomena are not unique to the northern latitudes, although the prevalence of ice crystals especially near the surface in the cold season does give rise to far more interesting optical phenomena than are commonly observed in warmer climates.

Thursday, October 26, 2017

New ECMWF Seasonal Model

As the snow melts away this evening in Fairbanks-land under the influence of chinook flow (with temperatures generally in the 40s), long-range forecasters are watching the tropical Pacific to see if the incipient La Niña episode will continue to strengthen.  If it does, then we would not expect abnormal warmth to be the theme of the winter in southern and interior Alaska; instead, some notable cold episodes would be rather likely, but variability would also be high - perhaps not unlike the recent swing in temperatures.

A new tool that forecasters will be using this winter is an upgraded version of the seasonal model from the well-known European Centre for Medium-Range Weather Forecasts (ECMWF).  In addition to having better physics and higher resolution in both the atmosphere and ocean simulations, the system now includes a coupled sea-ice model, leading to a much better representation of inter-annual variability and trends in ice conditions.  In my view the lack of a sea-ice model in the previous version was a rather major shortcoming.

In preparation for the release of the model upgrade (version 5), I've been looking at the performance of the model using the historical retrospective forecasts that are provided for bias correction and calibration.  The figures below show summary statistics (correlation coefficient) for forecasts of sea surface temperature in the equatorial Pacific (Niño3.4 region) and in the North Pacific (PDO and NPM indices); I'm showing results for the NMME models as well as versions 4 and 5 of the ECMWF model.  (See this post from last year for some background information.)

It's encouraging to see that ECMWF's version 5 model is better than version 4 for the PDO and NPM forecasts.  Also, the ECMWF's PDO and NPM forecasts are now generally better than any of the individual NMME models, and for the PDO at 1-3 month lead times the ECMWF is even better than the NMME ensemble mean.  Oddly, however, my results show no improvement in the Niño3.4 forecasts.

I also looked at forecasts of several of the most important atmospheric teleconnection patterns that meteorologists tend to monitor, including the PNA (Pacific/North American) pattern and the EPO (Eastern Pacific Oscillation).  Both of these atmospheric circulation patterns are closely connected to Alaska's winter weather; for example, the 4-panel of maps below shows the 500mb height and surface temperature patterns associated with the positive PNA (top) and negative EPO (bottom) phases, both of which tend to bring unusual warmth to most of Alaska.

The opposite PNA and EPO phases are shown below (negative PNA on top, positive EPO on bottom).  To appreciate the significance of these two patterns, consider that the November-March PNA and EPO index values (which are actually independent of each other) jointly explain about two-thirds of the November-March temperature variance in Fairbanks; so we can "predict" the Fairbanks winter temperature with a mean absolute error of only 2.0°F if we know the PNA and EPO index values.

In view of the significance of these atmospheric "modes", it's of interest to see whether the models can predict them on seasonal time scales.  The leftmost sets of columns in the charts below show that the answer is yes, and the PNA and EPO forecasts are actually quite good compared to several other teleconnection indices such as the North Atlantic Oscillation and the Arctic Oscillation.

It's also encouraging to see that the ECMWF's upgraded model is considerably better at predicting the EPO pattern, and this bodes well for seasonal winter forecasts in Alaska.