Monday, October 20, 2014

Fairbanks Forecast Performance

For some time I've been meaning to take a look at the long-term performance of the National Weather Service temperature forecasts for Fairbanks, and particularly with one question in mind: do the forecasts show enough variance at the end of the short-term forecast period, i.e. 5-7 days in the future?

The question is motivated by the idea that sometimes the computer models indicate a pronounced temperature anomaly from about a week in advance, but the early NWS forecasts for the same time show only a small departure from normal.  A recent example was seen in the early October cold spell, when the ECMWF and GFS deterministic forecasts of September 29 both showed a notable cold anomaly in place by October 5, but the NWS forecast for the high temperature on October 5 was 38 °F, only 3.6 °F below normal.  In this case, as time went on and the forecast became more certain, the forecast dropped and the observed high temperature was 31 °F.  However, there are many cases when the computer forecasts are badly wrong from 7 days out, and so it is entirely justifiable for the official forecast to show only a small anomaly at longer lead times.  Indeed, it would be most undesirable for the raw model forecast to be reflected in the official outlook, because the numbers would often swing wildly from day to day.  The question is, does the NWS have the right balance?

It's possible to answer this question using a history of NWS forecasts that I have collected for Fairbanks airport since November 2011.  First, here is the basic "skill" of the forecasts for lead times of 1-6 days, i.e. the forecasts for "tomorrow" through "6 days from now".  Averaged over all seasons, the average error of the high and low temperature forecasts is similar and rises from just over 4 °F to nearly 8 °F over the six days.  Not surprisingly, the errors are much larger in winter, but it is interesting to see that the winter low temperature forecasts improve more significantly at shorter lead times, whereas the winter high temperature forecast error remains over 7 °F even for "tomorrow".

Here's a similarly-formatted chart showing the bias of the forecasts, i.e. the mean difference between the forecast and the observed temperatures.  Negative values indicate that the forecasts were too cold on average.  We see that the winter high temperature forecasts have been several degrees too cold on average in the past 3 years, even at shorter lead times, but the bias is much smaller for the low temperatures.  It would be interesting to investigate this further in search of a possible explanation.

Let's now consider the scaling of the temperature forecasts.  I've examined this by calculating the mean absolute error (MAE) that would result if the NWS forecast anomaly (departure from normal) were multiplied by values ranging from 0 to 2.  On the low end of this range, the forecasts would deviate very little from climatology and the forecast would just show normal values each day; but on the high end, the forecasts would show greater deviations from normal than they currently do.  The chart below shows the results of this experiment for day 7 temperature forecasts from all seasons of the year.

The data from the last 3 years show that (on average through the year) the high temperature forecasts are perfectly scaled at day 7, i.e. there is no way to improve the MAE by arbitrarily reducing or increasing the forecast anomaly.  We conclude that the NWS shows just the right amount of variance on average in the day 7 high temperature forecasts; this is not to say that we can't improve on any given forecast using additional information, but we can't reduce the error by simply adjusting the departure from normal across the board.

The day 7 low temperature forecasts are not quite optimally scaled, according to these results, as the NWS shows marginally too much variance.  In other words, the forecasts would be marginally (but only very slightly) better if they showed smaller departures from normal.

There is one other aspect of the problem that interests me, and that is whether we can show that the forecast variance is too small when the computer models show a large anomaly (as opposed to any size anomaly) and/or when the computer models agree with each other.  I'll return to this idea in a subsequent post.

Thursday, October 16, 2014

Barrow Temperature Update

The midpoint of October has been reached, and the mean temperature so far this month in Barrow is 25.4 °F, which is right in the middle of the pack in terms of the "new normal" since 2002.  Of course it is well above normal relative to the climate in former decades.

The chart below shows the daily mean temperature anomalies in Barrow between August 1 and November 30, for 2014 (black line) compared to the past 12 years, which were all very warm in September and October.  As we noted here, the anomalous warmth in recent years has tended to peak at the end of October; we can see that daily temperatures were above the 1981-2010 normal almost all the time in the second half of October in the past 12 years.  This year we've been following the same script since mid-September, and with sea ice still a good distance offshore, there seems to be no reason to expect a sudden change (see the ice analysis below, courtesy of the NWS in Anchorage).

Tuesday, October 14, 2014

Snow Pack Onset

Fairbanks has reported at least 1" of snow on the ground every day since October 4th. Today makes 11 consecutive days with measurable snow depth. However, the total for today was down to 1" and the forecast for the next few days suggest that it might dip below 1". If that occurs, the snow pack onset date will the date when the snow depth exceeds 1" again and stays that way for the rest of the season. Figure 1 shows the snow pack onset date for Fairbanks.

Figure 1. Date where snow depth stayed at or above 1" for the remainder of the season.

In is not unheard of for snow to be measured for at least 10 days and then drop below 1" before being reestablished some number of days later. Here are the instance when it occurred:

In 1919, at least 1" of snow was on the ground from October 1st through October 11th (11 days), but then the snow went below 1" for all but one day until October 29th.

In 1940, at least 1" of snow was on the ground from October 20th through October 30th (11 days); the snow pack was established for good on November 8th.

In 1949, at least 1" of snow was on the ground from October 9th through October 24th (16 days); the snow pack was established for good on October 31st.

In 1951, at least 1" of snow was on the ground from October 6th through October 25th (20 days); the snow pack was established for good on October 27th.

In 1955, at least 1" of snow was on the ground from October 7th through October 22nd (16 days); the snow pack was established for good on October 24th.

In 1972, at least 1" of snow was on the ground from September 29th through October 16th (18 days); the snow pack was established for good on October 19th.

In 1981, at least 1" of snow was on the ground from October 4th through October 14th (11 days); the snow pack was established for good on October 27th.

Monday, October 13, 2014

Arctic Ice Minimum - Part 2

This is a follow-up to last month's post on the seasonal Arctic sea ice minimum, which occurred on about September 17 based on spatial extent.  It's worth looking also at estimates of sea ice volume, thickness, and age, to see how this melt season compared to recent years.

First, volume: the University of Washington's Polar Science Center uses a numerical model to estimate Arctic sea ice volume based on various observations and calculation of known physical processes.  There is obviously a good deal of uncertainty in the volume estimates, but the results are a lot better than nothing.  The chart below shows the monthly mean volume estimates for April and September since 1979 (blue and red lines) along with the difference, i.e. the estimated volume of melt from April to September.

We see that the estimated volume in September increased this year for the second consecutive year and reached a level similar to 2009.  However, the volume remains very much below the volume of earlier decades, and the volume in April 2014 was comparable to the lowest values of recent years.  The relatively small difference between April and September this year indicates that the amount of melting this summer was significantly smaller than in recent years; in fact the melt volume was less than the 1981-2010 average.  It will be interesting to see if this translates into any recovery in April ice volume next spring.

The PIOMAS estimated ice thickness is shown in the following chart from U-Washington.  It's interesting to note that the estimated thickness this summer was considerably higher than in the past four years, which are closely clustered together at the bottom of the chart.  If the model is correct, this suggests that ice thickness remained very low during the 2013 recovery in ice extent; and but this year has seen a recovery in ice thickness, although total extent remained similar to 2013.

Finally, the estimated sea ice age compared to last year, courtesy of NSIDC:

The extent of 3+ year-old ice appears to be similar to last year, but there was more second-year ice this year in September.  This indicates that a significant fraction of last year's first-year ice (formed in the winter of 2012-2013) survived through its second summer, and this appears to be perfectly consistent with the volume and thickness changes shown above.

Saturday, October 11, 2014

Freeze-Up Progress

After a much colder start to October (first 10 days) than in the past five years, freeze-up is getting under way on area rivers a little early this year.  Through October 10, Fairbanks airport has seen 37 freezing degree days, compared to a 1981-2010 median of 9.5 by this date.  The total of over 3 times the normal for the date sounds like a significant anomaly, but owing to the typically rapid drop-off in temperatures at this time of year, it is only 4 days ahead of normal.

Here are a few webcam images of freeze-up progress at various locations, starting with today and going back about a week:

Tanana River at Nenana today:

Koyuk River from Koyuk today:

Koyuk River on Thursday Oct 9:

First widespread ice on Teshekpuk Lake, close to the Arctic coast, on Tuesday Oct 7:

 Teshekpuk Lake the next day, Wednesday Oct 8:

First ice cover on the lake at Inigok, between Umiat and Tesh Lake, last Saturday Oct 4:

And finally, a grainy shot of Toolik Lake (near the haul road just north of the Brooks Range) freezing over last Friday Oct 3:

Here's a chart showing the lake and air temperatures during the freeze-up of Toolik Lake.  I'm not sure of the depth of the temperature sensor in the lake, but it shows nicely how the temperature stabilized at about 2 °C in tandem with the freeze-up.  No ice was evident on the lake until this temperature threshold was reached, but since freezing began there has been only a tiny amount of additional cooling.  This nicely illustrates the heat exchange processes that are so important in the freezing of fresh water lakes around the world.

Friday, October 10, 2014

Fairbanks Temperature Trends - Part 3

In a couple of recent posts (here and here) I looked at changes in mean seasonal temperature in Fairbanks since 1950 and compared the changes to several nearby stations and to conditions at 850 mb.  Reader Gary asked if changes in surface winds might have contributed to the surface temperature changes; this is a plausible idea because wind over the Alaskan interior in winter disturbs the semi-permanent surface-based inversion and raises temperatures near the surface.

Unfortunately for this study, I'm not at all confident that historical surface wind speed measurements were made accurately and consistently enough at Fairbanks to allow an investigation of long-term trends; part of the problem is that the averaging time and measurement height of surface wind observations can vary widely.  However, balloon soundings from Fairbanks have measured wind speed aloft since 1948 and - in my view - are more suitable for this kind of analysis.  In earlier decades the balloon's position was tracked with radio direction finding or radio navigation aids, and now of course it's done with GPS.

To begin with, I interpolated the balloon-measured wind vectors to various heights above ground, so that a consistent average could be obtained for each height.  This is necessary because the sounding data is not reported at standard heights except for various "mandatory" pressure levels such as 850 mb.  After interpolating, and excluding any soundings from non-standard reporting hours, I calculated the December-February mean wind speed at each height for each winter since 1950-51.  See the chart below.

The black line shows the mean wind speed at 850 mb and suggests that the average wind speed at that level may actually have decreased since about 1975.  However, at each of the lower levels there was a significant jump in mean wind speed in the winter of 1990-1991, if the data are to be believed.  Interestingly the wind speed has trended down since then at the lowest levels (100-300 m AGL), but at 500 m the winter wind speed has remained elevated.

It's not quite clear what to make of these results.  It certainly looks as if low-level wind speed and mixing have increased over time in Fairbanks, although the 100 m wind speed has dropped back in recent years to levels similar to pre-1990.  The puzzling aspect is that 1990-91 does not show up as being particularly significant in the temperature time series (see below), although the inversion was certainly weaker than normal that winter.  With apparently higher mean wind speed throughout the 1990s, we would expect to see a jump in temperatures, but actually it was colder than in the decade following the 1976 PDO shift.

Here's another view of the change in wind speed at 500 m: the chart below shows the changing frequency of wind speed categories.  The long-term trend and especially the change after 1990 is quite striking.  It is tempting to blame all of this on changes in instrumentation or perhaps the vertical density of observations (which could affect the interpolation calculation), but neither of these changed significantly in the vicinity of 1990.

More investigation will be required to examine the physical causes of the apparent change in 1990 and to establish whether it is reflected in any other meteorological data.  As a first, simple step, I plotted up the differences in sea-level pressure, 500 mb height, 850 mb temperature, and sea surface temperature between the pre- and post-1990 winters (see below).  Lower pressure over western Alaska and more southerly flow over the interior is certainly consistent with warmer temperatures, greater mixing, and increased low-level wind speed, but it's not yet clear if the weather pattern changes were in any sense tied to 1990.

Wednesday, October 8, 2014

Rapid Autumn Transition

The hardy residents of the Fairbanks area are used to rapid seasonal changes and dramatic temperature swings, but the change in the past 3-4 weeks has been remarkable even for the usually volatile interior climate.

As we noted here and here, a strong chinook event raised the temperature to 76 °F at Fairbanks airport on September 14, which was close to the warmest on record so late in the season, but only 3 weeks later there was 4 inches of snow on the ground with a high temperature of 31 °F.  Given that the peak normal high temperature in early July is 74 °F, we might call this a transition from summer-like to wintry conditions in only 3 weeks.

Is this a record for rate of transition?  Based on these specific criteria (last 75+ °F day, first sub-freezing day with 2+" of snow on the ground), yes.  See the chart below; click for a larger image.  It's interesting to note the break from the relatively uniform results of the past 18 years.

We can show a similar result for Keystone Ridge.  The highest September temperature of 66 °F did not quite reach the peak normal for summer, but the daily mean temperature on the 13th was above peak normal for summer.  Looking at the time from last 65+ °F day to first wintry day (same definition), we again see a transition that was shorter than in recent years (only 16 days this year).

We don't yet know for sure if the snow cover in Fairbanks is the permanent winter snowpack (although it's looking increasingly likely), but if it is, then we would tie the record for shortest time from last 70+ °F to arrival of the snowpack.  The record is 20 days and was set in 1974 and 1992.  The record for shortest time from last 75+ °F to snowpack is 25 days, set in 1965.