Tuesday, October 28, 2014

Thin Snowpack Survives

The snow cover that arrived in Fairbanks on October 4 failed to melt out sufficiently to bring the official snow depth below 1 inch, and so we can now say the likelihood is very high that the permanent winter snowpack was established on October 4.  This is tied with 1933 for 3rd earliest in the Weather Bureau/NWS era; the only years with earlier onset of permanent snow cover were 1992 (September 13) and 1956 (October 2).

It is an interesting curiosity that the snow depth was reported at 1 inch for 11 consecutive days ending October 23, and this is tied for fifth longest such period (consecutive days at 1" snow depth).  However, it's nowhere close to the record: in 1953, the snow depth remained at 1" for a remarkable 48 days ending December 9.  The chart below shows the years with at least 11 consecutive days at 1" along with the number of days in the period with a high temperature above freezing.  Most often a lengthy spell of 1" snowpack is accompanied by few or no days above freezing, but this year 8 of the 11 days saw the temperature rise above freezing.  However, we should note that the Fairbanks snow depth measurements are now taken (I believe) at the #2 airport location, which is sheltered by vegetation and often runs several degrees cooler than the official temperature site; only 5 of 11 days rose above freezing at the #2 location.

On another note entirely, here's a webcam shot of shallow steam fog over the Tanana River at Nenana early on Sunday morning, with an air temperature of 4 °F at the airport nearby.  The tendrils of mist form when air that has been warmed and moistened in contact with the water surface mixes with colder air a few feet above the surface.  A lot of interesting microscale physics was occurring within that shallow layer of intense temperature gradient!

Monday, October 27, 2014

Sub Zero Temps

Fairbanks nearly recorded their first sub-zero temperature of the season this morning. If they had, the date would have been only 1 day behind schedule. Figure 1 shows the low temperatures this morning. The first 0°F of the season at the official climate site for Fairbanks has occurred as early as October 3rd and as late as November 22. Figure 2 shows the annual date of the first 0°F day since 1904. Finally, Figure 3 shows the earliest 0°F observation for each climate station in the greater Fairbanks area for stations with at least 15 years of observations that extend into the 2000s. Every station has experienced 0°F temperatures before the end of October. North Pole is the winner with a 0°F reading on September 26, 1983.

Figure 1. October 27, 2014 low temperatures from the University of Utah's Mesowest site.

Figure 2. Date of first 0°F temperature for Fairbanks.

Figure 2. Earliest date of 0°F temperature for all GHCN stations around Fairbanks with at least 15 years of data ending no earlier than 2000.

Sunday, October 26, 2014

Point Barrow Freeze-Up

This is just a brief post to note the arrival of widespread sea ice around and north of Point Barrow in the past few days.  The NWS-Anchorage sea ice analysis from Friday showed a large gain in ice cover that connected the shore ice to the Arctic pack for the first time:

It's interesting to observe that cooler temperatures developed quickly at Barrow in the wake of the sea ice formation; the low was 4 °F yesterday, and a significant low-level temperature inversion was observed for the first time since September 16 (see yesterday's 3am and 3pm soundings below).  Prior to sea ice formation, strong surface warming from the adjacent ocean waters tends to produce a steep low-level lapse rate (warm below, cold above) in the lowest levels of the atmosphere, but the heat source is reduced after most of the nearby ocean surface is frozen.  These changes are evident in the climatological vertical temperature profile as discussed in an earlier post here.

Saturday, October 25, 2014

Fairbanks Forecast Performance - Part 2

In an earlier post I began looking at the performance of NWS temperature forecasts for Fairbanks, with a particular focus on whether the forecasts show enough of a "signal" at the end of the short-term forecast period.  On average through the year, the forecast errors at Day 7 are about 20 percent smaller than they would be if the forecast just called for "normal" every day, so the forecasts are clearly useful even out to Day 7.  But do the forecasts show "enough" departure from normal or are they too heavily weighted towards climatology?  The first post showed that the scaling is about right; the NWS forecasts are close to optimal in this regard.

There is more analysis that we can do, however, if we bring in the computer model forecasts and compare them to the NWS forecasts.  For this purpose, I've extracted the GFS and ECMWF computer model forecasts of 850 mb temperature for every day since mid-August 2013 (when I started collecting the data).  The NWS forecasts tend to track with the 850 temperature forecasts, as we would expect, but the following chart shows a hint of something interesting (detailed explanation is below):

The chart shows the average of the Day 7 temperature anomalies (departure from normal) predicted by the two models on the x-axis, and the error of the Day 7 NWS forecast on the y-axis; and the chart only shows days when the model anomalies have the same sign and agree to within 4 °C.  So I've excluded many cases when the models disagreed, because I'm attempting to isolate what happens when the models agree reasonably well.

There is a lot of scatter, of course, and the overall correlation is very weak, but notice the frequency of points above the horizontal zero line when both models expect very cold conditions; the NWS forecast tends to be too warm (not cold enough) in these cases.  On the right-hand side of the chart, there are far fewer cases with comparable warm anomalies in the model forecasts, but in the top five events it seems the NWS was too cold (not warm enough).

My interpretation of the results is that the NWS forecast has a tendency to be too conservative when both of the leading computer models agree in predicting a very large temperature anomaly.  If both models are very cold, then the NWS forecast ought to be lower; and if both models are very warm, then the NWS forecast ought to be warmer.  The conclusion is tentative because of the scatter in the data, but it does make sense: when the two independent models both show a large signal, then this considerably raises the chance that something very unusual will occur; and it seems the NWS forecast anomaly should be amplified accordingly.

For comparison, it's interesting to look at the same charts using the two models individually, see below.  When either model by itself shows a large cold anomaly, there is no obvious bias of the NWS forecasts, although the data on the warm side still suggests an error pattern in the most extreme warm events.

What do I conclude from this analysis?  A general conclusion - and one that is well known - is that having access to independent model forecasts is very useful for assessing the likelihood of extreme events.  This is obviously one justification for running model ensemble systems such as the GFS ensemble forecast, but using a completely independent system like ECMWF provides even more valuable information.

The more specific conclusion is that there is some potential to improve the Day 7 temperature forecasts in Fairbanks when the GFS and ECMWF forecasts are closely aligned in showing a large temperature anomaly.  In other words, the degree of agreement between the models is itself a useful predictor and should be part of the forecast process.  Each model by itself has limited skill at day 7, but when the models line up, then this sends a signal that predictability is higher, and the forecaster would do well to pay attention.

Thursday, October 23, 2014

Disappearing Sun; Barrow Update

Today marks the day when the sun's angle above the horizon at solar noon has declined half way from the equinox to the winter solstice; or equivalently we have traveled three-quarters of the way from the summer to winter solstice in terms of the sun's elevation at noon.

What does this mean?  It means we hereby enter the dark third of the year in the northern hemisphere; and of course this fact is more inescapable the farther north you go.  We can illustrate the lack of solar energy across Alaska in winter by calculating the theoretical solar insolation under clear skies.  From this theoretical standpoint, the total solar energy received in Fairbanks over the next 4 months is less than is received in 6 days in the height of summer.  In Bettles the dark third of the year receives less radiation than in 4 days in summer.  However, in Anchorage the winter sun provides the equivalent of about 10 days in summer.

Here's the view at close to solar noon yesterday from the Alaska Climate Research Center webcam on UAF West Ridge.  The weakness of the sun is illustrated by the fact that some snow remains on the ground despite most days getting above freezing in the past two weeks; the official snow depth in Fairbanks has been at 1 inch for 10 days now.

On another note, Barrow has had a chilly and very windy time of it in the past several days, with a very strong pressure gradient importing cold Arctic air from the northeast.  Here's the surface analysis from Monday afternoon when winds were sustained at about 40 mph for a time.

The high temperature on Tuesday was only 17 °F in Barrow, which is the coldest day so early in the season since 2002.  As we've mentioned many times before, Octobers since 2002 have been extremely warm in Barrow compared to previous decades, and so this kind of chill would have been completely normal in the last century.  For example, the 1930-2000 normal for coldest high temperature to have occurred by October 21 in Barrow was 10 °F; and a high temperature of 17 °F would normally have been observed by October 12.  In 1996, when sea ice was firmly established from Barrow eastward by late September, the high temperature was -7 °F on October 11!

So, it's been a little cooler in Barrow in the past few days - but even this is only approaching normal from earlier decades.  October 2014 is still running well above both the 1981-2010 and 1971-2000 normals.

Wednesday, October 22, 2014

Anchorage Forecast Performance

This is a follow-up to Richard's excellent post on Fairbanks' forecast performance. What I want to focus on is the comparison of the official forecast to both climatology and persistence. Unlike Richard, I have not been proactively saving forecast products. Instead, I utilize the Iowa State text product finder. One of the forecast products is called the State Forecast and represents point forecasts for first-order stations. The product is issued twice daily and includes a minimum and maximum for the next 7 days (morning issuance) and 6.5 days (afternoon issuance). Figure 1 shows a sample State Forecast product.

Figure 1. State Forecast product issued by the Anchorage NWS Office on October 22, 2010.

In Figure 1, you will notice a series of minimum, maximum, and precipitation probability forecasts. Unfortunately, only the Anchorage and Juneau offices issue State Forecasts; hence, our analysis will focus on Anchorage (I know, not very Deep Cold). A choice must be made as to whether to use the morning or afternoon issuance. In this case, we used the afternoon forecast product, which effectively gives us a 6 day forecast. This option provides the shortest time window for assessing the Day 1 forecast (best case scenario) and makes the day-to-day comparison more meaningful. It also allows for a baseline to conduct a persistence forecast comparison. On the negative side, it eliminates Day 7 as a forecast period.

Forecast temperatures vs. Actual Temperatures

When looking at days 1 through 6, we see decreasing skill in the temperature forecast. This is not surprising, as we would expect this result for every forecast issued anywhere in the world. The question is how much value do we get from the forecast compared to another method. In Figures 2 and 3, we see the 2009-2013 difference between the forecasted temperature and the actual temperature for Day 1 through Day 6. Figure 2 is a summary by year and Figure 3 is a summary by month.

On both charts, we see decreasing skill the farther out in time we get; however, the actual forecast always exceeds the no skill forecast (climatology). The most dramatic forecast skills are in the winter months when the skill for Day 1 through Day 3 is especially high. A noticeable drop off is observed by Day 4. We see in Figure 2 that on average, the Day 1 forecast provides 3°F of improvement versus climatology and the Day 6 forecast provides 1.3°F of improvement.

A caveat regarding 2009 and 2013 in Figure 2. Those years had very large temperature anomalies and so the forecast skill for those years suffered accordingly.

Figure 2. Difference (absolute value) between forecasted temperature and measured temperature for Day 1 through Day 6 in Anchorage, Alaska, between 2009 and 2013. All months are aggregated for each year.

Figure 3. Difference (absolute value) between forecasted temperature and measured temperature for Day 1 through Day 6 in Anchorage, Alaska, between 2009 and 2013. All years are aggregated for each month.

Forecast temperatures vs. Climatology

If you had no access to television, radio, or the Internet, one option for generating a 6-day forecast is to predict that each day will be exactly normal; i.e., just use the numbers from the NCDC normals table. As it turns out, climatology appears to be a factor in the NWS forecast. As a forecaster, you would feel comfortable predicting that Day 1 is 20°F above or below normal based on the numerical models is the situation was warranted; however, the comfort level with a forecast of 20°F above or below normal for Day 6 is much reduced. Therefore, the forecast is tempered somewhat by trending it toward climatology. Figures 5 and 6 show the difference between the forecasted daily temperatures and the NCDC published temperature.

Looking at Figure 4, the Day 1 forecast is slightly more than 5°F different than the published normal temperature. However, by Day 6, the forecasted temperature is 3.7°F difference than the published normal temperature. In Figure 5, we see the breakdown by month. In every month and in each year, the forecast trends toward the climatological daily normal.

Figure 4. Difference (absolute value) between forecasted temperature and the NCDC published normal temperature for Day 1 through Day 6 in Anchorage, Alaska, between 2009 and 2013. All months are aggregated for each year.

Figure 5. Difference (absolute value) between forecasted temperature and the NCDC published normal temperature for Day 1 through Day 6 in Anchorage, Alaska, between 2009 and 2013. All years are aggregated for each month

Which Forecast is the Best?

So how do the point forecasts for Days 1 through 6 compare to a no skill forecast? For this analysis, we add a second type of no skill forecast called persistence. This is where you forecast that the temperature tomorrow will be the same as the temperature today. This can be extended all the way out through Day 6. When we do this, the results are shown in Figure 6.

We see that over the course of out 5-year period, the NWS forecast for the Anchorage International Airport is off by slightly more that 2°F (see Figure 3 for an NWS forecast breakdown by month). This is 1.1°F less (better) than a strict persistence forecast and 3.3°F less (better) than a climatology forecast (see orange line in Figure 3 for a climatology breakdown by month).

By Day 3, the climatology forecast catches up with the persistence forecast. Looking out to Day 6, the NWS forecast sill exceeds climatology by 1.3°F. The maximum differential between the NWS forecast and the no skill forecast is at Day 3.

Due to high variability of temperatures from year-to-year, it is impossible to assess the relative forecast improvement over this short time period. However, there is a large skill improvement when using the NWS forecasts for temperatures as compared to the alternatives.

Figure 6. Difference (absolute value) between actual temperature and three forecast methods for Day 1 through Day 6 in Anchorage, Alaska, between 2009 and 2013.

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.