Saturday, January 26, 2019

Satellite Radiation Data

In many previous posts, Rick and I have commented on the seasonal changes in solar radiation that are such a key part of understanding Alaska's climate, and I've also discussed the importance of "longwave" radiation, for example:

As a reminder, "longwave" radiation is the infrared radiation that is constantly emitted and absorbed naturally by all objects, including blog readers and their immediate surroundings at this very moment.  This is in contrast to "shortwave" radiation, which is only emitted in significant quantities by very hot objects like the sun.  The earth's climate system is fundamentally driven by the ever-changing spatial distribution of shortwave radiation from the sun, but longwave radiation transfers between the surface and atmosphere (and out to space) are also a critical part of the energy balance.

I recently took some time to acquire a very nice satellite-derived radiation data set from NASA's CERES project ("Clouds and the Earth's Radiant Energy System").  Based on data from polar-orbiting satellite instruments, CERES provides global coverage with radiation data ranging from hourly to monthly time scales.  With this, we can take a look at how the radiation energy budget works in Alaska and surrounding areas.

First, here's the most tangible element of the radiation transfer - the incoming shortwave at the surface - in mid-winter and mid-summer.

There isn't much to say about winter, but the July map highlights just how cloudy it is across a huge area of the North Pacific.  It's also interesting to note the relative maximum in available solar energy in July over the high Arctic owing to 24-hour daylight.  Of course this is part of the reason for concern over ice-albedo feedback in the Arctic Ocean; as ice extent decreases in summer, the albedo (reflectivity) of the ocean surface decreases dramatically and more solar energy is absorbed.  Here's a map of July average albedo in the CERES data.

As a result of high albedo in the recent climate, the Arctic Ocean surface absorbs less solar radiation than surrounding land areas, so the net shortwave gain is relatively low:

Now let's look at longwave radiation, which is less intuitive but equally important for understanding the climate.  First, the downward longwave; this is the infrared radiation emitted by clouds, water vapor, and other radiatively active gases in the atmosphere like carbon dioxide, methane, and ozone.  The amount of longwave radiation emitted by the atmosphere is closely tied to the average temperature and absolute moisture content (e.g. precipitable water), so a warm and humid air column emits much more than a cold, dry one.  Over Alaska, the coldest and driest atmospheric column is over the Brooks Range in January, but more extreme cold is found over eastern Siberia as well as the high Canadian Arctic.

The upward longwave emission from the earth's surface is just a direct reflection of the average temperature of the surface - see below.  The relative warmth of the far North Atlantic (the Norwegian and Barents Seas) in winter is pretty striking.

Adding up the downward and upward longwave at the surface gives the net longwave radiation.  The surface loses longwave energy on balance, but much more so in places (a) where it's warmer and so there is more longwave going around, and (b) where clear skies and dry air tend to prevail (more upward than downward emission).

Finally, putting all the pieces together to get the total net radiation balance yields the maps below.  As expected, the high latitudes lose energy via radiation on average in January, with the largest losses found in relatively warmer regions, i.e. over the oceans (even if ice-covered).  But note the relative maximum near the pole, where upward emission is low (thick ice cover and very low surface temperatures) but cloud and moisture aloft (probably derived from the North Atlantic influx) provides a bit more downward radiation than we might expect.

An interesting aspect of the July map above is that much of mainland Alaska gains less energy from radiation overall than the Gulf of Alaska and parts of the Bering Sea; I would not have guessed this, because I would have focused on the sunshine part of the equation.  In reality the ocean surface is colder than the land in summer, so longwave emission is reduced over the water; and the summer ocean has a very low albedo that allows it to capture most of the available shortwave.  Evidently these two factors more than compensate for the higher cloud cover over the ocean - so even though it's cloudier, the ocean near Alaska captures more energy from the summer sun than does the land.

And for a final map, the annual average net radiation gain at the surface.  Apparently the region very near the pole actually gains a small amount of energy over the year - another surprising result.

With nearly two decades of CERES data now in hand, I'll plan to take a look at seasonal and year-to-year variations in Alaska in a subsequent post.

Sunday, January 20, 2019

Sea Ice Update

The past six weeks have seen some fairly persistent cold weather over the Bering Sea and western Alaska - much more so than over the rest of the state, where prior to the recent cold spell the winter had been relatively warm.

In response to the cold in the west, sea ice has expanded in the Bering Sea, and the ice area exceeded half a million km2 just the other day.  In the 1981-2010 period, this milestone was typically reached just after the new year.  So the ice growth is still a bit behind "normal", but it's a big recovery from last year.

NOAA's latest daily sea ice analysis shows the ice pack extending to the south of Nunivak Island but not yet reaching St Matthew Island farther to the west.

The NSIDC analysis is very similar.

The chart below shows the recovery off last year's record-low ice extent in the Bering Sea - we already have more than last winter's peak extent - but we're nowhere near the high ice coverage of just 7 years ago.  The Bering ice extent of 2011-2012 was the highest in the modern satellite record for most of the mid-winter and meltout period, so sea ice has been highly variable in the Bering Sea over the last decade (to say the least).

The NSIDC provides ice extent data for other regions of the Arctic basin, so it's interesting to compare the long-term trends in both March (seasonal max) and September (seasonal min).  See the chart below, showing the 40-year linear trends in percentage terms, i.e. the trend divided by the 40-year mean ice extent for each region.  Of course some of the regions melt out completely every September, and many freeze up completely every March, so those cases have undefined and zero trends respectively.

Obviously the changes in ice extent have been much more dramatic at seasonal minimum, but nevertheless there have been highly statistically significant decreases in March ice extent in some of the basins - notably in the North Atlantic sector (e.g. the Greenland Sea and the Barents Sea).  However, it's the Chukchi Sea that has the dubious distinction of the greatest percentage loss in September - see below.  (Note that while the Bering Sea loss is also very high in September, the actual numbers involved are very small; another viewpoint would be to look at the absolute change in ice extent over time.)

Saturday, January 12, 2019

Cold Frequency This Century

The cold snap is on its way out across Alaska now, as Pacific low pressure approaches from the southwest and a blanket of clouds aloft provides a warming influence.  McGrath saw the temperature rise from -50°F this morning to "only" -18°F this evening as the sky turned overcast.  However, severe cold is hanging on across the north, with -40s in the Yukon flats, and Umiat is currently sitting at -52°F on the North Slope.

Here's a chart showing some statistics for winter cold in Alaska since the turn of the century.  The black markers show each winter's statewide minimum temperature, and the columns indicate the number of daily reports in which the daily minimum temperature reached -40°F and -50°F.  This is based on data from NOAA's ACIS tool, which includes daily temperature data from about 400 sites (but closer to 300 at the beginning of the century).

The frequency of extreme cold has been on quite a roller coaster, with (for example) -40° or colder showing up somewhere in the state on 89 separate days in winter 2011-2012, but only 25 days in 2015-2016.  (But this ignores the confound of mismatched observation times, such as morning coop reports.)

See here for an earlier post about statewide minimum temperatures on a longer time scale:

Wednesday, January 9, 2019

Cold Update

Here's a quick update on the colder reports so far in this notable but not yet extreme cold spell that is reminding much of mainland Alaska what real winter feels like.

Fairbanks airport made it down to -39°F and had a couple of days with midnight-to-midnight high temperature below -30°F, but cloud and light snow brought temperatures back up to near zero today.  Elsewhere in the central and eastern interior it's mostly colder, with -40s hanging on in Tok and the most sheltered valleys of the Fortymile region; and fresh, very cold air is working its way down across western Alaska, with lots of -40 reports showing up late today.  Click below to enlarge.

As is often the case, Chicken has been the "winner" for cold, with a -56°F low temperature reported as of Monday morning.

The hourly data from the Fortymile River HADS site at the Taylor Highway showed similar numbers for Monday, with afternoon temperatures rising just a couple of degrees to -50°F.  (As an aside, note the elevation: 1558 feet; this is part of the reason why the narrow valleys of this region get so cold.  Chicken is at 1800 feet.)

Other sites that have dipped below -50°F are Tok and Fort Yukon.

As impressive as these numbers seem in comparison to the last year or so, we don't have to go back even two years to see colder conditions - the map below shows the minimum temperatures on January 18, 2017 (courtesy of  But the next few days may rival this - will someone see a -60°F for the first time in several years?

Saturday, January 5, 2019

Colder Statewide, Snowy December in Bettles

A deep winter chill has descended upon much of Alaska this weekend, with temperatures dropping well below zero in the usual cold spots.  So far Fairbanks airport has dropped to -34°F, which is colder than anything that occurred last winter.  Of course it is a rare winter in Fairbanks that doesn't drop to at least -35°F at some point.

A few other cold spots today are:

Chicken  -49°F
Fortymile River HADS   -43°F 
North Pole  -39°F
Salcha RAWS  -38°F

The Fortymile River HADS site (on the Taylor Highway) did not rise above -40°F today, and with the cold expected to deepen somewhat in the next few days, I'd be very surprised if Chicken doesn't crack -50°F.  Only two winters in Chicken's climate record (1996-present) have failed to reached -50°F, and -60°F or colder occurs in about half of all winters.

Alaska's snow pack is below normal in many places, but one area with plenty of snow is the upper Koyukuk region; observers from Bettles up to Wiseman are reporting 30" or more of snow on the ground (click to enlarge maps below).

Remarkably, Bettles had its snowiest calendar month on record in December, with 62" of snow.  The typical annual snowfall in Bettles is close to 90" - fairly high for an interior valley location - but nevertheless 60" in a month is a lot.

A scatter plot of December snowfall and average temperature in Bettles shows a slight positive correlation, as the coldest months tend to be drier than normal; this is partly because the coldest Arctic air masses hold little moisture in winter, and partly because clear skies (associated with high pressure and dry weather) produce sustained temperature inversion and cold conditions at this time of year.

As we've noted before, the situation is quite different in Fairbanks owing to the effect of dry chinook winds from the south.  The proximity of the Alaska Range means that the warmest months, which are dominated by southerly flow, also tend to be dry, and so the snow-temperature plot shows a peak in snowfall at near-normal temperatures.