It turns out that a thing to do in Fairbanks when it drops below minus 40 is to get a selfie or usie in front of the UAF welcome sign with the temperature readout… after stripping down to something you might wear on a tropical beach. Nanook, the UAF mascot posted this one on UAF’s facebook page, on the occasion of groundhog day. If you are more social-media connected than I you’ve probably seen a bunch of these. So it does not appear to get too cold to go outside in shorts.
Before delving into the other question, here’s a screenshot showing 8am temperatures from around the state on February 2nd, in the last days of a ~2-week cold snap more severe than we’ve had in a few years. Click to enlarge.
The cold was not restricted to the interior. A surface high was parked right over the middle of Alaska, pushing the cold down into Bristol Bay, Southcentral and somewhat into the Yukon and even the panhandle. The temperatures on the north coast weren’t that extreme but look at the winds! Low wind chills, along with blowing snow for the east part of the coast. Atypical that Utqiagvig and Dillingham match at -6F (-21C) and 10 knots (16 km/hr) of wind. Not shown on this zoom level, but Kenai was a chilly -27F (-33C) and Seward dropped to -3F (-19C) late in the day when the north winds eased from 30 to only around 20 mph (50, 30 km/hr), not including gusts. At the Cordova airport, calm wind allowed a low of -8F (-22C), though you can bet there were windy areas nearby, such as closer to Prince William Sound or the Copper River. Which is worse, in your opinion, the super cold calm, or the just zeroish cold and strong wind?
OK already, is it ever too cold to snow?
Theoretically, it could snow at even the coldest surface temperatures found in Alaska.
Read on for the reasons, and for the answers to “Is it ever too cold to snow more than a trace? An inch or more?”
The common answer given for why snow is less common or absent at really cold temperatures, is the lack of moisture in the air at those temperatures. There is an important kernel of truth in this: The colder the air gets, the less likely any water content will remain in it as vapor. It is destined to be squeezed out along the way into either a cloud and/or precipitation, or fog, or dew or frost, or when it is really cold, directly into ice crystals. But what is happening down at the Earth’s surface, where we live and have our thermometers and are observing the snow, is not necessarily what is happening five, ten thousand or more feet above us where the snow actually comes from. So is it too cold to snow at -40F/C? Well, if you knew it was only, say,-5F (-21C) up where the snow factory is, would you still be asking the question?
But in reality — to continue in the probability mindset — even just a measurable amount of snowfall (0.1 inches, or, a better, slightly tougher standard of 0.01 inches of water equivalent – about 2.5mm or 0.25mm) at a -40F/C or colder surface temperature is very rare, and to get an inch or more in 24 hours would be about as common as a white raven. But not for the usual reason given above, again, because where the snow usually comes from (way above the surface) can often be much warmer than the surface. The temperature of the snow’s final resting place is not the determining factor. But here’s the better reason it almost never can happen: The kind of weather pattern that makes it really cold at the surface is divorced from the kind of weather pattern that makes it snow any significant amount.
Example of “snow” at extreme cold?
Snow was reported on Feb 1 & 2 at Fairbanks International for ~18 hours without a break, at temperatures between -29 and -49F (-34 & -45C). So, is this the very rare snow at extreme cold? The first thing to know is that although it fell for 18 hours — along with “mist” and “freezing fog” most of the time (read ice fog) — there was apparently never enough even to round up to 0.1 inches, the smallest so-called measurable amount. The station did however, in the 1st hour of the snow, report the minimum measurable precipitation (the liquid in what fell), 0.01 inches. That barely meets our criteria for measurable precipitation, but that part happened at “only” -35F (-37C), so it just misses the magic -40 extreme cold goal. Insignificant for “practical, real world purposes” but let’s examine the situation.
The recipe for extreme cold in Alaska is this, 1) a generally cold atmosphere as a jumping off point, which usually means a cold upper low positioned to give northerly or westerly flow to the area in question. 2) A dry, clear atmosphere to allow radiational cooling, which normally comes via the sinking air of low level high pressure. 3) minimal solar radiation. No problem there.
The recipe for snow often also includes an upper low, but positioned to give a more southerly fetch of upper winds to bring in moisture. This often brings in clouds in advance of any snow, and those clouds alone usually result in radiational warming at the surface, meaning by the time the snow arrives it’s already warmed some. Then often the surface high pressure weakens, and/or moves off to allow a low or a front to move in, bringing some wind to stir some warmer air down to the surface. All these things weigh against the probability of snow and extreme cold at the same time. But sometimes things work out to get the snow in aloft before the surface cold leaves. But that really was not what was happening on Feb 1 & 2, 2024, (but it did happen about a week earlier on Jan 28, which I’ll detail below.)
On the night of Feb 1-2 the surface map looked like this at 6 pm (3 hr model forecast which is about as good as any analysis):
Notice the moderate 1033 mb high over eastern Siberia with a lobe extending into Alaska. Lows around it are sucking that frigid air onto the coastal areas as mentioned above. The high is squashing all but very shallow clouds. The dashed blue lines indicate the 1000-500mb thickness which is equivalent to the mean temperature in the lower 1/2 (by mass) of the atmosphere. Numbers under 500 are very cold. You can see several contours colder than that centered over the golden heart city. Blue and green shading are the model’s calculation of where precipitation should be happening: basically nothing over the vast torso of Alaska. The model was right “for practical purposes” but this is when it was snowing in Fairbanks. The narrow trough over the Brooks range may have triggered off the snow in this case, but the model didn’t think it would, apparently. [graphic courtesy Tropical Tidbits].
The view from above
Nothing like seeing with your own eyes… and this satellite image is basically that. Beautiful day across Alaska. On the left side you can see the Seward Peninsula, Y-K delta and adjacent ice-covered Bering Sea. In the Lower right the Alaska Range, Talkeetna Mountains and Chugach Mountains really stand out, as does Prince William Sound in the corner. Many of the major rivers are clearly visible including the Yukon, and the Kuskokwim and the Tanana, where Fairbanks lies near the middle of the red oval. Click the image for a much bigger version. [MODIS image acquired courtesy of UAF/GINA]
While clear skies were the rule when this was made (about 2:30 pm, an hour before the snow started at Fairbanks), there are clouds in the northern part of the image. Somewhat spotty and not deep at all, but near the upper part there is what looks like a small circulation, about where the trough was indicated on the model above. It could be dynamic but also could be something forming in the lee of the Brooks Range. In general these clouds look too shallow for any meaningful amount of snow. How can I tell they are shallow? I also looked at the IR (infrared) satellite to try to judge cloud tops, plus the sounding shown below. I didn’t put the IR image on here since it’s a little hard to see much, because the temperature profile was so lacking in contrast. Jacking the contrast up helped confirm the lack of deep clouds. If you want to see it click here. You’re on your own for interpretation.
The image below is pretty telling, however. It’s a night visual shot from ~4 am on the 2nd. Yes, later than above, so clouds may have dissipated relative to the above image. But Fairbanks airport was reporting 3/4 mi visibility in light snow & fog, ceiling indefinite 500 ft, temp -44F (-42C). Yet the lights from not just Fairbanks but every small settlement you can think of are clearly visible. A half moon is enough for the sensor pick up some relief of the hills and mountains. Clouds with any amount of thickness would obscure or at least fuzzy both. BTW, The faint swish of lighter color clipping the upper right of the image is tiny bit of an aurora borealis display. Look at the full image here – it’s worth it. [NOAA-21 image acquired courtesy of UAF/GINA]
More good hard data comes via the balloon sounding for the afternoon of the 1st (00z on the 2nd), sent up about an hour before the snow started. I’d show you the one for the morning of the 2nd when snow was still falling but that sounding was cancelled due to…. extreme cold temperatures (an unfortunate issue with the automated equipment).
In case you are new to the Skew-T diagram, the two jagged lines trace the state of the atmosphere as one goes up from the surface. The X axis is temperature, labeled in Celsius, with -40 at the left. The thin blue lines every 5C slant up to the right, hence the “skew-T.” The Y axis is labeled in millibars (pressure) in the larger type, and meters of elevation with the smaller type inside the graph. The jagged line on the right is temperature, the one on the left is dew point. Along the right edge of the plot are the winds. Thanks to Univ. of Wyoming for the plots.
What this plot tells us is that starting from the very cold surface, the atmosphere warms around 6 degrees C as one goes up the first 1000 meters, then a couple more degrees the next km. This inversion, believe it on not, is smaller and certainly less steep than that during many very cold Fairbanks days. As mentioned earlier, it’s been really cold in the hills too — deep cold. In fact, the warmest part of the sounding was only about -29C (-20F). Compare this to the most productive snow producing temperature of around -15C (+5F).
The very jaggy dew point line (on the left) is mostly separated from the temperature line, meaning it is quite a dry atmosphere above about 500 m. Even where the dew point is near the temperature line, the cold air is very dry in an absolute sense, not ever getting above 0.36 g/kg of water on this sounding. That’s 0.036% water by mass. A hot, humid day in the tropics could see a water content almost 100 times as high! Also, notice that the upper level winds are from the west and northwest: cold snap direction, not snowstorm direction.
So how did it snow for 18 hours (if ever so lightly) given all the things against it: too dry aloft, clouds too thin, clouds at too cold a temperature for typical snow formation, etc? Was it the weak trough? The trough may have helped, but there is one more mechanism I’ve not mentioned. First, you might be wondering about the other weather stations in the region. Did they get snow as well, or was this a fluke or bogus observations? Yes, most did, but only Bettles and Nenana had a measurable precipitation, again, just 0.01 inches, the others just traces, including the coop stations in the area, some of which reported zero. At the closest hourly reporting stations of Fort Wainwright and Eielson the snow was shorter duration, quitting much earlier than at the airport. Why did this brief flurry elsewhere hang on so long at the Fairbanks airport? Tough question.
Here’s what I think may have been happening
Weather 101: precipitation results from condensation, which results from cooling the air. Rising (thus expanding) air is the most mentioned mechanism for cooling the air. But a cloud layer can cool from above by radiation just as the earth does. It happens all the time but is usually of little consequence. But in an arctic environment, with light winds and super dry air above a low cloud layer or fog layer, the super strong cooling can cause spontaneous ice crystal formation and growth enough for them to fall out. In simpler terms, the crazy low and yet lowering temperatures are squeezing the last bits of moisture out of the air. This typically produces what is commonly called diamond dust, or technically, un-branched ice crystals, and are minute in amount. Ice crystals were being reported at Eielson during much of the time snow was reported at FAI. Typically as the ice crystals fall, that process clears the air. In nature anyway. Add to this scene the considerable amount of moisture and particulate matter (which facilitates precipitation) injected into the air by Fairbanks’ power plants, buildings, vehicles and aircraft, relative to the small amount the air can hold at these temperatures, and the squeezing can’t clear the air because the moisture is being replenished. This sustains the ice fog too. But all this is happening quite low, and un-branched ice crystals aren’t really snow. But perhaps they grow in this unusual low level, artificially moist environment and become enough like snow to be called snow. Or perhaps the automated observation system is confused and reporting as snow something that really isn’t snow. I welcome other thoughts on this. I wasn’t there.
More serious, widespread cold snow
Now, for a more clear cut cold snow case, let’s back up a bit and look at Fairbanks less than a week earlier. Here’s a screenshot of hourly surface obs (time-wise they go from bottom to top):
The snow was coming down a respectable rate for Fairbanks at any temperature, especially for -27, -28F (~-33C). The airport picked up 0.6 inches (1.5 cm) of snow containing 0.04 inches (1 mm) of precipitation in a little over 4 hours. But by the time it started snowing the temperature had risen 12 F or 7C degrees from the -40 of early morning, the benchmark I’ve been throwing around as extreme cold and very rare snow. But it is an interesting case and still not common to get this much snow at those temperatures.
Here’s what was going on upstairs:
Although the surface was still very cold, there was a large and deep inversion, meaning it gets a lot warmer as you go up. At 1860 m (6100 ft) the air was a balmy -12.5F (9.5F). Where the 2 lines are close together is where the moisture is, and likely clouds. Deep enough clouds to produce significant snow. There is a dry layer right around the warmest part, but moisture below and above…but it was not snowing quite yet. The calm or light winds near the surface become first 15-20 kts east winds then 30-55 kts south winds mid-level, stronger still higher up. These winds quickly brought in warmth and moisture forcing it to slide up over the cold dense air at the surface without disturbing it much. The snow started a couple hours after this balloon was sent up (so, just like the 2/2 case above). By the next sounding (and probably well before) the dry layer was filled in with clouds/snow and the temperature at the top of the huge inversion was even warmer, eventually reaching -1.5 C (29F) (while still -29C (-20F) at the surface!)
Meanwhile, a similar thing was happening in Anchorage (and throughout a large part of the middle of the state), with the snow starting only a few hours earlier, illustrating not snow at extreme cold but similarly rare heavy snow at moderately cold temperatures. The airport there received 16.6 inches in a day and a half at temperatures around 0F for much of the event, rising to the teens above zero by the end. The water equivalent was 0.80 inches, (20 mm) so quite a bit of precipitation, though certainly less than would typically be contained in that much snow if it fell with the temperature near freezing. Here’s the sounding during the early, yet heavy part of that snowstorm:
Thanks for hanging in there through this long post. I’d love to hear your thoughts, literal or figurative observations, questions… and whether you’ve been outside in shorts at -40 or colder.