'Tis the season of -20°C

As -20°C looms, today we're going to look at when throughout the winter we are likely to get extreme temperatures.

Here is a recent history of when the temperature first dips below -20°C each winter:

First -20°C Each Winter

For the last few years, the first -20°C has shown up in mid-to-late November. 2010-2011 is roughly as long as I've been biking year-round, and I'm always wary of the end of November because it often brings a sudden blast of real-winter.

But that hasn't always been the case, and going back to 1995 we see a lot of years where -20°C didn't show up until December or even January. So this year is a little late in terms of recent history, but it's not unusual even going back 20 years.

For the rest of today we're going to look at the whole winter, and at how the likelihood of really cold days changes as we move from November through March. And because I don't want today to be entirely discouraging, we'll be looking at the extremely warm days as well as the extremely cold ones.

Likelihood of Warm & Cold Days

Here I've broken the winter down into weeks, and the orange and blue lines show the likelihood of experiencing at least one day with a High above 0°C or a Low below -20°C in each week. This is based on the temperatures for Blatchford for the last 20 years.

I've underlined "at least one day" up above because it's an important part of this. This isn't trying to count all of the cold days, or looking at how long a coldsnap or warmspell typically lasts. This is just trying to give a timeline of when we might expect warm or cold weather, even if only briefly.

Using the first week of January as an example: for Highs above 0°C the number is 75%, and for Lows below -20°C it is 60%. That doesn't mean that the week will be 75% above 0°C and 60% below -20°C. It just means that in the last 20 years, during the first week of January, in 15-of-20 years there was at least one day above 0°C, and in 12-of-20 years there was at least one night below -20°C.

So why did I do this chart? The goal is really to psych myself up for the winter. Days below -20°C mean wearing a whole bunch of extra gear (including putting boots on the dog when she goes for a walk), and this gives me an idea of how long that season might last.

-20°C can arrive as early as the second week of November, and it might not leave until April. But it's only in late-December and early-January that there's more than a 50/50 chance of dipping down below -20°C. So with this I can see that I need to keep the mitts and longjohns and dogboots handy for 5 months, but they'll get most of their use in December, January, and then the cruel joke that is early-March.

The flipside is the days above 0°C. Here we can see that for any given week of the winter we are always more likely to see a warm day than a cold one. Even in January there's a 70% chance that any week will pop up above the freezing mark. The one oddball is the first week of December, which is only at the 55% mark for some reason.

I've said before that I like using 0°C and -20°C as ranges, because they're representative of a pretty nice winter day, or a pretty chilly one. But now lets look at some of the real extremes.

Cold Days

Here's the same data, but this time looking at only the cold days: the Lows below -20°C, -25°C and -30°C. The really cold days follow -20°C pretty closely, with the peak in January, and that same annoying hump in early March.

For lows below -30°C the range is between 0 and 20% depending on the week. So even for the worst weeks - the 2nd and 4th of January - only 4 of the last 20 years have had any days that cold. Although the 2nd week of January has dropped below -25°C in 10 out of the last 20 years.

We shouldn't be too depressed though, because the chances of extremely cold days are much lower than the used to be:

This is an update of a chart that I used for one of my very first posts. It shows how the number of very cold days that Edmonton gets each winter has fallen over the years.

Nowadays the low drops below -20°C an average of only 23 times a year. Although that can range from only 6 times in a winter like 2015-2016, to 46 in 2010-2011.

Warm Days

Here we have the warm winter days. As we saw earlier, for most weeks of the winter there has been at least a 70% chance of a day above 0°C, with the only exception being the first week of December.

Even for days above 5°C, the lowest likelihood is 30%. So 6 out of the last 20 years have had at least one day above 5°C in the last week of December, or in the middle of January.

Winter days above 10°C are definitely pretty rare though, which we saw when we looked at chinooks. For most of November through February the likelihood is 20% or less. And in the last 20 years there have been no days above 10°C in the 3rd week of December, the 2nd-4th weeks of January, or the 1st week of February.

At the beginning of November this year we had a few weeks of very warm weather, which were pretty unusual. In the first week of November the chances of days above 10°C are only 40%, and for the second and third weeks it's about 20%.

Here we have the warm days and the cold days combined, which looks neat, even if it's is pretty hard to read.

Generally the warm days dwarf the cold ones. For most weeks of the winter - except for late-December and early-January - we're more likely to see a day above 5°C than a low below -20°C. And the highs above 10°C and lows below -30°C trade-off which each other, with the +10s disappearing at the same time that the -30s peak.

Everything today has been about the likelihood of any given week getting a warm or a cold day. Another way to look at this would be the number of days each week that are warm or cold, and so here is one final chart:

Average Days per Week

This is the average number of days each week that are above 0°C, or below -20°C.

Averages are a bit tricky, because they're not necessarily representative of any given year. So when we see an "average" of 2.4 Lows below -20°C in the first week of January, what that really means is that there are some cold years when most of the week is below -20°C, balanced against the warm years with no -20s. That's just something to keep in mind when looking at this chart, and it's why I started today with the at-least-one-day approach.

The general shape of these timelines is pretty similar to what we saw in the first chart today: winter peaks in early January; things ease off in February; and there is an early-March bounce to keep us on our toes. And here we see that in the darkest depths of winter at the beginning of January we get just about as many total days above 0°C as we do nights below -20°C.