Online Research Tips: Historical Weather Data
Yesterday’s post was all about spring-time melt, but the blanket of new snow covering most Labrador communities today surely takes our minds in another direction!
Having been tipped off to the snowfall warning last night by a friend, I straight away took to Environment Canada’s weather page, which I’m sure is a familiar resource for many. However, not everyone may be aware of the sheer quantity of weather data available through the site. Local forecasts are just the beginning!
This post is an illustration of the kind of data that can be gleaned from this excellent resource—and hopefully we’ll uncover some interesting Labrador snowfall trivia along the way.
On the Environment Canada web site, the archivally-minded will take particular interest in one link at the bottom right-hand corner of every local forecast page: “Historical Weather.” If you have a serious or semi-serious question about recent weather history anywhere in Canada, this is the first stop on your online research journey.
First up is a user-friendly splash page with daily data for the current month and options to look up the weather on any historical date of your choice. This is handy if you’re writing a historical novel or trying to date an anecdote about a famous blizzard, but I thought I would pose some more general, top-of-mind questions about snowfall in Labrador, prompted by today’s weather:
- How common are May snowstorms in central Labrador?
- When do we actually get the most snow? Anecdotally, I’ve heard that March is our snowiest month, but has that been true in the past, or at least since modern records have been kept?
For questions like these, the real jewel is the “Get More Data” link on the right-hand side of each local Historical Weather page, underneath the banner of station metadata. This link takes you to a Google Drive with documentation on how to download a lot more weather data. Now, despite the very good how-to guide provided by the site administrators, the next few steps require a bit of doggedness and/or confidence with behind-the-graphical-interface computer work.
This is a common theme in online research: the really good stuff often requires a bit of digging and a commitment to figuring out how information systems work.
Fortunately, however, resource providers generally want you to succeed in your research, and they try not to make things too complicated or technical. Within an hour, I managed to download all the daily weather data for the “Goose A” weather station from its beginnings in 1941 to 2020. (You can also get data at other timescales, including hourly, and for many different weather stations in the country, including several others in Labrador.)
Our First Question: May Snowstorms
Environment Canada has snowfall figures for Goose Bay on 2419 different dates in May, going back to 1941. We have had 10 cm snow or more on just 32 of those dates. Therefore, roughly speaking, if you wake up on a random day in May, you have about a 1.3% chance of seeing that much new snow (with better odds earlier in the month, and worse later).
This may be a good time to mention some caveats! For one, history isn’t necessarily predictive, especially in a context of climate change. Also, I’m not a statistician, meteorologist, or climate researcher. This blog post is purely illustrative! Shout-out here to actual experts like Happy Valley-Goose Bay’s own Robert Way at Queen’s University.
Our Second Question: The Snowiest Month
Using the Environment Canada Data, I have also made some charts to illustrate the bigger picture of daily snowfall patterns in Goose Bay.
Here the number of measurements per day mainly varies from 77 to 79, because of outages, errors, etc. (which may skew the results a bit, if they are non-random). The exception is February 29, which has only 20 measurements—you can likely guess why! Even with all those measurements, there is a lot of variability day to day. The data is so chaotic, in fact, that the overall curve was hardly discernible until I started averaging values across multiple dates. Therefore each point on the line above shows the mean snowfall not just for a given date, but for that date and the five days immediately before and after it. The blue bars show the average daily snowfall for each month.
The verdict: January is our snowiest month!
January (2.45 cm/day) is followed by December and March, which are tied to the nearest tenth of a millimetre (2.24 cm/day). Then February, November, and April follow in order, with a steep drop-off to October and finally May.
So much for daily snowfall, but there is another angle we can take.
What about total snow on the ground?
Environment Canada has us covered here too.
In this chart, the number of measurements per date (n) drops to a still-respectable 62-65 (or 17 for February 29), because snow depth was not measured until 1955. Fortunately, however, the data is much cleaner, presumably because total snow depth is less dramatically impacted on a daily scale by individual storms. I used 5-day averages nonetheless to smooth out the curve a little more. Notice that it leans to the right, showing that our snow melts faster than it accumulates (I would guess that compaction also contributes to reducing snow depth and therefore slowing accumulation, even at lower temperatures).
In March we lose on average 12.1 cm of snow depth, then in April another 46.1 cm, and nearly all of the remaining 27.1 cm melts away in May.
So not only does March receive less snow than January, it also sees a net decrease in snow depth! But a March day, on average, has 85.39 cm of snow, compared to 81.88 for a February day, so March is our snowiest month in that respect. Also, since 1955, annual snow depth has peaked, on average, on March 7 (again using 5-day average values). So maybe that is the origin of the March-is-the-snowiest belief.
Isn’t it amazing how much top-quality hard data one has at one’s fingertips, thanks to the power of public information sharing?