The weather in Vancouver this April was atypical, wasn’t it? A lot of my people were saying so. My mom says April showers bring May flowers, but that’s got to depend on location. Wikipedia says April showers are native to Ireland and Scotland. Regardless, it didn’t rain much in Vancouver this April.
This year I’ve transitioned from managing a web product to working with data. With that shift in my work life, I’m feeling extra motivated in my personal life to find complimentary outlets for my analytical aspirations. I find it refreshing to be analytical in my down time. I think I miss the years of locking my bedroom door during house parties so I could study math and physics.
I’ve also got climate change on the mind. It’s always there. So I’m doubly inclined to sharpen my sense of what’s actually happening. I feel uneasy in conversations about how unusual things seem. I don’t really trust my ability to intuit what’s happening, maybe because I don’t trust my memory. I’m much more comfortable talking about facts, especially when it comes to something important like the state of the planet.
So I want to take a deep dive into this weather question: Was the weather this April unusual?
The government’s weather records are a good place to start, specifically data recorded at YVR. It only took me an hour to download and format all the daily weather records between January 1937 and April 2016.
Ready for statistics? Here we go:
Figure 1 shows the number of days in April in the last 80 years where the daily high broke thirteen degrees Celsius. So high points on the chart are warm months.
- 27 days this April broke the thirteen degree mark. That ties the eighty year record set in 1940. Damn.
- In April months, daytime highs of thirteen degrees or higher happen on average about 14 days of the month
- The standard deviation is 5.7
So 27 days broke the thirteen degree mark in April 2016. Since 27 days is more than two times the standard deviation from the average (14 + 2 * 5.7 = 25.4 < 27), this is pretty unusual, statistically speaking. That’s enough for me to be comfortable with the statement that this past April was super warm.
For anyone who just choked on their salad, a standard deviation is a kind of ruler stick, a tool which tells you how far away from the rest of the data – and therefore how unusual – a particular data point is. Let me explain.
Let’s say you did choke and coughed out a bunch of half-chewed lettuce. If you tracked where those bits had landed, you could calculate the average distance of the bits. Say it’s 50cm in front of you. The bits wouldn’t all be at that average distance of course. The spread from that average could be roughly quantified by calculating the standard deviation. Say it’s 10cm, and say that 70% of the lettuce is found between 40 and 60cm in front of you. (Yes, I’m assuming the spray of lettuce follows a normal distribution.)
So now imagine you find one or two pieces out at 90cm. These would be outliers, defined statistically as data points more than two standard deviations away from the average. That’s the beauty of the standard deviation. You can use it to picture how things are spread out around the average.
Alright, let’s forget about lettuce and get back to weather.
So let’s map the data from the previous chart in a different way – without the time scale – let’s look at its distribution. Figure 2 shows how many years (the height of the bars) had Aprils with a certain number of days (x-axis) with highs that broke thirteen degrees. Notice that the distribution kind of looks like a normal distribution.
Near the middle you see a peak around 17 days, the bar there showing that 10 years in the past eighty had Aprils with exactly 17 days breaking thirteen degrees. Two Aprils, this recent one and one from 1940, saw 27 days breaking thirteen degrees, so there’s a bar at 27 (which I’ve circled) showing those 2 years.
I’ve also drawn three blue lines to show you (from left to right) the average, one standard deviation above average, and two standard deviations above average. The crux of the story here is that the circled bar is more than two standard deviations from the average, which indicates statistically that it is an outlier – that it is very unusual to see 27 days all above thirteen degrees in April.
Let’s talk about rain next. Take a look at Figure 3, which shows the number of days in April that saw rain, for all the years since 1937. Low points show dry months.
The story again with numbers:
- 9 days in April 2016 saw rain
- On average (last eighty years), April has 13.9 days with rain
- The standard deviation is 3.9
What this says is that the rainfall this past April was below average, but not unusually so, since it’s only one standard deviation from the average (see circled bar in Figure 4).
So what have we learned. Well I think I’ve convinced myself (and hopefully you too) that this last April was abnormal. It was unusually warm (very unusually!), and to boot it rained less than average. I ran the same analysis for May, and it’s the same. Not great news for the water supply this year, but it does feel good to have confirmed reports about the weather so far this Spring. Now you can tell people that not only was April 2016 unusual, it was two standard deviations unusual, with respect to the number of days with highs above thirteen degrees.
Stay tuned for my proposal that we start dividing the year into eight seasons instead of four. And thanks for reading! Feel free to drop any questions you might have in the comments.