I Pulled 30 Years of Weather Data to Calculate the Perfect Wedding Date

Here's what happens when you get engaged: you experience roughly 24 hours of heady bliss, champagne, and congratulations. Then reality descends and you remember you now have to plan a wedding.

According to common sense, Pinterest, and Emily Post, the first step is to pick a date, which should be fairly simple. But if you want to get married in the cheap showiness of nature (and in Canada) like I do, weather is a pretty big consideration. Since my fiancé is a data journalist and I'm a science journalist, we naturally decided we'd choose the date based on data.

We started by narrowing down the dates. We're planning to get married in my hometown, in midwestern Ontario, around the 45th parallel. Summer there typically only lasts from mid June to mid September, and I've seen snow at both ends. July is the driest month of the summer: June can still get late spring showers and August is prone to summer thunderstorms. We limited it to July, with a couple weekends in late June and early August for good measure.

Like most rational people, we wanted to choose a Saturday, and July 14 is one of my best friends' birthdays (I'm not gonna steal her day), so that left us with eight potential dates in 2018:

• June 23
• June 30
• July 7
• July 21
• July 28
• August 4
• August 11
• August 18

First we pulled the rainfall from those dates for the last five years and averaged them, but after calling a meteorologist—the only other person in the world who considers this a perfectly reasonable way to choose a wedding date—I realized I needed more data.

"My wife and I kind of did a similar thing; she's also a meteorologist," said Charles Roop, a meteorologist at WCTV in Florida. "But I would go a little bit further back. Go back 30 years. You have a wider range to deal with in terms of climatological data."

So I pulled the daily precipitation data for my hometown for each of the prospective dates every year from 1984 through 2016. (Environment Canada has an online database, but you can find similar data for the U.S. from the National Oceanic and Atmospheric Administration.)

Here's what it looks like:

As you can see, a lot of the time, these days have little to no rain at all. The few big spikes are all those August thunderstorms I mentioned (including nearly 3 inches of rain in 1984, and more than two inches in '99). A few millimeters of rain doesn't have to be a day-ruiner, but the risk of random massive storms in August made me write those dates off completely.

Then I calculated the average precipitation for each date based on the records from the last 32 years, here's what I found:

(Apologies to my fiancé for the ugly charts.)

July 21 is the most appealing date, with an average rainfall of just 0.78 millimeters compared to, say, August 4, with an average of 4.4 millimeters. And over the past few decades, July 21 hasn't been subject to any freak storms: the most it ever rained was in 1991, when less than half an inch fell.

Of course, a lower average history of rain doesn't ensure perfect weather. It could be too hot, or too cold, or cloudy, or foggy—but at least we have a better chance of being dry, right?

Even though there are no guarantees, to me, and my fiancé, it makes more sense to use the data at hand than to ignore it. But I still wondered: is this really a useful way to pick a date, or are we fooling ourselves thinking we can outsmart mother nature?

"That's the best you can do this far in advance," Roop told me. "But always have a contingency plan."

Get six of our favorite Motherboard stories every day by signing up for our newsletter .