Cold Coffee

With the winter solstice just a couple of weeks away it is probably no surprise to find that coffee gets cold much, much faster than expected. This has certainly been the case in my house recently. Fortunately a person can get used to drinking lukewarm and "desk-temperature"1 coffee relatively quickly.

A Coffee Refill

Despite the preamble, I'm not really thinking about the temperature of my coffee beyond the immediate circumstance. The mug was mostly empty when I picked it up and the remaining liquid had been left standing for at least an hour. Of course it's going to be cold. The reason it was left idle for so long is the unfortunate reality that comes with working late at night: it is really easy to lose track of time.

That said, while I was putting the finishing touches on some work-related project updates, a temperature-related question crossed my mind that could be answered with a series of numbers I recently acquired. The local community centre made their weather data available to the public a couple of weeks ago and I jumped at the chance to build a queryable database2 showing the temperature, wind, and precipitation data for the last century from the weather station 450m north of my house. A lot of the data from before the 40s is really hit and miss with gaps all over the place, but from 1943 the numbers became far more consistent and specific. So, just for fun, I wanted to know how the general temperature in this tiny corner of Japan has changed.

Looking at just the minimum and maximum temperatures for each year from 1950 to 2019, this is what we can see:

Temperature

The coldest winter saw the mercury drop to -14.9˚C in 1950 while the warmest winter couldn't even freeze water with 1.557˚C in 2013. The hottest summer hit 45.702˚C in 2018 while the coolest was 32.074˚C in 1976. Just these four figures can be misconstrued, so what if we take a look at what the chart can show really well; the variance in a given year.

The smallest temperature variance was 32.102˚C in 1976, with a winter low of -0.028˚C and a summer peak of 32.074˚C. The largest variance was 53.578˚C in 1964, when winter temperatures never went below -11.758˚C and summer hit 41.81˚C. Since 2014 there has consistently been a temperature variance of greater than 40˚ between summer and winter.

A common refrain that I've heard every summer from people is that "the temperature was never this high while growing up". This might certainly be the case given the number of times this area had a string of cool summers, but seven decades of data shows that this part of the country most certainly had July and August temperatures go past 40˚C in six years before 1970. One trend that is clear, though, is the warmer winters. Only twice has the temperature dipped below -10˚C since 2000.

Examined in isolation, this data cannot reveal common patterns across the country. That said, as more public offices across the prefecture make their historical data available in a semi-workable digital format3, it will be interesting to build some models that look for patterns despite the noise.


  1. I would posit that "desk-temperature" coffee is several degrees colder than room-temperature … despite the desk being in the room.

  2. And, maybe later, an API for the community to use.

  3. I could only get the data in a giant CSV file that needed some pre-processing to handle all the unescaped strings that contained commas. Fun!