These scatter plots of heat and rain helped me understand our changing climate

Hi, this is Gregor, co-founder and CTO of Datawrapper. With this week’s Weekly Chart, I’m returning to my favorite topic: the climate crisis.

Last week deadly wildfires were raging at the U.S. west coast in California, Oregon, and Washington (and they still are). At the same time, Hurricane Sally hit Florida and Alabama, causing major floods and damages. This bizarre episode reminded me of a paradox of the climate crisis we’re now living in. While temperatures are rising and droughts and wildfires become the new normal, there is more rain in a lot of places than there has been in the past.

I first read about it this phenomenon in the German national weather service (DWD) yearly report about climate change[1], and almost couldn’t believe it. How can there be more rain now than thirty years ago? After the extremely hot summers of 2018 and 2019, this year has again been very dry. As a hobby gardener, I follow the weather closely and remember the long stretches of days without rain.

That’s why for this week’s Weekly Chart, I set out to find a visualization that shows this paradox. And I found a chart with the weird Latin name thermo-pluvio diagram (thermo for heat, and pluvio for rainfall)[2]. It shows anomalies[3] in air surface temperature and precipitation for each month of a year in a scatterplot:

I like these kinds of four-quadrants scatterplots because they describe a clear contextual “map” in which we can quickly find each data point. The added lines make it clear that we’re talking about anomalies.[4]

The diagram shows that January, March, and October 2019 where hotter and wetter than the base period, while April, February, and June were hotter and dryer. We also see May jump out as the only month in 2019 that has been colder.

When looking at the current year, it seems as if most of the rain that was “missing” in other months fell in February 2020:

As we see, there is much variation between the years. It makes sense to “zoom-out” a little further and look at seasons instead of months. The next view allows us to look at an entire decade at once. Unsurprisingly, we see most seasons on the “hotter” side of the chart, but most seasons often appear on both sides of the dry/wet spectrum:

But since it’s still hard to interpret this “confetti explosion” chart, I’m ending this post with a much more simplified view of the average seasonal temperature and precipitation anomaly over the past 20 years.

And indeed, on average, three out of four seasons have been more rainy compared to the 1961-1990 base period, especially the winter months between November and January. The typical new dry season in Germany is not the summer, but spring (from March to May).

That’s it for this week. As usual, you can find the R script for this analysis on Github if you want to play around with it (feel free to re-use it).

  1. The DWD report says that “in the past hundred years, we’ve seen an increase in the average precipitation. In the future, there too will be a likely increase in yearly total precipitation.” ↩︎
  2. Found it on page 5 in the 2019 edition of the Climate Status Report for Germany (in German) by the DWD. ↩︎
  3. The DWD defines temperature anomalies as differences between a month’s average temperature and the average temperature of the same month over the base period 1961 to 1990. Precipitation anomalies use the same base period but use percentage-change instead of absolute differences. ↩︎
  4. I would have preferred arrows to show shifts, but that’s not possible with our Datawrapper scatterplots (yet!). ↩︎