September 16th, 2021
Fitting 6 decades of monthly data into a tiny chart
This week we’re looking into a fascinating monthly timeseries on carbon dioxide concentration in the atmosphere, published by the U.S. National Oceanic and Atmospheric Administration (NOAA).
The chart shows a couple of things:
So why did I decide to split the time series into separate lines per decade? To explain this let’s take a look a the typical form used for this dataset, like the one below that was published on NASA’s website in 2013:
The problem with this is that it’s not very effective to compress almost 60 years of monthly data into such a tight space.
A simple way out would have been to only show the seasonally adjusted version of the time series, shown as a black line in the above chart. But I’d argue that in this case, the seasonal pattern is actually very interesting. If you don’t believe me, just look at this fascinating NASA video “A Year in the Life of Earth’s CO2” (hint: the video also explains the seasonal highs and lows).
But if we want to keep the seasonal pattern in the chart the line gets squeezed so much that it just ends up looking like a funny snake texture. Unless we give it more space by splitting it up into multiple lines per decade!
By the way, this problem of finding the “right” aspect ratio for a line chart has been researched in the information visualization field a little bit. If you’re interested, I recommend Robert Kosara’s summary, but the short version is: we still don’t really know how to find the ideal aspect ratio, so it’s often best to just eyeball it.