Carbon in Context: The Logpile Chart
Introducing an alternative to log scales for comparing numbers
Welcome to Earth Week on Nightingale, the journal of the Data Visualization Society. In honor of Earth Day on April 22, we are publishing earth-related data-visualization content all week. Data viz can enhance our appreciation of the planet, illuminate our relationship to it, and call us to action to preserve it. After all, we only have one and it means the world to us. You can keep up with all of our Earth Week articles here.
A couple of months back, an idea came to me in a dream. That sounds ridiculous, I know, but it’s true. I woke up one morning with a clear-as-crystal picture in my head of a kind of chart that I hadn’t seen before.
The problem that it turns out my unconcious mind had been working on was finding an alternative to logarithmic scales. Log scales are great for those who understand them, but the vast majority of the general public do not. Surely there was a better alternative for allowing the easy comparison of vastly different quantities?
My solution, which I call a logpile chart, stacks a series of left-to-right linear scales in a pile on top of each other. Each scale represents just a tiny fraction of the one below it. This allows the viewer to come to a graphic with a quantity in mind and quickly find others that are comparable — putting that quantity in context.
As I’m publishing this idea during Earth Week, I decided to illustrate the concept with different carbon emissions figures — which I’ve found people have a very loose handle on. How much CO2 is emitted by producing wheat versus beef burgers, or by taking a transatlantic flight, or by different countries each year, or by the global shipping industry? The media throws around kilograms, tonnes, and gigatonnes loosely, but it’s hard for non-experts to understand how they compare.
So I wanted to create a visualization that lets people easily do that. I scoured the internet and the media for examples of CO2 quantities, and then narrowed those down to a nice even spread across the different scales in question. Here’s the result, which I’ve also published at duncangeere.com/carbonincontext.
Challenges
The process of developing it was pretty lengthy, partly just because I kept getting distracted by other projects. I went through several iterations of the design, some of which are below.
As you can see, the biggest difficulty I had was representing the points — originally they were circles with area scaled to the quantity of emissions. But even a square root scale couldn’t handle the difference between the largest and the smallest. Then I went to large same-sized circles, but people I tested it with found that confusing.
Eventually I resorted to simple points, which also had the benefit of giving more space for the annotations. It took a while to get to a stage where I was happy with these too. I tried several different versions, and also gathered feedback on Twitter.
Style A was the audience’s favourite, so I went with that, after trying out and rejecting E and B (below). I liked C (and so did several voters), but felt like it would be a pain to arrange things. Not a single person voted for D.
Design Elements
I’d like to pick out a few of the more subtle design aspects of the chart that I think make it particularly effective. First is the increase in axis line thickness down the page — this gives a nice structure to the chart, as well as a visual cue that each axis is “larger” than the previous one. (Thanks to Silfa for suggesting this idea!)
Second is the large but faded background text, which lets a reader who has a quantity in mind very quickly find the right place for it in the chart. Finally, there are the hourglass-like curves below each axis, which visually reinforce the idea that each scale represents only a teensy chunk of the one below it.
Final thoughts
I think this chart works very well for its core purpose — which is to allow for the comparison of widely-varying quantities, and for a reader to quickly get context for a number they’ve found somewhere else. As well as carbon emissions, I think this would work nicely for dollar amounts, distances, weights or spans of time, among others.
I don’t think it works so well for a single quantity increasing over time (e.g. cases of a disease rising), as data points that jump from one axis to another can’t be so easily connected. I feel like a log scale is better for that use case. But hey, I might be wrong. If you have an idea to make it work, then go ahead and remix the chart with my blessing.
This article was originally posted on my personal blog, Bar & Line.
