How are we doing in getting rid of the dreaded rainbow colour scheme in scientific visualisations?
On a whim, I asked for people to count presentations or posters with Rainbow palettes at this year’s meeting of European Earth sciences mega-meeting, the EGU general assembly.
Many thanks to those that stepped up. This was clearly a non-scientific survey, but should at least give us an idea of the proportion of figures with inaccessible and perception-skewing colour schemes that are out there in the world. Call it a reasonably well informed prior.
Some noticed that certain fields or sessions were much worse than others, and there’s clearly still plenty to do. But I’m heartened that we got a good response to the request, and that there are a good number of people on board with #endrainbow.
The next logical step is a proper study (suggested here). We’ve thought about this before, but never got it off the ground. One easy-but-time-consuming-and-expensive way would be to track the accessibility all of the figures in a selection of Earth sciences journals, over time. It would be great to know if our campaign was getting any traction, or if this is dwarfed by the changes in the matlab and matplotlib defaults (to parula and viridis respectively). If anyone would like to help with this, and maybe be a co-author, then please let us know at betterfigures@gmail.com.
Other things that came up in the discussion were the idea of a short course skills building course, and a rainbow-free-graphics icon or watermark.
On to the results …
Posters
| Volunteer | Rainbow | Non-Rainbow | Both | Neither |
|---|---|---|---|---|
| @philipheron | 24 | 9 | 42 | 25 |
| @niallhrobinson | 21 | 5 | NA | NA |
| @rarohde (1) | 23 | 41 | 28 | 24 |
| @rarohde (2) | 38 | 113 | 38 | 46 |
| anon | 6 | 27 | 10 | NA |
| @fcrameri (1) | NA | 22 | 83 | 44 |
| @fcrameri (2) | NA | 19 | 43 | 34 |
| Total | 112 | 236 | 244 | 173 |
Talks
| Volunteer | Rainbow | Non-Rainbow | Both | Neither |
|---|---|---|---|---|
| @spinokel | 1 | 5 | NA | NA |
| anon | 2 | 4 | 4 | NA |
| Total | 3 | 9 | 4 | NA |
Which means that of the 608 posters or presentations with colour figures, 363 (60%) of them had at least one figure with a Rainbow-type palette.
There are lots of good places to get advice on colour palettes for scientific figures, but you could do worse than to start with the Better Figures guide.
UPDATE 17/04/2018: Thanks to @fcrameri for adding some more data, which took us from 54 to 60% with at least one Rainbow palette. Also @TDaviesBarnard for pointing out that many of the Rainbow plots at EGU were “legacy” plots – older plots not created by the presenter, and used to set context for the discussion.
Here are the tweets with the raw data – if I’ve interpreted your data incorrectly, or you have more data please let me know.
Better Figures 