September 11, 2013

Over the Rainbow

About a year ago posted on their blog an open letter to NASA asking them to avoid using the rainbow (spectral) colour scale for representing continuous data. The letter listed five problems with the Rainbow colour scale:
  • Colour-blind people can't perceive the scale properly
  • Divisions between hues produce false visual artefacts
  • The order of the hues has no inherent meaning
  • Yellow appears much brighter than the other hues
  • It is more difficult to see detail than with scales that vary in brightness
I'm ashamed to admit that I've used the rainbow colour scale in my own work, often in response to pressure from users who are used to seeing the scale used elsewhere and want to apply it to their own data.

NASA has responded to's open letter in the form of Robert Simmon from the Earth Observatory. Robert has been on a similar crusade within NASA to eradicate use of the rainbow colour scale, apparently with some success.

Robert's response has been a series of six blog posts on how to use colour for data visualization. It is the best tutorial I've come across on this subject. The posts are:
The use of colour is a vital but under-appreciated aspect of data visualization. It's all too easy to use the default colour scales provided by the tools we use to create visualizations. Unfortunately, these defaults are often inadequate. Rather than using the defaults, spend some time thinking about how you are using colour to represent data. If you refer to Robert Simmon's "Subtleties of Color" tutorial when doing so then you can't go wrong.

[ 2013-09-24 ] Robert posted this addendum