I’ve seen lots of discussion lately about the dangers of using rainbow colour scales (or Jet colour scales) in maps and graphs. See #endrainbow on Twitter and Ed Hawkins articles:
- Why rainbow colour scales can be misleading.
- The end of the rainbow
- Graphics: scrap rainbow colour scales (Nature 519, 291)
and this article by Light and Bartlein from 2004:
The issue is that rainbow scales can ‘distort, mislead and confuse’ by creating false boundaries between values. For design guides, the use of rainbow colour schemes is likely to increase error as users do not get an intuitive feeling if a value is high or low. Rainbow colour scales come up in some hydrologic publications and many of the issues identified by Ed Hawkins, and others, are relevant.
There is some good examples in Australian Rainfall and Runoff. The first figure below is a map of Baseflow Peak Factor using a rainbow scale. It is not visually obvious that the reds are low values and the blues are high. The second figure shows similar data using a sequential pallette. High values are dark, low values are light. This figure is much easier to interpret.
There is some good advice on picking a suitable colour scale for maps and graphs (here). In summary if you wish to show order in the data, then choose an ordered palette. People expect that darker means more. For departure from a reference value i.e. values can be either positive or negative (or above or below a value), a diverging palette may be suitable.
This issue has been around a long time but should be easy to fix, let’s start by improving the figures in Australian Rainfall and Runoff while it is still a draft.