Colourmaps
When you are using multiple colours in your figures, you can use a colourmap. Colourmaps are a premade selection of colour gradients that you can apply easily in your figures. Generally, three categories of colourmaps exist13:
- Sequential colourmaps are appropriate for visualising sequential data. They can consist of different hues of one colour, or an intuitively increasing sequence of colours.
- Diverging colourmaps indicate deviation from a standard, or median value. They generally consist of two contrasting colours decreasing in hue to the middle.
- Qualitative colourmaps are appropriate for visualising different categories (discrete numerical data or categorical data) instead of continuous numerical values. Therefore, they consist of different colours without a gradient in between.
Depending on what software you use, you can add colour libraries to apply colourmaps to your figures. Always use a colourmap that is appropriate for your data. This is especially important when you are visualising gradients in maps or graphs. Be aware that although many colourmaps exist, not all of them are good to use.
For categorical data, the exact type of colourmap you use does not matter as long as it is suitable for visionally impaired people and acknowledges any possible intuitive colours for your data. However, for sequential data there are many more things to consider. Often, scientist still use rainbow colourmaps to visualise sequential data. Although these might be visually attractive, this type of colourmap has several problems.
- The arrangement of colours does not follow an intuitive order: to the human eye, yellow is the brightest colour. Therefore, we automatically interpret values with this colour as highest or most important. However, in the rainbow spectrum, yellow indicates intermediate values20.
- Not all shifts between colours in this type of colour appear equal as some look more abrupt than others. This results in an inaccurate representation of underlaying (continuous) data with a consistent rate of change.
- Abrupt changes in the rainbow colour scale are intuitively interpreted as a meaningful boundary, while in the data there is none21.
- The uneven colour gradients in the rainbow colourmap are unreadable to people with colour-vision deficiency20.
As graphs and maps are not only used to communicate information, but also to analyse data, this can lead to wrong interpretations. You might also miss important details in your data. Therefore, choose a suitable colourmap for your data and figure.
Good colour bars for sequential or diverging data need to contain colours in an intuitive order and even colour gradients, i.e. a homogeneous gradient without any abrupt changes. Scientifically derived colour maps with a balance of intuitive sequences of hue, luminance, and saturation now exist. These accurately represent data variation, reduce complexity, and are appropriate for colour-blind people20. Do not change the scale of a scientifically derived colour bar as this would distort the data, similar to when you would change the spacing between some of the axis ticks of an x-y graph20. Remember to always add a colour bar in figures that use a colour scale.
We recommend you to read this paper by Crameri et al. (2020) on the misuse of colour in science communication. It contains several examples of how to (not) use colour in scientific figures, and a flow-chart to decide what type of colourmap or scheme you need for your figure. It also contains several links to downloads of scientifically derived colourmaps to use in your software of choice.
