Robust Color Maps That Work for Most Audiences (Including the U.S. President)
- University of Innsbruck, Digital Science Center and Department of Statistics, Faculty of Economics and Statistics, Innsbruck, Austria (reto.stauffer@uibk.aca.at)
Color is an integral element in many visualizations in (geo-)sciences, specifically in maps but also bar plots, scatter plots, or time series displays. Well-chosen colors can make graphics more appealing and, more importantly, help to clearly communicate the underlying information. Conversely, poorly-chosen colors can obscure information or confuse the readers. One example for the latter gained prominence in the controversy over Hurricane Dorian: Using an official weather forecast map, U.S. President Donald Trump repeatedly claimed that early forecasts showed a high probability of Alabama being hit. We demonstrate that a potentially confusing rainbow color map may have attributed to an overestimation of the risk (among other factors that stirred the discussion).
To avoid such problems, we introduce general strategies for selecting robust color maps that are intuitive for many audiences, including readers with color vision deficiencies. The construction of sequential, diverging, or qualitative palettes is based on on appropriate light-dark "luminance" contrasts while suitably controlling the "hue" and the colorfulness ("chroma"). The strategies are also easy to put into practice using computations based on the so-called Hue-Chroma-Luminance (HCL) color model, e.g., as provided in our "colorspace" software package (http://hclwizard.org), available for both the R and Python programming languages. In addition to the HCL-based color maps the package provides interactive apps for exploring and modifying palettes along with further tools for manipulation and customization, demonstration plots, and emulation of visual constraints.
How to cite: Stauffer, R. and Zeileis, A.: Robust Color Maps That Work for Most Audiences (Including the U.S. President), EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7173, https://doi.org/10.5194/egusphere-egu2020-7173, 2020
Comments on the display
AC: Author Comment | CC: Community Comment | Report abuse
A very "illuminating" presentation! I learned a lot about rendering of colors.
Great, thanks for the positive feedback!
Thanks a lot Siri!
in case you have questions in the future do not hesitate to contact us (contact information e.g., on ), we are happy to help, if needed.
To play around with our tools and get more information, you may also have a look at and/or watch Achims talk from the useR! conference last year () with some more detalis than in today's short PICO :).
All best,
Reto
Hoppala. Seems that the tool kills hyperrefs. Next shot:
More information about the colorspace package:
* http://colorspace.r-forge.r-project.org/
The useR! talk by Achim Zeileis:
* https://www.youtube.com/watch?v=6bv2IAcNE_c
Link to the hclwizard website:
* http://hclwizard.org/
Sorry about that!
Dear authors,
Thanks for your presentation. How do your color maps compare to those of the presentation of Shepard, Crameri and Heron? While your color maps look good, it is unclear to me what principles are followed that lead to perceptually uniform color maps.
Regards,
Taco Broerse
Thanks for the feedback. Close approximations of various of Crameri's scientific color palettes are shipped as part of the "colorspace" package. See:
http://colorspace.R-Forge.R-project.org/articles/approximations.html#approximations-of-crameris-scientific-color-scico-palettes
Similarly, many other well-known palettes like those from ColorBrewer, Matplotlib/Viridis, or CARTO can be closely approximated (also shown on the same page).
The construction principles are not unsimilar but we give more flexibility to the user in creating either relatively uniform palettes that try to distinguish small to moderate changes along the entire scale (like Crameri's) or to distinguish extremes etc. The construction principles are shown here:
http://colorspace.R-Forge.R-project.org/articles/hcl_palettes.html
Thanks for your answers. More on an elementary level, do you also provide background to what makes a good color map? Not only from a intuitive viewpoint, but more quantitative?
Yes and no.
Yes, because our work is based on a lot of research regarding the perceptually-uniform HCL color space, incorporates best practices for color selection, and comes with various tools to assess the appropriateness and robustness of the chosen palettes (in different contexts, on different backgrounds, with different color vision deficiencies, etc.).
No, if you were looking for a quality criterion that is optimized by the palettes. In my experience this is very hard to do because it is unclear what you really want to bring out. Sometimes uniformity is mentioned but in many cases this is not the prime purpose of a color coding. In other situations a minimum distance between all colors in a discrete is mentioned while in others a smooth and monotonic luminance gradient is more important. Yet in other applications the robustness for color vision deficiencies is essential. I don't think there is a color scale that excels in all these criteria - thus, there are always tradeoffs to be made.
The colorspace web page and accompanying paper on arXiv (https://arxiv.org/abs/1903.06490) try to give an overview and discussion with lots of references. One other recent paper I particularly like in this regard is: https://doi.org/10.18637/jss.v090.c01
Thanks. This is the answer I was hoping for.
Thanks bringing up this discussion and the nice online chat! :-)
Thanks bringing up this discussion and the nice online chat! :-)