The choice of color to represent information in scientific images is a fundamental part of communicating the findings. However, many of the color palettes widely used to display important scientific results are not only dangerously misleading, but also unreadable for a proportion of the population.
For decades, scientists have been pushing for a permanent change to remove such pallets from public consumption, but the fight for universal access to science communication continues.
A color map is a palette of many different colors that assign values to areas on a plot. An example of a deceptive color map is rainbow, which typically starts with blue for low values, then goes through cyan, green, yellow, orange, and finally red for high values. This color combination is neither divergent, which would allow us to see a central value, nor sequentially, which would make the values low to high intuitive.
Color brings life to data
Using colored bar graphs can allow scientists to turn their collected data into something meaningful to share widely. It could be the first direct impact of a black hole, mapping votes cast in political elections, planning an expensive rover route over Mars’ topography, essential communication of climate change, or a critical diagnosis of heart disease.
Despite the apparent importance of color, scientists often choose the default palette setting of the visualization software being used.
rainbow – Or jet — Color palettes are often the default setting on software, but the beautiful sweep of blue to red is misleading when displaying scientific data.
Basically, switching between colors in the palette is not seamless. For example, the transition between blue and green and then between yellow and red occurs over a short distance. wiki And batlo, are examples of even color palettes, where the color bars smoothly change colors.
To put this in context, having a palette that changes wildly between colors is one position along the x or y axis of a number that are not evenly spaced. In jet color maps, this number would be closer to one than four and eight to 10 away. Such an uneven color gradient means that some parts of the palette will naturally be highlighted over others, causing the data to distort. The RGB color space on which such disparate color gradients are built is mathematically simple, but is not consistent with how we perceive colors and the differences between them.
Another problem with uneven color palette like rainbow is that data presented using these colors may be unreadable or inaccurate for people with vision deficits or color blindness. Color maps that include both red and green with equal lightness cannot be read by a large segment of the population.
The general estimate is that 0.5 percent of women and eight percent of men worldwide suffer from color-vision deficiency. While these numbers are small and have almost disappeared in populations from sub-Saharan Africa, they are probably quite high in populations with a large fraction of white people, for example, in Scandinavia.
Needless to say, scientific results should be seen by as many people as possible, and the shortcomings of such color-vision should be taken into account.
the winding road to the end of the rainbow
with issues jethandjob rainbow And other uneven color palettes have been known for years. Although some areas of science have made significant changes to best practices on color policy, other areas have stuck with their default settings.
As researchers become interested in more effective data communication, we outline approaches scientists can make to communicate their findings more efficiently: jet Or rainbow default color palettes; If it is necessary to use red and green, make sure they are not the same brightness for accessibility; And use a palette that changes between colors evenly.
There is an increasing recognition of the challenges associated with rainbow stripes. Some academic publications – eg nature geology – Adopted an even more color palette policy for new submissions. The Intergovernmental Panel on Climate Change has colour-blind friendly guidelines for statistics.
Software packages such as MATLAB and Python have been removed rainbow as their default color palette for data visualization features. However, old habits die hard and vigilance is still necessary – it’s important to call out bad color choices when not noticed (otherwise trends keep repeating).
Better science communication, better results
The importance of accurately sharing scientific data in an accessible way cannot be underestimated. Uneven color gradients are often chosen to artificially highlight potential danger areas, such as storm track boundaries or current virus spreads.
can generate judgments based on data being misrepresented, for example, a Martian rover being sent over terrain that is too steep because the topography was viewed incorrectly, Or a medical worker was making a misdiagnosis based on uneven color gradients.
Science accessible to all begins with moving away from the default. This can start with learning to choose similar color gradients for term projects, to dismiss papers for misleading figures by international publishers. One day, it may even involve Canada’s Meteorological Service breaking away from dramatic uneven stripes to highlight weather changes.
Basically, using the wrong color map is the equivalent of deliberately misleading the public by distorting the data, and has significant potential consequences.