Rainbow colors distort and mislead research in hydrology &ndash;
guidance for better visualizations and science communication

Stoelzle, Michael; Stein, Lina

doi:https://doi.org/10.5194/hess-2021-118

Preprints

https://doi.org/10.5194/hess-2021-118

Preprints

09 Mar 2021

Review status: this preprint is currently under review for the journal HESS.

Rainbow colors distort and mislead research in hydrology – guidance for better visualizations and science communication

Michael Stoelzle¹ and Lina Stein² Michael Stoelzle and Lina Stein Michael Stoelzle¹ and Lina Stein²

Show author details

¹Faculty of Environment and Natural Resources, University of Freiburg, Freiburg, Germany
²Department of Civil Engineering, University of Bristol, Bristol, UK

Received: 27 Feb 2021 – Accepted for review: 08 Mar 2021 – Discussion started: 09 Mar 2021

Abstract. Nowadays color in scientific visualizations is standard and extensively used to group, highlight or delineate different parts of data in visualizations. The rainbow color map (also known as jet color map) is famous for its appealing use of the full visual spectrum with impressive changes in chroma and luminance. Beside attracting attention, science has for decades criticized the rainbow color map for its non-linear and erratic change of hue and luminance along the data variation. The missed uniformity causes a misrepresentation of data values and flaws in science communication. The rainbow color map is scientifically incorrect and hardly decodable for a considerable number of people due to color-vision deficiency (CVD) or other vision impairments. Here we aim to raise awareness how widely used the rainbow color maps still is in hydrology. To this end we perform a paper survey scanning for color issues in around 1000 scientific publications in three different journals including papers published between 2005 and 2020. In this survey, depending on the journal, 16–24 % of the publications have a rainbow color map and around the same ratio of papers (18–29 %) use red-green elements often in a way that color is the only possibility to decode the visualized groups of data. Given these shares, there is a 99.6 % chance to pick at least one visual problematic publication in 10 randomly chosen papers from our survey. To overcome the use of the rainbow color maps in science, we propose some tools and techniques focusing on improvement of typical visualization types in hydrological science. Consequently, color should be used with more care to highlight most important aspects of a visualization and the identification of correct data types such as categorical or sequential data is essential to pick appropriate color maps. We give guidance how to avoid, improve and trust and color in a proper and scientific way. Finally, we sketch a way to improve the communication of rainbow flaws between different status groups in science, publishers, and the media.

Michael Stoelzle and Lina Stein

Status: final response (author comments only)

RC1:
'Comment on hess-2021-118', Fabio Crameri, 22 Mar 2021

GENERAL COMMENTS

The study by Stoelzle and Stein is clearly written and carefully performed (e.g., elaborate and somewhat conservative classification scheme and cross-checks between reviewer team to reduce bias) and addresses an important topic from a new angle: This is the first time that a systematic review of unscientific rainbow colour maps is performed in an environmental journal covering almost a thousand papers in total and opens an important usage-over-time perspective with the previous review by Borland and Taylor (2007). This study is therefore useful not only to understand the current (mis)use of faulty colour maps and colour combinations, but also provides valuable information on the community-wide usage through time. Moreover, the study offers a valuable up-to-date overview over current resources needed for science-proof data visualisation.

Even though visualisation is a fundamental, and widely-used scientific method, the results found by the authors underline the lack of education in visualisation, and hence awareness, amongst scientists from an empirical standpoint. Even more dramatic is the finding that some editors and authors have disregarded expert review comments on figure accessibility and accuracy.

The authors have done a fine job to cover and refer to previous work on the topic and provide an insightful new angle. Given the widespread lack of understanding about the importance of colour in scientific data representation, and the relevance to the specific readership, I highly recommend this manuscript, after minor revision, for publication in Hydrology and Earth System Sciences.

SPECIFIC COMMENTS

Figure 6: The figure could be made more intuitive by clearly labelling the top row of panels too (which are currently not labelled). It should be clear (either by additional on-figure text or via the caption and added panel labels) that the ‚original‘ panel is the not suited version to improve, and the three panels to its right-hand side are options to make it suitable. So, I suggest to add (a), (b), (c), … labels to all panels or clarify the on-figure explanation, e.g., by „story-telling graph titles“.

lines 257-259: These points were also made in Crameri et al., 2020 (already in reference list of the current manuscript), to which the reader could be pointed to here as well.

lines 310-311: The reference for the actual Scientific colour maps is:

Crameri, F. (2018). Scientific colour maps. Zenodo. http://doi.org/10.5281/zenodo.1243862

lines 312-318: The authors might consider clarifying that there are two other potential data types/options (which however might be less common): multi-sequential (e.g., bathymetry + topography with a centric value but not diverging) and circular/cyclic (e.g., river orientation; with repeating colours for e.g., 0 and 360°) as outlined in Crameri et al., 2020.

line 320-321: This is a valid point to make.

line 322: Crameri et al. (2020) provides a handy flow chart to select a colour map based on the data to be visualised, which could be referred to here as well.

lines 343 and 391: I suggest to change: „Marie Curie“ to „Marie SkÅodowska-Curie“

line 341: The reference for „Shephard et al., 2017“ should be:

Crameri, F. (2017), The Rainbow Colour Map (repeatedly) considered harmful, edited by G.E. Shephard, EGU-Geodynamics Blog, https://blogs.egu.eu/divisions/gd/2017/08/23/the-rainbow-colour-map/, last access: 13 January 2021

or something like that, as it was written by myself and edited by Grace Shephard without contribution from others.

line 374: A good reference to back up this statement would be:

Moreland, K. Why we use bad color maps and what you can do about it.

Electron. Imaging 2016, 1–6 (2016).

who concludes that the widespread use of the rainbow is the main reason scientists (and others) propagate it further.

line 405: Not to leave out the other important aspect investigated here, the authors may rewrite to: „…to banish the rainbow color map, and simultaneous red and green usage, …“ or something along these lines.

TECHNICAL COMMENTS

lines 127-128: That sentence sounds unclear to me; consider clarifying. E.g., the term „vision deficiency scale“ sounds somewhat arbitrary.

line 132: If it is actually the case, consider clarifying that it means „all papers published in HESS“.

line 212: Consider informing the (potentially non-hydrologist) reader that the following suggestions are specific to/from the field of hydrology, as some of the used terms (e.g., ‚response surfaces’) are likely not familiar to readers from outside the discipline, and a potential source of confusion.

line 234: „point“ instead of „points“

line 318: Point D needs to be clarified grammatically.

lines 369-371: I do miss some critical commas throughout the manuscript. Here, for example, after „perspective“ and „With that“.

line 398: „alarming“ to „alarmingly“?

Citation: https://doi.org/10.5194/hess-2021-118-RC1
- AC2:
  'Reply on RC1 (F. Crameri)', Michael Stoelzle, 22 Apr 2021
  Thank you, Fabian Crameri, for the valuable comments on our manuscript. We identified following main suggestions:
  We will revise Fig. 6 to gain more clarity about the suggested line graph improvements.
  
  Some suggestions about more specific references are given. We will revise those references.
  
  We will add information about multi-sequential and circular/cyclic color maps to complete the color tasks in section 3.2
  
  The misuse of sequential and diverging color maps (L320-321) is supported by our paper survey. As also reviewer #2 (RC2) suggests to provide more information about sequential vs. diverging color maps, we will revise the manuscript and plan to add an example of a diverging color map into Fig. 1.
  
  We agree on the list of specific and technical comments from RC1 and will consider them to improve the manuscript.
  
  Citation: https://doi.org/10.5194/hess-2021-118-AC2
CC1:
'Comment on hess-2021-118', J.C. Refsgaard, 24 Mar 2021

As a person belonging to the 5% minority with color-vision deficiency (CVD), I clearly recognize many of the issues addressed by Stoelzle and Stein, and I highly appreciate that the authors put focus on use of rainbow colors and colors where red and green are difficult to distinguish. Both in papers and particularly at presentations, where time to study colors is limited, I have over the years had great difficulties reading figures when they have been prepared with colors such as rainbow (upper panel of Figure 1) or graphs such as the “Original” in Figure 6. I have therefore always preferred use of black/white/grey instead of colors, which was relatively easy to argue for in the old days before color printers became common, but I acknowledge that the world today has moved beyond black/white figures.
As it is clear from the paper that it is relatively easy to choose colors and figure layouts that do not pose problems to the CVD minority and still ensure efficient visualization and communication, I am convinced that all scientists would like to do so, if there were aware of the problem. Given the knowledge that apparently exist on how easily to overcome the problem on the one side and the total lack of progress on the other side, there appears, however, to be a lack of attention in the scientific community.
I therefore hope that the message in this paper will be spread across the scientific community, so that many of the useful practical suggestions presented in Section 3 will be adopted by future authors. In my view scientific journals/publishers have an important responsibility in promoting this – and a responsibility that goes beyond putting recommendations into author guidelines without checking compliance. I suggest that the authors consider making recommendations on the roles and responsibilities of the journals and publishers.

Citation: https://doi.org/10.5194/hess-2021-118-CC1
- AC1:
  'Reply on CC1', Michael Stoelzle, 22 Apr 2021
  We appreciate the positive comment from Jens C. Refsgaard on our paper. It is good to have a first-hand account how much colour issues in science communication impact people with CVD. From this CC we extract three important points to improve our paper:
  Rainbow color map and red/green color use issues are not only relevant for published papers but also in (online) presentations where the audience has less time to encode data variability represented by color variation. We will incorporate this point in our reasoning against the rainbow color map/red-green encoding.
  
  We absolutely agree that progress in this topic could only be done if more and more scientists are aware of the flaws of the rainbow color map and other colour issues. We think this paper contributes to this aim. However, we will revise the manuscript to give clear guidance how awareness can be increased.
  
  The discrepancy between recommendations and compliance is an important point that is also mentioned by other reviewers of this paper. We will make more precise suggestions for journals, editors and reviewers (and their roles) in order to ban visualizations with the above mentioned colour issues instead of having only (vague) recommendations to avoid those color maps.
  
  Citation: https://doi.org/10.5194/hess-2021-118-AC1
RC2:
'Comment on hess-2021-118', Anonymous Referee #2, 26 Mar 2021
General comments

This paper tackles the question of color palettes, which is important for decodability of figures by individuals with color vision deficiency and more generally, for scientific interpretability.

The authors find that “36% of publications in HESS in 2020 had visualisations that were not scientifically correct, not perceptually uniform, and difficult to access for around 4% of the readership due to color vision deficiency”. I particularly like the way the authors evaluate reviewer comments, and highlight the lack of awareness among both reviewers and editors.

Overall, I fully support this paper. It raises an important issue; the analyses are novel and technically sound; the paper is well written, and it provides helpful solutions. Figure 7 is particularly useful as it highlights examples of “poor” and “good” color palettes clearly.

Specific comments

The paper focusses mainly on color vision deficiency (CVD) and people with low/reduced vision, however one might also argue that good visualisation and labelling is equally important for people with other cognitive differences, such as (I’m guessing) dyslexia. Has there been research on this? If so, this aspect might be worth including in your literature review.

The discussion of color palette type (negative-to-positive, strictly negative, or strictly positive) comes a little late in the manuscript (Figure 7c-d). It might be worth describing the type of color gradient that is most suited for negative-to-zero, negative-to-positive, and zero-to-positive scales sooner; e.g. a red-white-blue palette, which is currently missing from Figure 1.

It might be helpful to provide the readers with a “checklist” of items to verify when creating a readable scientific figure (e.g. “the data-ink ratio”; “a white mid-point at zero for negative-to-positive palettes”).

Some repetition could be avoided, e.g. section 3.4 also contains some repetitions about CVD etc; perhaps it could be condensed a little.

It was useful to read about the colorblind options in R packages. Are there similar options for Python users?

I wondered if the paragraph about preprints (l.89-95) was really useful. It seemed to me this was a small sample compared with the analyses in subsequent paragraphs, so the utility wasn’t entirely clear.

Lines 212-218 and elsewhere mention various types of visualisation (e.g. heatmaps at l.252), but it might be helpful to see examples (especially examples of good hydro-climatological visualisations).

Finally, the title focusses on the hydrologic community but there were large parts of the text that were not specifically hydrological. Perhaps this could be strengthened a little. For example, Figure 1 could provide examples of hydroclimatic variables (i.e. highlighting which types of palettes are particularly suitable for specific variables).

Technical corrections

1.14 raise awareness how -> raise awareness of how

1.14 the rainbow color maps still is -> the rainbow color map still is

1.23 we sketch a way to improve the communication of rainbow flaws -> we outline an approach to ?? (unclear what is meant by ‘improve the communication of rainbow flaws’)

l.31 10 millions of unique -> 10 million unique

l.35 In terms of correct encoding, (comma needed for meaning)

l.36 “we are stronger in encoding..”: meaning could be clarified

l.42 “uses” -> “used”

l.47 the word “shares” is used instead of “percentages” (here and elsewhere); perhaps consider replacing for clarity

l.61 the term “perceptual uniform” needs to be explained. I would recommend replacing “perceptual uniform” with “perceptually uniform” throughout the paper. It is explained at line 309, but this comes too late.

l.87 notable -> notably

l.114 “a graph with two lines encoding continuous variables over time without any annotations… is classified as rainbow-related”: worth providing examples alongside A-D for clarity?

l.127 “a vision deficiency scale” – terminology could be clearer.

l.139 “two cross checks… led (not lead) to minor deviations”: if this information is included, then it might be worth specifying what “minor deviations” means and how many people are in the cross-checking and original reviewer teams.

l.149 It might be worth justifying the choice of journals – why were Sci Rep and NComms selected?

l.168 “a current redistribution of disappearing black and white papers into papers with and without color issues”. I think this means something like ‘coincidence between the decline of black and white papers and the emergence of papers with color issues’

l.186 less -> fewer

l.190 73-92% of how many? A little unclear why two numbers here.

l.193 four “suggestions” perhaps

l.233 ScientistS

l.237 “a pointedly use of” is unclear

l.246 “luminance” is unclear (also used elsewhere in manuscript). Does it mean transparency? Shading?

l.84 rise awareness -> raise

l.289 parts of science -> areas of science

Figures

Figure 1. “The same delta changes in values”: this could be rephrased for clarity; it is not entirely clear what the “+1” on the figure or in the caption refer to. Also, is “perceptual uniform” “perceptually uniform”?

By this point in the manuscript (line 70), I think if would be helpful to distinguish the colors used for scales that range from negative to positive (e.g. “red white blue”) and those that are “strictly positive” or “strictly negative”.

Figure 2. I wonder if it might make more sense to show the % of red-green or rainbow color maps as a fraction of the total number of papers (instead of just the papers with color issues).

Figure 4 seems clear to me.

Figure 5 is a little unclear. I wonder if examples (of alternatives, properties, tools etc.) could be provided for clarity?

Figure 6.

Dark background – is this supposed to be easier or harder to read?

Panel c. is it brightness or shades?

Panel i. beyond the graph title, good labelling can also be helpful. Historically, many journals have discouraged the use of labels on figures; but for some people, clear panel/facet labelling can help greatly. Perhaps this is worth a mention.

Also worth making sure that all panels are referred to in the main text.

Figure 7.

OrangeRed and Batlow are almost too small to read; would recommend deletion.

Do you mean “white strokes decrease the data-ink-ratio” (rather than increase)?

P.S. Is the correct technical term “color map”, “color scale”, “color palette”, or “color gradient”?
Citation: https://doi.org/10.5194/hess-2021-118-RC2
- AC4:
  'Reply on RC2', Michael Stoelzle, 06 May 2021
  We thank the reviewer for their supportive comments and advice.
  Specific comments in RC2
  The paper focusses mainly on color vision deficiency (CVD) and people with low/reduced vision, however one might also argue that good visualisation and labelling is equally important for people with other cognitive differences, such as (I’m guessing) dyslexia. Has there been research on this? If so, this aspect might be worth including in your literature review.
  
  The research in regard to cognitive difference and perception of data visualization is relatively limited (Lee et al, 2020; Wu et al, 2019) and mostly focused on general data visualization and not scientific data visualization. The main point when designing a data visualization for intellectual development disabilities seems to be the chart type (bar chart vs line chart vs pie chart) than visualization color (Wu et al, 2019). While we recommend scientists take advice on making their visualizations more accessible, the aim of our paper is the choice of color and less the choice of chart type.
  The discussion of color palette type (negative-to-positive, strictly negative, or strictly positive) comes a little late in the manuscript (Figure 7c-d). It might be worth describing the type of color gradient that is most suited for negative-to-zero, negative-to-positive, and zero-to-positive scales sooner; e.g. a red-white-blue palette, which is currently missing from Figure 1.
  
  It might be helpful to provide the readers with a “checklist” of items to verify when creating a readable scientific figure (e.g. “the data-ink ratio”; “a white mid-point at zero for negative-to-positive palettes”).
  
  Some repetition could be avoided, e.g. section 3.4 also contains some repetitions about CVD etc; perhaps it could be condensed a little.
  
  It was useful to read about the colorblind options in R packages. Are there similar options for Python users?
  
  I wondered if the paragraph about preprints (l.89-95) was really useful. It seemed to me this was a small sample compared with the analyses in subsequent paragraphs, so the utility wasn’t entirely clear.
  
  We will restructure, shorten and clarify the manuscript as recommended by the reviewer. This will include a checklist of recommended steps and available tools to reach an accessible data visualization as also suggested by Thorsten Wagener in RC3. The checklist aims to include recommendations for the three main coding languages used in Hydrology: R, Matlab and Python.
  Lines 212-218 and elsewhere mention various types of visualisation (e.g. heatmaps at l.252), but it might be helpful to see examples (especially examples of good hydro-climatological visualisations).
  
  Finally, the title focusses on the hydrologic community but there were large parts of the text that were not specifically hydrological. Perhaps this could be strengthened a little. For example, Figure 1 could provide examples of hydroclimatic variables (i.e. highlighting which types of palettes are particularly suitable for specific variables).
  
  For some often-used variables, such as temperature, precipitation or to indicate change in a variable, the IPCC Visual Style Guide for Authors (https://www.ipcc.ch/site/assets/uploads/2019/04/IPCC-visual-style-guide.pdf) offers some guidance for color choice. We have mentioned this issue in the manuscript (L327-330) but will revise the paragraph to give more guidance for variable-specific color maps. Beyond the regularly used variables, no obvious color association exists for some hydrological variables. For example, isotope concentration charts do not have obvious colors associated with them, as long as different concentration time series can be distinguished. Instead of providing rigid choices of color, it would therefore be more useful to offer tools for an appropriate choice of color.
  Of course, we will address the technical and minor suggestions in the revised manuscript. We thank the reviewer also for the valuable comments on the Figures.
  References
  Lee, B., Choe, E. K., Isenberg, P., Marriott, K., & Stasko, J. (2020). Reaching broader audiences with data visualization. IEEE computer graphics and applications, 40(2), 82-90.
  
  Wu, K., Tanis, S., & Szafir, D. (2019). Designing communicative visualization for people with intellectual developmental disabilities.
  
  Citation: https://doi.org/10.5194/hess-2021-118-AC4
RC3:
'What role does human laziness play?', Thorsten Wagener, 29 Mar 2021

This is a nice study by Stölzle and Stein which add to the growing guidance literature on how to best visualize data. Given that the earlier reviews already stressed several aspects, I’d like to suggest three aspects for the authors to consider. They are in no particular order:

[1] Maybe I missed this in the paper, but how many software packages (Matlab etc.) offer a rainbow colour scheme as the default? Do the authors simply use the default and not think about it? This would be my personal hypothesis based on my own past mistakes. If many software packages offer this as default scheme, then is there just a straight mapping of default schemes and schemes used? If so, then would the best strategy to approach the software producers to change their schemes (rather than focus on the users)? How much does the use of rainbow colour schemes correlate with the default colour scheme in the software used (do the authors have the data to calculate this)?

[2] The authors discuss in section 3.3. that tools like colorbrewer2.0 and others can be used to avoid issues for colour blind people. Tools like these offer a much wider help to avoid a wide range of colour issues discussed in this paper. Do the authors not think that a general use of such tools would avoid most errors they discuss? Basic use of such tools for all colour choices would solve most of the problem, why not suggest this as a standard? Would this be easier than a list of things that the scientist has to check separately?

[3] Point two leads me to my third point. The authors state at the end of their paper that “As a guide we presented manifold visual techniques…”. This is great for those highly motivated to do the right thing in terms of publishing visualizations, but there is a risk that this will be too much for many scientists. Is there a simpler step-wise guide the authors could propose? My personal strategy is to require all my students to use colorbrewer 2.0 to ensure that major errors are avoided, but maybe the authors could summarize their suggestions into a few key points?

Overall, this is a very nice and valuable paper and my suggestions do not require more than minor revisions to the actual manuscript. I personally think (and the authors do not have to share this opinion of course) that much is to do with laziness. Hence thinking about how we have a better starting point (change software defaults?) and what a simple strategy might look like that scientists can easily adopt (e.g. use existing tools as a standard) would be my suggestions to fight this.

Thorsten Wagener

Citation: https://doi.org/10.5194/hess-2021-118-RC3
- AC3: 'Reply on RC3 by T. Wagener', Michael Stoelzle, 29 Apr 2021
  
  We thank Thorsten Wagener for his well-conceived comments on our manuscript. Find our answers [1-3] below:
  [1] This is a very interesting point. We do not have a data set on the correlation between rainbow color map occurrence and the used software to produce the corresponding visualizations. But from our paper survey it is obvious that authors often use the standard color scheme, e.g., in R statistics to generate graphs such as line or point plots. As we mentioned in L41-43 the color preset of R version 3.x uses “black, red, green and dark blue” for the first four data sets, and color confusion is hence a logical consequence. MATLAB changed its default color system in 2014. Instead of blue, green, red and turquoise for the first four data points the colours are now blue, orange, yellow, purple. And instead of “jet”, the MATLAB rainbow equivalent, “parula” is used. However, “parula” is not considered as 100% perceptual uniform either as the luminance gradient is steeper at the edges and flatter in the middle of the color map. The parula color map could be found in a lot of papers from our survey. Same presets might exist in other software or visualization products/tools. However, advocating the companies to change a non-colorblind safe preset might force authors to implement their former choice as they are used to it. We assume that often the reason for the fame of visualization with the rainbow color maps is – as we wrote (L38-40) – that these figures look really colorful, appealing and attract attention. We guess that the authors also think that rainbow color boost their visualizations in a way to be more impressive or outstanding. However, a lot of software products and programming languages offer the possibility to load your own color maps. In other words, we think you cannot forbid the scientist to use a specific color map, but we might have possibilities (or at least make some efforts) to forbid the authors to publish papers with rainbow visualizations (and also red-green-figures) as the rainbow color map is considered to be scientifically incorrect.
  [2] Thanks for this important comment. Yes, it seems to be reasonable to add a precise statement on potential standard tools (like colorbrewer2.0). We will revise the Section “Improve color” accordingly and will also revise the Section “trust color” arguing that a cross-check (e.g., with a CVD emulator) helps the authors to be more confident about the use of color in their visualizations.
  [3] We think ColorBrewer 2.0 (https://colorbrewer2.org) is great tool to start with when thinking about figures and creating scientific visualizations with appropriate and valid color maps. However, it focuses on cartography and might be not sufficient for color choices for other graph types. ColorBrewer 2.0 is also limited in some ways, e.g., for qualitative color maps with 5 or more distinct colorblind-friendly colors the tool fails to present an appropriate color map. However, we agree on the point that more color choice guidance should be given to the reader and the proposed step-wise guide seems to be promising. Therefore, we will add a table (or similar) to the revised manuscript that can be used as a checklist or a step-wise guide.
  
  In our opinion laziness is an understandable explanation for having that much color issues in scientific visualizations, but it is not good reason. With recent updates to the default colour map in MATLAB, R or others it seems that often the reason is a laziness to change anything (i.e., if others have used these colours in their publications it must be okay). We think such laziness could be fought with more rigorous obligations to check your own visualizations for CVD. Here we see the role of the journal (compare also our answer to CC by J.C. Refsgaard, https://doi.org/10.5194/hess-2021-118-CC1) in moving from a recommendation to an obligation by updating the author guidelines regarding the use of color. It is then the role of editors (and may be also the reviewers) to insist on this journal obligation later on. This would encourage authors to overcome “laziness” or personal preferences more reliably. We will strengthen this point in the revised manuscript.
  
  Citation: https://doi.org/10.5194/hess-2021-118-AC3

Michael Stoelzle and Lina Stein

Data sets

HESS Paper survey data Michael Stoelzle and Lina Stein https://github.com/modche/rainbow_hydrology

Michael Stoelzle and Lina Stein

Viewed

Total article views: 2,369 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
1,646	698	25	2,369	10	7

HTML: 1,646
PDF: 698
XML: 25
Total: 2,369
BibTeX: 10
EndNote: 7

Views and downloads (calculated since 09 Mar 2021)

Month	HTML	PDF	XML	Total
Mar 2021	887	411	14	1,312
Apr 2021	230	110	7	347
May 2021	275	93	3	371
Jun 2021	254	84	1	339

Cumulative views and downloads (calculated since 09 Mar 2021)

Month	HTML	PDF	XML	Total
Mar 2021	887	411	14	1,312
Apr 2021	230	110	7	347
May 2021	275	93	3	371
Jun 2021	254	84	1	339

Viewed (geographical distribution)

Total article views: 1,899 (including HTML, PDF, and XML) Thereof 1,899 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 18 Jun 2021

Short summary

Here, a paper survey scanning 1000 papers found that 45 % of the scientific papers used rainbow color maps or red-green visualizations. Those rainbow visualizations, although attracting media's attention, will not be accessible for up to 8–10 % of the people due to color vision deficiency (CVD) and distorts and misleads scientific communication. The study gives guidance how to avoid, improve and trust color and how the flaws of rainbow color maps could be communicated in science.


Total:	0
HTML:	0
PDF:	0
XML:	0