際際滷

際際滷Share a Scribd company logo

Effective Use of Color in Scientific Visualization

Download as PPTX, PDF
R
Riley X. Brady

This document discusses effective use of color in scientific visualization. It outlines different types of color maps: sequential for ordered data, diverging to show anomalies from a reference value, and qualitative for categorical data. Examples are provided. The document advises against using the rainbow color map due to perceptual and interpretability issues. It recommends colorblind-friendly colormaps from resources like ColorBrewer and CMOCEAN. Masking, showing uncertainty ranges, and anomalies from a reference are also discussed.

Convert to study guideBETA

Transform any presentation into a summarized study guide, highlighting the most important points and key insights.

1 of 21
Downloaded 17 times
EFFECTIVE USE OF COLOR IN SCIENTIFIC VISUALIZATION Riley X. Brady
THEORY: COLOR CHANNELS Munzner 2014
THEORY: SEQUENTIAL COLOR MAPS Sequential color maps are useful for data with some regular interval, e.g., raw temperature or salinity. They should have a monotonic increase in luminance and can have hue changes. These color maps are unambiguous when printed in black and white. Grayscale Monochromatic Multi-hue Thyng et al. 2016
EXAMPLE: SEQUENTIAL COLOR MAP Krumhardt et al. 2017 @zlabeonTwitter
THEORY: DIVERGING COLOR MAPS Diverging color maps are useful for data with reference to some critical value, e.g., anomalies from some climatology. They should be composed of two sequential color maps, but with mirrored luminance values. These color maps are ambiguous when printed in black and white. This can be avoided by overlaying solid and dashed contour lines. Blue to Red Blue-green to Brown Spectral Thyng et al. 2016
EXAMPLE: DIVERGING COLOR MAP Differing contours for grayscale readability Sarmiento and Gruber 2006 Combination of diverging and sequential @rileyxbrady on Twitter (from E3SM model run)
THEORY: QUALITATIVE COLOR MAPS Qualitative color maps are useful for data with distinct categories (e.g., political parties, species) and are generally distinguished by different hues of equal luminance Accent Paired Set1
EXAMPLE: QUALITATIVE COLOR MAP Krumhardt et al. 2017
EXAMPLE: QUALITATIVE COLOR MAP Berliner Morgenpost
ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. Borland and Taylor 2007Thyng et al. 2016
ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. Hawkins2016
ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. 3. Its difficult to interpret for people with colorblindness. Staufferetal.2015
ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. 3. Its difficult to interpret for people with colorblindness. FRIENDS DONT LET FRIENDS USE JET
BE MINDFUL OF COLOR BLINDNESS Use a colorblindness simulator Avoid using red and green together in line plots, choropleths, etc. As many as 8% of men and 0.5% of women have some form of colorblindness. Think about your audience!
COLORS SHOULD BE INTUITIVE sea ice precipitation changes Pendergrass et al. 2017
RESOURCE: COLORBREWER Strings work natively in matplotlib (use e.g. BuGn_r to reverse) Download the Brewermap package Download the RColorBrewer package
RESOURCE: CMOCEAN Available via packages for python, MATLAB, R, GMT, Paraview, https://github.com/matplotlib/cmocean Although designed with the oceanographer in mind, these colormaps are useful for all fields. They are perceptually uniform and colorblind friendly.
MORE COLOR MAP RESOURCES Color maps http://soliton.vm.bytemark.co.uk/pub/cpt-city/ https://sciviscolor.org/ Color palettes https://colorhunt.co/ https://colordrop.io/ https://color.adobe.com/explore Design your own http://tristen.ca/hcl-picker http://gka.github.io/palettes Testing https://projects.susielu.com/viz-palette
Mask out missing or uncertain data IPCC AR5
Show a range of uncertainty Jamie Scott / Climate Change Web Portal
Show an anomaly relative to reference cases Maarten Lambrechts

More Related Content

Effective Use of Color in Scientific Visualization

  • 1. EFFECTIVE USE OF COLOR IN SCIENTIFIC VISUALIZATION Riley X. Brady
  • 3. THEORY: SEQUENTIAL COLOR MAPS Sequential color maps are useful for data with some regular interval, e.g., raw temperature or salinity. They should have a monotonic increase in luminance and can have hue changes. These color maps are unambiguous when printed in black and white. Grayscale Monochromatic Multi-hue Thyng et al. 2016
  • 4. EXAMPLE: SEQUENTIAL COLOR MAP Krumhardt et al. 2017 @zlabeonTwitter
  • 5. THEORY: DIVERGING COLOR MAPS Diverging color maps are useful for data with reference to some critical value, e.g., anomalies from some climatology. They should be composed of two sequential color maps, but with mirrored luminance values. These color maps are ambiguous when printed in black and white. This can be avoided by overlaying solid and dashed contour lines. Blue to Red Blue-green to Brown Spectral Thyng et al. 2016
  • 6. EXAMPLE: DIVERGING COLOR MAP Differing contours for grayscale readability Sarmiento and Gruber 2006 Combination of diverging and sequential @rileyxbrady on Twitter (from E3SM model run)
  • 7. THEORY: QUALITATIVE COLOR MAPS Qualitative color maps are useful for data with distinct categories (e.g., political parties, species) and are generally distinguished by different hues of equal luminance Accent Paired Set1
  • 8. EXAMPLE: QUALITATIVE COLOR MAP Krumhardt et al. 2017
  • 9. EXAMPLE: QUALITATIVE COLOR MAP Berliner Morgenpost
  • 10. ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. Borland and Taylor 2007Thyng et al. 2016
  • 11. ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. Hawkins2016
  • 12. ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. 3. Its difficult to interpret for people with colorblindness. Staufferetal.2015
  • 13. ASIDE: AVOID THE RAINBOW 1. It is not perceptually or logically ordered, despite it being a sequential color map. 2. Sudden luminosity changes create false boundaries in data. 3. Its difficult to interpret for people with colorblindness. FRIENDS DONT LET FRIENDS USE JET
  • 14. BE MINDFUL OF COLOR BLINDNESS Use a colorblindness simulator Avoid using red and green together in line plots, choropleths, etc. As many as 8% of men and 0.5% of women have some form of colorblindness. Think about your audience!
  • 15. COLORS SHOULD BE INTUITIVE sea ice precipitation changes Pendergrass et al. 2017
  • 16. RESOURCE: COLORBREWER Strings work natively in matplotlib (use e.g. BuGn_r to reverse) Download the Brewermap package Download the RColorBrewer package
  • 17. RESOURCE: CMOCEAN Available via packages for python, MATLAB, R, GMT, Paraview, https://github.com/matplotlib/cmocean Although designed with the oceanographer in mind, these colormaps are useful for all fields. They are perceptually uniform and colorblind friendly.
  • 18. MORE COLOR MAP RESOURCES Color maps http://soliton.vm.bytemark.co.uk/pub/cpt-city/ https://sciviscolor.org/ Color palettes https://colorhunt.co/ https://colordrop.io/ https://color.adobe.com/explore Design your own http://tristen.ca/hcl-picker http://gka.github.io/palettes Testing https://projects.susielu.com/viz-palette
  • 19. Mask out missing or uncertain data IPCC AR5
  • 20. Show a range of uncertainty Jamie Scott / Climate Change Web Portal
  • 21. Show an anomaly relative to reference cases Maarten Lambrechts

Editor's Notes

  • #3: Munzner textbook Luminance brightness/darkness Saturation intensity or purity of the color Hue the color itself
  • #4: First example just uses luminance channel for ordered data Second example selects a single hue and uses luminance channel Third example uses multiple hues and luminance channel
  • #5: Krumhardt et al. 2017 Coccolithophore growth rate
  • #6: Note that the values of your plot should actually center around that critical value
  • #7: NOTE: Make sure the diverging color map centers around an intelligible critical point.
  • #8: Note that the values of your plot should actually center around that critical value
  • #9: Different species of coccolithophore Side note: think about other vis channels for communicating information, e.g. hatching.
  • #10: Different species of coccolithophore Side note: think about other vis channels for communicating information, e.g. hatching.
  • #11: http://www.climate-lab-book.ac.uk/2016/why-rainbow-colour-scales-can-be-misleading/ https://journals.ametsoc.org/doi/pdf/10.1175/BAMS-D-13-00155.1 https://jakevdp.github.io/blog/2014/10/16/how-bad-is-your-colormap/ https://eagereyes.org/basics/rainbow-color-map http://medvis.org/2012/08/21/rainbow-colormaps-what-are-they-good-for-absolutely-nothing/
  • #12: http://www.climate-lab-book.ac.uk/2016/why-rainbow-colour-scales-can-be-misleading/ https://journals.ametsoc.org/doi/pdf/10.1175/BAMS-D-13-00155.1 https://jakevdp.github.io/blog/2014/10/16/how-bad-is-your-colormap/ https://eagereyes.org/basics/rainbow-color-map http://medvis.org/2012/08/21/rainbow-colormaps-what-are-they-good-for-absolutely-nothing/
  • #13: http://www.climate-lab-book.ac.uk/2016/why-rainbow-colour-scales-can-be-misleading/ https://journals.ametsoc.org/doi/pdf/10.1175/BAMS-D-13-00155.1 https://jakevdp.github.io/blog/2014/10/16/how-bad-is-your-colormap/ https://eagereyes.org/basics/rainbow-color-map http://medvis.org/2012/08/21/rainbow-colormaps-what-are-they-good-for-absolutely-nothing/
  • #14: http://www.climate-lab-book.ac.uk/2016/why-rainbow-colour-scales-can-be-misleading/ https://journals.ametsoc.org/doi/pdf/10.1175/BAMS-D-13-00155.1 https://jakevdp.github.io/blog/2014/10/16/how-bad-is-your-colormap/ https://eagereyes.org/basics/rainbow-color-map http://medvis.org/2012/08/21/rainbow-colormaps-what-are-they-good-for-absolutely-nothing/
  • #17: Click on link and go to website. Demo sequential, diverging, qualitative color maps. Show colorblind safe.
  • #18: Cmocean Perceptually uniform http://tos.org/oceanography/assets/docs/29-3_thyng.pdf
  • #22: Maarten Lambrechts