Quick Guide: Creating Plots For Science and Accessibility

Introduction

There are a wide range of tools and libraries available to create data plots for quick sharing and for scientific outputs (e.g. plots for journal articles). However, the way that they are prepared can have an impact on how they are interpreted and their accessibility.

This quick guide seeks to provide some aspects to consider to support plot interpretation, and thus clarity on the science drawn from the related results, and to enhance their accessibility by considering aspects such as colour choices, use of symbols etc.

What to include on a plot

• Data providers will know best how their datasets can be visualised to be of most use to potential users and so NCAS is only providing guidelines rather than formal rules.
• Axes should be suitably labelled and units should be included.
• In order to allow like-for-like comparisons of plots from the same dataset, it is helpful if axes have fixed limits or fixed-width limits 
• Not all parameters available in a data file have to be included.
• Derived parameters, as opposed to those that are explicitly available in a data file, may be included.
• The data may be shown in a single plot or split across multiple, thematically-grouped plots.

Ensure that colours are easily distinguishable for all

• When using discrete colours in plots, make sure that they are distinguishable for people who have various forms of colour blindness
• When using continuously-varying colour palettes, make sure that these are also “perceptually uniform”, i.e. the brightness of the colours varies in a continuous way.
  • The commonly used rainbow palette is neither suitable for colour blind people nor for perceptual uniformity.
• Refer to Crameri et al. (2020) “The misuse of colour in science communication”  for more details about this subject.
• Suitable colour schemes can easily be found for common plotting packages.

Ensure that plots can be traced back to the underlying datasets

• Don’t rely on the availability of the plots through CEDA to indicate the datasets from which they are derived. This context will be lost as soon as the plot is downloaded.
• There is a limit to how much text information can be included on a plot so aim to include key details that will allow someone to trace it back to the underlying dataset from a web search, e.g.:
  • The official NCAS name of the instrument
  • The location where the instrument was operated
  • The version of the dataset used (if more than one is available)
  • A reference to NCAS and/or to CEDA
• Be cautious about including web addresses since these can cease to be valid after just a few years.  

The gold standard of traceability

• Image files have the capacity for text-based information to be embedded within them
• The NCAS-IMAGE metadata standard exploits this capacity to ensure that image files (including plots) contain a comparable level of metadata to netCDF data files.
• The types of information include details of:
  • the person responsible for creating the plots
  • the instrument from which the dataset was derived 
  • the CEDA catalogue page for the underlying dataset
  • how to acknowledge use of the plot (and the licence covering its use)   

Examples and useful tools

Perceptually uniform plotting colour schemes

• Crameri, F., Shephard, G.E. and Heron, P.J. The misuse of colour in science communication. Nat. Commun. 11, 5444 (2020): https://doi.org/10.1038/s41467-020-19160-7
• Coblis Color Blindness Simulator: https://www.color-blindness.com/coblis-color-blindness-simulator/
• Cameron Homeyer colormap for weather radar: https://github.com/ARM-DOE/pyart/issues/713

Example colour schemes (many more can be found from a web search)

• Fabio Crameri Scientific colour maps: https://www.fabiocrameri.ch/colourmaps/
• Kenneth Moreland Color Map Advice for Scientific Visualization: https://www.kennethmoreland.com/color-advice/
• Paul Tol’s Notes - Colour schemes and templates: https://personal.sron.nl/~pault/
• Coloring for Colorblindness: https://davidmathlogic.com/colorblind
• Matplotlib: https://matplotlib.org/stable/users/explain/colors/colormaps.html
• ProPlot colormaps (for python): https://proplot.readthedocs.io/en/latest/colormaps.html

Follow this link for access to plots created from datasets collected at the NCAS Capel Dewi Atmospheric Observatory (which was previously known as the NERC MST Radar Facility). Note that:

• These plots do not use colour blind friendly colour palettes
• They make use of embedded metadata (only those created after 2022 follow the NCAS-IMAGE metadata standard)

More information

Speak to the NCAS data team for more information.

Also the following presentation has some useful slides that cover the use of colours etc. in plots to aid scientific interpretation and accessibility:

Hooper, D. A. (2024, December 12). Data Standards for Atmospheric Science. iMST Lidar and Radar School associated with the16th International Workshop on Technical and Scientific Aspects of iMST Radar and Lidar (MST16/iMST3), Kühlungsborn, Germany. Zenodo. https://doi.org/10.5281/zenodo.14419170