Scientific visual presentation Oerlemans [2005]. What makes for a good scientific figure? Accurate Informative Easily understood Appropriate for anticipated.

Scientific visual presentation

Oerlemans [2005]

What makes for a good scientific figure?

• Accurate• Informative• Easily understood• Appropriate for anticipated audience/medium

– What’s best for an oral or poster presentation is not necessarily ideal for a manuscript, and vice versa

– Has a natural heritage

• No fluff

ENSO pattern

GRACE estimates of ice-sheet mass loss

How do you make a good scientific figure?

• Data visualization is an extension of your scientific self• Accept that making a good figure will require numerous revisions• Good figures, liked good papers, are longer lived• Is the figure a hypothesis test or a summary of data?• Is the figure a cartoon or reality?

• Have all the data ready to be plotted beforehand• Sketch it• Plan plotting stages• Recognize its heritage and shamelessly appropriate others’ good ideas• Shamelessly re-use your own good ideas• Develop a framework that ensures repeatable figure output• Use an image editor as a last resort only

Technical approach

Philosophy

More direct advice

• There is only one font: Helvetica

• There is only one font size: bigger

• There is only one system of units: S.I.

• Plotting defaults are often poor choices

• Plan figure and axis size

• Produce raster output as a last resort only

• Use the breadth of the color palette

• Three dimensions only when absolutely necessary

Color choices good and bad

• Positive–negative changes warrant a contrasting color palette• Discretize color ranges as often as possible (e.g., cbarf)

Maintain some semblance of realism in cartoons

Useful figure features that are challenging

• Multiple subplots

• Multiple axes

• Logarithmic or non-linear color maps

• Geographically accurate maps

• Transparency

• Multiple color maps

• High-quality animations

• GUI elements

incre

as

ing

diffic

ulty

The easiest/best things I can teach you re: Figures

• set(0, ‘DefaultFigureWindowStyle’, ‘docked’) always docks figures• set(0, ‘DefaultFigureWindowStyle’, ‘default’) always pop figures• Command–~ switch between current application’s windows• Command–tab switch between applications• epstopdf command-line conversion of eps figures to pdf• redblue red/blue color map• ‘CO_2’, ‘SO_4^{2-}’, ‘Temperature (\circC)’ simple LaTeX commands• linkaxes match multiple axes• again with the cells

Multiple subplots

• subplot(212)• subplot(‘position’, [0.08 0.05 0.90 0.44])

• Most subplots will require manual adjustment

x0 y0 w h

Multiple axes

• axes(‘position’, get(gca, ‘position’), ‘color’, ‘none’)• Set(gca, ‘yaxislocation’, ‘right’)

• line vs. plot• Avoid the hack that is plotyy

Logarithmic or non-linear color maps

• imagesc(x, y, log10(new_climate_mode), [0 1]) colorbar(‘ytick’, 0:0.5:1, ‘yticklabel’, {‘1’ ‘’ ‘10’})

• Cheat

GUI elements• Load data / assign variables• Design figure• Assign behavior, i.e., write sub-functions