Since the popularization of the term “infographic” on the Internet, this whole discussion around “good” and “bad” visualization has reached another level, thanks to the massive amount of poorly executed designs being produced out there. Among the most common errors, you’ll see the misuse of color, something that indicates that there’s still a lot of confusion when it comes to pick the right palette to help convey the message, instead of polluting and creating additional noise. And that’s why the work done by Robert Simmon – our first interviewee of 2015 – is so noteworthy.
Simmon is a data visualizer and designer currently working with Planet Labs. He had previously spent 20 years at NASA, co-founding the Earth Observatory web site, where he also wrote in the Elegant Figures blog. It was here that Robert published, in 2013, a six-part post untitled Subtleties of Color. This series quickly became a reference for both experts and beginners, and led to several keynote appearances by Robert in events such as the second OpenVis Conference and Malofiej 22.
As an expert at creating clear and compelling imagery from satellite data, Robert’s focus is on producing visualizations that are elegant and easily understandable, while accurately presenting the underlying data. He helped create some of NASA’s most widely-seen imagery, including the Earth at Night and the 2002 Blue Marble. His imagery has appeared in newspapers, web sites, and advertisements, and was featured on the login screen of the original Apple iPhone.
We had the chance to do some catching up with Robert just before the end of 2014, to talk about visualization, the challenges of scientific communication and his time at the Earth Observatory.
Visualoop (VL) – “Careful use of color enhances clarity, aids storytelling, and draws a viewer into your dataset. Poor use of color can obscure data, or even mislead”. This is a quote from your talk at the latest OpenVis Conference, and it reflects well the importance you give to this subject. When did this concern with the use of color began?
Robert Simmon (RS) – My first task at NASA was to help develop a CD-ROM about Earth’s ozone layer. The CD had more than 15 years of satellite data and a tool to visualize it, month by month. I had a sense the existing imagery was bad (by bad I don’t just mean unattractive, but also a poor representation of the underlying numbers), but I didn’t know why it was bad.
Over a year or two I created a handful of new palettes, but they weren’t much better. It wasn’t until I saw Edward Tufte show a map of ocean depth and elevation in one of his lectures (the same map is in Envisioning Information) that I understood there was an underlying theory behind using color in visualization—that’s when I began to get serious about it. I also learned that cartographers had solved many of the dataviz problems I’d been struggling with decades ago, and I should start paying attention to the history of cartography.
Unfortunately, I lost the battle with ozone data (It’s still shown with the rainbow palette (http://ozonewatch.gsfc.nasa.gov), but I’ve had some influence over the colors used at NASA (http://neo.sci.gsfc.nasa.gov) and NOAA (http://climate.gov/maps-data).
VL – Apart from color, what other misuses and common errors you see, when presenting scientific information visually?
RS – One of the things I see too much of is extraneous detail — instead of simply filling in areas of no data with a neutral color, visualizers use a true-color map or exaggerated terrain. If data is worth showing, it can stand on its own (with appropriate scales and annotations, of course). Stuffing white space with extraneous detail just distracts from the important information.
Another is 3D. Our visual systems are very good at determining precise locations in two dimensions: left and right, up and down. We’re much less able to perceive depth (see Colin Ware’s Information Visualization: Perception for Design). This is even more true when three-dimensional space is compressed into the two dimensions of a page or screen. Without stereo vision we’re reduced to secondary cues like shading and motion parallax. Yet many many scientific visualizations—even (especially?) award-winning ones—naïvely attempt to render data in full 3D. The results are often impressive, but rarely communicate much quantitative information.
Serious multi-dimensional data visualization remains a huge challenge, and requires considerable thought.
VL – Historically speaking, who’s groundbreaking work would you recommend someone in the beginning of his/her journey into the world of scientific data visualization, to explore?
RS – My favorites are a pair of papers by Bernice E. Rogowitz and Lloyd Treinish “Why Should Engineers and Scientists Be Worried About Color?” and “How NOT to Lie with Visualization” and, of course, Cynthia Brewer’s Color Brewer.
In the past year, there’s been a series of articles attempting to improve the use of color in scientific visualization: “The End of the Rainbow” by Ed Hawkins, Doug McNeall, David Stephenson, Jonny Williams & Dave Carlson. “Somewhere over the rainbow: How to make effective use of colors in meteorological visualizations” by Reto Stauffer, Georg J. Mayr, Markus Dabernig, and Achim Zeileis, published in the Bulletin of the American Meteorological Society (they also made a tool for creating new palettes: HCL Wizard. And Mathworks even replaced the default color palette in Matlab with a new, perceptually-based one. My series of color posts is even mentioned in the whitepaper explaining the rationale (pdf), so that’s one item off my bucket list.
VL – Rob, there’s this general sense that scientists still struggle a bit, when the time comes to communicate their findings and studies to a non-scientific audience – either it’s in text, graphics, slides. In your opinion, what can be done to solve this – or at least, evolve more rapidly?
RS – Communicating science is hard. There’s an even an entire journal, Public Understanding of Science, devoted to the topic.
That said, I think many (maybe most) scientists could do a better job explaining their research to non-scientists, without too much extra work. It’s fundamentally a problem of familiarity. If you’ve been thinking about and working on something for year or decades, it’s hard to put yourself in the position of someone new to the topic. I certainly find myself assuming people already know the concepts and jargon I use when I talk about visualization—the precise difference between hue, saturation, and lightness, for example. You just need to remember to step back and explain things. It’s especially helpful to give concrete examples, and put new information in a familiar context.
Scientists should also listen to the advice of people who specialize in science communication, especially if their approach is research-based.
VL – And what about some nice examples, what good scientific visualizations have caught your eye, recently?
RS – So many. XKCD’s map of the California drought . A Hovmöller diagram is one of the more esoteric plots, but this example explains what is being shown effortlessly. It doesn’t even need a caption.
Chloe Whiteaker (at Bloomberg) working with NASA’s Gavin Schmidt recently did this fantastic graph of changes in global temperatures. Instead of plotting the annual average, she plotted the 30-year trend for every year since 1880. It elegantly removes all of the short-term noise that can give the false impression that the rate of warming is slowing. (It’s not—the Earth is getting hotter as fast as ever).
Another thoroughly depressing subject, but I find this visual exploration of extinctions by Anna Flagg (ProPublica) delightful.
Although I’m often critical of NASA visualizations, I like a lot of what they publish. Especially these small multiples showing the decline in Arctic sea ice. Make sure you check out the full-resolution version.
VL – Open Vis Conference was great, I imagine. In fact, you also attended another amazing event, Malofiej, earlier last year! Both have their obvious differences, but were there any similarities, interesting topics or trends that you saw discussed or at least, mentioned, in Pamplona and in Boston?
RS – My primary impression of both was how incredibly gracious and friendly everyone was. (Thanks to Alberto Cairo for the invitation to Malofiej, and Lynn Cherny for asking me to submit an abstract to the Open Vis Conference.)
Almost everyone I’ve met in the visualization community is eager to share ideas and help out. Criticism is common, but it’s constructive criticism. I think that’s really helped the field grow in recent years, and improve the overall quality of data visualization.
Another commonality was the emergence of data visualization in the newsroom. That’s obviously a focus at Malofiej, but 3 speakers at OpneVizConf also work in journalism: Lena Groeger, Kennedy Elliot, and Jen Christiansen.
VL – We’re seeing more and more events, year after year. Was there any particular one you missed, just to regret it later?
RS – All of the ones I didn’t attend? I say that partially tongue-in-cheek, but If I could afford the time I’d go to one conference a month. I most regret missing the 2013 Gordon Research Conference (GRC) in Visualization in Science & Education. I had been going since 2005 and couldn’t in 2013 because of NASA travel restrictions. I should have just taken the time off. It’s particularly interesting to me because most of the presentations are based on research into human perception and cognition.
VL – Over six months ago, you announced that you’d be leaving NASA to embrace a new challenge, with Planet Labs. What can you tell us about your responsibilities over there?
RS – My primary job at Planet Labs is currently to make sure the the imagery — particularly our forthcoming global mosaics (a much, much higher-res version of the NASA Blue Marble) — looks right. By “right” I mean not how the satellites see the Earth, or how an astronaut would, but how we expect the Earth to look. Think of the view a low-flying bird has, rather than the fuzzy and faded view from above the atmosphere.
I’m also working a little bit on internal data presentations, and will likely shift focus to data products derived from Planet Labs imagery when we have a full constellation of dives up and running.
VL – And how hard it was to leave the Earth Observatory team?
RS – Honestly, I thought I’d be balling as I walked out the door (and maybe I did get a little dust in my eyes), but I didn’t leave myself any time off so I was too busy to think about it much. Fortunately, I was able to catch up with some of my co-workers at AGU (the year’s biggest Earth science conference), which is in San Francisco every December.
NASA provided me the time and space to learn data visualization on the job (I have degrees in materials science and engineering; not design, visualization, or computer science), which I don’t think I would have had anywhere else. They also have an incredible amount and variety of important data, and scientists who knew it backwards and forwards, which made it easy to go into work.
I’m currently looking forward to what Josh Stevens (http://www.joshuastevens.net/) does in my old position: he started on January 5th, I think.
VL – To close, you shared the images and visualizations created during your time at NASA that stroke you as your favorites. Revisiting that list, care to add any other?
I should have made it 11: I forgot to add this photo from the ISS, with a quote from Earthshine, by Neil Peart. I’ve been a fan of Rush since elementary school (I’m sure my 6th-grade classmates can still confirm this) and I was thrilled to get permission from the band to use the lyrics. They matched perfectly, and gave me the opportunity to explain some cool science.
VL – Thank you so much, Rob!
RB – Thank you!
We really appreciate the time Robert dedicated answering to our questions – specially in this time of the year. You can connect with him on Twitter (@rsimmon).