Imagery Does Speak, But What Does Yours Say?
Good Data Visualization Matters
It’s October 1st, and as always it’s shaping up to be a very busy autumn. There’s a lot going on, between the floods we had here in Boulder, and sequestration-strangled science funding in the US now brought to a halt with the federal government shutdown. Today is also when acceptance notifications go out for AGU Fall Meeting. I’m confident about getting in, but I don’t have a guess as to whether I’ll get a poster or a talk. I like both, and they each have their upsides from scientific and business standpoints, so that at least is a no-lose situation. Our blog title here, “Imagery Speaks”, is a key part of presenting geospatial work. Imagery, and all data visualization, does speak. But if you’re not careful with your vocabulary, you’ll have your presentation saying the wrong thing. Accurately presenting data, from graphs and tables to images and maps, is the critical final step in any geospatial project.
Maybe by coincidence, but I’ve been seeing a lot of great articles, papers, and advice on doing just that. As I wait for notice on my own presentation, I figured it’s a great time to review the foundations of good geospatial communication. What graphs, data scaling, color tables, band combinations, and even what map projection you choose makes a big difference on how your results look. Even The Onion has weighed in on this before.
1) Choose the right map projection for your work. If you’re looking at changes in land cover (i.e. area is an important metric),then use a projection preserving area, like Lambert Azimuthal Equal Area. For large scale atmospheric or oceanic circulation patterns, an equal angle projection might be a better choice. Working at a governmental scale, survey-compatible projections like UTM or State Plane might be the way to go. There’s no such thing as a perfect map projection, but there are always better and worse ones for any given project.
2) Human perception is narrow and has some interesting shortfalls. Colors are not all equal in the eye of the beholder. It’s now well known that rainbow color tables present a distorted picture to your audience. Tables like Color Brewer let viewers get an accurate picture. Rob Simmon, at NASA EO, has an excellent series of articles on data visualization that are must-read material for all scientists and presenters. Edward Tufte is a staple of solid data visualization principles as well. If you do data visualization, you should follow them on Twitter, too.
3) Whether it’s a talk, poster, or presentation, never use Comic Sans font. A lot of your audience will immediately disqualify your project as serious work. Yes, it sounds trivial, but you don’t get to choose your audience’s biases and you may as well avoid the easy ones.
4) GSA Today ran a great article by Eric Cheney, No More Lousy PowerPoint Slides, which should be used as a checklist as we get ready for Fall Meeting, or any conference or presentation. Choose your images and graphics carefully. It’s far too easy to put too much detail, whether text,graphics, or images on a projected slide. And always keep your audience first in mind. Your presenting to them, you want them to understand your work. If it helps, even think of it like you’re selling your project to them (really, you are).
Best of luck to you if you’re waiting on abstract acceptance, and I hope you get the presentation you want, whether it’s a poster,talk, or seminar. Keep an eye out for good data visualization principles, and share them with us here, or on Twitter or other media. Got some great IDL or ENVI visuals? We’d love to see them!