Josh is the lead data visualizer and cartographer at NASA's Earth Observatory. Prior to coming to NASA, he was working on a PhD in geography at Penn State while on an NSF IGERT fellowship in Big Data Social Science. One time he made an eclipse map. More about Josh and his work can be read on his personal website.
Q: How did you get into GIS?
A: My undergrad studies started out all over the place, and I had no idea what GIS even was until I was almost through with college. As a freshman I studied graphic design, following a lifelong interest in all things visual. But after the first year I got interested in photography, but shortly thereafter I switched majors again, this time to computer science. I briefly considered what graduate school in comp sci might be like before being a little “homesick” from more artistic work; design, ultimately, was where my heart was.
During my junior year I stumbled upon a thing called Geographic Information Science in the list of majors at Michigan State University. Analysis and design, with a side of engineering? I changed my major that semester and have been hooked ever since.
While I bounced around between those majors, the bits and pieces I picked up were like little drops of experience that coalesced into the perfect preparation for a career in cartography and visualization. I didn’t know it at the time, but I couldn’t have planned it better if I tried.
Q: What do you do for NASA? Please describe your typical day on the job.
A: I sort of wear two hats. To tell a new visual story every day, I have to quickly analyze data, create maps and charts, and help our editorial team craft articles to communicate Earth science, primarily from or related to NASA missions. I am always scrambling to get or find data and then visualize it the best I can in a very short amount of time.
Over the longer term and in-between the daily articles, I lead the development of our style guide that establishes the overall look and feel of Earth Observatory visuals. This involves defining typographic styles, color palettes, base maps, and workflows. The workflows could be anything from a set of scripts to tutorials that enable us to go from raw data to public-ready graphics in an intuitive and consistent way.
Our bread-and-butter publication is the Image of The Day, which we put out 7 days a week, 365 days a year. So my typical day usually involves creating one or more visualizations to ensure that keeps happening, while carving out time to refine our style, identify new data sources, learn new technologies, or develop tools to help our team quickly publish press-ready visuals.
Q: Your PhD thesis is titled “Cues and Affordances in Cartographic Interaction”. Could you tell us about your research, and what spurred you to focus on this particular topic? Does what you learned feed into your work at NASA?
A: This research was a lot of fun! I was primarily interested in how to communicate varying “layers” of interactivity within maps. Sometimes a map symbol might only reveal a tooltip, while other features allow analytical functions, queries, or other capabilities. Some symbology has no interactivity at all. That’s information that should be clear to the user, and the visual design of map symbols can help clue users in to whether or not (or how much) a symbol is interactive.
I started my PhD before the major UI shift toward flat design, which was a good time to have a front-row seat to the backlash that followed that trend becoming commonplace. Early popular skeuomorphic designs were a bit heavy-handed with aesthetic ornamentation. As a response, designers sort of swung (too far) in the opposite direction: many interfaces became so flat that buttons were not distinguished from other design elements. This sort of design philosophy gives people chicken hands: they are constantly pecking, trying to discover which elements on screen can be clicked.
I wanted to humanize that experience, enabling users to do more thinking and less pecking.
My research was predicated on the belief that there’s a sweet spot in the middle: many, many interfaces could benefit from subtle cues that make interactive UI components a bit more obvious.
This research helped me think more clearly about hierarchies and designing with a purpose. Every map or visualization is a layering of information, and even if there’s no interactivity in a graphic, there’s still a competition for your attention and focus. Careful design ensures the viewer is drawn to the important bits, without totally removing less important elements. I like maps that communicate a key point quickly, then draw you in, revealing more insight as you study them.
Even if your fingers aren’t pecking a screen in different spots, your eyes might be. Good design settles things down and enables readers to focus on—or be guided to—the important information.
Q: There’s a lot of kids out there who want to work for NASA someday (including my own), although most of them are probably dreaming about space shuttles. If NASA has data visualization and cartography jobs, how wide does the variety get?
A: The variety is out of this world! (That was lame, wasn’t it? But it’s true!) You’ll find people working as everything from biologists to seamstresses at NASA.
I work in the Earth Science Division, and while the exact job title of “cartographer” is not a thing as far as I know — I don’t even have it — there’s an enormous amount of geospatial analysis and mapping going on. A lot of colleagues of mine have backgrounds in other fields — oceanography, geology, etc. — but we all make maps with the same data (perhaps with different software; the geologists really love GMT). But if it happens on Earth, NASA probably has an instrument that measures it, and handfuls of people with diverse expertise analyzing it and mapping it.
Q: What kind of technology do you use on the job? Mostly open source, or mostly proprietary, or an even mix?
A: It’s a mix. I’m a bit of a generalist: I use what gets the job done. That said, it is with some privilege that I am able to make those sorts of decisions. If there’s software out there, paid or otherwise, I probably have access to it.
That said, my go-tos by and large tend to be open source. GDAL is the real MVP of my workflow, and I use QGIS daily.
My top 5 most-used tools include QGIS/GDAL, Photoshop, After Effects, Python (with matplotlib, pandas, and NumPy), and Bash.
Q: Which systems are the most common sources of satellite imagery for your work?
A: We like to show things in true color when we can; readers really enjoy seeing satellite imagery that is as easy to understand as a photograph. That places a lot of emphasis on MODIS, VIIRS, and Landsat imagery.
Q: How often is it that a new system or source of imagery becomes available?
A: All the time! While the instrument construction projects and big launches make the news a few times a year, there are thousands of scientists around the globe developing new data from all the satellites already in orbit. Algorithms are improved, data sources are combined, and new applications emerge almost around the clock.
Q: Your website has dozens of examples of beautiful and informative maps. I’m guessing it takes quite a bit of work to pull the data together into a publishable product. Can you give us an example of a workflow, going from raw satellite data to polished map?
A: Thanks! I appreciate that.
One thing I have to admit being most proud of is that these projects are done super quickly. We publish daily, so I often only have a few hours, maybe 12 hours for larger stories, to get all the data that goes into something, process it, analyze it and find the story, and then design a map (or other visuals). In the last three years, there’s only one project that I worked on for longer than a week, which was the 2012 and 2016 updates to NASA’s Black Marble maps of nighttime lights.
The biggest effort has gone into developing the styles and workflows that make it possible to publish these visualizations so quickly.
- Generate a best-pixel mosaic of the area using five years of Landsat data in Google Earth Engine. While that was running:
- Download SRTM data for the area and merge the tiles with gdal_merge.py
- Hillshade the elevation data in QGIS (or GDAL)
- Color-correct and reproject the finished Landsat mosaic
- Blend the Landsat data with the hillshade
- Finish the map up with boundaries, water bodies, and labels, export for the web
To get even more out of the data, I also used the Landsat mosaic and elevation data to render a true color view at an oblique angle. The whole story finishes with a recent, individual Landsat scene. (You can read about how to color-correct and pan-sharpen Landsat scenes in tutorials from me and Rob Simmon.)
That all came together in about four hours. There’s always so much more I wish I could do with imagery, but our tight deadlines force us to be quick and lean.
Q: You’re a moderator for the esteemed Reddit community Data Is Beautiful. Last time I logged in there were 12+ million subscribers. How long have you been moderating, and what exactly does moderating entail?
A: I’ve been moderating Data Is Beautiful since early 2014. Geeze, thinking back, it is hard to believe I am the second longest running mod in the subreddit. Back then we had about 50,000 subscribers and we were not a default shown to all visitors. We’d see maybe one popular post a week. It has grown quite a bit, and that has been awesome to witness over the years. We now have subscribers posting insanely well-done work that makes the front page of the entire site almost daily.
Each mod contributes to keeping things organized and spam-free, but most take on a labor of love depending on their interests. Early on I established the CSS design for the subreddit and our visual flair system to separate different types of posts. Other mods organize AMAs, run contests, or code up sweet bots that quantify the number of original content (OC) pieces a user has posted.
These days I am not super active as a mod; we’ve brought on a bunch of fantastic mods that really keep the sub running and growing.
Q: What do you do for fun? Any hipster traits we should know about?
A: I spend a lot of time playing with my kid (4). We’re really into Lego right now (see that, Ken? No ‘s’). My wife made the mistake of giving me Monster Hunter: World as a gift, and I haven’t been able to put it down.
I am looking forward to getting back into fishing this spring and summer. That’s a hobby I’ve neglected recently and I can’t wait to get back into it.
If a beard, a love of craft beer, and a fixed-gear bike are the criteria, I might be a hipster on paper. But in real life, I’m less like The Decemberists and more like Dexter (the awkward part, not all the other stuff…)
Q: Would you consider yourself a geohipster? Why / why not?
A:. Even though I use a lot of tools that might not be mainstream, I don’t see that as a goal or accomplishment that sets me apart. These are things that just help me do the job I need to do. So I wouldn’t say so, but maybe others might.
I’m also a bit wary (and weary) of an over-emphasis on the tools used by cartographers and the GIS community. How those tools are used, and the goals they achieve, is much more important. There’s a bit of a ‘library name drop culture’ on social media, where a long list of tech and libraries will be highlighted, and then it’s like “oh, and by the way this app ensured the four-pronged butterfly did not go extinct.”
That’s wrong, and we as a community should strive to fix that. There are social media accounts devoted to bashing this file format or that software. Why? That’s as useful as posting instagram photos of food you plan not to eat.
It was Michelangelo, not chisel brand X, who made David.
Q: On closing, any words of wisdom for our readers?
A: “[You] absolutely have to have dark in order to have light. Gotta have opposites, dark and light, light and dark, continually in painting. If you have light on light, you have nothing. If you have dark on dark, you basically have nothing. It’s like in life: you gotta have a little sadness once in a while so you know when the good times come.” –Bob Ross