Science on the Map

TOOLBOX SCIENCE ON THE MAP

Easy-to-use mapping tools give researchers the power to create beautiful visualizations of geographic data. ILLUSTRATION BY THE PROJECT TWINS THE PROJECT BY ILLUSTRATION

BY MARK ZASTROW tracing a map off an atlas so that I had some- by the company Mapbox, which has offices in thing a bit cleaner to work with.” The one after San Francisco, California, and in Washington hen linguist Lauren Gawne roams that — “using Google Earth and dropping pins DC. It lets users create maps from their data the valleys of Nepal documenting on it” — was generic, ugly and “looks horrible and pre-existing online cartographic databases. endangered Tibetan languages, in a PowerPoint”. TileMill is just one tool in the emerging field Wshe takes pains to distinguish each dialect’s So in 2013, she jumped at the chance to join a of customized mapping, where a bevy of open- geographical origin. But when it came to pro- workshop on mapping and visualization at the source technologies and start-ups have given ducing maps of her results, for many years her University of Melbourne rise to an abundance of offerings for research- cartographic methods were somewhat crude. in Australia, where she NATURE.COM ers and enthusiasts. These tools are more “My old maps were [made] using MS Paint was working on her PhD. For more on scientific approachable for novices than the conventional on top of some copyrighted map that I really There she discovered the software, apps and geographic information systems (GISs) that shouldn’t have been using,” she says. Her free, open-source pro- online tools, visit: geographers have long used for analysis of geo- next solution wasn’t much better: “My mum gram TileMill, created nature.com/toolbox spatial data sets. They allow non-specialists

to easily visualize, manipulate and share their basemap in TileMill and data layers in CartoDB, data in formats that are as slickly browsable as then wrap them in an online interface using Google Maps but with greater power and flex- MORE ONLINE Leaflet, a mobile-friendly visualization package ibility. “TileMill allows you to be a complete that runs in the program JavaScript and meshes control freak,” says Gawne, now at Nanyang For more with other JavaScript visualization packages Technological University in Singapore. From on scientific such as D3. Duncan Smith, a geographer at line styles to font spacing and kerning, “I can mapping, University College London, has made one such really manipulate all the variables quite easily.” including a combination: an online map of UK census data Until recently, Google, which is based in gallery and called LuminoCity that uses Leaflet to display Mountain View, California, itself had staked links to tools, the map data over basemaps produced in Tile- see go.nature. the biggest claim in this space, providing various Mill, and a variant of D3 called Dimple to show DUNCAN A. SMITH, CASA UCL ways to access and decorate its maps through com/u76knj graphs of the data onscreen. application programming interfaces (APIs). But as demand grew, the tech giant began lim- STORAGE HUBS iting public access to its APIs in 2011 — and of global basemaps — ranging from familiar Researchers can also store their data sets in a this allowed slightly more sophisticated open- geopolitical and satellite-image formats to CartoDB account, then access them (using the source tools to flourish, says Oliver O’Brien, more stylish black-and-white and even pencil- ubiquitous SQL database language) for other a geographer at University College London. and watercolour-themed renditions. Some are online applications, notes Desmet. For one Today, a fully fledged ecosystem of start-ups produced by TileMill’s maker Mapbox, which project, he used D3 to build a map depicting with open-source technology at their core offer boasts a growing list of corporate and media radar observations of bird migration as wind- platforms that many say have surpassed Goog- clients — in many cases supplanting Google in like flowing curves. The source code is stored le’s offerings. a growing ‘battle of the basemaps’. in the repository GitHub, but the map pulls the “Google really nailed down having maps Mapbox first released TileMill in 2011. The scientific data from his CartoDB account. on the web,” says Javier de la Torre, a founder team took a powerful but complex open-source Despite the visual sophistication of these and current chief executive of one of Google’s cartographic renderer called Mapnik, built an tools, the level of computational analysis they emerging rivals, CartoDB of New York City. easy-to-use interface around it and created a provide is limited. But after using these pro- “What I think they didn’t see coming was that simple styling language, CartoCSS, to custom- grams to get to grips with the basic principles, there was going to be this explosion of new map- ize the maps’ appearance. researchers can progress to more-powerful makers.” “TileMill was a game-changer, absolutely,” GIS platforms. Many scientists — including says Bennett. It allowed non-experts to produce those involved in public policy, such as urban THE NEW MAPPING LANDSCAPE professional-looking maps — either for publica- planning and crisis mapping — use arcGIS, a In 2011, de la Torre was part of a team research- tion as static figures or for use as basemaps in suite of products maintained by Esri, based in ing biodiversity informatics. The group was other visualization tools — without the need for Redlands, California. But there is also an open- seeking an online platform to make a map of more-complicated GIS programs. source alternative: QGIS, a project of the Open all known species on the planet. “There wasn’t The landscape continues to shift rapidly. Source Geospatial Foundation. technology for doing that,” he says — no tool In January, Google announced that it would Researchers who already write code as part could handle the amount of data, nor visualize shut down some premium and paid forms of of their work can use programming languages how they changed over time. Google Maps and focus on its basic Maps API. such as Python and R, which already have The researchers decided to develop the tool In response, CartoDB introduced tools to help capable mapping packages that users may not themselves and created what became the open- users migrate their data to CartoDB, while still even be aware of, points out astronomer James source platform CartoDB. The company offers allowing them to integrate the Google Maps Davenport of the University of Washington in free and paid plans for hosting and visualizing APIs. Mapbox, for its part, has shifted develop- Seattle. He says that astronomers often “end up data through its website. Unlike TileMill, which ment from TileMill to its intended replacement, bastardizing scientific visualization software is primarily intended for drawing and design- Mapbox Studio. to make maps”. He now uses the Python pack- ing static maps, CartoDB specializes in visual- Cost of data storage is a potential stum- age matplotlib in tandem with the rest of his izing dynamic layers of data on top of basemaps. bling block for scientists with large data Python-based analysis to project his infrared Users can import their geo-located data into sets — although CartoDB is open source, its observations onto maps of the sky. CartoDB’s web-based interface and then filter convenience comes in large part from using it Even researchers who would rather not touch or cluster data points, change the colour or size on the company’s hosted web service. The firm a line of code can accomplish a lot with the help of symbols, and animate data changes over time. offers 75 megabytes of storage for free, but to of CartoDB and TileMill. “You don’t have to “CartoDB wants to be a place where your data store more than 1 gigabyte of data, the price rises be particularly technically competent,” says lives,” says Steve Bennett, a research-oriented quickly to hundreds of US dollars per month. Gawne, who produced the Tibetan-language technologist at the University of Melbourne CartoDB also charges to keep data and maps maps for her thesis in TileMill and now teaches who takes workshops on mapping, including private on the site. “We’ve had real problems,” mapping workshops herself. “You have to be not the one that Gawne attended. says Bennett. “If you’re a PhD student with no afraid to try it.” ■ Peter Desmet, who collaborates with a bird- funding, it just doesn’t work.” Mapbox works tracking research team at the Research Insti- with a similar pricing model for hosting maps Mark Zastrow is a science writer in Seoul. He tute for Nature and Forest in Brussels, was a on its servers, although TileMill itself is a free, reported this article from Washington DC. colleague of de la Torre and became an early downloadable program. However, CartoDB adopter of CartoDB. “I was never a desktop does work with academic users to try to find GIS person,” he says. But in CartoDB, “you can a solution, says de la Torre, and awards grants CLARIFICATION create and share a visualization in literally min- of up to US$3,500 to researchers studying the The Toolbox story ‘Adventures with R’ (Nature utes”. Being able to simply send a link to the map impacts of climate change, in recognition of the 517, 109–110; 2015) did not make clear online also makes it much faster to point out company’s environmental roots. what stopped Rabih Murr from practising R data-quality issues to colleagues, he says. Power users can daisy-chain these tools — he was preparing a paper for publication. Another strength of CartoDB is its selection together: for example, one could create a