How to Tell Science Stories with Maps

August 25, 2015   Greg Miller

This map, part of Audubon’s Birds and Climate Change report, depicts predicted changes in tree swallows’ summer (yellow) and winter (blue) ranges.

Maps are amazing for their ability to show us something we can’t see directly, from the path of the Curiosity rover on Mars, to the tangle of underground fracking wells in North Dakota, to clusters of unvaccinated schoolchildren in California. For journalists, maps can be both a powerful data-visualization tool and a reporting tool.

“Maps are some of the most information-dense ways of communicating data,” says Len De Groot, director of data visualization at the Los Angeles Times. People understand maps intuitively because they use them in their everyday lives, De Groot says. “You can do a lot in a map because people already understand the fundamentals—unlike, say, a scatterplot.”

Maps can also reveal relationships and stories that aren’t otherwise apparent. In the mid-2000s, De Groot was part of a team at the South Florida Sun-Sentinel that mapped FEMA disbursements after several hurricanes, including Hurricane Frances, which struck in 2004. “We didn’t start with any agenda, we were just doing the standard where’s-the-money-going thing,” he says. “To our surprise there was one zip code in Miami where we saw there was a spike in payouts in areas where we knew there was very little damage.” That led to a broader investigation by the paper, which revealed widespread fraud and got the paper nominated for a Pulitzer Prize and ultimately led to policy changes at FEMA.

Fortunately for cartography-curious journalists, making maps is easier than ever, thanks to a proliferation of (mostly) free and open-source software. Anyone with a computer can put together a nice-looking map in less than an hour. Making really good maps isn’t quite that easy, however. It requires more thought and a willingness to learn some technical skills. Here are some considerations to keep in mind, and some tips on how to get started.

Whether to Map

Just because you could make a map from a given dataset doesn’t necessarily mean you should. The key question is whether location is the most important thing about the story you’re trying to tell, says Virginia Mason, a senior graphics editor at National Geographic. For example, one upcoming story involves a Department of Agriculture program that culls wolves to protect farm animals. The graphics team had data on how many animals have been killed in each state, and their first thought was to make a map, shaded to indicate the number of animals killed. But they soon thought better of it. “Really the significance isn’t where the states are, it’s what the states are,” Mason says. So they decided to instead create a chart that makes it easy to see at a glance which states have killed the most wolves.

If the same data were displayed on a map, readers would have to scan the map and make comparisons between the colors or shading of different states. It’s something our brains aren’t especially good at, says Peter Aldhous, a self-taught mapmaker and science reporter for BuzzFeed News. There’s been a lot of perceptual science research on what visual features our brains are best at discriminating, Aldhous says. It’s found, for example, that we are much better at comparing the length of lines than we are at comparing area, color, or saturation. So, if comparing numbers is the point, a bar chart is a solid choice. (For a good introduction to using perceptual science to make better infographics, check out Alberto Cairo’s book, The Functional Art).

Mason and her colleagues at National Geographic did make a map, or rather a series of them, for another conservation story, a 2013 feature on African lions. The maps show the historical range of lions, which covers large swaths of Africa, and the scattered pockets where they still exist today. In that case, she says, place matters—the lions now survive almost entirely in wildlife parks and other protected areas. Also, most readers aren’t as familiar with African geography as they are with the 50 states. “If you did a chart of lion populations by wildlife reserve, people aren’t going to know where those reserves are,” Mason says.

This map, published in National Geographic in 2014, shows the historical range of lions in Africa and the scattered pockets where they still exist today.

What Is Your Point?

Once you’ve decided that a map is in fact what you want, the next question is what point you want your map to get across. Then, spend some time thinking about how to design the map so that every part of it emphasizes that point, and anything that doesn’t goes away, says Anthony Robinson, a geography professor who teaches cartography at Penn State. (Robinson also developed a mapping MOOCthat’s been taken by more than 100,000 people—it’s a great introduction, but the next installment hasn’t been scheduled yet).

“One common thing I see is satellite or aerial imagery on something like election maps,” Robinson says. If the point of the map is who voted for whom, the land cover is irrelevant. A neutral white or gray background is usually preferable.

Too much clutter is another pitfall, Robinson says. Mapping software puts every little road, stream, and populated place at your disposal. Resist the urge to include them all, even if you have to rebel against the default settings of the software or remove features by hand using Adobe Illustrator or asimilar program.

You can see a nice example of a strong visual hierarchy in the map below, from The New York Times.

This New York Times map documenting the movement of refugees mutes details of oceans and landmasses to minimize clutter.

The red lines show the movement of refugees around the world. The white oceans and gray landmasses are just visible enough to provide a reference. “We used a satellite image, but we took out all the color so the red arrows really popped,” says Tim Wallace, a cartographer and graphics editor at theTimes. This map also illustrates how making a map can reveal new stories—in contrast to recent media attention to refugees’ desperate attempts to cross the Mediterranean and Andaman Seas, the map shows a surprising amount of displacement within Africa.

Pick the Right Projection

In 2003 The Economist published a map to accompany an article on the threat of missile attacks from North Korea. They used a standard Mercator projection and concentric circles to indicate the regions potentially in harm’s way. Wrong! The Mercator projection grossly distorts distances near the poles, so the real reach of any missiles from Pyongyang would be far greater. The magazine published a correction, along with a revised—and much scarier—map a short time later (see both versionshere).

Picking the wrong projection is a rookie mistake. The problem is simple: The Earth is round, and maps are flat. Projections are the mathematical formulas that force a round planet onto a flat map. Each projection distorts some combination of area, shape, direction, and distance. Each has its strengths and weaknesses. Mercator, devised for 16th-century nautical charts, is very good at preserving direction. The azimuthal equidistant projection, which preserves distances relative to a central point, might have been a good choice for the North Korea map. If preserving area is important, say for mapping the impacts of climate change on bird habitats, the Albers equal-area conic projection is a good choice.

Normalize Your Choropleth

When I first met Anthony Robinson, at a mapping conference in 2013, he was wearing a little pin that said: “Normalize your choropleth.” I had no idea what it meant, but it turns out to be extremely important advice.

Choropleth maps are everywhere. They use colors or shading to map a variable—anything from election results to crop yields to cancer rates. But if you’re not careful, it’s easy to end up with a misleading or meaningless map. Take cancer. If you map the number of cancer cases in different areas across the U.S., you will likely see big dark splotches along the Eastern Seaboard and Southern California, with a few small blobs across the Midwest. Because that’s where the people are. You might as well have made a population map. What’s far more interesting is where the cancer rate—the number of cases per 100,000 inhabitants—is higher or lower than normal. Normalizing (by population, in this case) allows readers to make meaningful comparisons. In other cases, you might want to normalize by area—for example, a map of air-pollution sources might be normalized to show the number of factories or automobiles per square mile.

Color Matters

Nothing makes professional cartographers roll their eyes in dismay like a rainbow color scheme, the kind you often see on weather maps. “It’s like this trend of putting bacon on all kinds of food,” says Robinson. “It’s just gratuitous.” A rainbow palette may look pretty at first glance butit’s rarely a good choice for maps. It’s especially bad for mapping a continuous variable like temperature. If red is warmer than blue, why is violet (a mix of red and blue) colder than blue? And why is it that a two-degree difference in temperature can correspond to either a subtle change from light green to dark green or a dramatic jump from green to yellow? A more intuitive color scheme would use darker shades of a single color to represent higher values.

Robinson’s Penn State colleague Cindy Brewer has spent much of her career applying perceptual research on color to mapmaking. Her website, Color Brewer, is a great resource for beginners. It forces you to think about the data you’re trying to map. For divergent data—areas of above- and below-average rainfall, for example—Color Brewer might suggest a color scheme that transitions from brown to green, with neutral white in the middle. For mapping distinct categories, such as the dominant invasive species in different areas, the site will suggest a combination of unrelated colors that are easy to tell apart. It also limits you to 12 categories—any more than that and readers will have a hard time telling the colors apart. Color Brewer also lets you click boxes to limit yourself to color schemes that can be read by color-blind people, or that look good on paper as well as on screen.

For a recent anniversary story on Hurricane KatrinaBuzzFeed News science reporter Peter Aldhous took storm surge data from a NOAA computer model (left) and tried to “make it look less like an acid trip in a paint factory.” His more subdued but more intuitively readable version is on the right.

Getting Started

Just a few years ago, getting started with digital maps meant diving into the deep end with the kind of complicated geographic-information-system (GIS) software typically used by natural-resource managers, city planners, and other experts. It’s much easier these days. One good option isCartoDB, an online mapmaking tool that allows you to upload datasets and make basic interactive web maps using a point-and-click interface. To make more sophisticated maps, or to customize your maps so they look exactly as you want, you’ll need to learn a bit of coding. You won’t become an expert overnight, but you can quickly pick up enough to be useful, especially if you take advantage of CartoDB’s online courses, which were designed with journalists in mind. Pricing varies according to how much data you’re allowed to upload—a free version allows up to 50MB, which is plenty to play around with.

For some of my very first efforts at making maps, I used an even easier option: ArcGIS online, from Esri, the Microsoft of GIS mapping software. Journalists can request free access through Esri’smedia relations team. The point-and-click interface includes easy access to census data and other public databases, and includes several templates for incorporating photos and text into “story maps,” like this one I made to go with a story on a wildlife die-off in Florida. The free version offers less customization and less data analysis than CartoDB, but it’s a good option for the code averse. Investigative Reporters and Editors recently posted a helpful gallery of video introductions to several mapping and data visualization tools.

CartoDB, Esri, and other makers of mapping software often offer workshops at journalism conferences. As do the journalists who use them. Aldhous and De Groot have taught a mapping workshop at the Knight Digital Media Center at the UC Berkeley Graduate School of Journalism. Aldhous has also taught mapping to journalists at the annual meeting of IRE, an organization he credits with fostering his own interest in mapping and in data journalism more broadly. “They talk a lot about a data frame of mind, where you’re not just thinking about who can I interview for a story, you’re thinking about how can I use data to help me find a story or bring context to a story,” he says. “The maps are really part of that.”

Greg Miller

Guest contributor Greg Miller is a science journalist based in Portland, Oregon. Previously, he was a senior writer at Wired, where he co-founded the cartography blog Map Lab. Follow him on Twitter @dosmonos.

Image credits:

Tree swallow map: National Audubon Society / Stamen Design.

Lion map: Virginia W. Mason / National Geographic.

Refugee map: Sergio Peçanha and Tim Wallace / The New York Times.

xkcd map projections: xkcd (CC BY-NC 2.5).

NOAA and BuzzFeed maps: Peter Aldhous / BuzzFeed News and NOAA.

Photo of Greg Miller: courtesy of Greg Miller.

How to Tell Science Stories with Maps的更多相关文章

  1. USER STORIES AND USE CASES - DON’T USE BOTH

    We’re in Orlando for a working session as part of the Core Team building BABOK V3 and over dinner th ...

  2. Thoughts on an Article from Science 'A network framework of cultural history'

    Apparently, this is an excellent interdisciplinary study. This paper drawn on a simple but large dat ...

  3. 【Repost】A Practical Intro to Data Science

    Are you a interested in taking a course with us? Learn about our programs or contact us at hello@zip ...

  4. Competing in a data science contest without reading the data

    Competing in a data science contest without reading the data Machine learning competitions have beco ...

  5. 15 Most Read Data Science Articles in 2015. So far …

    15 Most Read Data Science Articles in 2015. So far … We've compiled the latest set of "most rea ...

  6. 11 Facts about Data Science that you must know

    11 Facts about Data Science that you must know Statistics, Machine Learning, Data Science, or Analyt ...

  7. How To Build Compelling Stories From Your Data Sets

    How To Build Compelling Stories From Your Data Sets Every number has a story. As a data scientist, y ...

  8. 【转】The most comprehensive Data Science learning plan for 2017

    I joined Analytics Vidhya as an intern last summer. I had no clue what was in store for me. I had be ...

  9. 【转】Go maps in action

    原文: https://blog.golang.org/go-maps-in-action ------------------------------------------------------ ...

随机推荐

  1. 404 Note Found团队-项目UML设计

    目录 团队信息 分工选择 课上分工 课下分工 ToDolist alpha版本要做的事情 燃尽图 UML 用例图 状态图 活动图 类图 部署图 实例图 对象图 时序图 包图 通信图 贡献分评定 课上贡 ...

  2. lr常见问题汇总(持续更新版)

    在使用mms协议的时候,会碰到报错 Action.c(): 错误: C 解释器运行时错误: Action.c (): Error -- Unresolved symbol : mms_play. 解决 ...

  3. Genymotion-ARM-Translation_v1.1安装报“an error occured while deploying the file”

    如上图,在将Genymotion-ARM-Translation_v1.1.zip拖动Genymotion虚拟机中报了图中错误,在百度后找到了解决办法,下面是我的操作步骤 1.下载re管理器之类的ap ...

  4. [转帖]Ubuntu 安装VNC的一个方法

    来源: https://blog.csdn.net/CSDN_duomaomao/article/details/75270271 Ubuntu 16.04 LTS 安装VNC,在百度和谷歌找了很多教 ...

  5. php连接池 php–cp

    原文地址:http://blog.sina.com.cn/s/blog_9eaa0f400102v9fd.html 数据库连接池php-cp介绍时间 2015-01-23 11:53:05 数据库连接 ...

  6. HDU4622_Reincarnation

    题目给出一个长为2000的字符串,和10000询问,每次询问从第l到第r个字符中间有多少个不同的子串. 其实,全部预处理.f[i][j]表示从i到j个字符的子串数.重构2000遍SAM. 对于新加入的 ...

  7. DAY7-Flask项目

    1.cookie用于用户登录: Flask用于登录的插件:flask-login 插件初始化: 使用插件:  2.访问权限: 访问某个页面要登录之后才能访问: 为插件编写函数,写在user模块中:

  8. BZOJ3244 NOI2013树的计数(概率期望)

    容易发现的一点是如果确定了每一层有哪些点,树的形态就确定了.问题变为划分bfs序. 考虑怎样划分是合法的.同一层的点在bfs序中出现顺序与dfs序中相同.对于dfs序中相邻两点依次设为x和y,y至多在 ...

  9. elasticsearch使用More like this实现基于内容的推荐

    基于内容的推荐通常是给定一篇文档信息,然后给用户推荐与该文档相识的文档.Lucene的api中有实现查询文章相似度的接口,叫MoreLikeThis.Elasticsearch封装了该接口,通过Ela ...

  10. 点击--》java9 新特性 详解

    引言: 点击-->java9 新特性 详解 点击-->java8 新特性 详解 正题: 1.局部变量var 将前端思想var关键字引入java后段,自动检测所属于类型,一种情况除外,不能为 ...