At the Edge of the Map

Alison Balmat
May 01, 2000
graphic depicting colorful global map

Alan MacEachren smiles at the question, leans back in his chair. His eyes focus for a moment on a map of the Netherlands, created by remote sensing, that is tacked to his corkboard. In a thoughtful and unhurried motion, his fingers graze over his long wiry beard. An animated map of climate change flashes across the screen of his laptop, which sits in the midst of a desk piled high with books.

"What is a map?" he repeats slowly. "Well, that's the very first question I ask my students in the introductory geography course I teach." He chuckles softly, a gleam in his eye. "So if I told you the answer, they'd have it too easy."

After a pause, he chooses to elaborate a bit anyway. A map, he says, is both an abstraction and a representation.

"A portrait is an abstraction of a person," he explains. "If it is drawn well, the viewer can sense the character of the person in the portrait. The artist can emphasize certain attributes while eliminating others—it's all in how the artist wishes to portray the person. And so it is with maps.

"Maps allow us to see a world that is too large and complex to be seen directly," MacEachren continues. "We are not revealing knowledge as much as creating it."

MacEachren drew his first map as a second grade project—a map of the roads, houses, and trees between his home and his school. As a child, he collected road-maps "from every place I ever visited." During his junior year of college, a course on South Asian regional geography and another on map use led MacEachren to become a geography major and eventually to obtain two more geography degrees. His first research project involved studying how people understand concepts of distance and how it influences where they shop. He found that the perception of distance between locations is more closely related to travel time than to actual mileage.

"My research throughout my career has continued along this same track, having to do with mapping and people's cognition of space," he says. "It goes beyond just geography and cartography—the science of making maps. It draws together information from computer science, cognitive psychology, linguistics, semiotics, engineering . . ." MacEachren smiles, as if the list could easily continue.

A faculty member in the geography department since 1985, MacEachren received the 1998 Wilson Research Award from Penn State's College of Earth and Mineral Sciences for his book How Maps Work: Representation, Visualization, and Design.

Imagine. With the click of a button, you can watch a cold front move across Russia. You can slow it down or speed it up, manipulating the time frame over which this climatic phenomenon occurs. You can observe cloud cover moving across the globe or look at the temperatures of water and land at different points during the day.

For a group of schoolchildren, this scenario may soon be a reality. As part of a research project led by Penn State psychologist Lynn Liben, MacEachren has designed visual tools to teach spatial concepts in the earth sciences. The project, "Visualizing Earth," lets children learn about spatial-temporal relationships in an interactive way. "We are interested in how students differ in their development of spatial, logical, and symbolic thinking," MacEachren explains.

airplane navigation map

The computer-animated maps allow students to isolate single variables of globally scaled weather phenomena. They contrast a given day's temperature and the amount of water vapor, for example, or view the cyclical nature of climate changes. They can pinpoint specific moments in time: the temperature patterns at midnight in Seattle or the springtime climate of Boston.

"Visualizing Earth" was created to experiment with visual tools and to address questions of spatial cognition—what underlies a child's ability to use and understand images, Liben explains. "If red spots on the map symbolize hot temperatures, can the child understand that red means hot?" Liben asks. "If one object takes up more room on a map than another object, can a child perceive the difference in scale?

"I'm looking at the spatial skills children need to understand maps," Liben adds, "and then seeing how we can create maps that are easier for them to use."

In the mid-1970s, health officials examining maps of mortality data noticed an unusual number of women were dying from oral cancer in southeastern United States. Because most of the women worked in the textile industry, the officials hypothesized that fiber from the fabrics was causing their disease. But after obtaining the women's personal histories, they discovered their deaths were not work-related; rather, many of the women in this area of North Carolina were taking snuff. After these surprising results, the National Center for Health Statistics decided to use maps more often to assess health patterns.

The NCHS supplied health statistics to MacEachren and associate professor of geography Cynthia Brewer. Their task was to find the most effective color schemes to portray a particular map's data, as well as a way to depict data that was less reliable, for instance, because of a small sample size. They tried eight color schemes, ranging from one that used completely different hues at each end of the spectrum with a dull median color, to another that used dark and light shades and tones of the same color. They produced maps following each scheme and asked volunteers to answer questions about them.

Next, they combined their successful color schemes with various methods to represent data reliability. After measuring the accuracy of the volunteers' responses (and therefore how well each color scheme communicated the information), they found that the most effective scheme combined color hue and lightness for mortality and used a hatching pattern to overlay information about the data's reliability.

Brewer and MacEachren's maps appeared in the 1996 Atlas of United States Mortality—but MacEachren was only getting started. He took the color scheme work and translated it into an interactive animated version.

color-coded U.S. map indicating incidence of heart disease in white males

One map, using data from the Centers for Disease Control, tracks heart disease in U.S. white males from 1979 to 1993. "Because it is an interactive map, you can focus in on a subset of high mortality values, then go through the time series and watch patterns of heart disease in this country," MacEachren says. "You can see an evolution unfold before your eyes."

Deaths begin along the eastern seaboard, the Northern Appalachians, and some other scattered areas. "The interactive animated map shows the change in where the men are dying over time; the pattern is sliding to the south and west." He changes a setting on the computer, and a new visual pattern is suddenly evident. "When you adjust the pace of animation, a pattern pops out at you," he says.

"Speed it up and watch the progression."

Add in another factor: per capita income. Click a few buttons and a new animated map comes on screen. The colors clearly show that some men in the top 20-percent income bracket have a high rate of death from heart disease, yet they are not sliding across the country, as are the lower-income males.

"The rich guys are around the periphery of the eastern zone," MacEachren says, "and they stay there over time. The blue—the lower-income males—are the main core and are moving south and west. We can suggest a reason for these relationships among income, smoking, jobs, and death due to heart disease," MacEachren says. "The theory must be tested, but it is visual tools like this one that let health workers generate a hypothesis."

MacEachren clicks a button on his laptop and the computer screen comes to life. A colorful animated map flashes across the screen: wispy cloud formations swirl up the eastern coastline of the United States. He allows the storm system to run its course several times, silently observing the seemingly aimless movements of the clouds. Clicking the STOP button, MacEachren manipulates the time frame, then sets the storm in motion again, at a quicker pace this time. Suddenly patterns emerge from the whirling cloud masses, and MacEachren smiles, satisfied. He has proved once again that changing the pace of animation reveals details about the data that would be difficult to detect otherwise.

The idea for this animated map, he says, stemmed from a February 1995 snowstorm that moved up the eastern coastline from North Carolina, dumping large amounts of snow on some areas and completely bypassing others. MacEachren tracked the storm with the help of climatologists and then mapped out the various predictions made by computer models forecasting the storm's possible course.

The three models he charted disagreed markedly as to where the storm would be at a particular time, so MacEachren used various colors to denote the inconsistencies. Blue represented agreement among the models and bright red splotches indicated discrepancies. "The system allowed for surface pressure patterns and storm centers to be examined as they corresponded or diverged over time," MacEachren says. It became readily apparent that the three weather forecasters agreed on where the storm was headed, yet disagreed both on its time of arrival and on the location of the eye of the storm. These differences led to disagreements in how much snow was predicted. "Simultaneously animating each of the storm tracks showed discrepancies in time estimates," MacEachren explains. "As the storm moved across the Appalachians, the uncertainty was not in its predicted path, but rather in the rate at which the storm would move along that path.

"This map illustrates that there are both spatial and temporal components to whether models agree or not," MacEachren says. "These models did much better on spatial predictions and less well on the timing of events."

Earth System Visualizer

A cyclone--whirling across the North Atlantic--is captured in the Earth System Visualizer, developed as a science education tool for the “Visualizing Earth” Project. Users can explore changes in three layers of information, cloud cover, land temperature, and water temperature, over a period of time that they determine.

Because different types of models are best at predicting different aspects of the weather, predictions are typically averaged to produce a forecast. However, MacEachren's map demonstrates that animated visual tools could be very useful to climatologists and forecasters.

Exploring data to gain insight and understanding, MacEachren says, is the essence of scientific visualization. "It's a research area that has applications in, for example, environmental analysis, medicine, aeronautical design, and mineral exploration. And it's developing worldwide."

MacEachren chairs the International Cartographic Association's Commission on Visualization and Virtual Environments. The commission emphasizes applying dynamic, interactive maps to science, education, and policy analysis. "It links researchers around the world," MacEachren explains, "allowing them to share ideas and collaborate on new developments in the visualization of geospatial data." Other commissions within the ICA deal with topics such as cartographic education and international mapping standards, MacEachren says. They train individuals to use mapping technologies, including the new visualization tools developed by MacEachren's commission, to further economic development and to manage the environment.

For example, visualization techniques can help policy makers foresee the effects of development. "If a high-speed railroad ran through the middle of this area in the Netherlands," MacEachren asks, pointing to the map on the corkboard in his office, "what would the impact be?" One way to see what might occur is through virtual reality simulators like those developed by MacEachren's commission co-chair, Menno-Jan Kraak, and his colleagues in the Netherlands. "You are hooked up to head-tracking equipment, wearing 3-D goggles, and viewing a scene in front of you," MacEachren explains. "As you move, your perspective changes just as it would in the real world."

MacEachren's efforts internationally were reflected locally with his creation of the Penn State Geographic Visualization (GeoVISTA) Center in October 1998. Collaborating with Penn State colleague Donna Peuquet and with the University's Center for Academic Computing, MacEachren is working with virtual reality equipment, creating 3-D databases that incorporate time into their mapping schemes.

"Ultimately, we want to create virtual reality simulators that are on your desktop computer," he says. "Any number of people around the globe will be able to communicate, while observing and manipulating spatial and temporal processes."

In the modern state of Oaxaca, Mexico, lies the Apoala Valley. Before the Spanish conquest, the Mixtec people living in the valley produced manuscripts depicting both the landscapes and lineages of their communities over time. These "cartographic histories" combine a location (the Apoala valley) with a point in time (pre-Hispanic influence).

computer screen with various types of data plotted

Climatological data from the Gulf of Mexico can be mined and represented in several ways--as a coordinate plot, a scatter plots, and as a map.

MacEachren and Peuquet adopted the name "Apoala" for a project they began together three years ago combining components of space, time, and visualization. "The name embodied our project," MacEachren says. "Not only does the Apoala art show both space and time in the same representation, but it shows that people have been combining the two concepts for centuries."

MacEachren and Peuquet are developing and linking database and visualization tools to show a time series of 100 years of climate change in the Susquehanna River Basin. Peuquet's specialty is geographic information systems—computer software that stores, manipulates, analyzes, and produces graphical output for earth-related data. "Geographic databases are currently static, showing only a snapshot in time," Peuquet explains. "This is obviously very constraining since the world, whether you are studying urban systems or biological systems, is a very dynamic place."

"Factoring the progression of time into the equation will be vital in seeing patterns," MacEachren adds. "It's just a natural combination—GIS, space-time, and visualization."

To search for patterns in the large data set, Peuquet and MacEachren are developing flexible methods to ask the database a question—they enter a query and the computer finds corresponding patterns. "'Show me all the data for time periods in which the Susquehanna River Basin had unusual amounts of precipitation.' We want to ask the database questions like this," MacEachren explains, "questions that incorporate what, where, and when."

MacEachren and Peuquet extract that information, feed it into visualization tools, run it as an animation, and watch the changes in precipitation patterns. "We relate the precipitation patterns to other variables too," MacEachren explains. "For example, was it warmer or cooler when this precipitation occurred? What weather systems passed through the area that caused the event?"

MacEachren and Peuquet are also using a concept called data mining to find interesting patterns in the data without posing a specific query. They then use the same visualization methods to understand the pattern that the system outputs

The human aspect will be factored in as the project progresses. "We'll add more layers to our tools concerning the people—what are their incomes, do they live in a flood plain, where are their water sources—to discover which people might be affected by a flood, for example. Their type of water source, their ability to upgrade their water system to deal with the flood, their capacity to pay for insurance—it all connects."

But insight about climate patterns is not MacEachren's goal. Insights about the design of computer tools and how scientists might better use them is. "We're building systems that allow experts to analyze patterns and make hypotheses about them," MacEachren says. "Essentially, we are creating new ways for people to understand and visualize their world."

Alan MacEachren, Ph.D., is professor of geography and director of the GeoVISTA Center in the College of Earth and Mineral Sciences, 310 Walker Bldg., University Park, PA 16802; 814-865-7491; He is a Faculty Fellow with the Center for Academic Computing. His work is supported by the U.S. Environmental Protection Agency, the National Institutes of Health, and the National Science Foundation. The "HealthVis" project is sponsored by the U.S. Centers for Disease Control and the National Center for Health Statistics. Lynn Liben, Ph.D., is distinguished professor of psychology in the College of the Liberal Arts, 450 Moore Bldg.; 814-863-1718; Cynthia Brewer, Ph.D., is associate professor of geography, 325 Walker Bldg.; 814-865-5072; cab38@psu. edu. Donna Peuquet, Ph.D., is professor of geography, 305 Walker Bldg.; 814-863-0390; Graduate students working on these projects include: Mark Harrower, Amy Griffin, Dan Haug, and Robert Edsall. Writer Alison Balmat is majoring in English and psychology at Penn State. Maps courtesy of Alan MacEachren and the GeoVISTA staff.

Last Updated May 01, 2000