"Modeling our world," geography returns to Harvard.
Take your geographic information system (GIS) for a spin around the block. Its easy. Sit at your computer, which you have loaded with GIS software, and call up on the monitor a street map of Greater Boston. Superimpose on it a second map showing the household incomes of the citizenry. Those with the best dental work are in the suburbs, struggling scholars in Cambridge, the poor in the inner city. Add a third layer of census ethnicity information. All these data are public and readily available. Top it off with lines and dots showing Massachusetts Bay Transportation Authority bus routes and subway stations. Almost everyone is near a stop except in one big chunk of the city south of downtown. Your multilayered map will reveal to you what you might not have otherwise perceived, that the poorest, blackest Bostonians, in Roxbury and Dorchester, the ones least able to afford cars, are the least well served by public transportation.
Thats an easy demonstration of GIS, says Peter K. Bol, Carswell professor of East Asian languages and civilizations. Another is to look at high-voltage electricity lines and higher cancer rates and see that they dont go together. A college student could easily learn to do this in a course on GIS and figure out the answers to all sorts of questions, which may have to do with the distribution of populations in 1102 in China, something Im working on at the moment, or very contemporary problems in the world we live in. If you wonder what happens when the polar ice caps melt and ocean levels rise, GIS can show you within minutes just how much of the coast will be inundated if you raise the water level by two feet.
After we see the new coastline, we can look at population, Bol continues. How many million people are going to have to move? How much infrastructure is going to disappear? Enabling people to use GIS, to have it as part of their intellectual toolkit for understanding the world around them, is a very important thing.
|Wendy Guan and Peter Bol|
|Photograph by Jim Harrison|
That is one of the goals of the Universitys new Center for Geographic Analysis, established with the support of the provost earlier this year and launched ceremoniously at a gathering on May 5. The occasion was heralded as a consequential reversal, the return of geography to a University that in 1948 had shocked academe by dumping its geography program (see "Geographers See Death, Birth, and Job Prospects"). President James B. Conant, a chemist, said then that the subject had no place within a university curriculum. That is the past, said President Lawrence H. Summers on May 5. Geography is a very different field today, and it is increasingly at the center of a very wide range of intellectual concerns.
The new center is housed at the Institute for Quantitative Social Science in the north building of the spiffy terracotta-and-glass quarters of the Center for Government and International Studies on Cambridge Street (go to www.gis.harvard.edu/icb/icb.do). Its affairs are overseen by an eight-person faculty steering committee composed of GIS pioneers in the Faculty of Arts and Sciences and the Division of Engineering and Applied Sciences, the Graduate School of Design, and the Harvard School of Public Health. One of them, the first director of this hotspot of high-tech geospatial analysis, is Bol, an historian who has spent most of his intellectual time in medieval China.
Things exist in time and space, Bol says. Historians have been very good on temporal change. We havent been good on spatial variation.
Think of a chronology as the basic tool of historians. If you were to do a chronology of the past 20 years in the United States, youd probably focus on presidential leadership and the Congress, longer-term economic trends, and things like that. But if you were to cut across the picture at any one moment, you would think of government as quite dispersed; its not only what happens in Washington, its the state governments and local governments. Economic activity is stronger in certain areas than in others. Incomes are distributed unequally by regions and within regions and among groups of people. Pollution, climate, water resourcesall these things vary spatially.
So does the inside of the brain or the inside of the earth. An estimated 80 percent of all data has a spatial component, says Bol, and it is a component that has been largely neglected. Spatial analysis is a way of thinkingabout how things are distributed and about the relationships among things.
A map is the basic tool of a geographer, and a map and a chronology have a peculiar similarity: both are false. A chronology picks out whats important in terms of whats going to happen later, and misrepresents actual things by picking out certain things as worthy of attention. Its a scaled representation where the scale is always false, says Bol. The map is also a representation, of the spatial arrangement of things, and any map, to be useful, also is false. On a gas-station road map, the roads are not represented to scale. If they were, you wouldnt be able to see them.
One problem with that gas-station map is that you have to take it on faith. You dont know where the data that went into it came from or how accurate they are. A geographic information system lets you disaggregate all the kinds of data you might have on a map, ask what the precise sources of that information are, and then choose which data to represent. Making and using a map require judgment and interpretation.
I happened to have some very large data sets for medieval China of persons who passed the civil-service examinations, entered government, and held higher office, says Bol about his own use of GIS in research. I can show on a GIS map that from the eleventh to the thirteenth century, the geographic distribution of high officials narrows. In the eleventh century, most areas have some representation in government. But by the thirteenth century, there are only three places producing most of the officials, and government is dominatedjust as it is about to fall to the Mongolsby three regional elites. That is not something I knew before I made the map.
Theres more to geotech than GISspecifically, there are data gathered by remote sensing from satellites and aircraft and by global positioning systems (GPS), the devices some of us have in our automobiles to tell us exactly how lost we are. A team of researchers at the School of Public Health has used wearable GPS units attached to members of the public to measure to what extent community trails and paths contribute to the promotion and maintenance of physical activity. Professor of biology Brian Farrell, who heads a team from the Museum of Comparative Zoology that is mapping the insect biodiversity of Hispaniola, has his field researchers keep GPS units in their backpacks so that they can determine and record precisely where in the cloud forest they met a particular beetle (see Brian Farrell in Bugdom, September-October 2003, page 66). Paul Cote, geographic data wrangler at the design school, has offered a course there in remote sensing. He has also demonstrated that theres more to GIS than flat-looking maps. For instance, he has made three-dimensional topographic maps of downtown Boston (with video tours in which the viewer flies among and around the buildings) and shown in an animation that the Eiffel Tower could be moved to fit nicely into Government Center Plaza.
Users testify that GIS is not only a technology for displaying information, but an instrument of discovery. Heres evidence from two other historians who have latched onto GIS.
Professor of history Jill Lepore writes in the appendix of her recent book, New York Burning: Liberty, Slavery, and Conspiracy in Eighteenth-Century Manhattan (see Witness to Violence, September-October 2005, page 42), that she began her research by constructing a three-part database that included a list of city residents, a detailed coding of events from a trial that was central to her story, and an inventory of buildings, streets, and meeting places in the city. With [these] data in hand, I then reconstructed the city spatially, using GIS mapping. My initial aim was to detect patterns in the conspiracy, the fires, the confessions, and the trials that were otherwise unobservable.As it turned out, I did discover important patterns, many of which informed my argument.But reconstructing the city proved as much an end itself as a means to another end. The database took me to the streets; it introduced me to the people and places of eighteenth-century New York. When I set about to write, I found myself referring to my database constantly: how long did it take to walk from Hoggs to Hughsons; what kind of people lived on Broadway; how many coopers worked in the West Ward? The database helped shape and refine my argument but also it helped me to understand the city.
Michael McCormick, Goelet professor of medieval history, speaks of a more distant era. I was making a GIS map of communication networks to go with a paper Id written, and I began by writing a legend describing what the map was about. I had to rewrite the legend three times because as I made the map of communications microzones and local ceramic distribution in Byzantine Galilee and Judaea, I kept discovering relationships among the data that I had not noticed before.
McCormick, who will teach The Fall of the Roman Empire next spring, is now making a series of 50 maps with an underlying GIS architecture for that course. One or two maps, for instance, will spotlight the fate of the city of Trier, in Germany: home of the Celtic Treveri tribe; conquered by the Romans in the first century B.C. and founded as a city by Augustus; attacked by barbarians and rebuilt in the third century; a capital of the Roman Empire in the fourth century; finally abandoned by the Romans in the early fifth century, to be sacked and captured by the Franks and possibly the Huns, before ending up as a medieval archbishops town. The maps will show whats going on in Trier in this century or that, as well as abundant information about the empires changing borders, trade routes, settlements, natural conditions, and so on. McCormick will build databases keyed to his 50 maps, so that a student can hit a button to see all the shipwrecks of the fourth century, for instance, or watch short videos made by their professor of monuments, excavations, and other sights in Rome, Constantinople, Milan. He hopes that the new maps will show his students how historians can use GIS as an analytical tool to help understand the avalanche of new information bearing on the Roman Empire coming in from written sources, archaeology, and the natural sciences. And he offers this pedagogical observation: Our undergraduates today are certainly as smart as they used to be, but they are now more visually aware than students 20 years ago. They are quicker to get into visual material than texts.
McCormick is supported in this project by the staff of the Harvard Map Collection, housed in Pusey Library and repository of more than half a million maps (see http://hcl.harvard.edu/libraries/#hmc). David Cobb, its head, has been very helpful digging out relevant historical maps, and he is a whiz at digitizing them, says McCormick. In preparation for his class, the role of the professional staff at the Center for Geographic Analysis will be to search for, evaluate, and obtain data layers, such as digital elevation models or satellite images, from both public and commercial sources; to manage data-conversion contracts with external service providers who will massage the data in various ways to make it suitable for GIS use; to design a database structure for organizing the spatial data; to prepare training materials and deliver the training to the research assistants who will enter data into the database and maintain the database in the long term; to design and build a dynamic data-viewing and -mapping tool that can be downloaded from the course website and run on personal computers without network or license constraints; and to provide instruction to students on how to use that tool. Thus equipped and instructed, McCormicks students will be able to do GIS work on their own, and make their own discoveries about that great historical enigma, the fall of Rome.
But what about other undergraduates? The fact of the matter is that in the Faculty of Arts and Sciences we have no formal GIS curriculum, zilch, said Bol last spring. The problem is very serious. Wendy Guan, director of GIS research services at the center, says that five courses are needed to grasp the field fully: in GIS and computer cartography, in remote sensing, in both introductory and advanced GIS, and in spatial analysis.
The problem for students who just want to put a bit of GIS in their intellectual toolkit eased on September 1 when the government department appointed Harvards first preceptor in GIS: Sumeeta Srinivasan. The center conducted the job search and found Srinivasan in the Division of Engineering and Applied Sciences, where as a postdoctoral research associate she taught a course on the spatial analysis of environmental and social systems, a course with a good deal of quantitative content and some fairly stiff mathematical expectations. The first time she taught it, in the spring of 2005, she had only one undergraduate among her students, Lee Murray 06, who won Harvards Howard T. Fisher Prize in Geographical Information Science for a class project, Spatial Analysis of Redistricting of Congressional Districts in New York State Following the 2000 Census. In 2006 Srinivasan had nine undergraduates in class and one of them, Frances C. Moore 06, won the Fisher Prize for A Spatial Analysis of the Causal Factors of Nepals Maoist Insurgency.
This fall Srinivasan is teaching a less mathematically daunting introduction to GIS and its applications in both the social sciences and environmental sciences. It is one of the prerequisites for her spatial analysis and modeling course in the spring. In the spring she will also offer an advanced GIS workshop in which students can explore some aspect of GIS in depth and do a semester-long project. She plans a three-week intensive introduction-to-GIS workshop to be taught in January in the Longwood Medical Area; it will feed into spatial-statistics course offerings at the School of Public Health.
Guan has a staff of two GIS specialists full time and one GIS applications engineer, who splits his time between the center and the University Library. Two types of projects come to us, she says. Those like Mike McCormicks, in which a scholar understands the possibilities of GIS but lacks the equipment or time to do whats wanted with it, and those in which a scholar says, Im doing this research. How can you help? People in lots of disciplines are starting to recognize that theres something going on with geospatial analysis thats important, says Bol. If they have grant money available for a project, the center will charge for its services. If not, we wont.
Getting the data is often the most challenging step in a project, says Guan. Collecting quality, precise data can be so hard that some researchers become discouraged. But the center itself is collecting and disseminating spatial data sets from diverse sources inside and outside of Harvard. That is a critical part of its mission. Some are public goods and free, shared by government agencies through the Web. Some come from third-party vendors who repackage raw data, add some value, and resell them. The University maintains a marvelous data source, the Harvard Geospatial Library. It keeps and catalogs data sets, so far numbering in the thousands (and many more will come to it, from the center and elsewhere). Go to other libraries on the University Libraries website (http://lib.harvard.edu) for a look. Bewildered newcomers to the site should head at once to the general help menu. The volume of data in the world is increasing exponentially, its price dropping linearly, says Guan. Just think of the number of satellites circling the globe.
Yes, think of those satellites, many equipped with cameras capturing plain or fancy images of the earth. Jason Ur, assistant professor of anthropology, has used declassified intelligence-satellite imagery, in a GIS framework, to reconstruct state-sponsored irrigation systems in the Assyrian heartland from 702 to 681 B.C.
Joseph A. Greene, assistant director of the Semitic Museum, is considering deploying what he calls a slightly theoretical technology that uses low-level aerial scanningfrom helicopters, sayin wavelengths other than visible light, to detect ancient structures now covered by vegetation. The data come back as numbers that are processed into pictures. The scans tell you where to look first. You spot something and go onto the ground and, yes, its a road and it was built by the Romans.
So far, theres been more talk than action at Harvard about using remote-sensing data in research, but, says Guan, there are many ideas. She is a seasoned user of remote-sensing and GPS data. Born in China, she became interested in the facts of geography as a child. But when she went to collegewhen the government, for the first time after the Cultural Revolution, allowed students to take an examination and, based on their scores, go on to collegeshe studied biology. There were still remnants of political censorship, and geography was classified as a state-secret-related major, she says. I wasnt trustworthy enough to study geography.
When China opened its door, Guan emigrated to Canada, where she studied the geography of recreational resources in Ontario at Trent University and got a masters degree. She went on to earn a Ph.D. at the University of Georgia in the late 1980s in an interdisciplinary program in forestry, economics, and geography. Her dissertation was on water-quality modeling with GIS. Her first job was at a government agency in Florida, using GIS applications in Everglades restoration, land and natural-resource management, and regulatory programs. Then she migrat ed to Seattle to work for the Weyerhaeuser Corporation, an international forest-products company. As part of the forestry technology team, she managed GIS, remote-sensing, and GPS technology to keep track of the companys treeswere they diseased? were they drought-stressed?to regulate the fertilization plan, to plot helicopter routes, and much else. Black boxes containing GPS units mounted in the logging trucks promoted efficiency. (They also allowed headquarters to see whether a truck was speeding, which didnt sit well with all drivers, but there was a potential upside to the technology as far as they were concerned: if a truck showed up on a head-office monitor as motionless for two hours on a remote dirt road, trouble could be deduced and help dispatched.) Data from remote sensing and global positioning systems feed into GIS, the core technology. Harvard scholars mired on a dirt road with such data will often require the analytical help of trustworthy specialists.
Guan and colleagues, jointly with the Harvard University Center for the Environment, will host a public workshop later this fall to showcase research at Harvard that uses remote sensing and to provide a high level, up-to-date overview of the technology and its applications.
Alongside its other responsibilities, the center hopes to advance the field of GIS and spatial analysis intellectually. For starters, Bol and his colleagues have proposed that the University raise the money for two professorships, one in the Faculty of Arts and Sciences and one at the design school, to bring to the University persons with a strong background in geographic information science, with broad vision, who are likely to see ways in which geographic information can be relevant in many disciplines and provide a platform for tying interdisciplinary knowledge together. Word is awaited on the realization of that initiative.
|Photograph by Martha Stewart|
Jack Dangermond, M.L.A. 69, of Redlands, California, was on hand for the centers launch and spoke there. Trained as a landscape architect, he came to Harvard originally to earn his masters degree, but spent most of his time working in the design schools Laboratory for Computer Graphics and Spatial Analysis, founded by Howard T. Fisher, a geographer and mathematical cartographer. Dangermonds wife, Laura, worked there, too. The lab was in the process of inventing the very first computer map displays. He and Laura, full of youthful enthusiasm to do good, took what they had learned back to California and founded the Environmental Systems Research Institute (ESRI), an infant nonprofit firm that applied the new computer-mapping techniques learned at Harvard to the analysis of such things as air pollution. ESRI morphed into a booming business, the leading provider of GIS software to business, government, and education. Jack Dangermond remains its influential president. About 140,000 organizations use the companys tools as their foundation for building and applying geographic knowledge. The Dangermonds made a gift of all ESRIs software to the University, so that academics could get access to the technology easily. He also provided the initial endowment for the Fisher prize.
GIS is becoming pervasive as an information infrastructure for managing geography, he said in an interview. As we increasingly affect the planet, the human footprint becomes bigger and more invasive, and we gradually are moving into a role of managing the world, instead of being participants in a natural world. We are going to have to take more responsibility for all the processes of geographywater, climate, the forests, those natural aspects of the landscape that persist. When we weave the various GIS systems together into a kind of nervous system for the planet, we will see and increasingly participate in its evolution. Now GIS brings geography back to Harvard. It left for unfortunate reasons, and that was a big loss to Harvard and to academic geography in general because it signaled that geography was a fading science. Now it returns, not as a traditional place-based geographywhere is Nigeria?but as a science of the planet, modeling our world as a whole, its processes, and their relationships.
Christopher Reed is executive editor of this magazine.
The China Historical GIS
The project maps aspects of Chinese history from 222 B.C.E to 1911 C.E. This map concerns Liangzhe province in 1077. The background shows population density, the colored circles represent the tax quotas for commercial tax collection offices. By 1077, in contrast to earlier periods, explains Peter Bol, Carswell professor of East Asian languages and civilizations, the hierarchy of economic central places, as represented by the tax quotas, no longer corresponded to the hierarchy of administrative central placesthe counties and, above them, the prefectures. This was a sign that a commercial revolution was taking place, because the economy was no longer being subordinated to the administrative system. Some non-administrative towns have more economic activity than their county seats, some county seats are more active than the seats of the prefectures to which they belong. If this can be extended to all of Chinaand it can be, since we have the figureswe will be on the way to showing how economic activity was distributed and where the important trade routes were.
Air Travel and Pandemic Control
Influenza epidemics often begin in the New York City area, as shown by the red area in the map on top, built of data from 122 cities for the five years prior to September 11, 2001. Because of reduced flying activity around New York City after 9/11, the 2001-2002 influenza epidemic began in the Midwest, as shown below. That season was also characterized by slower spread and a later national peak. Says instructor in pediatrics John Brownstein, of the Childrens Hospital Informatics Program at the Harvard-MIT Division of Health Sciences and Technology, and lead author of a just-published paper about this research, For the first time, weve been able to showusing real datathat air travel spreads the flu, suggesting that reducing the number of air passengers might ameliorate a flu pandemic.
Unemployment Rates in Germany
Nicola Fuchs-Schndeln, assistant professor of economics, joined forces with Rima Izem, assistant professor of statistics, and with the help of the Center for Geographic Analysis undertook a spatial analysis of unemployment rates in Germany. Their goal is to explain the stubbornly low labor productivity in the former East Germany, and their findings could have public-policy implications. The project is a work in progress. The map at far left shows the sharp break in unemployment rates at the former East-West border; the unemployment rate in the West is about half that in the East, even after reunification. The simulated unemployment rates are based on a model developed by Fuchs-Schndeln and Izem that explains the gap in unemployment rates by differences in job characteristics and workers' skills in East and West. The model shows the effects of individuals commuting behavior between counties.
Maximizing the Benefits of Urban Bus Retrofits
Urban buses belch unhealthful fine particulate matter, and installing diesel particulate filters is a common retrofit considered by transit authorities. Susan L. Greco, a doctoral student in environmental health at the Harvard School of Public Health, and colleagues did a case study of 25 Massachusetts Bay Transportation Authority bus routes and determined the intake fractionan expression of the emissions-to-public exposure relationshipfor all road segments in the Boston area. They imagined that funds were available to retrofit only half the buses and compared the public-health benefits of retrofitting half the buses on each route to the benefits of retrofitting the same number of buses but exclusively those on the high-intake-fraction routes. The latter course of action increased net benefits by 77 percent.
The Record of American Democracy
Political scientists have known that local context affects political behavior, but the detailed information needed to test hypotheses about this relationship has heretofore been unavailable. This project, managed by the Harvard government department, presents data on election returns, and socioeconomic summaries and demographic measures of the public from the 1990 census, for every state in the country. Shown here are four thematic maps of Cleveland generated from this data.