Clark University Research
950 Main Street • Worcester, MA 01610
Tel: 508-793-7711 • academicaffairs@clarku.edu

Active Learning and Research
Active Learning and Research
Geographer Ron Eastman develops geographic information system software that helps urban and rural planners make informed decisions about how to allocate land for different purposes.

Meet the researchers: Co-participants in learning and investigation

Interview with Professor Ron Eastman
Geographer Ron Eastman is a pioneer in the application of computer technology to geographic analysis. Since 1987 he and his research team have been developing and refining the capabilities of IDRISI, a geographic information system (GIS) that provides tools for sophisticated spatial analysis and the processing of remotely-sensed imagery. In a recent conversation, Eastman discussed IDRISI's special capabilities for decision support and uncertainty analysis in the context of land allocation, and the link between his research interests and classroom teaching.

You received a bachelor's degree in psychology. What inspired you to pursue graduate work in geography?

Travel! I was living in Europe after I finished my bachelor's degree and traveling a fair amount in Europe and Asia.

In the process of exploring, I had access in London to some fabulous maps. To me the most beautiful ones were the air-navigational charts that featured wonderful renditions of topography. I found I could spend hours looking at these maps, learning about geography and topography. This experience led into a fascination with maps, and ultimately a decision to learn how one made maps. I enrolled in a Ph.D. program in cartography in London.

The switch from psychology was easy, because cartography at that time was focusing on the process of visual communication, and how to craft the map to serve the purpose of thematic, rather than topographic mapping. My background in perceptual and cognitive psychology fit perfectly into the research agenda of cartography.

And when the digital age came along in the late 1980s, you followed the transition from maps to computers as a more flexible medium for storing spatial information?

Yes. I think the real appeal of computer technology has been its ability to help us understand and facilitate the map use process. The initial application of computer technology was to map production. It was natural to look at the computer as a way of speeding up the process of constructing the map, and then facilitating the revision of the map as time went on. What the computer also offered, however, was a way to focus on what people were doing with maps. So the whole focus changed from the depiction of spatial information to the analysis of that information. My primary focus since moving into the GIS realm has been to see how we can move the technology further into the analytical realm.

For example, consider the practice of constructing a road between two locations. There are a lot of steps involved, and you want to do it as quickly and inexpensively as possible. Traditionally you might take some maps showing elevation and land cover and try to estimate slope and the route that could be most easily constructed through the given landscape. An exacting engineering analysis might take months.

A geographic information system (GIS), on the other hand, provides many tools that can be brought to bear on the problem. Slope estimation is an example. In the computer, elevation can be stored, not just as a set of contours, but also as a grid or matrix of height values. By examining each height value in the context of its neighbors, the software can quickly and accurately calculate slope at any location. Something that would have taken months to do using paper maps can be done in a matter of seconds with GIS. Another example is the process of choosing an optimal route for that road. Geographers developed software routines for GIS to determine, based on a number of variables, the least cost route from one point to another.

IDRISI is unique as a GIS in that it incorporates decision support and uncertainty management tools. What is decision support? What is uncertainty management?

In 1990 we had a request from the United Nations to host sensitization sessions with different groups about the potential of GIS for spatial decision making. Over the course of about five years, we traveled all over the world and talked to many, many decision makers. What we found universally was that their decision making focused on the allocation of land for various purposes. We realized that GIS as it currently existed was not very well developed for that task. GIS had developed with information-retrieval methodologies, but it did not incorporate the logic of what makes a good decision.

As a project for the United Nations Institute for Training and Research (UNITAR), we took on the task of creating GIS routines specially targeted to the needs of these decision makers. One of our colleagues here at Clark, Sam Ratick, specializes in decision making, and a group of graduate students and I decided to get together with Sam. We explored decision making in a broad context, drawing upon literature from economics, business, regional science, and so on, and discussed how that might be incorporated into GIS. It got to the point where we were meeting several times a day! It was the most intense learning experience I've ever been through. Likewise for the graduate students, all of whom developed Ph.D. dissertations related to different topics in decision making.

We learned that there was great potential for developing decision making tools for GIS, and we set out to do that. We eventually constructed a set of tools for multi-criteria and multi-objective analysis in land allocation. These tools allow the analyst to incorporate many different criteria into the decision process, prioritize them, and create a map showing relative degrees of suitability for the particular land use. The tools also allow the criteria to be evaluated from the perspective of more than one objective.

For example, town planners might want to develop a zoning map that identifies areas suitable for different types of use. The criteria might include proximity to the road infrastructure, slope, land cover, and restricted areas. There might be several groups of stakeholders with different objectives. One group might be aiming to increase the tax base, while another might have land conservation and protection as their primary goal. The tools incorporated into IDRISI have the ability to help reconcile conflicting objectives when necessary, and to optimize solutions where there are complementary objectives. The development of these tools is ongoing.

How does uncertainty management fit in?

Uncertainty management has an important relationship to decisions about land allocation. To make good decisions, we need information that's accurate. But accuracy is not a yes or no proposition in geographic data, it's a matter of degree. Each set of data brought to bear on a problem has associated levels of accuracy and precision. Depending on the quality of data, the end result can have a substantial amount of uncertainty.

At IDRISI we've examined how error propagates through, and combines with, other sources of uncertainty in the decision making process for land allocation. All of those inputs--the criteria we choose, the data we use, our inference structure, and how we link all those things together--factor into an end result about which we say: that's the best location--at least we think so.

But what we also need to know is the reliability of our decision. The challenge that we've taken on at IDRISI is to create GIS tools that not only allow us to choose the best location for an activity, but to specify the degree of uncertainty in that decision (i.e., decision risk). And in most cases, the only way to overcome that uncertainty is to spend money for better quality data. That's something that politicians don't want to do. Nonetheless, a review of decision risk confronts them with the problem.

So one reason we study decision making at IDRISI is to facilitate sound decisions. But another reason is to provide decision makers with tools that will allow them to say 'I made the most reasonable decision for this situation, given the information that I had.' As computer technology becomes more a part of the decision making process, the more that process becomes explicit and transparent, and the more decision makers are likely to be held liable for their decisions. For example, it's possible now to ask a forest manager how he or she decided where to start a controlled burn that later got out of control. An examination of the decision making process might reveal that the person made a poor decision. But it might just as well show that he or she made the best decision possible given the quality of the information available.

Understanding data uncertainty also allows the decision maker to evaluate the risk that accompanies a decision choice. For example, one of the application areas where we've worked is food security, particularly the distribution of food aid in Africa. Food security analysts want to be able to map out, in advance, where the food supply is likely to be most vulnerable, so that resources can be targeted most effectively. Of course, if resources were unlimited, every household could be given food for a coming crisis. But since resources are finite, you don't want to risk giving food to areas that won't need it. Knowing the level of uncertainty associated with the data used to predict areas of vulnerability allows you to allocate your resources more intelligently and efficiently. I think managing data uncertainty is the next big horizon in GIS.

How are you able to incorporate opportunities for research and active learning into your teaching?

There are two ways. One of the things I require of students in all of my classes is a project. I call it an exploration, not a research project. Students are asked to take one of the topics from class and apply it to an area of interest to them. It works really well because invariably these students have a data set they want to explore, or a question they want answered. The project gets them involved in using GIS in ways that aren't already neatly set out for them. In many cases, students want to do things that maybe have never been done in GIS before. We talk through how the technology could be applied. The range of analytical tools in GIS is now extensive enough that we can usually figure out how to do it. Sometimes that process will lead to new software developments. One of the luxuries that we have at Clark is that we own our own software system--IDRISI--and we can make it do whatever we want.

So that's a very active learning process in a way that I can never anticipate, nor can they, but it drives it home. I teach them the technology and I provide information. They can take that in to a certain extent, but when they work on a project, they have made it their own. They come in as students and they leave as analysts. It's a great experience all around and just one of the things that I really enjoy about being here at Clark.

Secondly, technology has given us the ability to do multi-media presentations in the classroom. Students like it, and I like it, too. When I teach, I typically show a PowerPoint presentation that provides a structure and a variety of illustrations. Then, for each major topic, I provide a case study example using IDRISI. By combining in class a multi-media projector with my laptop, I can do live case study analysis. I use real data and perform the analysis in real time. Often during this process a student will raise his or her hand and ask what would have happened if I'd made a different choice in the analysis. I can say, truly, I don't know, let's find out. And we'll go back and do that.

For example, in one class I was demonstrating a rather esoteric technique that produces odd-looking images. A student asked what would have happened if I'd done such and such instead. I said I didn't know, but we tried it and produced a result. Then the student asked how that result was different from the previous one. We used a simple GIS tool to look at the difference between the two results and produced a most bizarre pattern. Immediately, we had a room of students throwing out possible explanations. The best explanation came from an undergraduate who likened the result to the pattern generated in a vibrating pan of water. We thought perhaps it could be vibration in the satellite platform. I don't know what that noise was in the image, but vibration was a pretty good guess, a guess that came from a discussion involving both graduates and undergraduates. Although the class consisted of about 40 people, it was like an intimate lab experience.

So I find that environment works extremely well. My wife asked once why I'm still writing lectures for classes I've been teaching for years! It's because every year I'm learning new things, and part of that learning happens in the classroom. It works two ways: I learn and the students learn as well.

What opportunities are available for undergraduates who want to get involved in research?

I think we have a culture in geography and at Clark of trying to involve undergraduates in research, and we each do it in different ways. For me it's easy, because in that process of making the project assignment, I will invariably have some students who need data, or who just aren't sure what they want to do. That invariably leads to a discussion between the student and me. We'll meet after class and talk about the student's interests and coursework. Many times I can identify a faculty member with related interests, and I encourage the student to meet with him or her. Often, within a day or so, the student is back with the needed data or a problem defined, and in the process he or she has suddenly become involved in research. It helps the faculty member and it helps the student.

Initially, I think undergrads tend to participate as research assistants in the traditional sense of the word; that is, they're learning first-hand by observing and participating. It's an entry point that's very unpressured. They can observe, they can see what research is like, and in many cases they can get very substantially involved. By the time they graduate, the stronger students have moved from being research assistants to research associates, such that they're really very much co-participants in the process of learning and investigation.

I had one student, for example, who was interested in physical geography and wanted more background in climatology. I got her involved in one of our projects focusing on drought in southern Africa. She assembled a lot of the climatological research for that, moved on to a year-long internship with Harvard Forest, and then entered graduate school. It was part of a continuum where she went from being a student learner, to being a student taking charge of her own destiny, to carving out her own research career. That's exciting. I enjoy facilitating that.

Contact Information Site Search

Additional Resources
Search by student
Search by professor
Search by department
Fund it
Present it

Ron Eastman
Professor Ron Eastman

 Learning through exploration QuickTime

 Involving undergrads in research QuickTime

Download software.



© 2014 Clark University·