To do so, it relies on a set of interacting discrete choice models that read and write data into a SQL database. For a specific example, Borning cites residential location choice, "Where a household moves to is a probabilistic choice, which depends on both variables in the household, such as income, number of school age children, et cetera, as well as characteristics of target neighborhoods, such as housing costs, zoning, and other factors. So it's about choices, and probabilities for making those choices calibrated into the actual data."
"Different people are interested in different aspects of the output," says Borning, emphasizing that the hope of a tool like UrbanSim is that it provides neutral projections that all parties can use for their analysis, no matter where they stand on issues like development versus preservation. "The hope is that we get people to agree on the data," says Borning, "Then they may say, 'Here's what's important to me about it,' rather than disputing the facts themselves. There's a lot of emphasis in our research on transparency, on documenting the indicators carefully. We say, here's exactly what it's measuring, and in the online documentation we even give the database code that's producing the values so people can look at it."
UrbanSim may be able to help with future scenarios of large scale evacuations of a city. Current transportation models can simulate how to evacuate a city today, but UrbanSim can alert planners to whether existing infrastructure could handle possible future evacuations, and what part of a region might be most compromised if changes are not made.
The team is currently developing the fourth version of the software, "We've actually had four different versions of the system as we discover new ways to write the models or improve methods and the like," he says, "We've used the system experimentally in a number of cities, including Honolulu, Salt Lake City, Houston, Puget Sound, and because it's open source, people have used it in Phoenix, Detroit, and Paris and Tel Aviv and elsewhere."
Although Houston has made some operational use of it, Borning says that so far most of the use has been experimental, "It turns out it's really a hard problem, we've recently rewritten the system to accommodate the changes to the models, so that slows down the deployment somewhat." His team has worked closely with the nearby Metropolitan Planning Organization (MPO) for the Seattle area, the Puget Sound Regional Council, to better craft a product that meets the needs of real-world practitioners, particularly in terms of how the model displays results. "They preprogram a lot of standard output menus, but give you instructions on how to tweak them and make them your own," says Mark Simonson, Principal Planner for the PSRC.
The PSRC and the U-Washington researchers have made good use of their proximity. Joint development meetings have resulted in a request that the planners email all their spreadsheets that they use to examine model output, so that the programmers can incorporate the indicators and better tailor the program to the council's use. "They've been in our office and had us talk out loud while using a prototype of the interface to see how easy it is for us to get results from the model, and what types of improvements we'd like to see," says Simonson.
For the Puget Sound area, UrbanSim is the right tool at the right time, says Simonson, "Our previous land use forecasting models did not directly take into account land use plans, so they were limited for policy analysis. We were directed by our boards to improve our ability to model 'what if' scenarios, especially land use scenarios."
UrbanSim may hold even greater promise for planners, "You could potentially simulate what would happen with interfering with the natural processes of the Mississippi and its surrounding wetlands," says Borning. "So you could imagine different alternative futures for New Orleans or another city. For example, if we built a road system here, what would the subsequent evolution of the city be." In fact, this may not just be a researcher's hypothetical. Borning cannot go into details, but there have been some preliminary discussions about potentially using UrbanSim for just such a reason. For the present, he Waddell and the rest of the team will continue to refine UrbanSim, although they have no plans to include radioactive monsters.
|This site is maintained by the Digital Government Research Center at the University of Southern California's Information Sciences Institute.|| CONTACT POLICIES|