Exploring the real "SimCity"
They craft development policies, study growth and design transportation systems, government agencies and infrastructure with the goal of creating a more civilized place to live.
Yet, any given city grows in an organic, intricate and usually idiosyncratic fashion, often becoming a huge, complex organism that nearly defies control, let alone management or growth projections.
Planners use a variety of tools to help understand the long-term effects of new policies. But existing tools are sorely inadequate to model the broad range of impacts that a zoning change, a subway line, or a new freeway might have on city-dwellers' lives.
Fueled by a recent Digital Government ITR grant, researchers at the University of Washington are exploring methods of computer-simulating how cities grow - and refining technology that could help remove much of the guesswork from urban planning.
The UrbanSim project is researching ways of building a complete, flexible and scalable microsimulation of urban growth.
Led by UrbanSim project directors Paul Waddell and Alan Borning and the team is building and testing systems of interdependent simulations. The "microsimulation," as they call it, attempts to map the way that cities' choices in transportation, zoning, environmental regulation and other domains affect development and the quality of urban life.
"We're interested in the potential to use the modeling system to mediate conflicts that are locking many urban areas up in litigation over management policies," says Waddell, an associate professor at the University's Daniel J. Evans School of Public Affairs and Department of Urban Design & Planning.
"All these politics are still there - there's nothing about what we're doing that will make it go away," Waddell says. "But we have the naive hope that (UrbanSim) will cause local stakeholders to be aware of the spillover effect that their development preferences have on others, and of the effect that others' preferences have on their development decisions."
And planners in Seattle have earmarked some $300,000 a year in the next two years toward calibrating UrbanSim to help with major transportation and land-use plan updates, says Larry Blain, Principal Planner for the Puget Sound Regional Council.
Until now, the council had been relying on limited, coarse-grained growth models such as DRAM/EMPAL for crafting transportation and land-use policy for the region of 3.5 million people,and then using the Delphi Method for refining projections, says Blain.
"This results in reasonable forecasts, meaning everyone agrees to them, but it doesn't allow us to test things like doing a good test of sensitivity to transportation accessibility. And it doesn't have land pricing in it at all, so we can't analyze the housing prices."
The council plans to work with researchers on using UrbanSim to use more fine-grained datasets to achieve disaggregate projections that take into account such issues as the decisions of developers, local governments and even homeowners.
"By being sensitive to a number of variables, including local land-use plans ... we can test alternatives and actually see different outcomes," Blain says. "We can't model that now, but we expect to when we use Urbansim, to see different outcomes from different policy assumptions."
UrbanSim began in 1995 when Waddell - then a doctoral student working with the North Central Texas Council of Governments - began experimenting with computer-modeling urban growth. He built it on a Java platform - fortuitously chosen in the language's commercial infancy - which has provided a solid and scalable foundation.
Together, they drafted computer science students to help rework Waddell's original, very rigid model code into a more robust simulation platform - one that could accept different kinds of data and modeling parameters - and that could survive use by large cities with huge datasets and complex questions. And they began experimenting with ever more complex real-world problems.
One current question the team is trying to answer is "How do we represent things in the urban environment in sufficiently fine detail that we can get the kind of simulation results we want," says Borning. Current land-use and transportation models are too coarse to accurately simulate the effects of certain policies - such as ones to create more "walkable neighborhoods."
One of the thornier problems has been trying to present the massive amount of simulation data in a useful and understandable way. "Different people bring strongly held values about urban development to the table," says Borning.
"For example, one person might be particularly concerned with minimizing traffic congestion, another with the impact of our decisions on the environment. One of the things we're trying to do is let different stakeholders - planners, business people, members of advocacy or neighborhood groups, and other citizens - understand and evaluate the results of the simulation according to what's important to them."
To do this, the team is applying an emerging methodology, called Value Sensitive Design, that seeks to incorporate consideration of human values in a principled and comprehensive way throughout the design of information systems.
Working with cities such as Salt Lake City, Honolulu and Eugene, Oregon, the researchers have been testing the models - now incorporating some 15 to 20 different configurations - with real-world problems.
In one case, the UrbanSim toolset is even being used under court order to help settle disputes in a lawsuit by the Sierra Club over projected pollution effects of a Salt Lake City freeway plan.
Indeed, the open-source UrbanSim model has now been downloaded by people in more than 60 nations, including Taiwanese transportation planners and Parisians working on a rail study, says Waddell.
While the scientists are interested in tech transfer, they are still wrestling with issues of configuring the system to accept heterogeneous data from disparate sources, and of providing a solid graphical user interface, Waddell says. "Over the course of the next year, I expect that we'll see a higher level of transferability," he says.
|This site is maintained by the Digital Government Research Center at the University of Southern California's Information Sciences Institute.||" CONTACT " POLICIES|