Introduction to Urban Simulation

This chapter provides an introduction to urban simulation, which we interpret broadly to mean operational models that attempt to represent dynamic processes and interactions of urban development and transportation. Its intent is to provide the reader with a reasonable understanding of the context and objectives for urban simulation modelling within metropolitan areas in the United States, the limitations and challenges of urban simulation models, the design choices involved in developing operational models, and how such models are applied. THE CONTEXT AND OBJECTIVES FOR URBAN SIMULATION Urban systems are becoming ever larger and increasingly complex as urban economies, social and political structures and norms, and transportation and other infrastructure systems and technologies evolve. Scarce resources make efficiency critically important, and in a democratic context that involves many stakeholders with conflicting values and priorities, it is neither feasible nor appropriate to deal with major land use and transportation policies and investments as isolated choices to be decided by planners or bureaucrats within the bounds of a single organization. Mathematical and theoretical models have long been used to attempt to reduce complexity and encode a clear and concise understanding of some aspects of urban Introduction to Urban Simulation 2 structure and transportation, as exemplified by the classic work on the Monocentric Model of the city (Alonso, 1964; Mills, 1967; Muth, 1969). While the value of theoretical models is facilitating a broad understanding of some underlying principles of urban development and transportation, much of this work remains too simplified in its assumptions and too abstract to be of direct value to agencies needing to inform decisions about specific policies and investments in particular urban settings. To begin to address more operational needs in planning and policy decisions, computerized models representing urban travel and land use began to be developed and used from at least the 1960’s in the United States, with the advent of the Urban Transportation Planning System for travel demand forecasting (Weiner, 1997), and the subsequent work on spatial interaction models for predicting locations of households and jobs across urban landscapes (Putman, 1983), which emerged out of earlier work on the Lowry gravity model (Goldner, 1971). A separate branch of applied urban modelling developed along the lines of the Input-Output model of the macroeconomy developed to describe the structure of economic flows between economic sectors (Leontief, 1966), adding a spatial component and transportation costs to represent economic and transport flows between zones in a region (de la Barra, 1989; Marcial Echenique & Partners Ltd., 1995). The objectives for much of the work on land use and transportation modelling in the United States from the 1960’s through the 1980’s were focused on the planning problem of determining transportation capacity needs—mostly focusing on roadway capacity—to accommodate expected demand generated by predicted land use patterns represented by the spatial distribution of households and jobs within a metropolitan area at some future planning horizon. Over the 1970’s and 1980’s, increasing pressure from environmental groups, proponents of transit, and others, led to a substantial shift in policy objectives, reflected in the passage of the Clean Air Act Amendments (CAAA) of 1991 and the Intermodal Surface Transportation Efficiency Act (ISTEA) of 1990. By 1990, a significant degree of attention had emerged on the effects of transportation improvements on land use changes, the potential for long-term induced demand from highway expansion that might significantly undermine the expanded capacity through additional travel, and increasing environmental consequences in the form of emissions and loss of open space due to stimulation of low-density development at and beyond the urban fringe. The passage of the CAAA and ISTEA legislation set the stage for lawsuits by the Sierra Club and the Environmental Defense Fund and other environmental groups in the San Francisco Bay area, Chicago, Salt Lake City, and other metropolitan areas, on the grounds that the computerized transportation and land Introduction to Urban Simulation 3 use modelling and planning processes did not adequately account for these complex feedbacks between transportation improvements, land use, and air quality (Garrett and Wachs, 1996). Simultaneously, the policy environment began to shift towards a more multi-modal approach to transportation, including non-motorized and transit modes, other demandside policies began to emerge as alternative ways to match capacity to needs, including a range of transportation system management techniques (ramp metering, traffic light signalization, and so forth), travel demand management (ride-sharing, staggered work hours, parking pricing policies, congestion pricing, etc.), and land use policies (jobs-housing balance, urban growth boundaries, transferable development rights, concurrency requirements or adequate public facilities ordinances). The range of policies and strategies under potential consideration by metropolitan areas to address transportation needs has essentially exploded over the past two decades, from a fairly narrow focus on highway capacity expansion to a multi-modal transportation capacity and demand management and land use policies. The objectives for operational urban land use and transportation models have consequently grown. Besides the growing need to test the effects and effectiveness of an ever-more diverse range of land use and transportation policies, and their interactions, pressures on operational modelling have grown from a very different perspective. From the earliest efforts to develop operational urban models, critics have raised serious concerns about the viability of such models. Lee’s “Requiem for Large Scale Urban Models” in 1973 (Lee, 1973) cogently argued that efforts to develop operational urban models had failed, and would likely continue to fail for a variety of reasons ranging from insufficient theory to computational and data demands. Much of the operational work in urban transportation and land use modelling has been criticized as being too much akin to a ‘black box’, meaning that its theory and implementation were not clear enough to an observer attempting to understand and evaluate it. While some of the criticism was aimed at problems that have since clearly been addressed, such as computational requirements, other concerns, such as insufficient behavioural theory, still require substantial attention and are not widely addressed even in many current operational simulation models. It is valuable to keep these critiques in mind when examining current and emerging modelling approaches. Combined with this kind of skepticism on a technical level, growing pressures have emerged on the planning and policy-making arenas to become more open and participatory—in short, more democratic. The tradition that has emerged within planning agencies of having technical staff run models to support planning processes, without clear and open access to the models, their assumptions, their theoretical Introduction to Urban Simulation 4 foundation and their practical application, has become very inconsistent with the current context demanding more democratic analysis and decision processes. In summary, the context and objectives for urban modelling have grown far more complex over the past two decades, and combine to shape the needs for urban model development in ways that are sensitive to a range of land use and transportation policies and their interactions, that build on clear and defensible foundations in behavioral theory, and that facilitate participation in the testing of alternative policy strategies and their evaluation. These are formidable challenges to address in a satisfactory way. THE DESIGN AND IMPLEMENTATION OF AN OPERATIONAL URBAN SIMULATION SYSTEM In the balance of this chapter, we explore recent advances and experience in the design and development of urban models that attempt to address these requirements and contextual factors. General questions of model design and application discussed below are grounded in a case study of the development of the UrbanSim system in the Puget Sound region, and the rationale for each design choice is presented. UrbanSim has been developed since the late 1990’s to address many of the concerns identified above, and represents an ongoing interdisciplinary research development effort to provide operational tools to support the assessment of land use, transportation and environmental policies and plans within metropolitan areas (Noth, Borning and Waddell, 2001; Waddell, 2000; Waddell, 2002; Waddell, et al., 2003). We propose that models be considered within a broader context in which they will be used to guide or inform policy choices, and that this be considered a participatory and iterative process. Figure 1 depicts the proposed policy development process as one that begins with a visioning, or goal-setting phase, and proceeds through development of objectives, identifying policies, formulating policy packages or scenarios, using models to examine the effects of these policy scenarios on important outcomes, and developing indicators and evaluating the effectiveness of the policy scenarios in achieving the original policy goals and objectives. The process is likely to be iterative for several reasons, chiefly that different stakeholders will disagree about the relative weight to place on each goal, and there may be many possible policy scenarios that could be evaluated. Ultimately, the process should lead to a convergence of agreement on a set of goals and on the preferred policy strategy for achieving them. Our hope is that a well-designed policy process that integrates use of models in a participatory decision process will increase the likelihood of a cooperative resolution, Introduction to Urban Simulation 5 as compared to the frequently observed political gridlock now observed in many metropolitan regions.