In lieu of an abstract, here is a brief excerpt of the content:

C h a p t e r 8 Urban Growth Models: State of the Art and Prospects John D. Landis Once known as urban activity models, urban growth models (UGMs) emerged in the mid-1960s out of advances in regional science, huge improvements in computing speed and storage capacity (or so they seemed at the time), a newfound surplus of detailed activity data, and federal mandates coupled with funding for metropolitan planning organizations (MPOs) to back up transportation funding requests with careful projections and hard-headed analysis. Overhyped and underdeveloped , early urban models soon proved unreliable. Their epitaph, ‘‘Requiem for Large-Scale Models,’’ published in the Journal of the American Institute of Planners (Lee 1973), is still one of the most widely read and cited journal articles in planning. By the early 1990s, UGMs had staged a small-scale comeback, buoyed by new modeling techniques—especially discrete choice models— further improvements in computing speed and storage capacity, the migration of GIS from mainframe to desktop, advances in data distribution channels (first CDs, later the Internet), and by the desire of local governments and MPOs to go beyond linear trend lines to investigate alternative planning scenarios. By 2000, a survey by the U.S. Environmental Protection Agency had uncovered more than twenty urban forecasting and simulation models, most of which had not existed just ten years earlier (U.S. EPA 2000). Recent and dramatic increases in the availability of high-resolution satellite imagery have given even greater impetus to urban modeling. Recent developments in urban modeling have been chronicled by Batty (1994, 2005) and Wegener (1994, 1998b). For the most part, UGMs have been developed and applied in the Global North, where their data needs can be more easily met. Recent improvements in remote sensing offer potential for their use in the Global South. Urban Growth Models 127 The allure of UGMs remains as siren-like today as in the 1960s. Foremost , they offer the promise of reducing huge volumes of local data into a few robust and understandable generalizations about processes of urban change. Second, they promise to facilitate thinking about the future of cities in ways that are new, nontraditional, and spatial. Third, with the addition of discrete choice modeling, they offer the ability to link the decisions and behaviors of individual agents to metropolitan outcomes; and this, more than any other capability, is the key to planning for sustainability. Fourth and most important, they provide a key analytical bridge between envisioning alternative urban development patterns and evaluating their impacts. Promise and performance, however, don’t always go hand in hand. Fifteen years after their resurrection, questions about UGMs still remain. Are they facilitating better metropolitan investment and policy decisions? Are they leading to a richer, more nuanced, and more comprehensive understanding of the effects of those investment and policy decisions? Are they helping open up local and metropolitan discussions about development to new ideas, approaches, and participants? Are they making the practice of local and metropolitan planning noticeably better ? The answer to these questions for the most part is ‘‘no’’: UGMs are still too complicated, still too untested, and still too slow to adapt to local circumstances to be used regularly and reliably. Nonetheless, progress continues, and the state of the art of urban modeling advances daily. This chapter takes a hopeful but critical look at UGMs through the lens of four archetypes: (1) the IRPUD (Institute for Regional Planning and Urban Development) model developed at the University of Dortmund , the best heir to the spatial interaction models of the 1960s; (2) the SLEUTH (Slope, Land Use, Exclusion, Urban extent, Transportation , Hillshade) model, an example of the use of cellular automata (CA) procedures to model urban change; (3) the California Urban Futures (CUF) family of urban growth models, the first to employ GIS; and (4) UrbanSim, the leading example of agent-based urban modeling.1 All four share two essential characteristics: they are calibrated, which means their coefficients are estimated from historical experience and are therefore not ad hoc; and they are spatially explicit, capable of looking at urban change at the level of an individual parcel (or comparably sized grid cell). This chapter is organized in four sections. The first, ‘‘Why Urban Growth Models?’’ explains UGM functions. The second, ‘‘Four Archetypal Urban Growth Models,’’ introduces the four selected UGMs. The third, ‘‘Model Comparisons,’’ looks at their respective structures, calibration procedures, ease of data assembly, and usefulness for simulation . The fourth, ‘‘A UGM Research and Development Agenda,’’ offers...

Share