Calibration of the SLEUTH urban growth model for Lisbon and Porto, Portugal

Abstract

The SLEUTH model (slope, landuse, exclusion, urban extent, transportation and hillshade), formerly called the Clarke Cellular Automaton Urban Growth Model, was developed for and tested on various cities in North America, including Washington, DC, and San Francisco. In contrast, this research calibrated the SLEUTH model for two European cities, the Portuguese metropolitan areas of Lisbon and Porto. The SLEUTH model is a cellular automaton model, developed with predefined growth rules applied spatially to gridded maps of the cities in a set of nested loops, and was designed to be both scaleable and universally applicable. Urban expansion is modeled in a modified two-dimensional regular grid. Maps of topographic slope, land use, exclusions, urban extents, road transportation, and a graphic hillshade layer form the model input. This paper examines differences in the model's behavior when the obviously different environment of a European city is captured in the data and modeled. Calibration results are included and interpreted in the context of the two cities, and an evaluation of the model's portability and universality of application is made. Questions such as scalability, sequential multistage optimization by automated exploration of model parameter space, the problem of equifinality, and parameter sensitivity to local conditions are explored. The metropolitan areas present very different spatial and developmental characteristics. The Lisbon Metropolitan Area (the capital of Portugal) has a mix of north Atlantic and south Mediterranean influences. Property is organized in large patches of extensive farmland comprised of olive and cork orchards. The urban pattern of Lisbon and its environs is characterized by rapid urban sprawl, focused in the urban centers of Lisbon, Oeiras, Cascais Setúbal, and Almada, and by intense urbanization along the main road and train lines radiating from the major urban centers. The Porto Metropolitan Area is characterized by a coastal Atlantic landscape. The urban pattern is concentrated among the main nuclei (Porto and Vila Nova de Gaia) and scattered among many small rural towns and villages. There are very small isolated patches of intensive agriculture and pine forests in a topography of steep slopes. These endogenous territorial characteristics go back in time to the formation of Portugal — with a “Roman-Visigod North” and an “Arabic South” [Firmino, 1999 (Firmino, A., 1999. Agriculture and landscape in portugal. Landscape and Urban planning, 46, 83–91); Ribeiro, Lautensach, & Daveau, 1991 (Ribeiro, O., Lautensach, H., & Daveau, S., 1991. Geografia de portugal (4 Vols., published between 1986 and 1991). Lisbon, Portugal: João Sá de Costa)]. The SLEUTH model calibration captured these city characteristics, and using the standard documented calibration procedures, seems to have adapted itself well to the European context. Useful predictions of growth to 2025, and investigation of the impact of planning and transportation construction can be investigated as a consequence of the successful calibration. Further application and testing of the SLEUTH model in non-Western environments may prove it to be the elusive universal model of urban growth, the antithesis of the special case urban models of the 1960s and 1970s.

Introduction

Modeling is essential for the analysis, and especially for the prediction, of the dynamics of urban growth. Yet the successful application of a model in one particular geographical area does not necessarily imply its successful use in another setting where local characteristics, territorial constraints and the classic site and situation properties of economic geography ensure that different development paths have been followed. Urban and environmental models need to be adapted to or able to “learn” the endogenous characteristics of the particular milieu that they explain and predict. Models are often judged by their predictive power. Yet, to model urbanization across locales, it is just as important to test the efficacy of the model's algorithms at capturing and simulating the land transformations that are specific to a place (Batty & Xie, 1994b, Clarke et al., 1996, Li & Yeh, 2000).

This paper focuses on calibrating the SLEUTH model, formerly the Clarke Cellular Automaton Urban Growth Model (Clarke & Gaydos, 1998, Clarke et al., 1997) for two Portuguese metropolitan areas. SLEUTH is an acronym for the input layers that the model uses in gridded map form: Slope, Land Use, Exclusion, Urban Extent, Transportation and Hillshade. The basic growth procedure in SLEUTH is a cellular automaton, in which urban expansion is modeled in a spatial two-dimensional grid. Diffusion, breed, spread, slope and road coefficients control the behavior of the cellular automaton, and four types of growth behavior can take place: spontaneous, diffusive, organic and road-influenced. Self-modification of the rules changes the control parameters when modeled growth rates are exceeded, so that the model's behavior includes feedback (Clarke et al., 1997). In cellular automata simulating artificial life, self-modification is equivalent to adaptation or evolution, and the calibration method used allows the model to “learn” its local setting over time (Clarke et al., 1996). This learning is quantified by the variation during calibration of the five control parameters. Calibration of the model has taken place for many cities within North America, but not elsewhere. Application of the model to Lisbon and Porto in Portugal is the first application to European cities, and indeed the first major application outside of the United States.

The SLEUTH model was applied to the metropolitan areas of Lisbon and Porto. These Portuguese cities present very different environmental and geographic characteristics that test the model's flexibility to adapt to different urban realities. Lisbon is the capital of Portugal, and the administratively defined metropolitan area includes large patches of farmland comprised of olive, cork, and fruit orchards surrounding the mouth of the Tagus River. The urban pattern of Lisbon and its environs is characterized by recent rapid urban sprawl, focused in the urban centers of Lisbon, Oeiras, Cascais Setúbal, and Almada, and by intensive urbanization along the main roads and train lines radiating from those major urban centers. By contrast, the Porto Metropolitan Area is characterized by a coastal Atlantic landscape at the west-facing mouth of the smaller River Douro and is surrounded by mountains. The urban pattern is concentrated at main nuclei (Porto and Vila Nova de Gaia) and settlements are scattered among many small rural towns and villages with small patches of intensive agriculture and pine forests.