Simulating sprawl: A dynamic entity-based approach to modeling North American suburban sprawl using cellular automata and multi-agent systems

The aim of this research is to develop new spatial models for studying complex urban systems. The models are designed to simulate entity-level dynamics in cities, extending cellular automata and multi-agent systems in a patently spatial fashion. The suitability of the approach for modeling urban systems is demonstrated with an application to sprawling suburban growth in the context of North American cities. The thesis discusses recent trends in urban simulation, with emphasis on a new wave of approaches to modeling urban systems. The application of that methodology to the study of suburban sprawl is demonstrated through the development of simulations based on the idea of Geographic Automata Systems (GAS). GAS act as a framework for extending automata-based methodologies from the computing sciences with spatial functionality. Two GAS models are built and discussed in the context of sprawl formation. The first focuses on sprawl as a growth-based phenomenon, simulating the geographic mechanisms that give rise to sprawl in hypothetical and actual metropolitan regions. The second model approaches the idea from a community-level standpoint, simulating dynamics within a residential submarket hypothesized to be in a sprawling urban area. The results of the research demonstrate the applicability of the modeling framework for simulating urban systems across a variety of scales. The models also reveal several insights regarding the geographical nature of sprawl in a North American context