Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Malczewski, Jacek


Conventional models for simulating land-use patterns are insufficient in addressing complex dynamics of urban systems. A new generation of urban models, inspired by research on cellular automata and multi-agent systems, has been proposed to address the drawbacks of conventional modelling. This new generation of urban models is called geosimulation. Geosimulation attempts to model macro-scale patterns using micro-scale urban entities such as vehicles, homeowners, and households. The urban entities are represented by agents in the geosimulation modelling. Each type of agents has different preferences and priorities and shows different behaviours. In the land-use modelling context, the behaviour of agents is their ability to evaluate the suitability of parcels of land using a number of factors (criteria and constraints), and choose the best land(s) for a specific purpose. Multicriteria analysis provides a set of methods and procedures that can be used in the geosimulation modelling to describe the behaviours of agents.

There are three main objectives of this research. First, a framework for integrating multicriteria models into geosimulation procedures is developed to simulate residential development in the City of Tehran. Specifically, the local form of multicriteria models is used as a method for modelling agents’ behaviours. Second, the framework is tested in the context of residential land development in Tehran between 1996 and 2006. The empirical research is focused on identifying the spatial patterns of land suitability for residential development taking into account the preferences of three groups of actors (agents): households, developers, and local authorities. Third, a comparative analysis of the results of the geosimulation-multicriteria models is performed. A number of global and local geosimulation-multicriteria models (scenarios) of residential development in Tehran are defined and then the results obtained by the scenarios are evaluated and examined. The output of each geosimulation-multicriteria model is compared to the results of other models and to the actual pattern of land-use in Tehran. The analysis is focused on comparing the results of the local and global geosimulation-multicriteria models. Accuracy measures and spatial metrics are used in the comparative analysis. The results suggest that, in general, the local geosimulation-multicriteria models perform better than the global methods.