Urbanization has profoundly transformed many landscapes throughout the world, and the ecological consequences of this transformation are yet to be fully understood. To understand the ecology of urban systems, it is necessary to quantify the spatial and temporal patterns of urbanization, which often requires dynamic modeling and spatial analysis. In this paper, we describe an urban growth model, the Phoenix Urban Growth Model (PHX-UGM), illustrate a series of model calibration and evaluation methods, and present scenario-based simulation analyses of the future development patterns of the Phoenix metropolitan region. PHX-UGM is a spatially explicit urban landscape model and is a modified version of the Human-Induced Land Transformations (HILT) model originally developed for the San Francisco Bay Area. Using land use and other data collected for the Phoenix area, existing growth rules were selectively modified and new rules were added to help examine key ecological and social factors. We used multiple methods and a multi-scale approach for model calibration and evaluation. The results of the different evaluation methods showed that the model performed reasonably well at a certain range of spatial resolutions (120–480 m). When fine-scale data are available and when landscape structural details are desirable, the 120-m grain size should be used. However, at finer levels the noise and uncertainty in input data and the exponentially increased computational requirements would considerably reduce the usefulness and accuracy of the model. At the other extreme, model projections with too coarse a spatial resolution would be of little use at the local and regional scales. A series of scenario analyses suggest that the Metropolitan Phoenix area will soon be densely populated demographically and highly fragmented ecologically unless dramatic actions are to be taken soon to significantly slow down the population growth. Also, there will be an urban morphological threshold over which drastic changes in certain aspects of landscape pattern occur. Specifically, the scenarios indicate that, as large patches of open lands (including protected lands, parks and available desert lands) begin to break up, patch diversity declines due partly to the loss of agricultural lands, and the overall landscape shape complexity also decreases because of the predominance of urban lands. It seems that reaching such a threshold can be delayed, but not avoided, if the population in the Phoenix metropolitan region continues to grow. PHX-UGM can be used as a tool for exploring the outcome of different urban planning strategies, and the methods illustrated in this paper can be used for evaluating other urban models.
