Today’s urban settlement structures are developing and changing at a faster rate than ever before. Applied geographers and planners are frequently posed questions such as: What will be the effect upon the existing retail structure of a town of the development of a new out-of-centre grocery store? What might be the effects of renewal policies on house values in an urban area? How will the density of population across a city change if more central ‘brown-field’ sites are developed in preference to peripheral ones? And what would be the implications of either of these developments for non-car-based transport policy? Computer-based urban models and simulations provide answers to these and other important, explicitly geographical, questions.
These questions are fundamentally about changes in the form of urban areas and also the ways in which they function (Batty and Longley 1994). Applied and other geographers have long sought to understand these problems by thinking of towns and cities as systems, in which interrelated geographical units or objects form a ‘set’ that together function as a whole (Haggett et al. 1977:6). In reality, individual towns and cities (like most all other geographical systems) do not develop in isolation from the rest of reality, but it is often helpful to screen out these more general considerations in order to ‘bound’ the specific problems that we are interested in. Limiting the extent of what we consider for practical purposes to be an ‘urban system’ is an important first step in urban modelling.
Many of the other chapters in this book have used models as practical tools to understand aspects of real-world systems. In the most general terms, a ‘model’ can be defined as a ‘simplification of reality’, nothing more, nothing less. In order to answer the sorts of specific and focused questions that were posed at the start of this chapter, it is necessary to disregard information about aspects of the urban system that are irrelevant (or only partially or indirectly relevant) and focus on those remaining system characteristics that have the greatest impact upon outcomes. Thus a good model selectively retains all of the aspects of a system that are important from a particular point of view and discards those that are not. This process of selection is central to the art of model building.
In practice, the selected aspects of the system are represented in an urban model principally using quantitative data. Although such data are used to represent system characteristics, they rarely if ever do this in a way that is either perfectly accurate or precise. This arises for a variety of reasons which are common to all quantitative analyses of socio economic systems. There is not enough space here to explore these in much detail (but see Martin 1997 for an introductory overview of some important aspects). Briefly, socioeconomic data (such as those from censuses) are usually averaged across administrative zones prior to being made available to researchers for