In recent years a number of studies have sought to demonstrate that a range of urban functions are affected by the residential density of the urban area. However, several authors have raised questions about the computational validity of some of the calculations of urban density used in these studies. In particular, questions have been raised as to what is a reasonable definition of the urban area over which the density should be calculated. If the urban area is defined very tightly to include only those built-up areas within the urban boundaries (however defined), then the density will be higher than if a liberal definition is adopted for the urban area which includes the fringe areas and undeveloped parts of the city. This paper seeks to define and illustrate a computational technique which takes most of the arbitrariness out of the definition of an urban area, such that the calculation of urban densities is relatively insensitive to the precise definitions used for the urban area of a city. Indeed, the geographic definition of the urban area can be relatively generous, and the computational technique will automatically adjust the density calculations so that they only apply to the populated urbanised area of the city. The adoption of such a technique is imperative when urban densities are being used to compare different cities of Australia for urban planning or socio-political reasons, or where Australian cities are being compared with other cities worldwide. The adoption of the techniques described in this paper will ensure that such comparisons are made on a firmer theoretical and computational foundation. (a).
Abstract