Validating Characterisations of Urban Form Using Tests of Association for Spatially Autocorrelated Data

The goal of this paper is to validate two spatially explicit conceptions of urban form through quantitative comparison with census population and housing data. The first measure of urban form is a spatially distributed representation of the intensity of the built environment based upon building values measured in dollars. The second measure is calculated from the first using the Moran’s I statistic, and offers both a local and global (a summary index of the entire study area) assessment of urban form. Validation is the process of determining whether these indices are sufficiently representative of the distribution of population and housing in the study area. The methods used follow the procedure for assessing correlation between two spatial processes, as articulated by Dutilleul et al (1993). Results show a consistently strong positive correlation between the two modelled conceptions of the residential built environment and the reference population and housing data. The importance of quantified measures of urban form is ultimately about the use of these indices to develop insights about the form and function of the built environment (Longely and Mesev (2000). The measures tested here are deemed suitable for quantitative research linking the pattern of the built environment with economic, environmental, and social processes