Modelling Urban Expansion: A Multiple Urban-Densities Approach

Most existing spatio-temporal urban expansion models consider urban land-use as a binary process, through the identification of urban versus non-urban areas. The main aim of this study is to analyze and model the expansion of multiple urban densities in Wallonia, Belgium. To this end, this study employs a multinomial logistic regression model that enables to visualize the consequence of different urban densities expansion. Cadastral datasets of years 2000 and 2010 are used to set four urban classes (non-urban, low-density, medium-density and high-density urban). Besides, several socio-economic, geographic and political driving forces dealing with urban development were operationalized to create maps of urban expansion probability for each urban density class. These probability maps are then utilized to predict future urban expansions for years 2020 and 2030. The model is validated using relative operating characteristic method for different urban classes. Our results suggest that different urban densities expansions are mainly linked to zoning status, neighboring areas that are urban and accessibility. Most importantly, this study highlights that the contribution of different driving forces to urban expansion process varies along with urban density. Keywordsmultinomial logistic regression; urban expansion; urban densities; driving forces.