The impact of neighbourhood size on the accuracy of cellular automata-based urban modelling
journal contributionposted on 27.09.2016 by Liu, Yan
Any type of content formally published in an academic journal, usually following a peer-review process.
Cellular automata are discrete dynamic models in which behaviour is specified in terms of local relations. This technique has recently been advantageously applied to modelling of the urban development process. However, the behaviour of the model is affected by spatial scale, including cell size and neighbourhood extent. Therefore, it is important to examine the impacts of various neighbourhood scales on the model’s behaviour and outcome. In this paper we configured a cellular automata model of urban growth in Sydney, Australia, using three different neighbourhood scales: a small neighbourhood of 1.5 cells radius, a moderate neighbourhood of 2.5 cells radius and a large neighbourhood of 3.5 cells radius, all with a fixed cell size of 250 metres. The moderate neighbourhood scale of 2.5 cells radius was found to best reflect those local mechanisms that have the most direct impact on urban development in Sydney. Hence this paper provides a useful reference in the search for a neighbourhood size that is suitable for cellular automata-based modelling of the processes of urban development.