Spatial-temporal specific neighbourhood rules for cellular automata land-use modelling

This paper explores the necessity of using spatial-temporal specific neighbourhood rules for simulation of urban dynamics at a regional level with the help of a Cellular Automata (CA) land-use model (Environment Explorer). Moreover, it explores a method for formulating these spatial-temporal specific neighbourhood rules. Therein, spatial metrics originating from the field of landscape ecology prove to be very useful, in particular the so-called enrichment factor. Analysis of historical data with the help of these spatial metrics revealed evidence of substantial differences in land-use developments between time periods and regions, for example, differences that should be incorporated in land-use models if applied on a more detailed regional scale. In a case study, micro-scale land-use data incorporated in a CA-model were used to analyse and simulate the urban morphological changes over two time periods in four urban regions in the Netherlands. In this study, regional and time-specific neighbourhood rules performed significantly better than a set of generic neighbourhood rules. In contrast, within current practice, most CA land-use modelling makes use of only one uniform set of neighbourhood rules for performing both large area and regional scale land-use simulations. This study questions this common practice and puts it into perspective. Moreover, it indicates some restrictions in the proposed regionalization of the neighbourhood rules; one which is particularly noteworthy is its time-specific dependency of local developments and circumstances.