Cellular Automata Model for the Urban Growth of the Metropolitan Area of Concepcion (CAMAC)

Cellular Automata (CA) have been intensively used in the last decades to simulate urban growth in cities. This is based on the fact that the suitability of a given location for built-up areas is highly dependent on the characteristics of the neighbouring locations, particularly on whether they are already urbanised or not. CA and Geographic Information Systems (GIS) have recently become an instrument for modelling the temporal dynamics of urban areas. Nevertheless, they have not yet been extensively used in Latin American cities. The prediction approaches for urban growth in cities are diverse. Some use probabilistic CA, while others define appropriate deterministic rules by hand, or use statistical tools to determine the CA transition rule, etc. Calibration parameters usually are the neighbouring radius, the random degree, or the influence of a given geographical feature. But up to our knowledge, the influence of the time scale, i.e., the number of years that a single CA iteration represents, has not yet been studied. Since accumulative effects of the CA dynamics are relevant in the CA theory, we think it important to explore the CA effect on urban growth models in different time scales. In the present work, we adopt the approach of Aguilera (2006), who runs a Geographical Logistic Regression to estimate the probability of a cell to become urbanised, and then it will urbanise only the most probable cells up to fill the real amount of surface growth. We repeat this methodology iteratively, in order to look at the CA effects. We apply it to predict the urban growth the Metropolitan Area of Concepcion, Chile (MAC), from 2000 to 2009, and we compare the results.