Partial validation of cellular automata based model simulations of urban growth: an approach to assessing factor influence using spatial methods

Abstract

Cellular Automata (CA) based models have a high aptitude to reproduce the characteristics of urban processes and are useful to explore future scenarios. However, validation of their results poses a major challenge due to the absence of real future data with which to compare them. A partial validation applied to a CA-based model for the Madrid Region (Spain) is presented as a proposal for determining the influence of given factors on the results and testing their spatial variability. Several simulations of the model were computed by different combinations of factors, and results were compared using flexible map comparison methods in order to study spatial pattern matches and similarities between them. Main and total effects of these factors were calculated for each method, by applying a simplified Global Sensitivity Analysis approach. Frequency maps showing the most frequent cells with changed land use in the results were generated.