An analysis of the effect of the stochastic component of urban cellular automata models

Urban cellular automata models have proved useful tools in urban growth prediction because of their simplicity and their ability to reproduce complex emergent dynamics. Complex emergent dynamic systems involve processes that are difficult to predict, in which randomness plays a key role. In view of the fact that randomness is particularly relevant to complex processes, the aim of this paper is to analyze the sensitivity of the results of urban cellular automata models to the different methods used to incorporate the stochastic component in the models. The urban growth patterns obtained using different stochastic components are analyzed and compared using a number of spatial metrics. The results show that the differences observed in the simulated patterns are sufficiently relevant to justify the need for this type of analysis, which allows for the selection of the stochastic component that best suits the dynamics of the area.

Research highlights
► The Monte Carlo method produces more stochastic uncertainty in the models.
► The Monte Carlo method controls better the degree of randomness introduced.
► The stochastic perturbation produces less stochastic uncertainty in the models.
► The stochastic perturbation controls worse the degree of randomness introduced.
► Results may improve scaling the stochastic perturbation with an exponential curve.