Calibration of cellular automata model with adaptive genetic algorithm for the simulation of urban land-use

Abstract

This paper presents a new method to simulate urban land-use changes by integrating adaptive genetic algorithms (AGA), cellular automata (CA) and GIS. Recently, cellular automata have been increasingly used to simulate land-use changes. The most important and difficult issue in the modeling process is to define and derive transition rules. Traditional logistic method has limitations for deriving the transition rules of CA models based on the assumption that the variables should be independent, which is not true in the actual situations. The limitations of logistic method can be overcome by using GA, which has a good global search capability in the parametric solution space and can find the best parameter combination. In this study, a CA model coupled with improved genetic algorithms is developed as AGA-CA model by Python behind the platform of ArcGIS to cope with the classic GA's disadvantages of easy premature convergence and low global optimal speed. The AGA-CA model is applied to simulate urban land-use of Suzhou city, a rapid urbanization area in the Yangtze River Delta in East China. The performance of the proposed model in simulating urban land-use is compared with that of the logistic-CA model. The results show that the proposed model outperforms the logistic calibrated CA models. Additionally, coupling Python with GIS provides a new way and short-cut for the simulation of land-use changes based on CA model.