Dynamic transition rules for geographical cellular automata

There is rapid development of CA models for simulation of land use patterns and urban systems recently. Traditional models use the uniform transition rules in large areas or at long times. These models have limitations because they use the same factors and weights in different places or times while these factors and weights vary spatially. This paper presents the method of using dynamic transition rules for producing more reliable urban simulation. The advantages of dynamic transition rules are apparent in this paper. First, it uses different factors and weights in different regions which are consistent with actual urban systems. Second, the main factors can be obtained from the dynamic transition rules in different regions which may produce different urban development patterns. Based on the stepwise logistical regression model, dynamic transition rules were obtained for simulating the evolution of the urban systems in five towns of the Dongguan city from 1988 to 2004. Compared with the general CA model, the simulation precision is improved by using this method. From the factors and weights of dynamic transition rules, three kinds of development patterns were obtained: center development pattern, road development pattern and center-road development pattern.