Abstract：Cities are dynamic intricate systems. The two characters of cellular automata(CA) should be utilized to resolve urban land-use problem. The first is that they are intrinsically spatial; and the second is that they can generate very complex forms by means of very simple rules, namely the local'or individualistic actions lead to global patterns with order at higher level. In this paper, the author employes CA-Urban model which is coupled with GIS and CA to simulate the land-use growth of Nanjing city in China in the past time. CA-Urban is a hierarchical model, in which macro and micro models are integrated; spatial model and socio-economical model are integrated. The land-use change from 1978 to 1990 was used to define the model parameters, and the land-use of 1990 was regarded as seeds. The roads were considered as an important influencing factor. The control background was defined by terrain, land use situation, city planning etc. The simulation results indicated that urban Nanjing has a trend to expand eastward and southward. Using the CA-Urban, the author evaluated the development situation for the three Development Zones:Nanjing New & High Technical Industrial Development Zone, Nanjing Economic & Technical Development Zone and Jiangning Economic Technical Development Zone; Some schemes as to how to develop the areas along Yangtse River is offered. Of course the model simulation is only a trend of the expansion of Nanjing. This is a characteristic of a complicated system. In the practice, the author found that the CA-Urban had some limits to prediction. The global rules and local rules conflicted. For example, globally, the housing land-use in Nanjing city developed more quickly than the industrial land-use, so we give higher weight to housing land-use than to industrial land-use; locally, in some industrial development zones, the industrial land-use should expand more than other land-use. Considering city as a organism, we should employ CA, nerve network and genetic arithmetic to simulate the urban dynamic process supported by GIS and virtual reality emulate technology. They can revise rules in real time and dynamic and achieve interoperate between transition rules and virtual cities, build up rules-storeroom analogous to expert system at last. That consummate transition rules can advance the urban dynamic models as well.
