1

This research is based on the idea that the simulation of urban growth may follow simple spatial rules, derived from our empirical knowledge, but nevertheless explaining the spatial dynamics resulting from social practice. The cellular automata program SpaCelle is built on a very general paradigm from outside of the field of urban models, that of spatial competition between diverse cellular populations, interacting with their environment. Each cellular individual uses its life force, which varies from its birth to its death, to resist the environmental forces resulting from neighbouring individuals acting in rings growing broader around it. The computer program, based on this general principle (the meta-model) must be input by the user from a knowledge base consisting of the different classes of cellular status, the rules governing the life and death of the individuals and the transition rules. The arrangement of the cells may be imported from an Arc View grid or entered directly. The geographical relevance of this model was tested through an experiment on the evolution rules of the urban space of Rouen over the last fifteen years. The results are very close to the observed situation, which validates the general model and the rule base for the urban area of Rouen.