Modeling urban evolution by identifying spatiotemporal patterns and applying methods of artificial intelligence.Case study: Athens, Greece.

While during the past decades, urban areas experience constant slow population growth, the spatial patterns they form, by means of their limits and borders, are rapidly changing in a complex way. Furthermore, urban areas continue to expand to the expense of "rural” intensifying urban sprawl. The main aim of this paper is the definition of the evolution of urban areas and more specifically, the specification of an urban model, which deals simultaneously with the modification of population and building use patterns. Classical theories define city geographic border, with the Aristotelian division of 0 or 1 and are called fiat geographic boundaries. But the edge of a city and the urbanization "degree" is something not easily distinguishable. Actually, the line that city ends and rural starts is vague. In this respect a synthetic spatio - temporal methodology is described which, through the adaptation of different computational methods aims to assist planners and decision makers to gain an insight in urban - rural transition. Fuzzy Logic and Neural Networks are recruited to provide a precise image of spatial entities, further exploited in a twofold way. First for analysis and interpretation of up - to - date urban evolution and second, for the formulation of a robust spatial simulation model, the theoretical background of which is that the spatial contiguity between members of the same or different groups is one of the key factors in their evolution. The paper finally presents the results of the model application in the prefecture of Attica in Greece, unveiling the role of the Athens Metropolitan Area to its current and future evolution, by illustrating maps of urban growth dynamics.