A vector-based cellular automata model to allow changes of polygon shape

In the last few years, cellular automata (CA) have been increasingly used to simulate geographic phenomena due to their computational simplicity and their explicit representation of space. However, recent researches have demonstrated that the classical raster-based CA models are sensitive to spatial scale. In attempts to overcome this problem, this paper presents a novel vector-based CA model, called VecGCA that defines space as a collection of geographic entities of different shapes and sizes that correspond to real-word entities. The model was tested with real data to simulate land-use changes in an agroforested area in southern Quebec, Canada. Its performance was assessed through visual and quantitative analyses of the shape and distribution of the spatial patterns that were generated when compared to the patterns produced by a conventional raster-based CA. The results obtained show that both models generate a similar trend in land-use change, but the landscape is considerably less fragmented with the VecGCA model compared to the raster-based CA model.