A probabilistic seismic hazard model based on cellular automata and information theory

International audienceWe try to obtain a spatio-temporal model of earthquakes occurrence based on Information Theory and Cellular Automata (CA). The CA supply useful models for many investigations in natural sciences; here, it have been used to establish temporal relations between the seismic events occurring in neighbouring parts of the crust. The catalogue used is divided into time intervals and the region into cells, which are declared active or inactive by means of a certain energy release criterion (four criteria have been tested). A pattern of active and inactive cells which evolves over time is given. A stochastic CA is constructed with the patterns to simulate their spatio-temporal evolution. The interaction between the cells is represented by the neighbourhood (2-D and 3-D models have been tried). The best model is chosen by maximizing the mutual information between the past and the future states. Finally, a Probabilistic Seismic Hazard Map is drawn up for the different energy releases. The method has been applied to the Iberian Peninsula catalogue from 1970 to 2001. For 2-D, the best neighbourhood has been the Moore's one of radius 1; the von Neumann's 3-D also gives hazard maps and takes into account the depth of the events. Gutenberg-Richter's law and Hurst's analysis have been obtained for the data as a test of the catalogue. Our results are consistent with previous studies both of seismic hazard and stress conditions in the zone, and with the seismicity occurred after 2001