A Probabilistic Cellular Automaton Approach for a Spatiotemporal Seismic Activity Pattern
Takayuki Hirata, Masajiro Imoto · Zisin (Journal of the Seismological Society of Japan 2nd ser ) · 1997
A probabilistic cellular automaton was used to reproduce spatiotemporal seismicity patterns. Based on the conditional probability distribution obtained by analyzing the coarse-grained seismic activity pattern of the observational data set of Kanto area, the rule of the probabilistic cellular automata was determined. The simulation was carried out by a Monte Carlo method. The mutual information of the model was calculated for our probabilistic cellular automata model for the spatiotemporal seismic activity to estimate the degree of information transfer from the past state to the future state: the mutual information is 0.173bit in our model. A correlation function was calculated for both the simulation patterns and the observational seismicity patterns. As a result, although the interaction considering the nearest neighbor regions and only one time step is not sufficient enough to reproduce the observational seismicity pattern in our naive approach, we succeeded by a heuristic manner in finding out the rule of probabilistic cellular automata only considering the nearest neighbor interaction and one time step that well reproduces the observational seismicity pattern.