Prediction of strong events based on the solution of an inverse problem in cellular automaton models
This work states an inverse problem where key parameters are to be recovered in a certain cellular automaton model. Similar to seismicity, the model generates a catalog of events belonging to different energy levels. A method is suggested to reconstruct model parameters from the catalog and to build an algorithm for predicting strong events. For the model imitating a simple linear fault, this approach allows to successfully predict 80\% of strong events with alarm time of about 1% and obtain a slight error in determining the epicenter. The same approach has been modified and tested with simple block models. About 80% of strong events were predicted within the alarm time of about 3% and small errors were obtained in determining epicenters.