Self-Organization and Correlations of the San Andreas Fault System through Interactions: Forecasting based on Recreating Great Earthquakes in the Computer

Rundle, J B - University of California, Center for Computational Science, Davis, CA 95618 United States
Rundle, P B - University of California, Center for Computational Science, Davis, CA 95618 United States
Turcotte, D L - University of California, Center for Computational Science, Davis, CA 95618 United States
Yakovlev, G - University of California, Center for Computational Science, Davis, CA 95618 United States
Van Aalsburg, J - University of California, Center for Computational Science, Davis, CA 95618 United States
Donnellan, A - Jet Propulsion Laboratory, Earth and Space Science Division, Pasadena, CA 91125 United States
Tiampo, K F - University of Western Ontario, Department of Earth Science, London, ON N6A 5B7 Canada
Klein, W - Boston University, Department of Physics, Boston, MA 02215 United States

Advanced computer models and simulations, combined with new data sets and data mining techniques, together with ideas about complex nonlinear systems, enable a new approach to the problem of earthquake physics and forecasting. Modern computational technology allows us to construct models such as "Virtual California" that include many of the physical processes known to be important in earthquake dynamics. These include elastic interactions among the faults in the model, driving at the correct plate tectonic rates, and frictional physics on the faults using the physics obtained from laboratory models with parameters consistent with the occurrence of historic earthquakes. An important consequence of the elastic interactions in the model is the appearance of correlations and space-time patterns of occurrence of events. Without the interactions, each fault element would behave independently; with the interactions, cooperative phenomena and patterns are observed. Both the statistics and the patterns can be used in earthquake forecasting. For example, we find that our results for interval times are fit well under most circumstances by the Weibull statistical distribution, and we compute waiting times to future earthquakes based upon our simulation results. A contrasting approach to the same problem has been adopted by the Working Group on California Earthquake Probabilities, who use observational data combined with statistical assumptions to compute probabilities of future earthquakes. In present work, we are emphasizing two lines of research. The first is an overall optimization of the parallel structure of the code, enabling much larger scale simulations. The second is data assimilation, in which we ingest new data into the codes in an organized way, so that we can steer the models to produce an improved forecasting capability. This talk will summarize these ideas and results, along with future priorities and needs.