We are building numerical models to calculate the history of resolved stress at any point on any fault and the earthquake potential resulting from the stress history. The stress model would include co-seismic static and dynamic stress changes, tectonic stress accumulation (from individual faults where their geometry and slip rate are known) and anelastic relaxation. We are deriving earthquake probabilities from the stress history using extended Epidemic Type Aftershock Sequence (ETAS) models. Also, we will use a more physically-based model, the rate-and-state seismicity model, which provides a relation between stress history and seismic activity.
We are using this stress model and high-quality earthquake data to address several questions about earthquake triggering. These include whether static or dynamic stress is more influential, whether small (more frequent) earthquakes dominate triggering, whether rate-and-state friction explains earthquake rates in detail, and whether detailed stress calculations significantly improve on simpler ETAS models. We are using the Harvard CMT catalog for earthquakes with magnitudes over 5.5 and regional catalogs in the western USA, Japan, China, New Zealand, Taiwan, and Greece, among other areas.
Based on the answers we find, we will construct an earthquake forecast model which combines the advantages of ETAS type clustering models and explicit stress calculations. We will then use the model to make testable forecasts first for magnitude 5.5 and larger quakes around the plate boundaries, then for smaller events in specific regions.
