The investigators are preparing, publishing, and testing earthquake forecasts in order to resolve some basic questions about shallow seismicity, and improve knowledge of global seismic hazard. Each forecast describes a large region (of global scale), or even the entire Earth, and consists of a digital map (or set of maps) of earthquake probabilities for particular minimum/threshold magnitude(s) and for particular period(s) of time. The emphasis on large areas and moderate threshold magnitudes means that large numbers of actual earthquakes can be compared with each forecast, and relative success (compared to a null hypothesis or simpler model) can be evaluated quantitatively. This work builds on two different forecasting methods, each based on an elementary insight about seismicity: Earthquakes cluster in space and time (ETES model); and earthquakes occur at locations of fault slip and other anelastic strain (SHIFT model). In this project, each model is first undergoing refinement based on guidance from the other. Short-term earthquake clustering and triggering are being studied in five tectonic zones based on a detailed kinematic model. Also, the assumed proportionality of long-term seismicity to tectonic rate (except on mid-ocean spreading rises) is being tested using statistics that take account of clustering. The next step is to merge the two forecast models, testing several different kinds of possible combination, to create time-dependent earthquake forecasts with a tectonic basis. Finally, the investigators will determine the time-scales over which the forecast method should shift from primarily clustering-based to primarily tectonics-based. (This time is likely to be controlled by the local rate of tectonic stress accumulation.) Each hypothesis about general controls on seismicity (and/or specific parameter values) is initially evaluated by retrospective testing against existing catalogs. Hypotheses which pass these tests are further investigated in formal prospective tests administered by the Collaboratory for the Study of Earthquake Predictability. This project is creating a new combined forecasting algorithm for consideration by the Working Group on California Earthquake Probabilities and other similar groups, which provides a "seamless" transition from 1-year and 5-year forecasts to 50-year and 500-year forecasts, meeting a variety of social needs. The potential downstream social benefits may be large if more accurate earthquake forecasts can be quickly implemented on a global basis. The project is also attempting to resolve one of the most fundamental questions about seismicity: Whether low-seismicity plate boundaries are permanent, or temporary? This will guide the next generation of hypotheses about what controls seismic versus aseismic slip on cool, frictional faults.
