Characterizing the Triggering Susceptibility of Characteristic Faults

Morgan T. Page, Nicholas J. van der Elst, & Bruce E. Shaw

Submitted August 10, 2017, SCEC Contribution #7477, 2017 SCEC Annual Meeting Poster #017

Statistical averages of earthquake triggering behavior, namely Gutenberg-Richter magnitude scaling, can produce foreshock probabilities that differ by orders of magnitude from fault-specific methods that employ characteristic magnitude distributions. Following a M4.8 earthquake near Bombay Beach in 2009, estimated probabilities for a subsequent M≥7 event varied by ~100-fold depending on whether or not characteristic magnitude distributions were used (Michael, 2012). In the September 2016 swarm, the public advisory included a range of probabilities (a factor of 30) to capture model uncertainty.

The characteristic model may be most easily tested in terms of its foreshock predictions. First of all, due to the spatial reach of aftershock triggering, these tests are not sensitive to the assumed width of the fault zone. Even more importantly, the predictions of the characteristic model with respect to foreshock triggering are dramatic. The UCERF3-ETAS (Field et al., 2017) model, for example, predicts that a potential foreshock near the southern end of the quiet San Andreas fault is 23 times more likely to trigger a M≥7 mainshock than a similar earthquake near the active San Jacinto fault. A model with Gutenberg-Richter scaling would predict the same foreshock probability for these two scenarios.

While the foreshock-related consequences of the characteristic model are sizable and have been used in public earthquake advisories, they have, to our knowledge, never been directly observed. We search for evidence of this effect both in California and global subduction zones. The global dataset allows us to maximize the data power of our tests, and while it is a different tectonic environment than where the characteristic model has been routinely applied to foreshock probabilities (namely, California), the essentials of the problem are the same – we look to see if fault activity rate, normalized by slip rate, has a linear effect on the triggering of large earthquakes.

Key Words
characteristic magnitude distribution, earthquake triggering, statistical seismology

Citation
Page, M. T., van der Elst, N. J., & Shaw, B. E. (2017, 08). Characterizing the Triggering Susceptibility of Characteristic Faults. Poster Presentation at 2017 SCEC Annual Meeting.


Related Projects & Working Groups
Earthquake Forecasting and Predictability (EFP)