Revisiting the Earthquake Interevent Time Distribution and the Poisson Model with the QTM Catalog for Southern California

Erin J. Hightower, & Jean-Philippe Avouac

Published August 15, 2019, SCEC Contribution #9761, 2019 SCEC Annual Meeting Poster #035

Existing probabilistic seismic hazard models often assume a stationary Poisson process. Such a process is often argued to be insufficient for the largest earthquakes, which have a greater impact on crustal stresses and seismicity patterns and may require a distribution with memory, such as the Brownian Passage Time model. Previous analysis of the earthquake interevent time distribution using the declustered NCEDC and SCSN catalogs for northern and southern California, respectively, showed a fat tail on the cumulative distribution of interevent times relative to an exponential distribution, suggesting that earthquakes do not follow a Poisson process. We revisit this analysis and the Poisson assumption using the new Quake Template Matching (QTM) Catalog for southern California (Ross et al., 2019), declustered using the Zaliapin and Ben-Zion (2013) algorithm. By randomly generating multiple sets of synthetic interevent times, each distributed according to an exponential distribution with the same Poisson parameter value as the QTM data, we test whether the observed fat tail is a statistically significant deviation from the Poisson model, given the mean error of the synthetic data. Our test reveals that the tail in the QTM data does not lie significantly outside the mean error of the synthetic data and that there is not enough evidence to reject the Poisson model at the 5% significance level. This holds when including all magnitudes and for only higher magnitude events. We are also testing a full synthetic catalog, with Poissonian event times, Gutenberg-Richter distributed magnitudes, and aftershock sequences generated using the ETAS model, declustered with the Zaliapin and Ben-Zion (2013) algorithm, to determine whether incomplete declustering, and hence an artificial excess of smaller interevent times, could be responsible for the tail seen in other catalogs. As has been observed in other studies, the shape of the interevent time distribution is dependent on the number of aftershocks remaining in the catalog. Non-Poissonian behavior is likely an artefact of incomplete declustering or catalog incompleteness.

Key Words
Interevent Time Distribution, Poisson Model, Aftershock Declustering

Citation
Hightower, E. J., & Avouac, J. (2019, 08). Revisiting the Earthquake Interevent Time Distribution and the Poisson Model with the QTM Catalog for Southern California. Poster Presentation at 2019 SCEC Annual Meeting.

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