The Recent Rate of Great Earthquakes: Global Clustering or Random Variability?
Abstract
Five M≥8.5 earthquakes have occurred since 2004 compared to only 11 events from 1900 to 2004. This observation has led some to suggest that we have entered into a period of heightened global earthquake activity (Bufe and Perkins, BSSA, 2005; Brodsky, AGU Fall Meeting, 2009; Ammon et al., SRL, 2010). I test the null hypothesis that the recent activity represents a random variation in a Poisson process plus localized aftershock activity using M≥7 events from 1900 to the present and M≥6 events from 1992 onwards from the PAGER and PDE catalogs. Aftershocks are defined by the Gardner and Knopoff (BSSA, 1974) algorithm modified for use with large earthquakes. I test for global triggering after M≥8, 8.5 or 9 events by comparing the rate of earthquakes in 0.25 to 5 year windows after the great events to the rate for the rest of the catalog. With 95% confidence, a significantly higher rate after the great earthquakes is only found for some of the shorter time windows when the aftershocks are not removed from the data. For all other combinations of parameters, and whenever aftershocks are removed, no significant variations in rate are found. Next, I test for general departures from Poisson behavior by comparing the distribution and autocorrelation of interevent times against the predictions from a Poisson model for M≥ 6, 6.5, 7, 7.5, 8, and 8.5. The presence of aftershocks can be detected for the M≥ 6, 6.5, 7 and M≥8.5 catalogs. For M≥7.5 and M≥8, and for all catalogs when aftershocks are removed, the tests cannot reject the Poisson model. Finally, I test whether the proposed clustering of moment in windows from 1952-1964 and 2004-2011 is significant by comparing a search for the maximum percentage of seismic moment released in two windows that have a combined duration W with the same search in synthetic Poisson catalogs. No significant clustering is found for any value of W. The search was repeated for Benioff strain (square root of moment) and, once again, no significant cluster was found. Thus, the temporal distribution of large, global earthquakes is well described by a Poisson process plus localized aftershock activity, but without global triggering, and the recent spate of great earthquakes can be explained as a random fluctuation without predictive power for the future. While the probability of future large mainshocks has not statistically significantly increased, neither has it decreased and, except in active aftershock zones, global hazards estimates should use the longest possible catalog rather than focusing on the recent past.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFMNH31A1525M
- Keywords:
-
- 7223 SEISMOLOGY / Earthquake interaction;
- forecasting;
- and prediction;
- 4302 NATURAL HAZARDS / Geological;
- 4313 NATURAL HAZARDS / Extreme events;
- 4318 NATURAL HAZARDS / Statistical analysis