DO EPISODIC TREMOR AND SLIP (ETS) EVENTS AFFECT SEISMICITY IN THE NORTHERN CASCADIA SUBDUCTION ZONE?

Episodic tremor and slip (ETS) events deform the surrounding crust, albeit at low strain rates. Slip events elsewhere appear to directly trigger earthquakes, or correlate with higher seismicity rates at later times. Could the deformation associated with ETS events influence seismicity rates in the northern Cascadia region? Alternatively, could the ETS events be one manifestation of a process that also influences seismicity?

To test this hypothesis, I analyzed the seismicity rate from the northern Puget Lowland to see if there is a 14-month periodicity that matches the 14-month ETS cycle. The northern Puget Lowland region has a large number of earthquakes occurring in a volume of crust that modeling shows is deforming during the ETS events. I selected earthquakes from the 1976 to 2006 catalog that were shallower than 35 km in a 2° by 2° region, and converted the catalog to a time series consisting of the number of earthquakes per day. The analysis was carried out several times using different minimum magnitudes. Autocorrelating these time series consistently showed distinct peaks at lag times of 416 to 430 days, suggesting a relationship between crustal seismicity and the 447±37 day ETS cycle.

The sharpness of the autocorrelation peak suggests that it is due primarily to the alignment of prominent aftershock sequences that appear to have a characteristic temporal spacing of about 14 months. A broad autocorrelation peak between 11 and 14 months remains even after removing the large aftershock sequences by limiting the number of earthquakes per day in the catalog. Surprisingly, the peak in the seismicity rate does not occur during the ETS events, but rather between events. This phase difference between ETS events and peak seismicity rates suggests that direct triggering is not involved. There is also a strong annual cycle, with about 50% more earthquakes occurring in the early summer than in the early winter. Combining the 12 and 14-month cycles results in an interference pattern with a 6.5-year period, and this interference pattern predicts peaks and troughs in the seismicity rate that superficially resemble those in the earthquake catalog. The presence of these cycles suggests that seismic hazard varies temporally, and that this variation is at least partly predictable.