Lack of Dependence of Dynamic Triggering on the Timing within the Seismic Cycle

Numerical models predict that dynamic triggering of earthquakes is more likely when faults are close to failure (e.g. late in their earthquake cycle), and laboratory experiments have supported this hypothesis. We attempted to test this idea by analysing data on three adjacent transform faults of the East Pacific Rise which have a relatively well defined, quasiperiodic seismic cycle with a median repeat time of 5 years. Moreover, the Gofar, Discovery and Quebrada transform faults share several seismicity properties with continental geothermal areas, including high geothermal gradients, high seismicity rates, and frequent earthquake swarms, that suggest they may be prone to dynamic triggering. We analyze an earthquake catalog of over 100,000 events recorded in 2008 by a network of 38 Ocean Bottom Seismometers. We extract Mw>6.3 mainshocks from the Global CMT catalog, and perform the β test for an array of time intervals covering from 5 hours before to 10 hours after the low-frequency Rayleigh wave arrival. To verify the presence of common seismicity patterns, β plots are also stacked for multiple earthquakes. We observe triggering after the May 12th Wenchuan earthquake. On the Quebrada transform a burst of seismicity starts during the wavetrain; in Gofar there is no response during the wave, but an increase in seismicity (β=5.08) starts about 2 h later; no triggering is visible on the Discovery fault. A Mw=6.0 earthquake ruptured the Gofar transform on September 18th, and triggered seismicity on Discovery: ~60 earthquakes (β=15.3), starting 1h after the wave arrival. We have no data from Quebrada for this period. Other instances of triggering are dubious. Stacked β plots suggest delayed triggering (Δt>1h) in Gofar and Discovery, but the statistical significance of these results is unclear. From a comparison of different fault segments, triggering does not appear to be more common at late stages in the seismic cycle. Instead, the events triggered by the largest dynamic stresses concentrate in the regions between rupture zones. This suggests that changes in rock composition or fluid content may make these areas act as barriers to rupture propagation as well as facilitating dynamic triggering. Using the Rate-and-State seismicity model, we estimate that the effective normal stress where triggering occurs: is extremely low (σ<0.1MPa in Quebrada and σ<0.5MPa on Discovery), implying a nearly lithostatic pore pressure.