Spatial-temporal clustering of large earthquakes on Oceanic Transform Faults

Progress towards a quantitative and predictive understanding of the earthquake behavior can be achieved by improved understanding of earthquake cycles. However, it is hindered by the long repeat times (100s to 1000s of years) of the largest earthquakes on most faults. At fast-spreading oceanic transform faults (OTFs), the typical repeating time ranges from 5-20 years, making them a unique tectonic environment for studying the earthquake cycle.

One striking observation on OTFs is the quasi-periodicity and the spatial-temporal clustering of large earthquakes: same fault segment ruptured repeatedly at a near constant interval and nearby segments ruptured during a short time period. This has been observed on the Gofar and Discovery faults in the East Pacific Rise (EPR). Between 1992 and 2014, five clusters of M6 earthquakes occurred on the Gofar and Discovery fault system with recurrence intervals of 4-6 years. Each cluster consisted of a westward migration of seismicity from the Discovery to Gofar segment within a ~ 2-year period, providing strong evidence for spatial-temporal clustering of large OTFs earthquakes.

Motivated by the observations at EPR, we conducted a global search for more repeating large earthquakes on OTFs, using surface-wave based determinations of the relative separations between earthquake centroids. We found that there are many more cases of repeating earthquakes on OTFs than previously reported.

We also built a numerical model with multiple fault segments in the frame of rate-and-state friction to understand the role of stress transfer on the spatial-temporal clustering of large earthquakes on OTFs. We used the Gofar and Discovery faults as a model. We found that static stress transfer alone can explain the clustering of earthquakes if these fault segments are close (< a few km). Otherwise, viscoelastic or dynamic stress transfer will play a more important role.