Supercycles and Synchronization Signatures in Synthetic Seismic Sequences

Abstract:

Synchronization is a key concept in nonlinear dynamics. Owing to the paucity and uncertainty of paleoseismic data, we can’t say much about the synchronization of large earthquakes in complex fault systems such as the San Andreas, but we know that substantial elastic strain has accumulated in the southern part of this system since its last major earthquake in 1857. One question is whether the distribution of long open intervals is consistent with the recently published Third Uniform California Earthquake Rupture Forecast (UCERF3, Field et al., 2015), which assumes that time-dependent event probabilities can be modeled by Reid renewal processes correlated across faults only by co-rupture. In a UCERF3 world, the probability of observing an open-interval distribution as extreme as the present day would be low, less than 1% according to one estimate (Jackson, 2015). An alternate hypothesis is that we are in a period of low overall energy release; i.e., near the minimum of a “seismic supercycle.” UCERF3 does not explicitly model supercycles, but they emerge from long runs of physics-based rupture simulators, such as the RSQSim model of Dieterich & Richards-Dinger (2010) and the ALLCAL model of Ward (2008). In these models, the synchronization of large events on different fault sections leads to variations in seismic energy release of ± 50% on time scales of about 200 years. Spectral analysis of a million-year RSQSim catalog shows synchronization harmonics with a fundamental period of 200 years and a corresponding depletion at longer event periods. This synchronization signature is absent in UCERF3 and randomized versions of the RSQSim catalog. We further investigate synchronization and its time dependence using two-dimensional “recurrence plots” (Eckmann et al., 1987) to map the temporal recurrence of proximate RSQSim states. We use the results to speculate on the hazard implications of the supercycle hypothesis.