Revisiting the Tidal Activation of Seismicity with a Damage Mechanics and Friction Point of View

University Of Southern California, Los Angeles CA

Investigators

Abstract

If earthquake nucleation were strictly a stress-threshold effect, then one would expect a correlation between times of peak tidal stress and regional seismicity. Failure of statistically rigorous tests to find such a relationship places important constraints on the nucleation process. Explanations of this failure to observe correlation generally involve an accelerating self-driven nucleation process that occurs on a time scale longer than the diurnal tide thereby destroying the expected synchronization. Although a general correlation between tides and earthquakes is not observed, it has recently been reported that a correlation of seismicity with tides is observed to develop during the few years before an earthquake in a finite sized region that scales with the magnitude. This emergent correlation has been interpreted as a consequence of the "critical point model" for regional seismicity. According to the critical point model, an earthquake, once nucleated, can jump barriers to grow large only when the crust is in a "critical state" that is characterized by long-range stress correlation. The rational is that the crust is more sensitive to tidal stress perturbations when it is near the critical state. If true, the emergent correlation of seismicity with tides offers another tool (in addition to accelerating seismic moment) for monitoring the approach of a fault network to the critical state, and the attendant possibility of a large earthquake. A preliminary attempt to reproduce the reported observations of emergent correlation between diurnal tides and seismicity before large earthquakes has not been successful. After reviewing published theoretical arguments in support of emergent diurnal correlation, there is no reason why, even in the critical state, it should not be destroyed by self-driven delayed nucleation. However, review of the theory leads to formulation of a new hypothesis for the tidal activation of seismicity: tidal activation occurs only when the sum of tectonic plus tidal stress exceeds the prior maximum value (analogous to the Kaiser effect in acoustic emissions). In this case the 15-day beat between solar and lunar tides predicts a strong 15-day correlation. The delayed nucleation that is hypothesized to destroy the diurnal correlation may not have as large an effect on the new longer cycle. This project is developing suitable observational techniques to optimize detection of this 15-day quasiperiodic signal and searching for it before a suit of large earthquakes in a variety of tectonic settings. also developing a suitable null hypothesis and tests for the statistical significance of any signals we find. Even if the 15-day signal is not observed, a statistically significant negative result places important constraints on the physics of earthquake nucleation. The effects of cyclic perturbations on a uniformly driven system as it approaches a critical state using a cellular automaton are being investigated. The Olami-Feder-Christensen non-conservative automaton develops long-range stress correlations leading to a critical state. Rate- and state-dependent friction is added to the automaton's failure criterion to produce self-driven delayed nucleation. Although this automaton is not realistic in its simulation of long-range elastic interactions between faults, it provides a suitable testbed for the two hypotheses being explored: 1) whether tidal activation becomes stronger near the critical state, and 2) whether delayed nucleation associated with rate and state-dependent friction is expected to attenuate the 15-day correlation to the same extent as the diurnal correlation.