TEMPORAL CLUSTERING OF MAJOR EARTHQUAKES ON INDIVIDUAL FAULTS VIA POSTSEISMIC RELOADING: A POSSIBLE EXPLANATION FOR NON-UNIFORM REPEAT TIMES IN PALEOSEISMIC DATA

On a single fault segment, paleoseismic evidence from locations such as the Basin and Range [Friedrich et al., JGR, in review] and Dead Sea Transform [Marco et al., JGR, 1996] indicate that the occurrence of major earthquakes in time is often extremely heterogeneous and may exhibit temporal clustering, which we define as the occurrence of multiple event sequences with intra-cluster inter-event times much shorter than the average time between clusters. Many factors may contribute to temporal clustering of major earthquakes. Over multiple event time scales, the role of time-dependent postseismic stress transfer has not previously been considered.

Major earthquakes generate large concentrations of stress in the lithosphere beneath the coseismic fault. As this postseismic stress concentration is dissipated via time-dependent viscoelastic relaxation, the coseismic fault is reloaded at a rate that is initially much higher than the background rate derived from far-field plate motions. Simultaneously, earthquake recurrence intervals are moderated by various sources of system noise, including stress perturbations due to neighboring earthquakes, crustal heterogeneities, and fault evolution. Depending on the relative timing and magnitude of earthquakes in a sequence, therefore, the postseismic stress stored in the lithosphere and available for transfer to the coseismic fault may be greater or less than average. If the amount is greater than normal, postseismic stress transfer can become a significant factor in controlling the time to the next event. In this situation clustering may result.

To investigate these long-term postseismic processes, we develop a spring-dashpost-slider model of time-dependent stress transfer in the earth. With this tool, we gain an understanding of how variations in rheology, fault slip-rate, and system noise affect the distribution of earthquake recurrence intervals. In tectonic environments with a weak lower crust/upper mantle, we find that small random variations in the fault failure criteria generate temporally clustered earthquake sequences. This effect is enhanced as the geologic slip-rate on the fault decreases. By analogy, areas such as the Basin and Range are particularly susceptible to clustering induced by long-term postseismic stress effects.