Where Have All the Paleoearthquakes Gone?

Abstract:

Many seismic hazard projects use earthquake rates estimated from paleoseismic data as a mainstay for estimating future earthquake probabilities. A prime example is the third Uniform California Earthquake Rupture Forecast (UCERF3), which relies on reported dates of observed paleo-seismic displacements at 31 sites on 13 named faults in California. However, there is a problem: recorded paleoseismic events ceased at about the beginning of the instrumental seismic era. The reported event rates for the ensemble of 31 sites sums to about 0.1 per year. Allowing generously for occurrences of earthquakes that rupture multiple sites simultaneously, the event rate is on the order of 0.04 per year. Yet the most recent paleo-event date is 1910. Such a long open interval would be extremely unlikely for a Poisson process with that rate, which might be expected statewide even if individual faults ruptured in quasi-period events. Quasi-periodic behavior for the whole ensemble would make the discrepancy worse. Possible explanations for the discrepancy include (1) extreme luck, (2) unexplained regional collaboration amongst faults, or (3) mistaken identification of near-surface displacements as evidence of large earthquakes. The first can be rejected with 99% confidence. There is no evidence for the second in the pre-1910 paleoseismic history nor in any theoretical models yet published. The third could explain the observed quiescence because mistaken identity would be prevented by instrumental seismic data.