IS MUCH OF WHERE GIANT TRENCH EARTHQUAKES OCCUR SIMPLY A SAMPLING BIAS?

Prompted by the great December 2004 Sumatra earthquake, we have been reanalyzing ideas about variations in the maximum size of trench earthquakes and seismic coupling as a function of convergence rate, plate age, and trench sediment thickness. Although it had been proposed that the largest (Mw=9) earthquakes occur when young lithosphere subducts rapidly and where trench sediments are thickest, much of the correlation vanishes using a new set of data. Hence it seems likely that instead of some trench segments but not others being prone to Mw=9 events, these apparent differences may simply reflect the short earthquake history sampled. This possibility is suggested by the variability in rupture mode along trench segments. For example, the plate convergence rate and slip in the Mw 9.6 1960 Chilean earthquake require that either the characteristic Chilean subduction earthquake is smaller than the 1960 event, the average recurrence interval is greater than observed in the last 400 years, or both. We explore this possibility using numerical simulations of Peru-Chile trench seismicity in which the spatial variation along strike results simply from magnitude 8 earthquakes being 10 times more frequent than magnitude 9s.