Seismological evidence for the brittle-ductile interaction hypothesis on earthquake loading

Abstract

We extended the analyses of temporal variation of coda Q⁻¹ and seismicity by Jin and Aki (1989, 1993) for central and southern California to year 2003. We use the relative frequency N(M _c ) of earthquakes in a certain magnitude range around M _c characteristic to a seismic region to quantify the seismicity. The cross-correlation function between coda Q⁻¹ and N(M _c ) is calculated using a 10-year moving time window. The correlation coefficient for the entire period of about 60 years is peaked at the zero-time shift with the value close to 0.8 for both regions. We found, however, the simultaneous correlation is disturbed before major earthquakes. The disturbance is, consistently, a delay in the change of coda Q⁻¹ relative to that of N(M _c ) before the occurrence of a major earthquake. We attribute the temporal change in coda Q⁻¹ to fractures in the ductile part of the lithosphere and that in N(M _c ) to the response of the brittle part to the ductile fracture. We believe that M _c characteristic to a seismic region is originated from a characteristic size of fractures in the ductile zone of the lithosphere. The observed delay of coda Q⁻¹ change relative to N(M _c ) before a major earthquake can be explained simply by the strain energy stored in the brittle part of lithosphere reaching a saturation limit and starting to flow back to the ductile part.