Aseismic-Seismogenic Coupling For A Long Strike-Slip Fault

Abstract

SUMMARY A 2-D mechanical model for the elastic coupling between the upper seismogenic and lower aseismic parts of a long strike-slip fault is presented. The model spans the quasistatic and dynamic stages of the earthquake cycle by predicting the dynamic dislocation at the fault trace in terms of the quasistatic build-up of average stress drop on the seismogenic zone Aa = ToH/W (where W is the width of the seismogenic zone, H is the thickness of the crust and To is the time-dependent remote stress). The stress drop Aa(z) is predicted to be nearly uniform with depth z, except near the base of the seismogenic zone where a stress concentration develops. If coseismic slip is constrained to the crust then faults which are at least four times longer in the strike direction, relative to the coseismic depth, attain a mean dislocation which is independent of fault length. The rise time may be as small as 6 s in the absence of coseismic movement beneath the seismogenic fault and increases to about 30 s for coseismic slip of the entire crust. Rapid post-seismic creep on the lower aseismic fault is predicted by the model, since the dislocation at the surface of the Earth exceeds the pre-seismic dislocation at the base of the plate. The application of the model to contemporary small earthquakes near the southern segment of the San Andreas fault leads to an interpretation that the 1857 seismic gap is locked at a depth of about 7-8 km by asperities which become unlocked by a maximum energy release rate of the order of lo7 J m-’. This estimate for the maximum energy release rate is equal to that of Tse, Dmowska & Rice (1985) and comparable to the estimate of Li (1987) of 0.6 to 3.8 x lo7 J m-2. For a typicul San Andreas great earthquake with a surface dislocation of 6m, an average recurrence time of 150yr, a relative plate velocity of 4cmyr-’, and a stress rate of 0.1 bar yr-’, the model predicts for W/H = 8/40 and 12/30, respectively, a depth of coseismic slip of 12 and 30 km, a stress drop of 75 and 38 bars and a critical energy release rate of 0.28 and 0.17 x lo7 J mp2. These lower energy release rates are more consistent with the estimate of Rudnicki (1980) of 0.38 x lo7 J m-’ and imply that a great earthquake may occur before the energy release rate reaches the maximum value of lo7 J m-2.