Source characteristics of major large deep-focus earthquakes based on finite fault inversions

We apply the finite fault inversion algorithm to systematically study the slip histories of global Mw > 7.0 deep-focus earthquakes. The finite fault inversion algorithm is based on waveform inversions in the wavelet domain, and inverts the fault slip, rake angle, rupture initiation time and slip rate function using a simulated annealing approach. Inversion results for an example earthquake, 3 March 2002 Mw=7.3 deep-focus event (depth = 196 km in GCMT solutions) in Hindu-Kush region, indicate that it is feasible to constrain the rupture process of large deep earthquakes through finite fault inversions. In the example, ninety teleseismic P and SH waveforms downloaded from the IRIS Data Management Center are used in the inversion. The fault planes are constrained based on the nodal planes of the GCMT solution. They have a size of 120 km × 120 km, which is further divided into 900 4km × 4km sub-faults. The high angle nodal plane, which has a strike of 108° and a dip of 67°, is chosen as the preferred rupture plane, because it fits the data better than its conjugated plane. The rupture propagated bilaterally with major slip occurring at about 25 km southeast of the hypocenter. The inverted seismic moment is 1.27e+27 dyne×cm/s, 95 percent of which occurred within the first 20 s. The peak slip amplitude is 2.3 m. The average rupture velocity is 2.2 km/s, about 50% of local shear wave velocity. We will present slip histories of major large deep earthquakes obtained through this procedure.