Mitigation of Bias in Inversion of Complex Earthquake without Prior Information of Detailed Fault Geometry

Rupture process of earthquake derived from geophysical observations is important information to understand nature of earthquake and assess seismic hazard. Finite fault inversion is a commonly applied method to construct seismic source model. In conventional inversion, fault is approximated by a simple fault surface even if rupture of real earthquake should propagate along non-planar complex fault. In the conventional inversion, complex rupture kinematics is approximated by limited model parameters that only represent slip on a simple fault surface. This over simplification may cause biased and hence misleading solution. MW 7.7 left-lateral strike-slip earthquake occurred in southwestern Pakistan on 2013-09-24 might be one of exemplar event to demonstrate the bias. For this earthquake, northeastward rupture propagation was suggested by a finite fault inversion of teleseismic body and long period surface waves with a single planer fault (USGS). However, surface displacement field measured from cross-correlation of optical satellite images and back-projection imaging revealed that rupture was unilaterally propagated toward southwest on a non-planer fault (Avouac et.al., 2014). To mitigate the bias, more flexible source parameterization should be employed. We extended multi-time window finite fault method to represent rupture kinematics on a complex fault. Each spatio-temporal knot has five degrees of freedom and is able to represent arbitrary strike, dip, rake, moment release rate and CLVD component. Detailed fault geometry for a source fault is not required in our method. The method considers data covariance matrix with uncertainty of Green's function (Yagi and Fukahata, 2011) to obtain stable solution. Preliminary results show southwestward rupture propagation and focal mechanism change that is consistent with fault trace. The result suggests usefulness of the flexible source parameterization for inversion of complex events.