Estimating the earthquake source mechanism using full waveform inversion

Abstract

The system of earthquake and tsunami early warning system has been realized in some countries to step forward towards a practical use for disaster mitigation purposes. The success of early warning of geohazards depends on the accuracy and the promptness of information issued to the public at the occurrence of such geohazards. Earthquake source mechanism is one of the most indispensable data to be informed of as precisely as possible for disaster mitigation. The centroid-moment tensor (CMT) solution, that represents the stress tensor centered at the hypocenter at the occurrence moment, could be provided as an instantly available data for the focal mechanism of an earthquake, for example. It is, however, necessary to consider the location of a fault plane and the time sequence of fault propagation on the plane, which could not be directly estimated from the CMT solution because of the necessity of aftershock observation to identify the fault shapes and locations. We, therefore, try to deploy an approach using full waveform inversion scheme (FWI) to specify the shapes and locations of the fault system immediately without the aftershock observation that would cost several hours after the earthquake. As a first step in this study, we investigate the applicability of FWI evaluating three fault parameters (strike, dip, and rake) assuming a known source time function for a fault composed of a single segment. We conducted some numerical experiments in order to evaluate the effect of each parameter to the L2-norm error function which is often used in FWI. Our numerical results indicate that fault parameters can be estimated by FWI with sufficient accuracy.