LOCATION AND CHARACTERIZATION OF CALIFORNIA NON-VOLCANIC TREMOR USING TIME REVERSAL, INVESTIGATION OF DIFFERENT IMAGING FIELDS

In 2008, Gomberg et al. reported that the passage of the Love waves associated with the 2002 Denali earthquake triggered non volcanic tremor (NVT) at seven sites along the San Andreas fault. Studying triggered NVT is important because it might help to map the faults locked zones leading to a clearer picture of a particular regions seismic risk. All of this assumes that we know where tremor originates. However, tremor locations are usually poorly resolved. Applying Time-Reversal (TR) to tremor may be a powerful approach to the location problem. TR exploits the full waveform in order to provide a highly-resolved source location, which is essential in order to establish the connection between the origin of the NVT and the fault. We have demonstrated (Larmat et al., 2006, 2008) that the TR wavefield defines the original source radiation patterns, and so TR potentially provides the means to characterize the physical processes producing NVT.

We report the results of a synthetic study of the application of time reversal for locating the ”Hemet” NVT source reported in Gomberg et al. (2008). This synthetic study is a first step in developing the method for actual data inversion. The subject volume is a segment of the Los Angeles basin that is modeled using the package SPECFEM3D (Komatitsch et al., 2002). Local subsets of the ANZA and CI seismic networks are used to rebroadcast the signal. We demonstrate that an important part of source location and characterization when using realistic 3D Earth models is the use of suitable imaging fields for interrogation of the time reversed broadcast. Low velocity zones act as seismic energy magnets that compete with the time reversal focus used to locate the source. Our study shows that this problem is solved by monitoring the divergence of the time reversed wavefield instead of the field itself when searching an isotropic source. Associating divergence with an isotropic source was intuitive. On the contrary, we demonstrated that the strain is the appropriate imaging field for a moment–tensor source, as predicted by the adjoint methods. Our study further demonstrates that working with different imaging fields provides a deeper understanding of the source and the wavefield propagation.