5:15 PM - 7:15 PM
[SSS09-P07] Centroid Moment Tensor analysis with a fully nonlinear waveform inversion using the Neighbourhood Algorithm: Influence of radial anisotropy
Keywords:Centroid moment tensor, Moment tensor analysis, Radial anisotropy, Nonlinear inversion
Centroid Moment Tensor (CMT) analyses have been routinely performed for several decades to retrieve point source parameters through seismic waveform inversions. CMT catalogs have been provided by various institutions and agencies, such as the Global CMT (GCMT) catalog for global seismic events (Dziewonski et al., 1981; Ekström et al., 2012), and the JMA (Japan Meteorological Agency) catalog (Nakamura et al., 2003), and the F-net (by the National Research Institute for Earth Science and Disaster Prevention: NIED) catalog (Fukuyama et al., 1998) for the regional seismic events around the Japanese islands. These catalogs have provided fundamental information for seismological studies. Generally, it is known that discrepancies in the reference earth structure may lead to systematic errors in the estimated CMT solutions (e.g., Sawade et al., 2022). Since these conventional CMT analyses employ a 1-D Earth model (or with phase-shift corrections based on global phase velocity maps for GCMT), the resultant CMT may be biased to some extent in regions with strongly heterogeneous region like the Japanese islands. To overcome such a problem, the 3-D Green’s functions have recently been used to compute synthetic seismograms for the CMT inversions. Although the full 3-D treatment of seismic wavefield allows us to incorporate a complex 3-D earth model, it is computationally demanding, limiting its application for a routine analysis. In addition, the influence of seismic anisotropy on the source estimation has been pointed out that anisotropic structure near the source may introduce apparent non-double-couple (NDC) components even in a case of pure shear faulting (e.g., Menke & Russell, 2020). Moreover, Kawakatsu (2024) has recently suggested the potential to constrain anisotropy parameters near the source region using NDC components. Therefore, a proper estimation of the moment tensor with less biased NDC components is becoming more important. However, no-volume-change condition (or no-isotropic components), presumed in conventional routine CMT analysis, may result in a biased CLVD (compensated linear vector dipole) component.
In this study, we have developed a new flexible method for the CMT analysis employing the surface-wave WKBJ method, considering the effects of localized structures in forward computation of theoretical seismograms, which enables us to take account of the effects of localized anisotropy and a large-scale 3-D structure without heavy computational costs for forward modelling.
At first, we investigated the effect of radial anisotropy on the retrieved CMT solutions through synthetic experiments. Theoretical seismograms were generated using the surface-wave WKBJ method, employing various combinations of independent seismic structures for the source area and the source-receiver paths. Then, these input seismograms were inverted using isotropic or anisotropic reference models based on our CMT analysis with nonlinear waveform inversion. This approach allows us to separately discuss the influence of seismic structures at the source and the ray paths (or wave propagation) on the retrieved CMT solutions. The results indicated that radial anisotropy in the source-receiver paths affected the centroid depth and time, while that in the source area generated apparent NDC components with less impact on the centroid location. These results indicate the importance of considering radial anisotropy in the CMT analysis.
We then applied the method to real data for selected seismic events in and around the Japanese islands and performed the CMT inversions with multiple reference models (with or without the radial anisotropy) using the F-net data. We also examined the effects of zero-trace constraints on the CMT solutions. Preliminary results indicate that incorporating radial anisotropy in the reference model affects the relative proportions of the DC and NDC components.
In this study, we have developed a new flexible method for the CMT analysis employing the surface-wave WKBJ method, considering the effects of localized structures in forward computation of theoretical seismograms, which enables us to take account of the effects of localized anisotropy and a large-scale 3-D structure without heavy computational costs for forward modelling.
At first, we investigated the effect of radial anisotropy on the retrieved CMT solutions through synthetic experiments. Theoretical seismograms were generated using the surface-wave WKBJ method, employing various combinations of independent seismic structures for the source area and the source-receiver paths. Then, these input seismograms were inverted using isotropic or anisotropic reference models based on our CMT analysis with nonlinear waveform inversion. This approach allows us to separately discuss the influence of seismic structures at the source and the ray paths (or wave propagation) on the retrieved CMT solutions. The results indicated that radial anisotropy in the source-receiver paths affected the centroid depth and time, while that in the source area generated apparent NDC components with less impact on the centroid location. These results indicate the importance of considering radial anisotropy in the CMT analysis.
We then applied the method to real data for selected seismic events in and around the Japanese islands and performed the CMT inversions with multiple reference models (with or without the radial anisotropy) using the F-net data. We also examined the effects of zero-trace constraints on the CMT solutions. Preliminary results indicate that incorporating radial anisotropy in the reference model affects the relative proportions of the DC and NDC components.