nti-waveguide effects in Pacific slab: evidence from high-frequency waveform analysis and numerical modeling

Speaker: Simanchal Padhy
“There have been many studies on travel time seismic tomography to investigate the velocity structure of the subduction zones in the Pacific plate beneath the Japanese islands. Obayashi et al. (2009), based on teleseismic P-wave tomography, have shown the absence of slab-related velocity anomalies due to bending or tearing of the subducting Pacific slab at the junction between the Japan and Izu-Bonin arcs. On the other hand, Kennett and Furumura (2010), using seismic waveforms, have observed slab tear at the junction between Honshu and Kuril arcs. To corroborate the tomographic evidence for slab tear, we studied the detailed lateral structure of the subducting Pacific plate near Honshu by analyzing waveforms from deep earthquakes recorded at fore-arc Hi-net network and F-net broadband stations in Japan. Such waveforms explain the low-frequency (f < 0.25 Hz) precursors followed by high-frequency (f > 2 Hz) energies with long coda due to the multiple scattering and diffractions of seismic waves in the stochastic waveguide of the Pacific slab (Furumura and Kennett, 2005). However, recent analysis shows that for some particular source-receiver paths, the waveforms exhibit loss of high frequency energy in P-coda, high attenuation, loss of low-frequency precursor and presence of converted phases in P-coda for deep earthquakes occurring in the subducting Pacific plate. Such complexities in the observed waveforms are difficult to explain due to the radiation pattern of P- and S-waves and/or by anomalous propagation of seismic waves in existing plate model, indicating sudden lateral change in the wave guiding properties of the subducting slab, such as expected to be caused by the thinning or tearing the slab in deeper part representing anti-waveguide properties of the slabs. To explain the observations, we employ two-dimensional finite-difference method (FDM) simulations of complete high-frequency P-SV wave propagation taking thinning of the Pacific slab into account.

The model is represented by 5000 x 5000 grid points using a uniform grid size of 0.15 km. We use a parallel FDM simulation technique to conduct such large-scale 2-D simulations. The FDM simulation uses a staggered-grid 16th order scheme in space and 2nd-order accuracy in time. Random heterogeneity is introduced into the plate model through an anisotropic form of stochastic distribution described by a von Karman correlation function with elongated structures with a longer correlation length of 10 km in horizontal direction and much shorter correlation length of 0.5 km in depth and standard deviation from background P- and S-wave velocities of 5% following the study of Furumura and Kennett (2007).

We expect that the observed guided wave energy must decouple from waveguide where the slab is deformed or thin. Low frequency energy leaks out of the slab and travels to the receivers along paths in the low-velocity and low-Q mantle surrounding the slab, while high frequency signal of shorter wavelength can travel through thin plate. The results of this study, along with the evidence for weak velocity anomaly as inferred from seismic tomography (Obayashi et al., 2009) and observations of slab tear in the Pacific plate (Kennett and Furumura, 2010) as mentioned earlier, we expect a local velocity anomaly or thinning in the oceanic lithosphere along the junction between Izu-Bonin and Honshu arc. It is necessary to examine these effects further with a 3D FDM simulation for different slab geometries and source depths, which is ongoing. Acknowledgments We acknowledge NIED for providing Hi-net and F-net data. SP acknowledges JSPS for supporting to carry out this work and all the group members for stimulating discussions.”