3-D S-wave velocity structure of the upper crust derived by ambient seismic noise tomography of off Ibaraki region in the Japan Trench subduction zone

Abstract:

We applied ambient noise tomography analysis to a dense network of short-period ocean bottom seismometers (OBSs) deployed in off Ibaraki region, northeastern Japan along the Japan Trench subduction zone. The network consists of 32 OBSs with station intervals of about 6 km. First, we measured phase-velocities of the fundamental and the first-higher modes of Rayleigh wave using spatial auto-correlation (SPAC) method (Aki, 1957; Nishida et al., 2008) and derived the S-wave velocity (Vs) structure using trans-dimensional Markov chain Monte Carlo (MCMC) method with assumptions of flat and isotropic layers. Next, we measured phase-velocity anomalies for inverting 2-D phase-velocity structure using Fast marching method (Rawlinson et al., 2005; Saygin 2007). Finally, we inverted the phase velocity for a 1-D Vs structure at each horizontal grid and constructed a 3-D model by collecting the local 1-D structures. Although a trade-off between the estimation of layer thickness and Vs is often problematic for surface-wave non-linear inversion, we successfully reduced such trade-off by using both fundamental and first-higher modes.

Our Vs structure includes the higher-resolution Vs structure of the sedimentary layers and that of the crust, thanks to the dense array and the availability of both fundamental mode and first-higher modes of Rayleigh wave. Our model has thick sedimentary layers. The S-wave velocities range from several-hundred meters per second to two kilometers per second. The crust has complex structure compared with sedimentary layers. Notably, the south region has a more complex crust structure than the north area. Some P-wave velocity structures derived along the existing active-source seismic profiles suggest that some seamounts have been subducting in the southern part of the array. Such subducting seamounts may cause the crustal structure complex seen in the southern region (e.g., Wang and Bilek, 2011; Sun et al., 2020). Detailed velocity structure of the sedimentary layers, in which S-wave velocities are extremely low, will make a significant contribution to waveform modeling of OBS data. Our future work includes finite element three-dimensional waveform modeling (Ichimura et al., 2017) to better resolve complex structure of the subduction zone.