11:15 AM - 11:30 AM
[SSS07-07] An estimation of seismic heterogeneous structures around deep slow-earthquake areas using receiver functions in multi-frequency bands
Keywords:Nankai subduction zone, Slow earthquake, Receiver function, Wave propagation modeling
We attempted to detect thin seismic heterogeneous structures in deep slow-earthquake areas in the Nankai subduction zone using both synthetic and observed waveforms. Seismic heterogeneity around a subducted slab reflects the environment hosting the slow-earthquake occurrence. A subducted seamount is one of the possible factors that yield heterogeneity, and previous studies showed that the subducting seamount may control shallow slow earthquakes (e.g., Sun et al., 2020). In contrast, in the deep slow-earthquake area, some studies have suggested the relationship between the ratio of P-wave velocity to S-wave velocity and around the deep slow-earthquake sources (e.g., Shelly et al., 2006; Nakajima and Hasegawa, 2016). Because of the not high spatial resolution of these studies, there is an ongoing debate about whether or how the seismic heterogeneous structure affects the occurrence of deep slow earthquakes. Therefore, estimating a detailed structure around the slab surface is essential for understanding the physics of a deep slow-earthquake occurrence. In this study, we examine the detectability in thin seismic heterogeneous structures, from wave propagation modeling of RAYSUM (Frederiksen and Bostock, 2000) and OpenSWPC (Maeda et al., 2017). We employed a model with a simplified low seismic velocity layer of 1 km thick between the continental plate and the dipping oceanic plate, and computed receiver functions (RFs) (Langston, 1979; Owens et al., 1984).
We recovered positive converted Ps phases at ~ 4 s for the slab surface. Later phases also included multi-reflected within the overlying plate. The arrival times and amplitudes varied with the back azimuth of rays, which was consistent with the slab geometry. However, synthetic low-frequency RFs up to 0.5 Hz, both by RAYSUM and OpenSWPC, failed to recover a thin low-velocity layer of 1 km thick. This is because the vertical resolution of the structure by the RF depends on the wavelength of the S-wave (Levin et al., 2016). Thus, we considered using RF with higher-frequency contents up to 2 Hz that may resolve localized heterogeneity around the slab surface.
As a result, we recovered thin seismic heterogeneous structures in RAYSUM, but not fully reflected in OpenSWPC. Computed wavefields by OpenSWPC may contain reflections from the boundaries of the model space if the width of the computational region setting is insufficient (Maeda et al., 2017). To overcome the calculation memory problem, we also perform high-frequency computations of OpenSWPC on the supercomputing system of the Earthquake and Volcano Information Center, the Earthquake Research Institute, the University of Tokyo.
Both low and high-frequency RFs experienced computational instability of deconvolution in the frequency domain. It generally comes from spectral holes in the power spectrum of the vertical component of the denominator in the deconvolution in the frequency domain. For a more stable RF calculation, we need to avoid the divisions in the frequency domain. To overcome the computational instability, we also computed the time-domain RFs (e.g., Ligorría and Ammon, 1999; Ruan et al., 2023). Based on synthetic results, we also performed multi-band RF analysis (e.g., Sawaki et al., 2021) for observed data in the Nankai subduction zone. Comparing observed and synthetic results, we discuss the presence of thin heterogeneous structures around deep slow-earthquake sources and the factors that cause the thin seismic heterogeneous structures.
We recovered positive converted Ps phases at ~ 4 s for the slab surface. Later phases also included multi-reflected within the overlying plate. The arrival times and amplitudes varied with the back azimuth of rays, which was consistent with the slab geometry. However, synthetic low-frequency RFs up to 0.5 Hz, both by RAYSUM and OpenSWPC, failed to recover a thin low-velocity layer of 1 km thick. This is because the vertical resolution of the structure by the RF depends on the wavelength of the S-wave (Levin et al., 2016). Thus, we considered using RF with higher-frequency contents up to 2 Hz that may resolve localized heterogeneity around the slab surface.
As a result, we recovered thin seismic heterogeneous structures in RAYSUM, but not fully reflected in OpenSWPC. Computed wavefields by OpenSWPC may contain reflections from the boundaries of the model space if the width of the computational region setting is insufficient (Maeda et al., 2017). To overcome the calculation memory problem, we also perform high-frequency computations of OpenSWPC on the supercomputing system of the Earthquake and Volcano Information Center, the Earthquake Research Institute, the University of Tokyo.
Both low and high-frequency RFs experienced computational instability of deconvolution in the frequency domain. It generally comes from spectral holes in the power spectrum of the vertical component of the denominator in the deconvolution in the frequency domain. For a more stable RF calculation, we need to avoid the divisions in the frequency domain. To overcome the computational instability, we also computed the time-domain RFs (e.g., Ligorría and Ammon, 1999; Ruan et al., 2023). Based on synthetic results, we also performed multi-band RF analysis (e.g., Sawaki et al., 2021) for observed data in the Nankai subduction zone. Comparing observed and synthetic results, we discuss the presence of thin heterogeneous structures around deep slow-earthquake sources and the factors that cause the thin seismic heterogeneous structures.