3:00 PM - 3:15 PM
[SSS06-05] Distinctive deep S reflections from the Pacific slab
Keywords:reflected wave, Pacific slab, Wave propagation simulation
Distinctive later phase from moderately deep earthquakes near the Ryukyu Islands.
From moderately deep earthquakes near the Ryukyu Islands at depths from 30 to 200 km, distinctive later phases are observed over a wide area from central Japan to Hokkaido. Such phases can be observed for earthquakes from Amami Isl. to Taiwan, but not for events near Kyushu. Fig.1 shows the radial (R) component of F-net records for an earthquake off the main island of Okinawa (2014-03-03, 115 km, M6.5). This record section shows two distinctive signals (*s1 and *s2) with an offset of about 30-40 s. Their fast apparent wavespeed (7 to 8 km/s), particle motions, and low-frequency (< 0.5 Hz) components indicate reflected S waves from the mantle. The detailed characteristics vary with source location and depth. From events in the same area Fukao et al. (1983) reported P reflections observed in central Japan, and suggested that P reflections at the Pacific slab interacted with the 410 km discontinuity. The observed S reflection appears to have a similar mechanism.
Anomalous ground motion due to S reflection
Fig.1a shows the peak ground velocity (PGV) obtained from F-net and Hi-net with broadband filtering (Maeda et al. 2011), and Fig.1b the attenuation function with the expected curve for this event (Si and Midorikawa, 1999). The PGV shows a large (>10) peak at 1500 km distance (Niigata-Kanto region). Large PGV are also observed at larger distances around 1700 km (Akita) and 2500 km (East of Hokkaido). The large-scale anomaly arises from S wave interaction with heterogeneous mantle structure.
2D FDM simulation
To investigate the cause of the remarkable S reflections and peculiar PGV distribution, 2D FDM simulation of seismic wave propagation was performed along a cross section from Ryukyu Island to Hokkaido (Fig. 1a; a-a’)., The Pacific and Philippine-sea slabs were included based on Slab 2 (Hayes et al. 2018) on top of ak135-F as a laterally homogeneous background model (Kennett et al. 1995). Since the Pacific slab is modeled only to a depth of 400 km in this cross section, deep slab structure was extended based on tomography (Gorbatov & Kennett 2002; Fig.1d). Lower attenuation (Qp/Qs=1200/600) and higher wavespeed (2 to 5%) than the surrounding mantle was set inside the slab, assuming a thermal structure for the subduction zone. A low-wavespeed (-10%) oceanic crust with a thickness of 7 km was placed on the top of the slab to a depth of 110 km, and continued as a 7 km low-wavespeed (-10%) dehydrated layer from 110 to 450 km. Stochastic random fine-scale heterogeneity was set based on Kennett & Furumura (2018). A point (line) source was used with the F-net CMT and a source-time function of 3.5 s delta.
Simulation results
Fig.2a shows seismic wavefield snapshots at 250 – 450 s from the earthquake onset. P-wavefield is shown in red and S in green. A record section [R component] is also shown in Fig.2b. The snapshots show wide-angle S wave reflection from the upper boundary of the Pacific slab at depths from 400-600 km. The interaction of the reflection with the refracted wave transmitted through the high-wavespeed slab and the direct wave, produces a large triplication in S that is superposed on the S triplication caused by the 410 km discontinuity to form a large *s1 phase near 1500 km epicentral distance. The *s2 phase is generated by the S triplication from the 660 km discontinuity. The attenuation pattern of the calculated PGV explains the observations well (Fig.1c, Fig. 2c), but the large PGV peak is displaced to about 150 km further away. The large S reflection and peculiar PGV pattern cannot be reproduced by simulations without slabs (Fig.2c, + mark). Large S reflections at 410/660 km discontinuities can be reinforced by the anti-waveguide effect of the subducted slab for low-frequency (< 0.5 Hz) wave.
From moderately deep earthquakes near the Ryukyu Islands at depths from 30 to 200 km, distinctive later phases are observed over a wide area from central Japan to Hokkaido. Such phases can be observed for earthquakes from Amami Isl. to Taiwan, but not for events near Kyushu. Fig.1 shows the radial (R) component of F-net records for an earthquake off the main island of Okinawa (2014-03-03, 115 km, M6.5). This record section shows two distinctive signals (*s1 and *s2) with an offset of about 30-40 s. Their fast apparent wavespeed (7 to 8 km/s), particle motions, and low-frequency (< 0.5 Hz) components indicate reflected S waves from the mantle. The detailed characteristics vary with source location and depth. From events in the same area Fukao et al. (1983) reported P reflections observed in central Japan, and suggested that P reflections at the Pacific slab interacted with the 410 km discontinuity. The observed S reflection appears to have a similar mechanism.
Anomalous ground motion due to S reflection
Fig.1a shows the peak ground velocity (PGV) obtained from F-net and Hi-net with broadband filtering (Maeda et al. 2011), and Fig.1b the attenuation function with the expected curve for this event (Si and Midorikawa, 1999). The PGV shows a large (>10) peak at 1500 km distance (Niigata-Kanto region). Large PGV are also observed at larger distances around 1700 km (Akita) and 2500 km (East of Hokkaido). The large-scale anomaly arises from S wave interaction with heterogeneous mantle structure.
2D FDM simulation
To investigate the cause of the remarkable S reflections and peculiar PGV distribution, 2D FDM simulation of seismic wave propagation was performed along a cross section from Ryukyu Island to Hokkaido (Fig. 1a; a-a’)., The Pacific and Philippine-sea slabs were included based on Slab 2 (Hayes et al. 2018) on top of ak135-F as a laterally homogeneous background model (Kennett et al. 1995). Since the Pacific slab is modeled only to a depth of 400 km in this cross section, deep slab structure was extended based on tomography (Gorbatov & Kennett 2002; Fig.1d). Lower attenuation (Qp/Qs=1200/600) and higher wavespeed (2 to 5%) than the surrounding mantle was set inside the slab, assuming a thermal structure for the subduction zone. A low-wavespeed (-10%) oceanic crust with a thickness of 7 km was placed on the top of the slab to a depth of 110 km, and continued as a 7 km low-wavespeed (-10%) dehydrated layer from 110 to 450 km. Stochastic random fine-scale heterogeneity was set based on Kennett & Furumura (2018). A point (line) source was used with the F-net CMT and a source-time function of 3.5 s delta.
Simulation results
Fig.2a shows seismic wavefield snapshots at 250 – 450 s from the earthquake onset. P-wavefield is shown in red and S in green. A record section [R component] is also shown in Fig.2b. The snapshots show wide-angle S wave reflection from the upper boundary of the Pacific slab at depths from 400-600 km. The interaction of the reflection with the refracted wave transmitted through the high-wavespeed slab and the direct wave, produces a large triplication in S that is superposed on the S triplication caused by the 410 km discontinuity to form a large *s1 phase near 1500 km epicentral distance. The *s2 phase is generated by the S triplication from the 660 km discontinuity. The attenuation pattern of the calculated PGV explains the observations well (Fig.1c, Fig. 2c), but the large PGV peak is displaced to about 150 km further away. The large S reflection and peculiar PGV pattern cannot be reproduced by simulations without slabs (Fig.2c, + mark). Large S reflections at 410/660 km discontinuities can be reinforced by the anti-waveguide effect of the subducted slab for low-frequency (< 0.5 Hz) wave.