1:45 PM - 2:00 PM
[SCG59-01] Detailed seismic structure across the Kuril Trench off Nemuro, Hokkaido, Japan, revealed by FWI of dense OBS data
Keywords:Kuril Trench subduction zone, Full waveform inversion, Plate boundary structure, The 17th century Kuril earthquake
The southern Kuril Trench subduction zone experienced a M9-class great earthquake in the 17th century, which generated a large tsunami by a large coseismic slip on the shallow part of the plate boundary fault. Since there is a high probability of another earthquake within the next 30 years, the study of the seismic velocity structure in the source area of the 17th-century Kuril earthquake is crucial for understanding the mechanism behind the occurrence of large shallow slip. Azuma et al. (in revision) have performed a traveltime inversion (TTI) analysis of dense OBS and MCS data and found that the boundary between the large-slip shallow and deeper parts of the plate boundary fault is characterized by a sharp trenchward rigidity reduction in the overriding plate, associated with a prominent reflective zone (RZ). While the rigidity-reduced RZ is thought to mark the position sharply contrasting slip behavior along the plate boundary, the current TTI model provided a poor understanding of the formation of the RZ and its relationship to the plate subduction. In this study, we investigate detailed structural characteristics of the RZ and its surrounding structure using an acoustic full waveform inversion (FWI) technique. We aim to elucidate the relationship between plate subduction and the evolution of the forearc structure, as well as the relationship with the cause of large shallow slip.
We applied frequency-domain acoustic FWI to a controlled seismic survey data using OBSs that were installed at intervals of ~2 km along the line across the Kuril Trench, conducted by the R/V Kairei of the Japan Agency for Marine-Earth Science and Technology (Azuma et al., in revision). We performed FWI following a multiscale layer-stripping workflow (e.g., Górszczyk+ 2017) with four different offset ranges and six different frequency bands (min=2.5 Hz, max=3.5-8.5 Hz expanding by 1 Hz). In general, the result of the first arrival tomography result is considered to be a good starting model for the FWI. However, in this case, the FWI converged in a wrong local minimum when we used all the data. We investigated the cause and found that the starting model is too smooth to explain the detailed structure that affects the near-offset data. Therefore, we suppressed the weight of near-offset data. By this tuning, we succeeded in obtaining an appropriate model that can explain the observed waveform data well.
The resultant FWI model imaged more detailed structural characteristics, such as shorter-wavelength Vp heterogeneities in the shallow sedimentary layers and the frontal wedge, which are consistent with the MCS profiling, while the overall structure of the overriding plate is consistent with the TTI model. However, the deep structure showed an evident increase in the dip angle of the incoming plate approximately 30 km from the trench, which could not be estimated by TTI and MCS analyses. The evident dip change spatially correlates with the RZ developing above the plate boundary, suggesting that the shape of the incoming plate may be closely related to the formation of the RZ and thus associated with the change in coseismic slip behavior along the plate boundary. On the other hand, no significant change in Vp was detected within the subducting oceanic crust, suggesting that the oceanic crust is subducting with little alteration, at least out to 80 km from the trench, where is the landward limit of FWI resolution.
We applied frequency-domain acoustic FWI to a controlled seismic survey data using OBSs that were installed at intervals of ~2 km along the line across the Kuril Trench, conducted by the R/V Kairei of the Japan Agency for Marine-Earth Science and Technology (Azuma et al., in revision). We performed FWI following a multiscale layer-stripping workflow (e.g., Górszczyk+ 2017) with four different offset ranges and six different frequency bands (min=2.5 Hz, max=3.5-8.5 Hz expanding by 1 Hz). In general, the result of the first arrival tomography result is considered to be a good starting model for the FWI. However, in this case, the FWI converged in a wrong local minimum when we used all the data. We investigated the cause and found that the starting model is too smooth to explain the detailed structure that affects the near-offset data. Therefore, we suppressed the weight of near-offset data. By this tuning, we succeeded in obtaining an appropriate model that can explain the observed waveform data well.
The resultant FWI model imaged more detailed structural characteristics, such as shorter-wavelength Vp heterogeneities in the shallow sedimentary layers and the frontal wedge, which are consistent with the MCS profiling, while the overall structure of the overriding plate is consistent with the TTI model. However, the deep structure showed an evident increase in the dip angle of the incoming plate approximately 30 km from the trench, which could not be estimated by TTI and MCS analyses. The evident dip change spatially correlates with the RZ developing above the plate boundary, suggesting that the shape of the incoming plate may be closely related to the formation of the RZ and thus associated with the change in coseismic slip behavior along the plate boundary. On the other hand, no significant change in Vp was detected within the subducting oceanic crust, suggesting that the oceanic crust is subducting with little alteration, at least out to 80 km from the trench, where is the landward limit of FWI resolution.