17:15 〜 19:15
[STT41-P02] 地震波速度構造のモホ速度不連続が日本海東縁部での震源深さに及ぼす影響
キーワード：震源深さ、地震波速度構造、モホ、日本海東縁、屈折波
The JMA catalog's focal depths in the eastern margin of the Japan Sea reach 40 km in some places. We interpret this deepening as the absence of velocity discontinuity at the Moho in the velocity structure used to locate earthquakes. Here, we show that the relocated depths using variable Moho depth (VMD) models become shallower and more consistent with those determined using data from the Ocean Bottom Seismometer (OBS) observation.
The eastern margin of the Japan Sea is one of the seismically active regions in and around Japan, including the source areas of the 1964 Niigata (M7.5), 1983 Japan Sea (M7.7), 1993 Hokkaido-Nansei-oki (M7.8), and 2024 Noto (M7.6) earthquakes. Thus, the information on earthquake locations in the region is important for understanding the seismotectonics and monitoring seismic activity. Some earthquakes in the region have deep catalog depths reaching 40 km, particularly in areas far from the land. Waveform paste-up of the deep earthquakes shows that the first phase is mainly refracted P-wave, followed by direct P-wave. This observation indicates the existence of a velocity discontinuity that refracts seismic waves, and the focal depth is shallower than the discontinuity. Compared with the known Moho depths, the abovementioned focal depths are too deep. The velocity structure for the hypocenter location by the JMA has no velocity discontinuity corresponding to the Moho. To account for the travel time of refracted waves by this model, hypocenters should be located deeper so that the seismic waves travel deeper, faster layers. This interpretation was confirmed by the shallowing of relocated focal depths using a velocity model with discontinuity at the Moho.
However, a model with constant Moho depth is unsuitable for the region. We tested a location using VMD models. We employed two models: the Japan Integrated Velocity Structure Model Version 1R (JIVSM_1R) and CRUST1.0. The former gives Moho depths at finer mesh points. We used the "power law" velocity model in which the velocity is expressed as the power of the radius from the Earth's center. The model's parameters are velocity values at the surface and the Moho, and Moho depth. By interpolating the Moho depth models, we give the average Moho depth for each source-station pair.
We evaluated the depth difference between the JMA catalog and VMD models by the difference in the D90 depth. The depths obtained by the VMD models are shallower than the JMA catalog. The difference is about 4 km at the location 40 km from the coastlines and more than 8 km further away from the coast. To evaluate the depth accuracy, we compared the focal depths with those obtained by the observation using OBS conducted after the 2024 Noto earthquake. In the area between the Noto peninsula and Sado Island, the maximum focal depth using the OBS data is about 18 km (Shinohara et al., 2024). Focal depths by the VMD models are consistent with this result, whereas the JMA depths reach nearly 30 km. The difference between the two VMD models is minor, indicating that including Moho discontinuity is essential for determining focal depths. Though the VMD model is a conventional technique and gives unrealistic focal depths for events far from the stations, it may be the second-best solution without a permanent OBS network.
The eastern margin of the Japan Sea is one of the seismically active regions in and around Japan, including the source areas of the 1964 Niigata (M7.5), 1983 Japan Sea (M7.7), 1993 Hokkaido-Nansei-oki (M7.8), and 2024 Noto (M7.6) earthquakes. Thus, the information on earthquake locations in the region is important for understanding the seismotectonics and monitoring seismic activity. Some earthquakes in the region have deep catalog depths reaching 40 km, particularly in areas far from the land. Waveform paste-up of the deep earthquakes shows that the first phase is mainly refracted P-wave, followed by direct P-wave. This observation indicates the existence of a velocity discontinuity that refracts seismic waves, and the focal depth is shallower than the discontinuity. Compared with the known Moho depths, the abovementioned focal depths are too deep. The velocity structure for the hypocenter location by the JMA has no velocity discontinuity corresponding to the Moho. To account for the travel time of refracted waves by this model, hypocenters should be located deeper so that the seismic waves travel deeper, faster layers. This interpretation was confirmed by the shallowing of relocated focal depths using a velocity model with discontinuity at the Moho.
However, a model with constant Moho depth is unsuitable for the region. We tested a location using VMD models. We employed two models: the Japan Integrated Velocity Structure Model Version 1R (JIVSM_1R) and CRUST1.0. The former gives Moho depths at finer mesh points. We used the "power law" velocity model in which the velocity is expressed as the power of the radius from the Earth's center. The model's parameters are velocity values at the surface and the Moho, and Moho depth. By interpolating the Moho depth models, we give the average Moho depth for each source-station pair.
We evaluated the depth difference between the JMA catalog and VMD models by the difference in the D90 depth. The depths obtained by the VMD models are shallower than the JMA catalog. The difference is about 4 km at the location 40 km from the coastlines and more than 8 km further away from the coast. To evaluate the depth accuracy, we compared the focal depths with those obtained by the observation using OBS conducted after the 2024 Noto earthquake. In the area between the Noto peninsula and Sado Island, the maximum focal depth using the OBS data is about 18 km (Shinohara et al., 2024). Focal depths by the VMD models are consistent with this result, whereas the JMA depths reach nearly 30 km. The difference between the two VMD models is minor, indicating that including Moho discontinuity is essential for determining focal depths. Though the VMD model is a conventional technique and gives unrealistic focal depths for events far from the stations, it may be the second-best solution without a permanent OBS network.