4:30 PM - 4:45 PM
[STT42-11] Relationship between the detailed site amplification estimated from DAS observation and near surface structure
Keywords:DAS, Site amplification, coda normalization method, Sakurajima
The seismic wave is influenced by Source, Path, and Site effects. In Japan, high-density site amplifications from NIED are used for disaster reduction planning in cities. DAS (Distributed Acoustic Sensing) is a new technology in geophysics that measures high-density seismic waves at intervals of a few meters along optical fibers. This method allows capturing detailed distributions of shaking. DAS data has suggested a negative correlation between site amplification and shallow speed structure (Yang et al., 2021). Additionally, DAS observations in the Azuma mountains indicated that site amplification, estimated by coda-normalization from DAS data, is influenced by the thickness of the near-surface sediment layer (Nishimura et al., 2020). Understanding the effects caused by near-surface structures like topography and geology could provide more detailed estimations of spatial distribution for site amplification. This study estimates amplification factors from DAS data in Sakurajima, examining the relationship between site amplification and various geological features like sediment and lava flows from different ages, as well as topographic changes.
From November to December 2023, DAS observation was conducted in Sakurajima. The ONYX interrogator by Sintela Ltd. was used, installed at the Sakurajima International Volcanic Sabo Center. Measurements were taken along a 38.8 km cable surrounding Sakurajima. The gauge length was 9.8 m, the channel interval was 4.98 m, and the sampling rate was 250 Hz.
A total of 23 earthquakes recorded by DAS, confirmed by the JMA catalog, were analyzed. Most earthquakes were located in Hyuganada, with others in the Amakusa region, near Tanegashima, and near Sakurajima. Amplification factors were calculated from a 10-second window of the S-wave direct part. All amplification factors were relative values. To compare with amplification factors, we estimated velocity by calculating the CCF from coda waves across various channel sections.
Amplification factors showed no correlation with the cable direction or distance but displayed differences between lava flows and sediment. Amplification factors ranged from 0.1 to 1 times on lava flows and 2 to 5 times on sediment, relative to the median. In general, sediments had velocities below 500 m/s, while lava flows had velocities above 600 m/s. A negative correlation was observed between amplification factors and velocity. However, when examining the Taisho lava flow between the 1300 to 2300 channels, a different relationship was found, with high amplification and high velocity.
When comparing the results with J-SHIS data, the cross-correlation coefficient between DAS and J-SHIS was less than 0.1. This is thought to be because, while overall has same tendency with J-SHIS, local amplification factor variations were observed in certain areas. This suggests the importance of evaluating localized effects not captured by differences in microtopography. Variability in the amplification factor was assessed using the bootstrap method, revealing significant variability around the 3000 channels and 5000 to 6000 channels.
We also investigated amplification factor differences based on the epicenter distance. In the latter part of the 5000 channels, the amplification factor was higher than the median when the epicenter was near Sakurajima, and lower when the epicenter was farther away. This suggests a path effect when the epicenter is close.
The amplification factor was also estimated using a 10-20 second window after the S-wave direct arrival and a time window for the coda part. No changes due to the time window were observed. Since Sakurajima exhibits strong fine heterogeneity underground (Shito et al., 2020), early waves from the direct wave to 10 seconds are scattered waves similar to the coda waves. The differences in the path from the epicenter to the observation point seem to have little effect.
Although Taisho lava has a high velocity, the amplification factor is still high. This may be due to the influence of structures deeper than Vs30, as suggested by Yang et al. (2021), where amplification factor and velocity calculated from the coda CCF may be sensitive to different depths. Additionally, in the case of strain measured by DAS, local heterogeneity could influence the results (Capdeville and Sladen, 2024).
From November to December 2023, DAS observation was conducted in Sakurajima. The ONYX interrogator by Sintela Ltd. was used, installed at the Sakurajima International Volcanic Sabo Center. Measurements were taken along a 38.8 km cable surrounding Sakurajima. The gauge length was 9.8 m, the channel interval was 4.98 m, and the sampling rate was 250 Hz.
A total of 23 earthquakes recorded by DAS, confirmed by the JMA catalog, were analyzed. Most earthquakes were located in Hyuganada, with others in the Amakusa region, near Tanegashima, and near Sakurajima. Amplification factors were calculated from a 10-second window of the S-wave direct part. All amplification factors were relative values. To compare with amplification factors, we estimated velocity by calculating the CCF from coda waves across various channel sections.
Amplification factors showed no correlation with the cable direction or distance but displayed differences between lava flows and sediment. Amplification factors ranged from 0.1 to 1 times on lava flows and 2 to 5 times on sediment, relative to the median. In general, sediments had velocities below 500 m/s, while lava flows had velocities above 600 m/s. A negative correlation was observed between amplification factors and velocity. However, when examining the Taisho lava flow between the 1300 to 2300 channels, a different relationship was found, with high amplification and high velocity.
When comparing the results with J-SHIS data, the cross-correlation coefficient between DAS and J-SHIS was less than 0.1. This is thought to be because, while overall has same tendency with J-SHIS, local amplification factor variations were observed in certain areas. This suggests the importance of evaluating localized effects not captured by differences in microtopography. Variability in the amplification factor was assessed using the bootstrap method, revealing significant variability around the 3000 channels and 5000 to 6000 channels.
We also investigated amplification factor differences based on the epicenter distance. In the latter part of the 5000 channels, the amplification factor was higher than the median when the epicenter was near Sakurajima, and lower when the epicenter was farther away. This suggests a path effect when the epicenter is close.
The amplification factor was also estimated using a 10-20 second window after the S-wave direct arrival and a time window for the coda part. No changes due to the time window were observed. Since Sakurajima exhibits strong fine heterogeneity underground (Shito et al., 2020), early waves from the direct wave to 10 seconds are scattered waves similar to the coda waves. The differences in the path from the epicenter to the observation point seem to have little effect.
Although Taisho lava has a high velocity, the amplification factor is still high. This may be due to the influence of structures deeper than Vs30, as suggested by Yang et al. (2021), where amplification factor and velocity calculated from the coda CCF may be sensitive to different depths. Additionally, in the case of strain measured by DAS, local heterogeneity could influence the results (Capdeville and Sladen, 2024).