10:45 〜 12:15
[SSS04-P01] Preliminary study of three-dimensional attenuation structure using S-net strong-motion data
キーワード：S-net, Japan Trench, Spectral inversion, Quality factor, Attenuation structure
A preliminary study was carried out to see if analysis of 3D attenuation structure is possible using S-net strong-motion data. S-net is a large-scale seafloor network of seismometers and pressure gauges in the Japan Trench area (Aoi et al. 2020). The data used in the present analysis were identical to those in Dhakal et al. (2023). The analysis was done using the S-wave parts of the records, having vector peak accelerations smaller than 50 cm/s/s. In this study, to avoid the effects of coupling problems between the sensor houses and the seafloors, the horizontal X-component records only were used and the analysis was limited between 1 and 10 Hz. The magnitudes (Mw values) of the earthquakes ranged between 3.8 and 7.0, and focal depths were shallower than 70 km. The selected data comprises at least three records at each station and for each earthquake. The final dataset contained more than 6000 recordings from 605 earthquakes, including recordings from five KiK-net surface stations. We refer interested readers to Dhakal et al. (2023) for details on the data and data processing.
Using the method proposed by Nakamura and Uetake (2002, 2004), we performed simultaneous inversion of the Q value, source spectra, and site characteristics using the Fourier spectra at every 1 Hz between 1 and 10 Hz. The mean spectral values computed from the values for five consecutive frequency points with central frequencies at 1, 2, ..., and 10 Hz were used in the inversion. The ground amplification was divided into two groups for S-net and KiK-net. However, since the purpose of this preliminary study was to see if a 3D Q structure could be obtained, the site or source characteristics were not constrained like Nakamura (2009). Therefore, a trade-off was allowed between the site and the source spectra.
By performing checkerboard resolution test (CRT), we investigated an appropriate mesh size with which a reasonable attenuation structure could be obtained. Initially, two cases (0.1° x 0.1° x 10 km, and 0.5° x 0.5° x 30 km) with different attenuation block sizes in the longitude, latitude, and depth direction were considered. The analysis for 0.1° × 0.1° × 10 km resulted in poor resolution. The result of 0.5° x 0.5° x 30 km was well, but we wanted to know if we could see a more detailed structure or not. So, we proceeded with the examination of the case with 0.2° × 0.2° × 20 km block sizes. It was found that the third case reproduced the CRT reasonably well as shown in the included figure. At depths of 0-20 km, the resolution is generally good up to near the trench axis. This is a result that cannot be obtained only from inland data, for example, recorded by K-NET and KiK-net. Even the results in the 20-40 km depth ranges appeared to be satisfactory for the given data set to farther offshore. The Low Q zone that seems to extend along the island arc may indicate the structure at the mantle wedge. We are currently working on increasing the amount of high-quality S-net data. In our future study, we plan to evaluate the Q structure with a constrained condition, allowing the comparison of the results with past studies. We investigate how the Q structure contributes to estimating strong motions for various applications.
Acknowledgements
The hypocenter information of the earthquakes were obtained from the Japan Meteorological Agency. Generic Mapping Tools (Wessel and Smith 1998) were used to make figures. This study was partially supported by JSPS KAKENHI Grant Number JP20K05055.
References
Aoi et al. (2020). Earth Planets Space 72, 126.
Dhakal et al. (2023). Earth Planets Space 75, 1.
Nakamura (2009). PhD Dissertation, The University of Tokyo, 1-206.
Nakamura and Uetake (2002). Zisin 54, 475-488.
Nakamura and Uetake (2004). Zisin 56, 447-455.
Wessel and Smith (1998). Eos Trans AGU 79, 579.
Using the method proposed by Nakamura and Uetake (2002, 2004), we performed simultaneous inversion of the Q value, source spectra, and site characteristics using the Fourier spectra at every 1 Hz between 1 and 10 Hz. The mean spectral values computed from the values for five consecutive frequency points with central frequencies at 1, 2, ..., and 10 Hz were used in the inversion. The ground amplification was divided into two groups for S-net and KiK-net. However, since the purpose of this preliminary study was to see if a 3D Q structure could be obtained, the site or source characteristics were not constrained like Nakamura (2009). Therefore, a trade-off was allowed between the site and the source spectra.
By performing checkerboard resolution test (CRT), we investigated an appropriate mesh size with which a reasonable attenuation structure could be obtained. Initially, two cases (0.1° x 0.1° x 10 km, and 0.5° x 0.5° x 30 km) with different attenuation block sizes in the longitude, latitude, and depth direction were considered. The analysis for 0.1° × 0.1° × 10 km resulted in poor resolution. The result of 0.5° x 0.5° x 30 km was well, but we wanted to know if we could see a more detailed structure or not. So, we proceeded with the examination of the case with 0.2° × 0.2° × 20 km block sizes. It was found that the third case reproduced the CRT reasonably well as shown in the included figure. At depths of 0-20 km, the resolution is generally good up to near the trench axis. This is a result that cannot be obtained only from inland data, for example, recorded by K-NET and KiK-net. Even the results in the 20-40 km depth ranges appeared to be satisfactory for the given data set to farther offshore. The Low Q zone that seems to extend along the island arc may indicate the structure at the mantle wedge. We are currently working on increasing the amount of high-quality S-net data. In our future study, we plan to evaluate the Q structure with a constrained condition, allowing the comparison of the results with past studies. We investigate how the Q structure contributes to estimating strong motions for various applications.
Acknowledgements
The hypocenter information of the earthquakes were obtained from the Japan Meteorological Agency. Generic Mapping Tools (Wessel and Smith 1998) were used to make figures. This study was partially supported by JSPS KAKENHI Grant Number JP20K05055.
References
Aoi et al. (2020). Earth Planets Space 72, 126.
Dhakal et al. (2023). Earth Planets Space 75, 1.
Nakamura (2009). PhD Dissertation, The University of Tokyo, 1-206.
Nakamura and Uetake (2002). Zisin 54, 475-488.
Nakamura and Uetake (2004). Zisin 56, 447-455.
Wessel and Smith (1998). Eos Trans AGU 79, 579.