Japan Geoscience Union Meeting 2015

Presentation information


Symbol S (Solid Earth Sciences) » S-SS Seismology

[S-SS26] Seismic wave propagation: Theory and Application

Tue. May 26, 2015 9:00 AM - 10:30 AM 103 (1F)

Convener:*Tatsuhiko Saito(National Research Institute for Earth Science and Disaster Prevention), Hisashi Nakahara(Solid Earth Physics Laboratory, Department of Geophysics, Graduate School of Science, Tohoku University), Jun Matsushima(School of Engineering, The University of Tokyo), Kiwamu Nishida(Earthquake Research Institute, University of Tokyo), Kazuya Shiraishi(JGI, Inc.), Chair:Hiroshi Takenaka(Department of Earth Sciences, Graduate School of Natural Science and Technology, Okayama University), Takayuki Miyoshi(Japan Agency for Marine-Earth Science and Technology)

9:45 AM - 10:00 AM

[SSS26-04] Three dimensional seismic wave-speed structure beneath the Kanto Plain based on adjoint tomography

*Takayuki MIYOSHI1, Masayuki OBAYASHI1, Yoko TONO1, Kazuto ANDO1, Seiji TSUBOI1 (1.JAMSTEC)

Keywords:Kanto Plain, seismic wave-speed structure, adjoint tomography

We have obtained the preliminary model of three dimensional (3D) structure beneath the Kanto Plain, metropolitan area of Japan. We applied the spectral-element method (e.g. Peter et al. 2011) and adjoint method (Liu and Tromp 2006) to infer 3D velocity model and to reproduce the observed waveform bandpass filtered between 5 and 20 second. We used the travel-time tomography result (Matsubara and Obara 2011) as an initial 3D model and used broadband records obtained at the NIED F-net stations. We selected 147 earthquakes based on the earthquake catalog by the F-net and the S/N ratio of their seismograms. The 3D model used for the forward and adjoint simulations is represented as a region of approximately 500 by 450 km in horizontal and 120 km in depth. Minimum period was 4 sec. The initial 3D model reproduced P-wave seismograms well, however it could not really explain S-waves and later arrivals. For the adjoint inversion, we picked up the windows of the body waves from the observed and theoretical seismograms. The forward and adjoint simulations were implemented by K computer in RIKEN. The performance of the original application was 5% of the peak performance of the K computer, however the modified code achieved 10% by optimization. One iteration requires about 0.1 million CPU hours at least. The model parameters of Vp and Vs were updated by using the steepest descent method. The revised model reproduces observed waveforms better than the initial model. Acknowledgements: This research was partly supported by MEXT Strategic Program for Innovative Research. We thank to Dr. Daniel Peter for his comments and suggestions. We also thank to the NIED for providing seismological data.