Solid Earth Sciences (S)
Session Sub-categoryTechnology & Techniques(TT)
Session IDS-TT46
TitleSeismic Big Data Analysis Based on the State-of-the-Art of Bayesian Statistics
Short TitleBayesian Analysis of Seismic Big Data
Main Convener NameHiromichi Nagao
AffiliationEarthquake Research Institute, The University of Tokyo
Co-Convener 1NameAitaro Kato
AffiliationEarthquake Research Institute, the University of Tokyo
Co-Convener 2NameTakuto Maeda
AffiliationGraduate School of Science and Technology, Hirosaki University
Co-Convener 3NameKeisuke Yano
AffiliationThe university of Tokyo
Session LanguageJ
ScopeRecently, a big seismic database has been being constructed that collects data of vibrators implemented in such as buildings, lifelines and smartphones, in addition to seismic data of the conventional continuous/temporal dense seismic observation arrays. Development of methodologies and algorithms, which are inadequate at this moment, optimized to comprehensively analyze the seismic big data is essential in order to utilize the big database as much as possible for prevention/mitigation of earthquake disasters and clarification of earthquake phenomena. On the other hand, recent progress of Bayesian statistics is significant, which is the mathematical basis of various methodologies, such as machine learning, especially deep learning, to extract valuable information from big data. The state-of-the-art of Bayesian statistics is expected to substantially advance seismic big data analyses.
This session mainly accepts presentations that focus on analyses of seismic big data, especially related to analysis methods based on Bayesian statistics such as machine learning, sparse modeling and data assimilation, and their applications to real seismic data. Presentations related to mathematical or statistical theories beneficial to data analyses, feasibility studies of algorithms eventually applicable to real seismic data, and the current status of seismic observations and analysis results are also highly welcome.
Presentation FormatOral and Poster presentation
Invited AuthorsYukitoshi Fukahata (Disaster Prevention Research Institute, Kyoto University)
Takahiro Omi (Institue of Industrial Science, the University of Tokyo)
TimePresentation NoTitlePresenterAbstract
Oral Presentation May 27 PM1
13:45 - 14:00STT46-01A bad inversion result suggests bad modelingYukitoshi Fukahata
14:00 - 14:15STT46-02A seismic wavefield estimation in a dense seismograph network: An optimization for the MeSO-netTakahiro Shiina
14:15 - 14:30STT46-03Uncertainty quantification based on 4DVar data assimilation for massive simulation modelsShin-ichi Ito
14:30 - 14:45STT46-04Bayesian approach for statistical modeling and forecasting of aftershock activitiesTakahiro Omi
14:45 - 15:00STT46-05Bias Correction for the Distribution of Aftershocks Within Short-Term Period Immediately After Large Main ShockKosuke Morikawa
15:00 - 15:15STT46-06Deep-learning-based Earthquake Detection for Continuous Seismic Network RecordsKeisuke Yano
Presentation NoTitlePresenterAbstract
Poster Presentation May 27 PM2
STT46-P01P-wave Polarity Determination of Waveform Data Observed in Western Japan, Using Deep LearningShota Hara
STT46-P02Trans-dimensional Bayesian inversion for a radially anisotropic S-wave velocity model in the crust and upper mantle: Application to the Australian continentToru Taira
STT46-P03Seismic wavefield imaging based on dense seismic networks using replica exchange Monte Carlo methodMasayuki Kano
STT46-P04Bayesian oscillator decomposition for seismic dataTakeru Matsuda
STT46-P05Uncertainty evaluation of source parameter estimates by MCMC in OklahomaNana Yoshimitsu