[S-TT46] 最先端ベイズ統計学が拓く地震ビッグデータ解析

[STT46-P03] Seismic wavefield imaging based on dense seismic networks using replica exchange Monte Carlo method

*加納 将行¹、長尾 大道^2,3、永田 賢二^4,5、伊藤 伸一^2,3、長谷川 慶²、酒井 慎一²、中川 茂樹²、堀 宗朗²、平田 直² (1.東北大学理学研究科、2.東京大学地震研究所、3.東京大学情報理工学系研究科、4.産業総合技術研究所、5.科学技術振興機構)

キーワード：MeSO-net、レプリカ交換モンテカルロ法、地震動イメージング

In recent years, dense seismic observation networks have been constructed. For example, the Metropolitan Seismic Observation network called MeSO-net has been observing ground motions since 2011 with the average intervals of ~5 km in the Tokyo metropolitan located on the Kanto sedimentary basin, Japan. Utilizing seismograms of MeSO-net, we have developed a wavefield reconstruction method by simultaneously estimating subsurface velocity structures and source locations using replica exchange Monte Carlo (REMC) method [Kano et al. 2017, GJI, JGR]. The reconstructed wavefield can be used as input ground motions to simulate seismic response of infrastructure for the purpose of rapid evaluation of seismic hazards. In this presentation, we will summarize the REMC-based wavefield estimation method and discuss future developments and applications of the method.