日本地球惑星科学連合2024年大会

講演情報

[J] ポスター発表

セッション記号 S (固体地球科学) » S-TT 計測技術・研究手法

[S-TT39] ハイパフォーマンスコンピューティングが拓く固体地球科学の未来

2024年5月26日(日) 17:15 〜 18:45 ポスター会場 (幕張メッセ国際展示場 6ホール)

コンビーナ:堀 高峰(国立研究開発法人海洋研究開発機構)、八木 勇治(国立大学法人 筑波大学大学院 生命環境系)、汐見 勝彦(国立研究開発法人防災科学技術研究所)、松澤 孝紀(国立研究開発法人 防災科学技術研究所)

17:15 〜 18:45

[STT39-P04] Prototype System for Forecasting of Plate Boundary Sliding Behavior Based on Sequential Data Assimilation

*堀 高峰1中田 令子1,2飯沼 卓史1 (1.国立研究開発法人海洋研究開発機構、2.東京大学大学院理学系研究科)

We are building a prototype system to forecast spatiotemporal variation in slip at plate boundaries in subduction zones. Comparing the consistency with daily observed on-shore and off-shore crustal deformation data and synthetic ones based on simulation results of spatiotemporal variation in slip at plate boundaries in a database prepared in advance, and quantifying the results according to the degree of consistency, we can evaluate the plausibility of the forthcoming plate boundary sliding behavior scenarios. Specifically, the Nankai Trough is targeted, and the data are GEONET's daily coordinate values (F5 solution) and the daily average values of the DONET's water pressure gauges. At present, the simulation database is based on earthquake cycle simulations that assume rate- and state-dependent friction laws and calculate spatiotemporal variation in slip at plate interfaces that take into account the three-dimensional shape in a semi-infinite homogeneous elastic body. As long as the crustal deformation result is given in a form that can be compared with observation data, there are no particular restrictions on the model used for simulation. So, it will be possible to add simulation results that incorporate realistic underground structures and different friction laws. It is also possible to add the results of data assimilation using other methods. Furthermore, it is also possible to perform prediction for multiple regions in parallel, with the region of data to be compared limited according to the region that can be handled by the model. By building such a highly flexible system, we would like to try forecasting where possible.