Community Tsunami Inundation Forecast using Next Generation Co-processor

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K12845
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Natural disaster / Disaster prevention science
• Research Institution: Tohoku University
• Principal Investigator: Koshimura Shunichi 東北大学, 災害科学国際研究所, 教授 (50360847)
• Research Collaborator: Musa Akihiro
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: シミュレーション / 高性能計算 / 広域被害把握

Outline of Final Research Achievements

Using a next-generation co-processor, we aim to develop a community tsunami inundation forecast system to be implemented in a community-level tsunami disaster response. To achieve tha goal, the research team conducted a research to optimize a tsunami inundation model running on a new computational platforms, to clarify the required computational resources, quality of the simulation, potential hardware configurations. As a result, we found that the TUNAMI code (Tohoku University's tsunami inundation simulation code) can be efficiently optimized on a new vector processor called SX-AuroraTSUBASA. It achieved a performance to execute the 6-hour inundation forecast with 10 m resolution in about 4 minutes using 256 cores. This research achievement leads to construct a community-based tsunami inundation forecast system for the end-users such as local municipality.

Free Research Field

津波予報

Academic Significance and Societal Importance of the Research Achievements

本研究成果の最大の意義は，高精度リアルタイム津波浸水予測を，安価な次世代プロセッサを用いることで実行可能にし，さらに大学等での大規模なスーパーコンピュータと同等以上のパフォーマンスを達成できたことである．これにより，自治体などコミュニティでの自立的・分散的な運用が可能なリアルタイム津波浸水・被害予測システムを構築することが可能になった．さらには，我が国だけでなく，津波リスクの高い諸外国への展開が可能になったことが重要な成果である．