Earthquakes and tsunamis are catastrophic geohazards that bring destruction to cities and

sometimes countries. The recurrence interval of mega-earthquake occurrence has long been a topic of focus due to its close association with the risk assessment of natural hazards. Goldfinger et al. (2003) proposed the use of seismo-turbidite as an indicator of earthquakes. Seismo-turbidite are deposits of turbidity currents induced by the seismic activities and can be identified as a potential marker for seismic events in sedimentary records. Even though seismo-turbidites proved to be a good indicator of earthquake records in sedimentary successions, there is one major problem that needs to be resolved. That is how to determine the scale of the seismic event that deposited the seismo-turbidite. The goal of hazard prevention is to identify the frequency of events that may affect human activities on land, not any seismic activity. This study proposes the inverse modeling of seismo-turbidites to reconstruct the scale of earthquake induced turbidity currents from the deposited seismo-turbidites. Japan Trench has a relatively accurate historical record of past mega-earthquakes (Ikehara et al. 2016), making it the ideal study area as the first location for the testing the inverse analysis of seismo-turbidites deposited in nature.

This study presents a preliminary test of the inverse analysis of seismo-turbidites using artificial datasets of cores at the proposed sites of IODP Expedition 386 in a Japan Trench topographic setting. In this study, we used 2D numerical simulation and the submarine topography of the region to examine the behavior of turbidity currents resulting from the large area of submarine failure. Then, DNN inverse analysis was tested to determine whether it can reconstruct the location and the scale of the initiation region of turbidity currents resulting from mega earthquakes.

Goldfinger et al. 2003: Deep-water turbidites as Holocene earthquake proxies: the Cascadia subduction zone and Northern San Andreas Fault systems

Ikehara et al. 2016: Documenting large earthquakes similar to the 2011 Tohoku-oki earthquake from sediments deposited in the Japan Trench over the past 1500 years