4:45 PM - 5:00 PM
[MIS01-05] Triggering mechanism and development of the turbidity current induced by the 2011 Tohoku-oki earthquake
Keywords:sediment gravity flow, machine learning, deep sea, inverse analysis
The 2011 Tohoku-Oki earthquake and its subsequent tsunami triggered turbidity currents in the Japan Trench, as evidenced by ocean-bottom pressuremetry and sediment core analyses. This study investigates the triggering mechanisms of these tsunami-generated turbidity currents and their implications for reconstructing past megathrust earthquakes along the Japan Trench. We conducted numerical experiments using a 2D shallow-water turbidity current model implemented in Python. Two triggering mechanisms were evaluated: (1) earthquake-induced sediment liquefaction and (2) tsunami-induced seabed erosion.
Our simulations exhibited that the earthquake-induced model can generate turbidity currents reaching the trench but fails to explain the widespread turbidite deposits. In contrast, the tsunami-induced model produces multi-surge turbidity currents capable of depositing sediments over extensive areas, consistent with observed turbidite distributions, including those associated with the 869 Jogan tsunami. However, the numerical simulation results for the tsunami using Delft3D show that even the tsunami of 2011 could not provide enough suspended load to generate such turbid flows. In other words, we may have to consider a hybrid mechanism in which the tsunami erodes the seabed that has been fluidized by the earthquake, combining the mechanisms described in (1) and (2) above.
To reconstruct past turbidity currents, we applied an inverse modeling approach using deep neural networks (DNN). Our results demonstrate that DNN-based inversion can accurately estimate flow parameters from turbidite deposits when sufficient core data are available. Preliminary inverse modeling of the Jogan turbidite suggests that increasing the number of core sites improves the reconstruction of initial flow conditions. This study highlights the importance of tsunami-induced turbidity currents in shaping deep-sea sedimentary records and provides a methodological framework for reconstructing megathrust earthquake histories. Future work will focus on refining the numerical model by incorporating high-resolution topography and expanding core sample analyses to enhance the accuracy of inverse modeling.
Our simulations exhibited that the earthquake-induced model can generate turbidity currents reaching the trench but fails to explain the widespread turbidite deposits. In contrast, the tsunami-induced model produces multi-surge turbidity currents capable of depositing sediments over extensive areas, consistent with observed turbidite distributions, including those associated with the 869 Jogan tsunami. However, the numerical simulation results for the tsunami using Delft3D show that even the tsunami of 2011 could not provide enough suspended load to generate such turbid flows. In other words, we may have to consider a hybrid mechanism in which the tsunami erodes the seabed that has been fluidized by the earthquake, combining the mechanisms described in (1) and (2) above.
To reconstruct past turbidity currents, we applied an inverse modeling approach using deep neural networks (DNN). Our results demonstrate that DNN-based inversion can accurately estimate flow parameters from turbidite deposits when sufficient core data are available. Preliminary inverse modeling of the Jogan turbidite suggests that increasing the number of core sites improves the reconstruction of initial flow conditions. This study highlights the importance of tsunami-induced turbidity currents in shaping deep-sea sedimentary records and provides a methodological framework for reconstructing megathrust earthquake histories. Future work will focus on refining the numerical model by incorporating high-resolution topography and expanding core sample analyses to enhance the accuracy of inverse modeling.