10:45 AM - 12:15 PM
[HDS06-P08] Numerical experiment to invert coseismic slip distribution from regional distribution of tsunami deposits
Keywords:Tsunami deposit, Tsunami source modeling, Paleotsunami
Tsunami deposit is a geological proxy of past tsunamis, including tsunamis that occurred before the period with instrumental observations. Several past tsunami sources have been reconstructed by the combination of the onshore distribution of tsunami deposits and tsunami numerical simulations. However, recent post-event surveys reported a gap between the inundation limit of tsunami and the distribution limit of onshore tsunami deposit. This leads to a concern about the potential underestimation of tsunami inundation due to using the inland limit of tsunami deposit as a constraint of minimum inundation. Recently, forward modeling of tsunami-induced sediment transport has been enabled by developing numerical models. Application of these models to paleotsunami source modeling has started in some cases.
The 2011 Tohoku-oki earthquake (Mw 9.0) ruptured inter-plate megathrust with a spatial extent of 300–500 km along its strike. Considering fault heterogeneity is essential to modeling such giant earthquakes. Fault heterogeneity can be expressed in kinematic fault modeling by giving different slips in each subfault and numerically solved by linear inversion of observation data in modern events. However, in paleotsunami source modeling using sediment transport modeling, fault heterogeneity must be solved by trial-and-error due to the nonlinearity that appears in near- or onshore tsunami behavior and sediment transport. Increasing the number of subfaults (i.e., the number of unknown parameters) means a more expensive computational cost to solve trade-off relations between slips of each subfault. Thus, an in-depth understanding of paleoseismic fault heterogeneity is difficult to be achieved by inverse analysis of tsunami sediment transport.
In this study, we propose a new framework for inverse analysis of fault heterogeneity using a regional distribution of tsunami deposit. This framework is achieved by parameter optimization of a forward model of tsunami-induced sediment transport. We use a numerical model TUNAMI-STM (Sugawara et al., 2014; Yamashita et al., 2016) in forward simulations of sediment transport. Unknown parameters to be solved in optimization are coseismic slips of 10 subfaults, which compose large megathrust fault mimicking the 2011 Tohoku-oki rupture. The sediment transport simulations are performed in five distant areas along the Japan Trench. The objective function is defined by the Euclidian norm of error between reference deposit thickness and simulated thickness and minimized by the sequential approximate optimization technique (Kitayama et al., 2011). The reference deposit data were artificially synthesized by sediment transport simulation. Thus, an objective of this numerical experiment is to reconstruct slip distribution of the reference fault used to generate the reference data by parameter optimization. In this presentation, we introduce the framework and current situation of optimization and discuss the relationship between tsunami source heterogeneity and the regional distribution of tsunami deposit.
The 2011 Tohoku-oki earthquake (Mw 9.0) ruptured inter-plate megathrust with a spatial extent of 300–500 km along its strike. Considering fault heterogeneity is essential to modeling such giant earthquakes. Fault heterogeneity can be expressed in kinematic fault modeling by giving different slips in each subfault and numerically solved by linear inversion of observation data in modern events. However, in paleotsunami source modeling using sediment transport modeling, fault heterogeneity must be solved by trial-and-error due to the nonlinearity that appears in near- or onshore tsunami behavior and sediment transport. Increasing the number of subfaults (i.e., the number of unknown parameters) means a more expensive computational cost to solve trade-off relations between slips of each subfault. Thus, an in-depth understanding of paleoseismic fault heterogeneity is difficult to be achieved by inverse analysis of tsunami sediment transport.
In this study, we propose a new framework for inverse analysis of fault heterogeneity using a regional distribution of tsunami deposit. This framework is achieved by parameter optimization of a forward model of tsunami-induced sediment transport. We use a numerical model TUNAMI-STM (Sugawara et al., 2014; Yamashita et al., 2016) in forward simulations of sediment transport. Unknown parameters to be solved in optimization are coseismic slips of 10 subfaults, which compose large megathrust fault mimicking the 2011 Tohoku-oki rupture. The sediment transport simulations are performed in five distant areas along the Japan Trench. The objective function is defined by the Euclidian norm of error between reference deposit thickness and simulated thickness and minimized by the sequential approximate optimization technique (Kitayama et al., 2011). The reference deposit data were artificially synthesized by sediment transport simulation. Thus, an objective of this numerical experiment is to reconstruct slip distribution of the reference fault used to generate the reference data by parameter optimization. In this presentation, we introduce the framework and current situation of optimization and discuss the relationship between tsunami source heterogeneity and the regional distribution of tsunami deposit.