5:15 PM - 6:30 PM
[MIS15-P08] Examining grain-size changes in sediment transport modeling of the 2011 Tohoku-oki tsunami in the Sendai Plain
Keywords:tsunami numerical simulation, multiple grain size, the 2011 Tohoku-oki tsunami
Information of past tsunami events inferred from tsunami deposits is useful in preparing for future tsunami disasters. Existing tsunami sediment transport modeling is intended to correlate characteristics of tsunami deposits with tsunami source from wave heights and source parameters. In such simulations, either single or multiple grain size class is assumed for particle composition. Although some studies conducted sediment transport modeling considering multiple grain-size classes, majorities have been performed with the single grain-size assumption. However, the single grain-size simulations cannot resolve lateral and horizontal variations of grain-size composition. Major sedimentological features of sandy tsunami deposits, such as landward and upward fining, cannot be used for comparison of field data and numerical simulations. In this study, we examined and compared the sediment transport simulations of the 2011 off the Pacific coast of Tohoku Earthquake Tsunami, considering single and multiple grain-size compositions.
The study area is located in the neighborhood of the right bank of the Nanakita River in the Sendai Plain. A detailed field survey for the tsunami deposit was carried out previously (Abe et al., 2012) along a shore-normal transect (transect B). The Delft3D, which has been developed by Deltales, was used for the modeling. It includes cross-sectional simulations along transect B (1-D model) and two-dimensional horizontal simulations (2-D model) covering transect B. The fault parameters proposed by Imamura et al. (2012) was used to generate the initial water level of the 2-D tsunami simulations. Time series of the water level at 2 km offshore of transect B, generated by the 2-D simulation, was used for the input of the 1-D simulations. We assumed an initial thickness of 5 m for movable bed, which corresponds to the entire seafloor and the land areas from shoreline to the Teizan canal; meanwhile the areas further inland were assumed to be fixed bed (initial sediment thickness of zero). Four grain-size classes (sediment 1 to 4) were used to synthesize the grain-size distribution of the movable bed, based on the grain size data of the tsunami deposits (e.g., Szczuciński et al., 2012). The Manning’s roughness coefficient map constructed by Sugawara et al. (2014) was used for the computation of the tsunami-induced flow.
Modeling results showed that inundation distance, flow depth and distribution of the tsunami deposit were generally consistent with those measured by Abe et al. (2012), although there were some discrepancy in the flow depth near the shoreline. The simulations with single grain size class (mode diameter) and the multiple grain-size classes were almost similar. This is probably because of the assumption of the normal distribution for the grain-size composition and higher percentage of the mode diameter. Comparison of the 2-D and 1-D simulations showed a slight difference, due to effects from topography that cannot be considered in the 1-D simulations.
The study area is located in the neighborhood of the right bank of the Nanakita River in the Sendai Plain. A detailed field survey for the tsunami deposit was carried out previously (Abe et al., 2012) along a shore-normal transect (transect B). The Delft3D, which has been developed by Deltales, was used for the modeling. It includes cross-sectional simulations along transect B (1-D model) and two-dimensional horizontal simulations (2-D model) covering transect B. The fault parameters proposed by Imamura et al. (2012) was used to generate the initial water level of the 2-D tsunami simulations. Time series of the water level at 2 km offshore of transect B, generated by the 2-D simulation, was used for the input of the 1-D simulations. We assumed an initial thickness of 5 m for movable bed, which corresponds to the entire seafloor and the land areas from shoreline to the Teizan canal; meanwhile the areas further inland were assumed to be fixed bed (initial sediment thickness of zero). Four grain-size classes (sediment 1 to 4) were used to synthesize the grain-size distribution of the movable bed, based on the grain size data of the tsunami deposits (e.g., Szczuciński et al., 2012). The Manning’s roughness coefficient map constructed by Sugawara et al. (2014) was used for the computation of the tsunami-induced flow.
Modeling results showed that inundation distance, flow depth and distribution of the tsunami deposit were generally consistent with those measured by Abe et al. (2012), although there were some discrepancy in the flow depth near the shoreline. The simulations with single grain size class (mode diameter) and the multiple grain-size classes were almost similar. This is probably because of the assumption of the normal distribution for the grain-size composition and higher percentage of the mode diameter. Comparison of the 2-D and 1-D simulations showed a slight difference, due to effects from topography that cannot be considered in the 1-D simulations.