4:00 PM - 4:15 PM
[MIS15-05] Analysis of the Response of Tsunami Sediment Transport Numerical Simulation to Parameter Uncertainty
Keywords:Tohoku-oki tsunami, Sendai Bay, sediment transport , numerical modeling
Numerical simulation of tsunami-induced sediment transport is a powerful technique for improving our understanding of sedimentological characteristics of tsunami. Recently, validation of the simulation has been conducted using the vast amount of data obtained from the 2011 Tohoku-oki Tsunami (TOT) as a benchmark (Sugawara et al., 2014). Numerical simulations are generally affected by uncertainty related to input parameters for the model, however, few studies investigated the uncertainty systematically and comprehensively. In this study, sensitivity analyses using the TUNAMI-STM numerical model are conducted on the southern part of Sendai Bay, Miyagi Prefecture, where large-scale geomorphological changes due to the 2011 TOT were observed (Udo et al., 2013), in order to establish a systematic framework for evaluating the uncertainty in the simulation.
Sensitivity analysis was performed with ten parameters in total, such as grain size and settling velocity. A combination between grain size and other parameters, such as critical friction velocity and coefficient of transport formula, was taken into account in several cases. The area of comparison was divided into six segments, in which volumes of the tsunami-induced erosion and deposition varied significantly (Udo et al., 2013). The sensitivity was examined using simulated volumes of erosion and deposition for each case normalized by the outputs of a reference case.
The sensitivity was high in grain size and Manning’s roughness coefficient, which exhibited larger change rates of the erosion and deposition volume to change rates of the parameter. Besides, change rates of the erosion and deposition volumes by single simulation showed variabilities among the segments. As a result of detailed analysis focusing on sensitivity to the grain size, significant correlations were identified among the changes of the volume of onshore deposition, offshore erosion, the survival rate of the coastal dike (i.e. the ratio of fixed bed and a total length of dike) and height of the dike. In addition, changes in the sensitivity corresponding to the dimension of the segments were analyzed using narrower segments. These results suggest that the magnitude of the effect of parameter uncertainty depends on the distribution of coastal dike and spatial scale of comparison between simulation and observation.
Sensitivity analysis was performed with ten parameters in total, such as grain size and settling velocity. A combination between grain size and other parameters, such as critical friction velocity and coefficient of transport formula, was taken into account in several cases. The area of comparison was divided into six segments, in which volumes of the tsunami-induced erosion and deposition varied significantly (Udo et al., 2013). The sensitivity was examined using simulated volumes of erosion and deposition for each case normalized by the outputs of a reference case.
The sensitivity was high in grain size and Manning’s roughness coefficient, which exhibited larger change rates of the erosion and deposition volume to change rates of the parameter. Besides, change rates of the erosion and deposition volumes by single simulation showed variabilities among the segments. As a result of detailed analysis focusing on sensitivity to the grain size, significant correlations were identified among the changes of the volume of onshore deposition, offshore erosion, the survival rate of the coastal dike (i.e. the ratio of fixed bed and a total length of dike) and height of the dike. In addition, changes in the sensitivity corresponding to the dimension of the segments were analyzed using narrower segments. These results suggest that the magnitude of the effect of parameter uncertainty depends on the distribution of coastal dike and spatial scale of comparison between simulation and observation.