The 2011 Heisei tsunami far exceeded the level previously anticipated, resulting in devastating impacts in Japan. This event made it clear that preparation for tsunami hazards, based on past historical data alone, is inadequate. It is because tsunami hazards are characterized by a lack of historical data – due to the fact tsunamis are rare, high impact phenomena. Hence, it is important to populate a dataset with more data by including events that might have occurred outside the recorded historical timeframe, such as those inferred from geologic evidence. The dataset can also be expanded with “imaginary” experiments performed numerically using proper models. Unlike historical data that directly represent actual tsunami events as fact, geologic evidence (for example, sediment deposits) remains a conjecture for tsunami occurrences, and tsunami runup conditions evaluated using geologic data are uncertain. Theoretical approaches require making hypotheses, assumptions, and approximations. Numerical simulations require not only the accurate initial and boundary conditions but also adequate modeling techniques and computational capacity. Therefore, it is crucial to quantify the uncertainties involved in geologic, theoretical, and modeling approaches.
Approximately 30 years ago, research on paleo-tsunamis based on geologic evidence was initiated and has been significantly advanced in the intervening years. During the same period, substantial advances in computational modeling used to predict tsunami propagation and runup processes were made. Understanding tsunami behavior, characteristics, and physics have resulted primarily from the well-organized international effort of field surveys initiated by the 1992 Nicaragua Tsunami event. Such rapidly advancing knowledge and technologies were unfortunately not sufficiently implemented in practice in a timely manner. Had this been the case, the disaster of the 2011 event would have been reduced, possibly avoiding the infamous nuclear meltdown at the Fukushima Dai-ichi Nuclear Power Plant.
Having learned lessons from the 2011 Heisei Tsunami, Japan is now attempting to develop a robust tsunami-mitigation strategy that consists of two-tier criteria: Level 1 Tsunami for structure-based tsunami protection and Level 2 Tsunami for evacuation-based disaster reduction. Tsunami intensities of Levels 1 and 2 are determined by experts’ analysis and judgments. In the United States, a probabilistic tsunami hazard analysis is now widely adopted: for example, the latest ASCE-7 inundation maps are based on the hazard level of a 2,500-year return period. But again, due to the lack of data, the probabilistic analysis must rely mainly on imaginary experiments and experts’ judgments.
The topic of this special issue focuses on the theme of uncertainty involved in tsunami hazard prediction. We review and examine uncertainties associated with tsunami simulations, near-shore effects, flow velocities, tsunami effects on buildings, coastal infrastructure, and sediment transport and deposits. Substantial uncertainty regarding tsunami hazards is likely the result of tsunami generation processes. This component, however, is not discussed here because it is closely related to the topic of probabilistic ‘seismic’ hazard analysis.
This special issue is a compilation of seven papers addressing the current status of predictabilities, and will hopefully stimulate continual research that will lead to further improvements.
Presenting numerically simulated examples, the paper by Lynett shows that the accurate prediction of tsunami-induced currents are much more difficult to achieve than the prediction of inundation depths. A small difference in an input parameter in the numerical model results in a very large difference in currents, especially the currents associated with the eddy formations. Keon, Yeh, Pancake and Steinberg demonstrate that significant temporal and spatial variations in tsunami effects are exhibited in the GIS-based IT tool called the Data Explorer. The Data Explorer provides the means to explore and extract pre-computed numerical time-series data at any grid point specified by the user. The concept is simple, but it has the unique ability to retrieve the data extremely quickly from massive datasets. This capability allows us to directly analyze the time-series data and to perform comprehensive sensitivity analysis. In order to generate realistic tsunami waveforms in the laboratory, Hiraishi describes a novel laboratory apparatus equipped with a hybrid wavemaker system capable of producing a combination of currents, a large heave of water, and waves. With the use of this apparatus, the tsunami waveform observed off Japan’s Kamaishi coast is modeled in the laboratory tank. To attempt to numerically simulate the local effects, Arikawa and Tominta present their hybrid numerical model, combining a depth-averaged 2D model and a fully 3D hydrodynamic model with the use of a multi-grid numerical scheme. This approach is crucial because tsunamis are multi-scale phenomena. A typical tsunami wavelength in deep water is on the order of several tens to hundreds of kilometers. When a tsunami approaches the shore, it may break, so refinement of the grid size is necessary, and three-dimensional flows become important when evaluating the local effects. Jaffe, Goto, Sugawara, Gelfenbaum, and LaSelle provide a comprehensive review of the models used to estimate tsunami sediment/boulder transport and deposits, thereby inferring the tsunami runup conditions (inundation depths and flow speeds) based on the tsunami deposits. They suggest that techniques for uncertainty quantification are crucial to advance the science of tsunami sediment transport modeling. Yeh and Sato analyze the failure mechanisms of buildings and coastal protective structures observed following the 2011 tsunami. Revealing the mechanisms, some engineering considerations to achieve resiliency are proposed to cope with the so-called “beyond-the-design-basis” tsunami hazards, in which its uncertainty is uncertain. Manawasekara, Mizutani, and Aoki investigate the effects of buildings’ openings and orientations on tsunami loading by performing laboratory experiments. This paper is complementary with the one by Yeh and Sato in demonstrating that the detailed changes in structure design could make a significant difference in tsunami loading on the buildings.
We express our sincere appreciation to the authors for their contributions, and to the reviewers for their time-consuming efforts. We hope you find the papers in this special issue interesting and useful.
Comparisons of Numerical Models on Formation of Sediment Deposition Induced by Tsunami Run-Up