Abstract
In this study, we quantify uncertainty and include it appropriately into tsunami scenarios for inundation analyses for nuclear power plants. The uncertainty in responses of tsunami analysis is affected by dominant factors involved in fault characterization and rupture distribution, relative to tsunami source and evaluation point, submarine topography and site-specific geometry in land, so it is worthy importance to classify and quantify the uncertainty in site-specific tsunami inundation analysis. Feature values representing uncertainties in tsunami analysis were defined and the influence analyses on tsunami height at shoreline using machine learning were implemented, for tsunamis along the Japan Trench. Features related spread of wavefield in a north-south direction and maximum initial water level were highly influenced to tsunami height at shoreline, in case of tsunamis along the Japan Trench. The results also showed the importance and validity of relative position to tsunami source and evaluation point when considering the multiple evaluation points are considered.
