Probabilistic tsunami damage assessment considering stochastic source models: Application to the 2011 Tohoku earthquake

A computational framework for probabilistic tsunami risk assessment due to a mega-thrust subduction earthquake is developed and is applied to the 2011 Tohoku Tsunami from retrospective viewpoints. The uncertain tsunami source characteristics are represented by multiple source inversion models and their stochastic variations that are generated using the spectral analysis and synthesis method. By conducting Monte Carlo tsunami simulation, stochastic inundation depth maps can be developed, which are subsequently integrated with tsunami fragility curves to develop stochastic tsunami risk maps. The stochastic tsunami risk maps display spatial variability of tsunami damage probabilities for a building portfolio, reflecting not only possible tsunami scenarios but also uncertain tsunami resistance of buildings. The numerical results indicate that both stochastic tsunami risk maps and risk curves are affected by the local terrain features, proximity to major tsunami sources, and building characteristics (material type and story number). Consideration of different reference tsunami source models in probabilistic tsunami risk assessment are identified as one of the critical contributors to the overall uncertainty of the tsunami risk predictions. Therefore, in determining critical scenarios for tsunami evacuation and risk mitigation, a wide range of possible tsunami scenarios should be considered in light of the current limited seismological knowledge for the mega-thrust subduction earthquake.