Probabilistic tsunami hazard assessment with simulation-based response surfaces

T. Kotani, K. Tozato, S. Takase, S. Moriguchi, K. Terada, Y. Fukutani, Y. Otake, K. Nojima, M. Sakuraba, Y. Choe

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


This paper proposes a novel response surface-based method for probabilistic tsunami hazard assessment (PTHA). Although recent advancements in numerical simulation have enabled the accurate characterization of tsunami hazards, the high computational cost of numerical simulation often prohibits its broad application to probabilistic hazard analysis. The proposed method addresses this challenge by constructing the response surface (RS) of a target output using the results of high-fidelity tsunami simulations. The proposed method quantifies uncertainties in key simulation variables and propagates those uncertainties to the target output through the RS in a computationally efficient Monte Carlo simulation (MCS). We illustrate and validate the proposed method through a case study of the tsunami induced by the 2011 Great East Japan earthquake. The case study focuses on Sendai, Ishinomaki, and Kamaishi in Japan as the target locations. The proposed method estimates coastal tsunami heights while considering uncertainties in the fault slip and rake as well as the modeling error associated with the numerical simulation. The MCS allows us to estimate the probability density functions of the tsunami height at the target locations. The proposed method quantifies the contribution of each source of uncertainty to the overall uncertainty in the target output and thus facilitates engineering decision-making.

Original languageEnglish
Article number103719
JournalCoastal Engineering
Publication statusPublished - 2020 Sept


  • Fault model
  • Numerical simulations
  • Probabilistic Tsunami hazard analysis
  • Response surface
  • Uncertainty analysis


Dive into the research topics of 'Probabilistic tsunami hazard assessment with simulation-based response surfaces'. Together they form a unique fingerprint.

Cite this