Probabilistic Assessment of Delayed Multi-fault Rupture Effect on Maximum Tsunami Runup along the East Coast of Korea

Abstract

In this study, the effect of delayed multi-fault rupture on maximum tsunami runup was investigated. Rupture of faults mostly occurs in a delayed pattern; however, conventional models assume that rupture of faults occurs simultaneously. We simulated the effect of delayed rupture of faults on the maximum tsunami runup on the east coast of Korea with reference to the 1993 Hokkaido earthquake. The Cornell Multi-grid Coupled Tsunami Model (COMCOT) was used to generate the maximum tsunami runup database on the east coast of Korea. The delay time between faults was examined with a 1-min interval up to 10 min. The maximum tsunami runup showed a tendency to increase significantly in a specific space and time interval. The middle part of the east coast of Korea (from Gangneung to Samcheok) showed a significant increase in the maximum tsunami runup. In particular, the maximum tsunami runup due to delayed rupture can be larger than that of simultaneous rupture of faults by up to a factor of 1.8. To quantitatively evaluate the effect of delayed rupture, we performed a data-driven analysis using Bayesian network with the database generated from numerical tsunami modeling. It was found that the maximum tsunami runup is amplified at a specific delay time (e.g., 4 min and 9 min) between rupture of faults and this is closely related to the superposition of tsunami component generated by each fault. Our results demonstrate that the effect of delayed rupture should be considered in the prediction of tsunami runup.