Probabilistic Tsunami Inundation Maps Assuming Various Interplate Earthquakes along the Nankai Trough

The coastal areas of southwestern Japan have repeatedly suffered from tsunamis caused by the interplate earthquakes along the Nankai Trough. Similar tsunamis, therefore, will occur with a high probability. Tsunami hazard maps useful in disaster mitigation indicate accessible routes to evacuation lots on tsunami inundation depth distribution estimated by worst-case earthquake scenarios. Because earthquakes and tsunamis contain uncertainty, probabilistic assessment of tsunami hazards is also necessary, especially in urban planning and disaster insurance. Fujiwara et al. (2020) proposed various slip models of 3480 cases for interplate earthquakes ranging from M 7.6 to M 9.1 along the Nankai Trough and constructed tsunami hazard curves at the coastline. However, probabilistic assessments of tsunami inundation depth distribution would be better for practical uses. This study aims to create regional probabilistic tsunami inundation maps using the 3480 fault models. Tsunami calculations were performed repeatedly 3480 times for the whole region of the Nankai subduction zone. Because tsunami inundation is sensitive to on-land micro-topography, we used high-resolution DEM of 10 m spacing for an area (Tokushima Pref.) of about 50 km x 70 km by a nesting grid system. We defined inundation depth as the difference between the calculated maximum water level and the elevation after crustal displacement due to earthquakes. Earthquake occurrence probability for each case came from the Gudenberg-Richter relation consistent with the marine seismic activity. The annual frequency exceeding a threshold in inundation depth was calculated from the earthquake occurrence probability and the probability density distribution of the maximum water level considering aleatory uncertainty defined by a lognormal distribution. Tidal variability was also included in the exceedance frequency by convolving the probability density distribution estimated by the time series with a local tide record. The obtained probabilistic inundation maps indicate correlations with topography as that high probability appears in areas along the rivers and at the bay end of the rias coast. These features are not noticeable in the tsunami hazard curves along the coast.