Abstract

The October 2023 tsunami from Sofu Seamount showed large amplitudes relative to the associated seismic activity, suggesting atypical non-tectonic source processes. Comparisons of pre- and post-event bathymetric data revealed substantial topographic changes at Sofu Seamount, including the formation of a crater-like depression at the central cone within the larger caldera. Previous studies have suggested complex source processes, including repeated tsunami-generating episodes and uplift-dominated deformation; therefore, the complete source mechanism cannot be represented by a single deformation process. This study investigates tsunami excitation by an idealized localized subsidence process associated with crater formation, using the October 2023 Sofu Seamount event as a reference case. The objective is to examine how the duration of localized subsidence affects tsunami amplitude and dominant period, rather than to uniquely determine the complete source process of the event. Tsunami generation in the source region was simulated with a three-dimensional incompressible model that resolves vertical water-column motion and non-hydrostatic response, and the resulting water-surface variations were propagated to Hachijō-jima with a tsunami propagation model. We prescribed the localized subsidence as a single time-dependent event and varied its duration from 1 to 13 min to evaluate how the tsunami response depends on the source time scale. The simulations show systematic trends: shorter durations generate larger-amplitude and shorter-period waves, whereas longer durations produce smaller-amplitude and longer-period waves. Tide-gauge records at Hachijō-jima were used as a reference to assess whether the simulated waves had realistic amplitude and dominant-period characteristics. The results indicate that localized subsidence associated with crater formation can excite tsunami waves with amplitudes and dominant periods comparable to those observed at the coastal stations for certain subsidence durations. These findings indicate that localized subsidence associated with crater formation should be considered when evaluating possible components of volcanic tsunami generation, together with other possible processes such as uplift, slope failure, and repeated source episodes. A quantitative reconstruction of the October 2023 event will require wider-domain modeling, multistage source scenarios, and comparison with offshore pressure-gauge observations.