Feasibility studies on simulating seismic‐wave propagation in media for a suboceanic earthquake, including both land and ocean‐bottom topographies and a seawater layer, are scarce. Some of the conventional staggered‐grid finite‐difference method (FDM) simulations use a simplified structure model without a seawater layer and with a flat ocean‐bottom topography. In this study, we apply our heterogeneity, oceanic layer, and topography scheme (HOT)‐FDM and a 3D structure model including land and ocean‐bottom topographies, a seawater layer, and a fluid–solid boundary condition to an aftershock (Mw 5.8) of the 2009 Suruga Bay earthquake. We attempt to reproduce observations at seismic stations near the coast and then simulate waveforms in the ocean‐bottom stations. Our results show that a large difference between the cases with and without topographies can be seen in the coda part after the S wave in the simulated waveforms in terms of amplitudes and elongations. The synthetic waveforms in the model with topographies are in agreement with the observed waveforms. Our results also show that a significant difference in the amplification of the coda part between the cases with and without the seawater layer can be found at the ocean‐bottom stations. This coda part is the S‐ and Rayleigh‐wave propagation associated with the ocean and the underlying sediment layers. Our results show that a realistic model with topographies and a seawater layer is needed in FDM simulations in order to precisely reproduce observed waveforms or predict seismic motion for a suboceanic earthquake.