3D numerical modeling of abyssal T-wave generation by scattering on rough ocean surfaces
Résumé
About 71% of the Earth's surface is covered by water. As a consequence, a significant portion of all earthquake energy gets released into the oceans. The term T-waves is associated with acoustic waves of seismic origin that travel in underwater waveguides at horizontal speeds close to the speed of sound in water. T-waves are the most common earthquake sounds in the ocean and make strong but transient contributions to the ambient sound field in the 1-100 Hz frequency band.\nT-wave generation scenarios described in the literature are unable to explain the common "abyssal" T-waves generated near the epicenter of earthquakes under flat abyssal plains.The hypothesis of T-wave generation by scattering of ballistic sound waves on the ocean surface roughness has been recently revisited by [O. A. Godin, J. Acoust. Soc. Am. 150, 3999–4017 (2021)] in the small perturbation approximation. Godin [2021] (conservative) estimates predict that ocean surface scattering could provide T-waves of significant energy (with respect to downslope conversion on large topographic features) and of realistic characteristics.\n\nThis article presents 3-D full-wave numerical modelings of the T-phase in a fluid-solid ocean model, which combines a large bathymetric feature with realizations of rough ocean surfaces (swell and wind wave).Specifically, after computing 3-D swell surfaces and sea state realizations from two-dimensional wind variance spectra, we simulate surface scattered T-phases at long ranges by coupling a time-domain spectral-element solver to a modal propagator by means of a mode projection technique. The results of these simulations are compared to 3-D simulations featuring a seamount.Our approach confirms Godin's calculations on most aspects and, for realistic swell and wind conditions, produces realistic T-waves whose amplitude is indeed significant compared to T-waves generated by downslope conversion.