Hybrid numerical modelling of T-wave propagation: application to the Midplate experiment

Abstract

A hybrid method, coupling a ray tracing method and a finite differences approach, is proposed for modelling T-wave propagation from an underwater source to an on-land seismic station. The long-range hydroacoustic wavefield, estimated in the SOFAR channel by the Maslov approach, shows many triplications of propagation with an increasing number of caustics as the range increases. Ray tracing approaches lead to a straightforward analysis of the SOFAR propagation: we find that the duration and the amplitude of the hydroacoustic T waves generated by a source close to the SOFAR axis may be respectively eight times longer and almost seven times higher than the duration and the amplitude of hydroacoustic T waves generated by a source close to the SOFAR limits. The finite difference modelling handles the complex hydroacoustic-seismic T-wave conversion on atoll shores with an illustration of seismic T waves recorded during the Midplate experiment in 1989. Two different seismic stations, FGA on the Fangataufa Atoll and DIN on the Mururoa Atoll, both in French Polynesia, have recorded the seismic T waves due to an underwater chemical blast at a distance greater than 900 km. Synthetic seismograms computed by our proposed hybrid method are close enough to the real data for quantitative interpretation. We believe that the input model structure is accurate enough to allow such analysis of the seismic T waves. The numerical simulation shows that the seismic T waves recorded at both stations are mainly composed of P phases and Rayleigh phases. The simulation shows that the seismic T-wave duration is often linked to the source depth, although other factors (the continental slope or the distance between the top of the continental slope and the seismic station) may also affect the signal duration.