Bayesian geoacoustic inversion of seabed properties for a continuous-gradient sediment using ship noise in deep water

This paper presents a Bayesian geoacoustic inversion method for estimating seabed properties using ship noise in deep water. Two bottom-reflected paths are observed in the processing of low-frequency data recorded by a bottom-moored vertical line array. To accurately represent the observed reflection paths and characterize the thick sediment layer influenced by gravitational compaction, the seabed is parameterized using a continuous-gradient model. Bayesian inversion based on the interference striations is applied to data from the periods before and after the closest point of approach. The Bayesian information criterion is employed to determine the optimal number of sediment layers. Inversion results indicate a total sediment thickness of approximately 460 m, comprising an upper layer with a gradual sound speed gradient (0.46 s-1) and a lower layer with a steep gradient (1.90 s-1). Parameter estimates show consistent agreement with seismic profiles, with the two sediment layers corresponding to the Pleistocene and the Pliocene, respectively. These results demonstrate that the proposed method provides reliable geoacoustic inversion in deep water. Furthermore, they indicate that the sediment structure is better represented by continuous gradients than by discrete homogeneous layers.