Trans-dimensional geoacoustic inversion on the New England Mud Patch using modal dispersion data

Abstract

This paper presents trans-dimensional geoacoustic inversion of two low-frequency broadband signals recorded during the Seabed Characterization Experiment (SBCEX). The considered sources are a chirp emitted by a towed J15 source and an underwater impulse created by a combustive sound source (CSS). Corresponding received signals are recorded on single hydrophones, and the two source/receiver configurations are reciprocal, on the same track. Input data for inversion are modal time-frequency dispersion curves, estimated using warping. Inversion is then performed within a Bayesian framework. A trans-dimensional inverse algorithm is used to estimate the model parametrization (i.e., number of seabed layers), and the mode covariance matrices are estimated as a first-order autoregressive error process. Inversion results on the two reciprocal tracks are consistent with the modal information available in each dataset. Results are also consistent with what is known about the area.