Bayesian maximum entropy for geoacoustic inversion using ship-radiated noise

Abstract

Several acoustic signals from merchant ships, also known as ships of opportunity (SOOs), are employed for geacoustic inversion of the seabed on the New England Mud Patch Area using a Bayesian maximum entropy (BME) approach. The SOOs were recorded by three vertical line arrays deployed at different locations in the Seabed Characterization Experiment in 2017 (SBCEX2017). An effective seabed composed of a mud layer on top of a sandy layer is used for this work. For each SOO, the space of possible hypotheses H is uniformly sampled using Monte Carlo. For each hypothesis in H containing a set of geoacoustic parameters, a SOO spectrogram is simulated using the range independent normal mode model ORCA where the squared error between the measured and simulated SOO is calculated. BME is then applied to compute the marginal conditional posterior probability distribution of the geoacoustic parameters where the probability density function (PDF) provides the current state of knowledge. The inferred PDFs are consistent across all the measured SOOs used for this research and agree with previous geoacoustic inversions in the SBCEX2017 [Work supported by ONR Contract No. N00014-19-C-2001.]