Sequential bottom parameter estimation using blind deconvolution of sources of opportunity in ocean waveguide

Abstract

This paper investigates the performance of sequential bottom parameter estimation based on ray-based blind deconvolution (RBD) of sources of opportunity using both numerical simulations and the 2016 Santa Barbara Channel (SBC) experimental recordings of shipping noise for ranges up to several kilometers. The RBD algorithm relies on estimating the unknown phase of the random sources to approximate the source-to-array channel impulse responses (CIR) by wideband beamforming along a well-resolved ray path [Sabra et al., JASA, 2010, EL42-7]. The power ratio of the direct and bottom-bounced arrivals is processed to infer the bottom reflection loss and is utilized to invert for the bottom parameters. Usually, the estimated bottom reflection loss is not accurate enough as the estimated CIR is noisy. Sequential parameter estimation uses a state space model for predicting and correcting the bottom parameters as the estimated bottom reflection loss values become available. This approach is a robust estimation tool employing predictions from previous estimates and corrects stemming from models that relate bottom parameters to the bottom reflection loss. Inversions results for the SBC experiment were also performed with conventional active sources to validate the inversion obtained with RBD of sources of opportunity.