Extracting tomographic arrival time information from ships of opportunity in the Santa Barbara channel using blind deconvolution

Abstract

This paper summarizes the ongoing investigations surrounding the use of a ray-based blind deconvolution algorithm to recover arrival time information from sources of opportunity, such as shipping vessels, recorded on vertical line arrays (VLAs) in ocean waveguides. The deconvolution is primarily performed by using an estimate of the unknown source phase, obtained through wideband beamforming, to essentially match filter the VLA recordings and recover the channel impulse response (CIR). This paper will focus on results from an experiment performed in 2016 in the Santa Barbara shipping channel (water depth ~550 m). Four VLAs, with both short (~15 m) and long (~56 m) apertures, were deployed between the north and south bound shipping lanes and continuously collected acoustic data during one week. With the ultimate goal of passive acoustic tomography in mind, this paper aims to discuss (1) the robustness of the algorithm to extract differential arrival times along VLA elements using ships as sources of opportunity, (2) the achievable accuracy of blind arrival time measurements in comparison to the time-of-flight precision required for tomographic inversions, and (3) the ideal parameters (e.g., frequency bandwidth, snapshot duration, beamforming methodology...) for which to perform this ray-based blind deconvolution method in SBC-like ocean environments.