Broadband ML-approach to environmental parameter estimation in shallow ocean at low SNR

Ocean acoustic parameter estimation using a horizontal hydrophone array is addressed in this paper. Shipping noise is exploited as a source of acousti c energy. Shipping noise shows large relative bandwidths: signal energy is spread over a couple of octaves. Locations of the acoustic sources are not assumed to be known and have to be estimated simultaneously with ocean dept h. We develop a procedure based on conditional maximum-likelihood estimatio n for broadband signals in a shallow ocean. Monte Carlo simulations based o n the bootstrap paradigm demonstrate feasibility of our approach at low sig nal-to-noise ratio and allow to analyze the estimator accuracy. A simple pa rametric approach is used in this paper which generates bootstrapped sample s of the estimates of the array cross Spectral density matrices. Bootstrapp ed parameter estimates are obtained by applying the proposed estimator to t he bootstrapped samples. The joint distribution of the bootstrapped estimat es shows low correlation between source bearing, source depth, and environm ental parameters. Estimates of source range, however, are highly correlated with sound speed and ocean depth. If the source range estimate is larger t han the true range value, the ocean depth estimate is found to be smaller t han the true depth. The proposed broadband processor is applied to the sens or data obtained from an experiment with a towed horizontal hydrophone arra y in the Baltic Sea. (C) 2001 Published by Elsevier Science B.V.