Bayesian tracking predictions in an uncertain ocean environment.

Abstract

This paper considers probabilistic prediction of the future locations of a moving acoustic source in the ocean based on past locations as determined by Bayesian source tracking in an uncertain environment. The Bayesian tracking approach considers both source and environmental parameters as unknown random variables constrained by noisy acoustic data and prior information and numerically integrates the posterior probability density (PPD) over the environmental parameters to obtain a time‐ordered sequence of joint marginal probability surfaces over source range and depth. The integration is carried out using Markov‐chain Monte Carlo sampling methods which provide a large collection of track realizations drawn from the PPD. Applying a probabilistic model for source motion to each of these realizations produces a sequence of source range‐depth probability distributions for future times. These predictions account for both the uncertainty of the source‐motion model and the uncertainty in the state of knowledge o...