A comparison of acoustic uncertainty approximations in a Pekeris waveguide.

Abstract

Uncertainty in environmental parameters is often the dominant source of error in underwater acoustic field predictions. This presentation provides comparisons of three techniques for assessing uncertainty in a predicted acoustic field caused by uncertainty in environmental parameters: field shifting, polynomial chaos expansion, and coarse uniformly sampled direct simulations. The uncertainty assessments are performed in a 100‐m deep Pekeris waveguide with an uncertain water‐column sound speed for frequencies of 100 Hz to 1 kHz at ranges of 1 to 10 km with a variety of common bottom types. The accuracy and computational efficiency of each approximation are quantified in terms of an absolute‐difference error norm for the probability density function (PDF) of acoustic field amplitude and the number of acoustic field calculations necessary to predict this PDF, respectively. In all cases, the true field‐amplitude PDF is determined from numerically converged direct numerical simulations. The strengths and limitations of each technique are highlighted and the applicability of these results to other sound channels and uncertain parameters is discussed. In particular, polynomial chaos expansion produces the highest accuracy but does so at the highest computational cost, while field shifting produces acceptable accuracy for the least computational cost. [Work sponsored by the Office of Naval Research.]