Sonar Performance Predictions Incorporating Environmental Variability

Perfect spatial/temporal knowledge of the ocean environment is rarely available for evaluation of sonar performance. Instead, performance prediction is often done with acoustic models assuming idealized inputs. The current capability of many such models is superb; realistic inputs, however, often are not available, and this is the central focus of the present paper. The prediction of sonar performance using a probability density function (PDF) based on environmental variability is presented. The PDF describes the distribution of the predictive capability of an acoustic model with respect to measurements of actual performance and, therefore, represents the uncertainty in one's ability to model the actual performance of the system. The PDF accounts for the inherent variability of the environment not contained in the model inputs, and is a useful probabilistic description of the environment’s intrinsic variability. As examples, two littoral transmission loss data sets are invoked, and other passive sonar inputs are assumed, from which curves of predictive probability of detection (PPD) versus range are presented.