Ocean acoustic field matching: normal mode filtering and non-Gaussian sources

Two qualitative results concerning statistical sonar signal processing and acoustic field matching are obtained. First, normal-mode field predictions are integrated with statistical signature analysis by constructing a boundary-value problem in the acoustic waveguide. From this construction it is found that the normal-mode filter is the unique acoustic preprocessor which does not confound deterministic waveguide correlation structure with stochastic source covariance structure. Second, the origin of deterministic, Gaussian, and non-Gaussian source signatures is investigated by associating physical parameters with the classical Lindeberg central limit conditions. From construction it is found that there are important objects that are not adequately represented either by infinitesimal points or by infinite surfaces. If receiver resolution is inadequate to resolve source complexity, these objects will exhibit a non-Gaussian acoustic signature via an entirely linear progression from internal excitation, to source radiation, through waveguide propagation, and finally to reception