| The motivation for this work is a desire to exploit ship-radiated noise in shallow water environments for the purpose of passive range estimation. The method presented utilizes range-frequency intensity striations, arising from multi-path interference in shallow water waveguides and measured across a horizontal line array, to estimate range between a source and a receiver array. This method is based upon the premise that the effects of waveguide geometry on multi-path interference may be represented by a single parameter, beta, called the waveguide invariant, which relates striation slope to source range.;Much of the literature on striation-based passive range estimation is focused on range-independent environments or relies on the assumption that beta ≈ 1 in shallow water waveguides in order to eliminate the requirement of knowing the specific values of propagation parameters a priori. However, prior work has shown that range-dependent effects from bathymetry or sound speed profiles can cause the value of beta to vary from its canonical shallow water value and introduce significant bias to range estimates. This work specifically focuses on the effects of range-dependent bathymetry in shallow water waveguides and discusses issues surrounding normal mode propagation and mode coupling.;Beyond discussing range-dependent modal propagation, several signal, array, and image processing techniques are presented, which enable estimation of range-frequency intensity images from acoustic data and permit the images to be characterized in terms of striation angle and slope distributions. Using these techniques, a method for estimating range from striation patterns is developed and applied to simulated striation patterns and acoustic data from a 2007 experiment performed off the coast of Port Everglades, Florida. To improve range estimates obtained from the application of this method, an analytical formulation of a waveguide invariant distribution is derived, which takes into account range-dependent propagation parameters using the assumption that propagation satisfies the adiabatic approximation. This generalized waveguide invariant distribution provides a more accurate estimate of beta for use in the range estimation algorithm.;The results of applying the improved range estimation method to simulated and real data are compared, and practical issues of applying this method to acoustic data, including potential performance limitations, such as signal to noise ratio and effective array aperture, are also discussed. |
