An active sonar system is used to image wide areas of the continental shelf environment by long-range echo sounding at low frequency. The bistatic system, deployed in the STRATAFORM area south of Long Island in April-May of 2001, imaged a large number of prominent clutter events over ranges spanning tens of kilometers in near real time. Roughly 3000 waveforms were transmitted into the water column. Wide-area acoustic images of the ocean environment were generated in near real time for each transmission. Between roughly 10 to more than 100 discrete and localized scatterers were registered for each image. This amounts to a total of at least 30000 scattering events that could be confused with those from submerged vehicles over the period of the experiment. Bathymetric relief in the STRATAFORM area is extremely benign, with slopes typically less than 0.5 degrees according to high resolution (30 m sampled) bathymetric data. Most of the clutter occurs in regions where the bathymetry is locally level and does not coregister with seafloor features. No statistically significant difference is found in the frequency of occurrence per unit area of repeatable clutter inside versus outside of areas occupied by subsurface river channels.

The sonar equation rests on the assumption that received sound pressure level after scattering can be written in decibels as a sum of four terms: source level, transmission loss from the source to the target, target strength, and transmission loss from the target to the receiver. This assumption is generally not valid for scattering in a shallow water waveguide and can lead to large errors and inconsistencies in estimating a target's scattering properties as well as its limiting range of detection. By application of coherent waveguide scattering theory, the sonar equation is found to become approximately valid in a shallow water waveguide when the object's complex scatter function is roughly constant over the equivalent horizontal grazing angles +/- delta psi spanned by the dominant waveguide modes. This is approximately true (1) for all objects of spatial extent L and wavelength lambda when 2delta psi

Spectral and normal mode formulations for the three-dimensional field scattered by an object moving in a stratified medium are derived using full-field wave theory. The derivations are based on Green's theorem for the time-domain scalar wave equation and account for Doppler effects induced by target motion as well as source and receiver motion. The formulations are valid when multiple scattering between the object and waveguide boundaries can be neglected, and the scattered field can be expressed as a linear function of the object's plane wave scattering function. The advantage of the spectral formulation is that it incorporates the entire wave number spectrum, including evanescent waves, and therefore can potentially be used at much closer ranges to the target than the modal formulation. The normal mode formulation is more computationally efficient but is limited to longer ranges. For a monochromatic source that excites N incident modes in the waveguide, there will be roughly N2 distinct harmonic components in the scattered field. The Doppler shifts in the scattered field are highly dependent upon the waveguide environment, target shape, and measurement geometry. The Doppler effects are illustrated through a number of canonical examples.

No abstract

In a previous paper [Makris and Cato, J. Acoust. Soc. Am. 96, 3270 (1994)], it was shown that a vocal member of a humpback whale herd can be used as a source of opportunity to locate nonvocal members with a passive towed array. That analysis employed full-field but narrow-band propagation and scattering models to emphasize the high spatial array gains available. In the present paper, full-field simulations are performed to determine the structure of actual broadband humpback whale vocalizations after scattering from whales in a shallow-water waveguide. The simulations show that the time signature of a whale vocalization is significantly altered during each of the three stages of (1) propagating from vocal to nonvocal whale, (2) scattering from the nonvocal whale, and (3) propagating from the nonvocal whale to a receiver. The large time-bandwidth gains available in humpback vocalizations then cannot be optimally exploited without first modeling broadband propagation and scattering of the whale vocalization for the given waveguide and bistatic geometry. This raises serious questions about whether the humpbacks themselves, who have limited spatial gains over the noise, can actively detect nonvocal herd members with their vocalizations, as was discussed in the above reference.