Page 66 - 2017Winter

Underwater Acoustics
guide to and from the target. The reverberation problem is concerned with determining ways to detect targets while screening out environmental reverberations caused by the interface roughness, variations in water temperature, schools of fish, and other sources of geological and biologi- cal clutter. One example of a high-frequency application is the detection and identification of underwater unexploded ordnance (UXO). Sea disposal of munitions was an accepted practice until it was recognized as an environmental haz- ard and banned in the 1970s, after an estimated 200 million pounds of UXO had been dumped into the world’s oceans. Advancements in synthetic aperture sonar, automatic tar- get recognition, and high-fidelity models have made rapid wide-area surveys for UXO possible. In another application, low-frequency sound has been used to image fish popula- tions over thousands of square kilometers. This technique provides an advantage over conventional methods that use slow-moving research vessels to monitor highly localized transects, which significantly undersample fish populations in time and space.
Propagation and reverberation of acoustic fields in shallow water depend strongly on the spatial and temporal variability of water column and seabed properties, and lack of knowl- edge of environmental variability is often a limiting factor in acoustic modeling applications. Although water column properties can be directly measured, seabed properties are more efficiently estimated through remote sensing tech- niques. Since the 1990s, matched field inversions have been applied to measurements of acoustic signals for geoacoustic parameter estimation and probabilistic inference for uncer- tainty estimation. In more recent work, the focus has been on Bayesian methods, which are characterized by the use of distributions to summarize data and draw inferences. Ad- vances have been made in optimization algorithms, such as simulated annealing and genetic algorithms, and methods of automated environmental parameterization.
Ambient noise in the ocean is the sound field against which signals must be detected, and scientists have sought to char- acterize it since the 1940s. Sources of ambient noise include dynamic processes of the sea, biological sources, such as marine mammals and snapping shrimp, and anthropogen- ic causes such as ships, geophysical prospecting, and con- struction, with significant spatial variation throughout the world’s oceans (Carey and Evans, 2011). The construction and operation of offshore wind farms have become an im- portant source of underwater noise because the increased
demand for renewable energy drives their production. Anal- yses of historical data suggest low-frequency ambient-noise levels have been increasing at a rate of 3 dB per decade since the 1960s (Hildebrand, 2009). Future increases in ambient- noise levels are expected to be exacerbated by ocean acidi- fication, which will lower seawater attenuation. In Arctic regions, the dramatic reduction in sea ice is expected to change the overall character of the ambient noise from being dominated by ice-generated processes, such as ridging and cracking, to being controlled by human activities, including shipping, seismic exploration, oil and gas development, and fishing. Although ambient noise is often viewed as an inter- fering source that masks a signal of interest, it can be used as a remote sensing tool. Scientists also use ambient noise to image the seabed, estimate precipitation and wind speed, and monitor deep ocean temperatures.
A great number of signal-processing techniques have been developed for different applications in underwater acous- tics. As a widely used method for processing acoustic data recorded on a linear array of hydrophones, the adaptive beamformer is prime example of sonar signal processing. Although many variants exist, all adaptive beamformers work by combining signals in a manner that increases the signal strength from a chosen direction and combining sig- nals from undesired directions in a benign or destructive way. Methods for working with sparsely populated linear ar- rays, including the design of coprime arrays or application of compressive-sensing techniques, have also been devel- oped in recent years. In other work, signal-processing tech- niques from quantum mechanics, including path integral methods and random matrix theory, have been applied to explain sound propagation through a stochastic ocean. A va- riety of techniques for estimating ocean acoustic parameters have also been applied to ocean acoustics problems. Some examples are warping methods to estimate modal eigenval- ues, particle filters to infer geoacoustic properties, and the waveguide invariant to localize sound sources.
Underwater acoustic instrumentation represents an area of extraordinary engineering achievements. The sound sur- veillance system (SOSUS), which began in the 1950s, was an early example of cabled acoustic observatories. The initial objective of SOSUS was the long-range detection of diesel submarines, but today, retired SOSUS installations are used for research in earthquake and volcano seismicity, the moni- toring of marine mammal behavior, and the use of acoustic methods for ocean acoustic tomography and thermometry
64 | Acoustics Today | Winter 2017