Abstract:

The efficacy of empirical orthogonal functions (EOFs) in modeling ocean fluctuations for acoustic propagation predictions is well known, and recent work has demonstrated their advantages for sediment sound-speed variabilities. In this study, using EOFs for the direct estimation of sediment sound speed, density, and attenuation profiles is described, when relatively extensive data sets such as those from the Atlantic Generating Station (AGS) site are available. First, interpolations are generated based on minimizing mean-square errors. The resulting fields are appraised and compared thoroughly with those obtained from universal kriging, a method which is commonly employed by geostatisticians. Not only do the EOF interpolations have advantages in numerical efficiency, but also their construction is modified here to incorporate lateral terrain orientations as occur at the AGS site. Uncertainties in the geoacoustic fields are then modeled by stochastic variations in the EOF coefficients. The field representations are used for environmental input to propagation modeling of poro-elastic sediments in the AGS region. Statistics of transmission loss fluctuations are discussed in terms of geoacoustic variabilities. [Work supported by ONR.]