Acoustic provincing of ocean basins

Abstract

The three-dimensional sound speed structure of the ocean basin is an important component that is required as input to acoustic models which are used to investigate acoustic propagation. Long-range (ocean basin) acoustic modeling which incorporates sound speed profile range-dependent effects requires significant computer time since modeling run-times are dependent upon the number of sound speed profiles input into the model. It is often necessary to reduce the acoustic model run-time by minimizing the number of input sound speed profiles. In order to examine the spatial variability of acoustic propagation on ocean basin scales, a simplification was required that could be achieved through acoustic provincing. A hierarchical divisive cluster technique was used to calculate acoustic provinces of the General Digital Environmental Model (GDEM) sound speed profiles data set. Cluster analysis is a multivariate statistical procedure for detecting natural groupings in data. In this case, the clutter analysis algorithm groups sound speed profiles based on the similarities in statistical properties related to the sound speed profile structure. The Euclidean distance was used as a measure of the similarity between sound speed profiles. The groupings, or clusters, of sound speed profiles are then mapped backed to their geographic locations in order to determine the sound speed provinces. The sound speed profile most similar to the other sound speed profiles within a province was selected as the representative sound speed profile for that province. The acoustic provincing algorithm is discussed as well as the application of the algorithm to the North Atlantic and North Pacific Oceans.