Stochastic Sound Speed Profile and Transmission Loss Models for the New England Shelf

The proximity of the New England Shelfbreak to Gulf Stream warm-core ring eddies, along with its bathymetry, makes the area one of the most complex ocean environments in the world. Its physical oceanographic complexity extends to the shelf, which can be a valuable study case for algorithms designed to recognize stochastic patterns in large datasets. Empirical Orthogonal Function (EOF) analysis, or Principal Component Analysis (PCA), has historically been the standard statistical method for detecting spatiotemporal patterns in data. Recently, sparse dictionary learning methods like the K-SVD (singular value decomposition with k-means clustering) have begun to supplant EOF analysis for this application. Pattern recognition results from these techniques are particularly useful to underwater acoustic propagation models that require realistic environmental data input to produce useful approximations of the underwater soundscape. Using in-situ Conductivity-Temperature-Depth (CTD) measurements taken on the New England Shelf and archived in the World Ocean Database (WOD), two stochastic models were developed based on EOF analysis and the K-SVD method. Both models were used to produce stochastic realizations of sound speed profiles (SSPs) and compared with each other. These SSP realizations were further used to calculate range-dependent probability density functions of Transmission Loss (TL) over a given sound propagation path. The analysis has shown that the K-SVD method can produce SSP and TL ensembles with fewer bases compared to the EOF method. Hence, a stochastic K-SVD model can better inform probabilistic underwater acoustic propagation simulations.