By Paolo Franchi In Memorandum Reports
Algorithms for automatic spatial-temporal search of favourable, marginal and unfavourable environmental conditions in support of anti-submarine warfare (ASW) activities. Abiva, Jeannine; Fabbri, Tommaso; Vicen Bueno, Raul. CMRE-MR-2019-027. March 2021
An understanding of how sound travels in the underwater environment is necessary to gain tactical, operational and strategic advantages in maritime operations. The variability in the underwater domain makes difficult getting a reliable categorization of the field in a timely manner, impacting naval operations such as Mine Countermeasures and Anti-submarine Warfare. In order to gain such advantages, even when variability is present in the environment, this report proposes an automated algorithm of categorizing the underwater environment over time and space according to its Sound Speed Profile (SSP). Although sound propagation schemes are available to determine the Transmission Loss from a single SSP, these algorithms are often computationally expensive for use over thousands of SSPs in a given Area of Interest (AOI). Motivated by unsupervised Machine Learning and Shape Analysis (SA) techniques, the proposed algorithms will automatically characterize an AOI. In particular, this report proposes the use of SA techniques and the Kohonen Self-organizing Map clustering algorithm to quickly classify SSPs in a given AOI. The resulting representative SSPs from the classification algorithms facilitates the transformation of the SSPs in the AOI into a Favourable, Marginal, and Unfavourable map. This study incorporates the shape of the curve into the classification schema, resulting in an algorithm with improved performance. The proposed algorithms are tested in a real and challenging scenario, the Strait of Gibraltar. This Strait is in the boundary between the Atlantic Ocean and the Mediterranean Sea, representing an area where the variability of the SSP depends on the region from where it is analyzed (East: Mediterranean; West: Atlantic). The results show how the proposed algorithms can be successfully used to analyze the data from each area over time and space to characterize an underwater environment. Further work will be done towards developing an automated search engine of specific acoustic propagation profiles in large Meteorological and Oceanographic databases.