Estimation of range-dependent sound-speed profile with dictionary learning method

Abstract

A range-dependent dictionary learning method is proposed to retrieve the sound-speed profile (SSP) in a water body, with the whole computational domain decomposed into multiple range-independent subdomains. A dictionary is constructed by using the World Ocean Atlas depth profiles of temperature and salinity. The simulation results verify the efficacy of the dictionary in retrieving the range-independent SSP profile in each subdomain.