Parameter estimation using multi-frequency range-dependent acoustic data in shallow water

Abstract

The estimation of all forward model parameters - geometric, geoacoustic,

and ocean sound speed -by the inversion of acoustic field observations is considered. The data was taken at a mildly range-dependent shallow water site in the

Mediterranean Sea. The inversion is based on data from a vertical array and carried out using information at multiple frequencies. Global optimization using a directed Monte Carlo search based on genetic algorithms and the Bartlett objective function is used. All geometric parameters are well determined, a range-dependent geoacoustic model is determined, and the ocean sound speed is estimated. Comparison of the observed pressure field as a function of depth and the predicted field show good agreement. The use of observations at multiple frequencies provides considerable stability for the estimated parameters. Optimization of only geometric and geoacoustic parameters in a range-independent environment is found to be satisfactory at the lower frequencies (165-175 Hz), but for the higher frequencies (325-335 Hz) optimization of additional parameters by inclusion of either a range-dependent forward model or the ocean-sound-speed profile seems essential for successful inversion.