Abstract:
Broadband acoustic sources propagating in a hostile (noisy) ocean complicate the analysis of received acoustic data considerably. Normal-mode models are reasonable propagators for use in a shallow ocean environments. Our previous work developed the Gauss--Markov representation of the narrow-band normal-mode model and then recently extended it to the broadband case [Candy and Sullivan, J. Acoust. Soc. Am. Suppl. 1 98, (1995)]. In this paper the design and trade-offs of the processor are investigated when applied to both synthesized and experimental data in order to construct the desired broadband processor. The enhancement of broadband acoustic pressure-field measurements along with the estimation of underlying model functions using a vertical array is discussed. The model-based approach is developed and implemented for a broadband source using a normal mode propagation model. The structure of the processor is of interest, since it has an implied parallel structure due to the propagation physics, while the optimal estimation solution implies a ``full'' structure. The apparent discrepancy is resolved and the processor is implemented using a ``bank'' of narrow-band model-based processors. The resulting broadband pressure-field estimates are quite reasonable for both synthesized and experimental data.