Comparison of numerical predictions for transmission loss and reverberation levels in shallow waters.

Abstract

Since there are no models to currently predict transmission losses (TL) and reverberation levels (RL) for sound propagation problems in shallow waters, what is customarily done to estimate these parameters is to use various standard numerical codes in the same form as they were originally developed for deep waters. This introduces a variety of errors that one must live with, until suitable models are specifically developed for the various shallow water environments of interest. In order to begin to understand the nature and size of these errors (which could often be larger than 60 dB, i.e., that is a factor greater than one million!), half-a-dozen commonly used TL and RL prediction codes have been considered and have been applied to six geographical sites and to fourteen (conceptual but realistic) configurational cases. The models used are the generic sonar model (GSM), the BAM, COLOSSUS, NISSM, RAYMODE, RASP, and RUMBLE. An extensive comparison between the predictions of all these codes in all the cases and sites was carried out, and the results were compared to the (sparse) available measured data. An N.S.W.C. report with the details is in preparation. The study reveals that COLOSSUS provides the ‘‘best’’ TL predictions (i.e., closest to the data) and RASP the best ones for RL, in the shallow water environments found in the continental shelves.