STATISTICAL ANALYSIS OF SHIP GENERATED NOISE 



S. C. Daubin 



Rosenstiel School of Marine and Atmospheric Sciences 

 University of Miami 



Factors involved in understanding the features of ship- 

 generated noise are discussed. Data are presented and 

 some of the salient features are identified. 



A calculational model for ship-generated noise is pre- 

 sented. The model involves area distributions of ships, 

 source level distributions, deterministic transmission 

 loss versus range curves, and a method for combining this 

 information. Model results are compared with data. 



Historically, acousticians have been conditioned to think in a 

 deterministic fashion: they attempt to establish the physics of a 

 situation and derive deterministically the results of some experiment. 

 One of the interesting aspects of the Munk and Zachariasen (1976) , 

 Dyson, et al. (1976) , and Flatte and Tappert (1975) studies is that 

 the random component in the sound velocity structure will force us 

 to think in the future of transmission loss in terms of a probability 

 density function. In the case of ambient noise, the random nature 

 of the variables involved is so obvious that we have little excuse to 

 think about it other than as a random process. This paper presents 

 some data, statistical results derived from these data, and then a 

 suggested method of interpreting the observations with respect to one 

 component of the various potential sources of ambient noise — namely, 

 the ship-generated component. 



Figure 1 shows a time series of ambient noise taken in the Pacific 

 by Gerry Morris aboard the FLIP. Basically, four sets of single 

 hydrophone data in 1/3-octave bands were taken over a period of several 

 hours' duration. Average 1/3-octave spectrum levels are plotted. The 



843 



