CAVANAGH: AMBIENT-NOISE MODELS 



I have heard several comments over the last few days which should 

 help to put the noise-prediction question into perspective. One person 

 told me that he is amazed if he sees good agreement of a present-day 

 ambient noise prediction with an experimental measurement because of 

 the complexity of the inputs and the uncertainties involved. Another 

 person told me that as many as 10 years ago, when we had little infor- 

 mation about shipping densities and distributions, it was relatively 

 easy to predict noise levels to within a decibel or two. So this is 

 the point of departure. 



This paper addresses the question of ambient-noise prediction 

 modeling and we point out the obvious: anybody can construct an 

 ambient-noise model instantly. Given the sources and a receiver, all 

 we need are the characteristics of the sources, how they couple with 

 the medium, the transmission to the receiver, and a way to add up all 

 the results. Very simple. 



Consider first what we might call the ultimate ambient-noise 

 model, one for which we know: the locations and radiation character- 

 istics of all sources in the ocean basin, the mechanism by which 

 each source couples to the medium, and the properties of the sound 

 transmission from each source to the receiver. The contributions of 

 each source are then summed up in a way that is appropriate to the 

 application: whether it be array response or simply omnidirectional 

 reception. 



We could stop right here, since we do not have this detail in 

 the information about sources or sound transmission available to us 

 today, and we cannot expect to have it for years to come. Instead, 

 let us use the ultimate model as a basis from which to view what is 

 being done today and what we expect to be accomplished in the near 

 future. 



802 



