CAVANAGH: AMBIENT-NOISE MODELS 



suffice since the sources are distributed on the ocean surface. The 

 sources' dipole directivity induces the wind-noise directionality in 

 the sense that arrivals will usually be concentrated in steep angles. 

 Note that fluctuations in wind noise are not available from a model 

 like this, except for those long-term variations due to changes in 

 average wind speed. 



Let us turn now to ship-generated ambient noise. We would like 

 to know the precise location of each ship in the basin, its speed 

 and course, and its radiated noise level. We want to know how to 

 characterize the source. Is it a point source 20 feet below the 

 surface or is it a distributed source at some other depth? 



The model of sound transmission from source to receiver need 

 not have greater resolution than that inherent in the input informa- 

 tion for the sources. In other words, if we do not know the location 

 of a ship to within 30 or 40 miles, then there is no point in trying 

 to predict transmission loss any more carefully, unless we are 

 considering the special case of studying the fluctuations of ambient 

 noise resulting from movements of ships. For the usual application 

 in which we are estimating the average ambient noise level, we need 

 an accurate range-averaged transmission loss. 



Finally, we will sum the results of each ship's contribution 

 incoherently. If the transmission arrival structure is known, we 

 can extract both azimuthal directionality and vertical directionality. 



As far as the ship information for model input is concerned, 

 here is what is available today. We have estimates of the average 

 number of ships of each of several types, say merchants and fishing 

 vessels and tankers, in each X-degree cell (or square) in the ocean. 

 Typically, X = 5. Some ships make daily reports of their location. 



805 



