PEDERSEN, GORDON, AND WHITE: SURFACE DECOUPLING EFFECTS 



The accuracy of the CHI approximation lies between that of the 

 linear and isospeed approximations. However, the linear method is 

 a significant improvement over the CHI approximation as well as the 

 isospeed approximation. 



For the case of negative near-surface gradients, the isospeed 

 method will always predict too large a decoupling loss and too large 

 a decoupling depth. 



Frequencies of 10, 31, and 104 Hz have been investigated for 

 three different ocean profiles. For all of these conditions the 

 decoupling loss of the linear method agrees with the mode result to 

 within 0.4 dB. At lower frequencies this ray approximation eventually 

 breaks down. The frequency of this breakdown depends on the depth 

 excess in the sound speed profile. When only a few modes are trapped, one 

 encounters the situation where the ray corresponding to the last trapped 

 mode does not intersect the surface. For this case a completely different 

 approach will have to be developed. At a frequency of 1.4 Hz for the Sea 

 of Japan profile, the linear method was off by 3.3 dB. 



Figure 8 presents the surface decoupling depth for the isospeed 

 approximation. In this approximation the depth may be expressed in 

 terms of acoustic wave length. Values are given as a function of 

 the ray angle at the surface which is plotted here on a two-cycle 

 logarithmic scale. 



Although the isospeed approximation can degrade by as much as a 

 factor of 3 in extreme cases, it does provide a simple and useful 

 guide. For example, the horizontal lines in the figure designate the 

 surface angles involved for various ray paths for two typical deep 

 ocean profiles. The upper line represents a Pacific profile with 

 a 100- foot surface layer, while the lower line represents the Atlantic 



5 76 



