PEDERSEN, GORDON, AND WHITE: SURFACE DECOUPLING EFFECTS 



function of receiver depth. These computations are for the condi- 

 tions of Figures 1, 2, and 4, i.e., the Atlantic profile and 10 Hz 

 frequency. The linear approximation agrees with the mode theory to 

 within 0.3 dB, while the CHI and isospeed approximation agree to 

 within 0.6 dB. Although not shown here, the surface decoupling depth 

 of the isospeed approximation is 6.4 percent shallower than that of 

 mode theory. The corresponding value for the linear and CHI approxi- 

 mation is 4.0 percent shallower than that of mode theory. 



For this profile the CHI and linear approximations are only 

 slightly better than the isospeed computation. The reason is that 

 the Atlantic profile has a surface layer down to 460 foot depth. 

 Thus the profile is essentially isospeed (gradient of 0.02 sec ) 

 for the upper 75 percent of the decoupling depth. 



Figure 7 is similar to Figure 6 except the computations are made 

 for a sound speed profile for the Sea of Japan, in the month of July. 

 This profile is characterized by severe negative gradients starting 

 at the surface. This gradient is -1.2 sec for the first 67 yards. 



The isospeed approximation is very poor. As can be seen in 



Figure 7, this approximation to the decoupling loss can be off by 



more than 9 dB. Moreover, this approximation to the surface 



decoupling depth is 191 percent too deep. This depth is then off 

 by a factor of three. 



In contrast, the linear approximation is very good. As can be 

 seen in Figure 7, this approximation to the decoupling loss is off 

 at most only 0.3 dB. Moreover, this approximation to the surface 

 decoupling depth is only 6 percent too deep. The accuracy of the 

 linear approximation then is essentially the same here as it was 

 for Figure 5. 



574 



