PORTER: SOFAR PROPAGATION OF WIDE-BAND SIGNALS TO LONG RANGES 



Figure 14 shows the propagation loss as a function of range for 



2 



the N - bilinear profile. The source and receiver are both on the 



axis of the sound channel. For this profile, the WKB and diffraction- 

 corrected losses are nearly identical. On the other hand. Figure 15 

 for the CHAIN 82 profile shows as much as 5 dB difference between the 

 WKB and diffraction-corrected loss curves. The two curves agree 

 pretty well for ranges less than 100 km and greater than 350 km. Also 

 shown is experimental data for CHAIN 82 , source and receiver both on 

 the channel axis. The frequency is 100 Hz. Interestingly, the WKB 

 loss calculations are within 2 dB of the experimental data, while 

 the diffraction-corrected theory overestimates the loss by as much 

 as 5 dB. 



Nonetheless, the agreement between theory and experiment appears 

 to us to be pretty good. We have been able to estimate the loss with- 

 out any artificial ass\imptions of mode incoherence. Further, the 

 wide-band loss calculations have about the same computation require- 

 ments as do narrow-band mode calculations. 



SUMMARY 



We have illustrated phenomena present in wide-band, long-range 

 transmissions. Group-velocity versus frequency (dispersion) curves 

 are revealed by spectrum analysis of shot records. Dispersion is 

 observed at frequencies as high as 300 Hz and for mode numbers as 

 high as 70. It has been shown that at frequencies between 25 and 

 300 Hz that both ray and mode structure is apparent in the pulse 

 signature. An analysis based on the WKB approximation shows that this 

 combined structure should appear in the data and, further, predicts 

 the field at long ranges from the source. Good agreement between 

 theoretical and experimental- dispersion curves has been obtained 

 with the WKB approximation to mode theory. 



655 



