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



Now, that's more of a scientific interest than it is an operator's 

 interest. I agree with you, Chuck, that the thing to do is to try the 

 model and test its limitations against what we already have. 



Dr. Munk: 1 want to ask a naive question of Bob Porter and Fred 

 Tappert. It seems to me that if you are analyzing an initial pulse 

 problem, there ought to be an easier way to put it on a computer than 

 to add a million frequencies. And it seems one should go back and do 

 the physical integration for the initial pulse rather than using our 

 traditional dependence on Fourier synthesis and trying to make that 

 work. 



Am I mistaken on that? 



Dr. Tappert: I can respond to that. It is possible to include 

 time dependence in the parabolic equation: to keep the time variable 

 t as another independent variable. It is simply a question of pro- 

 gramming technique whether you prefer a million points in time or a 

 million frequencies. 



I don't think that many will be necessary. But it's just a 

 tradeoff in the approach that you use. The Fourier space representa- 

 tion is equivalent mathematically to the temporal representation for 

 linear equations. 



Mr. M. A. Pedersen (Naval Undersea Center): Well, this last 

 summer I talked to Morris Daintith and Ian Roebuck at AUWE, and they 

 have developed what they call the impulse method which does perhaps 

 what you are thinking about. Rather than looking at a discrete fre- 

 quency, then integrating over frequencies, they do the opposite 

 problem of picking a response function for a given time and integrate 

 in the other direction. 



664 



