Clark and Yarnall 



which reaches the receiver; i.e. , all of the trajectories of energy 

 flow are accounted for. This feature of the model, coupled with the 

 fact that the travel time across the Straits is less than one minute 

 (much shorter than characteristic periods of many geophysical phenom- 

 ena of interest) , is the theoretical basis for the claim that the 

 received acoustic signal provides a synoptic measure of the medium 

 along the propagation path (Fig. 7 illustrates that the aggregate pat- 

 tern of either type of ray provides for substantial space coverage 

 along the propagation path) . 



The use of the Jacobson approach requires that parallel surf- 

 ace and bottom boundaries be assumed. This is a gross over- simplifi- 

 cation of the Fowey Rocks to Bimini path; and, in an attempt to par- 

 tially overcome this deficiency, a number of models have. been studied 

 with varying depths ranging from very shallow to deep with respect to 

 the mean depth of the actual path. These preliminary studies have re- 

 sulted in the selection of the medium representation of Figure 6 and 

 the SRBR and RBR models of Figures 7a and 7b as the best fit to all 

 available experimental data. (Only the first four ray arrivals are 

 shown in Figures 7a and 7b) . 



This "fitting" includes an independent prediction by the RBR 

 model of both acoustic travel time and travel time dispersion, which 

 are gratifyingly close to measured values. Project MIMI multipath 

 studies give a measured value of roughly 750 ms for maximum signal 

 time dispersion at H43.* The major portion of the acoustic energy 

 appears to arrive with a time spread of 250 ms or less. The RBR 

 model predicts a travel time dispersion for all ray arrivals of 

 600 ms. The first ten arrivals are time dispersed over 440 ms. 



The problem of how many ray arrivals to include in a given 

 calculation is one of the difficult questions associated with the 

 propagation model. Jacobson' s RBR analytical results ■'■ provide for 

 an infinity of ray arrivals with, however, a finite travel time dis- 

 persion. Inclusion of a plane wave reflection coefficient at bottom 

 contact eliminates, to a reasonable order of accuracy, the problem of 

 the infinity of arrivals. •'■^ In practice liberal use of the digital 

 computer permits one to sequentially "add in" ray arrivals until sa- 

 tisfactory resolutions are obtained, or until a decisive and physical- 

 ly meaningful trend is established. A total of 10 ray arrivals has 

 been used in the simulation studies to be discussed. 



A detailed comparative study of the SRBR propagation models 

 indicates that the RBR rays will provide the dominant contribution to 

 a cw signal. There are four reasons for this: 



Experimental results obtained in communication with Dr. T.G. 

 Birdsall, University of Michigan. 



314 



