Clark and Yarnall 



The resultant, R, will exhibit a sharp amplitude fade and a rapid 

 phase change of close to 180° during the time interval between dia- 

 gram I and II. Figure 10a is a simulation of events commonly observed 

 in the Straits of Florida cw data. Figure 11 is a photograph of a 

 section of record containing several of these nulls. 



Two significant observations have been derived from the model 

 study of the swinging profile: 



1) Over the range of values studied, acoustic phase can 

 be taken as a linear measure of the change in gradient 

 except in the vicinity of the interference nulls. 



2) Over the range of values studied, the rapid phase 

 changes at the interference nulls provide no immediate 

 information about the variation in the medium. If our 

 purpose is to use acoustic phase as a measure of the 

 change in the gradient, then, as a matter of practical 

 necessity at this stage of our understanding, we must 

 regard these events as "noise" introduced into the 

 measurement. 



It is appropriate here to speculate briefly on the conse- 

 quences of adding a true noise component to the multipath model, 

 i.e. , an influence which perturbs the multipath ^ructure randomly. 

 To create physical significance for this concept we might think in 

 terms of turbulent eddies that create random nonuniformities in the 

 medium stratification. One very likely consequence would be a ran- 

 dom occurrence of interference nulls such as the events presented in 

 Figure 11. However, if the "noise level" is not too high, complete 

 incoherence of the acoustic signal will not be observed. (In Fig. 11, 

 for example, the phase can be tracked easily through the peak occur- 

 ring at 1 hour, 20 minutes, even though the continuity of the 9(t) 

 display is interrupted by the interference nulls) . Studies that have 

 been made of the interference nulls do in fact point to a random oc- 

 currence in time, suggesting the prevalence of random rather than 

 systematic multipath effects of this kind. 



We now refer to Figure 12 which presents the counter exam- 

 ple to the very stable transmission data of Figure 9. The "noisy" 

 character of the phase, 0(t) , and, in particular, the role played by 

 phase discontinuities at the interference nulls can be seen. In 

 spite of this, the expanded display of 9(t) does exhibit a recogni- 

 zable trend. In Figure 12 this trend has been tracked over a time 

 period of only about 12 hours; in the example of stable transmission 

 (Fig. 9) the trend took the form of a half -sinusoid. Both of these 

 data samples have been extracted from the long cw transmission test 

 previously designated as LCT-1. It is convenient at this point to 

 display certain of the data from this long test over its full time 

 range from September 1966 to January .1967 (Fig, 13). Data from the 

 thermistor string are shown, and phase at the H3 and H43 hydrophones. 



318 



