TRANSMISSION KUNS 



83 



■M- 



Figure 14. Signal with end spikes. 



tudes which differ from those obtained by the other 

 method, but probably by no more than the internal 

 spread of either method. 



The next step consists of the assignment of an 

 amphtude to each member of the selected sample. If 

 the received signal were square-topped, like the 

 emitted signal, this step would raise no questions. 

 However, received signals are often far from square- 

 topped. It was decided at UCDWR to assign to each 

 signal the peak value of the amplitude registered any- 

 where during one signal, with two qualifications. One 

 concerns noise received simultaneously with the 

 signal. At low signal levels, the noise which is re- 

 ceived continuously shows up as a very striking 

 "spiny" record (shown in Figure 13). Such spines 

 superimposed on the signal are disregarded. This 

 rule presupposes that noise spikes can be distin- 

 guished from rapid signal fluctuations. It has been 

 found that all persons competent to evaluate record 

 film are able, with but little practice, to make that 

 distinction. The other qualification concerns "end 

 spikes." Frequently, there is interference between 

 sound traveling via two different routes, for example, 

 direct and surface-reflected sound. As the two paths 

 do not have exactly the same length, the intensity at 

 the beginning and end of the signal may be markedly 

 different from the intensity during the signal. In the 

 case of destructive interference, the signal then as- 

 sumes the shape shown in Figure 1-1. The end spikes 

 appearing in such signals are also disregarded. 



The rules just outlined have certain advantages 

 and certain drawbacks. The principal advantage is 

 that the peak amplitude of a signal can be read much 

 more rapidly than such quantities as mid-signal 

 amplitude; also, it is unambiguous. The drawbacks 

 appear when the signal envelope is not smooth. In 



that case the sound emitted during a short signal 

 interval arrives at the receiver during a much longer 

 period of time, with the result that the energy re- 

 ceived during an interval equal to the signal length 

 is substantially less than all the energy received. This 

 effect will be quite conspicuous for very short signals, 

 but negligible for continuous transmission (10-sec 

 signals). As a result, the average amplitude is very 

 definitely a function of ping length, when peak 

 amplitudes are used; it very likely would not be a 

 function of ping length if the amplitudes of individual 

 signals were defined in a different manner. One 

 possible solutioji has been suggested by the group 

 which is carrying out transmission experiments at 

 WHOI. They have constructed an integrating cir- 

 cuit. If the received signal is squared and fed directly 

 into this integrating circuit, the recording instru- 

 ment shows the total energy received. This would be 

 strictly proportional to the signal length and would 

 thus provide a measure typical for the ocean and its 

 overall transmission properties. Any deviation from 

 strict proportionality would be indicative of non- 

 linear transmission and would, therefore, be of the 

 greatest importance. At the time of this writing, no 

 such experiments had been carried out. 



Once individual amplitudes have been assigned to 

 the five signals that comprise one sample, the average 

 amjilitude is found by taking the arithmetical mean 

 of the five individual amplitudes. This procedure has 

 the advantage of simplicitj'-. Alternatively, one could 

 compute the mean level or the mean intensity 

 (squared amplitude). A very rough estimate shows 

 that in a typical record the averaging of amplitudes 

 and of intensities will lead to results which are dif- 

 ferent by about 1 db. While this difference depends 

 on the assumed distribution function of amplitudes. 



