18 EWINf! [chap. 1 



indication of Layer 3 from refracted waves, and computation of the structure 

 A\itli standard formulae for refraction data would give an excessive value of 

 tliickness for Layer 2. 



Owing to velocity gradients in the ocean, the Ri arrivals do not follow 

 straight paths as sho^^'n in the preceding figures, but travel the curved paths 

 shown in Fig. 12. For shot 1 the paths are slightly bent, but can be closely 

 ajDin-oximated by straight lines. The two rays emerging from shot 2 indicate 

 the limits of a bundle of rays observable at that range. The deep ray grazing 

 the bottom (RSR ray in Ewing and Worzel, 1948) marks the maximum range 

 of a first-order reflected wave. The Ri ray paths for shot 3 are combined 

 surface sound-channel paths and deep refracted paths (the leakage arrivals 

 described by Hersey et ah, 1952). Beyond the range of the grazing ray, the Rj 

 arrivals diff"er from shot to shot only in the length of the path in the surface 

 channel. Hence, they plot on a line on the time-distance graph which is parallel, 

 rather than asymptotic, to D. 



DETECTOR SHOT I 



/y////' / ' /'V/'/'y/' /'/'/'/'///'/'/'//'/'/' /'/'/'/'/'/' /'/'/ ' /'-^'^^ 



Fig. 12. Ray paths in deep water for shots and detector located in the surface sound 

 channel. For average Nortli Atlantic areas, shot 2 would correspond to a shot at 

 40-45 miles range. 



The ray diagram shown in Fig. 12 is appropriate for most deep-water areas. 

 In areas of shallower water it is possible for the long range Ri arrivals to re- 

 enter the surface channel after having been reflected from the sea floor. In this 

 case the Ri curve is also asymptotic to a line parallel to D. Whether the asymp- 

 tote is above or below D on the time-distance graph depends upon the speed 

 of sound in the surface channel relative to the average speed in the water 

 beneath the channel and upon the depth of water. In high latitudes particularly, 

 the speed in the surface channel may be lower than the mean vertical velocity, 

 causing the Ri curve to cross line D. The existence of the surface channel 

 depends upon the presence at the surface of a layer of isothermal water or of 

 one in which the normal thermocline is inverted, i.e. temperature increasing 

 with depth. Without one or the other of these temperature structures, there is 

 no channel and the direct wave, which is so useful for measuring distance in 

 seismic work, cannot travel for any appreciable distance between two near- 



