TRANSMISSION IN ISOTHERMAL WATER 



107 



ent sound measured in shadow zones) makes this 

 highly probable. 



At frequencies below 15 kc the attenuation of 

 sound is largely conjectural. The shallow water meas- 

 urements discussed in C'hapter G .show that a is not 

 greater than about 1 db per kyd at frequencies below 

 2 kc. On the other hand, if the attenuation in this 

 frequency range were as low as 0.1 db per kyd, high- 

 speed warships could be heard consistently many 

 hundreds of miles away. Since such listening ranges 

 are apparently not obtainable, it may be inferred 

 that the attenuation coefficient in the surface layers 

 of the ocean is greater than 0.1 db per kyd at all sonic 

 frequencies. Such a high value is not necessarily in- 

 consistent with the much lower value observed in the 

 deep sound channel since the attenuation at low 

 frequencies near the ocean surface may result pri- 

 marily from scattering of sound out of the isothermal 

 layer into the thermocline, where it is bent sharply 

 downward, and is lost. 



5.2.3 Variation of Transmission Loss 



The previous section has discussed average values 

 of the attenuation coefficient at each frequency, but 

 has ignored changes in this coefficient. It has already 

 been noted that from a single run out to 6,000 yd the 

 attenuation coefficient can be determined with a 

 probable discrepancy of only about }/^ db per kyd 

 from its true value at that particular time and place. 

 Since the scatter of the observed values exceeds this, 

 it may be inferred that the attenuation coefficient in 

 sea water is probably not constant. 



In the measurements reported in reference 16, for 

 example, half the attenuation coefficients at 24 kc 

 differed by more than 1 db from the average value of 

 4.4 db per kyd. Corresponding variations also ap- 

 peared at the other two frequencies (17.6 and 30 kc). 

 However, those runs in which more than one fre- 

 quency was used show a good correlation (correlation 

 coefficient r between 80 and 85 per cent) in the varia- 

 tion of the coefficients for the three frequencies. 



A good illustration of this correlation is provided 

 by the runs made during one 24-hr period (February 

 14-15, 1944), analyzed in reference 13. Seven meas- 

 urements of the vertical temperature structure were 

 made during this period. In each case, no variation of 

 temperature of more than 0.2 F was noted down to 

 depths of 120 ft, but this was also the limit of ac- 

 curacy of the thermometer on these days. The surface 

 temperature changed appreciably during the period, 



however, probably as a re.sult of the changing position 

 of the vessels. The variation of the attenuation 

 coefficients for .sound at 17.6, 23.6, and 30 kc is 

 plotted in Figure 18 with the measured surface tem- 

 perature. Evidently the attenuation coefficients at 

 the three frequencies changed very substantially; 

 however, the difference in the attenuation coefficients 

 between the different frequencies was more nearly 

 constant. 



Under some conditions, however, the attenuation 

 coefficient during a 48-hr period is less variable. A 

 series of transmission measurements was made by 

 UCDWR in the deep water off Point Conception, 

 California, where a persistent well-mixed layer was 

 to be expected. Measurements were carried out dur- 

 ing and after a storm, with winds of force 3 to 6 

 (Beaufort scale). The surface layers of the sea were 

 probably better mixed during these transmission runs 

 than for any other reported transmission experiments. 

 A typical temperature-depth record taken during 

 these measurements is shown in Figure 19. 



54 



56 



58 



60 



62 



64 



66 



WATER TEMPERATURE IN DEGREES F 

 Figure 19. Depth record for Point Conception runs. 



The cumulative distribution of attenuation coeffi- 

 cients for the data taken with the shallow and deep 

 hydrophone is shown in Figure 20. These are plotted 

 on probability paper, so that a normal, or Gaussian, 

 distribution of plotted points would lie on a straight 



