TRANSMISSION IN ISOTHERMAL WATER 



97 



that the results do nut cast much light on the prob- 

 lems of transmission of low-frequency sound in deep 

 water at ranges greater than a few hundred yards. 



A much more complete set of measurements on low- 

 frequency transmission in deep water is given in a 

 progress report from UCDWR." That report sum- 

 marizes the first results obtained in a long-range pro- 

 gram designed to investigate sound transmission at 

 frequencies of 200, BOO, 1,800, 7,500, and 22,500 c. 

 Since short pulses of sound were used in this work, 

 the direct and bottom-reflected sound can be distin- 

 guished by the difference in travel time, provided the 

 range is not too great. Also, specially designed trans- 

 ducers were employed with the result that the power 

 output was high and usually remained constant dur- 

 ing each transmission run. 



Sample results for individual transmission runs at 

 200, 600, and 1,800 c are shown in Figure 11. As is 

 evident from the bathythermograph code given with 

 each plot, these data were obtained with isothermal 

 water at the surface. Each point in these plots repre- 

 sents the transmission anomaly found for a single 

 sound pulse. The curved lines represent the values 

 found from equation (3) with the amplitude reflec- 

 tion coefficient 7a chosen to give the best fit to the 

 observations. For each frequency, the measured 

 transmission anomalies at different ranges are moder- 

 ately accurate relative to each other, while the abso- 

 lute values are less reliable; hence each set of ob- 

 served anomalies in Figure 1 1 has been shifted verti- 

 cally to give the best agreement with the theoretical 

 curves. 



The agreement between the plotted points and the 

 theoretical curves is typical of the results generally 

 obtained in deep water with an isothermal layer. At 

 200 c, image effect is usually marked, and at 1,000 to 

 2,000 yd it reduces the sound level about 15 db on the 

 average below the inverse square value; this corre- 

 sponds to an effective reflection coefficient ya of 0.8, 

 somewhat lower than the value of 0.9 for the 200-c 

 curve in Figure 11. At 600 c, the effect is less marked, 

 corresponding to an effective reflection coefficient 7a 

 in the neighborhood of 0.7, and an average transmis- 

 sion anomaly of only about 10 db at 1,000 yd. How- 

 ever, the dip in intensity shown in Figure 1 1 at about 

 60 yd, in agreement with theoretical prediction, sug- 

 gests that image effect is in fact present. At 1,800 c, 

 the predicted minimum between 150 and 200 yd is ap- 

 parently present, but the reduction in intensity at 

 ranges between 1,000 and 2,000 yd is quite small, 

 corresponding to a value of about 0.5 for 7^. Thus at 



frequencies above about 1,000 c, it appears that image 

 effect is relatively unimportant. This is in general 

 agreement w-ith theoretical expectations.'- At ranges 

 greater than 2,000 yd, bottom-reflected sound is 

 usually dominant, even in water several thousand 

 fathoms deep. 



The reflection coefficients to which the different 

 curves in Figure 11 correspond should not be taken 

 as a measure of the amount of sound reflected by the 

 surface; It is, of course, virtually certain that most 

 of the sound reaching the ocean surface is reflected 

 back into the ocean in some direction, except possibly 

 when strong winds produce absorbing bubbles close 

 to the surface. However, some of this sound may be 

 reflected in directions quite different from the sound 

 reflected at an ideally flat, horizontal surface. Also, 

 the relative phases of the direct and surface-reflected 

 sound may be altered by the irregularities in the 

 ocean surface. As a result of these two factors, the 

 image effect to be expected for a flat, perfectly re- 

 flecting surface may be modified. Equation (3) is 

 then useful as a semi-theoretical, semi-empirical 

 formula for fitting the observed data. 



A somewhat different manner of presentation, 

 which includes all the data available at the time refer- 

 ence 11 was written, is shown in Figure 12. Here, all 

 available anomalies at certain fixed ranges are plotted 

 on a linear range scale. In such a plot, the short ranges 

 are too compressed to show the interference patterns 

 characteristic of the image effect. However, such plots 

 are very suitable if emphasis on the data at longer 

 ranges is desired. 



These data provide supporting evidence for the 

 general statements made in the discussion of Figure 

 11. The rise of the median curves for 600 and 1,800 c 

 is probably not real, but simply a result of observa- 

 tional selection; at the long ranges, the received sig- 

 nals are difficult to distinguish from noise, and only 

 those few signals can be measured which, because of 

 fluctuation, rise far above the noise level. Thus the 

 median curves in Figure 12 are probably considerably 

 higher at long range than they would be if all the runs 

 yielding data at short range could have been con- 

 tinued successfully to long range. 



5.2.2 



Absorption 



At frequencies above several thousand cycles, im- 

 age effect is usually unimportant, and in the absence 

 of temperature and salinity gradients, absorption be- 

 comes the chief effect modifying the inverse square 



