SUPERSONIC TRANSMISSION 



147 



computation of a curve of median transmission 

 anomaly. To obtain sucli a curve, values of the trans- 

 mission anomaly are read off at a number of prede- 

 termined ranges, such as every 500 yd. At each range, 

 the median transmission anomaly is noted. If these 

 median values are plotted against the range and the 

 resulting points connected, the resulting curve will 

 have the property that at each of the ranges at which 

 values were read it separates the actual curves into 

 two equally numerous portions. In a similar fashion, 

 upper and lower quartile curves of transmission 

 anomaly may be obtained. 



In general, the median anomaly curve will look 

 smoother than the individual curves making up the 

 sample, and "bumps" in the individual curves will 

 not show up if they do not all appear at the same 

 range. However, the median curve will be repre- 

 sentative of average transmission conditions, and the 

 quartile curves will provide a graphical measure of 

 the spread of the sample. 



Obviously, median curves of transmission anomaly 

 are valuable only if they are based on a fair-sized 

 sample. For this reason, separate median curves were 

 not constructed for all the classifications which were 

 established in the analysis reported here. 



Not all individual transmission curves extend out 

 to the same range. When transmission conditions are 

 relatively poor, the reading of the traces must be 

 stopped at a rather short range. When an appreciable 

 number of curves in the sample cannot be read at 

 the longer ranges, the median curve for the remainder 

 apparently turns upward; since this upward turn has 

 no physical significance, the median curve is stopped 

 short in such cases. 



In the computations summarized in this chapter, 

 median and quartile transmission anomalies were 

 determined at 1,000 yd and every 500 yd from there 

 on; the number of curves in the sample was noted at 

 each of these ranges. 



As a first problem, the degree of consistency be- 

 tween the UCDWR and the WHOI runs was investi- 

 gated. For this purpose, all runs over SAND with 

 downward refraction (NAN and CHARLIE), re- 

 gardless of hydrophone depth, were collected for each 

 institution separately, and median and quartile 

 curves of transmission anomaly plotted. These curves 

 are shown in parts (A) and (B) of Figure 3. There is 

 some evidence that the discrepancy of about 5 db be- 

 tween the curves is due, in part, to a different method 

 of calibration. While UCDWR has usually referred 

 transmission anomalies to the transmission level re- 



1000 2000 



RANGE IN YARDS 



Figure 4. Comparison of UCDWR and WHOI trans- 

 mission data over sand-and-mud bottoms with strong 

 downward refraction. 



corded at about 100 yd, WHOI has frequently ob- 

 tained the reference level by extrapolating the meas- 

 ured relative transmission anomalies backward. To 

 illustrate the relatively good fit which results from 

 vertical shifting of the curve, part (C) of Figure 3 

 shows the two median curves shifted so that they co- 

 incide at a range of 1,500 yd. 



