3.9 degrees 



Figure 11. Plot of y^ versus Rj for rock 

 outcroppings. 



A minor problem with the imagery of Figure 10 

 is that the return signal to each fish was produced by 

 the outgoing pulse from the near fish/ This was, 

 presumably, the result of mutual interference. 



DISCUSSION 



The results given in the preceding section clearly 

 indicate that stereoscopic mapping of the seafloor 

 from sonar data will require continuous and precise 

 measurement of fish depths, separation, and lateral 

 position in a fixed earth's reference frame. Because of 

 the low rate of scanning the seafloor (the result of an 

 energy propagation velocity of 5,000 ft/sec, com- 

 pared to the velocity of light of 186,000 mi/sec), the 

 only signal-processing approach to solving the slow- 

 scan problem is some kind of automatic bridging 

 process from one line scan to the next. This would 

 only work for continuous targets extending through- 

 out the swept area, for example, a long trench or 

 cliff. Moreover, the circuitry needed for a bridging 

 process of this sort would be more sophisticated than 

 the circuitry and the acoustic transducers, combined, 

 needed for measuring fish depth, fish separation, etc. 



It is within the state-of-the-art to measure fish 

 depths and fish separation with sufficient precision so 

 that, with same-side stereo scanning, the relative 

 target elevations are obtained with errors under 3 feet 

 in 100 feet (see Appendix A). However, because of 

 the likelihood of near-B points and the consequent 

 large errors in y^, it might be better to design a 



stereo-sonar system with a fish separation sufficiently 

 large to allow opposite-side scanning. With 

 continuous measurement of fish separation, the 

 rigging and auxiliary structures for such a towing 

 operation need not be too complicated. The idea of 

 two manned catamarans connected by a crosstrack 

 tension line might prove feasible. Large variations in 

 separation distance during towing would be tolerable 

 as long as the separation were precisely known for 

 each line scan. The problem with opposite-side stereo 

 scanning is, of course, that image correlation may be 

 impossible for asymmetrical seafloor features, such as 

 a cliff. 



From the data obtained at the 100-foot-depth 

 site the imagery problem can be avoided as long as 

 fish heave, peak-to-peak, is less than 4 feet for echo 

 ranges on the order of 100 feet. At greater depths, 

 say 600 feet and more, as the result of greater 

 tow-cable lengths vertical fish oscillations could be 

 kept under 4 feet even in sea states higher than zero. 

 Crosstalk or interference are no problem if tow 

 distances are unequal; and, in low-flying sweeps in 

 very shallow water (approximately 25 feet), control- 

 lable fins could be used to allow unequal trailing 

 distances without danger of grounding. 



Appendix C presents a description of a proposed 

 analog signal-processing system which eliminates the 

 need for interfacing a stereo operator's manipulations 

 with a digital computer. The proposed system com- 

 pletely eliminates the need for fusing two sonar 

 charts into a three-dimensional illusion; instead, the 

 system produces in real time a readout equivalent to a 

 stereo pair obtained photographically. 



Computation of target range, Rj, requires the equation: 



•^2 = 2R2 - Rj 

 where R, is the value read directly from the chart. 



12 



