towed side-scan 

 sonar fish 



Figure 5. Stereo-sonar tow system schematic. 



Although the manufacturer's specifications state that 

 the vertical sonar angle, d^ , is 20 degrees, it was 

 assumed that in the 100 to 200-foot range adequate 

 returns would be generated within a vertical angle of 

 about 50 degrees. 



To maximize stereo target-detection probability, 

 the fish-vector B was oriented toward the assumed 

 target area (marked by surface floats as shown in 

 Figure 7). This situation, of having B slanted down 

 instead of up (see Figure 8), creates a bad situation 

 with respect to the precision of determining target 

 elevation, y^. If the target is on the line passing 

 through the two fish (that is, on the extension of B), 

 the theoretical random error in y^ becomes infinite. 

 This is intuitively clear and is shown quantitatively by 

 the error-propagation equations of Appendix A. It 

 was believed, however, that the risk of having large 

 errors in y^ was worth taking because of the small 

 stereo target-detection probability associated with a 

 slanted-up fish vector. 



It was also suspected that a relatively large value 

 of the elevation angle, |3; (see Figure 8) would 

 decrease the one-fish target-detection probability 

 because of the small shadow associated with a large 

 j3j. A low-/? (low-flying) situation would cause large 

 errors in y^. 



Hence, with all of the above considerations, the 

 geometry of Figure 8 appeared to be optimum. A 

 better geometry might have been opposite-side stereo 

 (assuming that image blending is possible) with a 

 horizontal B component of around 200 feet. A 



Figure 6. Starboard section of towing outrigger. 



