ANDERSON: VERTICAL NOISE DISTRIBUTION 



Fortunately, all of the 100-Hz beams steered just off the horizon- 

 tal had nulls in the horizontal (see Figure 6) so that only the horizon- 

 tal beam (90 degrees) should be contaminated. This is more apparent at 

 the shallower receiver depth of 3781 meters (Figure 10) . The high 

 noise plateaus are essentially independent of frequency. Still shallower, 

 the plateau at the lower frequencies continues to grow and is quite 

 discernible (Figures 11 and 12) . At 100 Hz, however, the wider plateau 

 overlaps with its images from the grating lobes. Returning to the deep 

 receiver on the first day (Figure 13) , the splitting is apparent. It 

 should be remembered that the out-of-band noise associated with the band- 

 pass filter characteristics makes it impossible to infer the attributes 

 of the low-noise directions near the vertical. 



Figures 14 through 19 represent the second day's measurements for 

 a denser set of depths. Again, broadening of the noise is apparent in 

 shifting from 100 Hz for deep receivers (Figures 14, 15, and 16) to the 

 lower frequencies for the shallow receivers (Figures 17, 18, and 19) . 

 There are some differences between the results on these two days. 

 Notice the more rounded peak at 23 Hz on the second day versus the 

 squarer peak on the first (Figure 12) . On the second day, a freighter 

 nearby may have influenced the noise. 



Figure 20 illustrates the response of the system to narrowband 

 signals rather than broadband noise. There is sufficient signal-to- 

 noise in this case that out-of-band noise is not a problem. When the 

 filter sidelobes are not important, the beamformer has a much greater 

 dynamic range. 



871 



