Well, I did a bit of experimentation this morning, and my narrow signals are definately NOT coming from the satellite. I aimed my BUD at the sat, and although my BUD is so warped that it doesn't work well on Ku, I got a reasonable spectrum for the vertical side of G16. See the first image.
In the first image, I've labeled the 14321 SR CBS Newspath station as "A", I labeled a couple of the 3978 SNG signals at "B", and it looks like there are a couple more to the right of the 14321 peak. Also labeled one of those wide 30,000SR Hughesnet type internet transponders at "C". Although this isn't the cleanest of scans, it's somewhat typical, showing the widths of various SR transponders.
In the 2nd image, you can see what I was (and still am) getting on G16V with my Fortec dish, and Invacom lnbf. You can see that the whole spectrum is just covered with interferrence from very narrow peaks. These peaks are NOT visible in the scan of the horizontal side, and are not visible if I move the dish over to the next sat on either polarity. It's just on G16 on the vertical side.
Anyway, I went outside and checked out what the dish was aiming at, and sure enough, it's looking straight at my power pole, which has a high voltage transformer on top. So I'm guessing that the power pole is putting out some strange noise that extends all the way into the Ku band somehow.
Anyway, it wasn't coming from the sat.
BTW, both spectra show a significant drop off in signal at the high end of the band. This is due to the fact that both went through a bad port on my DiseqC switch. The other ports aren't quite that bad, although they still lose some high end signal due to my long cable run.