On page two of this pdf, it shows a 608 caption data waveform as it would be on an analog composite video signal. No caption data, just the static waveform.
https://wiki.millersville.edu/downlo...=1393615787752
(The other info in the pdf is related to 708 Hi-def captions, and does not apply here.)
Your images could show a signal from some closed-caption encoder run amok.
If captions, the waveform should be on line 21, field 1. There are synchronizing run-in clock pulses on the left, and data bits on the right. The waveform should be from zero to 50 on the waveform scale. An active closed-captioning waveform will have smaller 'dashes' in the right hand area that change every frame.
If the encoder is unterminated, it would be roughly twice that level, and could cause other downstream interference.
Some cable and satellite boxes can have internal caption encoding chips, that re-constitute line 21 captions from a data stream, rather than from the station's video signal. This encoder could lose termination thru internal component failure, or lack of termination in the external connection to another device. Or, it could be fed an unterminated signal. Keep in mind that Sony, for example, had (mechanical) auto-switching BNC termination setup that occasionally didn't work.
The varying levels you see may be due to automatic gain control function, or malfunction.
So, set your scope to show you line 21, field 1, and see what you've got. (Hopefully you have a line-select scope.)
Other things can can populate the vertical blanking interval are:
VITS (vertical interval test signals -- color bars, etc. on a single line)
VITC (vert. int. timecode -- which will have smaller dashes than you see)
Nielsen rating information codes
Macrovision (basic, early version)
and stuff like that.
Chip