Adding to the study of Francesco, these channels, which also pop-out in the coherent channel study, did not qualify as unsafe for the previous safety channel study using O3b data (channel list).

More precisely, here is the list obtained in the coherent study:

• V1:ASC_SR_TX_SRCL_I_FS_mean
• V1:ASC_SR_TX_SRCL_I_10Hz_mean
• V1:ASC_SR_TX_DCP_phi_B1_mag_10Hz_FS_mean
• V1:ASC_SR_TX_DCP_mag_B1_Q_10Hz_mean
• V1:ASC_SR_TX_DCP_mag_B1_Q_FS_mean
• V1:ASC_SR_TX_DCP_mag_B1_mag_10Hz_FS_mean

These channels did not exist for O3, the SR mirror was not installed. They have a high chance of being unsafe, they are based on the main differential arm length DC read-out photodiodes (B1), so the same channel that serves as the starting point for having h(t). The channels are just B1 demodulated at ~491Hz (DARM line) and then at 4-5Hz (SR angular dither lines in TX/TY). One would need to do a safety study to check if this double demodulation is sufficient to render these channels safe. But without a study I would assume by default that they are not safe. Or at least I expect that whenever there is a glitch that covers the 480Hz-500Hz band, there will be a glitch in these V1:ASC_SR_{TX,TY}_DCP_{phi,mag}_B1_* channels too.

It is likely that the maximum frequency of these 25-minute glitches doesn't arrive as high as 490 Hz. In the attached histogram, I've plotted the maximum frequencies of all the omicron triggers (not the clusters) with GPS time within one second around the central time of each of the 25-minute glitches. I'm tempted to conclude that there is not much energy above ~400 Hz related to these glitches. Which of course is not an argument to claim that the identified channels are "" safe.