I found the symmetric non-Gaussian feature around 340-400 on Gauch. (For high resolution, here) So, I follow-up this frequency band.

Whitened spectrogram shows same shape as non-Gaussianity.

On the series of ASD (see attached plots), we can see the symmetric feature between 340 and 400Hz. From the ASD, this is not related to line noise and looks like the bump with time-variation. By Yokozawa-san's comment, this is related to PR3 alignment. So, I'm monitoring DARM signal of frequency band at the control room.