I looked at a few times pointed out to me by Aidan and Betsy and didn't find any evidence that the flow rate sensor glitches are coupling into h(t). I've attached one example where there are two fairly large glitches in the flow rate readout with no visible effects in the h(t) spectrogram from the same times.