Thanks. I decided to run the data through the light curve analysis tool, Tangra, even though I don't have super accurate timing data. Yep, it definitely disappears and comes back again.
Keeping in mind that the only timing I had was the file creation time of each image, so only accurate to the nearest second.
Predicted:
Event Time: 21:32:54 (+/- 7 secs)
Max Duration: 4.9 secs
Observed:
Event Time: 21:32:53
Max Duration: 4.2 secs
I'll call it a win.