Short glitches centered at a frequency of 50Hz. At LIGO Hanford, these were found to be related to air compressor motors switching on and off at the end stations. These were solved on September 29, 2015 (we think) at LIGO Hanford by replacing the vibration isolators (rubber feet) on air compressors. However, a few weeks' worth of these glitches were in the O1 data set prior to the fix. LIGO electronic logs with more information are available here:
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=21436
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=22081
https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=22119
During later observing runs, other causes have produced glitches at similar frequencies, often of somewhat longer durations. See, for example, Slow Parallels.
Audio of Air Compressor (with associated spectrogram) Make sure to wear good headphones!