Correcting the vertical component of ocean bottom seismometers for the effects of tilt and compliance

Typically there are very high noise levels at long periods on the horizontal components of ocean bottom seismographs due to the turbulent interaction of bottom currents with the seismometer package on the seafloor. When there is a slight tilt of the instrument, some of the horizontal displacement caused by bottom currents leaks onto the vertical component record, which can severely increase the apparent vertical noise. Another major type of noise, compliance noise, is created when pressure variations associated with water (gravity) waves deform the seabed. Compliance noise increases with decreasing water depth, and at water depths of less than a few hundred meters, compliance noise typically obscures most earthquake signals. Following Crawford and Webb (2000), we have developed a methodology for reducing these noise sources by 1-2 orders of magnitude, revealing many events that could not be distinguished before noise reduction. Our methodology relies on transfer functions between different channels. We calculate the compliance noise in the vertical displacement record by applying a transfer function to the differential pressure gauge record. Similarly, we calculate the tilt-induced bottom current noise in the vertical displacement record by applying a transfer function to the horizontal displacement records. Using data from the Cascadia experiment and other experiments, we calculate these transfer functions at a range of stations with varying tilts and water depths. The compliance noise transfer function depends strongly on water depth, and we provide a theoretical and empirical description of this dependence. Tilt noise appears to be very highly correlated with instrument design, with negligible tilt noise observed for the 'abalone' instruments from the Scripps Institute of Oceanography and significant tilt observed for the Woods Hole Oceanographic Institution instruments in the first year deployment of the Cascadia experiment. Tilt orientation appears relatively constant, but we observe significant day-to-day variation in tilt angle, requiring the calculation of a tilt transfer function for each individual day for optimum removal of bottom current noise. In removing the compliance noise, there is some distortion of the signal. We show how to correct for this distortion using theoretical and empirical transfer functions between pressure and displacement records for seismic signals.