Problems with OBS data and how to handle them
IN PROGRESSThe instrument disk spins up every XXX samples and writes for approximately XXX minutes. This creates a step function in the data that should not be confused with the data (provide a sample plot)
The broadband OBSs have hourly glitches that interfere with long-period studies. They mostly affect the vertical channel. Nicolas Leroy has developed a method for removing these glitches, which should be published in the next year or two. If you are affected by these glitches, please contact us and we will show you the current state of Nicolas’ work.
The seafloor environment has several noise sources that mostly affect wideband/broadband data, most notably microseisms (0.1-5 Hz), infragravity waves (0.001-0.03 Hz) and currents(0.001-0.1 Hz). The infragravity wave noise can be removed from the seismometer signal using the differential pressure gauge data (Webb and Crawford , 1999) and the current noise can be removed from the vertical seismometer data using the horizontal channel data (Crawford and Webb, 2000). No way has been yet found to remove the microseism noise, which corresponds to Rayleigh waves, but they do provide a nice source for ambient noise studies!
Like most recent seismometers, our OBSs use Delta-Sigma digitisers that oversample the data, then decimate it after applying FIR anti-alias filters. This creates a time delay in the signal and some ringing near the Nyquist frequency. Our digitisers (Cirrus Logic CS5321/22 chipset) use a linear phase FIR filter in the last stage, which becomes zero-phase (acausal) when corrected for the time delay (done in the OBS software). This is good for waveform-based analysis, but bad for first-arrival picking or for identifying first motions.
The group delay is 29 samples (for example, 232 milliseconds at 125 sps), which is corrected for in the OBS software. The figure shows an example of the impulse response for a sample rate of 1000 sps: the response is the same in time and smaller in amplitude at lower sample rates (for example, at 62.5 sps, the peak value is -5,240,723).
If you need to pick first arrivals or identify first motions, you can post-process to change the linear-phase filter to a minimum-phase filter, using the procedure outlined by Scherbaum and Bouin (1997) and Chapter 8 of Scherbaum (2007). We have the filter correction parameters and the Scherbaum code necessary to do this. Below is the result of this correction applied to 62.5 sps data recording a local earthquake.
Webb S. C. and W. C. Crawford, Long period seafloor seismology and deformation under ocean waves, Bull. Seis. Soc. Am., 89(6), 1535-1542, 1999,
Crawford W. C. and S. C. Webb, Identifying and removing tilt noise from low frequency (<0.1 Hz) seafloor vertical seismic data, Bull. Seis. Soc. Am., 90(4), 952-963, 2000, doi:10.1785/0119990121
Scherbaum F. and M. P. Bouin, FIR filter effects and nucleation phases, Geophys. J. Int., 130(3), 661-668, 1997.
Scherbaum F., Chapter 8: The Digital Anti-Alias Filter, in Of Poles and Zeros, 2007. pp 117-136.