The data are acquired and stored in units of digital counts. The transfer functions of the sensors are given in "Response" files that are downloaded from the DMC at the same time as the data. (There is an option to apply the transfer function to the data while downloading the data from the DMC, but we chose not to do this. Instead, we used the IRIS-supplied function "evalresp" to compute the transfer functions and applied the transfer functions directly to the spectra.) Spectra in counts are converted to spectra in ground motion units (for example, acceleration in units of nanometers per second squared) by multiplying by the inverse transfer function. Figure F2 summarizes the inverse transfer functions for the 12 channels of data discussed in this report. The Guralp channels are "flat to acceleration" from ~10 mHz to the low-pass antialiasing filters at ~8 Hz. Geophones are designed to be "flat to velocity" at frequencies above their natural frequency, which in this case is 4.5 Hz. Because the antialiasing filters are applied at ~8 Hz, these geophones are not being used in this "flat to velocity" band. The gains for the vertical channels of both the Guralp and geophone were originally set 6 dB higher than for the horizontals. The gain for the vertical Guralp was decreased by 6 dB on JD360/2001 to reduce the chances of clipping during the water gun shooting as the JOIDES Resolution arrived on site. After this time the gains on all three channels of the Guralp were the same.
As a check on the transfer functions, we compared spectra between Guralp and geophone sensors for the same time window. Figure F3 compares vertical channel spectra at 1300 hr UTC on JD360/2001 before the gain change on the vertical Guralp. Within the band 0.1–10 Hz the low- and high-gain channels have identical spectra but the Guralp is ~3 dB (a factor of 2 in power) higher than the geophone. Vertical channel spectra computed just after the gain change show a similar offset between Guralp and geophone data (Fig. F4). It appears as if the gain change on day 360 is being handled correctly but that there is a discrepancy between the Guralp and geophone transfer functions for the vertical component. A similar comparison of horizontal component data shows identical spectra in this band for Guralps and geophones, as expected (Fig. F5). (Note that all spectra have been doubled to allow for energy in the negative frequencies. This is necessary to satisfy Parseval's relation.)
Instrument noise can be inferred from Figures F3, F4, and F5.