A VSP experiment was conducted during the logging run at Site 1115. The goals of the VSP experiment were to give accurate depth estimations of reflectors identified in the multichannel seismic (MCS) data, thereby allowing correlation of lithostratigraphic units, and to provide parameters with which to improve velocity processing of existing MCS data. A 300-in3 air gun was used to generate a source signal, which was received in the hole by the WST. The source signal was also recorded on a hydrophone close to the air gun (see "Vertical Seismic Profiling" in the "Explanatory Notes" chapter).
As part of the preparation for recording the VSP, the end of the drill pipe was positioned at 100 mbsf. A restriction in the hole prevented the WST from being lowered past 554.0 mbsf. The WST was raised to 552.2 mbsf, at which point the clamping arm was engaged, pushing the instrument against the wall of the hole. A total of twelve stations were occupied. Clamping intervals were variable (Table T16), largely because of poor hole conditions at the first four stations, or a better location was found on the caliper log from the triple combo run (see "Downhole Measurements"). A recurring (1.2-1.5 s) signal, with a peak at ~30 Hz, was a problem at all stations, and attempts to limit it largely governed the location of subsequent stations. High noise levels eventually caused the abandonment of the VSP. Shots that did not have a clear first arrival and showed high noise levels were flagged. At each station, the air gun was triggered a number of times. Each good shot was stacked, with firing continuing until the stack contained seven shots. Movement of the WST after it had been clamped against the side of the hole was minimal, or masked by external noise.
The recurring 30-Hz signal had a similar amplitude and power spectrum to the air gun source signal, and was received on both the geophone in the hole and the hydrophone close to the sea surface, even when the air gun was not fired. The noise was not seen on the WST signal when the instrument was returned to the bottom of the hole and positioned on the restriction at 554.0 mbsf, possibly because of better clamping conditions at that location. Identification of a source originating from the JOIDES Resolution was unsuccessful. Reorienting the ship relative to the prevailing currents and wind direction provided a minimal reduction in the amplitude of the noise. Although the signal was of a similar frequency to that of the ship's main propellers (judging from the revolutions per minute), no changes in operation had taken place since the last VSP experiment that could explain the new signal.
Interval velocities were calculated for stations at 552.2, 532.1, 505.0, and 490.1 mbsf (Table T16; Fig. F79). The results are variable, but fitting a straight line by linear regression gives an interval velocity of 2044 m·s-1 (Fig. F80). Physical properties (see "Physical Properties") and logging measurements (see "Downhole Measurements") of velocities in the VSP interval vary from 1757 to 4702 m·s-1. The high degree of variability comes mainly from lithostratigraphic Unit VI, characterized by frequent calcium carbonate-rich horizons that locally elevate the P-wave velocity. Interval velocities determined from the VSP experiment may have averaged out the effect of these high-velocity layers.
To convert between seismic traveltime and depth below seafloor, we examined the velocity information available from laboratory measurements and sonic logs to develop a model of the variation of velocity with depth. In this model, the more densely sampled velocities (DTCO; see "Downhole Measurements") measured downhole were used preferentially to values determined from laboratory measurements. Exceptions were from 0 to 152.41 mbsf and 779.9 to 802.5 mbsf, where log sonic velocities were absent. Because of an offset between the laboratory velocities measured using the PWS3 transducer and those of the PWS1 and PWS2 transducers (see "Compressional Wave Velocity"), the consistently 40 m·s-1 higher PWS3 (transverse) measurement was used. Where velocities measured using the PWS3 transducer were absent, the vertical velocity measured using the PWS1 transducer was substituted with a + 40 m·s-1 offset. The resultant velocity-depth function (Fig. F80) was passed through a 2-m Gaussian filter and resampled to 1 m before being used to convert from depth to two-way traveltime (TWT). To confirm the viability of the depth conversion, it was compared with VSP check-shot information. The P-wave transit time to the WST geophone at depth (Table T16) was properly corrected for the geometry of the experiment (see "Vertical Seismic Profiling" in the "Explanatory Notes" chapter) to give a direct and absolute tie between TWT and depth. Comparing this to the MCS data correlates horizons in TWT to depth. The VSP is limited to depths from 490.1 to 552.2 mbsf; therefore, this comparison was made only over a short depth interval. Table T17 shows that the difference between the VSP depth and that calculated for an equivalent time from the velocity-depth function ranges from 2.72 to 7.53 m, with an average of 4.86 m. We infer that depth conversions using our velocity-depth relationship are accurate to ~5 m above 552.2 mbsf.
A synthetic seismic trace was generated using velocity and density data from Site 1115. Density data were compiled from downhole logging measurements (RHOM) and laboratory index properties measurements. Similar to the compilation of velocity measurements, in situ logging measurements were preferred to laboratory measurements, except where they were absent or of poor quality (0-100 mbsf and 776 mbsf to the bottom of the hole; Fig. F81). Additionally, RHOM densities <1.7 g·cm-3 were removed from the data set below 500 mbsf. A number of synthetic seismic traces were computed following the application of different filtering parameters to the velocity and density data. The synthetic trace that best matched the data was generated by applying a 0.5-m median filter (to remove spikes), and an Akima spline (to resample the data to 1 m) to both data sets. An Akima spline was used in preference to a simple linear extrapolation because of the variation in measurement interval between the laboratory (meter scale) and downhole (centimeter scale) measurements. A reflection coefficient was subsequently calculated from the two data sets. Using the relationship derived above, the reflection coefficient was converted from depth to time (Fig. F82). A source signal, comprising the first 500 ms of the direct wave from the migrated MCS coinciding with Site 1109, was convolved with the reflection coefficient to give the synthetic seismic trace. Using the depth-time function derived above, both the synthetic trace and a number of migrated MCS traces close to the site were converted to depth (Fig. F82).
There are obvious problems with the correlations between the synthetic seismic data and the MCS data (Fig. F82). For example, a prominent doublet seen in the MCS data between 270 and 300 mbsf is not reproduced in the synthetic data and may be of the wrong polarity. Given that we have shown above that our conversion between time and depth is accurate, this indicates that the problem is with the parameters used to generate the synthetic seismic trace. Further work on this problem will include adjusting the filtering parameters of the velocity and density data, as well as experimenting with the source signal.
Figure F83 shows the correlation between the MCS data in time, the depth converted MCS traces, and the lithostratigraphy column (see "Lithostratigraphy"). Lines are drawn at 30-m intervals connecting the depth-converted MCS data and the lithostratigraphy column to the MCS data in time. The top of lithostratigraphic Unit X, which corresponds to what is interpreted as the top of the Miocene forearc sequence, is estimated to be at 2.21 s TWT and corresponds to the brightest reflector on top of the forearc sequence.