None of the seismic profiles used in this study have directly measured seismic wavelets. Therefore, the seafloor reflector was used in each case to determine the wavelet. Because of the likelihood of shallow reflectors at Sites 1166 and 742, the seafloor reflector was extracted exclusively from shotpoints distant from the drill sites. This was not the case for Site 1165, which was cored in a sediment drift of laterally uniform sedimentation and was therefore unlikely to contain significant local reflectors.
The source wavelet used for the Site 1165 synthetic seismogram was extracted from an average of the seafloor reflections of two widely spaced traces (shotpoints 1000 and 1191, the latter near the drill site) (Fig. F4A). Internal consistency of the wavelet over this substantial distance confirms the reliability of using the seafloor reflection (Fig. F4A). Unfortunately, the seismic source was a water gun and so the wavelet is tripeaked, spans a width of ~90 ms, and consists of a ~23-ms precursor. It was resampled to 0.2 ms, the same as the Site 1165 composite impedance log, for convolution.
Seafloor reflections at three widely spaced shotpoints (300, 1102, and 2461) were initially considered for wavelet extraction (Fig. F4B). All three shotpoints show two peaks at consistent spacing. However, two of the shotpoints (300 and 2461) show a possible third peak, but at slightly different arrival times and with different shapes. The third shotpoint (1102) does not show this third peak, which could be caused by a shallow reflector, and it was therefore considered to have a character most representative of a simple seafloor reflection. The lithologic section recovered at Site 1166 contained ~2.6 m of hemipelagic sediments overlying diamictite. The contact between these two units may produce the third peak found near the seafloor reflection at the shotpoint nearest the drill site. The wavelet that was used, therefore, consists of two peaks and is ~37 ms long. Including the third peak degraded the match between the synthetic seismogram and seismic section. No low-amplitude precursor was recorded.
The wavelet for the Palmer line was extracted from the average seafloor reflection of shotpoints 2004-2006 (Fig. F4C). The shotpoints nearest both Sites 1166 (shotpoint 2637) and 742 (shotpoint 4999) were very ringy, possibly the result of a shallow reflector occurring at the contact between surficial hemipelagic sediments and underlying diamicts. This is consistent with the inferred shallow reflector based on extraction of the seafloor reflection from the AGSO line as well. The seismic profile gathered during Leg 119 over Site 742 recorded a direct arrival from the water gun aboard the JOIDES Resolution and shows a character similar (but of longer duration) to the inferred wavelet for the Palmer line (Shipboard Scientific Party, 1989b), whereas the seismic section is ringy in the near surface.
The seafloor reflection at shotpoints 2004-2006 consisted of two peaks and one trough over a duration of ~27 ms. At the drill sites, the seafloor reflections contained five peaks and lasted 50 ms. Synthetic seismograms created with the drill site-based wavelets appeared too ringy throughout, whereas synthetic seismograms created with the wavelet from shotpoints 2004-2006 closely matched the nearby seismic traces.
Downhole velocity logging was undertaken for only ~40% of Hole 1165, and most of this log was unreliable, characterized by abundant cycle skips (Shipboard Scientific Party, 2001b). As a result, we used a composite velocity log to create the impedance log for Site 1165. This composite was derived primarily from logging density data and from core index velocities above and below the logged interval.
For synthetic seismograms, there are a number of advantages to using logging rather than core data: (1) logs are continuous, whereas core data are limited to recovered intervals; (2) logging data are recorded in situ, whereas cores expand due to pressure release as they are brought to the surface (Hamilton, 1976); and (3) logging data are unaffected by drilling disturbance. Drilling disturbance in cores increases considerably whenever extended core barrel (XCB) or RCB drilling techniques are required. The latter two were used for a majority of coring at Site 1165, with XCB coring used to the base of Hole 1165B and RCB used entirely for Hole 1165C.
The density log for Site 1165 was much longer (175-985 mbsf) and more reliable than the sonic log. Nevertheless, the litho-density tool, which measures bulk density, requires contact with the borehole wall. Where the borehole is ragged or too wide for good tool contact against the wall, the bulk density data are typically unreliable. In contrast, the resistivity log is relatively insensitive to hole conditions. Both resistivity and bulk density primarily respond to changes in porosity, so comparison of those logs can indicate regions of suspect density data. We subjectively edited the density log to remove spurious intervals, based on comparison to the resistivity, caliper, and bulk density correction factor logs. The edited regions were generally no more than a few meters thick, and linear interpolation was used to span regions of deleted density data. From the seafloor to the top of the density log at 175 mbsf and from the bottom of the density log to the total depth of the hole (985-999 mbsf), index velocities were used. Vertical core velocities were typically used since this direction is more representative of the formation as seen by vertically penetrating seismic waves. From 94 to 175 mbsf, however, vertical velocities were not recorded, so horizontal velocities were used. The moisture and density-derived (index) densities above the log were corrected to in situ conditions (Fig. F5A) based on comparison between the index and logging densities within the logged interval.
Where index velocities were used, the values had to be converted to in situ conditions, as cores rebound when exposed to atmospheric pressure (Hamilton, 1976, 1979). This correction was based on comparison of logging velocity to index velocity in the several intervals for which sonic logging data were reliable. Reliability of the velocity log was determined based on character match of the P-wave velocity log to the more robust resistivity log, in a manner similar to the comparison made between the resistivity and density logs. As sonic velocity is also primarily responsive to changes in porosity in the marine system, the first-order sonic and resistivity logs should also show similar character. Nine index samples were obtained within intervals of reliable sonic log. In addition, three core samples, taken at depths of 685.8, 878.1, and 916.6 mbsf (well below the bottom depth of the velocity log), were measured in a velocimeter to compare velocities at atmospheric and in situ pressure (Fig. F5B).
The velocities of the core samples were measured in a New England Research velocimeter at in situ differential pressures after the samples were saturated in a fluid of seawater chemistry (Table T1). Pore pressures were atmospheric, so confining pressure was equal to differential pressure. Velocimeter accuracy was confirmed by replication of Amoco measurements (Sondergeld and Rai, 1993) on Ferron sandstone samples (Jarrard et al., in press) for both compressional velocity and shear velocity and for both saturated and dry states. These three core plugs were collected during the leg and selected for velocity measurement based on their retention of internal cohesion and cylindrical shape.
A plot of in situ corrected index density vs. in situ corrected index velocity (Fig. F5C), for all samples in which both parameters were measured, demonstrates a close relationship between these two porosity-sensitive parameters. We used this relationship to convert logging density values to pseudovelocities. Index density was at no point converted to pseudovelocity as logging densities were. For the portions of the composite velocity curve derived from core data, index velocities were directly used.
The composite velocity and density logs were multiplied together to create the impedance log (Fig. F6). The practical effect of using an impedance log rather than just the velocity log is minimal for this site because the majority of the velocity data are derived from the downhole density log. However, in the nonlogged intervals, some independent character is apparent between density and velocity, although as porosity is the dominant control on both, this is a minor effect.
The resulting synthetic seismogram (Fig. F6) underestimated the depth of the basal reflector by ~7 m in comparison to the original seismic section but matched the depths of the overlying major reflectors. This was corrected by applying a 3% velocity decrease to the pseudovelocity log below 5730 ms. Poor hole conditions may induce velocity shifts of a few percent (e.g., Shipboard Scientific Party, 1989a) through a variety of mechanisms such as borehole wall rebound, wash-outs of loose sediments, or averaging bias across very thin layers or individual cobbles of anomalous velocity.
A reliable P-wave velocity log was collected from 367 to 30 mbsf at Site 1166 and forms the main basis of the synthetic seismogram for this site. Since only two index samples were collected from the seafloor to the top of the log, the region of the composite velocity section above the log was based on an assumed seafloor velocity of 1.5 km/s, increasing linearly to 1.9 km/s at a depth of 2.6 mbsf, and then increasing linearly at a rate consistent with the average increase of the logged portion of the same diamict unit (Fig. F7). The change in velocity at 2.6 mbsf is based on observed hemipelagic sediments in the upper 2.6 m of the core, changing to diamict below.
Because of the poor hole conditions at Site 1166, the density log would have required extensive editing. As a result of poor core recovery, index densities were infrequently measured and were probably not representative of overall lithologic variations. Consequently, the composite velocity log was taken as the impedance log (Fig. F7) by assuming a constant density. Since porosity is the dominant control on both velocity and density, using only one as the impedance log is justified here.
Although the velocity log for Site 1166 is generally of good quality, it apparently still contains minor artifacts. After generation of an initial synthetic seismogram for the AGSO line, two spikes, from 114.8 to 120.5 mbsf (0.1087-0.1142 seconds below seafloor [sbsf]) and 304.7 to 309.9 mbsf (0.2722-0.2764 sbsf), were removed from the composite velocity log as well as a short region that produced an anomalous reflection coefficient from 295.1 to 296.2 mbsf (0.263-0.264 sbsf).
The Site 742 impedance log was also created from the velocity log only (Fig. F8). We assumed a constant density for essentially the same reasons as for Site 1166. Sonic logging data were collected from ~33 to 270 mbsf, whereas the total depth of Site 742 was 316 mbsf. The lower 46 m of the hole contains 21 index velocity measurements, but they were not included in the impedance log for two reasons. First, these index velocities were anomalously low. All of these index values were measured along the z-axis, which is most sensitive to core disturbance through loss of internal cohesion caused by separation along lithologic planes. Such separation is common in RCB or XCB sections through rotational shearing of semilithified sediments. Second, the sample spacing of index velocities is insufficient to pinpoint impedance contrasts over short intervals. Index values for the region above the log were also not incorporated into the impedance log, primarily because of low sampling rate. They were used, however, as a guide for the ramping function that bridged the interval from the seafloor to the first log value.
A synthetic seismogram was created for Site 742 during Leg 119 (Shipboard Scientific Party, 1989b). That synthetic seismogram also only used the velocity log as the impedance log, with no use of core samples either above or below. Above the log, a constant velocity of 2.2 km/s was used, which differs from our assumption. We assumed a ramping function that accounts for the ~5.7 m of surficial hemipelagic sediments recovered at the seafloor that overlie the diamict unit that extends beyond the top of the logged interval (Shipboard Scientific Party, 1989b). It is quite possible that the stratigraphic contact between the hemipelagic sediments and the diamict produces a reflector, but not enough data are available to characterize it. Therefore, the ramping function assumes a velocity of 1.5 km/s at the seafloor, linearly increasing to 2 km/s for the top of the diamict. The diamict velocity is then assumed to increase linearly from 2 km/s to the average measured value at the top of the logged section. This produces an extrapolated continuation of the overall trend (Fig. F8).