Site 1095 is situated on the distal northeast flank of sediment Drift 7 (Rebesco et al., 1996, 1997). The drift is perpendicular to the margin and is bounded by channels (Fig. F2; see "Background and Scientific Objectives"). Previous seismic stratigraphic interpretations of the Antarctic Peninsula drifts can be found in Rebesco et al. (1996, 1997) and in McGinnis et al. (1997). Rebesco et al. (1996) identify six acoustic units, M1 through M6, above oceanic basement. Sites 1095 and 1096 were selected to sample the four upper acoustic units, M1 through M4, interpreted to correspond to two stages in the evolution of the drift: a drift maintenance stage (M1 and M2) and a drift growth stage (M3 and M4). Site 1095 penetrated to a depth of 570 mbsf recovering sediments from Units M1 to M4.
The existing MCS profiles across Site 1095, collected by the Osservatorio Geofisico Sperimentale of Trieste (Fig. F1A; also see "Seismic Stratigraphy" in the "Explanatory Notes" chapter and "Appendix" and Fig. AF1, both in the "Leg 178 Summary" chapter), have been examined to establish the seismic stratigraphy. Physical properties measurements from cores and results from downhole logging are used to correlate the section at Site 1095 with the seismostratigraphic units and to assign true depth to seismic reflectors. At each site, the acoustic units are described and tied where possible to the other drill sites.
Bulk densities for the formation have been derived from MST GRAPE measurements with a 2-cm spatial resolution, index properties measurements (1.5-m spatial resolution), and lithodensity logging data (HLDS, measurement separation ~15 cm [Fig. F55]). The results from index properties measurements agree very well with the in situ properties from the downhole logging. Compared to these two data sets, the GRAPE densities show generally lower values. This difference increases dramatically (~0.3 g/cm3) when the coring method changes from APC to XCB at 205 mbsf at the top of Core 178-1095B-14X, probably from the larger air/water gap between liner and core and the higher degree of core disturbance produced by the XCB. Two different density models have been tested using (1) only index properties data and (2) a combination of GRAPE density (0-150 mbsf) and downhole logging data (150-560 mbsf).
Three differently derived velocities are available. The MST logger provided continuous data (4-cm spatial resolution) down to 200 mbsf (within the APC-cored part of the holes) and only sparse data between 200 and 280 mbsf. Single Hamilton Frame (PWS3) measurements (one every section) provide high-quality data for the deeper parts of Site 1095. In Figure F56, MST data, interval velocities derived from a downhole seismic experiment (using the WST and a 2500-cm3 two-chamber GI air gun), and the Hamilton Frame measurements are plotted for comparison. Most of the MST-derived P-wave velocities are slightly higher than the interval velocities from the downhole seismic experiment, believed to provide the most accurate results (Hardage, 1985). Two different velocity models have been tested. The first model uses MST data (0-209 mbsf) and Hamilton Frame data (209-543 mbsf); the second model also combines MST and Hamilton Frame data but from different depth intervals (PWS3 = 0-80 mbsf, 209-560 mbsf, and MST = 80-209 mbsf).
Two differently derived signals of the GI gun that was actually used during the seismic site survey were employed for the seismic models. Six randomly chosen traces were extracted from a digital far-field recording of multiple firing of the gun in water (recording distance ~300 m) (Fig. F57C). The signals were brought in phase, then stacked and resampled at 0.5 ms using a cubic interpolation function (Fig. F57A). The signal (wavelength = 10 ms) contains a continuous-energy spectrum up to 350 Hz (Fig. F57B).
A second source signal was generated by extracting strong, coherent seafloor reflections from the MCS profile across the drill site. The signal was processed as for the far-field signature (above) (Fig. F57D, F57F). The seafloor signal has a much longer wavelength (40 ms) and consists of a negative onset followed by two positive excursions and a final negative one (Fig. F57D). The derived signal has a dominant frequency range of 10 to 250 Hz (Fig. F57E). Differences between the frequency spectra of the signals could be explained by the low-pass filter effect of a long path through seawater and the uppermost seafloor sediments and by anti-alias filtering applied during data acquisition.
The results from 12 geophone stations of a downhole seismic experiment have been processed to a six-trace VSP section (Fig. F58). Additionally, the traveltime information was used to compare a standard traveltime/depth relationship (Carlson et al., 1986) to a second-order polynomial extrapolation, a linear extrapolation, and a combined linear-polynomial model (Fig. F59). The polynomial model, the Carlson et al. (1986) relationship, and the combined polynomial-linear model show good agreement and allow the depth to oceanic basement at 650 ms one-way traveltime to be estimated as 1250 mbsf (Fig. F58). VSP reflectors should be compared to the multichannel reflectors and tied to depth. Unfortunately, the low number of stations (12) and their wide spacing (average: 33 m) prevent this. The chosen station spacing introduces a low-pass anti-alias filter (~25 Hz) and greatly reduces the coverage up the hole. As a result, only strong reflectors from the lower part of the drilled section and below are visible in the stacked VSP section. One of these reflectors, for example, belongs to the sediment/basalt interface (Fig. F58). The WST deployment at Site 1095 was mainly designed to provide a checkshot survey (i.e., one-way traveltimes to stations at known depths in the hole and interval velocities between those stations).
The raw velocity and density data were carefully evaluated and corrected for obvious artifacts. In the case of the MST, the 1.5-m core section end-effects (in P-wave velocity and density data) and density lows caused by core disturbance were removed. The velocity and density data were preprocessed and cleaned in 20-m sections according to the methods described in "Seismic Stratigraphy" in the "Explanatory Notes" chapter. Special care was taken with anomalous density values within the downhole logging data caused by overwidened hole sections. These artificial density lows have been manually removed. Model A data (index properties density, PWS3 and MST velocity) were resampled at a 0.8-m spatial resolution. For Model B (logging and MST density, PWS3 and MST velocity), the data resolution was degraded to 0.6-m spacing. Each acoustic impedance model was convolved with the far-field and seafloor signals. The velocity/density data, the impedance curve, the reflectivity coefficients, and two unfiltered synthetic seismograms are displayed for each model (Figs. F60 [Model A], F61 [Model B]). A depth axis is displayed next to the time scale to allow convenient traveltime/depth conversions.
The most eye-catching difference between synthetic traces is not caused by the differences in the velocity and density data but by the different source signals used for the convolution. The higher frequency content of the far-field signal and its shorter signal length result in a higher frequency, more detailed synthetic trace. Figure F62 depicts a comparison of the two trace spectra from Model A and the corresponding spectra of the signals used. Because shape and frequency range are nearly identical, there was no major loss in information during data subsampling and interpolation.
All four calculated synthetic traces are interpolated to a 1-ms resolution (1000 Hz) and subsequently filtered using a high-order, zero-phase equi-ripple band-pass filter (pass band = 20-110 Hz, attenuation = 60-70 dB, filter order = 170-220 [Fig. F63]). Five of these traces are then shown together with 42 traces of the field seismic profile (Figs. F64, F65, F66, F67). A zero-phase Butterworth band-pass filter (pass band = 10-110 Hz) and a 150-ms automatic gain recovery window were applied to the sorted and stacked data set. Additionally, the total time window and the delay were reduced, and the 2-ms data were interpolated to 1 ms to match the time resolution of the synthetic trace. The optimal filter parameters were adjusted to values used during the actual processing of the precruise survey data. Different amounts of time-invariant gain were applied to equalize the overall amplitude appearance of the field and synthetic traces. Possible tie lines connect synthetic and survey reflectors. Because of polarity changes within the synthetic, correlation between positive and negative half-cycles are possible.
Each data set has its own problem zones that show poor or no correlation with the survey data. For example, a zone of strong reflections between 300 and 350 mbsf appears in all four synthetic runs but does not fit the survey data. The best working preliminary model is Model A (index properties density and PWS3 velocity) in combination with the seafloor reflection as a source signal (Fig. F65).
Both velocity models allow an accurate time/depth conversion. The two-way traveltime to the base of Hole 1095B (~654 ms) is defined within 20 ms (see Table T41). The extrapolated VSP curve of Figure F59 has been used to create the table. Most of the important reflectors and seismostratigraphic boundaries show up within the synthetic seismograms.
In Figure F58, the six-trace VSP section and the analogue seismic data are shown at the same scale. The boundaries of the major seismostratigraphic units and the maximum penetration time are marked. The slight time offset (10-20 ms) below seafloor between the analogue and the digital data can be partly explained by the negative onset of the seafloor reflector.
Direct comparison between the multichannel records and the synthetic seismogram, the physical properties data, and the lithologic units suggests subdivision of the seismic units (Figs. F58, F68, F69).
The acoustic stratigraphy, in conjunction with the drilling data, places constraints on the age of depositional events for the distal continental rise sequences (Fig. F70). Several general observations can be made from the seismic profile across the distal drift site.
Seismic Unit I has been interpreted to show hemipelagic drape. The upper 50 m of seismic Unit I correlates with lithologic Unit I (see "Lithostratigraphy"). Lithologic Unit I is characterized by fine-grained, diatom-bearing silty clay and clay, alternating with silty clay with sand grains. These deposits are believed to form by the slow sedimentation of biogenic-rich facies as hemipelagites in a regime of weak bottom currents. Seismic Unit I extends ~10 m deeper than lithologic Unit I, in a transitional interval characterized by an increase in ice-rafted debris in the laminated silt and mud facies. The base of Unit I is marked by a reflector that is locally an erosional unconformity. The depth of this unconformity correlates with a hiatus in the magnetostratigraphic record (see "Paleomagnetism").
Seismic Unit II includes much of lithostratigraphic Unit II (see "Lithostratigraphy"). Lithostratigraphic Unit II has repetitive successions of turbiditic sediments and structureless bioturbated intervals containing enhanced concentrations of ice-rafted debris (see "Lithostratigraphy"). Truncation and onlap of reflectors in seismic Subunit IIa, and the mostly horizontal parallel and subparallel reflectors below seismic Subunit IIa, are both compatible with overbank turbidite deposition from a nearby channel. At ~435 mbsf within seismic Unit III, there is a downward lithologic transition from turbidites to the finer grained parallel-laminated siltstones and claystones that characterize lithologic Unit III (see "Lithostratigraphy"). The type of ice-rafted debris also changes downcore from scattered sand, granules, and small pebbles (i.e., a few millimeters) in lithologic Unit II to large-size pebbles (centimeter size) in lithologic Unit III. The more uniform sedimentary succession of lithologic Unit III is compatible with the regular reflector pattern below.
The ages assigned to the seismic units recognized at Site 1095 (Fig. F70; Table T41) are based on preliminary paleomagnetic data from cores (see "Paleomagnetism"). Seismic Unit I is Pleistocene in age (0-1.7 Ma). The base of seismic Unit I corresponds to a possible hiatus in the paleomagnetic data (see "Paleomagnetism"). Seismic Unit II is Pliocene to late Miocene in age (1.7 Ma or older, depending on the extent of the hiatus, to 7.76 Ma). The Pliocene-Miocene transition occurs within Subunit IIc. Seismic Unit III is older than 7.76 Ma.
At Site 1095, seismic Units M1 to M4 of Rebesco et al. (1997) were drilled. The base of seismic Units IIa and IId correlate with the unit boundaries M1/M2 and M2/M3, respectively. The bottom of the hole reached the boundary between seismic Units M3 and M4.