METHODS

To avoid unnecessary duplication of text, we do not include in this paper detailed descriptions of the methods used to derive the physical parameters on board the JOIDES Resolution. We refer the reader to the appropriate chapter of the Leg 178 Initial Reports volume (Shipboard Scientific Party, 1999a).

All data are presented in meters below seafloor (mbsf) vertical scale. We decided to maintain this vertical scale for the following reasons:

1. Two meters-corrected depth scales have been generated for Sites 1095 and 1096—one was generated on board and was included in the Initial Reports volume, although it was not included in the Janus database (see "Composite Depths" sections in the site chapters of Shipboard Scientific Party, 1999b, 1999c). Subsequently, another corrected scale was proposed by Barker (Chap. 6, this volume) and was added to the Janus database. At the moment, we are not able to find criteria helpful to select one as the best.
2. For the purpose of this paper, the small vertical shifts (a few meters at most) imposed by the composite depth scale do not significantly affect the results.
3. By presenting the data vs. the original mbsf scale, we allow any user to apply his/her own preferred depth correction, if needed.

Multisensor Track Velocity

The output files from the multisensor track (MST) instrument typically contain errors resulting from velocity values that are unrealistically low (~1200 m/s), produced by voids in cores, or excessively high, resulting from the presence of dropstones in cores and core section edge effects (Fig. F2). To systematically clean the data, we manually removed the unrealistically low values and averaged the values at 3.5-m intervals. This interval spacing was selected to scale the measurements to a resolution that is appropriate to the available seismic reflection data, considering the nominal vertical resolution of seismic reflection data is equal to one-quarter of a wavelength (). Assuming that the generator injector (GI)-gun survey preserves a frequency of 130 Hz, /4 is 3.5 m for a formation velocity of 1800 m/s. In the process of averaging within each 3.5-m interval, the standard deviation is calculated and data points generating a standard deviation in excess of a certain value are excluded from the computation. The process of determining the discriminating standard deviation is iterative and is conducted with a critical eye by the operator. In our case, we decided to keep all values within eight standard deviations to reject only the values that have a very low possibility of belonging to the considered distribution.

Multisensor Track Density and Porosity

Bulk density is calculated as a function of gamma ray attenuation. Porosity is calculated assuming a constant specific gravity of the solid part, knowing the water content of the sediments. The dispersion of these data is much less pronounced than that of the MST velocity data. A procedure analogous to that applied for velocity data was applied to bulk density and porosity.

Hamilton Frame Velocity

This measurement is conducted manually by an operator who selects appropriate parts of the core. Data can be obtained with two transducers inserted as a fork in the soft sediment in two orthogonal directions (PWS1 and PWS2) or by using a pair of cylindrical transducers (PWS3) that are analogous to those mounted on the MST. To preserve homogeneity of the data, we used PWS3 data, complemented by MST data where possible. With the exception of evident errors, the data were used without any correction of the original file. Errors in the Site 1095 PWS3 data set included in the Leg 178 Initial Reports volume were identified between 56.522 and 82.523 mbsf and between 359.343 and 414.844 mbsf (Fig. F3). Velocity in these intervals is shifted systematically toward higher values because a problem occurred during acquisition. The error was identified on board and a correction was applied before generation of the plots presented in the Initial Reports volume. The data included in this paper are corrected.

Laboratory Density and Porosity

Porosity and bulk density were obtained as part of the index properties (IP) measurements in the laboratory. Because these measurements were conducted on discrete samples carefully selected by the operator, with a spacing that was usually every core section (1.5 m) and therefore well within the limit of 3.5 m selected as appropriate to the resolution of the seismic profiles, we decided to use these data preferentially with respect to MST data. With the exception of evident errors, the data were used without any correction of the original file.

Final Treatment Common to All Data Sets

Vertical profiles of velocity, density, and porosity derived from core logs or core samples were obtained by combining data from different instruments and different holes aiming at reconstructing composite profiles for each parameter at each site (Table T1). Finally, all the composite files were gently smoothed and resampled every 2 m after linear interpolation between data points.

Following the suggestions contained in pertinent sections of the Initial Reports volume (Shipboard Scientific Party, 1999b, 1999c), we utilized RHOM (lithodensity sonde [LDS] corrected) bulk density values and porosity values from the APLC (accelerator porosity sonde [APS] near-array limestone porosity corrected) files. No further correction was applied to the data. Sites 1095 and 1096 do not provide velocity information from sonic logs. Site 1101 was not logged at all. For evaluation of the quality of the data, we also used caliper logs.

To avoid duplication of text readily available in ODP and open literature, we suggest the interested reader consult Shipboard Scientific Party (1999a), Goldberg (1997), and Broglia and Ellis (1990) for an understanding of methods and acronyms used by ODP.

Only one vertical seismic profile (VSP) was carried out, at Site 1095 (Shipboard Scientific Party, 1999b). The spacing of the recording stations makes this data set more appropriate for a velocity check shot than for the generation of a proper vertical seismic profile. However, both on the ship and in this study, the data were processed to obtain a low-coverage VSP.

The interval velocities obtained on board are considered reliable. They were obtained as a straightforward division of the distance between stations and difference in traveltime. However, the VSP processing was delivered as a closed result, with no possibility of refining the onboard processing. Therefore, we reprocessed the data as follows:

1. Separation of upgoing and downgoing fields using a median filter;
2. Predictive deconvolution of upgoing events using 500-ms operator length and automatic prediction lag. However, because a GI gun was used, deconvolution is not critical;
3. Conversion of the upgoing field to two-way traveltimes (TWT);
4. Corridor stack;
5. Filter for display 12-24/60-80 Hz;
6. Prediction of the depth of the deeper reflector using a plot of TWT vs. depth; and
7. Plot of the interval velocities against stacking velocities at the nearest shotpoint (SP) (1250 m).

A note of caution on the available raw VSP data: the data were transferred via internet from the ODP Borehole Research Group database. We encountered three kinds of problem:

1. Data are in ASCII format rather than SEGY standard format. Additional information was added in manually created headers.
2. Traveltimes are neither total traveltimes from the source nor traveltimes from the seafloor. Evidently, the traveltimes were muted with no specification. A comparison with traveltime tables recovered from the shipboard recording and available in the published report allowed us to reduce, by application of a constant shift, the traveltimes of the database to traveltimes from the seafloor. Relative traveltimes were found to be coherent with shipboard data.
3. The total depths used on the ship were relative to the rig floor. In our calculations, we considered depth relative to sea level by subtracting 11 m (the distance of the rig floor from sea level). After this data reduction, we observed that the picking of first arrival is in agreement with that obtained on the ship, with a tolerance of one or two sampling intervals.

A velocity analysis was carried out on MCS profiles at Site 1096 prior to drilling to derive a detailed velocity structure useful to site evaluation and planning of drilling. In addition to the standard stacking velocity evaluation, an inversion of traveltimes of selected reflectors was applied to a 5-km-long segment of MCS line IT90-109 from SP 1880 to 1980 (centered on Site 1096) using an acoustic tomographic technique.

The reflection tomography algorithm adopted here is described by Carrion et al. (1993a, 1993b), and an example of its application to MCS data from the Antarctic margin is provided by Tinivella et al. (1998). A detailed description of the method is reported also by Tinivella et al. (Chap 16, this volume). The MCS data were collected in 1990 with the following parameters: the seismic source was a tuned 40-air gun array with a total capacity of 72 L towed 6 m below the sea surface; the 3000-m-long analog streamer had 120 traces, spaced every 25 m; and a 50-m shot interval provided 30-fold coverage.

Ten reflectors were picked from the seafloor to the base of Unit M4, including unit boundaries from M1 to M4 of Rebesco et al. (1997). Traces with the highest coherency of reflectors were found within the shorter offsets. Ten traces were selected for picking, from 932.5 to 1432.5 m from the ship. The picking was done on both common offset (such as that shown in Fig. F4) and common shot gathers. The tomographic inversion was applied to one of every four shots.

Seismic profiles shown across Sites 1095 and 1096 are the same as those described by the Shipboard Scientific Party (1999a). The two sections presented in this study were produced by rescaling the stack sections, sampling one of every four traces, and displaying the signal as variable density. We do not show MCS profiles across Site 1101 because we have no relevant information to add. We refer to Shipboard Scientific Party (1999d) for seismic displays across this site.