DISCUSSION OF RESULTS

The sedimentary successions at the two drill sites considered here differed mainly in sedimentation rate. There were only minor differences in lithology or measured physical properties between them, or in coring procedures. They therefore shared most correlation problems and are best considered together here.

The most useful physical property for correlating between cores was MS. The high terrigenous component and fine grain size of the sediments and high sampling rate assured MS variability, and the sediment remained largely (but perhaps not completely) free from diagenetic alteration. Occasionally, there was a dramatic difference in MS between the two cores, involving an extremely high value in one hole over a short interval, attributable to the influence of a single large high-MS glacial erratic. The second most valuable physical property was one of the chromaticity parameters (a* and b*) derived from the multispectral color scanner measurements (Shipboard Scientific Party, 1999a). The other available parameters showed lower variability or were less reliable. MS was used for almost all ties. Exceptions were the ties from Core 178-1095D-2H to 3H via Core 178-1095A-3H, for which no MST data were available, and the similar lack of MST data for most of Core 178-1095D-6H. In both cases, chromaticity parameter a* was used to effect the ties.

There are three minor exceptions to the general convention at Site 1096. Core 178-1096C-7X was planned as a short core to regain alignment after drilling 24 m but recovered 124% (7.11 m from a nominal cored length of 5.7 m). The additional core is assumed here to have sampled the sediment above the nominal core top. Core 178-1096B-24H and cores below and Core 187-1096C-18X and cores below, both appended, are each shifted up by 1.0 m within the spliced sections to avoid overlap associated with excessive recovery (see the changes in offset in Table T2).

The greater inherent uncertainty attached to correlations between only two cores, compared with correlations among three, has already been mentioned. In addition, there are other factors that affect or illustrate the reliability of the results that should be mentioned here.

First, SPLICER provides a numerical estimate of the correlation coefficient over a specifiable interval (usually maintained at ±1 m) surrounding each correlation. The utility of this estimate rests on the assumption that the cores being correlated are undeformed by the coring process or are deformed simply and to the same extent. However, it was clear that at Sites 1095 and 1096, the cores were deformed to different extents. It is now recognized that cores may be deformed by the coring process to different extents (e.g., Feary, Hine, Malone, et al., 2000; Wang, Prell, Blum, et al., 2000); the upper part of a single APC core may be expanded, whereas the lower part may be compressed. Ship heave, particularly in the upper part of a hole where the bottom-hole assembly (BHA) may not be fully supported by the sediments, may also be expected to affect both core recovery and distortion of the recovered core. In addition, a 10% expansion of the correlated meters composite depth (mcd) scale compared with the meters below seafloor (mbsf) scale is generally quoted and attributed by some to the elastic response of the sediment to release of ambient pressure (e.g., Moran, 1997).

Cumulative offsets at other ODP sites, although in many cases averaging 10% of the depth, can vary widely and may be dependent on compaction and lithology. Figure F3 shows the cumulative offset depths calculated for each core at Sites 1095 and 1096 as a putative measure of the quality of the correlation. They are neither uniform nor averaging 10% expansion, and correlation at Site 1095 ends at ~85 mbsf, whereas at Site 1096, multiple coring was continued intermittently to 260 mbsf in an effort to obtain a complete sample of the succession. The measured offsets at Site 1095 are cumulatively negative, reflecting nett compression rather than expansion. However, there is perhaps a similar change in offset with depth at both sites. Through the top 50-60 mbsf, offsets decline to a minimum from which there is a sharp rise (the slope at Site 1096 reaches a maximum of 8%), reducing to <1% at greater depth where correlation was less frequent. Some of the correlated cores below 160 mbsf at Site 1096 were obtained using the extended core barrel (XCB). The very different APC and XCB coring techniques would have interacted differently with the sediments, but this cannot explain the unusual results at both sites at shallower depth.

I propose an explanation for the apparent compression. Almost all the compression (negative offset gradient) occurs at shallow subbottom depth at both sites. All of Leg 178 drilling suffered from disturbance by swell; on the continental shelf, it was of sufficient amplitude to halt drilling. In deep water, it was less clearly noticeable and could be absorbed by the heave compensator above the rig floor once the BHA was fully supported by the sediments. However, for the uppermost four to five cores at each site, it is unlikely that the BHA was fully supported. It is possible that a mobile BHA, following ship heave to some extent, would have pumped sediment from the walls of the hole into the top of the space to be sampled in the next APC advance, compressing the "true" sediment section into <100% and may similarly (though less time would have been available) have added to the base of the core by "repenetration" (e.g., Ruddiman et al., 1987; Robinson, 1990) after advance and withdrawal. A contributing factor might have been the very fine grain size of the sampled sediments, which could have increased their mobility. Such factors as the change from in situ to laboratory pressure still have an effect, but at shallow depth this is assumed to be exceeded by the effect of ship heave. This explanation is offered in support of the contention that absence of the "expected" 10% expansion is not a sign of miscorrelation between cores.

Second, notable features of the MS record at these sites (and at Site 1101, also on a sediment drift) are the narrow regions of steady, very low (probably "background") susceptibility, as shown in Figures F1 and F2 and (at higher magnification) in Figure F4 at ~45 mbsf. The remanent magnetic intensity (before demagnetization) is reduced over the same intervals (Shipboard Scientific Party, 1999b, tables T5, T11, T19; 1999c, tables T4, T8, T13). There is a correspondence between these lows and low values of the chromaticity parameters a* and b* in some but not all intervals.

The lows may be the result of diagenesis; however, it is clear (e.g., Fig. F4) that they occupy the same position and have the same form in the record from each hole (holes at a single site are typically only 10-20 m apart horizontally), and they have been used in the correlations described here. It would be unjustifiable to use them either for correlating between sites or in such activities as spectral analysis aimed at extracting paleoclimate variation, unless they could be shown to be primary or uniquely related to primary core features.

Figure F4 also illustrates core distortion. Correlation between cores from the two holes is generally very good. However, the upper part of Core 178-1096A-6H is more expanded than the equivalent part of Core 178-1096B-6H (the upper 87 cm is recorded as disturbed by drilling: Shipboard Scientific Party, 1999c). In contrast, the region around 48 mbsf is relatively compressed in Core 178-1096A-6H, in a region remote from the end of the core in either hole. Differently again, part of the section appears to be omitted at the boundary between Cores 178-1096B-5H and 6H (even though nominal recovery in both cores exceeds 100%).

The existence of differential core distortion noted above indicates that care should be taken in determining the meters composite depth of samples from parts of cores not included in the optimal splice. It may be more precise to use a comparison of the MS records of the equivalent omitted and included cores rather than accept the computed composite depth of the omitted part.

NEXT