Multiple holes were cored at both sites with the Advance Piston Corer (APC) with the goal of recovering as complete a section as possible given time limitations. At Site 1098 three holes were cored through the roughly 46-m thick sedimentary fill overlying acoustic basement. Hole 1098A was drilled to 45.9 meters below sea floor (mbsf) with 45.78 m of sediment recovered (99.74% recovery); Hole 1098B was drilled to 43.0 mbsf with 44.66 m of sediment recovered (103.86% recovery); and Hole 1098C was drilled to 46.7 mbsf with 46.3 m of sediment recovered (99.14% recovery).

At Site 1099 two holes were cored through the upper 107.5 m of the approximately 260-m thick sediment fill overlying basement. Owing to time constraints, the interval was effectively cored only once, with the two holes only overlapping 2.3 m in the mbsf depth scale. The upper 62.3 m of section was cored in Hole 1099A and the interval from 60.0 to 107.5 mbsf was cored in Hole 1099B.

The primary goal of this study is to establish two common depth scales, one for the three holes at Site 1098 and the other for the two holes at Site 1099. As with all ODP drill cores, the depth to the top of each core is estimated from a drill-pipe measurement. These drill-pipe measurements along with curation procedures establish a unique meters below seafloor (mbsf) depth scale for the cores from each hole (e.g., Chapter 3 of Barker, Camerlenghi, Acton, et al. 1999). Uncertainties in the depth estimates, core expansion, incomplete recovery, and other factors result in deviations of the mbsf depths from the true depths. Thus, a continuous horizontal feature collected in several holes at a site, which in the absence of local bathymetric variations would have the same true depth in each hole, will likely have different mbsf depths. The offset of such features in the mbsf depth scale may only be several centimeters or could be a few meters, though rarely more than 10 m.

A common or composite depth scale overcomes such inadequacies, allowing cores from one hole to be compared directly with those from other holes (e.g., Hagelberg et al., 1992 and 1995; and Keigwin, Rio, Acton, et al., 1998). Similarly, data from the cores from one hole can be compared or combined ("spliced") with data from cores from other holes at the same site. Cores from a single hole will nearly always have some intervals with poor or no recovery or with recovery of sediments that are disturbed during drilling. By splicing intervals from cores from different holes at a site that has been multiply cored, a complete or nearly complete stratigraphic section can be constructed.
 

To determine the quality of the correlation, we used visual comparison as well as the correlation coefficient calculated and shown graphically within SPLICER. The correlation coefficient is determined by cross correlating the data from two cores. Its value ranges from +1 to -1, and the window over which it is calculated is adjustable. In nearly every case we used the default window length of ± 2 m on either side of a selected tie point. Occasionally we reduced this to ± 1 m to focus on specific features or to avoid features related to coring disturbance.

No expansion or compression of one core or core interval relative to others is permitted within SPLICER. Because of this restriction, it is not always possible to align every anomaly exactly with similar anomalies in other cores. Second order correlation of that nature can be best accomplished by using programs like ANALYSERIES (Paillard et al., 1996) to tweak the first order correlation accomplished with SPLICER.

In creating the mcd scale, we attempted to ignore data from intervals that were disturbed or distorted by the drilling process (Table 2). This typically includes the top 5 to 70 cm of nearly every core, which are commonly more water saturated or soupy relative to the sediment below. Only in the first core at each hole, particularly at the mudline, would we expect to see such poorly consolidated sediments. We also avoided other distortions, such as gas voids and deformed sediments. Hole 1098C contained additional data gaps caused by a 5-cm long interstitial water (IW) sample being removed from the end of each section prior to the sections being measured on the Multi-Sensor Track (MST).

The primary data sets used are those collected during Leg 178 with the MST, the Minolta Color Scanner, and the longcore cryogenic magnetometer (Barker, Camerlenghi, Acton, et al., 1999). Important additional constraints come from core photos and core descriptions.
 

Below we outline the features (lithologic units and anomalies from susceptibility, GRAPE density, color reflectance, and paleomagnetic inclination and intensity) that we think should be aligned within any composite depth scale. Our preferred composite depth scale achieves this goal.

We also avoid correlating certain features that are artifacts of drilling, IW sampling, or bad data points, such as those that occur when MST measurements are made at the ends of core sections or over voids. We have created a table that lists some of the more prominent coring disturbed zones and the voids (Table 2). In addition, any susceptibility or paleomagnetic data within 5 cm of the end of each section are ignored. This is done because the sensors that measure these parameters average over several centimeters of core, and so measurements near the end of the sections are biased.
 

Subunit Ia/Ib Contact
Subunit Ib contains more clastic material (ice rafted debris and pebbles) than the overlying subunit Ia. This subunit was defined as occurring from 1098A-6H-3, 60 cm to the base of Hole A and from 1098C-5H-4, 50 cm to the base of Hole C (Barker, Camerlenghi, Acton, et al., 1999). The boundary between subunits Ia and Ib is contained within a zone less than about 20 cm wide in which laminated diatomaceous mud at the base of subunit Ia grades downward into silty clay with ice rafted debris and pebbles. The subunit Ia/Ib contact, which can be correlated between cores with an accuracy of ± 15 cm, provides an adequate tie point for testing the composite depth scale.

Upper Contact of Lower Laminated Interval (LL)
A laminated diatomaceous mud, approximately 340 cm thick, occurs at the base of subunit Ia. This LL interval spans 345 cm in Hole 1098 (from 1098A-6H-1, 15 cm to 6H-3-60 cm) and 330 cm in Hole C (from 1098C-5H-2, 25 cm to 5H-4, 50 cm). In Hole 1098B, only 296 cm of the LL interval were recovered (from 1098B-5H-5, 102 cm to the bottom of the hole) during coring, which did not penetrate the lower portion of this interval. The upper boundary of the LL interval, which can be correlated between cores with an accuracy of ± 15 cm, provides an adequate tie point for testing the composite depth scale.

Homogeneous to Laminated (HL) Contact
This distinctive contact between a laminated diatom ooze and underlying very homogeneous massive diatom ooze (a turbidite unit) occurs within the upper portion of subunit Ia in all three holes (1098A-4H-2, 80 cm [23.20 mbsf]; 1098B-3H-6, 40 cm [23.40 mbsf]; and 1098C-3H-5, 8 cm [24.28 mbsf]). The contact is sharp and can be easily identified in the core photographs. The contact (at 24.92 mcd), which can be correlated between cores with an accuracy of ± 2 cm, provides a key constraint for the composite depth scale.

1. The interval with the largest decrease uphole in susceptibility from Hole 1098C, which occurs from 1098C-5H-4, 124 cm to 1098-5H-4, 138 cm, correlates with the similar interval within Hole 1098A, which occurs from 1098A-6H-3, 130 cm to 142 cm (Fig. 2). In this interval, the susceptibility decreases from 2040 to 50 (all susceptibility values are given in raw meter values, which can be converted to SI volume susceptibility units by multiplying by ~0.7 x 10^-5) in Hole 1098A whereas it decreases from 2860 to 920 in Hole 1098C. A second decrease occurs in Hole 1098C about 20 cm further upcore. A slightly better correlation coefficient is obtained by correlating the Hole 1098A decrease with this higher decrease in Hole 1098C (correlation coefficient of 0.86 versus 0.75 for our preferred correlation), but this misaligns the lithologic Subunit Ia/Ib contact by about 20 cm.
2. Susceptibility anomalies in the lower laminated intervals (43.19 to 43.50 mcd) are difficult to correlate with one another (Fig. 2). The susceptibility within this interval is about 10 to 50 times less than in the non-laminated intervals. The largest anomalies are associated with dropstones, e.g., the two peaks at 1098A-6H-1, 48 and 92 cm occur precisely where two dropstones are present (see Core Photos in Barker, Camerlenghi, Acton, et al., 1999).
3. Above 40 mcd, correlations between holes consistently give correlation coefficients greater than 0.6, and often greater than 0.8. Distinctive sets of anomalies occur throughout, which firmly establish the accuracy of the composite depth scale (Figs. 2-5). Difficulties within the 0-40 mcd interval arise mainly because some compression or expansion would be required to align every correlatable feature from one hole to another. An example where relative stretching within a core would be required occurs within the 34-40 mcd interval. Distinctive anomalies from 34-36 mcd from Core 1098A-5H correlate precisely with those from Hole 1098C-4H, but further downcore at 36-40 mcd, other distinctive anomalies are about 4-12 cm deeper in Core 1098C-4H than in Core 1098A-5H.
4. Particularly good correlations occur at the HL contact (24.92 mcd). As with the GRAPE density data and the a* data, the susceptibility data show little variation directly below the contact, and large variations above (Figs. 5).
5. The interval from about 16-22 mbsf is one of the more difficult to correlate (Figs. 4-5). Above and below this the correlation is clear, and so the composite depth scale is well constrained even within this interval.

1. Correlation coefficients are generally less than 0.4 for any correlation below about 30 mcd, with few exceptions (Fig. 6). One exception occurs at the very base of Holes 1098A and C, where a gradual decrease in density occurs from the base of both holes to about 3 m uphole. This long wavelength feature can probably only be correlated to within about ± 20 cm (Fig. 9).
2. A distinctive sequence of anomalies, including a large positive density anomaly, at 28.5-29.6 mcd correlates well between Holes 1098A and B (from 1098A-4H-4, 138 cm to 4H-5, 98 cm and from 1098B-4H-1, 132 cm to 4H-2, 92 cm).
3. The density within the very homogeneous massive diatom ooze interval (24.9 to 28.0 mcd) in each hole is nearly constant (1.3-1.5 g/cm3), whereas in the overlying laminated diatom-ooze interval the values show much larger variation (1.3-2.0 g/cm3). The HL contact is thus associated with a change in the character of the density, which can be correlated well between holes (Fig. 8).
4. Correlation coefficients are less than 0.4 for any correlation from about 16 to 22 mcd.
5. The density records from about 2 to 16 mcd correlate well (correlation coefficients of 0.4 to 0.6) between all three holes.
6. The anomalies near the mudlines of Holes 1098A and 1098B correlate well, but their correlation with 1098C is poor probably owing to some coring disturbance (Fig. 7).

1. No unambiguous match can be made with either long or short wavelength anomalies from the L*, b*, or 500-nm data, except at the mudline. At the mudline, all three color reflectance parameters correlate very well between Holes 1098A and B only. In general, the best correlation coefficients were less than 0.4. By smoothing the data, the correlations can be improved a little, but overall are quite poor.
2. Correlation of the a* data are mediocre, but good enough to illustrate that the composite depth scale aligns distinctive anomalies from one hole to another. In general, the a* data gives correlation coefficients of 0.3 to 0.6. Particularly good correlations are present at the HL contact and a few meters on either side. As with the GRAPE density data, the a* data show little variation below the contact, and large variations above (Fig. 12).

1. Correlation coefficients are typically greater than 0.5 for the upper 39 m, but poor below this (Fig. 10).
2. An extremely good correlation (correlation coefficients >0.9) of a distinctive set of intensity anomalies occurs in the interval from 33.3 to 39.5 mcd (Fig. 11)
3. Above 24 mcd, the correlation of intensity data between Holes 1098B and C is much better than that between either of these holes and Hole 1098A (Fig. 10).

1. The large uphole decreases in susceptibility which occurs in a narrow interval near the base of Holes 1098A and C are aligned. Similarly, this aligns the largest GRAPE density anomaly, which occurs within the same interval.
2. The boundary between Subunits Ia and Ib recovered in Holes 1098A and C agree to within 1 cm.
3. The upper boundary of LL interval in Hole 1098A is 7 cm higher than that from Hole 1098B and 15 cm higher than that from Hole 1098C. Because the thickness of LL varies between holes, improving the alignment of this boundary would misalign other features that are more distinctive.
4. An extremely good correlation (correlation coefficients >0.9) of magnetic intensity data in the interval from 33.3 to 39.5 mcd is obtained. Offsets of more than 15 cm would degrade this good correlation significantly (correlation coefficients would decrease by about 0.2).
5. The HL contact correlates within ± 2 cm in all three holes. Similarly, the distinctive anomalies in GRAPE density, a*, and susceptibility at this contact all provide confirmation that the composite depth scale is accurate and can be precisely constrained in this interval.
6. The composite depth scale requires overlap of 44 cm between Cores 1098B-4H and 5H. This overlap can be explained perhaps entirely by the 40-45 cm of disturbed sediment at the top of Core 1098B-5H. Often the top part of a core contains sediment that has fallen into the hole from above. This happens as the roller cone bit, which is part of the bottom-hole assembly (BHA), advances from the top of the core previously recovered to the top of the core that is next to be recovered. If the ship heaves upward as the piston strokes into the sediment, then the debris in the hole can be recovered. Additional expansion of the upper part of each core can occur because the top of the core is exposed to circulating water, particularly as the water jets from the BHA are cleaning out the hole.
7. Core 1098C-1H overlaps 14 cm with Core 1098C-2H. Again the top 70 cm of Core 1098C-2H is disturbed, so this small overlap is not unexpected.
8. Core 1098C-1H is shifted up 10 cm from its depth given in mbsf. This shift can be explained by the 30 cm of coring disturbance that occurs at the top of this core.
9. The color reflectance (L*, a*, b*, and 500-nm signal), susceptibility, and density data sets from the mudlines from Holes 1098A and B agree very well. Owing to coring disturbance of the mudline in Hole 1098C, which is visible in the core photo, the data sets from Hole 1098C correlate poorly with those from Holes 1098A or B.
10. Overall there are numerous characteristic anomalies in the magnetic susceptibility data that correlate well from one hole to another.

The splice, which was constructed from the susceptibility record and is not optimized for other data sets, should however give representative composite records for the other data sets. In rare cases, the other data sets may have missing values within the spliced intervals owing to the peculiarities of the instruments used during the cruise or the selection criteria used to cull data (e.g., the interval 36.9-38.8 mcd for the spliced intensity record in Fig. 11). For example, an interval that may have been slightly disturbed by coring may have been considered unacceptable for paleomagnetic remanence measurements, but acceptable for magnetic susceptibility, color reflectance, and other measurements. Another artifact which may occur in the composite records for the other data are discontinuities or jumps located at the splice tie points.

Spliced data sets for a*, GRAPE density, magnetic intensity, and susceptibility are included in Tables 5-8. Other data sets converted to the mcd scale can be retrieved over the internet from the ODP Janus database.

We found that none of the data examined from the upper part of Hole 1099B, which includes magnetic susceptibility, GRAPE density, and color reflectance data (L*, a*, and b*), appears to correlate well with the lower few meters of Hole 1099A. One of the better correlations, though not convincing by any means, is obtained with the b* data with no depth offset for Hole 1099B relative to 1099A. Given our inability to find any convincing anomalies to suggest that a depth shift was required, we consider that the best composite depth scale is that obtained by using the mbsf scale already available.

In the overlap region, the best splice is then obtained by using data from Hole 1099A because there is little or no coring disturbance as opposed to that visible in upper 5 m of core from Hole 1099B.
 

Coring in the two holes at Site 1099 provide only a few meters of overlap. None of the data sets within this limited overlap region provide convincing correlations. Thus, the preferred composite depth scale is the existing mbsf scale.

Composite scales, such as established here, represent first order correlation between holes at a site upon which higher order depth scales can be built. Smaller scale features, such as individual lamina or sets of laminae, may not correlate exactly given the restriction that neither compression nor expansion of the length of individual cores was allowed in developing the mcd scales. For Site 1098, sets of laminae, as well as centimeter-scale variations in physical and magnetic properties, will likely be correlatable between holes and should naturally lead to fine-scale adjustments in the composite scale.

Brachfeld, S., Acton, G.D., Guyodo, Y., and Banerjee, S.K., High-resolution paleomagnetic records from Holocene sediments from the Palmer Deep, Western Antarctic Peninsula, Earth and Planetary Science Letters, in press.

Hagelberg, T., Shackleton, N., Pisias, N., and Shipboard Scientific Party, 1992. Development of composite depth sections for Sites 844 through 854. In Mayer, L., Pisias, N., Janecek, T., et al., Proc. ODP, Init. Repts., 138: College Station, TX (Ocean Drilling Program), 79-85.

Hagelberg, T.K., Pisias, N.G., Shackleton, N.J., Mix, A.C., and Harris, S., 1995. Refinement of a high-resolution, continuous sedimentary section for studying equatorial Pacific Ocean paleoceanography, Leg 138. In Pisias, N.G., Mayer, L.A., Janecek, T.R., Palmer-Julson, A., and van Andel, T.H. (Eds.), Proc. ODP, Sci Results, 138: College Station, TX (Ocean Drilling Program), 31-46.

Keigwin, L. D., D. Rio, G. D. Acton, et al., 1998. Proc. ODP, Init. Repts, 172: College Station, TX (Ocean Drilling Program).

Moran, K., 1997. Elastic property corrections applied to Leg 154 sediment, Ceara Rise. In Shackleton, N. J., Curry, W. B., Richter, C., and Bralower, T. J. (Eds.), Proc. ODP, Sci. Results, 154: College Station, TX (Ocean Drilling Program), 151-155.

Paillard, D., Labeyrie, L., Yiou, P., 1996. Macintosh program performs time-series analysis, Eos, 77:379.