Stratigraphers have demonstrated that a continuous sedimentary section is rarely recovered from a single ODP borehole because core-recovery gaps occur between successive APC and XCB cores despite 100% or more nominal recovery (Ruddiman et al., 1987; Hagelberg et al., 1995). Construction of a complete section, referred to as a splice, requires combining stratigraphic intervals from two or more holes cored at the same site. To maximize the probability of bridging core-recovery gaps in successive holes, the depths below the seafloor from which cores are recovered are offset between the holes. This practice ensures that most intercore intervals missing within a given hole are recovered in at least one of the adjacent holes. On Leg 202, we found that two complete holes and a third partial hole were usually needed to recover a complete section in the APC portion of a site.
Our composite section and splice construction methodology follows one which has been successfully employed during a number of legs (e.g., Hagelberg et al., 1992; Curry, Shackleton, Richter, et al., 1995; Jansen, Raymo, Blum, et al., 1996; Lyle, Koizumi, Richter, et al., 1997; Gersonde, Hodell, Blum, et al., 1999, Wang, Prell, Blum, et al., 2000). The construction and verification of a complete composite stratigraphic section requires the construction of a variety of depth scales (Table T1) that are described below.
Before a splice can be constructed, the cores from the various holes must be stratigraphically correlated with each other. Such correlation enables development of a composite depth scale referred to as meters composite depth (mcd). This mcd scale differs from the traditional (hole specific) depth scale, called the meters below seafloor (mbsf) scale. The latter is based on the length that the drill string is advanced on a core-by-core basis but may be inaccurate because of ship heave (which is not compensated for in APC coring), tidal variations in sea level, and other sources of error. In contrast, the mcd scale is built by assuming that the uppermost sediment (commonly referred to as the mudline) in the first core from a given hole is the sediment/water interface. This core becomes the "anchor" in the composite depth scale and is typically the only one in which depths are the same on both the mbsf and mcd scales. From this anchor, core logging data are correlated among holes downsection. For each core, a depth offset (a constant) that best aligns the observed lithologic variations to the equivalent cores in adjacent holes is added to the mbsf depth in sequence down the holes.
Depth offsets are often chosen to optimize correlation of specific features that define splice levels in cores from adjacent holes. In a few cases (e.g., Site 1233), we constructed an alternate (secondary) splice to accommodate high-resolution sampling. Such secondary splices require the definition of secondary mcd scales to optimize splice levels. When calculating mcd of samples, it is important to use the composite depth scale that is compatible with the splice.
The length of the mcd scale at a given site is typically ~10% to 20% greater than the length of the cored section in any one hole as indicated by the mbsf scale. This expansion is commonly attributed to sediment expansion resulting from elastic rebound, stretching during the coring process, gas expansion during the core recovery process, and other factors (e.g., Moran, 1997).
For Leg 202, the mcd scale and the splice are based on the stratigraphic correlation of data from the whole-core Oregon State University (OSUS) Fast Track (hereafter referred to as Fast Track), the MST, and the AMST. Core-logging data were collected at 2.5-, 5-, or 10-cm intervals. We used magnetic susceptibility (the data are referred to as OSUS-MS if collected using the Fast Track or MST-MS if collected using the MST,) gamma ray attenuation (GRA) bulk density, natural gamma radiation (NGR), and reflectance (L*, a*, and b*). All of these measurements are described in "Physical Properties".
The raw stratigraphic data were imported into the shipboard Splicer software program (version 2.1) and culled as necessary to avoid incorporating anomalous data influenced by edge effects at section boundaries. Splicer was used to assess the stratigraphic continuity of the recovered sedimentary sequences at each drill site and to construct the mcd scale and splice. Splicer enables the construction of a composite depth scale for each hole at a given site by depth-shifting individual cores to maximize the correlation of core logging data.
Because depth intervals within cores are not squeezed or stretched by Splicer, all correlative features cannot be aligned. Stretching or squeezing between cores from different holes may reflect small-scale differences in sedimentation and/or distortion caused by the coring and archiving processes. The tops of APC cores are generally stretched and the bottoms compressed, although this is lithology dependent. In addition, sediment (especially unconsolidated mud, ash, sand, and gravel) occasionally falls from higher levels in the borehole onto the tops of cores as they are recovered, and as a result the top 20-100 cm of many cores is not reliable.
Correlations among cores from adjacent holes are evaluated visually and statistically (by cross-correlation within a 2-m depth interval). Depth-shifted data are denoted by mcd. A table is presented in each site chapter that summarizes the depth offsets for each core. These tables are necessary for converting mbsf to mcd scales. The mcd for any point within a core equals the mbsf plus the cumulative offset. Correlation at finer resolution is not possible with Splicer since depth adjustments are applied linearly to individual cores; no adjustments, such as squeezing and stretching, are made within cores. Such fine-scale adjustment is possible postcruise (e.g., Hagelberg et al., 1995).
Ideally, the base of the mcd scale is the bottom of the deepest core recovered from the deepest hole (e.g., Site 1233). In practice, however, the base often occurs where core recovery gaps align across all holes or the data quality does not allow reliable correlations between holes. Cores below this interval cannot be directly tied into the overlying and continuous mcd. However, below the base of the continuous mcd, cores from two or more holes can sometimes be correlated with each other (e.g., Site 1234) to create a floating splice. At other sites, we used the observation that the differential depth offset within the correlated interval was relatively constant (i.e., the growth in the cumulative offset, the difference between mbsf and mcd for a particular core, was linear) to assign cores below the splice a depth offset (e.g. Site 1238) by extrapolation using a growth factor (GF). GF is defined as the ratio of apparent length of sediment (using the mcd scale) to the drilled interval, as mcd/mbsf, within a defined interval. This extrapolation strategy was adopted when three conditions were met. First, the correlated interval represented at least one-half of the drilled interval. Second, the spliced interval had a linear growth in cumulative offset. Third, the entire sediment section had a shared lithology, because we found that lithology tended to determine the size of coring offset and associated change in GF. If any of these conditions was not met, then an mcd was assigned to the section below the splice by adding the greatest cumulative offset to the first core below the splice and, working downhole, appending cores using the drilled interval if recovery was 100% or the recovered interval if recovery was >100% (e.g., Site 1232).
We also provide corrected meters composite depth (cmcd) in our depth conversion tables. This scale is intended to correct the mcd scale for empirically observed core expansion. A cmcd datum is produced by dividing mcd by GF over a sufficiently long interval so that random variations in drill pipe advance due to ship heave, tides, and other factors are negligible. This operation produces a complete stratigraphic sequence that is the same length as the total depth cored. The cmcd scale is a close approximation of the actual drilling depth scale. Because the cmcd scales are not archived in the Janus database, in most cases we found it most practical to plot data on the mcd scale and to correct linear sedimentation rates (LSRs) and mass accumulation rates (MARs) by dividing by the average GF within each interval.
The splice is a composite core section representing the complete stratigraphy at a site. It is composed of core sections from adjacent holes so that coring gaps in one hole are filled with core intervals from an adjacent hole. The splice does not contain coring gaps, and an effort has been made to minimize inclusion of disturbed sections. The shipboard splice is ideally suited to guide core sampling for detailed paleoceanographic studies. A table and a figure are presented in each site chapter that summarize the intervals from each hole used to construct the splice. Additional splices may be constructed postcruise as needed. (e.g., Site 1233).
The choice of tie points (and hence of a splice) is a somewhat subjective exercise. Our method in the construction of a splice followed three rules. First, where possible we avoided using the top and bottom 1 m of cores, where disturbance resulting from drilling artifacts (even if not apparent in core logging data) was most likely. Second, we attempted to incorporate those realizations of the stratigraphic section that were most representative of the holes recovered. Third, we tried to minimize tie points in order to simplify sampling.
At Sites 1238, 1239, and 1241, logging operations produced data sets that were of sufficient quality to allow for core-log integration. Log data are provided with a depth scale referred to as meters logging depth (mld). This depth scale is constructed using the time at which the logging tool is raised from the bottom of the hole, the rate at which it is raised, and the time at which it crosses the sediment/water interface. Core-log integration produces yet another depth scale, estimated log depth (eld). This depth scale is constructed by correlating physical properties measurements of cores made using the MST with those made in the drill hole using logging tools. This depth scale has the advantage of correcting for stretching and squeezing within cores. It has the disadvantage of typically beginning at ~100 mbsf, where the drill pipe is positioned during logging operations. Where available and where logging data are of sufficient quality, eld is the best estimate of in situ depth and so is ideal for calculating MARs.
To determine eld, logging and whole-core MST data were imported into the Sagan software package (version 1.2) and culled and smoothed as necessary to ensure that bad data were not included in the integration (see explanation in the Splicer discussion in "Composite Depth Construction") and that the log and core data were compatible. Because core logging data, in general, have a higher resolution than downhole log data, it is necessary to smooth the core logs before comparing them with downhole logs. Sagan allows the correlation of individual cores within different holes with the data series recovered from logging. We used NGR and GRA bulk density measurements made in the borehole and in the laboratory for core-log comparison. Sagan allows the user to shift depths and to linearly stretch or squeeze the entire recovered sequence and then to shift depth, and to stretch or squeeze individual cores in user-defined increments until acceptable core-log integration is established. We attempted, whenever possible, to minimize the number of tie points used to tie a single core to the log. For sites where core-log integration was attempted (Sites 1238 and 1239), we provide supplementary data files, which are Sagan eld tables (see the "Supplementary Material" contents list). These tables, when loaded into Sagan, allow the user to assign eld to any sample(s) with a previously assigned mcd or mbsf.
The use of the Fast Track was a major innovation during Leg 202. Because throughput on the track was extremely high (15 min per core in its fastest mode or 45 min per core in high-resolution mode), it was able to keep pace with most drilling operations and thus allowed for real-time evaluation of the position of coring gaps in the composite section. This helped us to make rapid drilling adjustments aimed at ensuring the recovery of a complete stratigraphic section. It also allowed us to run the MST at a slower rate, to optimize data quality (longer measurement time yielding greater analytic precision and, in many cases, higher depth resolution) because we did not need its data to assess recovery of a complete section at most sites. We recommend a second whole-core measurement track optimized for speed of analysis on future high-recovery paleoceanographic legs. Ideally, this track would have a magnetic susceptibility and a GRA densitometer so that the data type most appropriate to the sedimentary setting could be utilized.
The depth of a core interval recorded for a tie point in a splice table is not always the same as the depth for the same core interval returned by most database queries. This is because the tie-point depth is based on the liner length, measured when the cores are cut into sections on the catwalk. The cores are analyzed on the MST almost immediately after this liner length measurement. At some later time, typically 10 to 36 hr after being analyzed by the MST, core sections are split and analyzed further (see "Core Handling and Analysis"). At this time, the section lengths are measured again and are archived as "curated lengths." General database queries return depths based on the curated liner lengths. Because the sections are usually expanding during the period between the two measurements, the curated length is almost always longer than the initial liner length. Thus, the depths associated with the MST data used to construct the splice table are not identical to the final depths assigned to a given interval by the database. This leads to small differences, usually between 0 and 5 cm, between the mbsf and mcd recorded in a splice table and the depths reported in other places for the same core interval. We have chosen not to change these depths to be compatible with Janus because this would not improve their accuracy. For consistency, we recommend that all postcruise depth models use or build on mcd values provided in the Janus database.