The mcd scale was constructed using the program Splicer (version 2.2, available on the World Wide Web from the Borehole Research Group at the Lamont-Doherty Earth Observatory of Columbia University at www.ldeo.columbia.edu/BRG/ODP). Splicer allows data sets from several holes at a given site to be uploaded and correlated simultaneously. Only one data set is shown in Splicer's main viewing window at a time. Multiple data sets can be loaded in a single session, however, and the user can switch among them in the main window. Correlations are first made visually by selecting a tie point from data in one hole and comparing it directly with data from another hole at the same site. Ties determined from the physical or magnetic properties data are intended to correlate matching data patterns and amplitudes. Distinct stratigraphic features, such as many of the volcanic ash layers, can also be correlated between holes (Fig. F3). Matching characteristics of these layers, such as their descriptions recorded in the visual core descriptions or spikes in the magnetic susceptibility data, assist in correlation of continuous individual ash layers over the region. Table T3 lists the correlations made for ash layers at Site 1151.
Importing multiple data sets into Splicer allows the cross-correlation coefficients for all data sets to be computed simultaneously for the given composite depth scale. Depth adjustments that provide the best correlation within a preferred data set or a compromise of correlation coefficients among all the data sets are chosen. The values of the cross-correlation coefficient vary from +1 to -1. A value of +1 indicates perfect correlation (such as would be obtained by comparing identical data sets), values near zero indicate poor or no correlation, and a value of -1 indicates anticorrelation (such as would be obtained by comparing a data set to its inverse). Each time a depth adjustment is made within Splicer, the coefficient is recalculated, allowing the user to determine the preferred correlation. The window over which the coefficient is calculated is adjustable. The default window length of ±2.00 m on either side of the selected tie point was used for most correlations. This window was reduced to ±1.00 m as needed to focus on specific features or to avoid anomalous features such as those biased by coring disturbances.
Correlation begins by selecting the core that has the most pristine record of the upper portion of the upper few meters of the sedimentary record, particularly the mudline (sediment/water interface). This first core is defined as the top of the composite section, and its mcd is the same as its mbsf depth. At Site 1151, the top of the composite section is Core 186-1151D-1H. A tie point that gives the preferred correlation is selected between data from this core and a core in a second hole. All the data from the second hole below the correlation point are vertically shifted to align the tie points between the holes. Once the appropriate tie is determined and the depth adjustment made, the shifted section becomes part of the reference section. The process continues downhole, vertically shifting the data one core at a time, relative to data from the other hole. By tying points of different mbsf depths, Splicer vertically adjusts the individual sections of core and brings the chosen features into the common mcd scale. The tie points are added to the Splicer "affine" table, which records all the depth adjustments that define the composite depth scale, in units of mcd. The depths to the top of each core (obtained in mbsf format from the ODP Janus database) are given in Table T1, and converted from mbsf to mcd scales using the offsets calculated in Splicer and recorded in the affine table.
No expansion or compression of the depth scale within a core or core interval is permitted within Splicer. Due to this restriction, it is not possible to correlate every feature exactly with features in other holes. In some intervals, relative expansion of sediments from one core to another was apparent. In these cases, the correlation was determined using the tie that visually minimized offset over the interval and showed the best overall graphical representation of the correlation coefficients.
In addition to the obvious visual improvement in correlation of data between holes by using the mcd rather than the mbsf scale, the overall improvement is quantified by using the program Analyseries (Paillard et al., 1996). The magnetic susceptibility data sets for Holes 1151C and 1151D were imported into Analyseries to determine an initial linear correlation coefficient between the data sets from 0 to 96 mbsf. Points of the same mbsf depth at the top and bottom in each data set were tied, and a linear correlation coefficient between 1 and -1 was obtained. The same data in the mcd scale, determined using Splicer, were then imported to Analyseries at various stages in the correlation process to provide a quantitative assessment of incremental improvement in correlation of the composite depth scale relative to the mbsf depth scale. In cases where the preferred correlation was ambiguous in Splicer, depth shifts that improved the overall linear correlation in Analyseries were retained in the composite depth scale, and those that degraded the correlation were discarded. The overall linear correlation coefficient for the magnetic susceptibility data for Holes 1151C and 1151D in the mbsf scale prior to any depth shifting was 0.650. After using Splicer to create a composite depth scale, the overall linear correlation coefficient improved to 0.845.
Additional constraints will become available as new data sets are acquired, such as U-channel paleomagnetic measurements and tephrachronology studies. The composite depth scale constructed already shows good correlation of ash layers (Table T3) between holes, and tephrachronology could confirm these correlations. Similarly, the composite depth scale results in excellent agreement in magnetostratigraphic constraints between Holes 1151C and 1151D. Specifically, the Brunhes/Matuyama boundary occurs at 82.26 mcd for both holes, even though this boundary was not used as a constraint in constructing the composite depth scale. In general, coeval features agree to better than ~10 cm as noted with the ash layers (Table T3), even though differences of ~20 cm are observed and can be expected given the resolution of the data used to construct the composite depth scale.
The data sets for Holes 1151A, 1151C, and 1151D were spliced to create a continuous downhole sedimentary section for Site 1151 (Figs. F4, F5, F6). The spliced record is a continuous stratigraphic section constructed from representative portions (or intervals) of data from the available holes at a site. The splice was created using the magnetic susceptibility data, but the tie points (Table T4) can be applied to the other data sets. Because the splice is optimized for susceptibility, gaps or artificial steps can occur in other data sets as shown in Figure F5. The splice is nonunique in that some intervals may be equally represented by more than one of the holes drilled at the site. When selecting tie points for the spliced record, as much data from a single core as possible was kept intact, minimizing the number of alternations between holes needed to provide the total spliced record. The exception to this approach was in intervals that contained large data spikes. Placing the tie points in these areas was avoided to reduce creating artificial steps within the spliced susceptibility data set.
Site 1150 provided only minimal overlap of cored intervals with recovery only from Holes 1150A and 1150B. Hole 1150A was cored to 713.00 mbsf (78.38% recovery) and Hole 1150B was cored continuously from 703.30 to 1172.00 mbsf (56.30% recovery), providing potential overlap only between 703.30 and 713.00 mbsf (Fig. F7). Within this narrow interval, only two depth shifts were required to align Hole 1150B to Hole 1150A. In both cases, Hole 1150B was shifted to align with Hole 1150A. As a result, the mbsf and mcd scales are identical for Hole 1150A. In Hole 1150B, Core 186-1150B-1R should be shifted 0.18 m downward, and all cores from Core 186-1150B-2R and below should be shifted 1.02 m downward from the originally recorded mbsf depths to place them within the mcd scale. Within the overlap interval, correlation of Holes 1150A and 1150B give a correlation coefficient of -0.125 in the mbsf scale. The slight adjustments made in the overlap region improve the correlation coefficient to 0.210 in the mcd scale. The overall correlation for Site 1150 is still poor, however, both in visual and mathematical terms. A spliced data set was not constructed for Site 1150, as the minimal degree of overlap would leave most of the data set essentially unchanged from the original sequence.