Our strategy for developing rmcd hole-to-hole correlations for Sites 1218 and 1219 was to keep the rmcd scale similar to the original shipboard splice mcd scale, where possible, in order to facilitate comparison with shipboard results and to ease the creation of alternative sediment sampling plans. Thus, where multiple copies of the same stratigraphic interval were recovered, our revised composite depths (rmcd) are identical with the shipboard composite depths of the core that is part of the original shipboard splice, whereas the other cores were finely adjusted to give the best match for all the holes on a centimeter to decimeter scale. We achieved this aim for Site 1218. For Site 1219, we had to correct some shipboard composite depth scale mismatches but kept most of the shipboard correlations intact. Following shipboard practice for intervals recovered in only one hole, we offset mcd and rmcd depths by a nominal ~1.1 m, according to the average observed length increase of the mcd/rmcd depth scale. The complete data sets, plotted on the new rmcd scale, are shown in Figures F3 and F4. In addition to the MST data, aligned on the rmcd scale, we also show the stacked records as superimposed dashed lines; the generation of these is discussed in a separate section. We also show the points where the original splice track switches between different holes. For Site 1218, all but one original splice point (toward the very bottom) were kept. The depth adjustments within each core were made by considering all data sets. However, depending on the lithology, different types of data were more useful for this purpose: the carbonate-poorer lithologic Units I, III, and IV (Shipboard Scientific Party, 2002a) were more easily matched using the magnetic susceptibility data, whereas for the carbonate-rich Unit II, all data proved equally useful. The depth tables that relate mbsf, mcd, and our new rmcd scales are given as Table T1 for Site 1218 and Table T2 for Site 1219. Depths were linearly interpolated between individual tie points. We note that the improved stratigraphic hole-to-hole correlation was relatively straightforward for intervals of intermediate carbonate content. For carbonate-poor intervals, characterized by low lightness and bulk density values and high magnetic susceptibility values, matches were based mostly on the magnetic susceptibility measurements.
One of the main reasons for generating a detailed new rmcd depth scale is that after alignment to this scale it is possible to stack data from different holes, thus increasing the signal-to-noise ratio. For example, color reflectance data are acquired by spot measurements with a footprint of ~4 mm2. Thus, processes such as bioturbation or uneven surfaces of split cores will produce noise in the data. The stacking process allows us to reduce this noise for all data sets and provides a more representative record. We generated stacked data sets for magnetic susceptibility, color reflectance (L*, a*, and b* color space), and GRA bulk density as follows. The pruned data, aligned on the rmcd scale, were first visually inspected for occurrences where data from one hole were significantly different from the other holes. If those data points were clearly outliers, they were removed. We then resampled data from all holes with a common 2-cm sampling step where available, using a Gaussian smoothing window with a half-width of 5 cm. The resampling is necessary, as the process of "stretching and squeezing" within cores implies that data measurements are no longer spaced equally in depth, and data from different holes are also not sampled at identical depths. The width of the smoothing window was chosen so as to preserve variations of the lithologic measurements that were common between data from separate holes. As the average sedimentation rates were between ~0.5 and 2 cm/k.y. throughout the Site 1218 and 1219 records, this smoothing process preserves any information with timescales longer than a few thousand years and is of the same order as that applied by the magnetic susceptibility loop during data acquisition. The smoothed data were then averaged if data were available from more than one hole or taken from a single hole. The stacked data sets are available in electronic form, as they are too large to be included as complete tables here. However, the format of the data tables is shown in Table T3 for Site 1218 and Table T4 for Site 1219.
It is important to stress that data correlated from one site alone might not be sufficient to identify hiatuses in the record, as a particular stratigraphic interval might only be recovered in a single hole. One of the strengths of stratigraphic correlations that were performed with data from ODP Legs 138 and 154 (Shackleton et al., 1995; Shackleton and Crowhurst, 1997; Shackleton et al., 1999) was that these correlations were not only achieved with data from one particular location, but with data from different sites. This approach allowed the detection of several hiatuses that otherwise would have remained unnoticed. We follow this pioneering work by performing additional site-to-site correlations for ODP Sites 1218 and 1219. Figure F5 illustrates how well the data between Sites 1218 and 1219 can be matched for an interval in the upper Oligocene/lower Miocene. Table T5 provides the tie points that match Site 1219 depths (in rmcd) to those of Site 1218. We applied a linear interpolation of depths between these tie points. The matching of data from 1219 to 1218 shows that individual features of the MST data can be matched on a centimeter to decimeter scale, an observation that is truly astonishing, given the large distance (~740 km) between Sites 1218 and 1219. We note that the comparison of data from Sites 1218 and 1219 suggests that both sedimentary sequences are complete and not interrupted by any hiatuses, at least on the scale of the lateral separation of the two sites.
Because data from the two sites match so well, we can now proceed to use this match to extrapolate magneto- and biostratigraphic datum events between sites. This approach yields two important benefits. First, we can evaluate, and reduce, the uncertainty in depth for datum events that cooccur at both sites. We can also evaluate small potential latitudinal differences for biostratigraphic events, as Sites 1218 and 1219 were at slightly different paleolatitudes at any given time (Lyle, Wilson, Janecek, et al., 2002). Second, for datum events that were only determined for one of the two sites, we can use the newly developed match of rmcd depth scales from Sites 1218 and 1219 to predict the position of these datum events for the site where it was not determined, thus improving the database available for generating detailed age models.
The depth determinations for magnetostratigraphic events presented here are based on shipboard magnetic data that were available at the time. We note that additional postcruise work has been carried out for some intervals of Sites 1218 and 1219 (Lanci et al., 2004), which will further reduce the depth uncertainty for these events. Table T6 lists magnetostratigraphic events interpreted for Sites 1218 and 1219 on our new rmcd scale, together with depth uncertainty estimates. We note that the correlation between Sites 1218 and 1219 is very good for the cabonate-rich interval (~55–240 rmcd for Site 1218). Above 55 rmcd, the identification and correlation of magnetic reversals is less certain. Table T6 also gives depths for Site 1219 as Site 1218 equivalents, using our site-to-site correlation. There are two examples where our site-to-site correlation gives information that data from the two sites on their own were unable to provide. Polarity Chron C11n, for example, was not identified with confidence in the shipboard magnetostratigraphic data for Site 1219; our match allows the determination of the equivalent stratigraphic interval for Site 1219 from data from Site 1218. Second, no magnetostratigraphic data are available from Site 1218 below Chron C12n, as the lithologic nature of sediments required a switch from the APC coring system to XCB. However, our site-to-site correlation results in depth estimates for Chrons C13n through C20n for Site 1218 by extrapolation from Site 1219, thus improving our ability to constrain age models. There are several additional examples where core breaks or core disturbance increased the depth uncertainty of magnetic reversals at one site that can now be further reduced by using our newly developed site-to-site correlation.
As was the case for the magnetostratigraphic data, we can also improve our available database of biostratigraphic datums by hole-to-hole and site-to-site correlation. The detailed determination of radiolarian datum events is given elsewhere (Nigrini et al., this volume). Calcareous nannofossil datum events, some of them improved from the original shipboard information, are listed in Table T7 for Site 1218 and Table T8 for Site 1219. For Site 1219, we also give equivalent depths for Site 1218, as determined by the tie points listed in Table T5. The integrated correlation of magneto- and biostratigraphic datum events to lithologic cycles from Sites 1218 and 1219 forms the basis for the development of an improved age model for the two sites.
The development of stacked MST data sets for Sites 1218 and 1219, together with shipboard and additional determinations of CaCO3 content for the two sites, allowed us to generate new high-resolution "estimated" carbonate content, based on a regression between lower-resolution direct measurements of CaCO3 and the stacked MST data. A high-resolution CaCO3 record is valuable, as it allows the investigation of depth- and latitude-dependent dissolution and productivity patterns as a function of the carbonate compensation depth (CCD). Using the stacked data sets of magnetic susceptibility, GRA bulk density, and the three color reflectance parameters L*, a*, and b*, we first extracted the dominant variation of these five parameters by principal component analysis. We then used a polynomial regression between the three leading principal components and measured CaCO3 values (e.g., Coxall et al., 2005), to develop high-resolution estimates of carbonate content from ~60 to 241 m depth on the Site 1218 rmcd scale, covering the earliest Oligocene to early/middle Miocene. Our results for Sites 1218 and 1219 are illustrated in Figure F6. Based on a comparison between the measured and estimated values, we determined the uncertainty of our estimated CaCO3 values to be on the order of 10%–15%. This uncertainty is lower for the interval from ~110 to 240 rmcd, but we note that the estimated values for Site 1218 are slightly (~10%) too low for the interval between 60 and 100 rmcd. Although it is possible to improve the match of measured and estimated CaCO3 values by calculating regressions over shorter subintervals, this also results in undesirable jumps in the estimated curves. A separate study (Vanden Berg and Jarrard, this volume) derived estimated calcite, opal, and terrigenous components by regression with light absorption spectroscopy data. We note that these two methods yielded similar results.
In combination with an astronomical age calibration of data from Sites 1218 and 1219 (Wade and Pälike, 2004; H.K. Coxall et al., 2005, unpubl. data), our estimated CaCO3 values combined with the stacked GRA bulk density estimates allow us to calculate more detailed mass accumulation rates of calcium carbonate for Sites 1218 and 1219. One interesting aspect of carbonate accumulation for the two sites arises by comparison with the tectonic subsidence histories of the two sites. Figure F6 illustrates that although CaCO3 is quite similar between Sites 1218 and 1219 throughout most of the Oligocene, Site 1219 shows a stronger decrease in CaCO3 values as this site approaches the depth of the CCD (also see Rea and Lyle, 2005, and Coxall et al., 2005, for more detailed discussion of the CCD development).
One final aspect of our integrated correlation effort is the calibration of our new rmcd scale to logging data. Similar procedures have been employed during previous ODP legs (Harris et al., 1995). The observation that mcd and rmcd scales are expanded compared to the mbsf scale arising from the length of the drill string has important consequences on properties that rely on absolute sediment thicknesses, such as sedimentation and mass accumulation rate estimates and synthetic seismogram studies. The hole-to-hole and site-to-site correlations have already demonstrated that the expansion in length of the rmcd scale compared to the mbsf scale is largely due to lithology-dependent core expansion caused by unloading. This view is confirmed by Rea and Gaillot (this volume), who computed core rebound coefficients by comparison of wireline in situ depths with core depths. Figure F7 illlustrates how stacked core data can be matched well to downhole logging data. The comparison of logging data with core data rests largely on measurements that represent the sediment porosity and/or density. Table T9 provides the tie points that match Site 1218 depths (in rmcd) to in situ logging depths (lmbsf). We observe that, on average, the in situ logging depth is ~89% of the rmcd scale, indicating that core expansion is on the order of 11%. One important result of our core-logging data comparison is the observation that both Formation MicroScanner and litho-density logging data show an intermediate step in the transition visible at ~213 lmbsf, at the base of Chron C13n. Sediment data also show this two-step transition but were only recovered in a core from a single hole (1218C). Thus, our match of logging data and core data can confirm that this step is an in situ feature at Site 1218 and not a coring artifact. We note that no correlation of logging data to core data was possible above ~70 mbsf, as the drill string was not removed completely during logging. We also observe that the logging data from Site 1218 are less noisy than those from Site 1219, and a match of Site 1219 to true depth is more easily accomplished by employing our site-to-site correlation in combination with the Site 1218 logging data than by using the Site 1219 logging data.