Polarity stratigraphies were obtained for Hole 986C (Fig. 1) and Hole 987E (Fig. 2). Borehole conditions were poor in Hole 986D below 400 meters below seafloor (mbsf); thus, extending the GHMT coverage was not possible. At Hole 987E, only the upper 500 m could be logged because of hole collapse. No interpreted polarity ages are presented with this data. The interpretations are included in Channell et al. (Chap. 10, this volume) as part of the paleomagnetic summaries and age models for Sites 986 and 987. The GHMT data for Hole 986C also provided important age control for a detailed study of Svalbard margin glacial history (Forsberg et al., Chap. 17, this volume).
At Hole 984B, only a magnetic susceptibility record was measured because the magnetic field readings from the basalt-rich sediments were so strong that they overwhelmed the NMRT sensor at that site. A detailed statistical analysis of the magnetic susceptibility was conducted to determine the empirical resolution of the SUMT (Kreitz, 1996). Understanding the resolution of these wireline geophysical tools relative to similar core measurements is critical to establish the correct correlation. The theoretical limits of the tool do not often apply to varying sediment types or enlarged borehole conditions. Also, the limited past usage of the tool provides little information on the typical "field" resolutions that can be expected.
The first step in core-log integration of the susceptibility data was to create a continuous core data set comparable to log data. A spliced section of core data for the upper 260 m was made by stacking all the susceptibility data from the multisensor track (MST) from all four holes (A-D) into the original shipboard meters composite depth (mcd) scale to create a revised mcd (rmcd) that included all the susceptibility data. The difference between the mcd and rmcd is that the former is a spliced record containing the "best representative" sections chosen from any of the drilled holes, whereas the latter contains all the magnetic susceptibility data from each hole that are computationally added to create one curve. This new stacked rmcd record represents the most complete estimate of magnetic susceptibility for 0-260 mbsf for Site 984 (Table 2, Table of Contents, ASCII Files, this volume). The overall depth scale of the rmcd and mcd is the same, although the relative depth of features (e.g., an obvious peak or trough) may have been shifted.
This stacked rmcd record was correlated with the log data beginning at ~90 mbsf (logging above this depth was not possible), and a mapping function was created to link the two data sets to a common depth scale (Fig. 3). The correlation function yielded a significant r2 value of 0.81. Cross-spectral analysis of these two time series provided coherency and gain spectra (Fig. 4A-D) that give tool resolution and signal attenuation information. The interval from 123 to 173 mbsf chosen for the analyses yielded a coherency close to the maximum theoretical resolution of ~50 cm (Fig. 4A, B). The maximum resolution observed in this interval was 53 cm. This represents a minimum of ~5 k.y. resolution at Site 984 where sedimentation rates average >10 cm/k.y. This is ample resolution to resolve orbital precession cycles (3-5 samples/cycle) and obliquity cycles (7-9 samples/cycle) without aliasing our results. Higher frequency climate signals <5 k.y. cannot be resolved with these data. The average resolution of the SUMT logs for all of Hole 984B was ~75 cm. In Figure 4C, the gain spectra for the interval (123-173 mbsf) showed that the logging data are actually amplified over the core data, which may reflect either good borehole conditions, poor core conditions or measurements, or some combination of these factors. However, because of the MST's higher sampling resolution of 5-7 cm (500- to 700-yr resolution), the gain spectra over the whole section from 90 to 260 mbsf shows that the core measurements are amplified relative to log measurements (Fig. 4D).
Analysis of core and logs below 260 mbsf where most of the cores were recovered using the extended core barrel (XCB) was complicated by sparser core recovery. The correlation factor drops to r2 = 0.65 from 0.81 but is still significant. The cyclicity in both core and log susceptibility measurements appears more pronounced in this lower section. However, there are cycles in the logs that do not appear in the core data. An example is shown in Figure 5 in the interval from 320 to 440 mbsf (note cycle at ~395 mbsf). This figure illustrates that poor core measurements, poor core condition, or, in the worst case, missing sediment may severely affect the susceptibility measurements. Despite their lower resolution, the log data successfully identify these cyclical changes, as well as potential core problems, and provide important constraints that improve both depth scales and estimates of sedimentation rate.