We used progressive alternating-field (AF) demagnetization of split-core sections and discrete samples along with rock magnetic experiments to characterize the paleomagnetic signal and resolve the magnetization components recorded in the recovered core. The component interpreted to record the magnetization at or near the time of deposition was then used to construct a magnetostratigraphy for Hole 1223A.
All split-core and discrete samples have a sizable drilling overprint, which is characterized by a steep downward direction and by a radial-horizontal component that points toward the center of the core. In the ODP core orientation system, the latter results in a strong bias in the declinations toward 0°. Initial natural remanent magnetization (NRM) measurements are thus characterized by inclinations >+60° and declinations of ~0°. In general, <30-mT AF demagnetization removes the drilling overprint but also reduces the magnetization by ~90%. Along the periphery of the core, the drilling overprint can be more demagnetization resistant, sometimes requiring AF demagnetization of up to 60 mT, particularly in the vitric tuff units. Small biases may thus persist in some of the split-core measurements for which AF demagnetization generally did not exceed 50 mT.
Following removal of the drilling overprint and excluding intervals disturbed by drilling, the sediments above 7 mbsf have stable remanent magnetizations that provide a record of the geomagnetic field at or near the time of deposition. Though more difficult to resolve, the underlying units similarly have stable remanent magnetizations, but the acquisition mechanism may be more complex than a depositional remanent magnetization (DRM) or postdepositional remanent magnetization (pDRM). Given the porous, permeable nature of the vitric tuff units and their degree of alteration (see "Alteration" in "Lithology"), some subsequent thermal or chemical remagnetization cannot be precluded.
The magnetostratigraphy provides age constraints that can be used to test hypotheses about the origin of the lithologic units. For example, are the volcaniclastic turbidites in Unit 2, the unconsolidated black sand in Unit 4, the vitric tuffs in Units 5 and 11, and/or other units coeval with previously dated landslide events? Alternatively, if the origin of the lithologic units can be established independently, then ages for each unit provided by the magnetostratigraphy will be the primary constraint on the age of depositional (landslide?) events. At this point, the ages provided by the magnetostratigraphy indicate that Units 2 through 14 are >1.8 Ma and <2.6 Ma. Thus all are roughly coeval with the range of ages for the Nuuanu Landslide, which is interpreted to have occurred between 1.95 and 2.15 Ma by Herrero-Bervera et al. (2002).
We made 28,062 remanent magnetization measurements along the archive-half sections from Hole 1223A. Measurements were made every 1 cm before and after AF demagnetization. All sections were progressively demagnetized in steps of 1-5 mT up to peak fields of between 50 and 70 mT. The high peak fields were necessary to resolve the characteristic remanent magnetization (ChRM) direction and allowed us to complete principal component analysis (PCA) on each interval measured along a section. Magnetic susceptibility was measured on whole-core sections every 2.5 cm (Fig. F56). Both remanence and susceptibility data for these sections are available from the ODP Janus database.
For interpretation of the data, we extracted the split-core results from the ODP Janus database and removed all measurements made within 5 cm of the section ends, as these measurements are biased by edge effects. We then removed all measurements made in regions disturbed by drilling or where there were gaps, such as those caused by removal of whole-round samples. The disturbed intervals and gaps used are given in Table T8. The resulting cleaned data set contains 22,843 measurements made at 1455 intervals along the archive halves (Table T9).
Prior to demagnetization, the inclinations display a very strong tendency for directions that point steeply downward and to the north; the inclinations are positive and typically >60° over the entire cored interval, indicating the presence of a steep downward-directed drilling overprint (Fig. F56). Also, the declinations are biased toward 0°, which for azimuthally unoriented cores (as are all the Leg 200 cores) indicates the presence of a radial overprint. The overprint is radially inward for the APC and XCB cores from Hole 1223A. Both radial and vertical overprints are observed during most ODP legs and are artifacts of the drilling process.
The overprint is similar, though not identical, to applying an isothermal remanent magnetization (IRM) of roughly 5-25 mT to the core. Thus, the intensity measured prior to demagnetization is a good proxy for concentration of magnetic minerals (mainly magnetite and titanomagnetite), as is the susceptibility, with both giving similar relative variations downhole (Fig. F56). The median intensity is 6.09 x 10-1 A/m prior to demagnetization but is reduced to 2.61 x 10-2 A/m after 20-mT demagnetization. Thus, over 90% of the magnetization of the samples is generally removed in order to remove most or all of the drilling overprint.
Orthogonal vector demagnetization plots illustrate that a ChRM direction can be isolated for most of the cored interval following removal of the drilling overprint by AF demagnetization of 10-30 mT (Fig. F57). For sediments and rocks from Cores 200-1223A-2H through 6X, the drilling overprint is more resistant to demagnetization, requiring demagnetization up to 50 mT to remove most or all of the overprint (Fig. F58). After the drilling overprint has been removed, the measured intervals generally display linear demagnetization paths that trend toward the origin or nearly to the origin. It is this magnetization component that is considered the ChRM, particularly the highest coercivity component following >30-mT demagnetization. In some cases the ChRM is not resolved, although the change in direction during demagnetization can often be used to infer the magnetic polarity of the interval. In the latter case, the directions obtained during progressive demagnetization tend to lie along a great circle path on a stereonet, with the beginning of the path representative of the drilling overprint and the directions after each demagnetization step trending closer and closer to the ChRM direction and possibly reaching it as shown in Figure F58.
In the vector demagnetization diagrams, the systematic offset from the origin of the linear demagnetization paths, such as is apparent by the decay of the declination in Figure F57, may be caused by a sensor for one axis of the magnetometer being slightly more noisy than the others, by small magnetic fields existing in the sensor region, or by real magnetization components that are unremoved by AF demagnetization. In some cases, the magnetization of the interval has been reduced to near the noise level of the magnetometer, but generally this is not the case. Furthermore, the offset is consistent over many measurements rather than being random noise. Given that many of the demagnetization paths trend to the origin, we think it is most likely that any consistent magnetization direction measured for an interval following AF demagnetization is related to unremoved components of magnetization in that interval.
In order to estimate the ChRM, we conducted PCA (Kirschvink, 1980) on the data using a program that iteratively searches for the demagnetization steps that minimized the size of the maximum angular deviation (MAD) angle, which is a measure of how well the vector demagnetization data fit a line. MAD values <10° are typically considered to provide lines that fit the observations well. The program requires at least four demagnetization steps be used, never uses steps lower than a user-defined value, does not require that the best-fit PCA line pass through the origin of the plot (the "free" option of standard PCA), and generally favors high-coercivity components over low-coercivity components in samples with multicomponent magnetizations. Because the drilling overprint persisted beyond 20-mT demagnetization in some of the intervals, particularly in the vitric tuffs though less so in the soft sediments, we only used results from 30-mT or higher demagnetization for Core 200-1223A-1H and from 40-mT or higher demagnetization for the other cores in the PCA. For comparison, the program also computes a Fisherian mean of the highest three or four demagnetization steps for each interval. This is referred to as the stable endpoint direction. Typically, only the highest three demagnetization steps are used in the average, unless the mean of these three directions has a precision parameter <200 (a measure of dispersion), in which case the fourth highest demagnetization step is included. In cases where the precision parameter is <200, the program will first search for outliers and remove them if they lie >10° from the mean. Both the PCA and stable endpoint directions for all 1455 intervals are given in Table T10. Comparison of the stable endpoint with the PCA direction can be useful for indicating where unremoved or partially unremoved magnetization components exist or where progressive demagnetization has been ineffective in revealing linear demagnetization paths. To assist with this comparison, we have computed the angular distance between the two directions and included it in Table T10.
Comparison of inclinations obtained from the methods show good agreement for most of the upper 7 m of the section (Fig. F59). The agreement is an indication that AF demagnetization has been successful in removing overprints and that a single component of magnetization remains. The mean inclination of this direction in the normal and reversed polarity intervals is consistent with the inclination expected for a geocentric axial dipole, which is 40.3° for normal polarity rocks and -40.3° for reversed polarity rocks at Site 1223 (Fig. F59). Thus, the ChRM provides a record of the magnetic field direction at the time the sediments were being deposited (a DRM) or shortly thereafter (a pDRM). Because AF demagnetization did not fully resolve the ChRM in some intervals below 7 mbsf, the PCA and stable endpoint directions sometimes do not coincide. In such cases, we think the stable endpoint direction will generally be closer to the true ChRM direction because it is an average of the directions from the highest demagnetization steps.
We measured the NRM of 25 discrete samples before and after AF demagnetization (Table T11). Of these, 11 samples (~7 cm3 volume) were collected from sediments from Cores 200-1223A-1H and 2H and the other 14 are from two 1-cm-thick intervals in the lowest vitric tuff samples are from interval 200-1223A-6X-1, 55-56 cm, and 11 samples are from interval 200-1223A-6X-2, 90 cm. The samples from the vitric tuff were cut into small pieces with volumes <1 cm3. The pieces were taken from different parts of 1-cm-thick slices of the working half to assess how the drilling overprint varied with location.
Following demagnetization, we conducted ARM and IRM experiments (see"Paleomagnetism" in the "Explanatory Notes" chapter) on the 11 sediment samples and on 8 of the vitric tuff samples (Tables T12, T13).
PCA and stable endpoint analyses of the NRM were conducted in the manner described above for the split cores, with results given in Table T14. The discrete samples were demagnetized in more steps and at higher peak fields, which allows the ChRM to be analyzed in more detail. Orthogonal vector demagnetization plots for sediments from Core 200-1223A-1H show the samples have two components, the low-coercivity drilling overprint, which is removed after 10- to 20-mT demagnetization, and the ChRM (Figs. F60, F61). The discrete sample directions agree well with what was obtained from split-core measurements, and the inclinations agree with the expected geocentric axial dipole inclination at Site 1223.
The few samples taken from the lowest vitric tuff unit illustrate that the ChRM can also be resolved by detailed demagnetization of samples taken from near the center of the core (Fig. F62), where typically 10- to 20-mT demagnetization removes the drilling overprint. Samples from near the periphery, however, are more severely affected by the drilling overprint and often require demagnetization up to 55-60 mT to remove the overprint. By 60-mT demagnetization, only a few percent of the initial NRM remains, making it more difficult to accurately estimate the direction of the remaining ChRM. This would explain the difficulty we had resolving the ChRM in some intervals from the split-core measurements (e.g., compare Fig. F62 to Fig. F58).
Although we have not had time to thoroughly evaluate the magnetic mineralogy that carries the remanent signal, the AF demagnetization behavior, IRM acquisition (Fig. F63), backfield IRM results, and volcanogenic lithology together provide strong circumstantial evidence that magnetite or titanomagnetite is the main magnetic carrier. A higher-coercivity mineral, perhaps hematite, could also be present as evidenced by a small component of the ChRM that remains after AF demagnetization of 80 mT and after applying a saturation IRM of 1 T and then demagnetizing the sample up to 80 mT.
The magnetic inclination derived from the ChRM is used as the indicator of polarity, where positive inclinations indicate normal polarity and negative inclinations indicate reversed polarity. Because the cores are azimuthally unoriented, we do not use the declination as it does not provide direct evidence of polarity, although it could be used to indicate relative declination changes. Because coring gaps of several meters are present within the cored interval and because the lithologic section contains units that indicate discontinuous sedimentation, such as the turbidites, some polarity intervals may not have been recovered or may be missing. No independent age constraints are currently available, so we have based the correlation of the polarity intervals with polarity chrons upon the assumption that the top of the lithologic sequence was 0 Ma. In addition, seismic data and previous piston coring in the area indicate that the sediments are likely to be a few million years old (Naka et al., 2000; Herrero-Bervera et al., 2002).
With the above constraints and caveats, the magnetostratigraphic record from Hole 1223A (Figs. F59, F64; Table T15) appears to contain all the major chrons and subchrons from Chron C1n (the Brunhes Chron; 0.0-0.780 Ma) through Chron 2r (1.950-2.581 Ma). Possibly Subchron C2r.1n (2.140-2.150 Ma), which falls within Chron C2r, is recorded in the bioturbated clay (lithologic Unit 6) near the base of Section 200-1223A-3X-2. The interval from the core catcher of Core 200-1223A-1H to the bottom of Core 200-1223A-6X is otherwise reversely magnetized and is interpreted to lie within Chron C2r.
Uncertainty in the polarity of the bioturbated clay unit (Unit 6) in Section 200-1223-3X-2 results from an ambiguity in the orthogonal demagnetization diagrams from split-core measurements. After removal of the drilling overprint, a linear demagnetization path indicative of a normal polarity interval is apparent, but the demagnetization path decays toward a shallow inclination, which with further demagnetization could be reversed polarity rather than the origin of the plot. As discussed above, there are multiple reasons why such behavior may occur, one of which is that there is an unremoved reversed polarity component. Discrete samples have been collected from this interval and will be measured postcruise in a less noisy magnetic environment, which should help resolve the ambiguity.
A notable anomaly in the magnetostratigraphy is the thinness of the upper normal polarity interval, which spans only the top 14 cm of Core 200-1223A-1H. This is thinner than expected by ~1 m based on prior piston coring in the vicinity (Herrero-Bervera et al., 2002). Furthermore, there is a distinct possibility that the BHA was slightly below the mudline when Core 200-1223A-1H was shot. The resulting recovery of <9.5 m (a full core) was taken to indicate that the mudline had been recovered. Partial recovery could have been caused instead by the cutting shoe encountering the turbidite in Unit 2, the base of which contained granules. The core was deformed through the base of the turbidite, indicating the piston corer did cut through the unit with difficulty. Additionally, penetration of the cutting shoe is often inhibited in sandy units, as was also the case for Core 200-1223A-2H. Thus, we may not have recovered the very upper meter or so of the sedimentary section or sedimentation rates may vary locally.
Given the magnetostratigraphic record, ages for the lithologic units and their boundaries can be estimated by interpolation between the observed reversals (Table T16). Linear interpolation implies constant sedimentation between the reversals, which is likely a poor assumption given the nature of the lithologic section. It does, however, give ages consistent with stratigraphic superposition and the magnetostratigraphic record. Thus, we provide ages based on linear interpolation, but the reader should be aware of the limitations of this simplistic age model.
Similarly, average sedimentation rates are calculated between reversals (Table T15). Coring gaps between cores and other intervals of poor recovery or suspect recovery, such as inferred here for the top of Chron C1n, may cause some inaccuracies in these estimates. Most of the reversals, however, occur within Core 200-1223A-1H, in which coring gaps are not an issue but hiatuses or erosional events may remain problematic. The average sedimentation rates are, of course, poor estimates of the true sedimentation rates for turbidites and landslide deposits.