Rock magnetism
The magnetic fabrics were analyzed via AMS. The directions of the maximum susceptibility (Kmax), intermediate susceptibility (Kint), and minimum susceptibility axes (Kmin) of the AMS ellipsoids were plotted on lower hemisphere equal area projections. Figure 2 shows the AMS data for depths of 67.85–66.11 m within the same core segment for which the directional continuity was confirmed. The data are not corrected for tilt. The directions of Kmax and Kint indicate a girdle distribution perpendicular to Kmin, which indicates a flattening fabric, and all results of the AMS analysis showed similar flattening fabrics (foliations). The mean inclination of Kmin between depths of 67.85 and 66.11 m was 74°, and the dip of the foliation plane formed by Kmax and Kint is therefore estimated to be 16°. The estimated dip angle of the bedding plane at the core site is approximately 12°, which suggests that the foliation plane reasonably coincides with the bedding plane. In the four sections of the core, bedding planes comprised of intercalated thin sandstone beds were directly examined. In each segment, the difference between the observed maximum dip direction and the direction of the foliation plane was less than 10°. Accordingly, we treat the foliation plane, as deduced from the AMS data, as a bedding plane at the site.
Thermomagnetic curves from samples at depths of 70.01 and 58.58 m are shown in Fig. 3. Both curves show an identical Curie temperature of approximately 570 °C, which indicates that magnetite (or Ti-poor titanomagnetite) is the predominant magnetic carrier. The heating curves in air as well as in vacuum show a slight increase above 450 °C, but the cooling curves do not show such inflections. When iron sulfide minerals such as pyrite are present in sediments, they are changed into magnetite and maghemite and, finally, into hematite through heating by oxidation that occurs primarily above 450 °C, which produces a heating curve that at first shows an increase in the magnetization and then a decrease (Passier et al. 2001). Therefore, although the thermomagnetic curves suggest a minor amount of iron sulfide minerals, (titano)magnetite is considered to be the main magnetic mineral in the samples.
A plot of anhysteretic remanent magnetization susceptibility (κARM) versus magnetic susceptibility (κ) for Core M and Core I is shown in Fig. 4a. According to the grain size estimates of King et al. (1983), the mean grain sizes of the magnetic grains in the specimens from Core M and Core I should primarily range from 1 to 5 μm. Figure 4b shows a Day plot (Day et al. 1977) using selected specimens from Core M and Core I. Both plots indicate that pseudo-single domain (PSD) or mixture of single domain (SD) and multidomain (MD) magnetite grains are the predominant magnetic carriers. The Core I data are from Kusu et al. (2014).
A plot of the magnetic susceptibility (κ) and anhysteretic remanent magnetization susceptibility (κARM) on a logarithmic horizontal scale is shown in Fig. 6g. The κ and κARM show similar gradual fluctuations, which range within an order of magnitude. Additionally, the fluctuation patterns of κ and κARM strongly resemble each other. Therefore, it is possible that the changes in the values of κ and κARM are related to changes in the magnetic mineral content in the specimen, and the changes in the magnetic mineral content with depth are relatively small.
Remanent magnetization
Typical orthogonal vector diagrams for NRMs during progressive AFD and THD are shown in Fig. 5. Most of the results show that the secondary magnetization components were removed by AF demagnetization at a peak field strength of 15 mT. Large fluctuations in the inclinations can be recognized between depths of 67.03 and 64.38 m (Fig. 5e), and steep upward inclinations are observed between depths of 64.36 and 63.20 m (Fig. 5d). In the lower part, below a depth of 67.05 m, the inclinations are positive (Fig. 5b, f), and the maximum angular dispersion (MAD) values for the ChRMs are generally less than 5°. However, values slightly over 10° are seen in a few specimens just below a depth of 67.05 m (Fig. 6d). In the upper part, above a depth of 63.18 m, the inclinations are negative (Figs. 5a, c), and the MAD values for the ChRMs are generally less than 10°, approximately 80 % of which are less than 5° (Fig. 6d). ChRMs were not obtained from five specimens (from depths of 66.61, 66.59, 66.55, 66.49, and 65.81 m), which had very weak remanences (Fig. 5g) and/or MAD values exceeding 25°. From these inclinations, we infer that the Olduvai normal polarity subchronozone occupies the lower part below a depth of 67.05 m, and the post-Olduvai Matuyama reversed polarity chronozone occupies the upper part above a depth of 63.18 m.
Orientation of the core
The core was not oriented, and absolute paleomagnetic declinations therefore could not be derived directly from the measured ChRMs. To obtain the declinations of the ChRMs, we reconstructed the core orientation via the following process. First, we calculated the dip azimuth of the foliation plane, which was deduced from the AMS in each core segment where the continuity of the cores was confirmed. We then adjusted the dip azimuth to the north direction. Using this process, we connected core segments between depths of 80 and 50 m. However, significantly different declinations of ChRMs were observed at three depth intervals: 79.78 to 76.59 m, 73.71 to 72.05 m, and 51.48 to 51.17 m (intervals are indicated by the gray bars in Fig. 6b). We then readjusted the azimuth directions at the three intervals to minimize the differences in the declinations between the three intervals and other depths. Next, the dip azimuth observed in the core was adjusted to the dip azimuth of the bedding plane observed around the core site (strike and dip: N63° W and 12° NE), and the tilt was then corrected. Finally, the mean declination of the normal polarity (below 67.05 m) was adjusted to 0°.
Paleomagnetic directional changes and relative paleointensity
The resultant declinations and inclinations of the ChRMs are shown in Fig. 6a, c. The reversed polarity declinations were generally aligned to 180°, and the inclinations after the tilting correction were aligned to ±53°, which would be expected given a geocentric axial dipole (GAD) at the core site. Large fluctuations in the declinations and inclinations were observed between depths of 67.03 and 64.38 m, and steep upward inclinations were observed at approximately 63.5 m depth. The VGP latitudes calculated from the ChRMs are shown in Fig. 6e. The VGP moves across the equator from the Northern Hemisphere to the Southern Hemisphere between 65.69 and 65.67 m depth. The VGP begins to move rapidly at a depth of 67.03 m, and the VGP settles at approximately 63.44 m after moving to the Southern Hemisphere.
The rock-magnetic measurement results indicate that the predominant carrier of magnetic remanences in the sediments is PSD, or a mixture of SD and MD-sized (titano)magnetite (Figs. 3 and 4), and the fluctuations in κ and κARM range within an order of magnitude (Fig. 6g). From these results, the sediment satisfies the criteria for relative paleointensity reconstruction proposed by Tauxe (1993). To obtain an appropriate proxy for relative paleointensity, we compared the NRM/ARM ratios from three different coercivity fractions, which were between 20 and 40 mT, 30 and 50 mT, and 30 and 40 mT. The variations of the three coercivity fractions were almost identical, and we therefore used the NRM/ARM ratio between 20 and 40 mT as a proxy for relative paleointensity in this study (Fig. 6f).
VGP and paleointensity
The VGP paths between 70.0 and 60.0 m are shown in Fig. 7a. In the upper Olduvai polarity transition, the VGP apparently did not move within a preferred longitudinal band but settled in several VGP cluster areas. The VGP can be observed in five areas (Fig. 7a): (A) eastern Asia near Japan, (B) the Middle East, (C) eastern North America (North Atlantic), (D) off southern Australasia, and (E) the southern South Atlantic off South Africa. The VGP apparently moved rapidly between the clusters.
In the relative paleointensity of the normal polarity, a maximum value is observed at 75.50 m depth (Fig. 6f) and it then decreases with the short-term oscillations. Before the onset of the paleomagnetic directional change, the relative paleointensity has a peak at 67.75 m depth (with a value approximately 56 % of the maximum value), and values then decrease at 67.15 m to approximately 11 % of the maximum value. Subsequently, the relative paleointensity slightly recovers up to 66.89 m. In this interval, the directional change begins at a depth of 67.03 m, and the VGP near the North Pole rapidly moves to cluster A (Fig. 6e). The relative paleointensity subsequently drops at 66.5 m depth. Between 66.53 and 65.81 m, the relative paleointensity is approximately 9 % of the maximum value. In the same interval, the VGP settles in cluster C. The relative paleointensity then remains low up to 64.56 m. In this interval, the VGP in the Northern Hemisphere moves into the Southern Hemisphere between depths of 65.69 and 65.67 m, and the VGP moves to cluster D at ~65.17 m. Afterwards, the relative paleointensity gradually increases. From 64.65 to 63.46 m, the VGP settles in cluster E, and then, the VGP moves to the area near the Antarctic.
During the polarity reversal, especially between depths of 67.03 and 63.46 m, the relative paleointensity drops to approximately 12 % of the maximum value. After the large movements in the VGP, the relative paleointensity remains low for a time and then gradually recovers. At 58.38 m depth, the relative paleointensity returns to the peak value observed at 67.75 m, just before the paleomagnetic directional change began, and at about 55 m, the relative paleointensity becomes comparable to the maximum value observed pre-reversal.
Age model
To construct an age model for the interval spanning the upper Olduvai polarity reversal (1.8–1.7 Ma) in Core M, we correlated our δ18O profile generated from tests of G. inflata in the core (Fig. 8g) with the LR04 global δ18O stack (Lisiecki and Raymo 2005; Fig. 8k) and determined the peaks of marine isotope stages (MIS) 64 and 63.
The peak observed at 75.00 m (1.61 ‰) is comparable to the peak of MIS 64 because it is just below the Olduvai termination, and the broad low at approximately 58.00–52.00 m (0.78–0.83 ‰) can be correlated to MIS 63 because it is just above the Olduvai termination. The maximum amplitude of δ18O in the core is 0.83 ‰. This is quite comparable to the amplitude of 0.71 ‰ observed in the LR04 stack between MIS 64 and MIS 63. We therefore assigned the peak at 75.00 m and the broad low at approximately 58.00–52.00 m to MIS 64 and to MIS 63, respectively.
It is, however, difficult to determine a particular position within the broad trough at approximately 58.00–52.00 m for correlation with the lowest point of MIS 63. To determine which position is suitable as the MIS 63 trough, we used Core I, which was drilled in the same study area (Fig. 1c). From Core I, the Olduvai termination boundary, magnetic susceptibility, and relative paleointensity were obtained, and a significant trough comparable to MIS 63 was recognized in the δ18O curve (Kusu et al. 2014). The relative paleointensity and δ18O curve for Core I were determined using the same methods as in this study for Core M. Figure 8 shows a comparison between the two cores and includes the horizons of MISs 64 and 63. The vertical axes are expressed as core depths. Because a coarse-grained tuff bed, a pumice-rich lapilli tuff bed, and a fine-grained tuff bed (A, B, C, respectively, in Fig. 8) are commonly recognized in both cores, these three tephra beds were used to correlate the two cores. In the interval between the coarse-grained tuff bed (A) and the fine-grained tuff bed (C), the fluctuating patterns in the two cores’ magnetic susceptibility records coincide well (Fig. 8i), as do the records of in situ paleomagnetic inclinations (Fig. 8j). We therefore used the magnetic susceptibility curves to more closely correlate the two cores. The red lines in Fig. 8 indicate the correlation between the two cores using representative tie horizons in the vicinities of MIS 64 and MIS 63.
From this correlation, we determined that the most likely position corresponding to the trough of MIS 63 in Core M is at a depth of 58.00 m. We assigned MIS 64 (1795.0 ka) to the peak at 75.00 m and MIS 63 (1772.5 ka) to the trough at 58.00 m, which serves as an age model for Core M. Based on the age model, the average sedimentation rate between MISs 64 and 63 is 73.9 cm/kyr.