Inclination records of the three major sites are shown in Fig. 3, which suggests that variations of a timescale of a few tens of 1000 years occur in common. Inclination records of each gravity-core site are presented in Fig. 4 and Additional file 1, Additional file 2, Additional file 3, Additional file 4, and Additional file 5 together with the records of the target sites of the inter-core correlations with magnetic susceptibility. The agreement of inclination records between the coupled sites is generally good. Inclination data of all nine sites are superimposed in Fig. 5a. Inclination variations with a timescale of a few tens of thousand years are visible. Sudden inclination decreases were observed at about 28, 35–40, 64 85, 113, and 188–195 ka in two or more cores. Part of them are close in age to known geomagnetic excursions, the Mono Lake excursion at ~33 ka, the Laschamp at ~41 ka, the Blake at ~120 ka, and the Iceland Basin at ~188 ka (Roberts 2008), and they may have recorded the excursions. On the other hand, low inclination spikes at about 128 and 180 ka, which were recorded in only one core (Site GC9 + PC5), may not be of geomagnetic origin.
We constructed a stacked inclination record of the Okhotsk Sea. The age interval of the stack was limited for the last 200 kyr because the number of available cores for older ages is small, three or less. First, the mean inclination of each site was calculated between 0 and 100 ka (Table 1). Inclination data from the uppermost 20 cm of each core were removed because of possible physical disturbance of surface sediments during coring. For calculating the mean inclination, incfish.py of the pmag.py programs for inclination-only data (Tauxe 2010) was used. Inclination anomaly (∆I) of each site, which is defined as observed mean inclination minus GAD inclination, ranges from −4.9° to 5.9° (Table 1), and the mean ∆I of the nine sites is 1.0° ± 3.2° (the mean ∆I is 2.1° ± 1.7° when calculated between 0 and 200 ka for four sites that cover this time interval). Next, we chose Site GC1 + PC7 as a representative location, and differences in inclinations expected from the differences in site latitudes were corrected using GAD inclinations. In addition, inclinations of each site were shifted slightly so that the ∆I of each site becomes zero. Then, the mean and standard deviation were calculated at 1-kyr intervals after resampling of each record (Fig. 5b). Long-term inclination variations are visible on the stacked record; intervals of shallower inclinations occur at about 25–45, 75–90, 110–135, and 185–200 ka in the Okhotsk Sea.
We then compare the inclination stack from the Okhotsk Sea with the stacks of relative paleointensity and inclination from the West Caroline Basin in the western equatorial Pacific (Yamazaki et al. 2008; Fig. 5). The inclination records of the two regions show parallel variations, although the two regions are about 5000 km apart; the periods of shallower inclinations in the Okhotsk Sea at about 25–45, 75–90, 110–135, and 185–200 ka coincide with those of negative steeper inclinations in the West Caroline Basin. These inclination shifts are synchronous with relative paleointensity lows in general (Fig. 5), as pointed out in the West Caroline Basin by Yamazaki and Oda (2002) and Yamazaki et al. (2008), although a period of paleointensity low around 100 ka contradictorily has steeper, but not shallower, inclinations in the Okhotsk Sea.
Yamazaki and Oda (2002, 2004) and Yamazaki et al. (2008) explained the inclination and paleointensity variations observed in the western equatorial Pacific that relative contribution of a persistent quadrupole component increased when the strength of the GAD field decreased. The western equatorial Pacific is known to have a large ∆I associated with a quadrupole component, and the sign of ∆I flips with polarity reversals (Johnson and Constable 1997). From sediment cores in the West Caroline Basin, ∆I of −6.5° ± 2.8° (N = 13) in the Brunhes Chron was reported (Yamazaki et al. 2008). The phase of the paleointensity–inclination correlation also flipped with polarity reversals, in-phase in the Brunhes Chron and anti-phase in the Matuyama Chron (Yamazaki and Oda 2002). The coeval inclination variations observed in the Okhotsk Sea, however, cannot be explained by this model because ∆I in this area is near zero, as observed in our cores.
Hoffman and Singer (2008) proposed that a magnetic field at Earth’s surface during polarity transitions and excursions is dominated by a field generated only in the shallower part of the core, designated the SCOR field. They also proposed that the SCOR field is represented by persisting higher-degree terms other than the GAD field and lasts for a timescale of ~106 years. Both the Okhotsk Sea and western equatorial Pacific are within a region of outward directed flux in the 1590–1990 time-averaged non-axial-dipole (NAD) field, which extends from Europe and Asia to the south of Australia (Hoffman and Singer 2008; Constable and Korte 2015). Thus, the observed inclination shifts around excursions, shallower in the Okhotsk Sea and deeper toward negative in the western equatorial Pacific, may be explained by a larger contribution of the SCOR field when the GAD was weak. In the western North Atlantic, the 1590–1990 time-averaged NAD field has opposite inward flux. In this region, periods of steeper inclinations than the site averages, in which rapid excursional directional swings are intercalated, were observed near the Laschamp excursion (Lund et al. 2001, 2005) and “excursions 13a, 15a, and 17a” (~510, ~573, and ~666 ka, respectively; Lund et al. 2001). This may support the SCOR field model. The coherent shifts in inclinations among the Okhotsk Sea, western equatorial Pacific, and North Atlantic may also be explained by hypothetical dipole wobbles. However, the correspondence of the inclination shifts to paleointensity lows prefers the SCOR field model.