SCIENTIFIC AND OPERATIONAL OBJECTIVES OF EXPEDITION 303

Climate-Related Objectives

Stratigraphy is the fundamental backbone of our understanding of Earth's history, and stratigraphic resolution is the main factor that limits the timescale of processes that can be studied in the past. Sub-Milankovitch-scale climate studies face the challenge of finding a stratigraphic method suitable for correlation at this scale (see Crowley, 1999). Even under optimal conditions, chronologies based on ¹⁸O are unable to provide sufficient stratigraphic resolution. Within the North Atlantic region, recent improvements in stratigraphic resolution have resulted in a new understanding of the dynamics of millennial-scale climate variability over the last ~100 k.y. (e.g., Bond et al., 1993; McManus et al., 1994; van Kreveld et al., 2000; Sarnthein et al., 2001). These stratigraphies have utilized chronologies from the Greenland Summit ice cores (GRIP/GISP2) and the recognition of regional lithostratigraphic linkages such as Heinrich events and higher-frequency ice-rafted debris (IRD) layers, ash layers, and susceptibility cycles combined with planktonic/benthic ¹⁸O, acceleration mass spectrometry (AMS)¹⁴C dates, and geomagnetic paleointensity data (e.g., Bond et al., 1992, 1993, 1999; McManus et al., 1994; Stoner et al., 1998, 2002; Voelker et al., 1998; Kissel et al., 1999; Laj et al., 2000).

The objective of the expedition is to integrate stable isotope and relative geomagnetic paleointensity data with paleoceanographic proxies and, in so doing, generate integrated North Atlantic millennial-scale stratigraphies for the last few million years.

Understanding the mechanisms and causes of abrupt climate change is one of the major challenges in global climate change research today (see Clark et al., 1999, p. vii) and constitutes a vital initiative of the Initial Science Plan of IODP. Ideally, the best approach to this problem would be to collect records of climate variability from a dense geographic network of sites, but this is impractical in paleoceanographic research. In the absence of dense coverage, the most viable approach is to obtain long, continuous time series from key regions and compare the response and timing of climate change among sensitive regions. We intend to develop PACs to establish the phase relationships among globally distributed millennial-scale records. Building global correlations on millennial timescales is an essential step to understanding the underlying mechanisms of abrupt climate change.

A persistent quasi-periodic ~1500 y cycle that is apparently independent of glacial or interglacial climate state has been observed for the past 80 k.y. (Bond et al., 1997, 1999). The millennial-scale cyclicity in the Holocene appears to be mirrored in the last interglacial (marine isotope Stage [MIS] 5e) and is defined by the same petrologic proxies in both interglacials. The presence of this cyclicity in interglacials and the IRD petrology that defines it indicate that the cyclicity does not reflect ice sheet instability or changes in calving of coastal glaciers, but rather changes in sources or survivability of drifting ice, driven by changes in the size and intensity of the subpolar cyclonic gyre (Bond et al., 1999). The Holocene cycles reflect a mechanism operating on at least hemispheric scale (Sirocko et al., 1996; Campbell et al., 1998; De Menocal et al., 2000), indicating that the MIS 5e and Holocene cyclicities have a common origin possibly related to solar forcing (Bond et al., 2001). The implication is that the 1500 y cycle may have been a dominant feature of the Earth's ocean-atmosphere climate over a very long time, which leads to the following questions:

Recently published evidence from earlier interglacials (MIS 11 and 13) in both the subpolar and subtropical North Atlantic suggests that interglacial cyclicity at those times may have had a significantly longer pacing, on the order of 5000 y or more. The interglacial records from MIS 11 and 13 in Oppo et al. (1998) and McManus et al. (1999), for example, show rather sporadic events that, regardless of age model, cannot occur every 1500 y. Similarly, an MIS 11 record from ODP Site U1063 off Bermuda indicates large shifts in benthic ¹³C on the order of 5–6 k.y. (Poli et al., 2000). In contrast, data from MIS 11 at ODP Site 980 implies the presence of 1–2 k.y. pacing (McManus et al., 1999), suggesting that the 1500 y cycle may be operating in MIS 11 and in other pre-MIS 5e interglacials. If this is true, then the interglacial climate variability may reflect a persistent, perhaps periodic, process operating continuously within the Earth's climate (rather than red noise resulting from a highly nonlinear climate system).

The best evidence for the 1500 y cycle during interglacials seems to be coming from IRD proxies that monitor changes in the subpolar gyre in the North Atlantic. Our drilling sites are positioned to monitor such changes. In contrast to ODP Site 980 (Feni Drift), our sites are located well within the main present-day routes of iceberg transport into the North Atlantic and are therefore well suited to capture faint interglacial signals that respond to shifting ocean surface circulation. If we can connect the 1500 y cycle to paleointensity records, we will have a means of directly comparing both signals with climate records from well outside the North Atlantic region.

Understanding the changes in the ice sheet-ocean-atmosphere system that gave rise to millennial-scale climate changes requires the precise long-distance correlation of ice cores and marine sediment cores. Geomagnetic paleointensity records from marine sediment cores have been shown to contain a global signal suitable for fine-scale correlation (see Meynadier et al., 1992; Guyodo and Valet, 1996; Channell et al., 2000; Stoner et al., 1995, 2000, 2002; Laj et al., 2000), at least for the last glacial cycle.

Beyond the range of AMS ¹⁴C dating, geomagnetic paleointensity data may provide the only viable means of sub-Milankovitch-scale long-distance correlation. Paleointensity records have been applied to stratigraphic correlation in the Labrador Sea for the last 200 k.y. (Stoner et al., 1998), throughout the North Atlantic for the last 75 k.y. (Laj et al., 2000), and for the Atlantic realm for the last 110 k.y. (Stoner et al., 2000). As variations in geomagnetic paleointensity control atmospheric production of ¹⁰Be and ³⁶Cl isotopes and the flux of these isotopes is readily measurable in ice cores, paleointensity records in marine cores provide an independent link between marine sediment and ice core records (e.g., Mazaud et al., 1994). The lows in paleointensity at ~40 and ~65 ka are readily identifiable as highs in ¹⁰Be and ³⁶Cl fluxes (Baumgartner et al., 1998; Raisbeck et al., 1987) in the Vostok and GRIP ice cores, respectively. Frank et al. (1997) showed that 10⁴–10⁵ y variability in ¹⁰Be production rate, as determined from globally distributed deep-sea cores over the last 200 k.y., can be matched to sediment paleointensity data. This observation and the similarity of globally distributed paleointensity records indicate that much of the variability in paleointensity records is globally correlative. The few high-resolution paleointensity records available beyond 200 ka also indicate that fine-scale features are correlative. For example, the paleointensity record for the MIS 9–11 interval (300–400 ka) from the Iceland Basin (ODP Sites 983 and 984) can be correlated to the sub-Antarctic South Atlantic (ODP Site U1089) at suborbital (millennial) scale (Stoner et al., 2003).

In addition to the practical use of magnetic field records for correlation of climate records, further drilling of high-sedimentation-rate drift sites will impact the "solid Earth" theme of IODP by documenting the spatial and temporal behavior of the geomagnetic field at unprecedented resolution. Such data are required to elucidate processes in the geodynamo-controlling secular variation and polarity reversal of the geomagnetic field. Recently derived records of directional secular variation and paleointensity from drift sites (e.g., ODP Legs 162 and 172) have substantially advanced our knowledge of magnetic secular variation, magnetic excursions, and directional/intensity changes at polarity reversal boundaries (see Channell and Lehman, 1997; Channell et al., 1998, 2002; Lund et al., 1998, 2001a, 2001b). Numerous directional magnetic excursions have been observed within the Brunhes Chron at ODP Leg 172 drift sites (Lund et al., 1998, 2001a, 2001b) and in the Matuyama Chron at ODP Leg 162 sites (Channel et al., 2002). These excursions (or brief subchronozones) correspond to paleointensity minima and have estimated durations of a few thousand years. From ODP Leg 162 records and records from the Pacific Ocean, it has been suggested that spectral power at orbital frequencies in paleointensity records may reflect a fundamental property of the geodynamo (Channell et al., 1998; Yamazaki, 1999; Yamazaki and Oda, 2002) rather than climate-related influences on paleointensity records (Guyodo et al., 2000).

There is no doubt that North Atlantic drift sites have revolutionized our understanding of the behavior of the geomagnetic field by providing Brunhes Chron paleomagnetic records at unprecedented resolution. The records can now provide useful constraints for numerical simulations of the geodynamo (e.g., Glatzmaier and Roberts, 1995; Gubbins, 1999; Coe et al., 2000). As a result of these parallel advances, our understanding of the geomagnetic field is on the threshold of substantial progress.

The Expedition 303 sites will provide high-resolution paleomagnetic records extending through the Matuyama Chron (to ~3 Ma). They will allow us to assess the temporal and spatial variability of the geomagnetic field in the Brunhes Chron and compare these records with reversed polarity records from the Matuyama Chron, and to address the following questions:

The nonaxial-dipole (NAD) components in the historical field vary on a centennial scale, and this has been interpreted to indicate similar repeat times in the past (Hulot and Le Mouël, 1994; Hongre et al., 1998). If this is correct, paleointensity records from cores with sedimentation rates less than ~10 cm/k.y. are unlikely to record anything but the axial dipole field. On the other hand, standing NAD components have been detected in the 5 m.y. time-averaged field, although the distribution of these NAD features remains controversial (Kelley and Gubbins, 1997; Johnson and Constable, 1997; Carlut and Courtillot, 1998). Refinement of time-averaged field models as the paleomagnetic database is augmented will lead to a better grasp of how the nonzonal terms in the time-averaged field may influence paleointensity records.