A Holocene paleosecular variation record from the northwestern Barents Sea continental margin



A high-resolution paleomagnetic and rock magnetic study has been carried out on sediment cores collected in glaciomarine silty-clay sequences from the continental shelf and slope of the southern Storfjorden trough-mouth fan, on the northwestern Barents Sea continental margin. The Storfjorden sedimentary system was investigated during the SVAIS and EGLACOM cruises, when 10 gravity cores, with a variable length from 1.03 m to 6.41 m, were retrieved. Accelerator mass spectrometry (AMS) 14C analyses on 24 samples indicate that the cores span a time interval that includes the Holocene, the last deglaciation phase and in some cores the last glacial maximum. The sediments carry a well-defined characteristic remanent magnetization and have a valuable potential to reconstruct the paleosecular variation (PSV) of the geomagnetic field, including relative paleointensity (RPI) variations. The paleomagnetic data allow reconstruction of past dynamics and amplitude of the geomagnetic field variations at high northern latitudes (75°–76° N). At the same time, the rock magnetic and paleomagnetic data allow a high-resolution correlation of the sedimentary sequences and a refinement of their preliminary age models. The Holocene PSV and RPI records appear particularly sound, since they are consistent between cores and they can be correlated to the closest regional stacking curves (UK PSV, FENNOSTACK and FENNORPIS) and global geomagnetic model for the last 7 ka (CALS7k.2). The computed amplitude of secular variation is lower than that outlined by some geomagnetic field models, suggesting that it has been almost independent from latitude during the Holocene.

1. Introduction

Sedimentary sequences with suitable lithological character and good paleomagnetic properties may provide valuable empirical inputs for the reconstruction of the geomagnetic field variability over geological times [e.g., Creer et al., 1972; Thompson, 1973; Verosub, 1977; Tauxe, 1993]. The reconstruction of the evolution of the magnetic field at various temporal and spatial scales is key to understanding the geodynamo models and the dynamics of the liquid outer core. The geomagnetic field varies over timescales from milliseconds to millions of years. Geomagnetic variations with timescales longer than 5 years are known as secular variation (SV) [Thompson and Barraclough, 1982; Bloxham and Gubbins, 1985]. Variations longer than 22 years, that is the duration of the turn-over of solar magnetic field, are of internal origin and reflect the magnetohydrodynamics of the Earth's iron-rich, electrically conducting, fluid outer core [e.g., Merrill et al., 1996]. Paleomagnetic and archeomagnetic data collected in the last decades allows reconstruction of geomagnetic paleosecular variation (PSV) for the past millennia, with the establishment of regional reference stacked PSV curves [e.g., Turner and Thompson, 1981, 1982; Hagstrum and Champion, 2002; St-Onge et al., 2003; Snowball et al., 2007; Barletta et al., 2010a, 2010b] and global geomagnetic field models ([e.g., Korte and Constable, 2005; Korte et al., 2005, 2009; Donadini et al., 2009] for an updated review see Donadini et al. [2010]). However, both reference PSV curves and global geomagnetic models are mainly based on the compilation of data collected from low- and midlatitudes.

As PSV data from high-latitude regions are still rare, there is a great interest to collect more widely distributed high-latitude PSV data to improve global geomagnetic field models [see Nilsson et al., 2010]. In particular, there is a lack of PSV data from sites within the surficial projection of the inner core tangent cylinder. This is a region of the outer core where a theoretical cylinder, tangent to the solid inner core equator and parallel to the axis of rotation, would separate distinct convective regimes. The tangent cylinder intersects the Earth's surface at a latitude of ±69.5°. The theory predicts that a different process might drive the geomagnetic field of the polar-regions: the flow within the tangent cylinder is thought to be moving as an upwelling polar vortex similar to that of a hurricane. Thermal winds and polar vortices within the tangent cylinder have been cited as the cause for low radial magnetic field over the north pole [Olson and Aurnou, 1999; Hulot et al., 2002; Sreenivasan and Jones, 2005, 2006] and an increased SV is predicted (4 times greater than the global average value for the time period 1870–1990 [Olson and Aurnou, 1999)]. Therefore PSV data from polar sites are of critical importance for geomagnetic field models. Some previous studies on late Pleistocene sediment cores from the Arctic region pointed out a geomagnetic variability larger than in intermediate and low latitudes, suggesting that the magnetic field has been strongly variable during at least the last 300 ka, with geomagnetic excursions more frequent and of longer duration than elsewhere [e.g., Nowaczyk and Antonow, 1997; Nowaczyk and Frederichs, 1999]. Anyway, recent rock magnetic studies pointed out that these features of apparent geomagnetic instability are due to partially self-reversed chemical remanent magnetizations acquired during the oxidation of detrital (titano)magnetite grains [Channell and Xuan, 2009; Xuan and Channell, 2010]. Furthermore, there are few Holocene relative paleointensity (RPI) determinations from marine sediments of the polar regions, limited to the North America margin for the Arctic region and to the Antarctic peninsula for the Antarctic region (Barletta et al. [2010a, 2010b] for the Canadian region; Lisé-Pronovost et al. [2009] for the Arctic Alaskan margin; Brachfeld et al. [2000, 2003] for the Antarctic peninsula) while a few other studies extend RPI reconstructions to the Late Pleistocene of the sub-Antarctic South Atlantic [Channell et al., 2000; Stoner et al., 2003] and the peri-Antarctic margins [Sagnotti et al., 2001; Macrì et al., 2005, 2006, 2010]. In conjunction with the PSV data, the reconstruction of detailed RPI records from the European high northern latitudes may complete the experimental evidence framework for a full understanding of the overall geomagnetic field variability.

In this study, we present the results of a high-resolution paleomagnetic and rock magnetic analysis on 5 sedimentary cores collected on the northwestern Barents Sea continental margin in the framework of the SVAIS and EGLACOM projects. The results allow us to reconstruct the directional PSV and the relative paleointensity changes of the geomagnetic field at high northern latitudes during the Holocene, thus providing original experimental constraints for testing models on the geomagnetic dynamo and the outer core dynamics. The data are supported by a robust, high-resolution, correlation and dating of the cores, and therefore have a valuable potential for establishing a firm temporal framework for the observed sedimentological changes and the inferred paleoenvironmental evolution.

2. Geological Setting and Age Constraints

The projects SVAIS (The development of an Arctic ice stream-dominated sedimentary system: The southern Svalbard continental margin) and EGLACOM (Evolution of a GLacial Arctic COntinental Margin: the southern Svalbard ice stream-dominated sedimentary system) were both conceived within the International Polar Year (IPY) in 2007–2009. The BIO Hespérides SVAIS cruise (29 July–17 August 2007) and the R/V OGS-Explora EGLACOM cruise (08 July–04 August 2008) investigated the Storfjorden glacial marine sedimentary system on the NW Barents Sea continental margin (Figure 1). The seafloor in this area was shaped by the action of several paleo-ice streams flowing into the Storfjorden and Kveithola glacial throughs, originating from the southern Svalbard archipelago and Spitsbergen banken.

Figure 1.

Location of the SVAIS and EGLACOM cores. (a) Location map of the study area in the NW Barents Sea. TMF: Trough-Mouth Fan; BY: Byørnøyrenna; KV: Kveithola; SF: Storfjorden. Arrows identify the location of the major paleo ice streams and paleo ice-flow direction. Bathymetry from GeoMapApp, http://www.geomapapp.org (Global MultiResolution Topography Synthesis [Ryan et al., 2009]). The inset at the upper right shows the northern polar region and the surficial projection of the inner core tangent cylinder at 69.5°N (red dashed line). (b) Detail of the square study area indicated in map A and location of the SVAIS-EGLACOM sediment cores. Color shaded relief bathymetric map of the Storfjorden and Kveithola Through Mounth Fans (TMF) area (from combined SVAIS and EGLACOM multibeam data) superimposed to the International Bathymetric Chart of the Arctic Ocean (IBCAO) gray scale shaded relief bathymetry after Jakobsson et al. [2008] (http://www.ibcao.org).

The overall objective of both projects is to contribute to the understanding of the evolution of glacial continental margins in response to ice sheet dynamics induced by climatic changes, in particular during the deglaciation phase following the last glacial maximum. During the SVAIS and EGLACOM cruises, 10 piston and gravity cores (SVAIS: 6 piston cores for a total of 26.70 m; EGLACOM: 4 gravity cores for a total of 9.23 m) were collected either on the continental slope and shelf (Figure 1), with a variable length from 1.03 m to 6.41 m (Table 1). We sampled 8 of these cores collected on the outer continental shelf and on the slope of Storfjorden and Kveithola throughs-mouth fans (TMFs), in water depths between 303 and 1839 m below sea level (bsl). The cores were retrieved at a latitude of about 75–76°N and are located within the Earth's core tangent cylinder region. Though paleomagnetic and rock magnetic properties were measured on the whole stratigraphic interval spanned by the cores, we focused our analysis on the Holocene interval, since it is characterized by excellent paleomagnetic properties. The Holocene record is well preserved in the continuous fine-grained homogeneous sediments of three cores from the mid slope (SV-04, EG-02 and EG-03) and of one core (SV-06) from the shelf, where however the sedimentary record contains hiatuses (Figure 2). An additional core (EG-01) from an upper slope gully was also analyzed, though the Holocene sedimentary record is very thin. Twenty-four AMS 14C calibrated ages are available for these five cores, and constrain the age of the sedimentary sequences (Figure 2 and Table 2). AMS 14C dating was performed at NOSAMS (Woods Hole Oceanographic Institution) on selected stratigraphic intervals, mainly using foraminifera tests, but also the bulk organic content of the sediments in the case of core EG-01 (see Table 2). Radiocarbon ages at NOSAM laboratory were calculated using 5568 years as the half-life of radiocarbon and the results were delivered as raw AMS 14C (yr BP) without reservoir corrections or calibration to calendar years. Ages calibration were then obtained through the calibration software Calib 6.0 [Stuiver and Reimer, 1993], using the marine09 calibration curve [Reimer et al., 2009], and applying an average marine regional reservoir effect ΔR of 84 ± 23 obtained from the Marine Reservoir Correction Database in Calib 6.0 for the northwestern Barents sea area (south of Svalbard). The mean values from the calibrated age range of ±1σ were then normalized to calendar year (conventionally 1950 AD) and are here reported as Cal. yr BP (Table 2).

Figure 2.

Lithologic logs and AMS 14C calibrated ages (red arrows) of the analyzed cores. The present study mostly refers to the Holocene interval of the sampled sequences, which consists of bioturbated and crudely stratified silty-clays with sparse ice rafted debris in the lower part of the sequence. The stratigraphic sequences of the cores collected from the continental slope appear continuous. Conversely the stratigraphic sequence of core SV-06, collected from the continental shelf (see Figure 1) is characterized by two distinct stratigraphic breaks (hiatuses) at depths of 102 (before 1900 yr BP) and 123 cm below seafloor (at 8200 yr BP).

Table 1. Location of the SVAIS and EGLACOM Cores
CoreLatitudeLongitudeWater Depth (m)Length (cm)
SV-0174° 58′ 2.82″13° 55′ 33.00″1813278
SV-0275° 13′ 42.42″14° 35′ 57.60″743641
SV-0375° 13′ 21.12″14° 37′ 14.94″761642
SV-0474° 57′ 25.50″13° 53′ 58.32″1839303
SV-0575° 06′ 42.18″15° 13′ 18.42″713632
SV-0676° 05′ 39.72″17° 43′ 31.92″303176
EG-0176° 06′ 12.08″13° 37′ 37.48″1069220.5
EG-0275° 12′ 54.44″13° 04′ 35.26″1722305.5
EG-0375° 50′ 36.92″12° 58′ 21.23″1432291.5
EG-0474° 51′ 53.78″16° 05′ 36.02″374105.5
Table 2. AMS 14C Dating and Calibrated Dates Using Software Calib 6.0a
CoreLab Ref.Depth (cm bsf)Sample TypeDescriptionProcessRaw AMS 14CAge Error∂13CCalibrated yr BP ± 1σCal. yr BP Applied to the Age Model
  • a

    Abbreviations: bsf: below sea floor; Nps: Neogloboquadrina pachiderma sx; OC: organic carbon; HY: hydrolysis.

EG-01OS-784093Sedimentpowdered sedimentOC483035−22.384890–50464968
 OS-78452102Sedimentpowdered sedimentOC28900190−24.4732428–3315632792
 OS-78453192Sedimentpowdered sedimentOC36700310−24.7641084–4163041357
EG-02OS-7838730ForaminiferaBenthic + planktonicHY4570130−254501–48294665
 OS-7838990Foraminiferabenthic + planktonicHY9460180010001–1046910235
 OS-78383182Forams and Pteropodsbenthic + plankt. + pterop.HY121001801.4113300–1366213481
EG-03OS-7838590.5Foraminiferabenthic + planktonicHY4910120−254957–52795118
 OS-78382160Foraminiferabenthic + planktonicHY85901300.018980–93149147
 OS-78324230.5Foraminiferabenthic + planktonicHY9740800.7310421–1059510508
 OS-8268562Foraminiferamixed planktonicHY7110300.57481–75577519
 OS-8268699Foraminiferamixed planktonicHY869030−0.449222–93629292
 OS-82687134Foraminiferabenthic + planktonicHY9790300.6410525–1059110557
 OS-82688187Foraminiferamixed planktonicHY12050400.0913328–1345013388
SV-06OS-777340Foraminiferabenthic + planktonicHY62030−1.22125–246186
 OS-7773660Foraminiferabenthic + planktonicHY207025−1.111500–15981549
 OS-7773783Foraminiferabenthic + planktonicHY239030−1.21869–19701920
 OS-77738123Foraminiferabenthic + planktonicHY677045−0.227161–72647213
 OS-77739124Molluscbivalve fragmentHY780050−0.428129–82768202
 OS-77740136Molluscbivalve fragmentHY8990601.579492–96379565

3. Sampling and Measurements

After dividing the cores into working and archive halves, each core section - about 1 m long -was subsampled with u-channel plastic holders for continuous paleomagnetic and rock magnetic measurements. U-channel samples were collected from the archive halves of the SVAIS cores (SV-03-04-05-06) in January 2008, at the Lithoteque of the Faculty of Geology, University of Barcelona. The working halves of the EGLACOM cores (EG-01-02-03-04) were sampled in July 2009 at the core repository facility of the Museo Nazionale dell'Antartide of Trieste.

Palaeomagnetic and rock magnetic measurements were carried out at the palaeomagnetic laboratory of the Istituto Nazionale di Geofisica e Vulcanologia, in Rome, in a magnetically shielded room. For each u-channel, we measured the low-field magnetic susceptibility (k) and the natural remanent magnetization (NRM) at 1 cm spacing. The NRM was measured on a small access (45 mm diameter) automated pass-through “2G Enterprises” DC 755 superconducting rock magnetometer (SRM), while k was measured using a Bartington magnetic susceptibility meter equipped with probe MS2C and mounted in-line with the SRM translating system. For the NRM measurements, we specify that the half-width of the response function of the three orthogonal Superconducting Quantum Interference Devices (SQUID) sensors of the SRM system varies between ca. 4.1 cm and ca. 6.7 cm for the transverse (X and Y axes) and the axial (Z axis) SQUID pick-up coils, respectively. It is well known that the different shape and widths of the response function curves of the three SQUID pick-up coils may result in fictitious effects on the paleomagnetic data, such as inclination shallowing or steepening [Roberts, 2006]. In our measurements, these spurious effects were corrected directly by the measuring software, by compensating the negative regions on the edge of the SQUID response functions for the X and Y axes and the broader width of the SQUID response function along the Z axis. The computed paleomagnetic data are therefore truly independent every ca. 5 cm and free from fictitious effects that may arise from uncompensated raw magnetic moment data. Moreover, we took particular care in avoiding any disturbance effects that may be introduced during the coring, cutting and sampling procedures and could result in remanence deflections due to plastic deformation of the soft sediments. In this study, we adopted a conservative approach, and disregarded the paleomagnetic data for ∼ 5 cm at both ends of each u-channel and stratigraphic gap.

After measuring the magnetic susceptibility, the NRM was progressively subjected to alternating field (AF) demagnetization in nine steps up to a maximum peak field of 100 mT (steps: 0, 10, 20, 30, 40, 50, 60, 80, 100 mT), by translating the samples through a set of three orthogonal AF demagnetizing coils in-line with the SRM, with NRM vectors measured after each demagnetization step.

After each NRM demagnetization cycle, an anhysteretic remanent magnetization (ARM) was imparted on each u-channel. For producing the ARM we used an in-line single-axis direct current (DC) coil coupled with the AF coils. We applied axial 0.1 mT bias DC field and symmetric AF peak of 100 mT along the Z axis, and translated the u-channel through the AF and DC coil system at a constant speed of 10 cm/s, that is the lowest speed allowed by the software running the measurements. This has an effect on the efficiency of the AF demagnetization and the intensity of the produced ARM [Sagnotti et al., 2003; Brachfeld et al., 2004]. The adopted procedure equals an AF decay rate of ca. 67 μT/half-cycle and results in the highest ARM intensity achievable with the employed instrumental setting and management software [Sagnotti et al., 2003]. From the AF demagnetization curves we computed the median destructive field (MDF) of the NRM (MDFNRM) and of the ARM (MDFARM), which are both almost single-component magnetic remanences. The MDF is defined as the value of the peak AF necessary to reduce the remanence intensity to half of its initial value.

4. Results

4.1. Rock Magnetism

The stratigraphic trends of the rock magnetic parameters are shown in Figure 3. The low-field magnetic susceptibility (k) and the ARM intensities mostly depend on the concentration of ferrimagnetic minerals. However, these two concentration-dependent rock magnetic parameters carry different information. The magnetic susceptibility values are determined by the contribution of all the rock forming minerals, in proportion to their relative abundance and specific magnetic susceptibility. The ARM is instead primarily sensitive to the concentration of fine, single-domain (SD), ferrimagnetic grains [King et al. 1982; Maher, 1988]. In the studied cores, the concentration-dependent magnetic parameters show values oscillating in a narrow range of variability especially for the Holocene interval (Figure 3). The magnetic susceptibility for the cores from the continental slope (SV-04 and the EGLACOM cores) fluctuates between 20 and 50 (×10−5 SI), with an average value between 30 and 40 (×10−5 SI), whereas it keeps distinctly lower values in the core from the continental shelf (SV-06), where the magnetic susceptibility profile is remarkably flat around an average value of 13 × 10−5 SI. Low magnetic susceptibility values are related to the composition of the terrigenous fraction containing diamagnetic minerals (mainly quartz and feldspars) and organic rich carbonate rocks deriving from the Mesozoic sequences of Svalbard, like the Agardhfjellet Formation [Sigmond, 1992]. In particular, in core SV-06 the homogeneously low k values in the upper 130 cm can be related to the abundant biogenic fraction including foraminiferas, nannofossils, diatomeas, pteropods, ostracods with large bivalve shells. A larger variability in magnetic susceptibility values is observed in the pre-Holocene intervals of the SV-04 and EG-02 cores (Figure 3). The ARM values mostly oscillates between 0.1 and 0.3 A/m, with a clear decreasing trend in the lower half of the cores, which is particularly evident in the pre-Holocene interval of all the cores from the continental slope and in the older part of the SV-06 core. Both MDFNRM and MDFARM mostly depend on the coercivity (composition and/or grain size) of the minerals carrying the remanence. For the cores on the continental slope (SV-04 and the EGLACOM cores), the MDFARM is remarkably constant throughout the whole stratigraphic sequence, with a mean value of 33 mT, while the MDFNRM is characterized by a step decrease at a depth of 20–30 cm, from a mean value of 43 mT above the step to a mean value of 31 mT below the step (Figure 3). For these cores, both MDFNRM and MDFARM show mean values and range of variability typical for magnetite grains [Maher, 1988]. For the core on the continental shelf (SV-06), the MDFNRM and the MDFARM show similar mean values (46 mT and 47 mT, respectively) and are both remarkably constant throughout the whole sequence. Finally, the ARM/k ratio depends on the grain size of ferrimagnetic minerals, with higher values for finer grained (single domain) ferrimagnetic particles and lower values for larger (multidomain) grains. The ARM/k ratio shows for all cores a downward decreasing trend, indicating a corresponding increasing trend in the magnetic minerals grain size, with values passing from ca. 1 × 103 A/m at the top to ca. 0.5 × 103 A/m at the bottom for the cores on the continental slope and from ca. 2–2.5 × 103 A/m at the top to ca. 0.8 × 103 A/m at the bottom for the core on the shelf (Figure 3). We conclude that the entire set of rock magnetic data indicate limited variations in the concentration, composition and grain size of the magnetic minerals, with a range of variability of the rock magnetic parameters within each core well within the same order of magnitude. Magnetic minerals are more abundant in the cores from the continental slope, whereas they are less abundant in the SV-06 core from the continental shelf. The MDF parameters show values typical for magnetite assemblages in the cores from the continental slope, whereas they likely indicate the presence of an additional magnetic component of higher coercivity in the core from the continental shelf. As discussed below, the stratigraphic trends of some rock magnetic parameters can be closely matched between cores and, together with paleomagnetic properties of the Holocene intervals, they can contribute to define a high-resolution correlation between cores. In Figure 4 we show the correlation of the ARM stratigraphic trends for the analyzed cores. For high-resolution core correlation and dating, all stratigraphic depths have been correlated to depth of core SV-04, which is the core with the highest number of available AMS 14C calibrated ages.

Figure 3.

Downcore variation of the main rock magnetic parameters measured for the analyzed cores. For each core, the plots show the stratigraphic trend of the intensity of the ARM after demagnetization in 20 mT AF, the magnetic susceptibility (k), the ARM/k ratio, the MDFNRM and the MDFARM values. The horizontal dashed lines indicate u-channel breaks. Lithologic logs and symbols as in Figure 2. See text for discussion.

Figure 4.

Correlation of the stratigraphic trends of anhysteretic remanent magnetization (ARM) for the analyzed cores. In the correlation procedure, all data have been transferred to the stratigraphic depth of core SV-04, which is the core with the highest number of available AMS 14C calibrated ages (see Table 2 and Figure 2). The ARM curves of the SV-04 and EG-02 cores match closely, with a correlation coefficient R = 0.94. The correlation looks also good for the SV-04 EG-03 and SV-04 EG-01 pairs, with R = 0.80 and 0.67 respectively, whereas it is poor for the pair SV-04 SV-06. The core SV-06, however is characterized by a lower content of magnetic minerals, as indicated also by the magnetic susceptibility values (see text and Figure 3) and the correlation between the SV-04 and SV-06 cores was mostly based on the radiometric ages and paleomagnetic data.

4.2. Paleomagnetism

The sedimentary sequence is characterized by excellent paleomagnetic properties. After removal of a viscous low coercivity remanence component at AF peaks of 10–20 mT, the paleomagnetic directions remain remarkably stable, with demagnetization vectors aligned along linear paths in orthogonal vector diagrams (Figure 5). This behavior allows precise identification of the characteristic remanent magnetization (ChRM), whose direction was computed by principal component analysis [Kirschvink, 1980] on the individual linear demagnetization paths, generally in the 10–60 mT or 10–80 mT AF demagnetization step intervals. The paleomagnetic results for the analyzed intervals are shown in Figure 6. The maximum angular dispersion (MAD) is generally very low (<2°) in the homogeneous fine-grained sediments, which is particularly valuable for PSV studies, whereas it reaches higher values (but less than 10°) for the lower intervals of cores SV-04, EG-01 and EG-02, characterized by variable and coarser-grained sediments (Figure 6). Since the cores were not azimuthally oriented, the ChRM declination of each u-channel has been arbitrarily rotated to align the mean value of the uppermost u-channel section with true north, and to line up the declination trends and values across consecutive u-channel sections. The ChRM inclination shows limited oscillations with arithmetic mean values around 70–80°, that are slightly shallower than the value expected at about 75° N latitude, where geometric considerations imply a geomagnetic axial dipole inclination value of 82.4°. The variation in the ChRM inclination values is less pronounced in the homogeneous silty clays of the Holocene intervals (with arithmetic mean ChRM inclination values of 75–78° for all cores), whereas it markedly increases for the coarser grained (ice rafted debris-rich) sediments of the lower intervals of the cores. A clear decreasing trend in ChRM inclination values is evident in the lower half of core EG-02 and is most likely due to lithological factors (Figure 6).

Figure 5.

Representative NRM demagnetization orthogonal vector diagrams [Zijderveld, 1967] for selected specimens subjected to AF demagnetization: open and closed symbols represent projections onto vertical and horizontal planes, respectively. The demagnetization data have been visualized and analyzed using the Remasoft program [Chadima and Hrouda, 2006].

Figure 6.

Downcore variation of the paleomagnetic properties of the analyzed cores. For each core, the plots show the stratigraphic trend of the intensity of the natural remanent magnetization (NRM), ChRM declination and inclination and maximum angular deviation (MAD). The horizontal dashed lines indicate u-channel breaks. The vertical red dashed line in the ChrM inclination plot indicates the value expected using the geocentric axial dipole (GAD) model. Lithologic logs and symbols as in Figure 2.

4.3. Relative Paleointensity

The NRM intensity may reflect the strength of the geomagnetic field during the time of the acquisition of the remanence but it also depends on the concentration of the natural remanence-carrying minerals. Provided that the basic requirements of a substantial uniformity in lithology and in concentration, composition and grain size of the magnetic minerals are met [King et al., 1983; Meynadier et al., 1992; Tauxe, 1993; Valet and Meynadier, 1998; Valet, 2003; Tauxe and Yamazaki, 2007], curves of relative paleointensity (RPI) variation are generally reconstructed by normalizing the NRM intensity for an opportune concentration-dependent rock magnetic parameter. To estimate RPI variation, we normalized the NRM remaining after demagnetization in 20 mT AF (NRM20mT) by magnetic susceptibility (k) and by the ARM intensity left after demagnetization in 20 mT AF (ARM20mT). Both methods resulted in a similar pattern (Figure 7) and therefore support a general coherency between the two normalization procedures. In particular, the two normalized curves match closely for the upper part of all the cores from the continental slope. We notice that the low magnetic susceptibility values measured in the SV-06 core from the continental shelf may result in unreliable oscillations in the NRM20mT/k curve (Figure 7). Therefore, for all the cores we used the NRM20mT/ARM20mT curves as the preferred RPI proxy.

Figure 7.

Downcore variation of the normalized relative paleointensity (RPI) curves NRM/k and NRM20mT/ARM20mT of the analyzed cores. For each core the two curves match closely for the Holocene interval, except for core SVAIS-06, due to the low, and poorly defined, values of the magnetic susceptibility in that core. The pre-Holocene intervals of each core are characterized by a marked lithological variability; in these stratigraphic intervals the two normalized RPI curves tend to show similar variations but with different values and amplitudes. Lithologic logs and symbols as in Figure 2.

5. Discussion

As formerly anticipated, in this study we limit our analyses to the Holocene interval of the sampled sequences. This choice relies on three main factors:

1. The Holocene interval is characterized by a homogeneous fine-grained lithology and a substantial magnetic homogeneity, as indicated by the rock magnetic parameters. This represents the ideal condition for continuous measurements on u-channel samples [see Roberts, 2006].

2. The paleomagnetic data for the Holocene may be correlated to the existing high-resolution PSV and RPI reference curves from stacking of regional paleomagnetic data or from global geomagnetic models.

3. The Holocene interval is also characterized by extremely well defined ChRM directions, with MAD values below 2° (Figure 6), which are therefore particularly suitable for PSV reconstructions and for high-resolution correlation with the available coeval reference PSV curves.

Keeping into proper account the constraints provided by the available AMS 14C dates, we tried to match the paleomagnetic record for each of the analyzed cores to the PSV and RPI variations expected at the cores location according to the global geomagnetic model CALS7K.2 [Korte and Constable, 2005] and to the closest Holocene PSV and RPI regional stack curves. For this purpose, we used the PSV and RPI stacks from 7 Fennoscandian lakes (FENNOSTACK and FENNORPIS of Snowball et al. [2007]) and the PSV stack from 3 British lakes (UK PSV stack of Turner and Thompson [1981, 1982]) (Figure 8).

Figure 8.

ChRM (a) declinations and (b) inclinations of the EGLACOM and SVAIS cores plotted as a function of age and compared with the UK and FENNOSTACK PSV stack curves and with prediction from the global geomagnetic main field model CALS7k.2 (see text for references and discussion).

These regional PSV stack records have been relocated to the EGLACOM and SVAIS core sites via the virtual geomagnetic pole (VGP) method [Noel and Batt, 1990]. In this method, the geomagnetic field is modeled by an inclined geocentric dipole and the remanence direction measured at a given site is converted to the correspondent direction observed at a reference site via a virtual magnetic pole. The method allows a direct comparison of the obtained paleomagnetic data with the paleomagnetic declination and inclination expected at the location of the analyzed cores according to the UK PSV and FENNOSTACK reference curves. The core correlation based on paleomagnetic data (PSV and RPI) has been also adjusted and checked keeping into account the constraints provided by the correlation of rock magnetic parameters (Figure 4). As a result, the PSV and RPI cross-correlation allows establishment of an improved age model for the cores. The available AMS 14C calibrated ages have been integrated with constraints derived from correlation with both PSV and RPI reference curves (Figures 8 and 9) and rock magnetism (Figure 4). The comparison of the ChRM declination records (Figure 8a) indicate similar trends for all curves, with larger variation in the interval between 2000 and 4000 yr BP. The VGP passed close to the cores site location at about 2350–2400 yr BP, when the ChRM is almost vertical (Figure 8b). A few sharp ChRM declination swings are indicated for ages older than 7500 yr BP; however, since they are recorded by a few measurements on single cores it is doubtful that they represent actual variation of the geomagnetic field. The comparison of the ChRM inclination records (Figure 8b) indicate also a good agreement between the curves and a remarkable match with the expected trend according to the global geomagnetic model CALS7K.2. A significant inclination shallowing was only recorded for the second u-channel, from the top, of the SV-06 core, spanning around 2500 to 3500 yr BP (Figure 8b). Similarly, the RPI records from the EGLACOM and SVAIS cores have been visually correlated to the available reference RPI curves (CALS7k.2 and FENNORPIS) (Figure 9). A good fit is observed between the RPI curve of each core (but SV-06) and the expected trend according to the CALS7k.2 model. The RPI data indicate relative maxima at about 2500 yr BP and at 9500 yr BP (Figure 9). RPI record of core SV-06 appears essentially flat for ages younger than 8500 yr BP (Figure 9).

Figure 9.

Relative paleointensity of the EGLACOM and SVAIS cores plotted as a function of age and compared with the FENNORPIS stack curve and with prediction from the global geomagnetic model CALS7k.2. For comparison purposes, the magnetic induction values expected at the core location according to the CALS7k.2 global geomagnetic model and the standardized RPI data of the FENNORPIS stack have been scaled so that a value of 1 was assigned to their maximum value (see text for references and discussion).

Paleomagnetic data have been furthermore used to reconstruct VGP paths. In Figure 10 we show the reconstructed VGP path for the time interval between 2000 and 5000 yr BP for cores SV-04 and EG-02. This is the period with the larger PSV variation (see also Figure 8). About 5000 yr BP the VGP was located over the Arctic Ocean, then a marked swing to lower latitudes is recorded between 4000 and 3300 yr BP, which brought the VGP over Siberia, followed by another fast westerly swing (known as “f-e event” in the UK reference PSV curve [Turner and Thompson 1981, 1982]), between 2600 and 2000 yr BP which brought the VGP over Greenland at about 2000 yr BP.

Figure 10.

Virtual Geomagnetic Pole (VGP) path reconstructed for cores SV-04 and EG-02 for the period 2000–5000 yr BP. This is the time interval, during the Holocene, characterized by the larger geomagnetic variation. The VGP path is similar for both cores and implies pronounced swings in paleomagnetic declination and inclination. The orange dashed circle indicates the surface projection of the inner core tangent cylinder (see text for discussion).

5.1. Stacking and Holocene VGP Path

The Holocene paleomagnetic data from the continental slope cores were merged in a stack curve for declination and inclination. The paleomagnetic stack curve was obtained by using Fisher statistics [Fisher, 1953] on data selected with an age sliding window of 200 yr. In particular, starting from 0 yr BP, the mean age of the sliding window was progressively increased in steps of 200 yr. At each step the Fisher statistics was computed on the ChRM directions from all cores, whose estimated age fall within a range of ± 100 yr relative to the mean age. This procedure ensured a number of data N higher than 3 for all the steps in the period 600–10000 yr BP (N values vary between 3 and 22).Of course, the paleomagnetic stack value was not computed for time intervals with a number of data N less than 3 (i.e., for age younger than 600 yr BP). Core SV-06 was not included in the stacking for its anomalous ChRM inclination and RPI records (see Figures 8 and 9). Figure 11 shows the excellent match of the SVAIS-EGLACOM stacked ChRM declination and inclination with the reference PSV curves (especially the UK curve for the f-e event and the FENNOSTACK and CALS7K.2 for ages older than 4000 yr BP) and the variations expected according to the CALS7K.2 model. Analogously, we reconstructed a RPI stack curve using the data from the same cores, by computing the arithmetic mean of the data falling within a sliding window of the same spacing (200 yr). There is a remarkably good match with the CALS7K.2 model over the whole time period (Figure 12). The comparison with the FENNORPIS stack indicate a remarkable match for the last 4000 yr, and a similar trend, but with different values, for ages older than 4000 yr BP (Figure 12). The RPI stack points out the occurrence of RPI maxima at 1.8, 2.4 and 8.8, 9.6 kyr BP. The younger RPI maxima matches very well the features outlined by the FENNORPIS stack and CALS7K.2 model. The occurrence of a RPI high for the early Holocene was also pointed out by the global analysis of Holocene dipole moment variation carried out by Ohno and Hamano [1993] and by the paleomagnetic data obtained from different sequences of the Canadian Arctic [Barletta et al., 2008, 2010b], which however indicated a single RPI maximum at about 8.5–9.0 krs BP. In Figure 13, we show the Holocene VGP path reconstructed form the EGLACOM-SVAIS PSV stack. The VGP path in the early Holocene (10000–8600 yr BP) tends to describe a counterclockwise (CCW) loop in the “Pacific” sector of the Arctic Ocean (Figure 13), with a VGP position close to the Bering Strait at ca. 9000 yr BP. After a period of low variation, between 8400 and 7000 yr BP, with VGP mostly clustered over the “Canadian” sector of the Arctic Ocean, another CCW loop is then repeated in the “Pacific” sector of the Arctic Ocean between 6800 and 5000 yr BP, with again a VGP position close to the Bering Strait at ca. 6000 yr BP. After another period of low variation between 5000 and 4000 yr BP, with VGP confined at high latitudes in the “Canadian” sector of the Arctic Ocean, a marked swing occurred between 4000 and 3000 yr BP (Figure 13), which rapidly brought the VGP position at relatively low latitudes over northern part of western (European) Russia. A rapid and pronounced westerly swing then occurred between 2600 and 2000 yr BP, which corresponds to the well-known “f-e” geomagnetic feature originally pointed out in the UK PSV stack. Finally, from 1800 to 600 yr BP the VGP maintained a nearly polar position, with a limited oscillation to lower latitudes at about 1200–1000 yr BP, when the VGP reached a position close to the northern tip of Novaya Zemlya (Figure 13). Overall, during the Holocene the reconstructed VGPs oscillate between 90° and 70°. This range of variability is larger than that expected by the CALS7k.2 model [Korte and Mandea, 2008], but, as for the model, the VGP always maintain within the surface projection of the inner core tangent cylinder (TC), with the exception of the large swing at 2.8–3.2 kyr BP when the VGP slightly exceeded the limit of the TC region (Figure 13). The swing to relatively low latitudes at about 2.8–3.2 kyr BP was also pointed out in the former Holocene global analysis of the geomagnetic field by Ohno and Hamano [1993], though the VGP latitude remains well within the TC in their model. The other major swing of the reconstructed Holocene VGP path is expressed by a fast westerly drift of the VGP between 2.6 and 2.0 kyr BP, which was also pointed out by both the former models of Ohno and Hamano [1993] and Korte and Mandea [2008], and brought the VGP position very close to the sampling sites at about 2.4 kyr BP (Figure 13).

Figure 11.

Comparison of the SVAIS-EGLACOM stacked ChRM (a) inclination and (b) declination curves with the reference UK and FENNOSTACK PSV stack curves and with the trends expected according to the global geomagnetic model CALS7k.2. For the stack curves, the red diamonds indicate the mean value computed with Fisher statistics on data selected with a sliding window of 200 yr, and the error bars indicate the uncertainty at the 95% confidence level.

Figure 12.

Comparison of the SVAIS-EGLACOM stacked RPI normalized curve with the FENNORPIS stack curve and with expected trends from the global geomagnetic model CALS7k.2. For comparison purposes, the magnetic induction values expected at the core location according to the CALS7k.2 global geomagnetic model and the standardized RPI data of the FENNORPIS stack have been scaled so that a value of 1 was assigned to their maximum value. For the stack curve the red diamonds indicate the arithmetic mean value computed on data selected with a sliding window of 200 yr, and the error bars indicate the standard deviation.

Figure 13.

The Holocene VGP path reconstructed form the EGLACOM-SVAIS PSV stack, shown in 6 time windows. The orange dashed circle indicates the surface projection of the inner core tangent cylinder. The data indicate periods of relatively low geomagnetic variation (i.e., between 7000 and 8500 yr BP) and periods of large geomagnetic variation (i.e., between 3500 and 4000 yr BP and between 2800 and 2000 yr BP; see text for discussion).

5.2. Implications for Holocene Geomagnetic Models

Finally, we analyzed the VGP scatter for the 4 cores with more than 100 data points through the Holocene. We computed the VGP scatter value (S) for the Holocene interval of each core, expressed as the angular standard deviation of the VGP data distribution [McFadden et al., 1988; McElhinny and McFadden, 1997], both considering all the available VGP data and by using the iterative cut-off method proposed by Vandamme [1994] (Scut-off). The cores are characterized by relatively low S values, which are remarkably similar around 16° for cores SV-04, SV-06 and EG-02 and slightly lower (12°) for core EG-03. The Scut-off values are lower and range between 7 and 14 (Figure 14). These VGP scatter values are considerably lower than those predicted for the latitudinal dependence of VGP scatter by various geomagnetic field models at such high latitudes (Figure 15). In particular, the phenomenological model G of McFadden et al. [1988] and McElhinny and McFadden [1997] defines a quadratic curve fitting a set of VGP scatter data plotted against paleolatitude and predicts that at the latitude of the SVAIS and EGLACOM cores the S values should be of about 21° (Figure 15). In the TK03.GAD statistical model of Tauxe and Kent [2004] the time varying geomagnetic field is described as a “Giant Gaussian Process” following Constable and Parker [1988] (i.e., assuming that the gauss coefficients glm and hlm, except for the axial dipole term g10 and in some models also the axial quadrupole term g20, have zero mean and standard deviations that are a function of degree l). The TK03.GAD model predicts that VGP distributions are circularly symmetric and the VGP scatter increases with latitude, with S values of about 23° when no cut-off is applied and about 19° with the cut-off criterion of Vandamme [1994] (Figure 15). Finally, Johnson et al. [2008] tried to reconstruct the characters of the time average geomagnetic field during the last 5 Ma on the basis of a synthesis of paleomagnetic data collected from various lava flows at 17 distributed locations and 8 additional regional data sets, and indicated that the Brunhes data are compatible both with models that predict a flat VGP dispersion with latitude [e.g., Constable and Parker, 1988] and with models predicting an increase in VGP dispersion with increasing latitude, such as those discussed above. Johnson et al. [2008] indicated that the Brunhes data show little variation of S with latitude and computed a mean Brunhes VGP scatter value of 16°, which is just the value computed for three of our cores. In any case, Johnson et al. [2008] pointed out the need of new paleomagnetic data from high northern latitudes to discriminate between the different models.

Figure 14.

Equal area plots of Holocene VGP positions computed for the four cores (SV-04 and 06, EG-02 and EG-03) with more than 100 data points (N > 100). The small circle indicates the cut-off angle estimated by the Vandamme [1994] method and the red points outside such small circles indicate the data discarded according to such cut-off angle. For each core, we indicate the number of data selected according to the Vandamme cut-off versus the total number of data and the computed VGP scatter with and without the Vandamme cut-off (Scut-off and S, respectively).

Figure 15.

VGP scatter values (S) of the SVAIS and EGLACOM cores versus latitude, compared with the values from the PSV database of Johnson et al. [2008] for the last 5 Ma, including standard deviation bars, and with predictions according to the model G of McElhinny and McFadden [1997] and the model TK03.GAD [Tauxe and Kent, 2004], computed with and without the cut-off criterion of Vandamme [1994]. For model predictions, the dashed lines denote 95% error bounds.

The paleomagnetic data collected from the SVAIS and EGLACOM cores hence indicate that they can be used to trace the PSV and RPI variation occurred at high latitude sites during the Holocene and allows a reliable correlation with existing reference curves and models. At the same time, they indicate that the range of geomagnetic field variation, as expressed by VGP dispersion during the same time interval, has been relatively low and comparable to that observed at low-latitude sites (Figure 15). This may be partly due to the smoothing effects associated with the acquisition of a depositional remanent magnetization, but it suggests that geomagnetic variability within the tangent cylinder was no higher than elsewhere during the last 10 kyr.

6. Conclusions

The paleomagnetic properties of the EGLACOM and SVAIS cores provide an excellent record that allows us to:

1. Reconstruct the paleosecular variation (PSV) and the relative paleointensity (RPI) of the geomagnetic field at high northern latitudes for the Holocene (extension to older ages is possible but it is hampered by the effects of lithologic changes and the lack of PSV reference curves). We propose new Holocene PSV and RPI reference stack curves for the region.

2. Achieve a high-resolution correlation and dating of the cores, with a substantial improvement of previous age models. This is critically important for the reconstruction of timing and rates of paleoenvironmental changes which followed the last deglaciation, as documented by sedimentological facies and trends. This allows us to develop a sound chronological framework for paleoclimatic studies that are the subject of ongoing and future researches on the same cores.

3. Recognize that the VGP scatter amplitude for the studied cores is lower than that predicted by some geomagnetic field models. This observation suggests that the amplitude of secular variation has been almost independent of latitude during the Holocene. This study provides direct evidence of geomagnetic field dynamics over a ten thousand year time scale at latitudes of 75°–76° N. The data provide constraints for the geodynamo models and do not support presumptions about an increased geomagnetic SV within the inner core tangent cylinder.


This study has been supported by Spanish IPY projects SVAIS (POL2006–07390/CGL) and IPY-NICE STREAMS (CTM2009-06370-E/ANT), and by IPY-related Italian projects OGS EGLACOM and PNRA MELTSTORM. The authors wish to acknowledge the cooperation of captains Pedro Luis de la Puente García-Ganges (BIO Hespérides), Franco Sedmak and Carmine Teta (OGS-Explora) and their crew, and of the technical staff at the UTM (CSIC, Barcelona) and the RIMA Department (OGS, Trieste). We thank the careful reviewers by Stefanie Brachfeld and Francesco Barletta. Their comments and suggestions allowed us to improve considerably the manuscript.