letters - MavDISK

Vol 435|9 June 2005|doi:10.1038/nature03674

LETTERS Geomagnetic dipole strength and reversal rate over the past two million years Jean-Pierre Valet1, Laure Meynadier2 & Yohan Guyodo3

Independent records of relative magnetic palaeointensity from sediment cores in different areas of the world can be stacked together to extract the evolution of the geomagnetic dipole moment1,2 and thus provide information regarding the processes governing the geodynamo. So far, this procedure has been limited to the past 800,000 years (800 kyr; ref. 3), which does not include any geomagnetic reversals. Here we present a composite curve that shows the evolution of the dipole moment during the past two million years. This reconstruction is in good agreement with the absolute dipole moments derived from volcanic lavas, which were used for calibration. We show that, at least during this period, the time-averaged field was higher during periods without reversals but the amplitude of the short-term oscillations remained the same. As a consequence, few intervals of very low intensity, and thus fewer instabilities, are expected during periods with a strong average dipole moment, whereas more excursions and reversals are expected during periods of weak field intensity. We also observe that the axial dipole begins to decay 60–80 kyr before reversals, but rebuilds itself in the opposite direction in only a few thousand years. Stacks of globally distributed relative palaeointensity records provide the best source of information regarding the evolution of the relative strength of the dipole field in the past. The remarkable consistency of the first stacks (Sint-200 and Sint-800) with alternative data sets derived from near bottom sea-floor magnetic anomalies4,5 and the good agreement with cosmogenic isotope records6,7 validated their geomagnetic origin. With the accumulation of recent studies from different oceanic basins, it has become possible to perform a first global stack for the past 2 Myr in order to document the evolution of the geomagnetic field during periods with field reversals. The most recent database8–21 of relative palaeointensity incorporates 15 records between 0.6 and 1.5 Myr ago. All display the same main features, but some discrepancies are caused by subtle changes in the response function of the sediment22 and by inaccuracies in the depthversus-time correspondence assigned to each sediment core. In fact, it has been shown23–25 that very few variations shorter than 20 kyr can be extracted with good confidence from composite curves. Guyodo and Valet25 recently correlated the intensity minima of 15 records spanning the time interval 0.75–1.25 Myr ago using the ODP site 983 (ref. 11) record as a target for the common timescale. The stack derived from this data set was calculated after normalizing the signals of all 15 records to both equal mean and variance over the interval common to all records. However, the variability of the signal decreases with temporal resolution. For this reason, it is difficult to combine different stacks obtained from different cores over different periods. In order to produce a new global stack for the time interval 0.6–2 Myr ago and then to be able to combine it with Sint-800, we thus restricted the present study to records with similar temporal resolution.

We selected 10 out of 15 records that were obtained with similar resolution and displayed coherent features. Records that were either poorly constrained in time or probably contaminated by lithological variations were not considered. Except for some intervals, core KS752 (ref. 8) is of rather poor quality and was not retained. Cores LC07 (ref. 9) and KK78 (ref. 10) show a very large increase of palaeointensity between 1.2 and 1.8 kyr ago, which is not present in the other records. Last, the records from ODP sites 983 (ref. 11) and 1101 (ref. 12) do not go beyond 1.1 Myr ago. In addition, they were obtained with much higher resolution than the rest of the database, and may have climatically contaminated intervals26. Before any calculation, the records were resampled at 1 kyr intervals. The record from ODP site 851, which covers the entire 2-Myrlong interval with relatively constant deposition rate, was selected as a target for establishing the common timescale. The influence of the

Figure 1 | Various stacks obtained for the period 0.6–2 Myr ago. a, b, Obtained by matching a subset of 10 cores tied to the reversals (a) or tied to reversals and intensity minima (b). c, Obtained using all 15 records. d, Obtained with 10 cores between 0.6 and 1.3 Myr ago (black curve) and 5 cores between 0.6 and 2 Myr ago (grey curve).

1 Geómagne´tisme et Paleómagne´tisme (UMR CNRS 7577), 2Geóchimie et Cosmochimie (UMR CNRS 7579), Institut de Physique du Globe de Paris, 4 Place Jussieu, 75252 Paris Cedex 05, France. 3Laboratoire des Sciences du Climat et de l’Environnement, Avenue de la Terrasse, 91190 Gif-sur-Yvette, France.

802

© 2005 Nature Publishing Group

LETTERS

NATURE|Vol 435|9 June 2005

Figure 2 | Composite Sint-2000 record depicting the evolution of the field intensity during the past 2 Myr. a, Curve obtained after stacking all individual records, shown with the 95% confidence interval in grey. The succession of the polarity intervals is depicted by the black (white) bars for normal (reverse) polarity at the top of the panel. b, Calibration of Sint-2000

correlations between cores was tested for the period 0.6–1.3 Myr ago, which incorporates the largest number of records. The stack (Fig. 1a) calculated after correlating the intensity minima corresponding to the reversals (Brunhes–Matuyama, onset and termination of the Jaramillo subchron, Cobb Mountain and Upper Olduvai) was actually very similar to the ‘optimal’ stack (Fig. 1b) obtained after improving the correlations by matching other minima of intensity without violating the original depth–time correspondence. We noticed also that there is no major difference between the curves tuned to different targets (Fig. 1d), and that the stacks involving all 15 records25 (Fig. 1c) or 10 cores only (Fig. 1a, b) do not significantly differ from each other. The oldest time period (2–1.5 Myr ago) is documented by 5 records only, but with relatively good geographical coverage. We verified that the stack calculated using this subset of 5 cores is identical to the one derived from all 10 cores over their common time interval, 0.6–1.3 Myr ago (Fig. 1d). We thus infer that it is representative of the field, but we are aware that additional records could reduce uncertainties for this period. In order to scale the 0.6–2-Myr-ago new stack with Sint-800, we assigned the same mean value to their common time interval, 0.6–0.8 Myr ago. The curve of the field intensity for the past 2 Myr shown in Fig. 2a is referred to as Sint-2000. As expected, larger dispersion is observed for the time interval 1.5–2 Myr ago because there are fewer records. Records of relative palaeointensity from sediments can be

using the absolute palaeointensity from volcanic records converted into virtual axial dipole moments (VADMs). The VADMs were averaged over successive time intervals (at least 0.1 Myr long). Error bars indicate the dispersion of the VADMs. c, Fluctuations of the dipole moment around its

Figure 3 | Time-averaged dipole moment calculated within sliding windows of different sizes, as a function of the number of reversals contained within each window. Window sizes are 40, 100, 200 and 300 kyr. The intervals immediately immediately surrounding the reversals have been removed from the calculation. The mean dipole moment decreases with the number of reversals. Because large windows can incorporate periods of low and high reversal frequency, the slope decreases with the size of the windows.


803

LETTERS


Figure 4 | Field intensity variations across the five reversals occurring during the past 2 Myr. In this figure we superimposed the changes in dipole moment during the 80 kyr and 20 kyr time intervals respectively preceeding and following each reversal. Note the 60–80-kyr-long decrease preceding the reversals, and the rapid recovery following the transitions.

calibrated to absolute determinations from volcanic records. However, this procedure is delicate, because volcanic records represent instantaneous measurements of the total field and their ages are associated with relatively large uncertainties. In order to remove the contribution of non dipole components and to reduce age uncertainties, we averaged the virtual axial dipole moments (VADMs) of the 2004 updated volcanic database27 over 0.1-Myrlong intervals. We then calculated the time-averaged VADM ((7.46 ^ 1.16) £ 1022 A m2) for the past 0.8 Myr and used this value for calibration. This interval was selected because all successive 0.1-Myr-long segments incorporate enough data points to calculate a reasonable estimate of the mean dipole moment. There is an overall satisfactory agreement between each successive time-averaged volcanic value and the calibrated record shown in Fig. 2b. A striking characteristic of Sint-2000 is the succession of periods with different mean values. During the Brunhes chron, the dipole oscillated around a value of (7.5 ^ 1.7) £ 1022 A m2, which is significantly larger than during the previous 400 kyr, (5.3 ^ 1.5) £ 1022 A m2. It is tempting to link the existence of a low (strong) dipole with the presence (absence) of reversals like during the Matuyama (Brunhes) or during the 0.5-Myr-long interval without reversals between the Gauss–Matuyama and the Cobb Mountain reversals. To provide a more quantitative picture of field strength as a function of reversal frequency, we calculated successive running averages of the field intensity using different rectangular windows (40, 100, 200, 300 kyr). The mean VADMs within all intervals were classified with respect to the number of reversals (N) that they contain (a 100-kyr-long interval embraces a maximum of two reversals). The mean of each category gives an estimate of the field strength for the intervals containing between 0 and 4 reversals. For each class of N reversals, the average field value and its standard deviation are reported in Fig. 3 as a function of N. The results show that the averaged field intensity linearly decreases with respect to the number of reversals. As expected, the slope decreases for increasing sizes of the windows as they involve longer time-averaged variations. The Cobb event was considered as a full polarity interval with two successive reversals separated by a 10-kyr-long interval of low intensity. If we were to see it as a single transition, the slope of the plot would be larger. We also removed the data contained within the 20-kyr-long intervals surrounding the transitions, and verified that they did not affect the calculations. From these results, we conclude that at least during this period the dipole field strength appears to be a dominant factor controlling the frequency of reversals. A similar but weaker dependence of average field intensity and length of polarity 804

intervals was previously reported28 in the study of an 11-Myr-long record of the Oligocene. We wondered whether the variability of the dipole (DVADM) is sensitive to its mean value. This parameter was approximated after smoothing Sint-2000 using a running average with a 40 kyr window size and subtracting each successive value from the corresponding initial data point. The plot in Fig. 2c shows that DVADM remains more or less constant. Thus minima of intensity are more easily reached when the averaged moment is weak than when it is strong. A direct consequence is that the ratio of the axial dipole field intensity to the non-axial dipole part is lower, which should generate more geomagnetic instabilities during weak field periods. One may then wonder whether this is consistent with the apparently frequent occurrence of excursions (roughly one every 100 kyr) that have been reported for the present polarity interval of rather strong dipole. In fact, a reduction by only half the present-day field29 is enough to generate anomalous (seen as excursional) directions covering one-fifth of the Earth’s surface. Because reversal positions were used as stratigraphic markers to correlate the records, timescale uncertainties are much smaller during these time intervals, which allows us to compare the pattern of the field variations immediately preceding and following each transition. In Fig. 4, we superimposed the field variations during these two phases for each polarity transition. The dominant feature is that each reversal was followed by a sudden and large field recovery. All curves show also that the dipole decayed before the transitions over a time interval of about 60–80 kyr. Because these features are present in sediments of various lithologies and physical properties, it is difficult to envisage that they could have been caused by postdepositional reorientation of magnetic grains30,31. Note that such processes could not account for the variations observed across the Cobb mountain event (1.19 Myr ago). Indeed in this case the restoration of the initial polarity after a rapid succession of two reversals is incompatible with the coexistence of directions with opposite polarities at the same stratigraphic levels. It is also significantthat similar asymmetry between the pre- and post-reversal stages has been reported in the most detailed volcanic records of reversals25. If confirmed, the respective durations of these two stages suggest that diffusive processes could dominate the pre-reversal episode while induction could drive the dipole recovery. Altogether, these results bring new constraints to models for the geodynamo, and emphasize the importance of palaeointensity studies for our understanding of the geomagnetic field. Received 23 July 2004; accepted 13 April 2005. 1.

2.

3. 4.

5.

6.

7. 8.

9.

Guyodo, Y. & Valet, J. P. Relative variations in geomagnetic intensity from sedimentary records: the past 200 thousand years. Earth Planet. Sci. Lett. 143, 23–-26 (1996). Laj, C., Kissel, C., Mazaud, M., Channell, J. E. T. & Beer, J. North Atlantic paleointensity stack since 75 ka (NAPIS 75) and the duration of the Laschamp event. Phil. Trans. R. Soc. Lond. A 358, 1009–-1025 (2000). Guyodo, Y. & Valet, J. P. Global changes in intensity of the Earth’s magnetic field during the past 800 kyr. Nature 399, 249–-252 (1999). Gee, J., Cande, S. C., Hildebrand, J. A., Donnelly, J. A. & Parker, R. L. Geomagnetic intensity variations over the past 780 kyr obtained from nearseafloor magnetic anomalies. Nature 408, 827–-832 (2000). Pouliquen, G., Gallet, Y., Dyment, J., Patriat, P. & Tamura, C. A geomagnetic record over the last 4 million years from deep-tow magnetic profiles across the Central Indian Ridge. J. Geophys. Res. 106, 10941–-10960 (2001). Frank, M. A 200 kyr record of cosmogenic radionuclide production rate and geomagnetic field intensity from 10Be in globally stacked deep-sea sediments. Phil. Trans. R. Soc. Lond. A 358, 1089–-1107 (2000). Baumgartner, S. et al. Geomagnetic modulation of the 36Cl flux in the Grip ice core, Greenland. Science 279, 1330–-1332 (1998). Valet, J.-P., Meynadier, L., Bassinot, F. C. & Garnier, F. Relative paleointensity across the last geomagnetic reversal from sediments of the Atlantic, Indian and Pacific oceans. Geophys. Res. Lett. 21, 485–-488 (1994). Dinares-Turell, J., Sagnotti, L. & Roberts, A. P. Relative geomagnetic paleointensity from the Jaramillo subchron to the Matuyama/Brunhes boundary as recorded in a Mediterranean piston core. Earth Planet. Sci. Lett. 194, 327–-341 (2002).


LETTERS


10. Laj, C. & Kissel, C. Geomagnetic field intensity and reversals for the last 1.8 Ma from a central equatorial Pacific core. Geophys. Res. Lett. 23, 3393–-3396 (1996). 11. Channell, J. E. T. & Kleiven, H. F. Geomagnetic paleointensity and astrochronological ages for the Matuyama-Brunhes boundary and the boundaries of the Jaramillo subchron: paleomagnetic and oxygen isotope records from ODP site 983. Phil. Trans. R. Soc. Lond. A 358, 1027–-1047 (2000). 12. Guyodo, Y., Richter, C. & Valet, J. P. Paleointensity record from Pleistocene sediments off the California Margin. J. Geophys. Res. 104, 22953–-22964 (1999). 13. Guyodo, Y., Acton, G. D., Brachfeld, S. & Channell, J. E. T. A sedimentary paleomagnetic record of the Matuyama chron from the western Antarctic margin. Earth Planet. Sci. Lett. 191, 61–-74 (2001). 14. Hayashida, A., Verosub, K. L., Heider, F. & Leonhardt, R. Magnetostratigraphy and relative paleointensity of late Neogene sediments at ODP leg 167 site 1010 off Baja California. Geophys. J. Int. 139, 829–-840 (1999). 15. Horng, C. S., Roberts, A. P. & Liang, W.-T. Astronomically tuned record of relative geomagnetic paleointensity from the western Philippine sea. J. Geophys. Res. 108, 2059, doi:10.1029/2001JB001698 (2003). 16. Kent, D. V. & Opdyke, N. D. Paleomagnetic field intensity recorded in a Brunhes epoch deep-sea sediment core. Nature 266, 156–-159 (1977). 17. Kok, Y. S. & Tauxe, L. A relative geomagnetic paleointensity stack from Ontong-Java plateau sediments for the Matuyama. J. Geophys. Res. 104, 25401–-25413 (1999). 18. Meynadier, L., Valet, J.-P., Bassinot, F. C., Shackleton, N. & Guyodo, Y. Asymmetrical saw-tooth pattern of the geomagnetic field intensity from equatorial sediments in the Pacific and the Indian oceans. Earth Planet. Sci. Lett. 126, 109–-127 (1994). 19. Sato, T., Kikuchi, H., Nakashizuka, M. & Okada, M. Quaternary geomagnetic field intensity: constant periodicity or variable period? Geophys. Res. Lett. 25, 2221–-2224 (1998). 20. Schneider, D. A., Kent, D. V. & Mello, G. A. A high-resolution marine sedimentary record of geomagnetic intensity during the Brunhes chron. Earth Planet. Sci. Lett. 111, 395–-405 (1992). 21. Valet, J.-P. & Meynadier, L. Geomagnetic field intensity and reversals during the past four million years. Nature 366, 234–-238 (1993).

22. Valet, J.-P. Time variations in geomagnetic intensity. Rev. Geophys. 41( 1), 1004 (2003). 23. Guyodo, Y. & Channell, J. E. T. Effects of variable sedimentation rates and age errors on the resolution of sedimentary paleointensity records. Geochem. Geophys. Geosyst. 10, doi:1029/2001GC000211 (2002). 24. McMillan, D. G., Constable, C. G. & Parker, R. L. Assessing the dipolar signal in stacked paleointensity records using a statistical error model and geodynamo simulations. Physics Earth Planet. Inter. 145, 37–-54 (2004). 25. Guyodo, Y. & Valet, J. P. A comparison of relative paleointensity records of the Matuyama chron for the period 0.75–-1.25 Ma. Phys. Earth. Planet. Inter. (in the press). 26. Guyodo, Y., Gaillot, P. & Channell, J. E. T. Wavelet analysis of relative geomagnetic paleointensity at ODP Site 983. Earth Planet. Sci. Lett. 184, 109–-123 (2000). 27. Perrin, M. & Schnepp, E. IAGA paleointensity database: distribution and quality of the dataset. Phys. Earth Planet. Inter. 147( 2–-3), 255–-267 (2004). 28. Constable, C. G., Tauxe, L. & Parker, R. L. Analysis of 11 Myr of geomagnetic intensity variation. J. Geophys. Res. 103, 17735–-17748 (1998). 29. Quidelleur, X., Gillot, P.-Y., Carlut, J. & Courtillot, V. The age and duration of the Matuyama–-Brunhes transition from new K–-Ar data from La Palma (Canary Islands) and revisited 40Ar/39Ar ages. Earth Planet. Sci. Lett. 168, 233–-242 (1999). 30. Meynadier, L. & Valet, J.-P. Post-depositional realignment of magnetic grains and asymmetrical saw-toothed pattern of magnetization intensity. Earth Planet. Sci. Lett. 140, 123–-132 (1996). 31. Mazaud, A. Saw-tooth variation in magnetic intensity profiles and delayed acquisition of magnetization in deep sea cores. Earth Planet. Sci. Lett. 139, 379–-386 (1996).

Acknowledgements This work was supported by the INSU-CNRS programme ‘Dynamique et Evolution de la Terre Interne’. Author Information Reprints and permissions information is available at npg.nature.com/reprintsandpermissions. The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to J.-P.V. ([email protected]).


805