Regionally Recurrent Paleomagnetic Transitional Fields and Mantle Processes

Transcription

1 Regionally Recurrent Paleomagnetic Transitional Fields and Mantle Processes Kenneth A. Hoffman Physics Department, Cal Poly State University, San Luis Obispo, California Brad S. Singer Department of Geology and Geophysics, University of Wisconsin-Madison, Wisconsin Whether Earth's lower mantle observably influences the geodynamo as it attempts to reverse polarity has been hotly debated for more than a decade. Although several paleomagnetic records of transitional events support this contention, analyses of worldwide Cenozoic reversal data, when considered en masse, appear inconclusive. With this controversy in mind we present transitional field data associated with a sequence of lavas on Tahiti-Nui, 40 Ar/ 39 Ar-dated at 578.6±8.8 ka, that record the Big Lost Event. The recorded field behavior is dominated by a cluster of virtual geomagnetic poles (VGPs) near the west coast of Australia, a common feature seen in previously published records of reversals and geomagnetic events obtained from lava sequences erupted at the Society Island hotspot over 2.3 Myr during the Plio-Pleistocene. This correspondence suggests that recurrent, or geographically preferred, VGP behavior is most apparent given regionally-partitioned paleomagnetic data. Synthetic removal of the axial dipole term from modern-day and historic field models produces a field configuration at the hotspot site compatible with these transitional paleodirectional findings. Various analyses of the core-surface modern-day, historic, and time-averaged paleomagnetic fields suggest the existence of a long-lived concentration of magnetic flux beneath Australasia. At those times when the axial dipole is weak or nonexistent, such a patch of mantleheld flux apparently can possess sufficient strength to regionally dominate the magnetic field structure. Lower-mantle seismic studies show the region beneath Australasia to be moderately anomalous, offering a possible clue toward an understanding of the physical conditions that may produce such a magnetic flux gathering. 1. INTRODUCTION The fact that Earth s magnetic field is capable of reversing its polarity and that transitional behavior can be recorded in rocks for later paleomagnetic investigation insures a unique source of data relevant to the problem of the geodynamo: observations of the field when the axial dipole is anomalously weak. Two particular questions that have been addressed by such paleomagnetic data are 1) whether or not there exists observable systematics to the process of field reversal [see Jacobs, 1994], and if so, 2) whether the lower mantle plays a significant role in influencing what is experienced at Earth s surface [see Merrill et al., 1996]. In the early 1990's scenarios involving mantle control were proposed, based on analyses of late-cenozoic transitional

2 field records. First, Clement [1991] and Laj et al. [1991] claimed that transitional VGP pathways followed preferred bands of longitude. Later, Hoffman [1992] argued for preferred locations where transitional VGPs cluster. In the years that followed a number of papers were published, some in opposition, others in support of geographically preferred VGP behaviors [see e.g. the review by Merrill and McFadden, 1999]. Of the supporting publications, the work of Constable [1992] was particularly intriguing as it appeared to tie elements of the modern-day field to the claim of preferred VGP bands. On the negative side of the issue Prévot and Camps [1993] tested the contention of preferential VGP behavior by amassing available transitional VGP data obtained from late-cenozoic lava sequences and found no hint of such behavior, whether in the form of banded or clustered VGPs. Love [1998] conducted a similar analysis and concluded otherwise, arguing that preferred VGP pathways were in fact a statistically significant feature of recorded transitional field behavior. These seemingly contradictory findings underscore the complexity of the problem, and suggest that the addition to the dataset of yet undiscovered transitional field records is unlikely to definitively solve it. Several of the most reliable and detailed paleomagnetic reversal records presently available contain one or more clusters of sequential transitional VGPs, many of which reside in either of the two claimed patches shown in Figure 1 (upper row left). For the case of the Brunhes-aged Blake Event recorded in Mediterranean sediments [Tric et al., 1991], the VGP is seen (Figure 1, upper row center) to first loiter within one patch and then move with apparent rapidity to the other before returning to normal polarity. Several other examples of clustered VGPs found in either the region of the Southwest Atlantic or Australasia are displayed in Figure 1, records obtained from various rocktypes and which span the Cenozoic. However, since these data account for only a fraction of the Cenozoic transitional field dataset, such a compilation in itself cannot be considered a robust confirmation of the existence of preferred quasistationary states of the reversing field. With the goal of shedding more light on the question of preferred transitional field configurations and, hence, the possible controlling presence of the lower-most mantle on the geodynamo, we now consider an approach in which paleomagnetic data is analyzed in a regional context. Specifically, we first revisit the case of the last reversal, the Matuyama-Brunhes polarity transition, for which by far the largest number of records are presently available. We then consider accounts of transitional field behavior obtained from lavas erupted at the Society Island hotspot over the past 3 Myr, and include in this dataset our newly-analyzed record of the Brunhes-aged Big Lost Event recorded in lavas from northern Tahiti-Nui. Later, we explore these 2

3 results alongside aspects of the modern-day, historic, and paleomagnetic time-averaged fields as well as seismic studies of the lower-most mantle. 2. THE MATUYAMA-BRUNHES REVERSAL Recordings of the last significant reversal of Earth s magnetic field, the Matuyama-Brunhes reverse-to-normal transition, have been obtained from various types of sedimentary materials as well as lavas. Hoffman [2000] analyzed records satisfying the set of reliability criteria imposed by Love and Mazaud [1997] and which these authors included in their select database. Love & Mazaud [1997] reported that the complete database demonstrated a statistically significant preference for transitional VGPs to lie within the two longitudinal bands claimed by Clement [1991] and Laj et al. [1991]. Following Hoffman s [2000] analysis, we separate those Matuyama-Brunhes data obtained from sites between latitudes 45 S and 45 N (region A) from those at higher latitudes (region B). Regardless of the overall complexity of transitional field behavior possessed by a given record, each of five examined region A recordings were found to contain a VGP cluster in Australasia (Plate 1a). In contrast, records included in region B contain significant VGP behavior in the South Atlantic. Most striking in this regard is the complex transitional field behavior recorded in three North Atlantic marine sediment cores at Hole 984 (Plate 1b) reported by Channell and Lehman [1997] at the time of the database publication by Love and Mazaud [1997]. Possessing remarkable similarity, each of these records display a back and forth diagonal transit of the VGP between the South Atlantic and Central Asia. Plate 1b shows the 3-record composite VGP geographic histogram for that part of the recorded reversal behavior. The regional dissimilarity in cluster location between the two Matuyama-Brunhes sub-datasets suggests, first, that during much of the last reversal the transitioning field was non-dipolar. More specifically, the correspondence of VGP cluster location associated with records from sites within region A sites that encompass much of the globe suggests that much of the power in the transitioning field resided in non-dipole terms of low-order. If so, this finding could imply that the spatial nature of the primary sources responsible for such apparently stationary transitional fields may be ultimately decipherable. 3. TRANSITIONAL EVENTS RECORDED AT THE SOCIETY ISLAND HOTSPOT We now turn to the recordings at a particular region A site, the Society Island hotspot. There presently exist four reasonably detailed transitional field recordings (each containing a minimum of 5 transitional VGPs) obtained 3

4 from lavas erupted at the site of the hotspot. Three of these come from a sequence of four Pleistocene reversal records obtained from Tahitian lavas exposed along the Punaruu Valley reported by Chauvin et al. [1990]: the Matuyama- Brunhes reversal (see Plate 1a), the upper and lower boundaries of the Jaramillo subchron, and what was assumed at the time of its publication to be the Cobb Mountain event. Each record was precisely dated by 40 Ar/ 39 Ar age determinations [see Singer et al., 1999]. Of these records only the lower Jaramillo contains too few transitional VGPs (3 in total) to be considered further. (The transitional record then thought to be a recording of the Cobb Mountain event was later found by Singer et al. [1999] to be some 100 kyr younger. Singer et al. [1999] renamed this recorded event the Punaruu after the site at which it was first discovered.) Each of the three remaining records contain a cluster of transitional VGPs in essentially the same location off the southwest coast of Australia [Chauvin et al., 1990]. In addition, Roperch and Duncan [1990] reported a 2.9 Ma transitional event obtained from lavas exposed on Huahine, another island in the Society Island hotspot chain. These transitionally magnetized Pliocene flows were found to be dominated by two dense concentrations of VGPs off the northwest coast of Australia. Although these four Society Island hotspot records possess varying degrees of complexity, all contain a number of sequential VGPs found near western Australia not unlike the composite Matuyama-Brunhes field behavior displayed in Plate 1a. Of course, the possibility that one or more of these clusters represents a period of rapid-fire lava eruptions cannot be discounted [e.g. Prévot and Camps, 1993; Riisager et al., 2003]. However, a number of examples of rather dense VGP clustering (see e.g. Figures 1 and 2a) is associated with sedimentary material. Hence, it is likely that such features often correspond to quasi-invariant states of the transitioning field spanning, perhaps, considerable time. 4. A NEW RECORD OF THE BIG LOST EVENT FROM TAHITIAN LAVAS 4.1 The Paleomagnetic Results We attempted to expand the Tahitian Matuyama- Brunhes dataset by sampling near a flow previously K-Ar dated by Le Roy [1994] at about 794 ka at the end of the Chemin des Milles Sources (CMS) in the northern part of the island some 14 km northeast from the Punaruu Valley exposure. Some 27 sites were sampled along the accessible part of this continuous spring-fed gully section. The number of cores drilled at a given site was dependent on the amount of exposure. Demagnetization was accomplished by both thermal and alternating field techniques. 4

5 Principal component analysis was employed to identify the primary thermoremanent magnetization component. All but two of the sites rendered a reliable mean paleodirection. It was sometimes difficult to determine in the field what constituted a distinct flow unit. However, in these cases the section contained sets of successive sites that possessed nearly identical paleodirections. These results were combined and the 25 successful site-mean analyses were reduced to 15 distinctive directional groups (Table 1). Plate 2 shows the path of the VGP, along with 95% confidence ovals, associated with the recorded paleodirections. As can be seen, the attainable record starts with a VGP at high northern latitudes in Siberia (frame a), transits across the equator to a location off the west coast of Australia, lingers there (frame b) for an unknown amount of time associated with the eruption of several lavas, returns along a longitudinally-confined path (frame c) to a location near that at the onset, and then rebounds (frame d) to the southwest coast of Australia near the cluster of VGPs seen in frame b. 4.2 The Geochronologic Results Given both the earlier-determined K-Ar date by Le Roy [1994] for a nearby flow at the end of the CMS as well as the fact that the grouping of transitional VGPs off the west coast of Australia is virtually identical to the findings for the Matuyama-Brunhes reversal recorded in the lavas exposed in the Punaruu Valley [Chauvin et al., 1990], we were convinced that the CMS lavas also recorded the last reversal of Earth s magnetic field. However, nine 40 Ar/ 39 Ar incremental-heating experiments on groundmass separated from 3 lavas within the transitionally-magnetized flow sequence yielded isochron ages with 2σ uncertainties of 576.5±11.0 ka, 578.8±22.0 ka, and 587.7±22.1 ka, respectively (Figure 2). The 40 Ar/ 39 Ar incremental heating experiments were carried out on ca. 100 mg aliquots of purified basaltic groundmass using a resistance furnace at the University of Wisconsin-Madison following analytical and data reduction procedures [Singer et al., 2002]. The neutron fluence monitor was either Ma Taylor Creek rhyolite sanidine or Ma Alder Creek rhyolite sanidine that have been intercalibrated against one another [Renne et al., 1998]. Because the resulting isochron ages are compared to 40 Ar/ 39 Ar ages from other lavas obtained using the same methods and standards [Singer et al., 2002], the 2σ uncertainties reported with the ages include uncertainties in the analytical procedure, and where appropriate, the standard intercalibration. From samples CMS-203, -100, and -115, gas increments that gave concordant plateau ages comprise more than 98, 83, and 99% of the gas released. The age spectra imply that sample CMS-110 lost a small frac- 5

6 tion of radiogenic argon, possibly due to petrographically undetectable post-eruption alteration. Notwithstanding, the isochron age of CMS-110 overlaps those of the over- and underlying lava flows. The combination of concordant analyses from two to five experiments on each lava flow improved regression of the isochrons such that ages have uncertainties of 1.3 to 2.4%. It is standard to determine the best age from a group of determinations with variable analytical uncertainty by calculating the inverse-variance weighted mean age of the population. Thus, samples yielding the highest quantities of radiogenic argon and the most precise ages are proportionally more influential than those which give much more uncertain results. The weighted mean isochron age of the three lavas, 578.8±8.8 ka, gives our best estimate of time elapsed since the eruption of this flow sequence. This age is about 200 kyr younger than the widely accepted ka age for the Matuyama-Brunhes reversal [Singer and Pringle, 1996; Tauxe et al., 1996], but is identical to the 580.2±7.8 ka age [Singer et al., 2002] recently determined for the Big Lost Event [Champion et al., 1988] using the same 40 Ar/ 39 Ar procedures on a lava flow sequence at La Palma, Canary Islands. 4.3 Discussion of Results Given the 40 Ar/ 39 Ar age determinations, Australianclustered VGPs are associated with each of five transitional field records obtained from Society Island hotspot lavas spanning 2.3 Myr from 2.9 Ma to about 0.6 Ma. If these correspondences are not simply fortuitous, they suggest the presence of similar structures of geomagnetic flux at least as seen from the site of the Society Island hotspot at times when the axial dipole field has been weak and/or when the geodynamo has been in a transitional state. Sources of such an enduring pattern of magnetic flux may reside either in close proximity to the hotspot volcanoes themselves or in the outer core. More specifically, the two most likely possible explanations are: 1) the existence of an anomalous crustal field source which locally dominates during times when the core field is very weak, and 2) longstanding control by the lowermost mantle of magnetic flux emanating from the outer core [e.g. Bloxham and Gubbins, 1987]. 5. THE MODERN-DAY FIELD AT THE SOCIETY ISLAND HOTSPOT If in fact a long-held pattern of flux emanating from the core-mantle boundary is responsible for the recurring transitional field behavior at the Society Island hotspot, one may attempt to find evidence of its presence in an analysis of the modern-day geomagnetic field. Since polarity reversal requires a change in sign of Earth's axial dipole, we 6

7 explore the above possibility by initially removing the axial dipole term, g 1 0, from the IGRF reference field for the years 1900, 1950 and 2000 [data listed in and proceed to determine the location of the respective VGPs at the Society Island hotspot if only this non-axial-dipole (NAD) field was present. Note that since the IGRF models have harmonic resolution to order and degree 10, they are not capable of resolving local crustal field sources. Hence, a finding in which modern-day NAD-field VGPs are compatible with the transitional paleomagnetic clusters from the hotspot site, would be supportive of far deeper, longstanding field sources that still exist. Since the dynamo process is "blind" to the sign of magnetic flux [see e.g. Merrill et al., 1996], it is equally valid to analyze either north or south VGPs which are antipodal. Such a procedure has been referred to as the "geomagnetic convention" [see Prévot and Camps, 1993]. For the Society Island hotspot site the north NAD-field VGPs associated with the twentieth century geomagnetic field are found in the Caribbean. The south NAD-field VGPs, however, lie near the coast of western Australia in close proximity to the clustered VGPs from the, now, 5 available transitionally-magnetized lava records obtained from the Society Islands (Plate 3). 6. DISCUSSION Although global databases of late-cenozoic transitional paleodirections have not found the Australasian VGP feature to be clearly resolvable [Prévot and Camps, 1993; Love, 1998], the correspondences seen in Plate 3 suggest that the repeatable nature of the feature may become quite evident when regional data alone are considered. This apparent inconsistency is compatible with either or both of the contentions that 1) the reversal process is, typically, complex [Coe and Glen, this volume] and 2) transitional fields are often non-dipolar [e.g. Prévot and Camps, 1993; Love, 1998]. Such a situation could arise if midway in the process at a time when the axial dipole field has nearly vanished a small number of long-lived, localized flux concentrations at the core surface largely define the structure of the residual field. Recorded field behavior, then, would be strongly site-dependent: those sites nearest to the coordinates of a significant flux concentration would be most affected by it, the influence of the feature greatly diminishing with distance from it on Earth s surface. In this way, regional fields above mantle-held flux bundles may take on more "dipolar characteristics." We now explore this idea: the three uppermost plots in Figure 3 show the absolute value of the radial component at Earth's surface of the 1900, 1950 and 2000 IGRF NADfields. Each plot is normalized to the strongest value present. The plots indicate two significant features, one in the 7

8 South Atlantic, the other about Australasia. Although the chronology shows both vertical NAD-field features to be growing in strength, over the last century the intensity of the Australasia has been increasing in relative strength to the South Atlantic feature. Figure 3 (middle three plots) shows the VGPs associated with the geomagnetic convention NAD-fields for the 1900, 1950 and 2000 IGRF's, respectively, for a 5 x 5 grid of 247 sites from within the indicated region in and surrounding Australasia (latitude range 60 S 0 ; longitude range 90 E 180 E). As can be seen, regardless of the relative significance at Earth s surface of the Australian flux patch from 1900 to 2000, the VGPs for sites in this region are found in a rather confined north-south sweep through western Australia, displaying negligible westward drift. Although the equatorial dipole has its pole at the equator in Australasia, the full explanation for the placement of the NAD-field VGPs appears to involve a more complex field source: Figure 3 (bottom plot) shows the VGPs for the same region of sites for the IGRF 2000 non-dipole field. As is seen, even with the removal of all dipole terms from the NAD-field, the VGPs still show a strong preference for the Australasian region. This positive correlation supports the contention that the principal field source causing this confinement of VGPs physically resides in the core beneath Australasia and has a significant influence on nondipole as well as dipole spherical harmonic terms. The observation of the low-to-mid latitude Austalasian NAD-field patch at Earth s surface is consistent with features of the core surface field modeled for the modern-day [Jackson et al., 2000; Johnson et al., 2003], historic [Bloxham and Jackson, 1992] and time-averaged paleomagnetic [Kelly and, Gubbins, 1997] fields (see e.g. Figure 4). Evidence for the existence of physically anomalous conditions beneath Australasia come from deep-earth seismic investigations: Kendall and Shearer [1994] report that reflections by S-waves indicate that the D -layer beneath Australasia is highly variable in thickness. In addition, anomalously fast seismic wave velocities deep within the mantle beneath Australasia have been documented, first, through normal mode analysis [Dziewonski and Woodhouse, 1987], and, later, through shear wave tomography [e.g. Li and Romanowicz, 1996]. Thus, findings from 1) transitionally magnetized lavas erupted at the Society Island hotspot since the Pliocene, 2) analyses of the modern-day, historic geomagnetic and time-averaged paleomagnetic fields, and 3) deep-mantle seismic tomographic studies, provide support for a model in which the observation of recurrent, regionally-controlled transitional field states are related to heterogeneity of the lowermost mantle and its effect on dynamo flux. Hypothetically, in the limiting case in which only the flux concentration beneath Australasia is present during the 8

9 reversal process, depending on sign, either north VGPs or south VGPs associated with sites over much of the globe would congregate in a region directly above it. As the significance of other flux concentrations may increase, however, the region that would be most influenced by the Australasian patch would clearly shrink. The Brunhes-aged event recorded in Mediterranean sediments (Figure 1: upper row center) may be an example whereby the South Atlantic flux feature first dominated the transitional field, only to be thoroughly overtaken by the Australasian feature. Although groupings of VGPs located near western Australia are seen to recur from one reversal to the next at the site of the Society Island hotspot, transitional paths of the VGP are often observed to be far more complex [see Coe and Glen, this volume]. Hence, it is likely that the degree of dominance by mantle-controlled flux beneath mid-tolow latitude sites like Australasia, largely varies over the course of a complete reversal; perhaps, only on occasion during the dynamo process may sustained field behavior be recorded. Further, even though an extensive region above such a flux concentration may display field characteristics that appear to be dipole-dominated (see Figure 3), the global field may be far more complex (see Plate 1). In addition, recurrent clusters of VGPs have been found in successive Pliocene reversal records from Oahu, Hawaii [Herrero-Bervera and Coe, 1999]. Interestingly, these groupings lie within Africa, suggesting that other mantlecontrolled localities exist at the core-surface that, on occasion, may regionally dominate transitional fields. Finally, Plate 4 shows the pattern of sites (shown in red) that, for the year 2000 NAD-field, are associated with south VGPs within the indicated Australasian cluster patch. The extent of coverage seen clearly indicates the considerable influence this flux feature has over the global NAD-field. At the same time, the Southern Hemisphere high latitude sites (shown in blue) are associated with north VGPs within the indicated South Atlantic cluster patch. Given a continuing descent in strength of the modern-day axial dipole, these findings suggest that field changes associated with the onset of the next geomagnetic reversal may be more predictable than previously thought. Acknowledgements. KAH wishes to acknowledge the several undergraduate physics majors D. Soukup, M. Jock, A. Battle, B. Bose and A. Demogines who assisted in the field and/or in the Cal Poly Paleomagnetism Laboratory. We also thank M. Relle for her 40 Ar/ 39 Ar age determinations conducted at the University of Wisconsin. This study was supported through National Science Foundation grants EAR , EAR , EAR , EAR , and EAR REFERENCES 9

10 Bloxham, J., and D. Gubbins, Thermal core-mantle interactions, Nature, 325, , Bloxham, J., and A. Jackson, Time-dependent mapping of the magnetic field at the core-mantle boundary, J. Geophys. Res., 97, , Bogue S.W., and R.S Coe, Successive paleomagnetic reversal records from Kauai, Nature, 295, , Champion, D.E. and M.A. Lanphere, Evidence for a new geomagnetic reversal from lava flows in Idaho: discussion of short polarity reversals in the Brunhes and Late Matuyama polarity chrons, J. Geophys. Res., 93, 11,667-11,680, Channell, J.E.T., and B. Lehman, The last two geomagnetic polarity reversals recorded in high-deposition-rate sediment drifts, Nature, 389, , Chauvin, A., P. Roperch, and R.A. Duncan, Records of geomagnetic reversals from volcanic islands of French Polynesia, 2, paleomagnetic study of a flow sequence ( Ma) from the island of Tahiti and discussion of reversal models, J. Geophys. Res., 95, , Clement, B.M., Geographical distribution of transitional VGPs: evidence for non-zonal equatorial symmetry during the Matuyama-Brunhes geomagnetic reversal, Earth Planet. Sci. Lett., 104, 48-58, Clement, B.M., L. Sierra, E. Smith, and P. Rodda.. A record of the upper Cochiti polarity transition from the southern hemisphere Fiji). Surv. Geophys., 17: , Coe, R.S., and J.M.G. Glen, The complexity of reversals, this volume. Constable, C., Link between geomagnetic reversal paths and secular variation of the field over the last 5 Myr., Nature, 358, , Dziewonski, A., and J. Woodhouse, Global images of the Earth s interior, Science, 236,37-48, Glen, J. M. G., R. S. Coe, and J. C. Liddicoat, A detailed record of paleomagnetic field change from Searles Lake, California: 2. The Gauss/Matuyama polarity reversal, J. Geophys. Res., 104, 12,883-12,894, Herrero-Bervera, E., and R.S. Coe, Transitional field behavior during the Gilbert-Gauss and Lower Mammoth reversals recorded in lavas from the Waianae volcano, O ahu, Hawaii, J. Geophys. Res., 104, 29,157-29,173, Hoffman, K.A., Dipolar reversal states of the geomagnetic field and core-mantle dynamics, Nature, 359, , Hoffman, K.A., Temporal aspects of the last reversal of Earth s magnetic field, Phil. Trans. R. Soc. Lond. A, 358, , Jackson, A., A.R.T. Jonkers, and M.R. Walker, Four centuries of geomagnetic secular variation from historic records, Phil. Trans. R. Soc. Lond. A, 358, , Jacobs, J. A., Reversals of the Earth s Magnetic Field, 2 nd ed. (Cambridge University Press, New York), pp. 346, Johnson, C.L., C.G. Constable, and L. Tauxe, Mapping Long- Term Changes in Earth's Magnetic Field, Science, 300, , Kelly, P., and D. Gubbins, The geomagnetic field over the past 5 Myr, Geophys. J. Int., 128, ,

11 Kendall, J-M., and P. M. Shearer,, Lateral variations in D" thickness from long-period shear wave data, J. Geophy. Res., 99, , Laj, C., A. Mazaud, R. Weeks, M. Fuller, and E. Herrero- Bervera, Geomagnetic reversal paths, Nature 351, 447, Leonhardt, R., J. Matzka, F. Hufenbecher, H. C. Soffel, and F. Heider, A reversal of the Earth's magnetic field recorded in mid-miocene lava flows of Gran Canaria: Paleodirections, J. Geophys. Res., 107(B1), /2001JB000322, Le Roy, I., Evolution des Volcans en Systeme de Point Chaud: Ile de Tahiti, Archipel de la Societe (Polynesie Francaise), thesis, l'universite Paris XI Orsay Li, X.D., and B. Romanowicz, Global mantle shear velocity model developed using nonlinear asymptotic coupling theory, J. Geophys. Res., 101, 22,245-22,273, Love, J.J., Paleomagnetic volcanic data and geometric regularity of reversals and excursions, J. Geophys. Res., 103, 12,435-12,452, Love, J.J. and A., Mazaud, A database for the Matuyama-Brunhes magnetic reversal, Phys. Earth Planet. Int., 103, , Merrill, R.T., M.W.,McElhinny, and P.L McFadden, The Magnetic Field of the Earth: Paleomagnetism, the Core and the Deep Mantle (Academic Press, San Diego) pp. 531, Merrill R.T., and P.L. McFadden, Geomagnetic polarity transitions. Rev. Geophys., 37, , Prévot, M. and P. Camps, Absence of preferred longitudinal sectors for poles from volcanic records of geomagnetic reversals, Nature, 366, 53-57, Renne, P. R., C. C. Swisher, A. L. Deino, D. B. Karner, T. Owens, and D. J. DePaolo, Intercalibration of standards, absolute ages and uncertainties in 40 Ar/ 39 Ar dating, Chem. Geol., 145, , Riisager, J., P. Riisager, and A.K. Pedersen, The C27n-C26r geomagnetic polarity reversal recorded in the west Greenland flood basalt province: How complex is the transitional field?, J. Geophys. Res. 108, NO. B3, 2155, doi: /2002jb002124, Roperch, P., and R.A. Duncan, Records of geomagnetic reversals from volcanic islands of French Polynesia 1. Paleomagnetic study of a polarity transition in a lava sequence from the island of Huahine. J. Geophys. Res., 95, , Shaw, J., Strong geomagnetic fields during a single Icelandic polarity transition. Geophys. J.R. Astron. Soc., 40, , Singer, B.S., K.A. Hoffman, A. Chauvin, R.S. Coe, and M.S. Pringle, Dating transitionally magnetized lavas of the Matuyama-Chron: Toward a new timescale of reversals and events, J. Geophys. Res., 104, , Singer, B.S., and M.S. Pringle, The age and duration of the Matuyama-Brunhes geomagnetic polarity reversal from 40 Ar/ 39 Ar incremental heating analyses of lavas, Earth Planet. Sci. Lett., 139, 47-61, Singer, B.S., M.K. Relle, K.A. Hoffman, A. Battle, C. Laj, H. Guillou, and J.C. Carracedo, Ar/Ar ages of transitionally magnetized lavas on La Palma, Canary Islands, and the Geomagnetic Instability Timescale, J. Geophys. Res., 107 (B11), 2307 doi: /2001jb001613,

12 Tauxe, L., T. Herbert, N.J. Shackleton, and Y.S Kok,, Astronomical calibration of the Matuyama-Brunhes boundary: consequences for magnetic remanence acquisition in marine carbonates and the Asian loess sequences, Earth Planet. Sci. Lett., 140, , Tric, E., C. Laj, J.P. Valet, P. Tucholka, M. Paterne, and F. Guichard,. The Blake geomagnetic event: transition geometry, dynamical characteristics and geomagnetic significance, Earth Planet. Sci. Lett., 102, 1-13, Valet, J.-P., and C. Laj, Invariant and changing transitional field configurations in a sequence of geomagnetic reversals, Nature, 311, , van Hoof, A.A.M., van Os, B.J.H. and C.G. Langereis, The upper and lower Nunivak sedimentary geomagnetic transitional records from Southern Sicily. Phys. Earth Planet. Inter., 77, , Kenneth A. Hoffman, Physics Department, Cal Poly State University, San Luis Obispo, CA USA Brad S. Singer, Department of Geology and Geophysics, University of Wisconsin-Madison, Madison, WI USA 12

13 Table 1. Paleodirectional data from Chemin des Milles Sources lavas, Tahiti-Nui Site # Group a N b INC DEC α 95 k Frame c VGP Long VGP Lat 195/ a / a/b / b / b /204/ b /101/ b / b / b/c / c / c / c / c / c / c/d / d a Directional group number b Number of specimens c Plot location in Plate 2 13

14 FIGURE AND PLATE CAPTIONS Figure 1. Examples of paleomagnetic transition records spanning the Cenozoic that contain a clustering of sequential virtual geomagnetic poles (VGPs) in the region of the South Atlantic and/or Australasia. Of the nine records shown, five and four were obtained from sediments and lavas, respectively. The claimed VGP preferred bands [Laj et al., 1991] and VGP cluster patches [Hoffman, 1992] are shown in the upper left plot. Plate 1. VGP systematics observed in records of the Matuyama- Brunhes. (a) Records containing a VGP cluster in the vicinity of Australia from the select database of Love and Mazaud [1997] associated with sites between 45 N and 45 S latitude [after Hoffman, 2000]. The corresponding sites and rocktypes are indicated. (b) Composite geographic histogram of VGPs (method as in Hoffman [2000]) contained in three parallel records obtained from Hole 984 in the North Atlantic [Channell and Lehman, 1997] during back and forth polar movement between the South Atlantic and Central Asia. Figure 2. Apparent age spectra (at center) and inverse isochron diagrams (at right) for transitionally magnetized basaltic lava flow sites CMS-203, CMS-110, and CMS-115. The isochrons combine 12, 27, and 10 analyses, respectively, from the age plateaux, and give the preferred age of each lava flow. Virtual geomagnetic pole latitudes of lavas comprising the stratigraphic sequence are shown (at left). Plate 2. Path of the VGP (flow mean poles along with α 95 ovals of confidence) as recorded in the lava sequence at the end of the Chemin des Milles Sources. The transitional field behavior begins in frame a with a Siberian VGP and ends in frame d with a west-australian VGP. [Note: Although the recorded field behaviors of the lowermost and uppermost flows (frames a and d, respectively) appear to be quite similar, we found no evidence of faulting that could have caused flow replication within the gully section.] Figure 3. (top) Absolute value of the vertical component of the NAD-field for the 1900, 1950 and 2000 IGRF, each normalized to the strongest intensity present. Regions in white have % of the strongest value. Note: the two most intense regions have opposite sign. (middle) Geomagnetic convention VGPs associated with a grid of 247 sites within Australasia (shaded region), each corresponding to the NAD-field shown immediately above. (bottom) VGPs associated with the same Australasian sites corresponding to the IGRF 2000 non-dipole field. Plate 3. Composite plot of all Australasian VGP clusters from available Society Island hotspot lava records and the south VGP (geomagnetic convention) for the non-axial dipole (NAD) part of the IGRF for years 1900, 1950, and 2000, as indicated. 14

15 Figure 4. The radial component of the geomagnetic field at the surface of the core (a) for the year 1990 [from Jackson et al., 2000], and (b) for the time-averaged paleomagnetic field over the last 5 Myr [from Kelly and Gubbins, 1997]. In both figures increased field intensity is displayed by an increased degree of darkness. Plate 4. Those sites about the globe that, given the modern-day NAD-field (year 2000 IGRF), would be associated with magnetic directions having a south VGP in Australasia (region indicated in red) or a north VGP in the South Atlantic (region indicated in blue). The site of the Society Island hotspot is also shown. Figure 1. Examples of paleomagnetic transition records spanning the Cenozoic that contain a clustering of sequential virtual geomagnetic poles (VGPs) in the region of the South Atlantic and/or Australasia. Of the nine records shown, five and four were obtained from sediments and lavas, respectively. The claimed VGP preferred bands [Laj et al., 1991] and VGP cluster patches [Hoffman, 1992] are shown in the upper left plot. Plate 1. VGP systematics observed in records of the Matuyama-Brunhes. (a) Records containing a VGP cluster in the vicinity of Australia from the select database of Love and Mazaud [1997] associated with sites between 45 N and 45 S latitude [after Hoffman, 2000]. The corresponding sites and rocktypes are indicated. (b) Composite geographic histogram of VGPs (method as in Hoffman [2000]) contained in three parallel records obtained from Hole 984 in the North Atlantic [Channell and Lehman, 1997] during back and forth polar movement between the South Atlantic and Central Asia. Figure 2. Apparent age spectra (at center) and inverse isochron diagrams (at right) for transitionally magnetized basaltic lava flow sites CMS-203, CMS-110, and CMS-115. The isochrons combine 12, 27, and 10 analyses, respectively, from the age plateaux, and give the preferred age of each lava flow. Virtual geomagnetic pole latitudes of lavas comprising the stratigraphic sequence are shown (at left). Plate 2. Path of the VGP (flow mean poles along with α 95 ovals of confidence) as recorded in the lava sequence at the end of the Chemin des Milles Sources. The transitional field behavior begins in frame a with a Siberian VGP and ends in frame d with a west-australian VGP. [Note: Although the recorded field behaviors of the lowermost and uppermost flows (frames a and d, respectively) appear to be quite similar, we found no evidence of faulting that could have caused flow replication within the gully section.] Figure 3. (top) Absolute value of the vertical component of the NAD-field for the 1900, 1950 and 2000 IGRF, each normalized to the strongest intensity present. Regions in white have % of the strongest value. Note: the two most intense regions have opposite sign. (middle) Geomagnetic convention VGPs associated with a grid of 247 sites within Australasia (shaded region), each corresponding to the NAD-field shown immediately above. (bottom) VGPs associated with the same Australasian sites corresponding to the IGRF 2000 non-dipole field. Plate 3. Composite plot of all Australasian VGP clusters from available Society Island hotspot lava records and the south VGP (geomagnetic convention) for the non-axial dipole (NAD) part of the IGRF for years 1900, 1950, and 2000, as indicated. Figure 4. The radial component of the geomagnetic field at the surface of the core (a) for the year 1990 [from Jackson et al., 2000], and (b) for the time-averaged paleomagnetic field over the last 5 Myr [from Kelly and Gubbins, 1997]. In both figures increased field intensity is displayed by an increased degree of darkness. Plate 4. Those sites about the globe that, given the modern-day NAD-field (year 2000 IGRF), would be associated with magnetic directions having a south VGP in Australasia (region indicated in red) or a north VGP in the South Atlantic (region indicated in blue). The site of the Society Island hotspot is also shown. 15

16 16

17 17

18

19 Site # / / / / / /101/ /204/ / d c b a Apparent age ka Apparent age ka Apparent age ka CMS 115 groundmass weighted mean plateau ± 13.9 ka 2 separate experiments CMS 110 groundmass weighted mean plateau ± 11.1 ka 5 separate experiments CMS 203 groundmass weighted mean plateau ± 7.3 ka 2 separate experiments Ar/ Ar Ar/ Ar Ar/ Ar air air air ± 22.1 ka Ar/ Ar i = ± 2.7 sums/(n-2) = 1.55 n= ± 22.0 ka Ar/ Ar i = ± 1.9 sums/(n-2) = 1.48 n= ± 11.0 ka Ar/ Ar i = ± 2.5 sums/(n-2) = 0.21 n= VGP latitude Cumulative % Ar released Ar/ Ar

20

21

22

23

24

25