Ooo

Fig. 4.14. Examples of some models for the transition field for a change from reverse to normal polarity, (a) Decrease in dipole moment which remains axisymmetric during the process, (b) Rotation of the main dipole without changes in moment, (c) A quadrupole transition model in which the polarity change is initiated in the southern hemisphere of the core, (d) An octupole transition model in which the polarity change is initiated in the low latitude zone of the core. From Merrill and McElhinny (1983), compiled from Fuller et al. (1979).

represented a conceptual leap and ushered in the era of "modern" polarity transition studies. See Fig. 4.14.

(iv) The transitional field is nondipolar but the transitional VGPs are longitudinally confined and are biased to fall within two (preferred) antipodal longitude bands of width about 60° (Clement, 1991). This model was based on sedimentary data.

(v) The transitional field is predominantly dipolar (although this aspect was later dropped) and the VGPs are not only longitudinally confined and biased toward two preferred bands but also, importantly, the same two bands are preferred for each of the most recent reversals (Laj et al., 1991, Trie et al., 1991). This hypothesis is particularly exciting for, if true, it may imply mantle control of processes in the geodynamo, at least during transitions. The preferred paths lie near the Americas or near an antipodal path (Gubbins and Coe, 1993) through western Australia and Asia. This model was also based on sedimentary data.

Fig. 4.14. Examples of some models for the transition field for a change from reverse to normal polarity, (a) Decrease in dipole moment which remains axisymmetric during the process, (b) Rotation of the main dipole without changes in moment, (c) A quadrupole transition model in which the polarity change is initiated in the southern hemisphere of the core, (d) An octupole transition model in which the polarity change is initiated in the low latitude zone of the core. From Merrill and McElhinny (1983), compiled from Fuller et al. (1979).

(vi) The transitional VGPs from lava flows clump in two patches, one on the American path above, lying between central South America and Antarctica, with the other cluster falling in western Australia and the eastern Indian Ocean (Hoffman, 1991a). The same clusters are occupied by VGPs for different reversals.

(vii) There is no consistent pattern for transitional VGPs from different reversals and the VGPs within a reversal are not longitudinally confined (Prévôt and Camps, 1993). This model emerged from analyses of 362 transitional VGPs from 121 volcanic units of Miocene age or younger and is an extension of an earlier suggestion by Valet et al. (1992).

If the transitional field were strongly dipolar then VGPs for the same time but for different sites around the globe would be consistent with each other (just as they are today). The transitional VGP paths would then be consistent for observation sites around the globe. Conversely, if different sites give widely divergent VGP positions for the same time then this is strong evidence that the field is not dominantly dipolar. Oddly enough though, consistency in transitional VGP paths from sites around the globe does not necessarily mean that the field was strongly dipolar (e.g., Merrill and McFadden, 1999). This exposes interpretation of transitional VGPs to several major difficulties, as is evidenced by the number of phenomenological models suggested and by their diversity (almost to the point of total contradiction).

The model probably most discussed today is model v, based on sedimentary data. In this model it is suggested that: transitional VGPs from a given site are confined in longitude; that for a given reversal there is a bias for the VGPs to fall into one of two preferred antipodal bands; and that the same bands are preferred for different reversals. Full testing of this model demands that each of its three components be properly tested. To test the hypothesis that the preferred bands are the same for different reversals it must first be confirmed that individual reversals do indeed exhibit a bias toward preferred bands for their VGPs, which itself requires confirmation of longitude confinement for individual VGP paths. Unfortunately, the paucity of data has meant that it has not been possible to separate out and resolve these individual questions. Consequently, investigators have tended to lump together all of the available data and then seek an overall clustering of some parameter (such as the longitude of the transitional VGP equator crossing). Thus, it remains difficult to interpret the results with any genuine clarity.

A major problem in assessing the reversal models is that the polarity transition records obtained from volcanics and from sediments appear to show different characteristics, as is illustrated in Fig. 4.15. The plot of 362 transitional VGPs from 121 volcanic records of reversals less than 16 Myr in age (Fig. 4.15a) does not exhibit any obvious clustering or preferred longitude sectors (Prévôt and Camps, 1993). The record for reversals less than 12 Myr in age recorded in

Transitional VGP equator crossings

Fig. 4.15. Comparison of the records of transitional VGPs observed in volcanics and sediments, (a) 362 transitional VGPs from 121 volcanic records of reversals less than 16 Myr old. From Prévôt and Camps (1993), reproduced with permission from Nature, (b) Plot of equator crossings for the available sedimentary VGP transition paths. From McFadden and Merrill (1995).

Transitional VGP equator crossings

Fig. 4.15. Comparison of the records of transitional VGPs observed in volcanics and sediments, (a) 362 transitional VGPs from 121 volcanic records of reversals less than 16 Myr old. From Prévôt and Camps (1993), reproduced with permission from Nature, (b) Plot of equator crossings for the available sedimentary VGP transition paths. From McFadden and Merrill (1995).

sediments is summarized in Fig. 4.15b, which shows a plot of the longitude at which each VGP transition path crosses the equator. The claimed clustering of the paths in two approximately antipodal positions is readily apparent. As shown by McFadden et al. (1993), this clustering is significantly greater than would be expected from a uniform random distribution.

However, it should be noted that both Valet et al. (1992) and McFadden et al. (1993) questioned whether model v is actually supported by the very sediment data on which it is based; they both note that the clustering might be linked to the poor distribution of the observation sites, which are themselves strongly grouped in two antipodal longitude bands with the angles between the corresponding site longitude and VGP path strongly clustering around ±90°.

During polarity transitions the intensity of the geomagnetic field decreases on average to about 25% of its value before or after the transition, although values as low as 10% have been observed (§4.4.3). The problem that arises is whether the magnetic torque on the settling magnetic particles has become too weak to overcome other effects so that there is inclination shallowing as discussed in §2.3.7. Quidelleur and Valet (1994) considered some aspects of this concept and then Quidelleur et al. (1995) carried out redeposition experiments to test it. Using the relation of (2.3.37), they obtained values of the inclination shallowing factor/as low as 0.23 with an ambient field of 4.6 nT, similar to the field in the middle of a transition. Barton and McFadden (1996) showed that, for this to be an important factor in the clustering of VGP transition paths, values of /below 0.3 are required and that the effect becomes pronounced when /approaches 0.1. Thus, there is the potential that clustering of transitional VGPs may have been enhanced in sedimentary records by a combination of rock magnetic problems and a poor distribution of sites.

Despite the conclusion by Prévôt and Camps (1993) that the lava data do not show any obvious clustering or preference for longitudinal bands, there is some evidence to the contrary. Constable (1992) analyzed nontransitional lava data from flows spanning the past 5 Myr to determine if there is any nonuniformity in longitudinal distribution. Her analysis included more than 2000 VGPs and showed a distinct bias toward two longitude bands, which were similar to the preferred bands identified in the sediment data. Some of this bias is almost certainly a consequence of the present orientation of the equatorial dipole. Love (1998) has repeated the test suggested by Prévôt and Camps (1993) but arrived at the dramatically different conclusion that the transitional VGPs do exhibit longitudinal preference for the Americas and for the antipodal Asian path (he did not, however, find evidence for latitudinal clustering). The initial data were similar (transitional VGPs younger than 20 Ma) but the selection criteria were different; Love (1998) considers VGPs from different lavas to be independent measurements, despite the counterview that lava flows typically bunch together in short intervals of time (Holcomb, 1987). The critical step is that Love weights each datum (VGP) according to its latitude and the number of transitional VGPs in that record. Histograms of the number of weighted datums versus longitude then show a distribution which, at the 99% level of confidence, is nonuniform. Despite the fact that the analysis fails to separate the critical issues noted above (Merrill and McFadden, 1999), the finding by Love (1998) that the two most occupied transitional longitudes are similar to those identified by analyses of sediment data is impressive.

Hoffman (1999) analyzed VGP data from the Love and Mazaud (1997) database for the Matuyama-Brunhes reversal transition. The data indicate the existence of four major groupings of virtual poles, leading to the suggestion of preferred locations for transitional VGPs (see also Hoffman, 1991a, 1992a).

Interpretation of the transitional field directions remains enigmatic. The problem is exacerbated by the fact that the claimed preference of transitional VGPs for certain longitude bands is small. For example, Love (1998) finds approximately one quarter of the weighted data fall into the two most populated longitudinal bands, whereas for a uniform distribution one might expect about one sixth of the data in those bands. This highlights the fact that the questions can only be resolved by observation, but the current paucity of relevant data precludes a clear resolution.

4.4.3 Intensity Changes

Reliable paleointensity information is typically much more difficult to obtain than reliable directional data. Thus, it is to be expected that even less would be known about the transitional field strength than about the transitional field directions. Nevertheless, there appears to be more agreement about the nature of intensity changes during a transition than about the directional changes, and some robust conclusions can be drawn.

Absolute paleointensities determined from lavas and relative paleointensity estimates from both sedimentary and igneous rocks recording transition directions indicate that the field intensity decreases substantially during a polarity change. These data indicate reductions sometimes to only 10% of the usual field intensity outside the transition. Figure 4.16 provides a summary of absolute paleointensities relating to polarity transitions for lavas less than about 10 Ma. Absolute paleointensities for this time interval have been analyzed by Tanaka et al. (1995) using only those determinations made by the Thellier (Thellier and Thellier, 1959a,b; see §1.2.4) or Shaw (1974) methods. The 323 values include many from the central part of polarity transitions that are characterized by low-latitude VGPs. In order to allow for the different site locations all intensity values have been converted to VDMs (see §1.2.4 and §1.2.5). Despite the attendant problems of using VDM during transitional times, the use of VDM (as with VGP) should be seen as a convenient mapping that attempts to take account of the effects of site location on the surface of the Earth. Average VDM for VGP latitude bands of width 20° between -90° and +90° are shown in Fig. 4.16. Note that when the VGP latitude lies between -45° and +45°, there is a dramatic drop in VDM values. These absolute intensity values indicate that on average the field at the central part of transitions (VGP latitude 0°) is about 25% its usual value.

The reduction in intensity values for low-latitude transitional VGPs provides robust evidence against a reversal model in which polarity reversal occurs by rotation of the dipole field with no change in magnitude (Merrill and McFadden,

VGP latitude

Fig. 4.16. Mean VDM versus VGP latitude for data covering the past 10 Myr averaged over 20° VGP latitude bands. The number of values averaged in each band is indicated and 95% confidence limits are shown. After Tanaka et al. (1995).

VGP latitude

Fig. 4.16. Mean VDM versus VGP latitude for data covering the past 10 Myr averaged over 20° VGP latitude bands. The number of values averaged in each band is indicated and 95% confidence limits are shown. After Tanaka et al. (1995).

1999). This is because a reduced intensity for low-latitude transitional VGPs demands a reduction in the dipole field strength during a polarity transition.

4.4.4 Polarity Transition Duration

In order to understand a process it is necessary to know the order in which events occurred and how long they took to occur. Consequently it is important to determine the duration of geomagnetic polarity transitions. Unfortunately estimates of the duration can vary by more than an order of magnitude for the same transition. Merrill and McFadden (1999) conclude that the duration of an average geomagnetic polarity transition is not well known but probably lies between 1000 and 8000 years (see also Bogue and Merrill, 1992).

There are four main methods for estimating the duration of polarity transitions, (i) Rapidly deposited sediments. This requires evidence of a constant rate of deposition, which is then estimated using absolute or relative dating methods to obtain ages before and after the transition. The duration of the transition can then be estimated from the depth interval in which the directions are transitional. A variant of this method combines relative intensity information with the directional information. The best estimates come from oceanic cores, for example, 4000 years by Harrison and

Somayajulu (1966), 4700 years by Niitsuma (1971) and 4600 years by Opdyke (1972). Other estimates also cluster around 4000-5000 years.

(ii) Statistical approach with lava flow data. This method uses the ratio of the number of transitional directions to the number of non-transitional directions in some time window and (under the assumption of a uniform random sample in time) apportions the times linearly to estimate the average time the field has spent in a transitional state. This was first undertaken by Cox and Dalrymple (1967) who obtained an average duration of 4600 years. Kristjansson (1985) obtained an average value near 6000 years. Unfortunately, magnetic field excursions (§4.4.5) also produce transitional directions. Lund et al. (1998) note that there is evidence for at least 14 excursions in the Brunhes alone, so they may be relatively common. This would tend to produce a bias toward an overestimate for the average duration of genuine reversal transitions.

(iii) Cooling rate in intrusive igneous rocks. If a transition occurs during the cooling of an intrusive igneous rock and the cooling is sufficiently slow, then the magnetic field changes are recorded as the cooling front sweeps through the intrusive (Dunn et al., 1971; Dodson et al., 1978). This method appears less reliable than the others because of uncertainties in estimating cooling rates and because the total magnetization at any point is not acquired instantaneously.

(iv) Precise absolute dating. This method is typified by Singer and Pringle (1996), who used precise 40Ar/39Ar dating on lava flows from different localities that have recorded transitional directions for the same reversal. There are significant difficulties in resolving such a short time span with radiometric dating: but this process may in the future provide our best estimates.

Theoretical considerations suggest that the reversal process proper probably takes longer than the time over which directional changes are seen. It has also been suggested that the intensity variations in a polarity transition occur over a longer time interval than the directional changes. These suggestions are based on transition records observed in plutons (Dunn et al., 1971), lavas (Mankinen et al., 1985; Bogue and Paul, 1993), and loess (Zhu et al., 1993). Using the statistical properties of the reversal time scale, McFadden and Merrill (1993) also suggested that the reversal process is longer than that observed in paleomagnetic directional data.

Lava flows at Steens Mountain in Oregon recorded a reverse to normal polarity transition about 16 Ma. Interpretations of directional and intensity data from some of these flows suggest that extremely rapid changes sometimes occurred in the transitional magnetic field (Mankinen et al., 1985; Prévôt et al., 1985; Coe et al., 1995); the astonishingly rapid change of 6° per day averaged over several days has been suggested. Merrill and McFadden (1999) query these interpretations, concluding that processes such as remagnetization associated with chemical changes could be responsible for the observations.

4.4.5 Geomagnetic Excursions

Wide departures from the geocentric axial dipole field direction have been observed to occur at a single locality, even though the field does not appear to change polarity but returns to its previous state. Such departures have been termed geomagnetic excursions. Verosub and Banerjee (1977) defined an excursion to have occurred when the VGP departs more than 45° from the geographic pole (but that departure is not associated with a polarity transition). Lund et al. (1998) have shown that there is evidence for at least 14 such excursions during the Brunhes, so it is likely they are a common feature of geomagnetic field variations. Sometimes it is difficult to distinguish when an excursion has occurred because short subchrons (~105 years in duration) are known to be present in the geomagnetic record (see §4.2.1).

Excursions are known to occur in lava successions, the best known of which is the Laschamp excursion observed in the Chaîne des Puys in France (Bonhommet and Babkine, 1967). They also occur in sedimentary sequences, but the interpretation is often equivocal because their time scale is short (<104 years) and only a narrow band of sediment is typically involved. Also, it can be argued that the departures are due to some effect of the sedimentation itself rather than the geomagnetic field (Verosub and Banerjee, 1977). Champion et al. (1988) identified 10 events in the late Matuyama and Brunhes (Fig. 4.17); they claimed that these events were actually field reversals and classified them as subchrons. However, the necessary evidence to conclude that these events are genuine reversals is lacking. Merrill and McFadden (1994), Langereis et al. (1997), and Lund et al. (1998) all classify these events as excursions. Indeed, the existence of several very short subchrons in the Brunhes would be at odds with the observed structure of the GPTS since the Cretaceous (§4.5.4).

Two possible explanations for geomagnetic excursions appear likely; they reflect either large amplitude secular variation or aborted reversals of the geomagnetic field. Distinguishing between thèse alternatives may not be possible unless aborted reversals can be shown to have a different signature from large-amplitude secular variation. It is theoretically possible for the nondipole field to become large enough that local reversals of the field could occur and thus both polarities would appear to exist simultaneously over a wide region (Coe, 1977; Merrill and McFadden, 1994). Gubbins (1999) notes that an excursion may represent a reversal of the magnetic field in the outer core but not in the inner core, so that the field would return to its original polarity. Consequently, an excursion could be global as suggested by Langereis et al. (1997) for six well-dated excursions during the Brunhes.

¡.asthamp

Blake Jumaica

[Mvantine Biwa III

Emperor

Big Lost Delta

Maluyama-Brunhes boundary

Kamikatsura

Jaramitlo subchron

Cobb Mountain

Fig. 4.17. Late Matuyama-Brunhes geomagnetic excursions (names in italics) identified by Champion el al. (1988). Black, normal polarity; white, reverse polarity; shaded, excursions. The duration of the excursions is <104 years.

Was this article helpful?

0 0

Post a comment