Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 18, 1854, doi:10.

1029/2001GL014543, 2002

Time-independent and time-dependent behaviour of high-latitude


flux bundles at the core-mantle boundary
Jeremy Bloxham
Department of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts, USA
Received 10 December 2001; revised 25 March 2002; accepted 29 March 2002; published 17 September 2002.

[1] Using a numerical dynamo model, we examine whether much longer, we would expect these features to be persis-
quasi-stationary high latitude flux bundles which are tent on much longer timescales than 300 years.
observed in the geomagnetic field over the last 300 years, [5] A number of studies have examined whether there is
and possibly over much longer time periods, result from evidence for flux bundles in paleomagnetic measurements
thermal core-mantle interactions. We do so by imposing of the field. The evidence is mixed: of recent studies of the
lateral heat flux variations at the core-mantle boundary based time-averaged field over the last 5 Myr, Kelly and Gubbins
upon the dominant pattern of seismic heterogeneity in the [1997] find flux bundles at approximately the same loca-
lowermost mantle. We find that the time-averaged field then tions (in the northern hemisphere) as the historical record
has a similar arrangement of flux bundles to that observed (see Figure 1b), Johnson and Constable [1997] find evi-
during the last few hundred years. However, the field is dence of non-zonal structure, but not unambiguous evidence
time-dependent, and this arrangement of flux bundles is of flux bundles, while Carlut and Courtillot [1998] fail to
not a feature of the field at all times. INDEX TERMS: find evidence of any non-zonal structure. Given the poor
1507 Geogmanetism and Paleomagnetism: Core processes (8115); distribution of measurements, persistent flux bundles cannot
1510 Geogmanetism and Paleomagnetism: Dynamo theories; 1560 be ruled out based on these studies [Constable et al., 2000].
Geogmanetism and Paleomagnetism: Time variations—secular and Additional support for non-zonal structure is found on
long term. Citation: Bloxham, J., Time-independent and time- intermediate timescales, namely from a 3000 year time
dependent behaviour of high-latitude flux bundles at the core- average of the field [Johnson and Constable, 1998].
mantle boundary, Geophys. Res. Lett., 29(18), 1854, doi:10.1029/ [6] A recent series of models of the field at 100 year
2001GL014543, 2002. intervals over the last 3000 years permits the persistence of
flux bundles to be studied more directly [Constable et al.,
2000]. Flux bundles are clearly seen at some points during
1. Introduction the interval but not at all times, with their number and
[2] Maps of the magnetic field at the core-mantle boun- locations varying.
dary over the last 300 years [Bloxham and Gubbins, 1985] [7] Thus, the observational evidence may be summarized
reveal the presence of high-latitude flux bundles [Gubbins as follows: throughout the historical record, i.e. on a time-
and Bloxham, 1987]. Figure 1a shows the average magnetic scale of a few hundred years (which is comparable to the
field at the core-mantle boundary for the period 1840 – 1990 convective turnover timescale of the core), nearly stationary
calculated from the ufm1 field model [Bloxham and Jack- flux bundles are clearly observed; on timescales of several
son, 1992]: concentrations of flux are located roughly thousand years, such flux bundles are transient; and on
symmetrically about the geographical equator, with two timescales of several million years, there is weak evidence
bundles in each hemisphere at approximately 60° north that the time-averaged field has flux concentrations at
and south of the equator. roughly the same locations at which they are observed on
[3] The stationarity of these persistent flux bundles is the several hundred year timescale. Can we reconcile these
unexpected, even on the timescale of a few hundred years. observations?
The westward drift paradigm of the secular variation [Hal- [8] Here we examine this question using a numerical
ley, 1692] predicts that they should drift through about 60° dynamo model [Kuang and Bloxham, 1997, 1999] on which
of longitude during this 300 year period. Instead, their we impose laterally-varying heat flux boundary conditions at
longitude remains roughly constant, leading to the sugges- the core-mantle boundary consistent with the dominant Y22
tion that they are held in place by either topographic pattern of seismic heterogeneity in the lowermost mantle
[Gubbins and Richards, 1986] or thermal [Bloxham and [Giardini et al., 1987]. We interpret seismic heterogeneity as
Gubbins, 1987] core-mantle interactions. The idea that resulting from lateral variations in temperature in the low-
thermal core-mantle interactions might influence the geo- ermost mantle, so that seismically fast regions correspond to
dynamo is not new [Cox and Doell, 1964; Hide, 1967]; here cold regions, and thus to regions of higher heat flux from the
though we address the specific possibility of persistent flux core. The peak amplitude of the imposed heat flux variations
bundles arising through thermal core-mantle interactions. is 37%. The amplitude of lateral variations in heat flux at the
[4] If the flux bundles are due to thermal core-mantle core-mantle boundary is very poorly constrained; the figure
interactions then, given that the lowermost mantle changes that we use here is not geophysically unreasonably large.
on Ta timescale of at least order 107 years and probably [9] The effects of laterally varying heat flux on numerical
dynamo models has been studied previously [Sarson et al.,
Copyright 2002 by the American Geophysical Union. 1997; Glatzmaier et al., 1999; Bloxham, 2000a, 2000b].
0094-8276/02/2001GL014543 One finding that has emerged is that flux bundles may be

1-1
1-2 BLOXHAM: HIGH LATITUDE FLUX BUNDLES AT THE CMB

paleomagnetic field. We note that model calculations with-


out lateral variations in heat flux result in an axisymmetric
time-average on timescales as short as a few thousand years.
[11] The flux bundles lie beneath regions of high heat flux
from the core. The time-averaged flow shows that they are
associated with vortex columns in which the fluid rotates in
the prograde sense about the axis of the column. Thus the
flow associated with the columns is eastward equatorward of
the column and westward poleward of the axis.
[12] Maps of the fluid flow at the core surface over the last
150 years [Bloxham, 1992] show a different pattern. From
roughly 270° east longitude eastwards to 90° east longitude
(and in particular beneath the Atlantic and Africa) the flow is
predominantly westward near the equator, while in the
opposite hemisphere (beneath the Pacific) the flow is weak
(or in some models eastward). The pattern of flux bundles
inferred at the core surface over this time period is not as
simple as the m = 2 pattern imposed on the dynamo model:
over the last three hundred years, the two flux bundles are less
than 180° apart, with their centers closer to 180° longitude
than to 0° longitude. In other words, they are displaced
towards the Pacific hemisphere. We might then expect that
the eastward flow associated with the columns would be
strongest beneath the Pacific (where the columns are closer
together). Now suppose that this part of the flow induced by
the columns is superimposed on an otherwise uniform west-
ward drift at the core surface. Then, we might expect little or
no westward drift beneath the Pacific hemisphere (where the
eastward columnar flow might cancel the westward drift) and
westward drift beneath the opposite hemisphere (where the
eastward columnar flow might not be strong enough to cancel
the westward drift). To address this possibility further we
would need to consider patterns of lateral heat flux variation
with a more complicated structure than m = 2.
[13] Constable et al. [2000] noted an absence of flux
bundles in the central Pacific hemisphere (i.e. around 180°
longitude) over the last 3000 years. The results presented here
Figure 1. Time-averaged radial component of the mag-
do show, on occasion, flux bundles around that longitude;
netic field at the core-mantle boundary: (a) 1840 – 1990
however, we noted in an earlier study [Bloxham, 2000a] in
[Bloxham and Jackson, 1992] and (b) 0 – 5 Ma [Kelly and
which we looked at the effects of a more complicated pattern
Gubbins, 1997]; and (c) derived from a 350,000 year
of heat flux variation (derived from a seismic model of the
average of the numerical dynamo model.
lowermost mantle Masters et al. [1996]) that the field within
the Pacific region was more nearly dipolar than elsewhere.
intermittent [Sarson et al., 1997; Bloxham, 2000a], a find- This suggests that the absence of flux bundles in the central
ing which is consistent with studies of locking of thermal Pacific over the last 3000 years may be the result of terms
convection to variations in boundary heat flux [Zhang and other than Y22 in the heat flux variation.
Gubbins, 1993, 1997]. Here we use a fully 3-dimensional
dynamo model (rather than a model with only one non-
axisymmetric mode [Sarson et al., 1997]) and integrate over 2.2. Time Dependent Behaviour
a long time period (the total model run is 24 dipole decay [14] The pattern of flux concentrations is not stationary.
times or nearly 500,000 years, though we exclude from this In Figure 2 we show and describe the evolution of the
study the initial 150,000 years which included a transitional magnetic field at the CMB over a 4000 year interval from
period between normal and reversed polarity) so that the the dynamo model. Two points are noteworthy. First,
time-average can be considered with some confidence. during this time-interval there are several intervals in
which the field morphology is not the simple pattern seen
in the time-average. Instead, there are intervals with one
2. Results dominant flux concentration in each hemisphere, with the
2.1. Time-Independent Behaviour concentration at 90° east sometimes dominant and at other
[10] In Figure 1c, we show the magnetic field from the times the concentration at 270° east dominant. At other
numerical model averaged over 17.5 dipole decay times (or times, the pattern is more complicated with several flux
roughly 350,000 years). The field is not axisymmetric and concentrations evident in each hemisphere. Similar behav-
has high-latitude flux bundles at similar locations to those iour was described by Constable et al. [2000] over the
observed in the recent field and in the time-averaged last 3000 years. Second, despite the time-dependent nature
BLOXHAM: HIGH LATITUDE FLUX BUNDLES AT THE CMB 1-3

Figure 2. Evolution of the radial component of the magnetic field at the core-mantle boundary obtained from the dynamo
model over a 4000-year interval. At year 0, the field is in a configuration with two flux concentrations in each hemisphere,
though with much stronger flux concentrations at 90° to 135° east than at 270° east. Over the next 700 years, this
configuration evolves into one with a more even distribution of flux between the two patches in each hemisphere. Then,
over the following 400 years, the pattern evolves into one with several flux concentrations in each hemisphere; this
evolution is driven in part by interaction of the flux concentrations with drifting patches of reversed and normal flux near
the equator. The pattern then evolves by year 1700 back to one with two approximately equal flux concentrations in each
hemisphere, though that pattern then evolves over the next 800 years to one with much stronger flux concentrations at 270°
east than at 90° east, the mirror image of the starting configuration. The evolution of this pattern is similar to that of its
mirror image with several flux concentrations forming by year 2900. Finally, the pattern evolves through year 3300 to year
4000 to a pattern with two approximately equal flux concentrations in each hemisphere. The final plot shows the time-
average over this time interval: even over an interval as short as 4000 years in which there is considerable time-dependency
of the flow, the time average is similar to that shown in Figure 1c (a 350,000 year time-average).

of the pattern, the time-averaged field over this 4000-year to obtain time-averaged field models which show a very
time-interval is similar to the time-average over the similar field morphology to the time-averaged field from
350,000-year time-interval in Figure 1c. This suggests field models over the interval 1840 – 1990 and 0 – 5 Ma.
that we may be able to obtain useful results by studying Second, the pattern is time-dependent so that the absence of
the field morphology from dynamo models over intervals persistent flux concentrations in the field models of Con-
of just a few thousand years (although a much longer stable et al. [2000] cannot be taken as evidence against their
interval may be required for any transient behaviour from presence in the time-averaged field.
the initial conditions to relax). We note that an analogous [16] Given that the flux bundles are not permanent, even
argument has been made for the interpretation of paleo- though they may be a feature of the time-averaged field, it is
magnetic data [Carlut et al., 1999]. perhaps fortunate that the last few hundred years, when we
have had direct observations of the field, has been a period
when these flux bundles are present in the field. In terms of
3. Discussion the morphology of the field at the core-mantle boundary, the
[15] There are two main results of this study. First, by last few hundred years may thus provide a better indication
imposing a Y22 pattern of heat flux variation at the core- of the time-averaged field than would ordinarily have been
mantle boundary on a numerical dynamo model we are able expected.
1-4 BLOXHAM: HIGH LATITUDE FLUX BUNDLES AT THE CMB

[17] The latitude at which these flux bundles occur has Bloxham, J., Sensitivity of the geomagnetic axial dipole to thermal core-
been previously argued to result from the effects of the inner mantle interactions, Nature, 405, 63 – 65, 2000b.
Bloxham, J., and D. Gubbins, The secular variation of the Earth’s magnetic
core and its tangent cylinder (the cylinder coaxial with the field, Nature, 317, 777 – 781, 1985.
rotation axis and tangential to the inner core boundary) Bloxham, J., and D. Gubbins, Thermal core-mantle interactions, Nature,
[Gubbins and Bloxham, 1987], while the longitude at which 325, 511 – 513, 1987.
Bloxham, J., and A. Jackson, Time-dependent mapping of the magnetic
they occur has been argued to be the result of thermal core- field at the core-mantle boundary, J. Geophys. Res., 97, 19,537 – 19,563,
mantle interactions [Bloxham and Gubbins, 1987]. In this 1992.
study, by imposing a Y22 pattern of heat flux variation, we Carlut, J., and V. Courtillot, How complex is the time-averaged geomag-
netic field over the last 5 million years?, Geophys. J. Int., 134, 527 – 544,
have fixed both the latitude and longitude of the heat flux 1998.
variations. Further work is required to determine whether Carlut, J., V. Courtillot, and G. Hulot, Over how much time should the
the latitude of the flux bundles is sensitive to the latitude of geomagnetic field be averaged to obtain the mean-paleomagnetic field,
the heat flux variations or whether it is contolled entirely by Terra Nova, 11, 239 – 243, 1999.
Constable, C. G., C. L. Johnson, and S. P. Lund, Global geomagnetic field
the inner core tangent cylinder; this could be addressed by models for the past 3000 years: Transient or permanent flux lobes?, Phil.
adding terms Y2l with l > 2 and even. On the other hand, and Trans. Roy. Soc. Lond. A, 358, 991 – 1008, 2000.
very much more speculatively, do the rotationally domi- Cox, A., and R. Doell, Long period variations of the geomagnetic field,
Bull. Seism. Soc. Am., 54, 2243 – 2270, 1964.
nated dynamics of the core and the presence of the inner Giardini, D., X.-D. Li, and J. Woodhouse, Three-dimensional structure of
core and its tangent cylinder influence the the pattern of the Earth from splitting in free-oscillation spectra, Nature, 325, 405 – 411,
lateral variation in the lowermost mantle, so that the heat 1987.
Glatzmaier, G., R. Coe, L. Hongre, and P. Roberts, The role of the Earth’s
flux anomalies necessarily occur at the correct latitude for mantle in controlling the frequency of geomagnetic reversals, Nature,
flux bundles to lie on the edge of the tangent cylinder? 401, 885 – 890, 1999.
[18] It is worth noting that had we interpreted seismi- Gubbins, D., and J. Bloxham, Morphology of the geomagnetic field and
cally fast regions as corresponding instead to regions of implications for the geodynamo, Nature, 325, 509 – 511, 1987.
Gubbins, D., and M. Richards, Coupling of the core dynamo and mantle -
low heat flux then our maps of the field would simply thermal or topographic, Geophys. Res. Lett., 325, 1521 – 1524, 1986.
have been shifted by 90° in longitude. Then, the flux Halley, E., An account of the cause of the change of the variation of the
bundles would have been in the wrong location with magnetical needle, with an hypothesis of the structure of the internal parts
of the Earth, Phil. Trans. Roy. Soc. Lond., 16, 563 – 578, 1692.
respect to the observational record. This places a con- Hide, R., Motions of the Earth’s core and mantle, and variations of the main
straint on models of chemical heterogeneity of the lower- geomagnetic field, Science, 157, 3784 – 3785, 1967.
most mantle [Karato and Karki, 2001]. Chemical Johnson, C., and C. Constable, The time-averaged geomagnetic field: Glo-
bal and regional biases for 0 – 5 ma, Geophys. J. Int., 131, 643 – 666, 1997.
heterogeneity will contribute to heat flux variations as a Johnson, C., and C. Constable, Persistently anomalous Pacific geomagnetic
result of variations in thermal conductivity. Any model of fields, Geophys. Res. Lett., 25, 1011 – 1014, 1998.
chemical heterogeneity (or more accurately of combined Karato, S., and B. Karki, Origin of lateral variation of seismic wave velo-
chemical and thermal heterogeneity) of the lowermost cities and density in the deep mantle, J. Geophys. Res., 106, 21,771 –
21,783, 2001.
mantle must satisfy this new constraint that seismically Kelly, P., and D. Gubbins, The geomagnetic field over the past 5 Myr,
fast regions correspond to high heat flux. Geophys. J. Int., 128, 315 – 330, 1997.
[19] Finally, we must caution that, given the complex Kuang, W., and J. Bloxham, An Earth-like numerical dynamo model, Nat-
ure, 389, 371 – 374, 1997.
behaviour of thermal core-mantle interactions reported Kuang, W., and J. Bloxham, Numerical modelling of magnetohydrody-
with 212-dimensional models[Sarson et al., 1997], addi- namic convection in a rapidly rotating spherical shell: Weak and strong
tional investigation is now called for so that the depend- field dynamo action, J. Comp. Phys., 153, 51 – 81, 1999.
Masters, G., S. Johnson, G. Laske, and H. Bolton, A shear-velocity model
ence of the results reported here on the amplitude of the of the mantle, Phil. Trans. Roy. Soc. Lond. A, 354, 1385 – 1410, 1996.
lateral variations in heat flux can be determined; modes in Sarson, G. R., C. A. Jones, and A. W. Longbottom, The influence of
addition to Y22 should also be considered. boundary region heterogeneities on the geodynamo, Phys. Earth Planet.
Inter., 101, 13 – 32, 1997.
Zhang, K., and D. Gubbins, Convection in a rotating spherical fluid shell
[20] Acknowledgments. This work was supported by NSF Awards with an inhomogeneous temperature boundary condition at infinite
EAR-0073988 and EAR-0112469. prandtl number, J. Fluid Mech., 250, 209 – 232, 1993.
Zhang, K., and D. Gubbins, Nonlinear aspects of core-mantle interaction,
Geophys. Res. Lett., 20, 2969 – 2972, 1997.
References
Bloxham, J., The steady part of the secular variation of the Earth’s magnetic
field, J. Geophys. Res., 97, 19,565 – 19,579, 1992.
Bloxham, J., The effect of thermal core-mantle interactions on the paleo-
magnetic secular variation, Phil. Trans. Roy. Soc. Lond. A, 358, 1171 – J. Bloxham, Department of Earth and Planetary Sciences, Harvard
1179, 2000a. University, Cambridge, MA 02138, USA. ( jeremy_bloxham@harvard.edu)

You might also like