You are on page 1of 18

Annu. Rev. Earth Planet. Sci. 1998. 26:50117 Copyright c 1998 by Annual Reviews.

All rights reserved

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

DYNAMICS OF ANGULAR MOMENTUM IN THE EARTHS CORE


Jeremy Bloxham
Department of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts 02138; e-mail: bloxham@geophysics.harvard.edu
KEY WORDS: Earths core, dynamo theory, geomagnetism, length of day

ABSTRACT
We examine the dynamics of the angular momentum balance of the Earths core. Not only is this balance of great importance to theories of the geodynamo process that is responsible for the generation of the Earths magnetic eld, but recent work has shown that angular momentum variations in the core have broad geophysical implications, ranging from studies of the travel times of seismic waves through the inner core to attempts to account for a possible phase discrepancy between atmospheric and oceanic angular momentum. In this review, we present a simple account of the underlying dynamics and review the relevant observations and their interpretation.

INTRODUCTION
The Earths core exhibits rich dynamics. Like the atmosphere and oceans, the outer core is a rapidly rotating uid in the sense that the effects of rotation are of importance in the leading order force balance, with viscous effects playing a secondary role at most. It differs dynamically from the atmosphere and oceans primarily in that the magnetic eld plays an important role and in that the outer core is a thick rather than thin shell, with a depth around two thirds of its radius. In the solid inner core, on the other hand, the dynamics are more like those of the mantle, with viscous effects of leading importance. The last few years have seen a number of exciting developments, both observational and theoretical, in our understanding of the dynamics of the Earths 501 0084-6597/98/0515-0501$08.00

502

BLOXHAM

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

core. These range from numerical studies of the geodynamo to seismological studies of differential rotation of the inner core to studies of gravitational coupling between the inner core and mantle. In this review we concentrate on these recent developments and attempt to tie them together within a common theme, namely the differential rotation or angular momentum balance of the Earths core. A number of recent reviews are of relevance. We mention reviews by Jeanloz (1990) for coverage of the material properties of the core, Braginsky & Roberts (1995) for a complete and careful derivation of the underlying equations describing the dynamics of the core, and Fearn et al (1988) and Roberts & Soward (1992) for reviews of dynamo theory. Finally, for an extremely comprehensive review of the Earths core, we highly recommend Jacobs (1987).

DYNAMICAL PREREQUISITES
Convection in the core is driven by the removal of heat from the core by the mantle. The heat that is removed from the core originates from two sources: the decay of radioactive isotopes in the core and the secular cooling of the core. The secular cooling of the core in turn results in growth of the solid inner core and the release of gravitational potential energy and latent heat (Verhoogen 1961, Braginsky 1963, Gubbins et al 1979). The geodynamo, the process by which the Earths magnetic eld is maintained against Ohmic decay, may be thought of as an instability that the Earths core exploits to aid in the transfer of this heat and gravitational potential energy to the core-mantle boundary: The generation of a magnetic eld assists the convective transfer of heat by providing a means to relax the constraint imposed on convective heat transport by the rotation of the Earth. In order to understand the dynamics of the core we must include the magnetic eld since it plays a central role in all aspects of the dynamics, although it considerably complicates the behaviour of the core. We start by building a relatively simple model of core dynamics. We include only those aspects of the dynamics that are essential. For simplicity, we adopt the Boussinesq approximation, in which density variations are only important to the extent that they result in buoyancy. We assume also that density variations result only from thermal expansion, thus ignoring compositional variations, and we assume that there are no internal heat sources. With these assumptions, the dynamics are described by the momentum equation, the magnetic induction equation, and an energy equation: 1 j B+ g + 2 u; (1) (t + u ) u + 2 u = p + 0 0 t B = ( u B) + 2 B; (t + u ) = 2 + Q diss . (2) (3)

DYNAMICS OF THE EARTHS CORE

503

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Here, u is the uid ow, B the magnetic induction, j the current density, and the temperature. is the Earths rotation, p the modied pressure (including the centrifugal contribution), 0 the mean density, g the gravitational acceleration, the kinematic viscosity, and the magnetic and thermal diffusivities, respectively, and Qdiss the Joule heating (the viscous heating is small in comparison and neglected here). We should emphasize that this description is too simple for many purposes, including understanding the energetics of the geodynamo. However, it is adequate for the purposes of this review where our emphasis is on the dynamics of the core, specically, the momentum balance of the core. Note that even these simplied equations are coupled and the resulting nonlinearity adds greatly to the complexity of the solutions that they admit, to say nothing of the difculty of nding a solution. Insight into the nature of their solutions can be obtained most readily if we nondimensionalize the system. To do so, we choose the core radius r0 2 as our length scale, the magnetic diffusion time r0 / as our time scale, B = 1/2 (2 0 ) as our magnetic eld scale, and hT r0 as the temperature scale, where hT is the temperature gradient at the inner core boundary. Then, using the same symbols as before but now representing nondimensionalized quantities, we have u = p + j B + Ra r + E 2 u; Ro (t + u ) u + z t B = ( u B) + 2 B; (t + u ) = +q 2 + J | j | 2. Ra (4) (5) (6)

The nondimensional numbers that appear are the Ekman number E, the magnetic Rossby number Ro, the Prandtl number q , the Joule heating number J, and the Rayleigh number Ra, dened by E Ra , 2 2 ro Ro , 2 2 ro q , J r0 T go , cp (7)

2 T go h T ro , 2

in which go is the gravitational acceleration at ro, cp is the specic heat capacity, and T the thermal expansion. In Table 1, we give typical values for the various parameters and nondimensional numbers that we have introduced. In the Earths uid outer core, the Ekman number and magnetic Rossby number are both small; in fact, even if turbulent values are adopted for the diffusivities, they do not exceed O(109 ), so neglect of viscous effects and inertial

504

BLOXHAM
Table 1 Physical properties of the core and values of nondimensional numbers, taken from Jeanloz (1990) and Stacey (1992) (owing to the great uncertainty in its value, we do not give a value for the Rayleigh number) Physical properties Radius Mean d ensity Magnetic diffusivity Thermal diffusivity Kinematic viscosity Thermal expansion r T Outer core 3485 km 1.1 104 kg m3 2 m2 s1 1 105 m2 s1 107 m2 s1 105 K1 Inner core 1221 km 1.3 104 kg m3 2 m2 s1 1 105 m2 s1 109 m2 s1 6 106 K1 Outer core Molecular E Ro q J 1016 109 102 0.5 Turbulent 109 109 1 0.5

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Nondimensional numbers Ekman Magnetic Rossby Prandtl Joule heating

effects appears reasonable as an initial approximation. Then, the momentum balance has the form u = p + j B + Ra r , z (8)

which represents a balance between the Coriolis acceleration, pressure gradients, the Lorentz force, and buoyancy. This magnetostrophic balance poses some challenging theoretical problems. First, there may not necessarily be a solution to Equation 8. Taylor (1963) explored this issue and showed that a solution will only exist if a set of conditions, known collectively as Taylors constraint, is satised. These conditions are given by ( j B) d S = 0, (9)

where is the surface of any cylinder within the core coaxial with the rotation axis (Figure 1). Such cylinders are called Taylor (or geostrophic) cylinders, and a dynamo satisfying Taylors constraint is said to be in a Taylor state. The integral in Equation 9 is the proportional to the axial Lorentz torque on the cylinder . First, this torque represents physically the contribution of the magnetic eld to the axial angular momentum balance of the core. In the magnetostrophic limit, there is nothing to balance the torque, and so it must vanish.

DYNAMICS OF THE EARTHS CORE

505

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Figure 1 The geometry for the integral in Equation 9.

Second, even if Taylors constraint is satised at a particular point in time and so a solution exists, the uid velocity is nonunique, remaining undetermined up to an arbitrary geostrophic ow u (s). This arbitrary ow is determined by requiring that Taylors constraint be satised at all times, i.e. by requiring that the Taylor torque and its rst derivative in time vanish. As elucidated by Taylor (1963), this geostrophic ow is responsible for shearing the magnetic eld so that the magnetic eld has precisely that geometry required to satisfy Taylors constraint. Finally, Taylor (1963) considered how the core would evolve if it started from a state in which the constraint is not satised. In order to balance the torque, we restore inertia to the system, giving Ro (1 s 2 ) 2
1

1 u (s ) = t 4 s

( j B) d S .

(10)

This system now admits oscillatory solutions. As described by Taylor (1963), the motions then consist of torsional oscillations about a Taylor state in which each cylindrical annulus rotates as a rigid body, linked to its neighbors by Bs (Figure 2). Damping of these oscillations results in evolution of the system to a Taylor state.

506

BLOXHAM

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Figure 2 Torsional oscillations of cylindrical annuli about the rotation axis.

Alternatively, or additionally, we could choose to restore viscous effects in the boundary layers to the torque balance. Following Roberts & Soward (1992) and Hollerbach (1996), we then have Ro (1 s 2 ) 2
1

1 u (s ) 1 1 + E 2 (1 s 2 ) 4 u (s ) = t 4 s

( j B) d S . (11)

The rst term on the left, as we have seen, gives rise to torsional oscillations; the second term represents viscous spin-up or spin-down of the Taylor cylinders. If inertia is unimportant, i.e. the system evolves on a time scale much longer than that of torsional oscillations, the rst term can be neglected and a balance can be achieved by mechanisms different from that described by Taylor. First, a balance is possible if the magnetic eld is O( E 1/4 ); this so-called Ekman state is of limited geophysical interest because it results, in the geophysical parameter regime, in very weak magnetic elds. An alternative balance was proposed by Braginsky (1975), with u (s ) = O( E 1/3 ), Bs = O( E 1/6 ), and B = O(1). This state is described as Model-Z because the scaling implies that Bs is very small, so the poloidal part of the eld in the core consists mostly of Bz, with the poloidal eld lines almost parallel to the rotation (or z) axis.

DYNAMICS OF THE EARTHS CORE

507

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

For the small values of E appropriate to the Earths core, this balance poses a possible problem because the resultant geostrophic ow will become very large. However, Jault (1995), by retaining inertia in Equation 11, has shown that as E becomes very small, a Model-Z solution evolves smoothly into a Taylor state solution. As E becomes small, the viscous spin-up time scale becomes much longer than the time scale of torsional oscillations, and a Taylor state takes over. The Taylor state provides a dynamical reference state for the core. In the next section, we seek observational evidence for this model and seek in particular to answer whether departures from the Taylor state are balanced primarily by viscous boundary layer effects or by inertial effects.

THE ANGULAR MOMENTUM BALANCE OF THE OUTER CORE Observational Constraints


Observations of the rotation rate of the solid Earth (the length of day) show that the rotation rate varies on a wide range of frequencies. At the zero frequency limit, it decreases slowly because of tidal dissipation (Munk & MacDonald 1960); the concomitant transfer of angular momentum can be observed though the gradual recession of the Moons orbit about the Earth (Christodoulidis et al 1988). At high frequencies, say those corresponding to biennial and shorter periods, as rst proposed by Starr (1948) and rst demonstrated by Hide et al (1980), it varies owing to the exchange of angular momentum between the atmosphere and the solid Earth (for a review, see Rosen 1993). In between these lie the so-called decadal variations in the length of day, or variations in the rotation rate on periods of a few years to several hundred years. It has long been proposed that these arise from the exchange of angular momentum between the core and mantle, largely because all other possible culprits have been found lacking (Munk & Revelle 1952). Two approaches can be taken to attempting to understand the angular momentum balance of the Earth. The rst approach is kinematic: The object is simply to demonstrate that the books balance; in other words, if the angular momentum of the solid Earth decreases in some particular interval, a matching increase in angular momentum can be found elsewhere. The second approach is dynamical in character: The object is to demonstrate that a torque exists that explains the rate of transfer of angular momentum. Until recently, efforts to understand decadal variations were concentrated on understanding the dynamics, in other words on attempting to nd a coupling mechanism to explain the transfer of angular momentum between the core and mantle on decadal time scales. Sadly, despite the existence of several candidate

508

BLOXHAM

mechanisms, none has yet been demonstrated unequivocally to be the cause of the decadal variations. Electromagnetic core-mantle coupling (Bullard et al 1950, Munk & Revelle 1952) has received the most attention. It arises from the electromotive force in the electrically conducting mantle, induced by the time-varying magnetic eld. Although little doubt has been expressed, at least until recent lower estimates of the electrical conductivity of the mantle (for example, Shankland et al 1993), that electromagnetic torques can be of sufcient magnitude (Rochester 1960), it has proven considerably more difcult to demonstrate that electromagnetic torques can account for the time dependency of decadal length-of-day variations. To do so requires a torque that has an amplitude of about 1018 Nm varying on the decadal time scale, with only a small mean component when averaged over, say, 100 years. The analysis of this problem is complicated by two factors: First, the part of the magnetic eld that is due to leakage of toroidal eld from the core into the mantle and that contributes to the electromagnetic torque is unobservable, and second, the part of the uid ow that results in an electromagnetic torque by advecting the poloidal part of the eld is also unobservable, at least without the introduction of additional assumptions. With different assumptions, Jault & LeMou el (1991) and Love & Bloxham (1994) have found electromagnetic core-mantle coupling to be decient, although Holme (1998) indicates that it may not be ruled out. Topographical core-mantle coupling was rst suggested by Hide (1969). It arises from the pressure associated with ow in the outer core acting on asymmetrical topography on the core-mantle boundary. Following Hides suggestion, a number of studies addressed this mechanism by taking simple models of the uid ow and magnetic eld at the core-mantle boundary and examining the topographical torque that arises from the perturbation due to boundary topography (for example, Anufriev & Braginsky 1975a,b, 1977a,b, Moffatt & Dillon 1976, Moffatt 1978b, Kuang & Bloxham 1993). These studies have, in general, found that the topographical torque is too small to account for decadal length-of-day variations. In an alternative approach, Hide (1986) described a method to calculate the topographical torque from observations. In his method, observations of the geomagnetic secular variation are used to deduce the uid pressure at the core-mantle boundary, which is then used to calculate the topographical torque using a map of core-mantle boundary topography determined from seismic observations. The resultant topographical torque is more than adequate to explain the decadal length of day variations (Jault & LeMou el 1989, Hide et al 1993). However, Kuang & Bloxham (1997a) have challenged the validity of this approach on theoretical grounds. In any case, it has been shown that errors in both the determination of the uid pressure at the core-mantle boundary and in the determination of core-mantle boundary topography are more than large

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

DYNAMICS OF THE EARTHS CORE

509

enough to prevent use of this method to calculate the torque (Jault & LeMou el 1990, Kuang & Bloxham 1993). Gravitational coupling was rst suggested by Jault & LeMou el (1989). It arises from the gravitational interaction between aspherical density variations in the mantle and in the core. Unfortunately, as the aspherical density structure of the core is unknown, the importance of gravitational coupling is hard to assess. Recently, Buffett (1996a) considered gravitational interactions between the mantle and solid inner core, suggesting a mechanism for length-of-day variations in which the mantle is gravitationally coupled to the inner core, which is in turn electromagnetically coupled to the outer core (Buffett 1996b). The gravitational coupling between the mantle and inner core arises as follows: Large density variations in the mantle cause asphericity in the gravitational potential throughout the Earth, including in the neighborhood of the inner core boundary; the inner core boundary is (to a good approximation) an equipotential surface and so is aspherical. However, the inner core boundary also corresponds to a density jump: The resultant asphericity in the density distribution is much greater than that which arises from convection in the outer core. Thus, as Buffett argues, the gravitational coupling between the mantle and the inner core is likely to be much stronger than that between the mantle and outer core. Finally, viscous coupling between the core and mantle must occur, but even with turbulent values for the viscosity of the outer core, it is almost certainly several orders of magnitude too small. The kinematic approach to studying core-mantle angular momentum exchange was pioneered by Jault et al (1988) and has been more fruitful. In order to pursue this approach, we need to calculate the relative angular momentum of the core. Although the uid ow at the core surface can be determined from the geomagnetic secular variation (for a review, see Bloxham & Jackson 1991), calculating the angular momentum of the core obviously involves an integral of the uid ow throughout the core. This was seen as the insurmountable obstacle to this approach, but as realized by Jault et al (1988), if variations in core angular momentum involve torsional oscillations (Braginsky 1970), then it is sufcient to know the ow at the core surface, since torsional oscillations, as discussed in the previous sections, involve motions of the form u (s), which by denition are independent of the z direction. Jault et al (1988) applied this theory to observations of the length of day and inferences of core surface ow for the interval 19691985 and found an encouraging agreement. Subsequently, Jackson et al (1993) extended the interval to 18401990, nding a good agreement, at least after 1900 (see Figure 3). These studies provide unequivocal evidence that decadal changes in the rotation rate of the mantle are due to the exchange of angular momentum with the core. More signicantly, given that this had long been suspected to be the case anyway, it suggests strongly that decadal variations in the length of day are due

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

510
3

BLOXHAM

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

msecs

-1

-3

-5 1840.0

1890.0

1940.0

1990.0

Figure 3 Change in the length of day derived directly from geodetic observations (solid line) and that are predicted from calculations of the core angular momentum derived from geomagnetic observations. The square symbols show the results of Jault et al (1988) and the diamond symbols the results of Jackson et al (1993).

to torsional oscillations in the core. Furthermore, there is no signicant phase discrepancy in this period range, suggesting that the magnetic diffusion time scale of the mantle is less than a few years. Does this provide any evidence of a nearTaylor state geodynamo? Of course, if departures from the Taylor state were balanced entirely by inertia in the core, the angular momentum of the core would be conserved, so strictly speaking observations of changes in core angular momentum demonstrate that the dynamics are more complicated than Equation 10 and must include a source of coupling. However, the results of Jault et al (1988) and Jackson et al (1993) show that the time variations of the two harmonics of the ow that contribute to the core angular momentum are almost in opposite phase, so that the net changes in angular momentum are smaller than would be expected from the amplitude of the oscillatory ow. In other words, the core is nearly conserving its angular momentum, with the discrepancy accommodated by core-mantle coupling. This suggests that the Taylor state may be a reasonable approximation in the Earths core.

DYNAMICS OF THE EARTHS CORE

511

Zatman & Bloxham (1997b,c) have pursued the inference that torsional oscillations are responsible for decadal length of day variations further. They nd a good t to the variations in core angular momentum using two oscillations, and they have inverted the oscillations for a model of the root-mean-square (rms) value of Bs over each Taylor cylinder in the core. The value of Bs that is obtained is comparable to the strength of the radial magnetic eld at the core surface, in other words much larger than O( E 1/6 ) predicted by Model-Z, unless there is some large constant of proportionality in the asymptotic relation. Unfortunately, this kinematic picture offers little insight into the dynamics of core-mantle coupling. Given that on the decadal time scale the Taylor cylindrical annuli behave rigidly, the kinematics give no clue even as to whether the mechanism of core-mantle coupling takes place at the core-mantle boundary (for example, electromagnetic or topographical coupling), in the body of the outer core (for example, gravitational coupling), or even at the inner core boundary (through the process described by Buffett 1996a,b, 1997). An interesting question that arises from this picture of length-of-day variations and torsional oscillations concerns the angular momentum balance of the core at subdecadal periods. On decadal time scales, the cylindrical annuli behave rigidly, but on shorter time scales, that must cease to be the case as the spin-up of uid comprising each annulus must occur on a nite time scale. Eventually, at very short time scales the core will be effectively decoupled from the mantle and will play essentially no role in the angular momentum balance of the Earth, as demonstrated by Dickman & Nam (1995) at a 9-day period. The interesting period range is that which is sufciently long so that the core is not fully decoupled from the mantle but short enough so that the cylindrical annuli do not behave rigidly, i.e. at shorter periods than those of torsional oscillations. Zatman & Bloxham (1997a) have sought evidence of the effect of the core in this frequency range by looking for evidence of a phase discrepancy between atmospheric angular momentum and the rotation rate of the solid Earth at subdecadal periods. Evidence of a phase discrepancy would indicate that in this period range, angular momentum is not perfectly conserved in the atmosphere solid Earth system. They nd evidence of a phase discrepancy at a period of around 100 days. The interpretation of this result is not straightforward because it is possible that it might be due to transfer of angular momentum from the atmosphere to the oceans. However, if the core is responsible, which is not implausible, this explanation will provide a useful means to examine the dynamics of angular momentum transfer within the core and may possibly help to elucidate the mechanism of core-mantle coupling even at longer periods.

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Numerical Dynamo Models


In the last few years, attempts to solve the full system of equations (Equations 13) in the Earths core have begun to yield dynamos that are apparently in

512

BLOXHAM

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

the strong magnetic eld regime, using either a fully three-dimensional system (Glatzmaier & Roberts 1995a,b, 1996a,b, Kuang & Bloxham 1997bd) or a two-and-a-halfdimensional system (Jones et al 1995). Owing to computational constraints, however, these models have not been run in the same parameter regime as the Earths core. Furthermore, the models are extremely expensive to run, especially in the fully three-dimensional case, and the magnetic eld evolves in a complicated manner in time. As a result, extrapolation of the results to the Earths core is uncertain. This uncertainty is all the more troublesome because the different models differ greatly in the structure of the magnetic eld that they maintain within the core. A particular source of difculty in numerical dynamo modeling is to construct a model in which the leading-order force balance is magnetostrophic (Equation 8) and in which departures from that balance take the form of torsional oscillations (Equation 10). This is difcult to achieve because computational constraints limit the range of Ekman numbers that can be employed to values effectively no smaller than 105. In practice, the Ekman numbers used are larger than this because the viscosity is articially increased with wavenumber to aid the convergence properties of the solutions (for details, see Glatzmaier & Roberts 1995a, Kuang & Bloxham 1997b). To help alleviate this problem, Kuang & Bloxham (1997b,c) apply viscous stress-free boundary conditions. This eliminates the viscous term in Equation 11, but it also then imposes a computational constraint on the values of Ro that can be studied. As a result, unlike the calculations of Glatzmaier & Roberts, Ro is much larger than in the Earths core, although the ratio E / Ro is one, as is the case for the Earths core if the dissipative processes are turbulent. The solutions obtained by these two methods are very different (Kuang & Bloxham 1997c). Before the dynamics of the core can be understood through numerical models, more needs to be done to understand this difference in results from the various numerical methods. Although the computational demands imposed by these calculations suggest that an understanding of core dynamics through numerical modeling will not be immediately forthcoming, these and similar calculations are nonetheless an exciting avenue of investigation.

THE ANGULAR MOMENTUM BALANCE OF THE INNER CORE


Recently, seismological observations have suggested that the inner core has a prograde rotation relative to the mantle at a rate in the range of 1 /year (Song & Richards 1996) to 3 /year (Su et al 1996), although the most recent results of Creager (1997) place the rate at only 0.2 0.3 /year. Rotation of the inner

DYNAMICS OF THE EARTHS CORE

513

core relative to the mantle is not a new idea (Gubbins 1981), but it has received renewed interest in large part because of recent numerical dynamo modeling that predicted a prograde inner core rotation of about 3 /year (Glatzmaier & Roberts 1995a,b). Although the small amount of angular momentum carried by the inner core is of little consequence to the angular momentum budget of the Earth, it is nonetheless of considerable dynamical interest. In large part, because of its small moment of inertia, the inner core reacts passively to the surface stresses and body forces to which it is subject, rotating so that these forces are in near equilibrium, and so it acts as an indicator of the dynamics of the deep interior of the core. Owing to its high electrical conductivity (most likely similar to that of the outer core), the inner core will potentially be subject to substantial electromagnetic stresses. Gubbins (1981) estimated the resultant electromagnetic stress at 3 1019 Nm, which corresponds to an electromagnetic spin-up time of 2 days. In the absence of other forces, the inner core will rapidly spin up, reaching a state where the magnetic eld produced (or possibly more accurately annihilated) by the shear at its surface reduces the electromagnetic stress to zero. In some respects, this is similar to the balance on a Taylor cylinder, where the ow u (s) distorts the magnetic eld to make the Taylor torque vanish. The inner core may then oscillate about this equilibrium position with a period of about 10 years (Gubbins 1981). Several authors have proposed a mechanism to explain the prograde rotation of the inner core (Aurnou et al 1996, Glatzmaier & Roberts 1996b), based on the premise that a prograde thermal wind in the outer core near the inner core boundary drags poloidal eld lines through the inner core, thereby accelerating the inner core until it has a similar prograde rotation to the uid around it. However, Kuang & Bloxham (1997c) have found that in the Glatzmaier-Roberts dynamo model, the ow in the outer core in this region may be controlled by the viscous Ekman boundary layer structure instead of by a thermal wind. Instead, Kuang & Bloxham (1997c) have found in their dynamo model that the inner core exhibits both prograde and retrograde rotation relative to the mantle, with peak rotation rates of about 1 /year and a periodicity of several thousand years, not inconsistent with the seismic observations. It is unfortunate that the seismic observations are unable to discriminate between these very different rotational behaviors of the inner core predicted by dynamo models: To do so would require several thousand years of seismic observations of inner core rotation. Su et al (1996) attempted to use the rotation of the inner core to estimate the viscosity of the outer core. In their model, the prograde rotation of the inner core is ascribed to the gradual slowing of the rotation of the mantle owing to tidal friction; the slowing of the rotation of the inner core lags behind that of

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

514

BLOXHAM

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

the mantle and hence in the mantle frame appears as a prograde rotation. Tidal friction results in an increase in the length of day of about 2 ms per century (Stephenson & Morrison 1995), from which they estimate that the inner core rotation rate lags that of the mantle by about 105 years. They compare this to the Ekman spin-up time for the outer core and hence estimate the viscosity of the outer core. In other words, in their model the inner core rotation rate lags because of the slow viscous transfer of angular momentum across the outer core. However, the results of Jault et al (1988) and Jackson et al (1993) demonstrate that the transfer of angular momentum within the outer core takes place on a subdecadal time scale, and so on the 105-year time scale considered here the mantle and outer core are undoubtedly tightly coupled. Thus it seems highly unlikely, given the tight coupling of the inner core to the outer core (Gubbins 1981), that the lag of the inner core rotation rate could greatly exceed one year. Rotation of the inner core relative to the mantle might appear to be ruled out by the analysis of Buffett (1996a,b), as the gravitational restoring torques would be too large. However, Buffett (1997) shows that this is not necessarily the case, as the aspherical surface topography at the inner core boundary can relax viscously on a sufciently short time scale, provided the viscosity of the inner core does not exceed 1016 Pa s.

CONCLUSIONS
The constraints imposed by the conservation of angular momentum play a central role in the dynamics of the Earth and especially of the core. On the one hand they complicate the study of the geodynamo, but on the other they enable a broader set of observations (specically observations of variations in the rotation rate of the solid Earth) to be brought to bear on understanding the geodynamo. The next few years will be particularly exciting. 1998 will see the launch of the rsted satellite, a Danish satellite that will provide observations of the magnetic eld of comparable quality to those obtained nearly 20 years earlier by Magsat. Such observations are sorely needed and will enable us to study the angular momentum of the core in greater detail. Advances in numerical modeling of the geodynamo are certain to add to our understanding of core dynamics, especially if additional dynamo models become operational so that detailed comparisons can be made with different model assumptions and parameters. Even advances in understanding the angular momentum of the atmosphere and, in particular, the oceans will be important if they enable us to separate the angular momentum of the core from that of the rest of the Earth at subdecadal periods.

DYNAMICS OF THE EARTHS CORE

515

ACKNOWLEDGMENTS This article has beneted greatly from discussions with Weijia Kuang and Stephen Zatman, though any errors are entirely of my making. This work was supported by the National Aeronautics and Space Administration and the National Science Foundation.
Visit the Annual Reviews home page at http://www.AnnualReviews.org.

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Literature Cited Anufriev AP, Braginsky SI. 1975a. Effect of a magnetic eld on the stream of liquid rotating at a rough surface. Geomagn. Aeron. 11:461 67 Anufriev AP, Braginsky SI. 1975b. Inuence of irregularities of the boundary of the Earths core on the velocity of the liquid and on the magnetic eld. Geomagn. Aeron 15:75457 Anufriev AP, Braginsky SI. 1977a. Inuence of irregularities of the boundary of the Earths core on the velocity of the liquid and on the magnetic eld. II. Geomagn. Aeron. 17:78 82 Anufriev AP, Braginsky SI. 1977b. Effect of irregularities of the boundary of the Earths core on the speed of the uid and on the magnetic eld. III. Geomagn. Aeron. 17:49296 Aurnou JM, Brito D, Olson PL. 1996. Mechanics of inner core super-rotation. Geophys. Res. Lett. 23:34014 Bloxham J, Jackson A. 1991. Fluid ow near the surface of Earths outer core. Rev. Geophys. 29:97120 Braginsky SI. 1963. Structure of the F layer and reasons for convection in the Earths core. Sov. Phys. Dokl 149:810 Braginsky SI. 1970. Torsional magnetohydrodynamic vibrations in the Earths core and variations in day length. Geomagn. Aeron. 10:18 Braginsky SI. 1975. An almost axially symmetrical model of the hydromagnetic dynamo of the Earth, I. Geomagn. Aeron. 15:14956 Braginsky SI, Roberts PH. 1995. Equations governing convection in Earths core and the geodynamo. Geophys. Astrophys. Fluid Dyn. 79:197 Buffett BA. 1996a. Gravitational oscillations in the length-of-day. Geophys. Res. Lett. 23:227982 Buffett BA. 1996b. A mechanism for decade uctuations in the length-of-day. Geophys. Res. Lett. 23:38036 Buffett BA. 1997. Geodynamic estimates of the viscosity of the Earths inner core. Nature 388:57173 Bullard EC, Freedman C, Gellman H, Nixon J. 1950. The westward drift of the Earths magnetic eld. Philos. Trans. R. Soc. London Ser. A 243:6792 Christodoulidis DC, Smith DE, Williamson RG, Klosko SM. 1988. Observed tidal breaking in the Earth/Moon/Sun system. J. Geophys. Res. 93:621636 Creager KC. 1997. Inner core rotation from small-scale heterogeneity and time varying travel times. Science. Submitted Dickman SR, Nam YS. 1995. Revised predictions of long-period ocean tidal effects on Earths rotation rate. J. Geophys. Res. 100:823343 Fearn DR, Roberts PH, Soward AM. 1988. Convection, stability and the dynamo. In Energy, Stability and Convection, ed. P Galdi, B Straughan, pp. 60324. Harlow, Engl.: Longman Glatzmaier GA, Roberts PH. 1995a. A threedimensional convective dynamo solution with rotating and nitely conducting inner core and mantle. Phys. Earth Planet. Int. 91:6375 Glatzmaier GA, Roberts PH. 1995b. A threedimensional self-consistent computer simulation of a geomagnetic eld reversal. Nature 377:2039 Glatzmaier GA, Roberts PH. 1996a. An anelastic evolutionary geodynamo simulation driven by compositional and thermal convection. Physica D 97:8194 Glatzmaier GA, Roberts PH. 1996b. Rotation and magnetism of Earths inner core. Science 274:188790 Gubbins D. 1981. Rotation of the inner core. J. Geophys. Res. 86:1169599 Gubbins D, Masters TG, Jacobs JA. 1979. Thermal evolution of the Earths core. Geophys. J. R. Astron. Soc. 59:5799 Hide R. 1969. Interaction between the Earths liquid core and solid mantle. Nature 222:105556 Hide R. 1986. The Earths differential rotation. Q. J. R. Astron. Soc. 278:314

516

BLOXHAM
in a rapidly rotating spherical shell I: weak and strong eld dynamo action. J. Comp. Phys. Submitted Kuang W, Bloxham J. 1997c. An Earth-like numerical dynamo model. Nature 389:371 74 Kuang W, Bloxham J. 1997d. Numerical dynamo modeling: comparison with the Earths magnetic eld. In The Core-Mantle Boundary Region, ed. M Gurnis, M Wysession, E Knittle, B Buffett, Geophys. Monogr., submitted. Washington, DC: Am. Geophys. Union. Love JJ, Bloxham J. 1994. Electromagnetic coupling and the toroidal magnetic eld at the core-mantle boundary. Geophys. J. Int. 117:23556 Moffatt HK. 1978. Topographic coupling at the core-mantle interface. Geophys. Astrophys. Fluid Dyn. 9:27888 Moffatt HK, Dillon RF. 1976. The correlation between gravitational and geomagnetic elds caused by the interaction of the core uid motion with a bumpy core-mantle interface. Phys. Earth Planet. Int. 13:6778 Munk WH, MacDonald GJF. 1960. The Rotation of the Earth. Cambridge: Cambridge Univ. Press Munk WH, Revelle R. 1952. On the geophysical interpretation of irregularities in the rotation of the Earth. MNRAS Geophys. Suppl. 6:331 47 Roberts PH, Soward AM. 1992. Dynamo theory. Annu. Rev. Fluid Mech. 24:459512 Rochester MG. 1960. Geomagnetic westward drift and irregularities in the Earths rotation. Philos. Trans. R. Soc. London Ser. A 252:531 55 Rosen RD. 1993. The axial momentum balance of Earth and its uid envelope. Surv. Geophys. 14:129 Shankland TJ, Peyronneau J, Poirier J-P. 1993. Electrical conductivity of the Earths lower mantle. Nature 366:45355 Song X, Richards PG. 1996. Seismological evidence for differential rotation of the Earths inner core. Nature 382:22124 Stacey FD. 1992. Physics of the Earth. Brisbane: Brookeld Starr V. 1948. An essay on the general circulation of the Earths atmosphere. J. Meteorol. 5:3943 Stephenson FR, Morrison LV. 1995. Long-term uctuations in the Earths rotation: 700 BC to AD 1990. Philos. Trans. R. Soc. London Ser. A 351:165202 Su W, Dziewonski AM, Jeanloz R. 1996. Planet within a planet: rotation of the inner core of Earth. Science 274:188387 Taylor JB. 1963. The magnetohydrodynamics of a rotating uid and the Earths dynamo

Hide R, Birch NT, Morrison LV, Shea DJ, White AA. 1980. Atmospheric angular momentum uctuations and changes in the length of day. Nature 286:11417 Hide R, Clayton RW, Hager BH, Spieth MA, Voorhies CV. 1993. Topographic core-mantle coupling and uctuations in the Earths rotation. In Relating Geophysical Structure and Processes, The Jeffreys Volume, ed. K Aki, R Dmowska, Geophys. Monogr., 76:10720. Am. Geophys. Union Hollerbach R. 1996. On the theory of the geodynamo. Phys. Earth Planet. Int. 98:16385 Holme R. 1998. Electromagnetic core-mantle coupling, 1: explaining decadal changes in the length of day. Geophys. J. Int. 132:167 80 Jackson A, Bloxham J, Gubbins D. 1993. Timedependent ow at the core surface and conservation of angular momentum in the coupled core-mantle system. In Dynamics of the Earths Deep Interior and Earth Rotation, ed. J-L LeMou el, D Smylie, T Herring, Geophys. Monogr., 72:97107. Washington, DC: Am. Geophys. Union Jacobs JA. 1987. The Earths Core. Cambridge: Cambridge Univ. Press Jault D. 1995. Model Z by computation and Taylors condition. Geophys. Astrophys. Fluid Dyn. 79:9924 Jault D, Gire C, LeMou el J-L. 1988. Westward drift, core motions and exchanges of angular momentum between core and mantle. Nature 333:35356 Jault D, LeMou el J-L. 1989. The topographic torque associated with a tangentially geostrophic motion at the core surface and inferences on the ow inside the core. Geophys. Astrophys. Fluid Dyn. 48:27396 Jault D, LeMou el J-L. 1990. Core-mantle boundary shape: constraints inferred from the pressure torque acting between core and mantle. Geophys. J. Int. 101:23341 Jault D, LeMou el J-L. 1991. Exchange of angular momentum between the core and mantle. J. Geomagn. Geoelectr. 43:11129 Jeanloz R. 1990. The nature of the Earths core. Annu. Rev. Earth Planet. Sci. 18:35786 Jones CA, Longbottom AW, Hollerbach R. 1995. A self-consistent convection driven geodynamo model using a mean eld approximation. Phys. Earth Planet. Int. 92:11941 Kuang W, Bloxham J. 1993. On the effect of boundary topography on ow in the Earths core. Geophys. Astrophys. Fluid Dyn. 72:16195 Kuang W, Bloxham J. 1997a. On the dynamics of the topographical core-mantle coupling. Phys. Earth Planet. Int. 99:28994 Kuang W, Bloxham J. 1997b. Numerical modelling of magnetohydrodynamic convection

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

DYNAMICS OF THE EARTHS CORE


problem. Proc. R. Soc. London A 274:274 83 Verhoogen J. 1961. Heat balance in the Earths core. Geophys. J. 4:27681 Zatman S, Bloxham J. 1997a. The phase difference between length of day and atmospheric angular momentum at subannual frequencies and the possible role of core-mantle coupling.

517

Geophys. Res. Lett. 24:1799802 Zatman S, Bloxham J. 1997b. Torsional oscillations and the magnetic eld within the Earths core. Nature 388:76063 Zatman S, Bloxham J. 1997c. A one dimensional map of | Bs | from torsional oscillations of the Earths core. Am. Geophys. Union Monogr. Submitted

Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

Annual Review of Earth and Planetary Science Volume 26, 1998

CONTENTS
Contemplation of Things Past, George W. Wetherill Volcanism and Tectonics on Venus, F. Nimmo, D. McKenzie TEMPERATURES IN PROTOPLANETARY DISKS, Alan P. Boss THE IMPORTANCE OF PAHOEHOE, S. Self, L. Keszthelyi, Th. Thordarson CHINESE LOESS AND THE PALEOMONSOON, Tungsheng Liu, Zhongli Ding
Annu. Rev. Earth Planet. Sci. 1998.26:501-517. Downloaded from arjournals.annualreviews.org by UNIVERSITY OF ALBERTA on 01/30/08. For personal use only.

1 23 53 81 111

STELLAR NUCLEOSYNTHESIS AND THE ISOTOPIC COMPOSITION OF PRESOLAR GRAINS FROM PRIMITIVE METEORITES, Ernst Zinner NOBLE GASES IN THE EARTH'S MANTLE, K. A. Farley, E. Neroda SATELLITE ALTIMETRY, THE MARINE GEOID, AND THE OCEANIC GENERAL CIRCULATION, Carl Wunsch, Detlef Stammer CHEMICALLY REACTIVE FLUID FLOW DURING METAMORPHISM, John M. Ferry, Martha L. Gerdes CHANNEL NETWORKS, Andrea Rinaldo, Ignacio Rodriguez-Iturbe, Riccardo Rigon EARLY HISTORY OF ARTHROPOD AND VASCULAR PLANT ASSOCIATIONS, Conrad C. Labandeira Ecological Aspects of the Cretaceous Flowering Plant Radiation, Scott L. Wing, Lisa D. Boucher The Re-Os Isotope System in Cosmochemistry and High-Temperature Geochemistry, Steven B. Shirey, Richard J. Walker DYNAMICS OF ANGULAR MOMENTUM IN THE EARTH'S CORE, Jeremy Bloxham FISSION TRACK ANALYSIS AND ITS APPLICATIONS TO GEOLOGICAL PROBLEMS, Kerry Gallagher, Roderick Brown, Christopher Johnson Isotopic Reconstruction of the Past Continental Environments, Paul L. Koch The Plate Tectonic Approximation: Plate Nonrigidity, Diffuse Plate Boundaries, and Global Plate Reconstructions, Richard G. Gordon LABORATORY-DERIVED FRICTION LAWS AND THEIR APPLICATION TO SEISMIC FAULTING, Chris Marone Seafloor Tectonic Fabric by Satellite Altimetry, Walter H. F. Smith

147 189 219 255 289 329

379

423 501

519

573 615 643 697

You might also like