The solid inner core of the Earth, a sphere of iron alloy with a radius of 1220 km (one-third the size of Earth's liquid core), has attracted the curiosity of a large community of scientists in various disciplines for at least a decade. One of the inner core's most remarkable properties is its anisotropy: Seismic waves propagate faster parallel to Earth's rotation axis than perpendicular to it. This 3 to 4% anisotropy in seismic velocities is probably due to the prevailing orientation of iron crystals that constitute the inner core, but the mechanism responsible for this orientation remains uncertain: Magnetic and dynamical processes have both been invoked.
Recently, three seismological studies (1-3) have suggested that the inner core rotates faster than the mantle, immediately provoking reactions from supporters and detractors alike. The search for this differential rotation was motivated by the predictions of some theoretical models of the dynamo that generates Earth's magnetic field inside the liquid core. However, the differential rotation rates inferred from seismology vary considerably among the studies, depending on the hypotheses and methods of detection. The problem is far from simple.
Song and Richards (1) and Su et al. (2) base their analyses on the assumption that the inner core exhibits a nearly cylindrical symmetry about an axis that is slightly tilted with respect to Earth's rotation axis (top part of figure). If the inner core does not rotate at the same speed as the mantle, it would then be possible to observe it from the apparent wobble of its symmetry axis. Song and Richards detected this wobble by studying seismic waves traveling from South Sandwich Island in the South Atlantic Ocean along a polar path to a seismometer in Alaska, whereas Su et al. considered the large, worldwide data set provided by the catalogs of the International Seismological Centre. The former found a rotation for the inner core that is 1.1º/year faster than that for the mantle over the 30 years of their observations, whereas the latter obtained a 3º/year faster rotation over the same interval. Yet both of these results raise some tough questions. In particular, the tilt of the symmetry axis is uncertain (4). It appears to be probably an artifact due to the uneven sampling of Earth by the seismic waves: Earthquakes occur mostly along subduction zones and mid-oceanic ridges, whereas stations are located mostly on continents. In addition, the uneven distribution of earthquakes in time may result in different axis positions when estimated over short time intervals. The anisotropy itself exhibits important departures from cylindrical symmetry or is superimposed on heterogeneities (5).
A more promising approach is to try to detect an inner core heterogeneity or anisotropy variation as it passes across a particular ray path, as attempted by Creager (3) (lower part of figure). Using South Sandwich Island events recorded at a network in Alaska, he first established a map of inner core anomalies in the region sampled by the rays and then used it to infer the inner core rotation rate from a specific path, where observations were available over a period of 30 years. He estimates values of 0.2º to 0.3º/year, but very low rates close to synchronous rotation are also possible, depending on how much of the signal is attributed to mantle heterogeneity.
Mantle heterogeneity may indeed introduce serious biases. For the path from South Sandwich Island to Alaska, the sources and receivers lie in the vicinity of subduction zones. Perturbing effects attributable to these structures are partly removed when the ray propagating through the inner core is compared with a nearby ray that stays inside the homogeneous liquid core, but residual effects still linger. It is thus hard to reach the extreme precision (0.3 s) required for detection. The effect of subduction zones is quite clear from the examination of the complete seismograms: For two events very close to each other, even phases unaffected by the inner core may exhibit substantial differences in travel times (6). Mislocating the earthquake hypocenter can also bias the travel times, in particular for the oldest events in the Southern Hemisphere, where the nearest station can be more than 2000 km away. For another polar path with simpler structures beneath source and station as well as accurate source locations--the Novaya Zemlya nuclear tests recorded in Antarctica--no travel time anomaly is observed over 25 years (7). Detecting such a subtle signal by seismology is likely to remain controversial.
Yet, the most serious problem with the differential rotation lies in its geodynamical consequences (8). Because Earth's mantle contains density heterogeneities, the resulting gravity field distorts the surface of the inner core, producing undulations on the order of 100 m in amplitude. This gravitational coupling is probably much larger than the magnetic coupling. If the inner core does not rotate at the same velocity as the mantle, its shape adjusts continually to the mantle-induced gravity field. This adjustment could occur by melting and solidification, because the iron at the inner core surface is close to its melting point. Or it could occur by viscous relaxation, if the inner core deforms easily. But in this case, one may question how the heterogeneities or anisotropy variations observed by seismologists could persist in the inner core.
Will seismology be able to resolve this debate? Inner core rotation, if it exists, induces only slight travel time perturbations over the interval spanned by instrumental seismology. The key then is to analyze many data series over long time intervals. Much future research lies in the past. Just as long-term recordings in astronomical observatories detected the irregularities of mantle rotation, old records gathered over past decades in seismological observatories are essential for detecting inner core differential rotation. No definitive conclusion about inner core rotation will be drawn until a large number of these data can be processed, a task made difficult by the limited number of stations having run continuously for 30 years and having accessible archives.
Right now, the differential rotation of the inner core is not yet firmly established. Even if today's results are contradicted tomorrow, they represent an important step for deep-Earth science, because they have already provoked exciting questions in geomagnetism, seismology, and geodynamics.
References and Notes