Mantle Anisotropy beneath Southern Africa from Shear-Wave Splitting: Preliminary Results from the Southern Africa Seismic Experiment


We present preliminary results on mantle anisotropy from shear-wave splitting measurements. Data were obtained from 55 stations of the Southern African Seismic Experiment, which have been deployed since April, 1997. We have obtained splitting parameters, the fast polarization direction φ and delay time δ using SKS and SKKS arrivals from about 100 events. Splitting has been detected at most stations, and there are systematic spatial variations in both splitting parameters which provide constraints on the orientation and layer thickness of mantle deformation, respectively. The data may be separated into three geologic regions: the southern Kaapvaal group with values of φ trending NE-SW, a second group further north within the Limpopo belt with directions more nearly EW, and group of stations on the Zimbabwe craton with directions ranging from NE-SW to NNE-SSW. In all three regions, values of φ follow closely the orientation of Archean orogenic structures, suggesting that the crust and mantle deform coherently during these ancient events, and that this mantle deformation is preserved as fossil anisotropy (Silver, 1996). Splitting delay times are small, averaging 0.65s, compared to a global average of about 1s for continents worldwide. The simplest explanation is a single layer that is either thin (about 60km) or possesses weak (about 2%) intrinsic anisotropy. Another possibility is a vertically heterogeneous medium, with the fast symmetry axis randomly distributed in the horizontal plane. This second model would still generate signifiant polarization anisotropy in surface waves, as proposed by Jordan et al (1998) for this region. Independent constraints on intrinsic anisotropy beneath southern Africa are available from the the petrofabric analyses of mantle nodules, which predict weak splitting anisotropy of about 2% for vertically propagating shear waves, but stronger anisotropy for horizontally propagating waves, given the most likely orientation of the nodules (Mainprice et al, 1998). Thus the two anisotropic data sets can be reconciled within the context of a single homogeneous layer. The average 0.65s splitting delay time corresponds to an effective mantle layer thickness of about 150km, so that the layer would extend to a depth of 190km, assuming the crust is isotropic. Finally, there is a systematic latitudinal dependence in the splitting delay times, which decrease as one approaches the latitude of -25⁰, both from the north and south. At this latitude, where splitting is undetectable, is found the 2.05 Ga Bushveld Complex, the largest layered igneous intrusion in the world. This association suggests that the magmatic process that formed the Bushveld, weakened the mantle anisotropic fabric, probably through recrystalization. If confirmed, this will constitute the clearest case of the pervasive altering of mantle crystal structure by magmatic processes, and may provide a key to understanding other regions that may have been subjected to the same process.

Meeting Name

AGU Fall Meeting (1998: Dec. 1, San Francisco, CA)


Geosciences and Geological and Petroleum Engineering

Document Type

Article - Conference proceedings

Document Version


File Type





© 1998 American Geophysical Union (AGU), All rights reserved.

This document is currently not available here.