INTERPRETATION OF SKS-WAVES USING SAMPLES FROM THE SUBCONTINENTAL LITHOSPHERE

The seismic properties of five South African kimberlite nodules have been calculated from petrofabric measurements. These garnet lherzolite and harzburgite composition nodules (56-80% olivine) are known from previous studies to have originated at depths of 120-170 km in the subcontinental lithosphere. Their in situ seismic properties have been calculated by extrapolating the elastic constants to the appropriate temperature (900-1050-degrees-C) and pressure (3.0-3.5 GPa) conditions. The average of the five samples has a maximum S-wave anisotropy (deltaV(s)) of 3.7% which is sufficiently high to explain previously reported teleseismic SKS delay times (deltat) in continental shield areas of up to 1.7 s. The biggest delay time is observed when the foliation is vertical, the lineation horizontal, and the fastest S-wave polarized parallel to the lineation, i.e. an orientation for transcurrent motion. Delay times of over 1 s can only be generated by this orientation. A strong mantle-crust mechanical coupling is suggested for such situations. The potential use of seismic anisotropy as an indicator of strain in the lithosphere has been investigated using experimental and simulated fabric data for olivine, the most abundant phase (70%) in the upper mantle. The V(p) and V(s) seismic anisotropy increases with fabric strength and finite strain. Recrystallization tends to reduce fabric strength and seismic anisotropy, resulting in saturation values for experimental and simulated fabrics which correspond to approximately 50-60% strain or 8-9% deltaV(s) for pure olivine aggregates. A survey of 25 naturally deformed peridotites of oceanic origin suggests an average maximum S-wave anisotropy for the olivine component of 9%, or about 8% for a lherzolite or harzburgite rock composition when the orthopyroxene component is taken into account. The ophiolite samples are twice as anisotropic as the kimberlite nodules. If an average anisotropy value is representative of the lithosphere, then the SKS delay times represent variations in anisotropic layer thickness for delay times over 1 s as the orientation is constrained to be foliation vertical and lineation horizontal. The magnitude of S-wave anisotropy is less sensitive than V(p) to variations in olivine volume fraction in range 50-100%, and to deformation or fabric intensity, further suggesting that S-wave anisotropy is particularly apt for the determination of the anisotropic lithosphere thickness. For smaller delay times there is some trade off between structural orientation and thickness.