Seismic anisotropy of the lithosphere/asthenosphere system beneath the Rwenzori region of the Albertine Rift

Abstract

Shear-wave splitting measurements from local and teleseismic earthquakes are used to investigate the seismic anisotropy in the upper mantle beneath the Rwenzori region of the East African Rift system. At most stations, shear-wave splitting parameters obtained from individual earthquakes exhibit only minor variations with backazimuth. We therefore employ a joint inversion of SKS waveforms to derive hypothetical one-layer parameters. The corresponding fast polarizations are generally rift parallel and the average delay time is about 1 s. Shear phases from local events within the crust are characterized by an average delay time of 0.04 s. Delay times from local mantle earthquakes are in the range of 0.2 s. This observation suggests that the dominant source region for seismic anisotropy beneath the rift is located within the mantle. We use finite-frequency waveform modeling to test different models of anisotropy within the lithosphere/asthenosphere system of the rift. The results show that the rift-parallel fast polarizations are consistent with horizontal transverse isotropy (HTI anisotropy) caused by rift-parallel magmatic intrusions or lenses located within the lithospheric mantle—as it would be expected during the early stages of continental rifting. Furthermore, the short-scale spatial variations in the fast polarizations observed in the southern part of the study area can be explained by effects due to sedimentary basins of low isotropic velocity in combination with a shift in the orientation of anisotropic fabrics in the upper mantle. A uniform anisotropic layer in relation to large-scale asthenospheric mantle flow is less consistent with the observed splitting parameters.

Appendix

The numerical simulations of seismic wave fields presented in this paper are based on the heterogeneous wave equation

$$\rho \partial_{t}^{2} u_{i} = \partial_{{x_{j} }} \left( {C_{ijkl} \delta_{{x_{l} }} u_{k} } \right),$$

where i, j, k, l = 1, …, 3, ρ denotes the density, u _i is the displacement and C _ijkl is the elastic tensor. Changes in material properties are restricted to the vertical $x_{1} ,x_{3}$ coordinate plane in a reduced 2-D Cartesian geometry. This implies that any spatial derivative $\delta_{{x_{2} }}$ vanishes. To discretize the wave equation on a grid, the finite-difference method is used up to the second-order approximation. The procedure is identical to Ryberg et al. (2002). An equidistant grid spacing of ∆x = ∆z = 0.25 km is assumed. A sampling rate of ∆t = 0.0175 s was found to be suitable to insure a robust discretization. For the comparison with SKS phases, a period of T = 8 s is used. As side effects are to be neglected, the grid size is chosen much larger than the evaluated model area.

In this work, anisotropy due to shape-preferred orientation of isotropic material is assumed. The elastic tensor of the effective anisotropic medium can be described by five independent elastic constants. The matrix of elastic constants for a horizontally transverse isotropic medium (HTI) is given by Ikelle and Amundsen (2005)

$$c_{ij} = \left( {\begin{array}{*{20}c} {c_{11} } & {c_{13} } & {c_{13} } & 0 & 0 & 0 \\ {c_{13} } & {c_{33} } & {c_{33} - 2c_{44} } & 0 & 0 & 0 \\ {c_{13} } & {c_{33} - 2c_{44} } & {c_{33} } & 0 & 0 & 0 \\ 0 & 0 & 0 & {c_{44} } & 0 & 0 \\ 0 & 0 & 0 & 0 & {c_{66} } & 0 \\ 0 & 0 & 0 & 0 & 0 & {c_{66} } \\ \end{array} } \right).$$

The effective elastic constants of the HTI-medium are derived by an analytical solution (Ikelle and Amundsen 2005).