A radial anisotropy model of the upper mantle from surface wave observations

Abstract:

Radial anisotropy within the upper mantle was first encountered in the early 1960's based on studies using surface waves. A disrepancy between the Love and Rayleigh wave data was observed which could not be explained using a simple isotropic model. This was later reconconciled using a transversely isotropic model now assumed in many modern day anisotropic models such as SAW642AN and S362ANI. Radial anisotropy is attributed to the lattice preferred orientation (LPO) of the anisotropic crystals believed to be organised by the flow in the upper mantle. These models are therefore important for analysing the geometry of the flow and the deformation of the mantle. Surface wave observations offer a unique way of studying the radial anisotropy. Rayleigh waves are sensitive to the vertical shear velocity (SV) and Love waves are sensitive to the horizontal shear velocity (SH). The ratio of these give the radial anisotropy parameter ξ. Although radial anisotropy models exist, they are usually limited to the fundamental mode measurments with poor path coverage due to the noise on the horizontal components. Higher mode Love wave measurements are difficult for oceanic paths. This is because group velocities of the higher modes are similar to the fundamental mode between a period range of 50-100 s. These therefore arrive and interferre with each other. The higher mode information therefore cannot be extracted easily. We modify the method of Debayle and Ricard 2012 which allow the extraction of information up to the 5th overtone by mimicking the interferrence from the arrival of the fundamental and higher modes. Synthetic tests show an excellent recovery of the fundamental and higher mode information from the Love waves. The inclusion of the the higher modes greatly increases the resolution to the SH velocity and increases the sampling to deeper structures. We apply this modified method to a large dataset and construct an SH model. This is then combined with an equivalent SV model for the radial anisotropy model which we present here with added higher mode information and improved path coverage compared to recent global models.