Effects of Isotropic and Anisotropic Structure in the Lowermost Mantle on High-Frequency Body Waveforms

Abstract:

It has been observed that vertically (SV) and horizontally (SH) polarised S waves crossing the lowermost mantle sometimes are split by a few seconds The splitting of such waves is often interpreted in terms of seismic anisotropy in the D” region. Here we investigate systematically the effects of elastic, anelastic, isotropic and anisotropic structure on shear-wave splitting, including 3-D variations in some of these physical properties. Taking advantage of accurate waveform modeling techniques such as Gemini and the Spectral Element Method we generate three-component theoretical waveforms in a wide set of 1-D and 3-D, isotropic and radially anisotropic earth models, accurate down to a wave period of T~5.6s. Our numerical simulations in isotropic earth models show that the contamination of S waves by other phases can generate an apparent splitting between SH and SV waves. In particular, in the case of very shallow sources, the sS phase can interfere with the direct S phase, resulting in split SH and SV pulses when the SH and SV (or sSH and sSV) waves have different polarity or a substantial amplitude difference. In the case of deep earthquake sources, a positive shear velocity jump at the top of the D” can cause the triplication of S waves and the ScSH and ScSV phases can have different polarity. Thus, when the triplicated S wave is combined with the ScS phase, the resulting SH-ScSH and SV-ScSV phases may seem split. On the other hand, in the absence of a sharp vertical variation in the shear wave velocity, the difference in polarity between ScSH and ScSV can make the SH pulse larger than SV and thus also lead to apparent splitting between these phases. This effect depends on the thickness of the D” and the Vs gradient within it. S waveforms simulated in radially anisotropic models reveal that a radial anisotropy of ξ=1.07 in the D” seems to be necessary to explain the 2-3s of splitting observed in waveforms recorded in Tanzania from an event in the Banda Sea. However, our analysis also shows that other factors such as sharp vertical variations at the top of D” and gradients of Vs and η within the D’’ may also affect the observed waveforms. This study suggests that caution should be taken when interpreting SH-SV splitting of deep mantle body waves exclusively in terms of anisotropy in the lowermost mantle.