[SIT23-P07] Radial anisotropy in the upper mantle using multi-mode surface waves incorporating the ηκ parameter: Application to the Pacific and Australian regions
Keywords:radial anisotropy, surface waves, inversion
Radial anisotropy can be described by five elastic parameters; four parameters related to seismic wave speeds (βv, βh, αh, αv) and an additional fifth anisotropic parameter (η). One of the anisotropic parameters, η, was originally defined by Anderson (1961), but its physical proprieties have been rather unclear compared with other four parameters related to elastic velocity.
The re-defined fifth anisotropic parameter ηκ by Kawakatsu et al. (2015) makes it easier to understand its physical proprieties compared with conventional η. The introduction of ηκ affects the shape of sensitivity kernels of Rayleigh wave phase speeds with respect to ηκ, PH-wave speeds αh and PV-wave speeds αv (Kawakatsu, 2016b). Since the sensitivity of Rayleigh wave phase speeds to ηκ becomes greater than that to η, we can expect a possibility of resolving the ηκ parameter. However, since the inverse correlation between the sensitivity kernels of SV-wave speed βv and ηκ becomes rather stronger, the trade-off between βv and may easily occur. Therefore, the independent determination of ηκ parameter from the inversions of surface-wave dispersion curves is not a straightforward issue.
In this study, incorporating ηκ, we performed inversions for five elastic parameters in the upper mantle, employing a classical iterativenonlinear least-squares inversion method (Tarantola and Valette, 1982). We used multi-mode dispersion data sets of surface waves in the Australian region (Yoshizawa, 2014, PEPI) and Pacific (Isse et al., 2019, EPSL) to reconstruct 3-D radially anisotropic model. In these inversions, we used isotropic PREM as a starting model to avoid the influence of reference anisotropy model.
The retrieved spatial distribution of the ηκ parameter can be mostly explained by trade-offs with SV wave speed βvat many locations in Australia and Pacific. Still, we can find structural features in our 3-D βv and ηκ models, particularly in some major structural boundaries such as ridges and the Tasman Line in eastern Australia, which cannot simply be attributed to the trade-off between βv and ηκ, indicating possible intrinsic character of ηκ in the upper mantle.
The re-defined fifth anisotropic parameter ηκ by Kawakatsu et al. (2015) makes it easier to understand its physical proprieties compared with conventional η. The introduction of ηκ affects the shape of sensitivity kernels of Rayleigh wave phase speeds with respect to ηκ, PH-wave speeds αh and PV-wave speeds αv (Kawakatsu, 2016b). Since the sensitivity of Rayleigh wave phase speeds to ηκ becomes greater than that to η, we can expect a possibility of resolving the ηκ parameter. However, since the inverse correlation between the sensitivity kernels of SV-wave speed βv and ηκ becomes rather stronger, the trade-off between βv and may easily occur. Therefore, the independent determination of ηκ parameter from the inversions of surface-wave dispersion curves is not a straightforward issue.
In this study, incorporating ηκ, we performed inversions for five elastic parameters in the upper mantle, employing a classical iterativenonlinear least-squares inversion method (Tarantola and Valette, 1982). We used multi-mode dispersion data sets of surface waves in the Australian region (Yoshizawa, 2014, PEPI) and Pacific (Isse et al., 2019, EPSL) to reconstruct 3-D radially anisotropic model. In these inversions, we used isotropic PREM as a starting model to avoid the influence of reference anisotropy model.
The retrieved spatial distribution of the ηκ parameter can be mostly explained by trade-offs with SV wave speed βvat many locations in Australia and Pacific. Still, we can find structural features in our 3-D βv and ηκ models, particularly in some major structural boundaries such as ridges and the Tasman Line in eastern Australia, which cannot simply be attributed to the trade-off between βv and ηκ, indicating possible intrinsic character of ηκ in the upper mantle.