Classification of Teleseismic Shear Wave Splitting Measurements: A Convolutional Neural Network Approach

Abstract

It has long been recognized that the P-to-s converted phases from the core-mantle boundary such as SKS, PKS, and SKKS (hereafter referred to as XKS) split into orthogonally polarized fast and slow components in azimuthally anisotropic media (Ando et al., 1983;Long & Silver, 2009;Savage, 1999;Silver & Chan, 1991). The two splitting parameters, the polarization orientation of the fast component (fast orientation or ϕ) and the time separation (splitting time or δt) between the two waves, reveal the orientation and splitting magnitude of the anisotropy, respectively. Over the past several decades, the shear wave splitting (SWS) analysis technique has been widely used to delineate azimuthal anisotropy in the upper mantle, where lattice preferred orientation of crystallographic axes of main constitute minerals such as olivine is the dominant cause of the observed anisotropy (Katayama &

Shear Wave Splitting Analysis using Deep Learning (SWSNet)

“…In this paper, we present a novel Deep Learning-based approach for the analysis of shear-wave splitting. In a recent study, Zhang and Gao (2022) utilized an Convolutional Neural Network (CNN) for waveform classification to automatically select reliable SWS measurements. However, a comprehensive analysis to infer anisotropic splitting parameters has not yet been presented.…”

Shear Wave Splitting Analysis using Deep Learning (SWSNet)

“…In this paper, we present a novel Deep Learning-based approach for the analysis of shear-wave splitting. In a recent study, Zhang and Gao (2022) utilized an Convolutional Neural Network (CNN) for waveform classification to automatically select reliable SWS measurements. However, a comprehensive analysis to infer anisotropic splitting parameters has not yet been presented.…”