Modeling shear wave splitting observations: Long-term data have the better idea

Abstract Seismic anisotropy provides a direct link to present and past geological and geodynamic processes in the Earth’s mantle. Measuring shear wave splitting (SWS) is a widely used and powerful tool to detect and quantify anisotropy. Therefore, the two splitting parameters, the fast polarization direction Φ and the delay time δt, are measured, and the splitting intensity is calculated. For studying anisotropy in the mantle SKS/SKKS/PKS and S/ScS phases are used. Especially phase pairs can be useful; e.g., SKS-SKKKS pairs help to recognize lowermost mantle anisotropy. The splitting observations are grouped into nulls (no SWS) and splits (SWS). Most studies focus on the splits in the modeling and interpretation but ignore the nulls. However, nulls contain crucial information on the anisotropy and should be considered in the interpretation.The majority of upper mantle anisotropy studies interpret the splitting observations assuming one single horizontal anisotropic layer (simple anisotropy) without separate modeling. However, several studies show clear backazimuthal variations of the splitting parameters, which are clear indications of more complicated anisotropy scenarios (complex anisotropy). In this case the observed splitting parameters do not directly represent the anisotropy in the subsurface, but they are apparent or effective values, and a separate modeling procedure is needed. Anisotropy scenarios with two layers with horizontal symmetry axes as well as one layer with tilted symmetry axes show characteristic backazimuthal variations. Long-term data are needed to gain observations from many BAZ directions to be able to recognize backazimuthal variations of the splitting parameters, including splits and nulls as well as observations on different phases. Ignoring or missing backazimuthal variations leads to wrong anisotropy models and interpretations. This talk focuses on backzimuthal variations of splitting parameters and why long-term data are essential for a reliable modeling and interpretation. Additional materials Animations shown on slides 7, 8, 9, 14, 15, 16: https://doi.org/10.5281/zenodo.15641348 Poster mentioned on slide 18 "Next steps": https://doi.org/10.5281/zenodo.18524083