A new method to test shear wave splitting: improving statistical assessment of splitting parameters
Date
2016
Authors
Corbalán Castejón, Ana, author
Schutt, Derek, advisor
Breidt, Jay, committee member
Aster, Richard, committee member
Egenhoff, Sven, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Shear wave splitting has proved to be a very useful technique to probe for seismic anisotropy in the earth’s interior, and measurements of seismic anisotropy are perhaps the best way to constrain the strain history of the lithosphere and asthenosphere. However, existent methods of shear wave splitting analysis do not estimate uncertainty correctly, and do not allow for careful statistical modeling of anisotropy and uncertainty in complex scenarios. Consequently, the interpretation of shear wave splitting measurements has an undesirable subjective component. This study illustrates a new method to characterize shear wave splitting and the associated uncertainty based on the cross-convolution method [Menke and Levin, 2003]. This new method has been tested on synthetic data and benchmarked with data from the Pasadena, California seismic station (PAS). Synthetic tests show that the method can successfully obtain the splitting parameters from observed split shear waves. PAS results are very reasonable and consistent with previous studies [Liu et al., 1995; Özalaybey and Savage, 1995; Polet and Kanamori, 2002]. As presented, the Menke and Levin [2003] method does not explicitly model the errors. Our method works on noisy data without any particular need for processing, it fully accounts for correlation structures on the noise, and it models the errors with a proper bootstrapping approach. Hence, the method presented here casts the analysis of shear wave splitting into a more formal statistical context, allowing for formal hypothesis testing and more nuanced interpretation of seismic anisotropy results.