Layer stripping of shear-wave splitting in marine PS waves

• Published in: Geophysical Journal International
• Main Authors: Haacke, R. Ross, Westbrook, Graham K., Peacock, Sheila
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: 2009
• Subjects: Svalbard; Alford
• Online Access: https://eprints.soton.ac.uk/340093/

The properties of split S waves can be used to infer: (1) the state of stress and strain in the Earth; (2) the directional dependence of hydraulic conductivity and (3) small changes in pore-fluid pressure in the rock mass that occur in response to dynamic processes, such as the earthquake cycle. Measurements of split S waves are particularly useful in shallow (<1000 m subseabed) marine sediments, where S-wave splitting from an azimuthal elastic anisotropy is typically produced by the presence of near-vertical aligned cracks. Here we present a method of measuring small amounts of S-wave splitting in marine P-to-S mode-converted waves, and illustrate the technique with data from an ocean-bottom seismometer (OBS) deployed on the west Svalbard continental slope. The analysis applies a modified version of the Alford rotation and layer-stripping technique developed for zero-offset S-wave sources and treats PS waves that undergo mode conversion at reflectors that are close to the seabed in comparison with the overlying water depth. When the seismic record contains coherent signal on both the in-plane and out-of-plane components, the layer-stripping technique is capable of decoupling the S-wave splitting from the effects of P-wave velocity anisotropy and reflector dip that influence the downgoing, P wave, part of the ray path. The amount of S-wave splitting in the data is small, however, and we find that this causes a greater practical problem for the analysis than the known theoretical limitations of the layer-stripping theory (such as use of a finite-offset source). For the analysis of the example data we develop a number of procedures that are necessary to mitigate the low signal-to-noise levels. These include using a wide range of shot-receiver azimuths to generate data redundancy, methods of identifying and rejecting poor measurements, and a predictive layer-stripping approach that minimizes the propagation of errors through the analysis that arise from scatter in the layer-by-layer results. With the PS waves of ...