Near-surface S-wave traveltime corrections and inversion: a raypath-consistent and interferometric approach
Removing near-surface effects in the processing of 3C data is key to exploiting the information provided by converted waves, particularly for the case of the PS mode where converted energy travels back to the surface as S-waves. The very low velocity of S-waves amplifies the distortions introduced by the near-surface in the PS-traveltimes. This is usually solved by computing the vertical traveltimes in the near-surface layer and removing them from the data in a surface-consistent framework. However, if the velocity change between the near-surface layer and the medium underneath is small, the vertical raypath assumption that supports the surface-consistent approach is no longer valid. This property results in a non-stationary change of the near-surface traveltimes that needs to be addressed to properly remove its effect. I show how the delays introduced by the presence of very low S-wave velocities in the near-surface can introduce raypath-dependent effects which can be larger than what can be considered a residual static. In this study, a raypath-consistent solution for removing nearsurface traveltime effects is proposed. This is achieved by transforming the data, organized into receiver gathers, to the τ-ρ domain and performing crosscorrelation and convolution operations to capture and remove the near-surface delays from the data. The τ-differences captured during the interferometric processing of the near-surface effects are then used in an inversion algorithm to estimate the S-wave velocities in the near-surface. This processing work-flow provides not only a set of corrections but also a velocity model that is based on them. I tested this method on synthetic and field data. In both cases, removing near-surface time delays in a raypath-consistent framework improved coherency and stacking power of shallow and deep events simultaneously. Shallow events benefited most from this processing due to their wider range of reflection angles. This approach can be useful in the processing of wide-angle broadband data, where the kinematics of wave propagation are not consistent with vertical raypath approximations in the near-surface. Additionally, this method provides a near-surface S-wave velocity model that can be used for building migration velocity models or to initialize elastic full waveform inversions.