Seismic velocity and polarization estimation for wavefield separation

We address the problem of estimating the shape parameters of seismic wavefields using linear arrays of three-component (3C) vector sensors with uncertain acquisition geometry. The goal is to separate the different seismic waves, which is of practical need for oil exploration and geophysics. We present a parametric model for multiple wideband polarized signals received by an array of three-component sensors with positional and rotational calibration errors, and derive the Cramér-Rao lower bounds on the performance of the model parameters for both the exact physical model and the model with uncertain acquisition geometry. We propose a method for jointly estimating the velocity and polarization parameters based on the shift-invariance properties of multiple wavefields impinging on the linear array. We then remove the interfering surface waves by using a beamforming filter designed to exploit the velocity and polarization diversity of the different seismic waves, after clustering of the shape-parameter estimates. Examples using simulated and experimental data illustrate the applicability of the proposed methodology.