Zero-offset VSP wavefield separation using two-step SVD method

With the development of seismic attribute analysis technique, the fidelity of waveform and amplitude becomes important to wavefield separation method. In this paper, we present a zero-offset vertical seismic profiling (VSP) wavefield separation method with good fidelity. The method is based on singular value decomposition(SVD) filtering and have four stage: first, aligning events of downgoing wave by static time shifting each trace of VSP; second, suppressing downgoing wave by high-pass SVD filtering; third aligning events of upgoing wave by static time shifting each trace of downgoing wave suppressed wavefield; fourth, extracting upgoing wave by low-pass SVD filtering. It is difficult to calculate the time shift of each trace for aligning events of upgoing wave, as the residual downgoing wave in downgoing wave suppressed wavefield is not very week. In this paper, we calculate the time shift via the largest singular value maximization algorithm. I demonstrate with synthetic data and real data example that the results of our method have good fidelity and are better than the results of traditional SVD method.