Stochastic noise attenuation in vertical seismic profile

Focusing on the problem that filtering method used in vertical seismic profile (VSP) data processing is not referred to the ambient noise, a wavelet shrinkage scheme which utilizes the statistical law of ambient noise is proposed. Based on the hypothesis that the ambient noise is statistically stationary, firstly the correlation between ambient noise and real data after the first seismic arrivals is estimated, when the correlation is small, the statistics of stationary noise can be derived from the data before the first seismic arrivals; then stochastic noise can be attenuated in wavelet domain using a threshold estimated by the statistics of stationary noise; and finally high fidelity VSP data is obtained. The simulation results of the synthetical seismic records and real data demonstrate that the signal-noise-ratio of the proposed method is about 2 dB higher than that of filtering method. The algorithm proposed can exhibit time-variant characteristic of seismic signal more sufficiently, and the cophase axis of the profile is more distinct after processing.