Seismic Wave Field Separation And Noise Attenuation In Frequency Domain Via Singular Value Decomposition (Svd)

Irregular interference noises are distributed widely without coherence. By normal band pass filtering, some interference signals might be reduced. Because of the narrow range of signals in main frequency band in the same area, that is to say, the spectrum of various seismic frequencies are coherent, we applied SVD technique to extract the singular value of target signals for the spectrum reconstruction, namely, transforming filtering parameters in normal frequency domain into the domain of SVD, then realized the separation of wave field and attenuation of noises. Compared with conventional band pass filtering, we can suppress interference noise effectively and avoid losing the effective signals of high-frequency and low-frequency, improve the SNR of seismic data, and guarantee the resolution on the premise of keeping original geological information. By this method, we are able to process seismic data with high fidelity, high resolution and high SNR effectively.