SVD (singular value decomposition) seismic wave field noise elimination in frequency domain
Shen Hong-yan1,2, Li Qing-chun2
1. Institute of Oil & Gas Resources, Xi'an University of Petroleum, Xi'an City, Shaanxi Province, 710065, China;
2. College of Geology Engineering and Geomatics, Chang An University, Xi'an City, Shaanxi Province, 710054, China
Abstract：Random noise has a wide frequency spectrum so that it does not have coherence, but the main frequency band for seismic signals in the same area are almost the same, especially for post-stack data, in other words, the frequency spectrum of the effective frequency band for seismic data has a better coherence than the random noise. Based on the conclusion above, SVD transform is conducted in frequency domain through Fourier Transform, the filtering factor was designed, and SVD values of the target function were extracted to reconstruct signals in frequency domain, then inverse Fourier Transform was conducted, and seismic wave field separation and noise elimination were realized. Compared with traditional band filtering the new method could completely suppress interference noise in main frequency band, protect effective signals in high and low frequency. The field seismic data processing results show that under the prerequisite in which the geological information is not damaged, the applying of the new method in seismic data processing could greatly raise the signal to noise ratio.
