The Research On Wavefield Separation Based On Polarization Filter In Shearlet Domain

Compared with single component seismic exploration, multi-component seismic exploration can acquire more wavefield information and help to reduce the non-uniqueness of seismic interpretation,which improvescomplex geological exploration and reservoir prediction. Since conventional seismic imaging methods are mostly based on P wave,separating different seismic wave is very essential to reduce imaging artifactsfor multi-component seismic exploration. Ground roll attenuation and P, S wave separation are two important jobs of seismic wave processing.This paper focuses on the application of polarization filter to seismic wave separation.Cross-correlating multi-component seismic data, the covariance matrix is built and it’s eigenvalues and eigenvectors are solved to obtain the reciprocal ellipticityand polarization direction of seismic wave. And a polarization filter function is employed to separate seismic wave based on polarization characters. Both fix time window polarization filter method and adaptive polarization filter method are applied to the P, S wave separation of Tunnel Seismic Prediction(TSP). The filtered results of synthetic data and field TSP data show that adaptive polarization filter method is more effective for separating reflected P and S wave.Based on the low frequency, low apparent velocity and elliptical polarizationcharacteristicsof ground roll, shearlet transform is adopted to decompose seismic data into different subbands and polarization filter is conducted on the shearlet decomposition coefficients. Shearlet transform, a multidimensional and multidirectionaltransform, decomposes seismic events into different coefficients in scale and direction. In this paper, polarization analysis and filter are implemented on the shearlet coefficients in different apparent velocity and frequency band. According to the dip of shearletcoefficients, a variable elliptical ratio threshold is provided to separate seismic wave. For the shearlet coefficients of large angle, non-aliased ground roll is dominated and a lower elliptical ratio threshold is set to attenuate the ground roll ascompletely as possible. For those of small angle, reflected body wave is dominated and a higher elliptical ratio threshold is set in these directions to preserve the reflected wave as much as possible.A synthetic seismic data with aliased ground roll and a field seismic data are used to test the proposed method. And a comparison is made among the filtered results with FK filter, wavelet polarization filter and shearlet polarization filter. The filter results of model and real data show that, the proposed filter method can suppress the aliased ground roll more obviously than FK filter method, and the effectiveness of shearlet polarization filter prevailsover that of wavelet polarization filter while ground roll overlaps with reflected body wave in time frequency domain.