INVERSION BASED DATA-DRIVEN ATTENUATION COMPENSATION METHOD

Seismic data resolution can be reduced because of heterogeneity and viscosity of subsurface medium. Inverse Q-filtering and Gabor deconvolution are able to effectively improve seismic data resolution. However inverse Q-filtering requires accurate Q values and it is always unstable or under-corrected for amplitude compensation, while Gabor deconvolution is based on the assumption of a minimum phase wavelet, which deviates from real conditions to some extent, therefore its application is limited. This paper combines merits of inverse Q-filtering and Gabor deconvolution: neglecting effects of wavelet and just compensating attenuation. The procedures include: 1) Extract an attenuation function by hyperbolic smoothing in Gabor domain; 2) Use non-combination theory and inverse strategy to restore effective frequency components of the compensated seismic data; 3) Perform an inverse Fourier transform to obtain compensated seismic data. This method, does not need an accurate Q value, is stable and accurate compared with traditional inverse Q-filtering methods; it is data driven and is applicable to different data sets because it avoids the assumption of a minimum phase wavelet compared with Gabor deconvolution; it is computationally efficient because only the effective frequency components are calculated compared with methods that need to calculate the whole seismic data series. The validity of the proposed method has been verified by tests on synthetic and real data.