Inversion based data-driven attenuation compensation method

Abstract

Summary 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 instable or under-compensation for amplitude compensation, while Gabor deconvolution is based on the assumption of minimum phase wavelet which deviates from real conditions to some extent, therefore its application is limited. We proposed a novel method which combines merits of inverse Q filtering and Gabor deconvolution: neglecting effects of wavelet and just compensating attenuation. The procedures include: 1) Extract attenuation function by hyperbolic smoothing in Gabor domain; 2) Use non-combination theory and inverse strategy to restore effective frequency components of compensated seismic data; 3) Perform inverse Fourier transform to obtain compensated seismic data. This method, which needs not accurate Q value, is stable and accurate compared with traditional inverse Q filtering method; it is data driven and it's applicable to different data sets because it avoids assumption of minimum phase wavelet compared with Gabor deconvolution; it is computational efficient because only effective frequency components are calculated compared with methods which need to calculate whole seismic data series. The validity of the proposed method has been approved by tests on synthetic and real data.