Near-surface Q Inversion And Application Based On Generalized S Transform

The attenuation of near-surface on land is relatively strong to valid seismic signal.However,due to the difficulty for Q estimation in shallow weathering zone and the adaptive problem in current compensation methods,the influence from near-surface has not been effectively considered into attenuation estimation and compensation in seismic data processing.In this dissertation,a novel method for near-surface Q estimation is proposed based on generalized S transform,which can provide time-frequency spectrum with high resolution for logarithmic spectral ratio algorithm.3D Q model of shallow weathering zone is established by use of a statistical relation between the estimated layer Q and layer velocity,overcoming inaccuracy of spatial interpolation with the limited values of the estimated layer Q.Finally,attenuation compensation with seismic data is achieved using the compensation method with time-varying Q values on basis of the built 3D Q model of near-surface.It raises the vertical resolution of seismic data effectively,maintaining Signal-to-Noise Ratio(SNR)and relative features of spectrum.Among the Q estimation methods,the logarithmic spectral ratio(LSR)is of high accuracy,but with weak anti-noise property.By contrast,the logarithmic spectral area-difference(LSAD)method owns better noise immunity.In this dissertation,we propose a novel Q estimation approach based on generalized logarithmic spectral ratio(GLSR),through the analysis between the difference and relation of the above two methods.Test of synthetic model and field data demonstrate that the proposed approach,combining the advantages of LSR and LSAD methods,is able to estimate Q factor with high accuracy and anti-noise property.