Seismic Time-Varying Wavelet Estimation And Blind Sparse-Spike Deconvolution

Seismic wavelet estimation and deconvolution is one of the important steps in seismic data processing,which influences the accuracy of high-resolution,reservoir inversion and seismic interpretation.However,the physics attributes of seismic wavelet are changing while propagation because of absorption and scattering,and the reflectivity is complex and diverse in real exploration industry.These make the results of estimated seismic wavelet and estimated reflectivity inaccurate.To overcome the problems above,we studied a nonstationary sparse spike deconvolution with anelastic attenuation,blind sparse-spike deconvolution with thin layers and structure,time-varying analytic wavelet estimation using local phase attributes,respectively.Firstly,because of the influence of absorption and scattering,the frequency and phase of the seismic wavelet change with time during wave propagation,resulting in an accuracy deconvolution result.To improve the accuracy deconvolution result of nonstationary seismic data,we assume the attenuation of the propagating seismic wavelet belongs to the anelastic attenuation.We present a nonstationary sparse spike deconvolution with anelastic attenuation,of which the subproblems are estimating the Q value,seismic source wavelet and reflectivity,respectively.The numerical results show proposed method performs better in synthetic seismic data and the real seismic data.Meanwhile,parameter setting,noisy seismic data and the estimation error of Q value to validate the effectiveness of our extended approach.Then,to increase the vertical resolution and lateral continuity of the estimated reflectivity,we improve the sparse-spike deconvolution by introducing the atomic norm minimization and structural regularization.Specifically,on the one hand,the conventional sparse-spike deconvolution methods preform not so well when there are some thin layers in the reflectivity,resulting in the loss of vertical resolution.To improve these methods,we apply the atomic norm minimization method to estimate the location of the reflectors which are further used as position constraints in the sparse-spike deconvolution.The numerical results show that we are able to vertically separate highly thin layers in the sparse deconvolution.On the other hand,most sparse-spike deconvolution methods are implemented trace-by-trace for 2D seismic data while estimating the seismic reflectivity,and thus,the lateral continuity of the 2D deconvolution result is not well preserved.We propose the sparse-spike deconvolution with structural regularization to preserve the lateral continuity of the reflectivity.The numerical results show that proposed method perform better than conventional sparse-spike deconvolution method.It should be highlighted that the improvements for vertical and lateral resolution we proposed are suitable for the general sparse-spike deconvolution methods.Last,the seismic wavelet is time-varying while propagation.Previous methods involved local time-frequency decomposition extract the amplitude spectrum of timevarying wavelet under the assumption of white reflectivity.However,the phase of the seismic wavelet is also an important attribute that influences the accuracy of estimated wavelet and the location of the reflectivity from deconvolution.Therefore,we propose a method by combining local time-frequency decomposition and local skewness to obtain the time-varying seismic wavelet.Numerical results regarding synthetic data and well-log data illustrate the effectiveness and practicability of our approach.