In practical seismic exploration, internal multiples generated when the wave impedance of medium is strong, and seismic records are recorded. The method of virtual event repress internal multiples is to move scattered points from underground to the surface, similar to the method of the surface-related multiple elimination (SRME). The method of SRME belongs to the prediction-subtraction approaches to eliminate internal multiples, prediction method is based on building a brand new way of seismic wave propagation (virtual reflection and virtual event), so that it has forward and backward wave propagation, and through convolution with significant wave to predict the internal multiples. Due to required data needing field information of full-wave, the authors use Seislet transform interpolating the missing data to ensure the premise of internal multiples prediction. The test data show that the above method has achieved good results.
In seismic data processing, random noise seriously affects the seismic data quality and subsequently the interpretation. This study aims to increase the signal-to-noise ratio by suppressing random noise and improve the accuracy of seismic data interpretation without losing useful information. Hence, we propose a structure-oriented polynomial fitting filter. At the core of structure-oriented filtering is the characterization of the structural trend and the realization of nonstationary filtering. First, we analyze the relation of the frequency response between two-dimensional（2D） derivatives and the 2D Hilbert transform. Then, we derive the noniterative seismic local dip operator using the 2D Hilbert transform to obtain the structural trend. Second, we select polynomial fitting as the nonstationary filtering method and expand the application range of the nonstationary polynomial fitting. Finally, we apply variableamplitude polynomial fitting along the direction of the dip to improve the adaptive structureoriented filtering. Model and field seismic data show that the proposed method suppresses the seismic noise while protecting structural information.
In seismic prospecting, fi eld conditions and other factors hamper the recording of the complete seismic wavefi eld; thus, data interpolation is critical in seismic data processing. Especially, in complex conditions, prestack missing data affect the subsequent highprecision data processing workfl ow. Compressive sensing is an effective strategy for seismic data interpolation by optimally representing the complex seismic wavefi eld and using fast and accurate iterative algorithms. The seislet transform is a sparse multiscale transform well suited for representing the seismic wavefield, as it can effectively compress seismic events. Furthermore, the Bregman iterative algorithm is an efficient algorithm for sparse representation in compressive sensing. Seismic data interpolation methods can be developed by combining seismic dynamic prediction, image transform, and compressive sensing. In this study, we link seismic data interpolation and constrained optimization. We selected the OC-seislet sparse transform to represent complex wavefields and used the Bregman iteration method to solve the hybrid norm inverse problem under the compressed sensing framework. In addition, we used an H-curve method to choose the threshold parameter in the Bregman iteration method. Thus, we achieved fast and accurate reconstruction of the seismic wavefi eld. Model and fi eld data tests demonstrate that the Bregman iteration method based on the H-curve norm in the sparse transform domain can effectively reconstruct missing complex wavefi eld data.
A key problem in seismic inversion is the identification of the reservoir fluids. Elastic parameters, such as seismic wave velocity and formation density, do not have sufficient sensitivity, thus, the conventional amplitude-versus-offset（AVO） method is not applicable. The frequency-dependent AVO method considers the dependency of the seismic amplitude to frequency and uses this dependency to obtain information regarding the fluids in the reservoir fractures. We propose an improved Bayesian inversion method based on the parameterization of the Chapman model. The proposed method is based on 1） inelastic attribute inversion by the FDAVO method and 2） Bayesian statistics for fluid identification. First, we invert the inelastic fracture parameters by formulating an error function, which is used to match observations and model data. Second, we identify fluid types by using a Markov random field a priori model considering data from various sources, such as prestack inversion and well logs. We consider the inelastic parameters to take advantage of the viscosity differences among the different fluids possible. Finally, we use the maximum posteriori probability for obtaining the best lithology/fluid identification results.
