Research And Application Of Bipolar Reflectivity Spectral Inversion Based On 3D Structural Geosteering Group Sparse

Sparse reflectivity spectral inversion method can reconstruct the frequency component of reflectivity outside effective frequency band from seismic data,and obtain impulse reflectivity with wide frequency band to improve the resolution and accuracy of seismic data interpretation.However,this method is an ill-posed problem for band-limited and noisy seismic data,regularization method is needed to add the a-priori knowledge of parameters to increase stability of the inversion,whitch means regularization terms is the key to the inversion objective function.On the other hand,structural morphology and distribution characteristics of seismic reflective interface are important a-priori information for seismic inversion,accuracy and stability of seismic inversion can be significantly improved by combining structural geosteering method.However,the practical reflective interface has the characteristics of large structural fluctuation,complex internal structure and strong lateral heterogeneity,whitch is difficult to establish accurate mathematical and physical model.In this context,this thesis conducts research and application of bipolar reflectivity spectral inversion based on 3D structural geosteering group sparse,which improves the resolution of complex structural seismic data effectively and provide high-quality seismic data for oil and gas exploration and development.This thesis firstly analyzes the characteristics of a-priori information in reflectivity spectral inversion method under the framework of Bayesian theory,methods of establishing multi-channel mixed regularization terms and group sparse regularization terms under structural geosteering,and establishes regularization terms based on the actual distribution characteristics of different a-priori information.Then investigates structural geosteering constraint methods based on local dip angle and layer interface,and proposes a structural geosteering constraint method based on shear backfill mapping by combining the advantages of both for structural description.After that,proposes a bipolar reflectivity spectral inversion method based on 3D structural geosteering group sparse,by combining the structural geosteering constraint method based on shear backfill mapping,with the bipolar reflectivity spectral inversion based on the principle of odd–even decomposition.Finally,experiments on synthetic seismic data and field seismic data demonstrate the superiority and reliability of the proposed methods in improving the resolution of complex structural seismic data.The main results and recongnition of this thesis are asfollows:(1)In inversion objective function,form of the regularization term depends on form of the a-priori distribution of noise and parameters to be inverted,and the regularization parameter depends on dispersion of the a-priori distribution of the parameters to be inverted and the signal-to-noise ratio of the observed data;Accuracy of the a-priori informations described by regularization term and regularization parameter directly determines the accuracy of inversion results.(2)Group sparse regularization term describes the fused distribution characteristics within groups and between groups based on the multichannel a-priori distribution characteristics,by grouping the multichannel reflection interfaces according to time or seismic channels.(3)A structural geosteering constraint method based on shear backfill mapping is proposed to adapt to complex structures,which can describe both small-scale and largescale structural features simultaneously by using local homophase axis dip and layer interface.On this basis,a 3D group sparse regularization term is established by matching the distribution characteristics after mapped,which can improve the spatial continuity of inversion results along structural geosteering and improve the stability and thin-layer resolution of reflectivity spectral inversion effectively.(4)A mapped domain bipolar odd-even decomposition method that can adapt to multi-channel complex structures is proposed.The proposed method can extract local structural information by using shear backfill mapping method,and the mapped model has typical group sparse distribution characteristics,which provides feasibility and rationality for the odd-even decomposition of multi-channel bipolar reflectivity in mapped domain,whitch enables the bipolar reflectivity spectral inversion method,not directly applicable to multi-channel complex structures,to function in improving the thin layer identification capability.(5)A bipolar reflectivity spectral inversion method based on 3D structural geosteering group sparse is proposed.The proposed method overcomes the difficulty in describing the a-priori information of reflectivity of complex structures by combining model transformation regularization with the group sparse regularization method based on structural geosteering shear backfill mapping.Experiments on synthetic seismic data demonstrate that the proposed method can achieve more accurate and stable reflectivity spectral inversion results by regularizing the inversion process with the a-priori information carrying structural features.(6)Experiments on field seismic data demonstrate that the proposed method is adaptable to complex structures,can reflect real contact relations of the layers,and portray the real location of pinch-out of the layers accurately.The inversion results at uphole trace are in are in good match with the logging data,which proves the reliability and rationality of the method,and also demonstrates that the method has the ability to identify thin layers and small-scale geological bodies with thickness less than the tuning thickness which cannot be distinguished by the original seismic and conventional reflectivity spectral inversion methods.The innovation of this thesis is mainly reflected in the following two aspects:(1)A shear backfilling mapping method that can adapt to complex structures is proposed.The mapping method does not change the horizontal position in 3D data,but only transforms the local occurrence to near-horizontal in mapped domain.The mapped data have the distribution characteristics of vertical sparse and horizontal near-horizontal continuous,which reduces the difficulty to describe the a-priori distribution characteristics of data,facilitates the establishment of the regularized inversion objective function,and improves the accuracy and stability of the inversion.(2)A bipolar reflectivity spectral inversion method based on 3D structural geosteering group sparse is proposed.The proposed method combines group sparse regularization with structural geosteering method based on shear backfill mapping to improve the accuracy of describing the a-priori distribution information of complex structural reflectivity,and improves the identification of thin layers by using a dipole spectral inversion method based on the odd-even decomposition in mapped domain.the proposed method can improve vertical resolution while preserving structural continuity of original data,and provide more interpretable high-resolution reflection interface results in comparison with conventional spectral inversion methods.