Study On Seismic Attribute Extraction And Application In Depth Domain

As one of the most influential techniques in oil and gas exploration of past 10 years,PSDM has become an important means to obtain underground structural data in seismic exploration.The PSDM data is more real,intuitive and convenient to reflect the structural features of the subsurface than the time migration data.However,the current research on depth domain data mainly focuses on processing.How to interpret seismic data in depth domain is rarely discussed.Seismic attributes are important features for describing geological information such as stratigraphic structure,lithology and hydrocarbon-bearing properties,and are indispensable parameters in seismic data interpretation.Therefore,this thesis focuses on the study of seismic attributes in the depth domain.The thesis compares and analyzes the differences in extraction methods and manifestations of seismic attributes in the depth domain,time domain and deep-time domain,and discusses the advantages of depth seismic attributes and their applicability in interpretation work.In addition,the thesis discusses the influence of noise,velocity grid and velocity field accuracy on the depth seismic attributes,so as to obtain the accurate attributes under the optimal migration parameters.Then,it will lay a solid foundation for the application of attributes in the interpretation.In the interpretation and application of seismic attributes,the thesis studies the application of plane-wave destruction algorithm in depth domain attribute analysis,including fault identification,as well as stratigraphic coordinate system data feature analysis.After two examples test,it is shown that the PCNN detection is performed on the basis of the dip attribute extracted by the plane-wave destruction algorithm,which can identify the fault more effectively.The seismic events of the stratigraphic coordinate system become flat,which reduces the complexity of window opening along the layer when extracting seismic attributes.Moreover,the extracted attributes are more stable and less affected by the overlying and underlying stratigraphic structures,which undoubtedly provides greater convenience and advantages for subsequent reservoir prediction and interpretation.In the last part,two seismic multi-attribute fusion methods based on trained dictionary are studied.According to the difference of the construction dictionary,the two dictionary construction algorithms of K-SVD and sparse auto-encoder are considered.The denoising factor is added to the fusion processing of sparse representation algorithm to realize the multi-attribute data fusion and denoising at the same time.The fused seismic attributes contain as much geological information as possible of the original single data and improve the signal-to-noise ratio of the seismic data.It can be proved from the above studies that the extraction and interpretation of seismic attributes directly in the depth domain has great potential and application value.