The Application Of Matching Pursuit Into Seismic Inversion

With the constant progress of seismic exploration technology, seismic reservoirprediction technology is more and more widely used, and also makes good profits inpractical exploration and production. Seismic impedance inversion as the coretechnology of in reservoir prediction, plays a more and more important role inreservoir prediction. This technology is an imaging (solving) process which appliesseismic data of surface observation, with known geological regulation, the drilling andlogging data as constraints, and responses to the spatial structure and physicalproperties of underground rock. It can reflect the underground geological structureand lithologic characteristics of reservoir, and it is the key technology of reservoirdescription and reservoir prediction.The present impedance inversion methods are linear. Most of the earth’s physicalinversion is more extreme value of the objective function optimization, thus, linearinversion method can lead to inversion result in local optimum. In order to find outthe optimal solution of the whole physical geography inversion, this thesis realizes thenonlinear seismic wave impedance inversion through matching pursuit and itsimproved algorithm. Accordingly, the author gets the following research results:First: On the basis of matching pursuit basic principle and process, the authormakes a further study of how to extract relatively reflection coefficient of the seismicdata with this algorithm, realizing seismic relative reflection coefficient inversion.The result of the model trial is that the algorithm can realize the adaptive sparsedecomposition of seismic signal, and the decomposition precision is relatively high.For the thickness of strata is less than14,the seismic inversion results still can reflectthe trend of wedge of the model, but the error may be larger.Second: According to the basic theory of matching pursuit algorithm and thealgorithm process, the author extracts the relative reflection coefficient of seismic data,realizing the higher precision of seismic reflection coefficient inversion. Through thecomparison of the new and old algorithms of the model trial, the author also verifiesthat the dipole matching pursuit algorithm is preciser and stabler-decoding than thetraditional matching pursuit decomposition algorithm of seismic signal decomposition,and the dipole matching pursuit algorithm keeps noise resistance. Third: In detail introduced the principle of wave impedance inversion, the waveimpedance inversion basic assumptions and main method on the base of waveimpedance inversion, mainly illustrates the use of relative reflection coefficient isextracted to realize the principle of wave impedance inversion and process. Modeltrial analysis indicates that the improved algorithm of wave impedance inversionresults than the traditional ones keeps higher resolution.Fourth: This thesis applies the matching pursuit wave impedance inversion andits improved algorithm into the actual seismic data inversion. And the author, basedon the comparative analysis of model, matching pursuit impedance inversion, anddecomposition of dipole matching pursuit impedance inversion, will verify the factthat matching pursuit and its improved algorithm keep higher accuracy and betterresult of seismic data of wave impedance inversion.Fifth: In the process of the using of the matching pursuit and its improved toextract relatively reflection coefficient, the author introduces three instantaneousparameters to restrain a complete atom library parameter selection in the guarantee theaccuracy of wave impedance inversion and also to improve the operation rate.Sixth: According to the research result of this thesis, we can know that thematching pursuit with improved algorithm is an effective inversion algorithm,especially suitable for nonlinear, more extreme value of geophysical inversionproblem, and it keeps widely application prospect in the geophysical nonlinearinversion. This method can also be applied to other fields about presented to solvenonlinear optimization problem and the algorithm has strong universality.