Research On AVO Seismic Parameter Pre-stack Inversion Method Based On Bayes Theory

AVO inversion can extract reservoir elastic parameters from the pre-stack gathers bystudying the relationship between the variation regularity of reflection amplitude and reservoirelastic parameters. Meanwhile, using rock physics as a bridge, it can convert elasticparameters into reservoir parameters in order to make the fluid identification and reservoirdescription happen. Because of the three-term AVO equation is typically too ill-conditioned,the traditional AVO inversion usually makes inversion stable by solving for two parametersrather than three ones or imposing a hard constraint, however, at the same time introducingthe bigger bias to the estimate. In order to balance the contradiction between the improvementof the stability of inversion problem and the losses of the inversion extent, the three-termAVO inversion based on the Bayes is being improved constantly, so that it can ensure thestability of inversion, and further improve the reliability of inversion results.Typically, using the Gauss consistency weighted prior distribution of model parameterscan improve the stability of three-term AVO inversion problem, but the resolution of theinversion results is relatively low; Using Cauchy long tail as a sparse constraint, the resolutionis relatively high, but such a prior constraint assume AVO three parameters is not relevantwhile precisely there is a certain correlation between the parameters, and decorrelation willintroduce theoretical error. As for the issues above mentioned, this paper introduces theconcept of multivariant Cauchy probability distribution of strong constraint inversion process.The method realizes the sparse pulse inversion effectively and enhances the reliability of theinversion results. For the corresponding information among the AVO three parameters, onecan get the statistical relation of rock geophysics in the target area through accurate analysisof the logging data, and then build the scale matrix into the inversion process. The AVOseismic parameters inversion based on trivariate Cauchy probability distribution is sufficientlyintegrated with the maximum a posteriori inversion process of logging, geological and otherpriori information constraints. At the same time, it brings the petrophysical analysis of thetarget area as an important constraint into the inversion process, improving the stability of thepre-stack inversion, thereby increasing the effectiveness and reliability of three AVOinversion. The inversion method has higher computing efficiency, moreover, model testing and practical application have accomplished some achievements.Pre-stack seismic inversion is based on analysis of AVO characteristics of pre-stackgather, but the existance of the amount of residual moveout in pre-stack seismic data makesthe inversion precision much lower. Therefore, in order to correctly extract AVO information,we must eliminate residual moveout, leveling the event. We discussed and verified the residualmoveout correction method based on the S transform in this paper. This method can extract thesame seismic at different times sampling points of residual time to get the shift of time bytime-frequency decompositing the seismic signal, and high precisely eliminating residualmoveout in signal reconstruction. And this residual moveout correction method has nothing todo with the residual velocity. The theoretical model and actual data processing results showthat the method to eliminate the residual moveout has a good result.