Adaptive Edge-preserving Non-linear AVO Inversion Based On Trivariate Cauchy Probability Distribution

AVO inversion can estimate the elastic parameters of subsurface rocks such as P-wave velocity,S-wave velocity,and density,and is an effective method for reservoir prediction and fluid identification.However,the following problems still exist in its practical applications:(1)The density term exhibits little sensitivity to amplitudes and,therefore,its inversion is unstable;(2)Due to the low-frequency components of seismic data are weakness,the low-frequency components are more easily to be disturbed by noise,and resulting in poor lateral continuity of the inversion results;(3)Seismic data are usually contaminated by noise and the conventional seismic inversion methods are sensitive to noise,causing the inversion results become oscillated near the interface boundaries.Therefore,the vertical resolution of inversion results is not high and does not clearly reflect the location of interface boundaries and geological body edges.Regarding the above-mentioned problems,on the basis of summarizing the research status and development trends of AVO inversion methods,this paper proposes a new method named adaptive edge-preserving non-linear AVO inversion based on Trivariate Cauvhy probability distribution.In order to stabilize the density term,we model the prior distribution of the Bayesian inversion via a Trivariate Cauchy distribution,which contains a scale matrix that provides correlation information between the three unknown AVO parameters.The introduction of the low-frequency information constraint effectively complements the low-frequency components of seismic data and increases the accuracy and lateral continuity of inversion results.By incorporating the AVO inversion into the adaptive edge-preserving smoothing filter framework,the interfaces and geological body edges in the inversion results are clearer,and the resolution of inversion results is also improved.