Abstract:Based on the framework of Bayes theory,the paper combines application of likelihood function and prior stochastic constrain information to carry out the inversion of AVO seismic parameter,and acquired inversed parameter is probable solution of posterior probability distribution function (PPDF). During inversion,assuming that the noise and model parameters are Gaussian distribution,using petrophysical study approaches and results to create model parameter covariance matrix in order to constrain inversion,the inversion adapts nonlinear iteration algorithm that do not rely on the selection of initial model;the speed of inversion is fast,which is suitable for prestack AVO inversion. After neglecting the influence of NMO stretching and tuning effect caused by bandlimited character of seismic data,the AVO inversion is carried out for synthetic records and real data,which obtained the reflection attributes of Pwave impedance,Swave impedance and density.Furthermore the other lithologic parameters can be further computed,which provides new tool for identification of porefluid nature.