Research On Nonlinear Post-stack Stochastic Inversion Method Based On Geostatistics Theory

At present,the exploration and development of oil and gas resources are facing many challenges,such as the insufficient resolution of seismic data to meet the needs of exploration and development of lithology and thin-layer oil and gas reservoirs.For seismic exploration,finding inversion and processing methods to improve the resolution as much as possible has become the focus of research.The conventional deterministic inversion method is relatively mature,and the algorithm is comparatively simple,but due to the band-limited nature of seismic data,the resolution is difficult to meet the requirements of fine reservoir description.The stochastic inversion method uses well logging data as hard data,and at the same time carries out inversion with seismic data,it can break through the limitation of seismic frequency band,thereby effectively improving the resolution of inversion.Based on stochastic inversion,this paper studies the nonlinear post-stack stochastic inversion method based on geostatistics theory.Starting from Bayesian theory,it uses geostatistics theory to fuse well logging information and geological information to establish a priori space of the model,which effectively supplements the lack of frequency band of seismic data and improves the resolution of inversion.Combined with the poststack seismic data,the posterior distribution is sampled by the nonlinear inversion method–Markov Monte Carlo method,and then the Bayesian solution is obtained through statistical inference,which avoids finding the analytical solution of complex highdimensional posterior distribution,and at the same time prevents the inversion result from falling into local extremum.The multi-point geostatistics theory is applied to stochastic inversion,which combines the advantages of multi-point and two-point geostatistics to better characterize complex spatial structures.