Study Of Seismic Stochastic Inversion Method Based On Sequential Co-simulation

Seismic stochastic inversion can combine the stochastic simulation with the stochstic optimazition.Compare to the determined inversion,stochastic inversion can get the more accurate inversion result,accurately describe the internal details of the reservoir.As a method of stochstic inversion,geostochastic inversion can integrate of geostatistical modeling and seismic inversion theory,combine the high-resolution advantages of logging data with the features of good transverse continuity of seismic data,can get the high-resolution inversion of reservoirs.Traditional geostatistical inversion is usual based on classical sequential Gaussian simulation,has the defect that the original data needs Gaussian transform,and the operation efficiency is low.Based on the previous studies,direct sequential simulation algorithm is applied to this paper.This method can simulate the data in the original field,there is no need to perform Gaussian transformation on the data,parallelization program is also applied to simulation method to improve simulation efficiency.This paper is under the framework of iterative global geostatistics theory,parallelized direct sequential simulation algorithm is used to perform post-stack and pre-stack geostatistical inversion.In the post-stack geostatistical inversion,direct sequential simulation and cosimulation are applied as disturbance method of model parameters.The objective function is set under the Bayesian theory.The synthetic seismograms obtained from the simulation results are matched with the actual seismograms,the simulation results are selected by the matchedresult.The cross-over principle of genetic algorithm is used as the global optimization technique,the iterative procedure continues until a stopping criteria is reached: frequently the objective function.Modeling data and seismic data are used to test the method,more accurate inversion results are obtained.The geostatistical seismic AVO inversion can be summarized in three main stages: stochastic sequential joint simulation of elastic models for the properties to invert—density,Pwave and S-wave velocities;forward modeling and mismatch evaluation between the observed and the inverted seismic data;and selection of the conditioning data for the generation of the next set of elastic models during the next iteration.In the stochastic sequential simulation stage,the sequential procedure starts by simulating,for the entire seismic grid simultaneously,density models followed by the co-simulation of P-wave velocity models from which S-wave velocity models are co-simulated.After the simulation of elastic models,synthetic pre-stack seismic volumes are calculated based on the Zoeppritz equation.After the forward modeling is complete,the resulting synthetic angle gathers are compared with the corresponding real ones in terms of a correlation coefficient.The correlation is performed trace-by-trace per angle gather allowing assessment of local mismatches between synthetic and real traces.After the mismatch evaluation is complete,the conditioning data used to constrain the co-simulation of the new set of elastic models during the next iteration is then generated.This step comprises the selection and generation of the best density,P-wave and S-wave velocity models along with the corresponding best local correlation cubes.The resulting local correlation coefficients are used by the global optimizer,a genetic algorithm,to converge the inversion into the solution.