The Study And Application Of Pre-stack Seismic Stochastic Inversion Method

With the deepening of the oil and gas exploration,the exploration difficulty is higher and higher,the exploration targets have been turned from structural reservoirs to lithological reservoirs.In the actual acquisition data,seismic data has good continuity,but it is limited;while the logging data has high resolution,but it can only display the change of formation lithology and elastic parameters on the position of well hole.Thus,in the actual industrial production,how to make full use of seismic and logging data of useful information has become the key themes of seismic inversion method.Compared with the elastic inversion(EI)and AVO inversion method,the stochastic inversion method can effectively use the high frequency information of logging data,so as to overcome with the limit of seismic data,and improve the vertical resolution.Meanwhile,compared with the post-stack inversion,pre-stack seismic inversion method can obtain more information,and improve the accuracy of reservoir characterization.Seismic stochastic inversion method is a kind of nonlinear inversion method,which combine the prior information construction with the optimization theory.Seismic stochastic inversion is based on the Bayes theory,combining the geologic statistical knowledge with the conventional deterministic inversion technique,which fully make the known seismic data,geological information and logging data combined.Firstly,we construct the reservoir parameter model reflecting the characteristics of the spatial uncertainty by the random simulation method,then using the seismic inversion and nonlinear optimization algorithm to filtrate the stochastic simulation results.Because of disadvantage of running slowly in the conventional stochastic inversion method.In this paper,we use the Fourier spectrum simulation and moving average(FFT-MA)gradual deformation algorithm(GDM)to build a priori information instead of the geo-statistics method.Moreover,we adopt the particle swarm algorithm based on the very fast simulated annealing algorithm(VFSA_PSO),instead of the original optimization algorithm,so as to improve the computing speed of seismic stochastic inversion.Model test results show that the method can effectively identify the thin layer.Therefore,we develop a method,pre-stack stochastic inversion,by combining the two advantageous inversions together which is formulated based on the Bayesian framework.The pre-stack stochastic inversion method include pre-stack EI stochastic inversion and pre-stack AVO stochastic inversion,which can get more parameters on the basis of improving the resolution.Stochastic simulation can generate multiple implementations,and complete the inversion with the constraint conditions of seismic data,which can select the optimal result of multiple implementations effectively,and you can also use these inversion results to do the analysis of uncertainty.We test the seismic stochastic inversion method by using the model and actual data.And the results show that the inversion method not only has theoretical significance,but it also has a certain practical value.