Reservoir Prediction Study Of Karamay Formation Seismic Inversion Method In Central Region

Through seismic inversion,formation lithologic information,pore fluid information and reservoir physical property information can often be obtained,so as to carry out reservoir prediction.In recent years,the exploration reservoir is more and more inclined to thin layer prediction,so the nonlinear inversion method with high precision and high resolution is more and more important.As a nonlinear probabilistic inversion method,the markov chain monte carlo(MCMC)method not only avoids the complex marginal integral problem in bayesian theory,but also obtains more accurate posterior probability distribution and optimal solution than bayesian inversion.In this paper,bayesian inversion is firstly carried out.Based on the layered patch saturation model and Zoeppritz equation in viscoelastic medium,the effects of bayesian inversion in inelastic medium and elastic medium are compared.Generally,the real part of seismogram is used in the calculation of likelihood function by bayesian inversion.In this paper,the real and imaginary parts of seismogram are considered,and the range of likelihood function is obviously reduced.By introducing the quality factor Q,the more accurate optimal solution is obtained.On this basis,the bayesian inversion of PP wave and PS wave is carried out.Then the propagation matrix theory is used to carry out forward modeling,and the MCMC inversion is carried out under the bayesian framework to obtain the optimal solution of reservoir thickness and porosity.Meanwhile,the bayesian inversion using the same prior information and the forward modeling theory is carried out.After the comparison of inversion results,the posterior probability distribution of MCMC inversion is more accurate.Based on the study of the adaptive strategy of the proposed distribution used in MCMC inversion,the proposed distribution was modified adaptively for the shortcomings of the inversion process,such as too long time and too much computation,and the proposed MCMC method was optimized to accelerate the convergence of markov chain,shorten the computational time of MCMC inversion,and improve the accuracy of the inversion.Due to the unique reservoir environment in one central region,we used the joint inversion technique of geostatistical inversion and prestack elastic inversion based on MCMC algorithm to predict the thin layer in the working area,and obtained a good prediction effect.