Pre-stack Seismic Inversion Based On Markov Chain Monte Carlo Method

Prestack seismic data with better fidelity and information of amplitude and travel timewhich related to the reservoir parameters, is more reliable to reveal not only the distribution ofreservoir, but also physical properties and hydrocarbon potential. The accuracy ofconventional geophysical inversion methods which are based on the linear approximationtheory cannot meet the need of current exploration, so we make use of non-linear algorithm tosolve the seismic inverse problem directly for the purpose of obtaining the elastic andreservoir parameters from the point of view of probability.This paper studies the prestack seismic inversion method based on Markov Chain MonteCarlo algorithm. Markov Chain Monte Carlo algorithm is a heuristic global optimizationmethod, exploring the model space by generating Markov chains. In the Bayesian framework,the optimal solution meets the statistical properties of the parameters with the constraints ofdata; the accuracy of solution is improved with the prior information joined in; optimizationprocess can jump out of local optimum, and ultimately obtain the global optimal solution.Using MCMC methods, we draw a large number of samples from the posterior distributionfunction. With these samples, we obtain not only the estimates of each unknown variable, butalso various types of uncertainty information associated with the estimation. In addition,because of not making use of a objective function with single optimal solution, the resultsobtained with MCMC method are independent of the choice of initial values.Based on the analysis of influence of elastic parameters and reservoir parameters onAVO, we develop the Bayesian frame based on MCMC algorithm to inverse elasticparameters and reservoir parameters. The algorithm to generate chains is Metropolis-Hastingsalgorithm, and the step length in proposal distribution function decreases as iteration goes onto ensure the speed to converge in the beginning and the precision of the estimation. Thenmake discussion of influencing factors, such as signal to noise ratio of seismic data and theinitial model, and make summary of the principle how to select parameters of the algorithmproviding a basis for actual data processing. Finally, by testing model and the application ofreal seismic data of A oil field, shows the prestack inversion based on MCMC method is available and can obtain good results.