Study Of The Fast Stochastic Inversion Method Based On The Lateral Constraint

Seismic stochastic inversion is an inversion method which is based on Geostatistics.Besides,it makes full use of the well logging data,and is also constrained by the seismic data at the same time.Compared with the deterministic inversion,stochastic inversion has high vertical resolution and can get multiple results.First,it uses the well data to analyze the space structure to obtain the variagram.Then,it conducts the stochastic simulation using the variogram to get the initial realization of the model.Last,we can get the inversion results combined with the optimization algorithm.Although there are many advantages of stochastic inversion,the conventional seismic stochastic inversion method is running the defects of poor lateral continuity and slow computing speed.This paper researches on two methods of constructing lateral constraints and three methods of improved optimization algorithms aiming to solve these problems.On the framework of Bayesian theory,this paper makes the priori information containing lateral information through adding the lateral correlated function into the covariogram,which represents the spatial correlation of the variable.Then we can get the lateral constrained simulation results combined with FFT-MA(Fast Fourier Transform-Moving Average).Besides,we introduce the second derivative constraint into the objective function according to the correlation between the seismic trace.It can improve the lateral continuity effectively.The model tests and the analysis of the real data show that the effectiveness of these methods.After that,we introduce and improve the appropriate optimization algorithm to optimize the intial model to get the inversion result.It demonstrates that the improved optimization algorithm enhances the computation efficiency and convergence correspondingly compared with the original algorithm.Reliable identification results of reservoir fluids can enhance the success rate of exploration and development,and reduce the input costs in some degree,which is meaningful to reservoir exploration and evaluation.On the basis of the research on lateral continuity and computation efficiency,we study the seismic fluids identification based on the prestack stochastic inversion.Firstly,we deduce the elastic impedence equation of the sensitive fluid factors based on the Russell approximation.Then we extract the fluid factors directly to reduce the uncertainty transmission and avoid the error accumulation.Lastly,we combine this process with the fast stochastic inversion method based on lateral constraint to provide the effective information.The practical application shows that this method we proposed not only has theoretical significance,but also certain practical value.