Study Of The Prestack Inversion Method Based On Geostatistics

Geostatistics can well describe the spatial correlation of subsurface model parameters and provide a wealth of prior information for seismic inversion.Compared with deterministic inversion,seismic stochastic inversion based on geostatistics often has higher vertical resolution and can obtain multiple equal probability inversion results,which is suitable for the uncertainty estimation of the inversion results.The main reason is that geostatistics uses the variogram as a basic tool to describe small-scale changes in model parameters and seismic stochastic inversion incorporates a large amount of logging data into the inversion process.Therefore,stochastic seismic inversion based on geostatistics has more advantages in thin reservoir research.Seismic stochastic inversion mainly uses the idea of stochastic simulation.Conventional stochastic simulation algorithms often require model parameters to obey Gaussian distribution.However,in practical applications,due to factors such as underground lithofacies or geological fluids,model parameters often present a multimodal non-Gaussian distribution.To sovle this problem,this paper proposes a seismic stochastic inversion method based on two different sequential simulation algorithms(Direct Sequential Simulation and Sequential Gaussian mixture Simulation algorithm under the framework of Bayesian.The method can overcome the shortcomings of the conventional stochastic inversion method that the model parameters need to obey the Gaussian distribution and enhance its ability in practical applications.The Direct Sequential Simulation algorithm can be applied to the prior distribution of any shape.The seismic stochastic inversion based on this simulation algorithm can obtain multiple equal probability high resolution inversion results.The Sequential Gaussian mixture Simulation algorithm can be applied to multi-peak Gaussian mixture prior distribution.The seismic stochastic inversion based on this simulation algorithm can simultaneously obtain multiple equal probability inversion results of continuous variables(elastic parameters)and discrete variables(lithofacies).The purpose of seismic inversion is ultimately for reservoir prediction and fluid identifycation that play an important role in exploration and development.This paper is based on these two methods.We use the petrophysical method to construct the sensitive fluid factor,then establish the prestack elastic impedance equation of the sensitive fluid factor,then extract the sensitive fluid factor,Finally,combine fluid factors and lithofacies for reservoir prediction and fluid identification.The practical application shows that the methods we proposed has practical value in reservoir prediction and fluid identification.