Capability of the Stochastic Seismic Inversion in Detecting the Thin Beds: a Case Study at One of the Persian Gulf Oilfields

The aim of seismic inversion is mapping all of the subsurface structures from seismic data. Due to the band-limited nature of the seismic data, it is difficult to find a unique solution for seismic inversion. Deterministic methods of seismic inversion are based on try and error techniques and provide a smooth map of elastic properties, while stochastic methods produce high-resolution maps of elastic properties with the same probability. The current paper studies a stochastic method of seismic inversion which was applied to one of the Persian Gulf oilfields. Joint posterior distribution of elastic properties was calculated using Bayesian principle; then a sequential Gaussian simulation technique was performed to decompose the global probability function of elastic properties into some local probability functions at each trace location. The sampling of the local probability functions was performed, and two hundred realizations of the elastic properties were generated. The results of the stochastic inversion were found to be capable of modeling heterogeneities of the reservoir. The generated realizations provided the possibility to uncertainties assessment by calculating the variance of the elastic properties. It was found out that the uncertainty increased in locations far away from the well. Moreover, stochastic inversion, unlike deterministic one, was found to be capable of detecting thin beds (3.5 to 5.7 m) embedded within the reservoir.

Keywords:

Stochastic Inversion , Deterministic Inversion , Bayesian Framework , Sequential Gaussian Simulation , Thin Bed Detection