Uncertainty estimation in seismic-basedporosity and saturation inversion: Examplesfrom unconsolidated and hard rockenvironments

摘要

Uncertainties associated with predicting pay, saturation, and porosity in the subsurface are related not only toseismic inversion accuracy, resolution, and confidence in rock-physics relations, but also to diagenetic processesassociated with porosity reduction processes and hydrocarbon migration into the reservoir. In this paper, we usestochastic simulation techniques and formal Bayesian inversion to explore the uncertainty associated withporosity and saturation estimates obtained in two prototypical environments: a consolidated, hard sand reservoirin east Texas and soft, unconsolidated sediments in marine environments offshore Asia. In hard rocks, brinesaturation is often correlated with increased cementation and reduced porosity, and hydrocarbons often migrateinto large-porosity areas. Therefore, porosity and saturation are often correlated. In soft unconsolidatedsediment environments, sands have high porosities, and porosity and saturation have a large degree ofindependence. A Markov chain Monte-Carlo (MCMC) simulation technique is used to exploit this geologicalinformation to simulate expected ranges of porosity and saturation associated with these geological processes.Baysian inversion is used to generate optimal estimates of porosity and saturation with uncertainty. We showthat although in hard matrix the response to changes in fluid saturation may be very small, a reliable saturationestimate can be obtained due to the diagenetic processes. In soft sediments, where matrix is more sensitive tochanges in fluid modulus and saturation, confidence in saturation estimates may be higher than the hardsediments example. These effects can be captured using joint inversion techniques that are based both on rockphysics and diagenetic characterization of the reservoir.