Integrating stochastic rock physics in seismic pre‐drill prospect risk and reservoir quality assessment

Abstract

This paper describes a methodology that predicts rock properties of a vertical profile of the earth and its associated uncertainty from seismic amplitudes. The methodology can be used to predict the probability of hydrocarbon presence, reservoir column and quality. It is well suited for pre-drill spot tests. The predictions are based on a Bayesian approach, which defines a prior stochastic model for sand beds in a background of shale, and potential fluid contacts. The rock properties are defined by a stochastic rock physics model, which also accounts for the vertical continuity of the rock properties within a lithology and fluid class. A tailored Markov chain Monte Carlo algorithm efficiently explores possible earth models to find a representative set that is consistent with seismic data as well as the prior stochastic model. We have applied the methodology to a case from the North Sea.

Seismic to Facies Inversion Using Convolved Hidden Markov Model

“…Later publications (30,31,32) use MCMC sampling in order to assess the exact posterior distribution, but with a Gibbs sampling strategy while the present thesis uses independent proposal Metropolis-Hastings sampling.…”