A sampling-based Bayesian model for gas saturation estimation using seismic AVA and marine CSEM data

We develop a sampling-based Bayesian model to jointly invert seismic amplitude versus angles (AVA) and marine controlled-source electromagnetic (CSEM) data for layered reservoir models. The porosity and fluid saturation in each layer of the reservoir, the seismic P- and S-wave velocity and density in the layers below and above the reservoir, and the electrical conductivity of the overburden are considered as random variables. Pre-stack seismic AVA data in a selected time window and real and quadrature components of the recorded electrical field are considered as data. We use Markov chain Monte Carlo (MCMC) sampling methods to obtain a large number of samples from the joint posterior distribution function. Using those samples, we obtain not only estimates of each unknown variable, but also its uncertainty information. The developed method is applied to both synthetic and field data to explore the combined use of seismic AVA and EM data for gas saturation estimation. Results show that the developed method is effective for joint inversion, and the incorporation of CSEM data reduces uncertainty in fluid saturation estimation, when compared to results from inversion of AVA data only