Bayesian Inference In CO2 Storage Monitoring: A Way To Assess Uncertainties In Geophysical Inversions

Abstract

Summary We present an integrated methodology for quantitative CO2 monitoring using Bayesian formulation. A first step consists in full-waveform inversion and CSEM inversion solved with gradient-based inverse methods. Uncertainty assessment is then carried out using a posteriori covariance matrix analysis to derive velocity and resistivity maps with uncertainty. Then, rock physics inversion is done with semi-global optimisation methodology and uncertainty is propagated with Bayesian formulation to quantify the reliability of the final CO2 saturation estimates.