Bayesian inversion for porosity and saturation using frequency-dependent rock physics models

The reliable estimation of gas saturation from seismic data is a difficult and longstanding problem. Rocks saturated with gas often show high attenuation and frequency-dependence of seismic properties, and previous work has indicated that such effects can be detected by considering frequency-dependence of reflectivity. Theoretical studies show that such analysis has the potential to improve estimates of gas saturation. This paper develops a Bayesian inversion scheme which allows probability distributions for porosity and saturation to be derived from pre-stack seismic data. We apply the method to well-log and 3D seismic data from the Vienna Basin. Analysis at the calibration well demonstrates the potential power of the method, and probability distributions for other zones of interest are derived.