Direct Bayesian seismic inversion for porosity estimation in a hard rock carbonate reservoir

• DOI: 10.3997/2214-4609.202112913
• Publication Date: 2021-09-29T06:06:02Z
• Authors: A. Heidari; N. Amini; T. Mejer Hansen; H. Amini; M. Emami Niri

We used a probabilistic inversion approach to quantify the porosity distribution and its uncertainty for a hard rock carbonate reservoir in Southwest Iran. We embedded a calibrated rock-physics model based on Nur's critical porosity model in the inversion engine to directly invert the post-stack seismic data into porosity. In contrast to the common practice in the probabilistic seismic inversion approaches, where the uncorrelated white noise is considered as data uncertainty, we assumed a correlated Gaussian noise model in the inversion setup. Assessment of the posterior mean, as well as the facies probability sections, indicates that the inversion is successful in resolving some thin layers and geological details within the key reservoir intervals. As well as porosity and facies, we converted the posterior mean model to critical porosity via the proposed rock-physics template. The statistical analysis of the porosity realisations at four well locations highlights the intervals with a mismatch with the measured porosity logs, which can be attributed to imperfect well-to-seismic ties, presence of shale, and inversion inability to resolve thin layers. The results also confirm that the inversion parametrizations such as prior information and noise model assumption were appropriate and representative of the reservoir properties and data uncertainties.