Statistical rock physics inversion for assessing source rock properties from seismic signatures: an application to the Canning Basin, Australia

Jiayuan Huang, Allegra Hosford Scheirer, Tapan Mukerji

Quantifying source rock properties is essential for subsurface characterization but remains a high-dimensional and nonlinear inverse problem. A statistical rock physics inversion workflow is implemented to efficiently estimate source rock properties from seismic and well-log data and quantify associated uncertainty. A thermal-maturation-dependent rock physics model is calibrated through Monte Carlo simulation to link source rock parameters with elastic properties. Weighted Approximate Bayesian Computation (ABC) integrates prior petrophysical knowledge, model calibration errors, and measured elastic data to estimate posterior distributions of source rock properties. The workflow is first validated through well-log inversion, showing posterior updates in source rock properties. Then the workflow is applied to seismic inversion after outlier detection using a robust Mahalanobis distance method, generating spatially coherent 2D distributions of rock properties in the Goldwyer III of the Canning Basin, consistent with well-log observations. Sensitivity analysis identifies porosity, kerogen, and illite as the most influential parameters. The workflow provides a robust, uncertainty-aware framework for source-rock property estimation.

This manuscript has been accepted for publication in the Journal of Applied Geophysics. Please cite the published version: https://doi.org/10.1016/j.jappgeo.2025.106026