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 petrophysical properties and potentials of source rocks is important for subsurface modeling and characterization. However, predicting these properties using seismic signatures and well-log information is a high-dimensional, nonlinear inverse problem, and is subject to uncertainty due to data ambiguities. In this study, a statistical rock physics inversion workflow is proposed to efficiently estimate source rock properties and quantify their uncertainty from seismic signatures. A thermal maturation dependent elastic rock physics model is implemented to link source rock properties with elastic properties by Monte Carlo calibration. Statistical rock physics inversion based on weighted Approximate Bayesian Computation (ABC) is proposed to combine prior information from petrophysical knowledge, rock physics model calibration error, measured elastic properties data from well log and seismic data to estimate posterior distributions of source rock properties efficiently.