Data assimilation-based inversion of rock rheological parameters in fault zones: A case study from a hydropower site in southwest China

Abstract

This study tackles the uncertainty and equifinality challenges in estimating heterogeneous rheological parameters and characterizing complex rheological behavior for rock masses in fault-influenced zones of a hydropower project in Southwest China, using advanced data assimilation (DA) methods. High-quality observational data were obtained from laboratory triaxial rheological tests and served as the foundation for subsequent model calibration. Heterogeneity within the rock mass was identified and partitioned into subregions using image-processing techniques, enabling localized parameter updates. Both Iterative Local Updating Ensemble Smoother (ILUES) and the Ensemble Smoother with Multiple Data Assimilation (ESMDA) were employed to invert rheological parameters for the homogeneous model. Uncertainty and correlation of posterior parameters were assessed. The results demonstrate that ILUES significantly outperforms ESMDA in this context, providing more accurate and reliable estimations of rheological behavior and a clearer representation of uncertainties. The integration of ILUES with image-based heterogeneity modeling offers a robust framework for addressing the challenges of equifinality and high dimensionality in rheological parameter estimation of fault-influenced rock masses. This work contributes to a deeper understanding of rheological processes in fault zones and offers scientific support for the stability assessment of hydropower infrastructure in complex geological conditions.