Constructive epistemic modeling of groundwater flow with geological structure and boundary condition uncertainty under the Bayesian paradigm

摘要: Summary Constructive epistemic modeling is the idea that our understanding of a natural system through scientific model mental construct continually develops learning about and from model. Using hierarchical Bayesian averaging (BMA), this study shows segregating different uncertain components BMA tree posterior probability, prediction, within-model variance, between-model variance total serves as tool. First, probabilities permits comparative evaluation candidate propositions each component. Second, systemic dissection imperative for individual contribution component to prediction variance. Third, representation facilitates prioritization overall uncertainty. We illustrate these concepts using groundwater flow siliciclastic aquifer-fault system. consider four components. With respect geological structure uncertainty, we three methods reconstructing hydrofacies architecture system, two formation dips. boundary conditions, having propositions. Through combinatorial design, with their result in 24 base models. The analysis helps advancing knowledge rather than forcing fit particularly or merely several