Multi-Scale Assessment of Prediction Uncertainty in Coupled Reactive Transport Models Conducted at the Florida State University

Abstract

IntroductionThis final project report summarizes research activities, findings, and deliverables obtained at the Florida State University (FSU) during the funding period of 2009 -2013. The goal of this project is: (1) to seek fundamental understanding of the factors causing and controlling predictive uncertainty in groundwater reactive transport modeling, and (2) to develop interdisciplinary tools for quantifying and reducing the predictive uncertainty. Uncertainty analysis in this project is focused on parametric uncertainty of individual models and model uncertainty between alternative models. The FSU project team was in charge of developing methods for uncertainty quantification and reduction. The methods have been applied to quantify predictive uncertainty in uranium reactive transport modeling at the Naturita Site.During this project period, we published twelve peer-reviewed journal articles (one is under revision and one under review) and four conference proceedings. In particular, the paper of published in Water Resources Research was selected as the Editor's Highlight for its new insights into faster computation of uncertainties. We presented our research results in multidisciplinary conferences such as geology, mathematics, geochemistry, and civil engineering. The publication on the SIAM News (Hill et al., 2012) introduced our interdisciplinary research results to members of the Society of Industrial and Applied Mathematics (SIAM). A master student (Geoffery Miller) and a doctoral student (Dan Lu) graduated with support of this project. The doctoral student received the 2012 Student Travel Fellowship and presented her research at the 2012 annual PI meeting; she is working as a post-doc at the Oak Ridge National Laboratory. We also collaborated with colleagues in the Los Alamos and Oak Ridge National Laboratories to help their research using the methods developed in this project. The collaboration produced one journal article (Dai et al., 2012) and one conference proceeding . Scientific Questions and Project FindingsThis project explored the scientific questions below related to various aspects of uncertainty analysis in coupled groundwater reactive transport models. Question 1: How does uncertainty behave for groundwater reactive transport models?This question was tackled by using the surface complexation model developed by Kohler et al. (1996). The model, denoted as C4 hereinafter, has two functional groups. The weak site, S 1 OH, is associated with one reaction: where S1 and S2 denote the adsorption sites. The uncertain parameters are the three equilibrium constants, logK1, logK2, and logK3 of the three reactions, respectively, and the fractions of the two sites. Since the fractions sum to one, the fraction of the strong site, logSite, is explicitly treated as an uncertain variable. Among the seven column experiments conducted by Kohler et al. (1996) to understand uranium reactive transport, three experiments, denoted as Expt_1, Expt_2, and Expt_8, were used in the analysis. They were performed unde...