UC Davis

Groundwater contaminant transport models rely on parameters that are often difficult or time-consuming to measure, generating uncertainty that complicates interpretation and decision-making. We apply global sensitivity analysis (GSA) to identify the parameters and interactions that most strongly control laboratory-scale reactive transport. Numerical simulations of one-dimensional, fully water-saturated systems governed by advection–dispersion–reaction transport of a single solute with sorption and decay are performed that systematically vary porosity, bulk density, dispersivity, decay rate, sorption distribution coefficient, and desorption rate. We employ GSA in three complementary ways to interrogate reactive transport behavior. First, we evaluate parameter influence across key breakthrough times (early arrival, peak concentration, and late-time tailing), offering a temporally resolved view of transport dynamics; as well as value of peak concentration to capture the worst case exposure. Second, we quantify both individual and interactive parameter effects, revealing how coupled processes control breakthrough behavior and how interaction-driven variance increases as the contaminant plume evolves in time. Third, we interpret sensitivity patterns within dimensionless transport regimes defined by Péclet and Damköhler numbers, allowing parameter importance to be evaluated across distinct advective–reactive conditions and facilitating comparison across systems and scales. Results show that, for typical laboratory-scale systems with constant pore velocity, sorption emerges as a consistently dominant driver across all scenarios. As the contaminant plume ages, additional processes and increasingly complex parameter interactions contribute to output variance. Decay dominates in reaction-dominant regimes and when evaluating parameter influence on the magnitude of peak concentration, whereas dispersion can unexpectedly dominate in advection-dominant regimes. By combining temporal resolution, interaction analysis, and regime-based classification, this work expands on how uncertainty propagates through contaminant transport models and provides practical guidance for prioritizing parameter measurements that most effectively improve predictive reliability.