On Multi‐Model Assessment of Complex Degradation Paths: The Fate of Diclofenac and Its Transformation Products

Abstract

AbstractWe present a methodology to quantify the impact of model structure and parametric uncertainty on formulations targeting biotransformation processes of Emerging Contaminants in subsurface water resources. The study is motivated by recognizing that modeling of bio‐mediated reactions of recalcitrant compounds in soil and aquifers is plagued by uncertainty. At the same time, process‐based models often require the parameterization of complex physico‐chemical processes, a situation which is exacerbated by the paucity of direct observations. Thus, assessment and formulation of modeling tools capable to balance complexity and reliability is a key challenge. The modeling strategy proposed here aims at pairing and applying a suite of quantitative tools starting from a prior diagnosis of multiple uncertainty sources and leading to parameter estimation and model selection in the presence of a limited number of observations. The methodology is illustrated through application to a multi‐step, reactive scenario involving biotransformation of the pharmaceutical diclofenac (DCF) in groundwater. Our framework includes four plausible models. These are obtained through successive simplifications of a recently developed highly complex model. Such simplifications are accomplished consistent with the results of a comprehensive Multi‐Model Global Sensitivity Analysis. The latter allows ranking the levels of influence of system processes on model outputs by incorporating the effects of model formulation and parametric uncertainties. The kinetic of the loop‐initiating process (DCF nitrosation, linked to the temporal evolution of N‐cycle components) is documented as dominating in explaining the variability of model outputs of environmental interest. Model discrimination criteria suggest that a simplified counterpart of the reference model is favored to interpret available data. Our modeling approach can assist interpretation and prototyping of a wide range of contaminant biotransformation models. The latter is a key objective also for the purpose of developing credible (environmental) risk assessment tools and designing experimental sampling campaigns.