MODELING IN THE FACE OF UNCERTAINTY: PAIRING NOVEL DATA STREAMS WITH REACTIVE TRANSPORT SIMULATIONS TO ENHANCE SCIENTIFIC UNDERSTANDING AND SUPPORT MANAGEMENT DECISIONS

From managing legacy contaminant plumes to mitigating saltwater intrusion, addressing modern groundwater challenges often requires a detailed understanding of the complex tapestry of chemical reactions that occur in the subsurface. Over the past 50 years, advances in reactive transport codes – computational tools that simulate both the physics of groundwater flow and the chemistry of water-rock-microbe reactions – have enabled the simulation of increasingly complex reaction networks at high resolution and across broad scales. However, an increase in model complexity comes with its own set of challenges; constraining simulation output to observed data, parameterizing large reaction networks and interpreting model results all become more difficult as model complexity increases.

Through two case studies, this presentation will show that these challenges can be overcome by pairing reactive transport simulations with novel data streams – such as in situ sensors and hydrogeophysical surveys – and by interrogating model output using modern data science techniques. In the first example, I pair in situ sensor data with stochastic reactive transport simulations to identify mechanistic controls on redox cycling in high-elevation floodplains. Sifting through a complex reaction network, global sensitivity analyses reveal that certain reaction rate thresholds must be surpassed in order to reductively immobilize contaminants in these environments. In a second example, I use geophysical surveys to understand the impacts of subsurface heterogeneity on nitrate leaching during managed aquifer recharge. Uncertainty quantification reveals that, even at an intensively studied field site, projected nitrate loads exhibit order-of-magnitude uncertainty, which may hamper management decisions and limit recharge implementation. These examples underscore the potential for stochastic reactive transport simulations to enhance process understanding and support decision-making when paired with the appropriate data and model analysis techniques.