ABSTRACT
Hydrologic exchange between river channels and adjacent subsurface environments is a key process that influences water quality and ecosystem function in river corridors. High-resolution numerical models were often used to resolve the spatial and temporal variations of exchange flows, which are computationally expensive. In this study, we adopt Random Forest (RF) and Extreme Gradient Boosting (XGB) approaches for deriving reduced order models of hydrologic exchange flows and associated transit time distributions, with integrated field observations (e.g., bathymetry) and hydrodynamic simulation data (e.g., river velocity, depth). The setup allows an improved understanding of the influences of various physical, spatial, and temporal factors on the hydrologic exchange flows and transit times. The predictors also contain those derived using hybrid clustering, leveraging our previous work on river corridor system hydromorphic classification. The machine learning-based predictive models are developed and validated along the Columbia River Corridor, and the results show that the top parameters are the thickness of the top geological formation layer, the flow regime, river velocity, and river depth; the RF and XGB models can achieve 70% to 80% accuracy and therefore are effective alternatives to the computational demanding numerical models of exchange flows and transit time distributions. Each machine learning model with its favorable configuration and setup have been evaluated. The transferability of the models to other river reaches and larger scales, which mostly depends on data availability, is also discussed.
ABSTRACT
Model-based decision making is commonly used in performance assessments to assure water resource protection for both human health and the environment for hundreds of years into the future. To make decisions regarding aquifer protection against potential contamination, a conceptual site model (CSM) describing the hydrodynamic behavior needs to account for subsurface heterogeneities in sufficient detail. When site-specific data are sparse, larger-scale geologic descriptions are adopted with the consequence of losing small-scale features (at the cm scale) that can control contaminant transport. In this study, a multiple lines of evidence approach is used to construct vadose zone CSMs based on an evaluation of several types of data, including geologic logs, borehole moisture content and concentration data, geophysical spectral gamma logging data, and groundwater concentration data for a tank farm at the Hanford Site in southeastern Washington State. The resulting CSMs of the unsaturated zone represent a synthesis of what is known about flow and transport processes at the site-scale and maintain consistency with knowledge that has been accumulated at the regional scale. Through a process of extensive data analyses, a systematic approach is described to create an evidence base that supports the evaluation and development of CSMs. Numerical models are then used to evaluate the impact that smaller-scale heterogeneities have on contaminant transport through the vadose zone for a performance assessment on waste tank closure. Together, the field data and the numerical experiments suggest that although small-scale features close to source releases can have an impact on horizontal spreading, overall there is a relatively minor impact on transport for the site under study as evaluated by differences in peak fluxes and arrival times for historical leak events, and for potential releases resulting from waste tank closure. Use of alternative CSMs, developed through careful examination of available characterization and monitoring data, provides confidence that geologic heterogeneities do not impact contaminant transport behavior significantly enough to alter the assessment of risk for closure at this site.