Transuranic (TRU) radionuclides buried 25 years ago in shallow unlined disposal trenches in a fractured shale saprolite had been detected in groundwater from downgradient monitoring wells and in surface water seeps. Field observations had suggested the actinide radionuclides were mobilized by natural organic matter (NOM) and rapidly transported with little retardation. A 73-day natural gradient tracer experiment injected trivalent lanthanides (Nd and Eu) as analogues to determine the mechanisms and rates of actinide transport at the field scale. Adsorption isotherms for 241Am and Eu with saprolite from the site confirmed a very high affinity for adsorption (R > 50 000) in the absence of NOM. However, reactive and nonreactive tracers arrived at approximately the same time along a 10-m long deep flow path, and anion-exchange chromatog raphy and filtration suggested that the mobile lanthanides in groundwater were a NOM complex. Although flow through a shallow flow path was intermittent, reflecting transient recharge events, large storms resulted in coincident peaks of both reactive and nonreactive tracers, suggesting that they migrated at similar rates over distances of 78 m. We conclude that NOM facilitated the almost-unretarded transport of lanthanide tracers and, by analogy, that NOM is facilitating the mobilization and rapid migration of the TRU radionuclides.

A 2D equivalent porous media (EPM) model was used to simulate transport of tritium for a field‐scale tracer experiment in a fractured and highly weathered shale saprolite. The tritium plume was characterized by rapid migration of the leading edge of the plume (up to 0.4 m/day), slower movement of the center of mass of the tritium pulse (0.009 m/day) and very slow decline of concentrations in the “tail” of the breakthrough curve (which has persisted at the site for at least 16 years). The EPM model successfully described the shape of the plume and the breakthrough curves for a monitoring well 3.7 m downgradient of the injection well using a flow velocity of 0.01 m/day and longitudinal and transverse dispersivity values of 0.8 m. An unusually low ratio of longitudinal and transverse dispersivity (αL/αT= 1) was needed to fit the nearly circular shape of the plume, which is believed to be caused by the water‐table slope being perpendicular to the orientation of the prominent bedding plane fractures (which are expected to correspond to Kmax). Simulated values for concentrations in the long “tail” of the breakthrough curve observed in a downgradient well were especially sensitive to the value of longitudinal dispersivity used. The best‐fit simulation, based on data over a 5 year period, was extrapolated to the most recent data point (16 years after the start of the injection) and the simulated concentration was very close to the measured value. Model predictions made with a slightly different value of longitudinal dispersivity resulted in very large errors at late time, indicating that duration of monitoring data (which strongly influences the fitting of longitudinal dispersivity) is a critical factor in accurate prediction. The experiment and simulations show that contaminant plumes can persist for long periods of time in fractured porous materials, presumably due to diffusive exchange between the rapidly moving water in the fractures and the relatively immobile pore water in the fine‐grained matrix.

No abstract