In-place management of contaminated sediments has the potential to provide an important alternative and complement to sediment removal. Despite the development and increasing availability of technologies on the market, their efficient application is hindered by uncertainties from a lack of scientifically substantiated causal relationships. While there is a substantial body of reference material on field performance for such technologies, in order to integrate and generalize those lessons in practical decision making at new sites, basic scientific relationships and uncertainties must be better quantified. For example, how is the system (sediment with monitored natural recovery or cap) response controlled by its biogeochemical and hydrodynamic properties? In addition, the ultimate effects of system characteristics on the pertinent risk end-points are not known in the context of in-place sediment remediation. These questions have not yet been addressed in an integrated fashion, with all components evaluated within the same system, from bench scale to field scale. The propagation of uncertainties associated with transitioning bench-scale measurements to field conditions needs to be quantified.

The overall objective of this project was to characterize and bound the uncertainties associated with the impact of sediment processes on the long-term performance of in situ capping strategies.

Surface sediment processes that affect natural attenuation and in situ remediation

This project included a combination of experimental work and modeling to enable evaluation of the impact of ebullition and advection on sediment bed stability and contaminant fluxes (in this case polycyclic aromatic hydrocarbons [PAH]) from the sediment. The experiments were focused on the quantification of ebullition metrics, PAH flux measurements, and sediment resuspension measurements conducted in batch systems and flume configurations using sediments collected from the Anacostia River capping project. These site-specific data were aimed at narrowing the process uncertainties as they are currently reported in the literature. A geostatistical model was developed and applied to enable comparison of microbial data collected at the field and laboratory scales. The integrative modeling approach evolved from a generic water quality model to a custom-designed sediment flux model (SFM) that allowed for the integration of data collected at various scales to link the flume and field aspects of the study.

For the Anacostia River, findings from site-specific field and laboratory studies and the uncertainty analysis conducted with the SFM have shown that microbial gas-generated ebullition is likely not a significant process affecting contaminant flux through cap layers at the site, while groundwater seepage is likely to be an important long-term process in areas where there is a net advective flux of water from the sediment bed to the overlying water column. The current findings suggest that additional site characterization of the Anacostia River capping site should consider studies aimed at reducing the uncertainty and spatial variability in groundwater seepage rates. Monte Carlo-based predictions of PAH flux obtained for the sensitivity and uncertainty analyses provide valuable insight into the relative importance and sensitivity of model input parameters and the impact of the basis for value distributions.

The outcome of these modeling analyses demonstrates that: (1) Model-predicted fluxes are strongly dependent on multiple process rates and physical parameters. Therefore, there is a clear need to evaluate the combined effect of input parameter uncertainties on uncertainties in expected fluxes and cap performance; (2) Partitioning and other chemical-specific characteristics of a particular contaminant play a critical role in determining the extent to which that contaminant can be mobilized and transported to the sediment-water interface; (3) Significant reductions in input parameter uncertainties can be expected by incorporating site-specific field studies and in situ/laboratory experiments, as opposed to relying exclusively on ranges available in literature for key process parameters; (4) Laboratory-based data can be used to reduce the spatial estimation variance of sediment attributes; (5) Significant (i.e., orders of magnitude) reductions in mean/median and interquartile range (IQR) for predicted total PAH flux can be obtained when relying on site-specific data sets to reduce parameter uncertainty for the Anacostia River; and (6) An integrated capping model framework, such as the SFM, can be used to effectively compare relative performance across multiple capping technologies, as well as to evaluate the effect of combined parameter uncertainties on predicted contaminant flux from the sediment bed to the water column.

The combination of targeted site-specific experimental and generalizable modeling tools represents a step forward in terms of tools available for evaluating the performance of various capping technologies. The SFM developed in this project contains a similar level of sediment process complexity embodied within other available models, but is a user-friendly tool that is specifically designed to simulate the effect of capping actions on contaminant flux. Particularly important and unique to the model structure is the ability to simulate multiple “sediment” materials, which allows for an explicit representation of the physical properties not only of the parent bed material but also sand cap, AquaBlok, coke breeze, and potentially other cap materials. (Project Completed – 2009)