On uncertainty in remediation analysis: variance propagation from subsurface transport to exposure modeling
Introduction
Hazardous-waste sites with subsurface contamination are among the environmental problems that are foremost in public awareness. The fate of organic and metal contaminants in the subsurface environment is an issue that is relevant to risk assessments for uncontrolled hazardous waste sites [1] and for classification of wastes that must be placed in managed hazardous-waste Resource Conservation and Recovery Act (RCRA) facilities [2]. Determination of human health risk due to environmental contamination is a multi-step process that begins with the evaluation of present site-specific conditions. For example, the source of contamination must be characterized in terms of its type, history and characteristic transport mechanisms. The magnitude of the source, and the rate and mode of contamination are a few of the most fundamental pieces of information required. Addressing long-term, potential human exposures to buried wastes requires the use of models. For example, models are used to evaluate the transport of the contaminant in the environment and its partitioning between environmental media (e.g. soil gas, groundwater, and soil). Similarly, the interaction between the environmental media and the human exposure media (e.g. outdoor vs. indoor air, groundwater vs. tap water) can be projected through modeling. Human uptake and exposure can then be determined based on the various possible exposure pathways. Finally, a dose-response relationship must be established and applied. Because models must project soil and contaminant behavior into the future, there is much uncertainty associated with these predictions.
Current efforts to perform this series of evaluations for common subsurface contamination problems are hampered by the inadequacy of the tools used to integrate the human health risk assessment with the transport assessment. For instance, human health risk assessment is often performed separately from subsurface contaminant modeling. Typically, ranges of contaminant concentrations derived from either data or from simulations are not used in human exposure and health risk evaluations. Instead, such risk assessments rely on upper-bound static concentrations derived from either limited site data or from simple models that use conservative assumptions about contaminant distributions. A range of values is needed to reflect the inherent uncertainties introduced in the modeling efforts. Ideally, an integrated uncertainty analysis should be used in the assessment of contaminated sites, such that information is passed from subsurface transport modeling to exposure and risk calculations.
The importance of adequately characterizing variability and uncertainty in fate, transport, exposure, and dose-response assessments for human health and ecological risk assessments has been emphasized in several U.S. Environmental Protection Agency (EPA) documents and activities 3, 4, 5. In these documents, EPA makes clear that there are a number of situations in which a Monte Carlo analysis can be useful. For example, a Monte Carlo uncertainty analysis may be useful when screening calculations using conservative point estimates fall above the levels of concern. Other situations involve the need to rank exposures, exposure pathways, sites or contaminants; when the cost of regulatory or remedial action is high and the exposures are marginal; or when there is a need to rank the importance of uncertain parameters. The EPA further recommends a “tiered approach”. In a tiered approach, one begins with a fairly simple screening level model and progresses to more sophisticated and realistic (and usually more complex) models only as warranted by the findings and value added to the decision. If screening calculations show exposures or risks to be clearly below levels of concern or when the costs of remediation are low, quantitative characterization of uncertainty is clearly not warranted.
In previous work by James and Oldenburg [6], uncertainty analysis for a three-dimensional numerical simulation based on an actual case study of the transport of trichloroethylene (TCE) through a relatively thick vadose zone near a residential area was performed. The analysis included the propagation of uncertainty owing to variance in major transport parameter values as well as variations in the site conceptual model. As a result, predicted TCE contaminant concentrations were represented by mean soil gas and groundwater values along with standard deviations for any point in space from 1960 to the year 2360.
Pelmulder et al. [7] integrated regional scale aquifer transport to human exposure from ingestion and dermal absorption of contaminants in tap water, inhalation of vaporized contaminants while bathing, and ingestion of foods irrigated partially with well water. This work made use of a three-dimensional model of water transport in the saturated zone but did not include soil gas transport in the unsaturated zone.
The purposes of this paper are two-fold: (i) we demonstrate model integration including uncertainty analysis; and (ii) we investigate the sources of parameter uncertainty in an integrated subsurface contaminant transport and human health risk model for a particular case study. The integrated model involves passing results from the three-dimensional subsurface contaminant transport simulation and uncertainty analysis of James and Oldenburg [6] as input to a calculation of human exposure and health risk. Our calculation addresses exposure concentrations, human uptake rates, doses and resulting lifetime potential risk owing to model predictions of groundwater and soil gas TCE concentrations. Parameters used in health risk assessment are uncertain and thus are represented by distributions. By considering the combined uncertainty attributable to both the transport and exposure/risk input parameters, we obtain total human health risk as a distribution of values. The integration of uncertainty analyses also allows comparison of the relative importance of uncertainty arising from the subsurface contaminant transport and the human exposure and health risk modeling.
Section snippets
Background
The study in this paper stemmed from the development of a risk-based remediation analysis framework called SELECT at the Ernest Orlando Lawrence Berkeley National Laboratory (LBNL). In the SELECT framework, hazardous-waste site remediation scenarios are assessed in terms of health risk and cost. The SELECT framework combines state-of-the-art subsurface contaminant transport and human health risk models thus making it possible to treat the uncertainty of the integrated analysis.
Although the
Methods
This study involved the use of two models, procedures for integrating these models, protocols for developing probability distributions, methods for propagating uncertainties through the assembled models, and a process for ranking uncertainty importance for parameters used in the combined models. In the sections below, we summarize the methods used to propagate and rank uncertainties.
The case study
The case study modeled in this paper is based on an actual site in which it is estimated that TCE was disposed of into shallow trenches over a period of approximately 10 years beginning in 1960. Detailed site geological characterization determined the subsurface to be comprised of a thick vadose zone of interbedded clay, silt, and sands with the water table at a depth of about 25 m. Although TCE is only one of seven identified volatile organic compounds (VOCs) found at the site, TCE was chosen
Conclusions
In this paper, we propagate uncertainty from a complex subsurface contaminant transport simulation through a sophisticated human exposure model resulting in a stochastic description of human health risk. This initial attempt at integrating uncertainty analyses between two models has also enabled the evaluation of which parameters from both models (T2VOC and CalTOX) exert the greatest influence on final risk variance for a particular case study. The actual risk calculation is hypothetical
Acknowledgements
This work was supported by the Laboratory Directed Research and Development Program of Ernest Orlando Lawrence Berkeley National Laboratory under the U.S. Department of Energy, contract No. DE-AC03-76SF00098. Funding was also provided in part by the State of California to the University of California at Berkeley through the Cal-EPA Department of Toxic Substances Control (DTSC) Contract Agreements 95-T1050.
References (38)
- et al.
Uncertainties in health-risk assessment: an integrated case study based on tetrachloroethylene in California groundwater
Regul. Toxicol. Pharmacol.
(1992) - U.S. Environmental Protection Agency, (U.S. EPA), Risk assessment guidance for superfund. Volume I Human Health...
- U.S. Environmental Protection Agency, (U.S. EPA), Test methods for evaluating solid waste (SW-846) physical/chemical...
- U.S. Environmental Protection Agency (U.S. EPA), Guidelines for carcinogenic risk assessment. Federal Register, Vol....
- U.S. Environmental Protection Agency (U.S. EPA), Guidelines for exposure assessment. Federal Register, Vol. 57, Issue...
- U.S. Environmental Protection Agency, (U.S. EPA), Summary report for the Workshop on Monte Carlo Analysis. U.S. EPA...
- et al.
Linear and Monte Carlo uncertainty analysis for subsurface contaminant transport simulation
Water Resour. Res.
(1997) - et al.
Regional scale framework for modeling water resources and health risk problems
Water Resour. Res.
(1996) - Falta, R. W., Pruess, K., Finsterle, S. & Battistelli, A., T2VOC User's Guide. Lawrence Berkeley Laboratory Report,...
- McKone, T. E., CalTOX, A multimedia total-exposure model for hazardous-wastes sites. Part I: Executive summary....
Numerical modeling of steam injection for the removal of nonaqueous phase liquids from the subsurface 1. Numerical formulation
Water Resour. Res.
Numerical modeling of steam injection for the removal of nonaqueous phase liquids from the subsurface 2. Code validation and application
Water Resour. Res.
Cited by (17)
Site-scale groundwater pollution risk assessment using surrogate models and statistical analysis
2024, Journal of Contaminant HydrologyAssessing human health risk of groundwater DNAPL contamination by quantifying the model structure uncertainty
2020, Journal of HydrologyCitation Excerpt :In T2VOC, hydrodynamic dispersion is neglected and molecular diffusion is simulated for the gas phase only (Basu et al., 2008; Falta et al., 1995). Mass transfer is depicted by advection in all phases (Bennett et al., 1998; James and Oldenburg, 1997). Fluid phases are assumed to be in local chemical equilibrium for each element, so that interphase mass transfer among all phases can be described by equilibrium partitioning relations (Gudbjerg et al., 2005; Neriah and Paster, 2019).
Quantifying uncertainty in environmental health models
2019, Cost-Benefit Analysis of Environmental Health InterventionsTemporal evolution of DNAPL source and contaminant flux distribution: Impacts of source mass depletion
2008, Journal of Contaminant HydrologySensitivity of key factors and uncertainties in health risk assessment of benzene pollutant
2007, Journal of Environmental SciencesAn integrated fuzzy-stochastic modeling approach for risk assessment of groundwater contamination
2007, Journal of Environmental Management