Uncertainty analysis using the multimedia environmental pollutant assessment system (MEPAS) computer algorithm for contaminant transport modelling through a groundwater system

To quantify the uncertainty associated with modelling the migration of a contaminant out of a waste site, through the ground water system and into a local surface water body, four steps were involved. These four steps included; sensitivity analysis to identify the input parameters causing the greatest variance in the calculated risk, research distributions for these parameters to propagate through the MEPAS model, uncertainty analysis to determine the range of the risk results for a waste site, and compare uncertainty analysis results with independent risk assessments to determine if the uncertainty associated with the MEPAS model is within acceptable limits. The sensitivity analysis was performed in two parts due to computer model limitations. The SENS 1.x model was used to create over 1000 independent scenarios of the input parameters. From the statistical analysis of the results, it was concluded that specific parameters could not be identified as being the most sensitive because of variance among contaminants. Therefore, the most sensitive groups of parameters were chosen. These groups included subsurface adsorption coefficients, layer thickness, saturated hydraulic conductivity, bulk density, field capacity, dispersitivities, effective porosity, contaminant velocity, travel distance in the saturated zone to receptors, mean monthly temperature and mean monthly precipitation. Although the percentage of the variance in the risk results changed depending on the contaminant being modelled, these groups could account for at least 20% of this variance. Many researchers have been involved in defining a majority of these parameters for the site of concern, WAG 5, and contaminants of concern, ⁹⁰Sr, ¹³⁷Cs, and PCBs (Aroclor 1254 and 1260). These individuals were interviewed and an extensive literature search was conducted to produce distributions for each of these parameters. Upon propagating these distributions through the MEPAS model, the uncertainty ranges for the chemicals of concem were from 4 to 7 orders of magnitude. These ranges can be directly related to the travel distance through the saturated zone and the adsorption coefficients, K_d. The variance in the travel distance proved to be responsible for 79.0% of the uncertainty in the public health risks. Even though the subsurface adsorption coefficients did effect the uncertainty as greatly, it is shown that the K_d values influence the uncertainty in the calculated risks. The retardation factor, which is greatly effected by the K_d value, can be used to estimate a maximum plume travel distance. For example, the maximum distance that the ⁹⁰Sr plume could travel is ft. Even with this uncertainty, it is confirmed that the baseline risk results based on PEIS methodology and the Bechtel remedial investigation risks were accurate. Both results fell within the upper portion of the uncertainty ranges.