In 2008, stochastic parameter uncertainty was incorporated into the TEAD model to better address the potential range of future plume migration. Probability distribution functions for model parameters were determined based on field data and prior studies. Monte Carlo analysis propagated uncertainty in model parameters to predictions of TCE transport. However, few realizations of model parameter values resulted in reasonable model calibration, calling into question the range and likelihood of predictions.
In 2009, the modeling analysis utilized PEST to (a) improve model calibration and (b) constrain the Monte Carlo analysis using the calibration data. The use of advanced features such as regularization and singular-value decomposition significantly reduced the statistical model-to-measurement calibration error. Regulatory priorities were addressed by adjusting observation weights in areas of potential concern. Also, the application of null-space Monte Carlo analysis in PEST improved the uncertainty analysis by automatically transforming parameter sets to achieve calibration. The application of these methods to the TEAD model resulted in a more complete assessment of possible plume migration than was possible with previous deterministic and stochastic models.