The Ensemble Kalman Filter for Groundwater Plume Characterization: A Case Study.

The Kalman filter is an efficient data assimilation tool to refine an estimate of a state variable using measured data and the variable's correlations in space and/or time. The ensemble Kalman filter (EnKF) (Evensen 2004, 2009) is a Kalman filter variant that employs Monte Carlo analysis to define the correlations that help to refine the updated state. While use of EnKF in hydrology is somewhat limited, it has been successfully applied in other fields of engineering (e.g., oil reservoir modeling, weather forecasting). Here, EnKF is used to refine a simulated groundwater tetrachloroethylene (TCE) plume that underlies the Tooele Army Depot-North (TEAD-N) in Utah, based on observations of TCE in the aquifer. The resulting EnKF-based assimilated plume is simulated forward in time to predict future plume migration. The correlations that underpin EnKF updating implicitly contain information about how the plume developed over time under the influence of complex site hydrology and variable source history, as they are predicated on multiple realizations of a well-calibrated numerical groundwater flow and transport model. The EnKF methodology is compared to an ordinary kriging-based assimilation method with respect to the accurate representation of plume concentrations in order to determine the relative efficacy of EnKF for water quality data assimilation.

Continuous Improvement of a Groundwater Model over a 20-Year Period: Lessons Learned.

Groundwater models developed for specific sites generally become obsolete within a few years due to changes in: (1) modeling technology; (2) site/project personnel; (3) project funding; and (4) modeling objectives. Consequently, new models are sometimes developed for the same sites using the latest technology and data, but without potential knowledge gained from the prior models. When it occurs, this practice is particularly problematic because, although technology, data, and observed conditions change, development of the new numerical model may not consider the conceptual model's underpinnings. As a contrary situation, we present the unique case of a numerical flow and trichloroethylene (TCE) transport model that was first developed in 1993 and since revised and updated annually by the same personnel. The updates are prompted by an increase in the amount of data, exposure to a wider range of hydrologic conditions over increasingly longer timeframes, technological advances, evolving modeling objectives, and revised modeling methodologies. The history of updates shows smooth, incremental changes in the conceptual model and modeled aquifer parameters that result from both increase and decrease in complexity. Myriad modeling objectives have included demonstrating the ineffectiveness of a groundwater extraction/injection system, evaluating potential TCE degradation, locating new monitoring points, and predicting likelihood of exceedance of groundwater standards. The application emphasizes an original tenet of successful groundwater modeling: iterative adjustment of the conceptual model based on observations of actual vs. model response.

A post audit of a model-designed ground water extraction system.

Model post audits test the predictive capabilities of ground water models and shed light on their practical limitations. In the work presented here, ground water model predictions were used to design an extraction/treatment/injection system at a military ammunition facility and then were re-evaluated using site-specific water-level data collected approximately one year after system startup. The water-level data indicated that performance specifications for the design, i.e., containment, had been achieved over the required area, but that predicted water-level changes were greater than observed, particularly in the deeper zones of the aquifer. Probable model error was investigated by determining the changes that were required to obtain an improved match to observed water-level changes. This analysis suggests that the originally estimated hydraulic properties were in error by a factor of two to five. These errors may have resulted from attributing less importance to data from deeper zones of the aquifer and from applying pumping test results to a volume of material that was larger than the volume affected by the pumping test. To determine the importance of these errors to the predictions of interest, the models were used to simulate the capture zones resulting from the originally estimated and updated parameter values. The study suggests that, despite the model error, the ground water model contributed positively to the design of the remediation system.