Space-Time Distribution of Trichloroethylene Groundwater Concentrations: Geostatistical Modeling and Visualization.

Area of Science:

• Geostatistics
• Environmental Science
• Hydrogeology

Background:

• Groundwater contamination poses significant risks to drinking water supplies.
• Large-scale contaminant plumes, like the trichloroethylene (TCE) site in Michigan, require sophisticated modeling for effective management.
• Understanding the space-time dynamics of contaminant distribution is crucial for remediation efforts.

Purpose of the Study:

• To develop and demonstrate a geostatistical methodology for modeling and visualizing the space-time distribution of groundwater contaminants.
• To apply this approach to a large TCE contamination plume in northern Michigan.
• To assess the predictive performance of different space-time models.

Main Methods:

• Projection of data into a new space using multidimensional scaling to handle non-stationary spatial covariance.
• Fitting non-separable space-time models to semivariogram data.
• Comparison of model performance using cross-validation.
• Interpolation of TCE concentrations onto a grid and visualization in a 3D space-time cube.

Main Results:

• The sum-metric and product-sum semivariogram models demonstrated comparable prediction accuracy, with a mean absolute error of 23% of the mean TCE concentration.
• The geostatistical model successfully visualized the plume's expansion and movement over time.
• TCE concentrations were observed to increase in the northern part of the study area, indicating plume migration towards the Chain of Lakes.

Conclusions:

• The proposed geostatistical approach is effective for modeling and visualizing complex space-time groundwater contamination.
• The method provides accurate predictions of contaminant concentrations, aiding in the understanding of plume dynamics.
• The visualization highlights the temporal evolution and spatial spread of the TCE plume, crucial for environmental management and remediation strategies.