Monday, April 12, 2010: 2:50 p.m.
Horace Tabor/Molly Brown (Westin Tabor Center, Denver)
Brian E. Caldwell, PG
,
Tetra Tech Inc., Oak Ridge, TN
Tiffany N. Downey, Ph.D.
,
Tetra Tech Inc., Phoenix, AZ
Patty Marajh-Whittemore
,
Southnavfacengcom, U.S. Navy, Jacksonville, FL
Complex interactions between chemicals of concern (COCs) and geochemical species become critically important when evaluating the success of active and passive in-situ remediation techniques. Site assessments and remedial actions, such as those undertaken at the Naval Air Station Pensacola Site 1, often rely on a broad suite of analyses to detect and characterize changes in groundwater quality as an indicator of remedial progress. The evaluation of the interdependencies and consequent elimination (via redundancy) of monitored analytes can result in significant savings over the course of the project.
Analytical results from ongoing natural attenuation groundwater monitoring at Site 1 were assessed to establish significant site-specific inter-species correlations. Using well-established statistical techniques such as Pearson’s linear relations and the multi-variate analytical technique of Principal Components Analysis (PCA), the significance of each currently monitored parameter was evaluated to their interdependencies and overall contribution relative to remedial progress.
A linear regression was performed on each pairing of 34 original variables, thus creating a matrix of approximately 600 unique correlation coefficients of which 19 were considered significant within a 95% confidence interval.
A total of nine PCA iterations were required to narrow the original dataset containing 34 unique analytes to an interdependent dataset containing 10 analytes. Each iteration eliminated independent analytes within the primary eigenvectors, and any analyte described by an insignificant eigenvector with an eigenvalue less than 1.
Pearson’s correlations consider only the relationship between any two analytes at a time, while PCA considers the interdependencies of all analytes simultaneously. Iterative PCA further restricts the relevant dataset to only those parameters which are correlated with multiple other analytes indicative of natural attenuation. The results of the iterative PCA suggest that analysis of only nine geochemical parameters, in addition to vinyl chloride, is necessary to monitor the progress of natural attenuation processes at Site 1.
