Multivariate Analyses to Improve Understanding of NAPL Pollutant Sources

Abstract

Current site assessment techniques do not always generate adequate information regarding the presence, type, or distribution of nonaqueous phase liquids (NAPLs) at sites with ground water contamination. Without this information, however, the design of remediation methods is uncertain, often resulting in costly and/or failed attempts to minimize risks associated with the contamination. In this work, it is proposed that a thorough multivariate analysis of data from ground water sampling efforts could improve our overall understanding of these sites. Multivariate analyses can provide considerable insight into the contaminant source characteristics by elucidating correlations in ground water concentrations that identify recurring chemical patterns or “signatures.” These correlations are related to the type of NAPL and proximity to the contaminant source. Multivariate plots and a principal components analysis (PCA) are used to interpret ground water data from a manufactured gas plant site in Iowa contaminated with both gasoline and coal tar. Conclusions from these analyses regarding the distribution of NAPL contaminants were generally consistent with those derived from direct physical evidence of the NAPL sources. The multivariate analyses, however, provide an additional level of interpretation regarding the distribution of coal tar in the subsurface that was not possible with the standard evaluation techniques used during the remedial investigation (RI). Comparison of chemical signatures of ground water samples among wells identified two distinct regions of coal tar contamination, and suggests which wells are impacted by each source providing greater confidence in the location of the DNAPL sources.