Data Analysis Methods for Chemical Fingerprinting of Environmental Media. Glenn W. Johnson and Robert Ehrlich, Energy and Geoscience Institute, Department of Civil and Environmental Engineering, University of Utah, Salt Lake City, Utah 84108
This paper presents considerations for chemical fingerprinting of multi-source media (groundwater, surface water, sediment) from the perspective of the project scientist. Given multiple contaminant sources in close proximity to each other, we commonly encounter co-mingled plumes and samples impacted by multiple sources. Given such a mixed system, chemical data from samples collected from such a system can be used to identify sources and establish the relative contribution from each source. Further, this can be accomplished with minimal a priori knowledge of source location and chemistry. Multivariate statistical methods have been developed that allow resolution of three parameters of interest in a mixed system: (1) the number of chemical fingerprints in the system, (2) the chemical composition of each fingerprint, and (3) the relative contribution of each fingerprint in each sample. In this paper, the rudiments of such statistical methods are presented, as well as brief case-studies where these methods have been applied to source-apportionment problems in different media.