Summary

The identification of chemical contaminant sources is a common problem in environmental forensic investigations. Successful inference of sources depends on sampling plan design, sample collection procedures, chemical analysis methods, and knowledge of historical industrial processes in the study area. However, in complex situations where multiple sources contribute similar types of contaminants, even careful project planning and design may not be enough. If sources cannot be linked to a unique chemical species (i.e., a tracer chemical), then mapping the distributions of individual contaminant concentrations is insufficient to infer source. If, however, a source exhibits a characteristic chemical fingerprint defined by diagnostic proportions of a large number of analytes, source inference may be accomplished through analysis of multiple variables; that is, through use of multivariate statistical methods. The objective of a multivariate approach to chemical fingerprinting is to determine (1) the number of fingerprints present in the system, (2) the multivariate chemical composition of each fingerprint, and (3) the relative contribution of each fingerprint in each collected sample.