Analysis of Water Quality Data to Locate Contaminant Sources

An innovative method is presented and demonstrated for analyzing spatially distributed water quality measurements to identify persistent sources of contamination within a waterbody. The method consists of first separating the monitoring stations into clusters that consist of stations whose measurements are substantially and significantly correlated as measured by the Spearman correlation coefficient and then applying principal component analysis to the correlated stations to quantify the relative magnitudes of the coherent concentration fluctuations within each cluster. Stations with higher relative magnitudes within each cluster are inferred to be closer to the contaminant sources that influence the contaminant concentrations within the cluster. The proposed procedure is demonstrated using a 10-year data set of synoptic monthly measurements of nutrient species, chlorophyll-a, and fecal indicator bacteria.