A comparison of multivariate statistical analysis and the use of an indicator ion for the interpretation of water quality data

The physical, chemical and biological factors that contribute to control the extent and character of ground water contamination at acid mine drainage sites are complex. Quantitative methods are needed to evaluate the various hydrogeologic and hydrochemical conceptual models required to interpret the details of water flow and quality variations. Indicator ions frequently are selected to characterize water quality data. This study shows that, in complex flow systems, multivariate statistical techniques constitute a significant improvement over the indicator ion approach for the purpose of identifying chemical processes that control the water quality over the full spectrum of observed water quality.;Cluster analysis, stepwise multivariate analysis of variance (MANOVA), canonical and discriminant analyses constitute the statistical analyses utilized. Cluster analysis identifies groups of underground sites that have internally consistent water quality characteristics. Subsequently, hydrochemical variables are identified that best differentiate among the groups.;The indicator ion approach identifies the zones of acid production, and describes the spatial and temporal variation of poor quality water. The use of zinc ion concentration and water discharge rate facilitates the calculation of metal load discharging from the acid producing zones. The multivariate statistical approach quantifies the full diversity of variation in water chemistry. In addition to identifying the acid producing zones in the mine, the statistical techniques identify subtle hydrochemical variations. The statistical techniques, however, are not an end in themselves. Instead, the statistical approach constitutes a tool for organizing and evaluating simultaneously the information carried by many related hydrochemical variables. The interpretation of the statistical results must be made in the context of professional judgement about the hydrochemical and hydrogeological system that generated the observed data.;A comparison of the utility of using an indicator ion to the utility of using selected multivariate statistical analyses is presented. Water discharge and quality data collected for a three year period at the acid producing Bunker Hill Mine near Kellogg, Idaho are used in the analyses. The spatial and temporal distributions of water quality and metal loading are analyzed.