Multivariate analysis in hydrochemistry: an example of the use of factor and correspondence analyses

ABSTRACT Statistical‐based, semiquantitative techniques [Factor Analysis (FA) and Correspondence Analysis (CA)] were applied to chemical data from two aquifers to assess the usefulness of such techniques at the initial stages of hydrochemical studies. For the first case, both FA and CA defined three factors to account for the major ion variations observed in the Milk River aquifer (Alberta, Canada), namely, sulfate reduction occurring in the recharge area and mixing of waters of different ages in the deeper portions of the aquifer, ion‐filtration processes, and ion exchange on clays. For the FA, the comparison between varimax and oblique rotations led to similar interpretations. For the second study area, the San Pedro River basin (Arizona, U.S.A.), FA (Q‐mode) extracted two factors, which helped segregate waters from the two aquifers in the area. The CA, applied only to samples from the confined aquifer, drew two factors which were related to dissolution of gypsum coupled with ion exchange (softening) as the main mechanisms operating in the aquifer, along with a hypothetical presence of fluorite in the aquifer sediments to justify the rather high [F‐] observed in local ground waters. Comparisons made between the results presented and the findings of earlier studies highlight the descriptive capabilities of FA and CA as exploratory tools in hydrochemical investigations.