Spatial Characterization of Water Quality in Seven Eastern Kentucky Reservoirs Using Multivariate Analyses

Abstract

We conducted water quality surveys along seven eastern Kentucky reservoirs three times during the 1994 growing season in order to understand the spatial variability within and between impoundments from a common geological setting. Sixteen parameters were measured at sites along the reservoirs representing each zone (riverine, transition, or lacustrine), and eight of these parameters displayed a statistically significant site/zone effect. Most of these were particle-associated parameters (total Fe, NH₄⁺, total suspended solids, and total phosphorus) that decreased significantly down-reservoir. Water transparency increased significantly down-reservoir, in accordance with this pattern. Temperature and dissolved oxygen also showed a significant spatial effect, as temperature was highest towards the riverine end and dissolved oxygen was highest at the lacustrine end. Six parameters showed significant variability among these reservoirs, including ammonium and chlorophyll a. The remaining four (conductivity, sulfate, alkalinity and pH) displayed similar patterns across reservoirs apparently as a result of local geologic attributes. Another goal of this work was to use a multivariate statistical approach to aid in understanding the relationships between water quality parameters in this region and to group zones within a reservoir according to their similarities in water quality. Principal Components Analysis reduced sixteen parameters to five principal components that accounted for 74 percent of the total variability in water quality. Two of these principal components contained all of the particle-associated and local-geologic parameters, and accounted for 42% of the total variability. Cluster analyses showed that the water quality signatures of two reservoirs were quite different from the rest. Interestingly, these represented the most pristine and most disturbed watersheds of all seven systems. Our findings are relevant to the management of these and other similar reservoir systems.