STATISTICAL ANALYSIS OF GEOCHEMICAL MARKERS IN KARST GROUNDWATER OF CENTRAL KENTUCKY

The Inner Bluegrass region of Kentucky covers roughly 5000 square kilometers of central Kentucky with the city of Lexington near the central part of the region. Most of the region is underlain by flat-lying Ordovician aged carbonates, and therefore contains an abundance of typical karst features such as sinkholes, caves, and springs. Studies of groundwater flow in the region began as early as the 1980’s (Thrailkill et al, 1982). Initial results suggested that individual ‘groundwater basins’ may exist, where groundwater flow tends to occur mostly within one sink –cave (conduit)-spring system with little to no flow mixing or crossing into other ‘groundwater basins.’ Gouzie (1986) proposed that the karst groundwaters might develop unique geochemical ‘signatures’ related to rock units and surface land uses within each ‘basin’ if there is no mixing or crossover between basins. In unpublished results, Gouzie (1986) found that a Discriminant Analysis function statistical method from the SAS (Statistical Analysis Software) program could yield roughly 90% accuracy in placing groundwater samples into the correct ‘basin’ based on basic geochemistry. This study revisits the earlier work by comparing a 2006 dataset collected as a ‘blind sample set’ to the samples reported in Gouzie (1986). The results from the new comparison are presented and discussed, with a focus on whether changes in land use over 20 years of urban growth around Lexington may mask accurate geochemical signatures or if the signatures remain useful even under conditions of urbanization. The geochemical signature method could become a valuable tool in other karst areas for environmental managers and remediation planning in contaminated areas, as such a tool may allow managers to quickly assess the areal extent of a drainage basin from simple groundwater analyses rather than more time intensive dye traces.