U.S. Geological Survey

Nitrate concentrations have increased in many Upper Floridan aquifer springs since the 1950s, exceeding 1 mg/L in recent years at some springs. It is necessary to understand the occurrence and biogeochemical transformation of nitrate in ground water to develop effective measures to reduce ground-water contamination. The objective of this study was to develop a statistical model to predict the presence of nitrate possibly attributable to human influences. A database of nitrate concentrations and related water-quality, land-use, and geologic parameters for 570 wells sampled from 1990-2006 in central and northeast Florida was analyzed. For these purposes, nitrate is considered to be present when its concentration is equal to or greater than 0.2 mg/L and absent (at background concentrations) below this level. Five statistical models were developed consisting of three different types: classification tree, regression, and neural network. The classification tree was chosen as the preferred model because of its good predictive ability and ease of interpretation. Many of the associations identified by the classification tree model are consistent with the nitrification and denitrification processes that describ