Logistic model of nitrate in streams of the upper-midwestern United States

Nitrate in surface water can have adverse effects on aquatic life and, in drinking-water supplies, can be a risk to human health. As part of a regional study, nitrates as N (NO3-N) was analyzed in water samples collected from streams throughout 10 Midwestern states during synoptic surveys in 1989, 1990, and 1994. Data from the period immediately following crop planting at 124 sites were analyzed during logistic regression to relate discrete categories of NO3-N concentrations to characteristics of the basins upstream from the sites. The NO3-N data were divided into three categories representing probable background concentrations (<3 mg L-1), elevated concentrations (3-10 mg L-1), and concentrations that exceeded the U.S. Environmental Protection Agency maximum contaminant level (MCL) for drinking water (>10 mg L-1). Nitrate-N concentrations were positively correlated to streamflow, upstream area planted in corn (Zea mays L.), and upstream N- fertilizers application rates. Elevated NO3-N concentrations were associated with poorly drained soils and were weakly correlated with population density. Nitrate-N and streamflow data collected during 1989 and 1990 were used to calibrate the model, and data collected during 1994 were used for verification. The model correctly estimated NO3-N concentration categories for 79% of the samples in the calibration data set and 60% of the samples in the verification data set. The model was used to indicate where NO3-N concentrations might be elevated or exceed the NO3-N MCL in streams throughout the study area. The potential for elevated NO3-N concentrations was predicted to be greatest for streams in Illinois, Indiana, Iowa, and western Ohio.