Parametric modeling of water quality and sampling strategy in a high-altitude appalachian stream

Multiple linear regression models were developed using log-transformed water quality data from a high-altitude stream in the Great Smoky Mountains National Park. The independent variables of cumulative Julian days, seasonality terms, stream pH, conductivity, and flow rate were used in the regression models to predict concentrations and loads for acid neutralizing capacity (ANC), chloride, hydrogen ion, nitrate, potassium, sodium, and sulfate. The regression analyses showed statistically significant declines in nitrate and ANC loads and concentrations with time, but did not show declines in hydrogen ion or sulfate concentrations. The water quality database and regression models were used to test weekly, bi-weekly, tri-weekly, monthly, bi-monthly, and quarterly sampling frequencies. The results showed overall that the weekly, bi-weekly, tri-weekly, and monthly sampling strategies should produce distributions that are statistically similar in mean and variance for stream water quality and loads.