QUANTIFYING UNCERTAINTY IN NITRATE POLLUTION FROM LAND APPLICATION OF SEWAGE SLUDGE

Uncertainty associated with nitrate-nitrogen pollution of groundwater from sludge applications was investigated.; Three alternative planning models were proposed. The models estimate annual loads and concentrations of nitrate-nitrogen in percolation. Published data from four field studies which included aerobic and anaerobic sludge and surface and soil incorporated applications were used for model testing. Point and stochastic validation procedures were used. The models were generally conservative. For three data sets average relative errors were approximately {dollar}pm{dollar}25% for all three models. The fourth data set had average relative errors of 100-500% Wilcoxon Signed Rank test results (at {dollar}alpha{dollar} = 5%) suggested point model predictions and observations were drawn from similar distributions. Stochastic validation tests also suggested that observations were samples from distributions of predictions. The models were tested as uncalibrated approximations of nitrate percolation. All three appear sufficiently accurate for planning application systems.; Two approaches to estimate relative sources of uncertainty in nitrate loads were developed. One involved analytical approximations of variances in loads. The other was based on Monte Carlo simulation. The analytical procedure was accurate for systems with mean annual loads {dollar}>{dollar}20 kg/ha. Both approaches were applied to example systems in the Eastern United States. Mineralization of soil organic nitrogen often was the major source of uncertainty in nitrate loads. Ammonia volatilization was the smallest source. Mineralization of sludge organic nitrogen was the major source for less stabilized sludge. Identification of major sources represents a first step in managing uncertainity. Site-specific measurements of mineralization rates could efficiently decrease uncertainty in predicted loads.; A planning procedure was developed which utilizes analytically approximated probability distributions of nitrate loads and concentrations. Distributions are specified with analytical moment expressions and a normal distribution function limited to be non-negative. Analytical distribution accuracy was illustrated with comparisons to simulation results. The planning procedure involves selection of application rates based upon acceptable frequencies of excessive nitrate concentrations. Severity of contamination is assessed with probability distributions of annual loads. This procedure was illustrated for representative systems.