Calibration Performance of a Two-Dimensional, Laterally-Averaged Eutrophication Model of a Partially Mixed Estuary

Eutrophication modeling of the estuarine portion of the Neuse River, North Carolina was conducted to predict the water quality improvement associated with a 30 percent nutrient loading reduction to the estuary. An existing two-dimensional, laterallyaveraged model (CE-QUAL-W2) was applied to predict water quality conditions in the lower 80-km of the estuary. During model calibration it was found that the extent to which the model explained observed variability in water quality parameters varied widely. Correlation coefficients between model predictions and observations were 0.93 for salinity, 0.77 for nitrite+nitrate, and 0.69 for dissolved oxygen, but only 0.25 for chlorophyll-a. The relatively poor chlorophyll-a calibration performance was ascribed to the model’s inaccuracy in predicting the timing and location of algal blooms. Good agreement was observed, however, between cumulative frequency distributions of chlorophyll-a observations and predictions. Predicted cumulative frequency distributions of chlorophyll-a concentrations matched observations to within a factor of two for frequencies between 0.01 and 0.99, although the maximum observed values were much higher than model predicted maximums. These peak chlorophyll values are of particular concern to the regulatory community as they result in water quality standard violations. A sensitivity analysis, performed to explore the model’s capability to predict these highest concentrations, indicated that changes in algal growth parameters alone would not increase predicted maximum concentrations to observed values. It appears from model results that residence time is an important factor in limiting the maximum predicted chlorophyll-a concentrations.