Assessing the Performance of Reaeration Prediction Equations

Abstract

The overall performance of 18 commonly used reaeration rate coefficient prediction equations was evaluated using statistical metrics of prediction accuracy and bias by comparing predicted reaeration coefficients to a database of values measured using gas tracer techniques. Adapting a commonly applied concept, predictive equations were evaluated in discrete regions of the velocity-depth space. Results indicate that rarely is there a single best prediction equation in a given velocity-depth region, rather there usually is a group of statistically indistinguishable top-performing equations. Also, no single reaeration equation performed well over all hydraulic conditions. Prediction equations, which include slope as a variable are more accurate and have lower bias than those that do not. However, even the top-performing equations exhibited large prediction errors of at least 40–50% and exceeded 100% in some regions. This level of error in predicting reaeration rate coefficients will continue to have a major impact on the uncertainty of dissolved oxygen forecasts from receiving water quality models.