An Evaluation of Dispersion Coefficient Models for Rivers

Abstract

This article intends to evaluate a few mathematical and empirical models of river dispersion coefficients from previous studies. Two problems were cited as the causes of their shortcomings: the significant discrepancy between measured and predicted values of the phenomenon. The models based on previous research fail to take into account some of the geometric and hydraulic facts of dispersive flows, such as dead zones and bend effects, because they were made under assumptions that are false in real rivers. The empirical models omit some of the most significant parameters known to affect dispersion, whereas the mathematical models demand cumbersome, time-consuming, and labour-intensive tracer experiments. Although the accuracy of more recent machine learning techniques has increased, they are still very expensive, prone to error, and require a high level of expertise. All the equations fall short of the two crucial criteria for scientific acceptance: reproducibility and strong predictive power. A form for a new equation is proposed that will take into account many of the omitted parameters and, as a result, improve accuracy. Poor prediction accuracy should be addressed by the new equation. It is possible to derive the equation using dimensional analysis.