Transverse mixing is a complex process and important in understanding the transport of pollution in rivers. This study presents a genetic programming (GP)-based model for estimating the transverse mixing coefficient (TMC) in flumes and rivers. More than a hundred of data points from previous studies, including datasets on laboratory straight rectangular flumes and field measurements in natural rivers, are collected and used to develop the final formulae for estimating TMC. During the analysis, TMC is separated into the transverse turbulent diffusion coefficient and the transverse dispersion coefficient given that they represent two different processes. Before formula optimization and search are performed using GP software, the target formulae are semi-defined to reduce search time and ensure the physical basement of the final formulae. The model presented in this study exhibits good improvement in terms...
