Flow discharge modeling in flooded rivers by using genetic programming

Flow discharge determination in rivers is one the key elements in design of river engineering projects. By flow entering into the floodplains and sudden changes in river geometry, the common resistance equations are not applicable. Therefore, many approaches have been developed for modification of flow discharge computations. Most of these methods have satisfactory results in laboratory flumes but extending those to all hydraulic and geometric conditions such as natural rivers are very limited. In recent decades, due to ability of complex phenamena modeling, the artificial intelligent methods have very wide applications in many fields of water engineering. Genetic Programming, a branch of evolotionary algorithms, is able to optimize the model structure and its components and to derive an explicit equation based on the variables of the phenamena. In this study, using 394 laboratory and field stage-discharge data sets of 30 compound channels, a new equation has been derived for calculation of flood discharge, based on geometric and hydraulics parameters, using Genetic Programming. The modeling is conducted in training and testing phases that at these two phases, the agreement between computed and measured flow discharge was very good. The determination coefficient for the training and testing phases was 0.99 and 0.96, respectively.