Modeling flood routing with genetic expression programming

Floods are natural disasters that cause huge losses and economic damages in many parts of the world. Flooding peak values are required in the design of bridges, canals, spillways for dams. The flood wave entering a particular section of the river or the reservoir causes the water level to rise and this change is investigated by flood routing method. Flood routing is a method used to determine flood replacement by using flood data in a river section. In this study, the Sütçüler flood, which caused the loss of life and property, which took place in Sütçüler Değirmendere, a line of the Aksu river on November 4, 1995, was modeled using flood rouring method. Progression in Artificial Intelligence (AI) offers opportunities in engineering studies (water resources, hydraulics and hydrological events) using new algorithms and models. This work presents the Genetic Expression Programming (GEP) method, an extension of genetic programming as an alternative to modeling flood routing. Thus, new models have been developed using the GEP method in flood routing prediction. In the GEP method, the inflow (I), outflow (Q) and time (T) parameters are used. The performance of the model was evaluated with two fitness measures, detection coefficient (R2) and root-mean-squared error (RMSE). It is included that the model based GEP has more performance in flood routing prediction than the other solution techniques based on the Muskingum model and finite difference methot.