Abstract

Our conclusive purpose of this research is to offer a procedure to study the sensitivity of flood routing models to the input discharge statistics of different hourly time steps. In some case studies, it could be obvious that after calibration of river flood routing modeling; some of the first output statistics in ascending branch of output hydrograph are not acceptable. This error occurs because of the influence of flood volume which has been passed the upstream boundary cross section before the chosen initial time step for modeling; consequently, it could not been utilized as the boundary conditions definition for flood routing modeling software. Other words, it could be noticed that time range of initial and boundary conditions is larger than modeling one. During an innovative procedure; we used MIKE11 software as an acceptable river flood routing model; and also, different structures of artificial neural networks (ANNs) to compute the sensitivity of river flood routing to the number of previous time steps discharge which are effective in modeling accuracy. In this way, we studied recorded flood discharge statistics during 30 years up to now in two hydrometric stations which are located in upstream and downstream of ZOSHK RIVER as our case study. Terminally, we recommended the best range of discharge time steps to use as the initial and boundary conditions of river flood routing modeling.