Design of a semi dynamical real-time flood forecasting model


Institute of Geophysics, University of Tehran, P.O. Box 14155-6466, Tehran, Iran


Based on the transfer function characteristics a semi daynamical real time flood-forecasting model has been derived which is mathematically stable. Furthermore an user-friendly software has been developed. Due to the incapability of models to perfectly portray complex natural systems, and due to faulty model input data, every forecast is subject to an error. Consequently, it is necessary to correct (update) the forecast in the light of recent model performance to minimize forecast error.
Generally speaking three simulated volume (µ), shape (e) and time (g) factors influence simulated runoff. None of them could cause the model to become unstable or fluctuating. By adjusting these three factors it is possible to make the updated model more accurate than the static one.
The model is made in the form of a software package. This program needs the input files including the rainfall and runoff data. In this investigation, eight events at Behbahan station in the Maroon basin have been used in calibration phase. The order and parameters of the model should be determined so that the realistic impulse response can be obtained. The volume, shape and time parameters can be updated automatically using recursive least squares method. If these parameters are not suitable, they can be chosen manually to achieve better results.
The last part of program allocates the flood forecasting. Again, four separate events at the Behbahan station have been chosen to investigate the capability of the model at the forecasting phase.