REAL-TIME FLOOD FORECASTING USING A HYDROLOGICAL GREY MODEL IN CONJUNCTION WITH A GLOBAL OPTIMIZATION METHOD

Real-time flood forecasting is the major element of a flood forecasting and control system that is a nonstructural method to reduce flood damage in flood prone areas. In this study, a hydrological grey model is developed to forecast runoff in real time, and the model’s applicability is evaluated by comparison with the observed and forecasted runoff. The model parameters are estimated with a global search method, the annealing-simplex method in conjunction with an objective function, HMLE. To forecast accurately runoff, the fifth order differential equation is adopted as the governing equation of the model. The statistic values between the observed and forecasted runoff in calibration and validation indicate that the simulated results are in good agreement with the observed. To evaluate the efficiency of the grey model, the results of the model are compared to these of an artificial neural networks (ANN) model. Comparing RMSE, R2, and REPF (Relative Error between the observed and forecasted Peak Flow) values of the ANN model and grey model results reveals that the grey model is a little superior to the ANN model.