Daily Rainfall-Runoff Prediction and Simulation Using ANN, ANFIS and Conceptual Hydrological MIKE11/NAM Models

Rainfall-Runoff modelling is considered as one of the major hydrologic processes with a key role in predicting flood forecasting and water resources. Furthermore, in order to prevent damages caused by the flood and control and inhibit as well as management and flood alarm, rainfall prediction is inevitable. In the present study, for a 10-year period (1999-2009), the rainfall has been predicted using historical rainfall data in Eskandari basin located in Iran using Adaptive Neural Fuzzy Inference System (ANFIS) and Artificial Neural Network (ANN). In addition, the best input combination was identified using Gamma Test (GT) for the rainfall prediction. Then, runoff discharge produced by the predicted rainfall and the observed rainfall were simulated by a conceptual hydrological model called MIKE11/NAM model and the results were compared together. The results indicated that ANFIS model might be better than ANN model for predicting rainfall. In this study, the efficiency coefficient of NAM model in runoff simulations was 0.7 based on observed rainfall and based on predicted rainfall was 0.58 in Eskandari basin. Moreover, the results indicated that NAM model simulated the base flow more accurate than simulating the peak flow in this basin.