Using Artificial Neural Networks (ANN) for Real Time Flood Forecasting, the Omo River Case in Southern Ethiopia
Lulseged Ayalew, Dietmar P.F. Möller and Gerhard Reik2
Summer Computer Simulation Conference 2007 (SCSC 2007)
San Diego, California (USA), July 15-18, 2007
Abstract
Abstract This study presents the application of artificial neural network (ANN) methodology for real time flood forecasting in Omo River, southern Ethiopia. Back propagation algorithms have been used for 1 to 6 hour runoff predictions with various combinations of flood events for training the ANN models. The performance of each model structure has been evaluated using common performance criteria, i.e., root mean square error (RMSE), coefficient of correlation (r), and coefficient of determination (R2). The criteria selected to avoid over training was the generalization of ANN through cross validation. Fairly accurate hourly runoff predictions have been obtained using the data of six flood events suggesting that the ANN models are particularly suited for flood forecasting purposes. The two important parameters, when predicting a flood hydrograph, are the magnitude and the time to peak discharge. It has been found that the ANN flood forecasting models have been able to predict this information with great accuracy. However, the forecasting efficiency decreases with increasing time.