Modelling flood prediction using Radial Basis Function Neural Network (RBFNN) and inverse model: A comparative study

Flooding has become prominent focus of hydrological studies because of increasing public awareness on this problem. The complex nature of floods and its responses make it the most challenging and important task of the researcher. Conventional methods for establishing the relationships between input and output data need to understand the behavior of the system, however the relationship is complex and highly nonlinear. To provide an alternative approach for accurate flood prediction, an artificial neural network which is capable of modeling nonlinear and complex systems is presented in this paper. A Radial Basis Function Neural Network (RBFNN) was developed for flood water level prediction at Kelang river located at Petaling Bridge. The peak water levels during flood events were used to train, test and validate the network. The result shows that the RBFNN model can be considered as a suitable technique for predicting flood water level. Nevertheless, with the implementation of Inverse Model cascaded with the RBFNN model the result show significant improvement.