COMPARATIVE STUDY OF ANN AND ANFIS MODELS FOR THE RAINFALL RUNOFF MODELLING OF PERIYAR RIVER BASIN

Abstract

Rainfall-runoff relationships are the most complex hydrologic phenomenon. In order to solve hydrology and water resource challenges, Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) are increasingly being used. In this study, ANN and ANFIS are utilized to develop rainfall-runoff models for the Periyar River Basin (PRB), and performance evaluation indicators like Root Mean Square Error (RMSE), Coefficient of Determination (R2), Nash-Sutcliffe Efficiency (NSE), and Percentage Deviation from Peak are used to compare the relative performance of these models. Without being in touch with the physical properties of the process, ANN, which resembles the human brain, can extract the nonlinear relationship between inputs and outputs of a complex process. ANFIS is the combination of the concept of neural networks and fuzzy logic. ANN model satisfactorily predicted the runoff values. ANFIS model outperformed ANN model for the same data set. ANFIS model predicted the peak runoff accurately than ANN. This study provides evidence that both ANN and ANFIS are useful tools for rainfall-runoff modelling. ANSIS performs better than ANN.