Rainfall-runoff modeling using artificial neural networks (ANNs) and multiple linear regression (MLR) techniques

Abstract

Runoff prediction has an important role in hydrology, water management, flood prediction and socio-economical concern. The effective flood management is always of great apprehension in the field of hydrology and water resources engineering. The present study shows the comparison of various training algorithms available for training multi-layer perceptron (MLP) in artificial neural networks (ANNs) and multiple linear regressions (MLR) for modeling the rainfall-runoff process. Gamma test (GT) is one of the non-linear modeling tools whereby an appropriate combination from input parameters can be investigated for modeling the output data as well as establishing a smooth model, to develop and evaluate the applicability of the MLP and MLR models by way of training and testing of developed models during monsoon period (June to September). The ANN models were trained using multi-layer perceptron with various types of algorithm namely Momentum, Quickprop, Delta-Bar-Delta, Conjugate Gradient and Levenberg Marquardt. The performance of the models were evaluated qualitatively by visual observation and quantitatively using different performance indices viz. root mean square error (RMSE), correlation coefficient (r), coefficient of efficiency (CE).