RUNOFF MODELLING THROUGH BACK PROPAGATION ARTIFICIAL NEURAL NETWORKWITH VARIABLE RAINFALL-RUNOFF DATA

In present study generalised multi layer Back Propagation Artificial Neural Network (BPANN) models have been developed for two sub-basins of Narmada (India) with three time scales (weekly, ten-daily and monthly) and the performance is compared with Linear Transfer Function (LTF) model. It is observed that the number of iterations required for development of BPANN model reduces with increase in variability of data set used for development of this model which was found superior.