Development of ANN Model for Simulation of the Runoff as Affected by Climatic Factors on the Jamuna River, Assam, India

Quantifying rainfall-runoff relationships in an orographic area is challenging. Meanwhile, estimation of runoff in a river is urgent as the river supports a human’s livelihood through various ways, including agriculture. Complex physiographic nature of watersheds limits the development as well use of physical based models to predict runoff in a river. Jamuna river is a sub-tributary of Brahmaputra river fed through snow and glacier melt. A barrage is constructed in the river for providing irrigation water to agricultural land. Artificial Nueral Network (ANN) model network has been developed for identification of the impacts of climate change in the runoff process of the Jamuna river using six predictors (surface upward latent heat flux, specific humidity, precipitation, wind speed, maximum air temperature and minimum air temperature) as input variables in the models. The results showed that the network was optimized with a network structure of 5–1–1. The RMSE and R2 value was found to be 0.352 and 0.88 respectively. Further, the results suggested that the study provides practical significance to the water resources managers under changing climate and long-term prediction of hydrological processes.