“DEVELOPMENT OF STAGE DISHARGE RELATIONSHIP AT BARMAN STATION ON NARMADA RIVER”

Abstract

NARMADA is the largest west flowing peninsular river, ranks seventh in discharge. A
number of dams have been constructed on the river and its tributaries. As such, a need arise to regulate
the reservoir for releasing the adequate water in the river throughout the year. River discharge is an
important hydrological parameter in any water resource management which requires collection of daily
discharge data. This is mainly in the form of forecasting or estimating the magnitude of hydrological
variable like rainfall or runoff. But, direct measurement of daily discharge in number of points in all the
streams is not only prohibitive in cost, but also very much time consuming, which can be best achieved
by developing stage discharge relationship. Prediction of stage discharge relation is of immense
importance for reliable planning, design and management of most of the water resource project.
The present study was carried out to develop a river stage and discharge modeling using Artificial
Neural Network (ANN) and Linear Regression. From the literature and the developed models, it is
clear that ANN models for river stage discharge are more efficient than the other traditional methods.
Ten ANN models were developed with two hidden layers and 10, 11, 12, 13, 14, 15, 16, 17, 18 and 19
numbers of neurons respectively. The developed models were trained, tested and validated on the data
sets of Barman station on Narmada River in Madhya Pradesh. The value of Coefficient of Correlation
and the value of Coefficient of Determination obtained from ANN were compared with those obtained
by Linear Regression. Comparing observed data and the estimated data through developed ANN
models, it has been proved that the developed ANN models show good results than the traditional
models.