Abstract:
For development of our agriculture based economy and to meet the demands of the growing population; in terms of irrigation, drinking water, hydropower generation and indu-3trial use etc. modelling and forecasting of river flows during non-monsoon months is essential. Efficient management of existing water resources projects and optimal planning of the future projects also attach great importance to low flow modelling and forecasting.
Ideally, a distributed model based on the principle of physical laws representing the movement of water through its different phases should be developed for any modelling exercise for low flow forecast. However, it remains a fact that the physical laws for representing the formation and propagation of runoff through its various processes have not yet been perfected. No doubt, for the more complicated problems the use of physically based models acquires a great importance. But the physically based models require extensive data input and enormous computational facilities. On the other hand, there are many problems for which necessary solutions can be obtained with desired degree of accuracy required for the purpose, with relatively less sophisticated, lumped, conceptual or statistical models, which require very limited data generally available in the field.
In this study, a statistical model has been used to forecast the low flows (November to May) for the river Narmada at Mortakka, using the monthly data of previous month. To begin with, the forecast of flow from the month of November to May is formulated on the basis of observed data of October for a specific year. The forecast is updated after each month when additional data become available. The observed discharge and forecasts have been compared for the four test years and percentage errors between them have been computed.
