• INFLOW FORECASTING FOR KAINJI DAM USING TIME SERIES MODEL

Abstract

A time series model of the inflow for a major river is formulated and an exponential smoothening Technique (EST) is developed in this work. The EST is a physically based method which can be devised in such a manner as to emphasize important and dominant features of a process and used to produce a model suitable for specific needs. Forecasts based on this technique can provide very useful results for the short-term and out perform more complex Box and Jenkins models with the added advantage of simplicity and model parsimony. The technique is applied in this paper to develop a non-causal representation of the inflows into the Kainji dam located in the west of central Nigeria on the river Niger. Inflow data from 1992 through 2008 were used to validate the models and provide short term forecasts. The results exhibit good prediction for periods of 6 months, which can be used to refine reservoir management. Longer term models require the ARIMA format in order to fully reflect seasonal and stochastic components.