*Research Scholar, Alternate Hydro Energy Centre, Indian Institute of Technology, Roorkee
**Principle Scientific Officer, Alternate Hydro Energy Centre, Indian Institute of Technology, Roorkee
***Former Professor, Department of Hydrology, Indian Institute of Technology, Roorkee
Online published on 2 April, 2015.
Stream flow prediction is required for the planning and operation of water resources projects. The data generation phenomenon is strongly recommended when the stream flow data is either inadequate or unavailable. The long term data generation facilitates flood mitigation and water resource project planning. Stream flow time series is a non-linear phenomenon, so stochastic models; Seasonal-ARIMA and X-12-ARIMA are mostly used prediction approach nowadays. The present study is focused on the monthly streamflow prediction for Manot Site in Upper Narmada River basin. Best Seasonal-ARIMA and X-12-ARIMA models are selected on the basis of AIC (Akaike's information Criteria), BIC (Bayesian information Criteria) and residual correlogram (ACF and PACF) analysis. Performance of different parameters for Seasonal-ARIMA and X-12-ARIMA are compared by statistical performance criteria. The study reveals that X-12-ARMA(1,0,2)(0,1,1) has better prediction accuracy as compared to seasonal ARIMA (0,0,4)(0,1,0).
Seasonal-ARIMA, X-12-ARIMA, Streams flow Prediction, Time series analysis, Hydropower
