عنوان مقاله [English]
Real time forecasting of river flow is an essential tool in optimum operation of water recourses systems espacially storage dams. Because of high complexity of hydro climatology events, real time inflow forecasting is a difficult task. Many researches have accordingly been carried out and various models have been developed for inflow real time forecasting.
Many storage dams, mainly hydropower dams, were constructed or are under construction in the Great Karun basin in south western Iran. The optimization of hydropower generation in this system is therefore of great importance. In this regard the inflow real time forecasting would also be an important item. In this study the Karun-III reservoir inflow real time forecasting was conducted using combination of autoregressive technique and the precipitation forecasts information available for a period of 22, November, 2004 to 20, March, 2005. In this research, various stochastic series data are generated based on the 5-day observed discharges. A series with the most agreement with the daily precipitation forecasts in the following days, is then selected to forecast the daily inflow in the next 5 days. Results showed that relative errors of the proposed method for rainy and non-rainy days was about %31.8 and %12.7, respectively. The average error for all the 109 days forecasted data was also reported as %21 which is quite below the average relative errors obtained in simple autoregressive technique which was about %30.