Forecasting by Stochastic Models to Inflow of Karkheh Dam at Iran
Abstract: Forecasting the
inflow of rivers to reservoirs of dams has high importance and complexity.
Design and optimal operation of the dams is essential. Mathematical and
analytical methods use for understanding estimating and prediction of inflow to
reservoirs in the future. Various methods including stochastic models can be
used as a management tool to predict future values of these systems. In this
study stochastic models (ARIMA) are applied to records of mean annual flow
Karkheh river entrance to Karkheh dam in the west of Iran. For this purpose we
collected annual flow during the period from 1958/1959 to 2005/2006 in Jelogir
Majin hydrometric station. The available data consists of 48 years of mean
Annual discharge. Three types of ARIMA (p, d, q) models (0, 1, 1), (1, 1, 1)
and (4, 1, 1) suggested, and the selected model is the one which give minimum
Akaike Information Criterion (AIC). The Maximum Likelihood (ML), Conditional
Least Square (CLS) and Unconditional Least Square (ULS) methods are used to
estimate the model parameters. It is found that the model which corresponds to
the minimum AIC is the (4, 1, 1) model in CLS estimation method. Port Manteau
Lack of fit test and Residual Autocorrelation Function (RACF) test are applied
as diagnostic checking. Forecasting of annual inflow for the period from 2006
to 2015 are compared with observed inflow for the same period and since
agreement is very good adequacy of the selected model is confirmed.
Author: Karim Hamidi
Machekposhti, Hossein Sedghi, Abdolrasoul Telvari, Hossein Babazadeh
Journal Code: jptsipilgg170029
