Abstract
Data of monthly volumes of discharges reaching Mosul Dam reservoir for a
period of (15) years, multiplied by (12) month and thus get a time series consisting of
(180) value for monthly volumes of discharges are organized and collected for years
( 1985 – 1999 ). A synthetic time series were obtained by applying multiplicative
decomposition of a time series method. Where normal method (which depends on the
values of the two components (trend and seasonal) only) was applied on the historical
data after conversion to the normal distribution , and the standard error (MAPE) of this
method was (39%). For the purpose of improving the results were an amendment in this
research on this method by using the modified method in this research which adopting
the values of components other than the series (cycles and random) and enter them in
the calculation the values of synthetic time series by multiplying their approximate
values in the results of the values of synthetic time series extracted by the normal
method above , in terms of improved the error measure value (MAPE) of the modified
method to (32%).
The results showed the effectiveness of the modified method used in this
research in the prediction of incoming water volumes for years ( 2000 – 2004 ), where
tests showed conformity (average, standard deviation, autocorrelation coefficient,
density analysis) and there is a match between the predicted values with historical
values.