-

The released energies from the earthquakes occured during April 1967-February 1969 around
Sefid-Rud dam with 4Okm radius were analysed as a time series for modeling. The magnitudes of
the earthquakes are less than 4 in Richter scale, so thc time series is sufficient for the return
period. The monthly release of energy was obtained from the average of total energy release per
month and the model of Auto Regressive Moving Average (ARMA) is fitted to it. The model
parameters are determined from mathematical relations and based on the correlation of the
residuals, Port Monto and Akaike tests, the best model is recognized. This model is found to be
the first order, which applies less parameters, simple and is showing seismicity of the region.
Based on this model for the next 23 months the energy release and magnitude of the earthquakes
are simulated.
Since the biggest earthquake occured in this period has a magnitude 4, the simulation is
restricted to a maximum magnitude 4. The simulation of the earthquakes shows that in August
and October 1969 and in January 1971, an earthquake with magnitude of 3.5 to 4 would occur. It
is obvious, if we have longer series and bigger earthquakes, we would be able to simulate bigger
earthquakes. This simulation can compensate the lack of the earthquake records, due to
equipment defect.
For future seismicity of the region, the model is able to predict the seismicity of a month
having the related data for the previous 21 months. By applying the data of 22 months we can
predict the seismicity of the 23rd month which can be compared with the original data of 22nd and 23rd months. Finally by using 23 months original data, the prediction for the 24th month is
applicable. So by these three month prediction in this region the trend of seismicity in the near
future can be evaluated. The trend in this case is constant.