Author: Meral BüYüKYILDIZ
Publishing Date: 2004
Volume 19 Issue 1
In this study, stochastic models were determined for annual mean streamflows of gauging stations operated by EIE and numbered as 1611, 1612 and 1622 in Central Anatolia Closed Basin.For these stations, mathematical expressions were obtained by using the methods of analyses of autoregressive (AR) models and autoregressive moving average (ARMA) models for annual streamflow data. A preliminary study about possible AR and ARMA model types was made after examining the correlograms and partial correlograms. In conclusion, the AR(1) model was found to be suitable for annual mean streamflow series of the selected gauging stations. The best ARMA(p,q) model was also found as ARMA(1,1) model for stations 1611, 1612 and ARMA(2,1) model for station 1622. Then 50 synthetic series having the same N length period were generated for each gauging station by using the developed models. Statistical characteristics (mean, standard deviation, skewness coefficient, correlogram) of these synthetic series were calculated and compared with the statistical charasteristics of the historical (original) series. Consequently, it was observed that stochastic models established for the gauging stations of 1611, 1612 and 1622, preserved the statistical characteristics of the historical series.
Key Words: Streamflow; autoregressive model; correlogram; partial correlogram; stochastic