Flood Frequency Analysis on the basis of extreme values theory (Case Study: Arazkuseh hydrometric station, Golestan)

Flood is an important hydrological event which has considerable outcomes in human’s life. The frequency analysis of prior period is one of the ways for assessing this event. However, the selection method of extreme values can have a significant impact on frequency analysis of such events. Several approaches have been proposed for this purpose, including the annual maximum series (AMS) and the peaks over threshold (POT). In the AMS method, only the greatest event occurred in each year will be selected. But in POT method, regardless of the time of occurrence of extreme events, a threshold is determined and the values over threshold are participated in the frequency analysis. The question is that, how to determine the appropriate threshold. For this aim, some constraints have been made, satisfying them, optimal point can be determined. Golestan province is one of the major poles of agriculture in Iran. Thus, Arazkuseh station in this province, with longtime daily recorded discharge data during 1344-45 to 1388-89, was selected as the study area. Drawing time-series diagrams for discharge data, individual events were selected to be used in the POT method.Comparing the results of the extreme values frequency analysis indicated that the POT method provides higher quantiles than the AMS, in different return periods. Also, by plotting confidence intervals for different return periods, POT method presents less uncertainty. The optimal threshold of 47 cubic meters per second was obtained for flood frequency analysis.