Possibilities of Using Regional Index-Flood Method with Annual Maximum and Partial Duration Series: A Case Study of Susurluk River Basin, Turkey

Among the natural disasters experienced in Turkey, floods, which cause the most loss of life and property after the earthquake, have increased their impact and frequency of occurrence over time, as well as unplanned urbanization caused by the increasing population, uncontrolled construction in stream beds, and changing climate. Therefore, it is important to accurately predict the magnitude and frequency of floods. This study investigated the possibilities of using the regional index-flood method and annual maximum series (AMS) and partial duration series (PDS) in the Susurluk River basin. Annual maximum flood series provided homogeneity in the Susurluk basin as a single region, and the Generalized Logistic (GLO) distribution fits the AMS. PDS was extracted according to the threshold levels determined using the variance-mean ratio and frequency factors. The PDS's most appropriate frequency factors (k) were determined according to the Poisson distribution, which makes the variance-mean ratio equal. k =3.5 was determined for only two stations. k =4 was suitable for seven stations, and k =5 was suitable for thirteen stations. The average number of peaks over the threshold level (lambda`) varies between 1.26 and 5.31. Since PDS is not homogeneous in a single region, cluster analysis divided the basin into three regions. After homogeneity was achieved, Pearson Type 3 (PE3) and Generalized Pareto (GPA) distributions were suitable with the PDS. The study concluded that instead of annual maximum flood series, partial duration flood series can be used in many stations in the short and medium term but can be used in fewer stations in the long-term estimations. Since Regions I and II are relatively lower and flatter areas than Region III, it was observed that the flows started to accumulate at the stations in these regions, and larger floods were predicted. Region III is close to the basin upstream, and smaller floods were predicted at the stations in this region. Since partial duration flood estimations are lower than annual maximum flood ones, they can provide advantages to engineering projects with lower costs. In addition, PDS can be useful in regionalizing floods, which are very common due to the data extraction process.