Flood frequency analysis and modeling with statistical methods

Floods occur as a result of the flow rate in rivers increasing and reaching the peak value. Floods are natural disasters that adversely affect economic and social life in the regions where they occur and cause deaths. In order to protect from floods and minimize the damages caused by floods, it is very important to create flood maps and to predict floods by estimating possible flood flows in advance. Flood flows are estimated by statistical methods, synthetic methods and deterministic methods. In this study, the flood flows for the rivers in the region where the Ağva neighborhood of Istanbul's Şile district is located were estimated using statistical methods for return periods of 50, 100, 200, and 500 years. These estimates were made using Normal, Log-normal, three-parameter Log-normal, Gumbel, GEV, Pearson Type III, and Log-Pearson Type III distributions, and the most suitable one was selected using the Kolmogorov-Smirnov (K-S) and Probability Plot Correlation Coefficient (PPCC) tests. Flood inundation maps were then developed for the flood discharges of different return periods by modeling with the MIKE 21 hydraulic modeling program developed by the Danish Hydraulic Institute (DHI). For the study, 43 years of flow data from Göksu observation station named D02A004 and 16 years of flow data from Sungurlu observation station named D02A149 located on the rivers in the region were used. As a result of the suitability tests, it was determined that the most appropriate distribution was Log-Pearson Type III. The flowrate using Log-Pearson Type III distribution was calculated as 589.00 m3/s, 708.00 m3/s, 837.00 m3/s and 1022.00 m3/s for 50, 100, 200 and 500 year return intervals, respectively. Then, using these flood flows, hydraulic modelling was performed with the MIKE 21 program and flood depth and velocity values for 50, 100, 200 and 500 years were determined. Maximum flood depth and maximum velocities at return intervals were compared by randomly selecting 10 points in the region.