Parent Flood Frequency Distribution of Turkish Rivers

Abstract

Determining the probability distribution (PD) of such random events as floods by estimating their behavior has always been a significant problem. Identifying the appropriate PD of floods is probably the most important step in flood frequency analysis. Accordingly, many studies all over the world have investigated how to determine an appropriate PD of floods [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]. The selection of techniques for assessing flood frequency in specific conditions depends on several dynamics, including national tradition, knowledge of the modeler, the aim of a study, governmental necessities, and data accessibility [17].Cunnane [18] investigated the flood frequency determination techniques adopted by 55 organizations from 28 countries. Gumbel, two-parameter lognormal (LN2), P3, and log-Pearson type 3 (LP3) distributions appear to be the most common of the six extant distributions, with the other two being extreme value type 2 (EV2) and GEV distributions.Pol. J. Environ. Stud. Vol. 27, No. 2 (2018) AbstractThis study investigates the existence of a single parent flood frequency distribution on a Turkish scale. In the design of hydraulic structures, estimating the project flood of a given period or probability is usually the first step. Determining an acceptable design criterion depends substantially on the probability distribution of floods. In this study, annual peak series from 268 Turkish rivers was collected and a database of L-Moment ratios was constructed. The best-fit probability distribution (PD) model was investigated among seven distribution models (generalized logistic, GLO, generalized extreme value, GEV, generalized Pareto, GPA, three-parameter lognormal, LN3, Pearson type 3, P3, Gumbel, GUM, and normal distribution, N) by using a framework of L-moment ratio diagrams. The L-Moment ratio framework used in this study consisted of two sequential procedures. The first graphical procedure evaluated the L-Moment diagrams visually while the second procedure is based on the hypothetical testing procedure of L-Moment diagrams. The results of a graphical inspection of the dataset show GEV distribution as a potential parent PD of the floods in Turkey. A more detailed hypothetical testing procedure comprises Monte Carlo simulations produced from a GEV model. In a hypothetical testing procedure the variability of L-skewness and L-kurtosis values of sample data are situated within the theoretical limits of GEV distribution. Consequently, the GEV distribution is accepted as a single parent PD for annual maximum flow series of Turkey.

Accounting for hydroclimatic properties in flood frequency analysis procedures

“…Prior work demonstrated that selecting one distribution that provides the best fit to annual maxima is difficult over a large hydrologically heterogeneous region due to the high sample variance of the L-moments (Asikoglu, 2018;Salinas et al, 2014a). To reduce the noise and guide model selection, we compute a weighted moving average (WMA) of neighboring L-skew and their corresponding L-kurtosis proportional to record length.…”

Regional Flood Frequency Analysis Using Linear Moments and Partial Linear Moments: A Case Study

“…First one refers to at-site flood frequency analysis while the second one as regional flood frequency analysis. In at-site frequency analysis, we analyze the results of each site or station individually instead of grouping the information from all sites (Zin et al, 2009;Seckin et al, 2011;Ahmad et al, 2015Ahmad et al, , 2016a; Afreen and Muhammad, 2012;Rahman et al, 2013;Asikoglu, 2018). At-site frequency analysis is not a promising technique in the presence of small data and non-availability of data at ungauged stations.…”