ORIGINAL RESEARCH
Parent Flood Frequency Distribution
of Turkish Rivers
Omer Levend Asikoglu
More details
Hide details
Ege University, Department of Civil Engineering, Izmir, Turkey
Submission date: 2017-04-04
Final revision date: 2017-07-12
Acceptance date: 2017-07-18
Online publication date: 2018-01-08
Publication date: 2018-01-26
Pol. J. Environ. Stud. 2018;27(2):529-539
KEYWORDS
TOPICS
ABSTRACT
This 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.