S. Chavoshi, S.s. Eslamian,
Volume 3, Issue 3 (10-1999)
Abstract
Designers of hydraulic structures are often faced with the problem of estimating flood frequencies at stream sites, where little or no flow information is available. A regional regression model is widely used which relates physical and climatological parameters to flow characteristics. In this study, a new method is used which is based on the station-year technique and combined records for several stream-flow gaging stations to make a single composite sample. This method, named ‘hybrid’, was proposed by Hjalmarson and Thomas (1992). It was applied to a group of records from 17 apparently homogeneous stream gaging stations to determine regional flood frequency equations. The study area consists of two adjacent basins, Gavkhoony and North Karoon in the central part of Iran. Using area and mean elevation of the catchments as the most important criteria in relation to peak discharge, the interactive process of the hybrid method was performed, resulting in two-parameter models of regional flood frequency. The performance of the hybrid method was evaluated by comparison with the regional relations determined from a multivariate regression. The comparison revealed that the accuracy of the hybrid method was significantly better than the regression method for low return periods.
S. S. Eslamian, S. Chavoshi Boroojeni,
Volume 7, Issue 1 (4-2003)
Abstract
Numerous methods are used in the investigation of floods in catchments such as regional flood frequency analysis. Regional flood frequency analysis relies on physical, climatic and ecological characteristics of catchments and applies statistical methods to study flow records. Hosking and Wallis developed Probability Weighted Moments and presented L-moments statistics as a new tool for flood frequency analysis. In this paper, the theory of L-moments was used to study the flood frequency of central catchments of Iran. A number of 27 sites each with more than 5 years of observed data were studied. In the first step, the diagram of L-kurtosis versus L-skewness was used and proper distributions for each site were applied. In order to eliminate the heterogeneous sites, homogeneous tests based on D, H1, H2 and H3 criteria were performed indicating that two sites appeared to be heterogenous. Next, using Goodness of Fit Test, the best regional distributions were determined which are GL, GEV, GN, PE3 and GPA, respectively. Finally, quantile estimates for distributions accepted at a 90% level were presented.
