Homogeneity testing for peak flow in catchments in the Equatorial Nile basins

Abstract

Regional flood frequency analysis deals with the identification of homogeneous regions of which the distribution of peak flows from sites from such a region are similar. Once a homogenous region is identified, standardized data from different sites within the region can be pooled together and a single frequency curve applicable to the region can be derived. A region can be considered homogeneous for flood frequency analysis if sufficient evidence can be established that data at different sites in the region are drawn from the same distribution (except for the scale parameter). Hosking and Wallis (1993) developed several homogeneity tests for use in regional studies. The aim of these tests is to estimate the degree of heterogeneity in a group of sites and to assess whether they might reasonably be treated as a homogeneous region. A method commonly used with flow data to determine regional homogeneity is the L-moment ratio diagrams. The method is based on the concept that all sites in a homogeneous region have the same population L-moments. Their sample L-moments will, however, be different owing to sampling variability. This has been taken into account by simulation. In addition to the L-moments, also the STU-index has been considered, which is based on the differences between the arithmetic averages of the at site data, including and excluding the largest value. From both the STU-index method and the L-moment diagrams, it is concluded that the Kenyan stations and the Tanzanian stations considered in this study (these are stations in Equatorial Nile basins) each belong to the same region. This homogeneity grouping forms the fundamental base for regional modeling in the study area