REGIONAL FLOOD FREQUENCY ANALYSIS FOR BLUE NILE RIVER BASIN, ETHIOPIA

Abstract

Regionalization is the most viable way of improving flood quantile estimation in index flood method of frequency analysis. Systematic ways of estimating flood quantiles for the design of hydraulic structures for the cases of both gauged and ungauged catchments are presented. These have been carried out on Blue Nile River Basin, which is one of the biggest basins in Ethiopia with ample of water resource potential. Blue Nile River Basin has been regionalized into similar flood producing characteristics based on established flood frequency procedures. The basin was delineated in to five homogeneous regions. Accordingly, region one comprisrs the widest portion of the basin, covering 37.6 % of the area. It includes the upper Guder catchment, middle and lower Didesa, lower part of the main Abbay basin and Dinder catchment. Region two covers Beles, Gilgel I Abbay, fin cha and lower Guder sub catchment. Region three includes upper . Didesa, Anger, and upper part of main Abbay ( Blue Nile Falls). Region four, which is the smallest portion of the basin area, about 3.3 %, embraces southern part of Gojam high lands, extreme upper course of Gilgel Abbay and Bir­ Temcha catchment. This basin is typically characterized by hilly and mountainous topography. Region five incorporates Gumera, Rib, Beshlo, Jema, Dabus and Muger catchments. Except one station (on Idris river) in region three, all regions have shown satisfactory results for homogeneity tests. · For the identified homogeneous regions four different best fitted distribution types and robust parameter estimation methods were selected to fit the standardized flow data for various return period. Generalized logistic type of distribution is found to be the best fitting distribution for region one and region four. Whereas, log Pearson Ill for region two, log normal for region three, and Vl generalized extreme value for region five are selected for fitting flood quantiles. For all distributions type probability weighted moment parameter estimation method is more efficient except for log Pearson Ill, in which method of ordinary moment is preferred. Based on this, regional flood frequency curves are developed for all regions using standardized flow data for the purpose of estimating flood quantiles in the ungauged catchments of the basin. This study is believed to benefit regional states, NGOs and the country at large in providing new information on flood quantiles for designing of hydraulic structures, culverts, bridges, dams, weirs, etc.