Repository logo
 

Flood estimation in developing countries with case studies in Ethiopia.

Thumbnail Image

Date

2017

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Extreme flood events have become more destructive in some parts of Ethiopia. Thus, accurate estimates of flood frequencies are vital for effective flood risk management. Yet, estimation of the peak flood is exceptionally complex requiring a wide range of methodologies. One of the approaches is the statistical (traditional) method, which determines the frequency of a flood value from the annual maximum discharge data. However, when such records are too short for flood frequency analysis, empirical formulae can be the option for peak flood estimation. But, most of these formulae are regional formulae based upon the statistical correlation of the recorded peak flood and one or two physical catchment characteristics, and they are unlikely to give reliable results of peak flood for other regions than those for which they were developed. On the other hand, when there are no streamflow observations at the site of interest, hydrological models such as Py- TOPKAPI are another option for modelling stream flows for flood frequency analysis. Thus, the main component of this study involves statistical data analysis and hydrological modeling aimed at finding out an appropriate method of flood frequency analysis for Ethiopian rivers. In this study, a broad overview of practical design-flood-estimation methods in Ethiopia along with international practices was carried out. The results revealed very large gaps in knowledge and in current design flood practices. The application of the PyTOPKAPI model in numerous catchments of the world was likewise reviewed including how the model has been used for flood prediction, forecasting of hydrological responses, etc. In this study, it was implemented in Ethiopia on Gilgel Ghibe and Mojo catchments, and promising results were obtained. This model was also combined with remotely sensed precipitation products for simulating stream flows which showed that the general streamflow patterns were well reproduced. Most importantly, the PyTOPKAPI model was applied in ungauged Ethiopian catchments using the Schreiber runoff ratio in an alternative model calibration approach. This shows how the PyTOPKAPI model can be used to predict runoff responses in ungauged catchments for water resources applications and flood predictions in developing countries. In addition, various flood frequency methodologies were evaluated on two Ethiopian rivers (Awash and Gilgel Ghibe). The aim was to find the most approprite method that best represents the statistical characteristics of the streamflow observations. In this case, the annual maximum discharge data from 14 stations of the two rivers (6 in Gilgel Ghibe and 8 in Awash) with 23 to 54 years of records were used. Seven flood frequency methodologies (TSPT, LN, LPIII, EVI, Chow’s, Stochastic and Weibull’s plotting position formula) were fitted to those data. Comparison of the results were made based upon probability plot correlation coefficient, normalized root mean square deviation and Nash-Sutcliffe fitting coefficient. The results showed that the TSPT technique was the best fit followed by Weibull’s Plotting Position formula, Chow’s, LPIII, EVI and Stochastic methods, in descending order of performance. Therefore, the TSPT method can be used for flood frequency analysis in Ethiopia. Moreover, flood frequency analysis was carried out based on the PyTOPKAPI modelled daily stream flows from the two case study catchments. The results were then compared with those of the traditional ones. It was found that simulation-based flood frequency analysis showed very good agreement with those from the traditional methods for both the case study catchments. It was thus concluded that PyTOPKAPI model-based flood frequency analysis could also be one of the appropriate methods of flood frequency analysis and peak flood estimation for Ethiopian rivers.

Description

Doctor of Philosophy in Civil Engineering. University of KwaZulu-Natal. Durban, 2017.

Keywords

Floods--Ethiopia., Flood forecasting--Ethiopia., Flood forecasting--Developing countries., Floods--Developing countries., Flood forecasting., Floods., Developing countries.

Citation

DOI