friend/nile - unesco · friend/nile final report i forward this is the final report of the capacity...

The FRIEND/Nile Project

Project 513RAB2042, Period 2001-2006

Flow Regimes from International Experimental and Network Data (FRIEND) of the River Nile Basin

FRIEND/NILE

Final Project Report Published by UNESCO Regional Office in Cairo

Cairo 2007

II

The FRIEND/Nile Project

Project 513RAB2042, Period 2001-2006

Flow Regimes from International Experimental and Network Data (FRIEND) of the

River Nile Basin

FRIEND/NILE

Final Project Report Published by UNESCO Regional Office in Cairo

Cairo 2007

Images on the cover page are in public domain. They represent:

Composite satellite image of the White Nile. This file is in the public domain because it was created by NASA. NASA copyright policy states that "NASA material is not protected by copyright unless noted".

(Source: http://en.wikipedia.org/wiki/Nile).

Final Project Report

Published by UNESCO Regional Office in Cairo 2007

Disclaimer

The designations employed and presentation of material through the publication do not imply the expression of

any opinion whatsoever on the part of UNESCO concerning the legal status of any country, territory, city, or its

authorities, or concerning the delimitation of its frontiers or boundaries.

http://en.wikipedia.org/wiki/Nile

FRIEND/Nile Final Report I

Forward

This is the final report of the Capacity Building and Networking of the Nile Countries: FRIEND (Flow

Regime from International Experimental and Network Data) Nile Project – Phase 1 (Budget Code:

513RAB2042). This project was successfully executed by UNESCO Cairo Regional Office and

implemented by University of Dar Es Salaam of Tanzania, National Water Research Center of Egypt,

UNESCO Chair in Water Resources of Sudan, University of Nairobi of Kenya and Ministry of Water

Resources of Ethiopia. The overall coordinator of this project is the Water Resources Research

Institute (WRRI) of Egypt.

The project was funded by the Flemish Government of Belgium through the Flanders – UNESCO

Science Fund-In-Trust cooperation. All project members express their great appreciation to the

Flemish Government of Belgium for funding the project and continued fruitful co-operation with

UNESCO. Furthermore, the project team appreciates the technical support and efforts of the Flemish

experts leaded by Dr. Rudy Herman, Economy, Science and Innovation Department, Flanders

Authority, Belgium.

This report is an achievement resulted from fruitful cooperation between the thematic coordinators

and research teams in the participating Nile countries. These experts played an active role in the

implementation of this project.

Thanks are to all project members, namely: project overall coordinator, thematic coordinators and

researchers who actively contributed to the implementation of this project.

Special thanks are due to UNESCO and UNESCO Regional Office in Cairo who had successfully

executed this project, namely Director of the office, FRIEND/Nile project director and manager and all

water unit staff.

Dr. Radwan Al-Weshah

FRIEND/Nile Director

Regional Hydrologist for Arab States

UNESCO Regional Office in Cairo

[email protected]

Prof. Dr. Mohamed Abd-El Motaleb

FRIEND/Nile Overall Coordinator

Director

Water Resources Research Institute

[email protected]

FRIEND/Nile Final Report II

FRIEND/Nile Final Report III

FRIEND/NILE Project Summary

Project Title Capacity Building and Networking of the Nile Countries: FRIEND (Flow Regime from International Experimental and Network Data) Nile Project (513RAB2042)

Project Site Five Nile Basin countries showed serious interests in implementing

this project (Egypt, Sudan, Kenya, Tanzania, and Ethiopia). The project is open to all other Nile basin countries.

Project Duration 2001-2006

Executing Agency UNESCO Cairo Regional Office, Egypt.

Government Sector Water Resources Ministries and Institutions in the Nile Countries

Project Partner University of Dar Es Salaam of Tanzania, National Water

Research Center of Egypt, UNESCO Chair in Water Resources of Sudan, University of Nairobi of Kenya, Ministry of Water Resources of Ethiopia and with contributions of other Nile countries

Donors The Flemish Government of Belgium

Project Budget Total: $929,700 (direct budget) with in-kind contribution from all

participating countries.

Brief description and Progress

The FRIEND/Nile project, initiated by UNESCO, aims at improving international river basin management of the Nile through improved cooperation amongst the Nile countries in the field of water resources management and regional-scale analysis of hydrological regimes. This has ultimately contributed to meet the basic human needs of safe and clean water supply, as well as to promote sustainable development of the region by securing sufficient quantity of water for the agricultural and industrial and other sectors. The project achieved networking and technical exchange in the water issues related to the project between the implementing institutions and the Flanders specialized universities. A network of water resources experts in the Nile basin and the Flemish community has been established and strengthened. Four research teams have been defined comprising researchers from the participating Nile Countries and Flemish experts. Different activities have been implemented comprising the organization of technical workshops and meetings, technical missions, equipment and software purchase and project technical publications in addition to the organization of the project international conference. Mutual trust, confidence and understandings have been developed among the research teams of the project from the participating Nile basin countries. Research capacities and networking between experts have been achieved. Enhancement of the south-south as well as north-south cooperation through research cooperation is strengthened through better working relationships. The sustainability of the established networking and technical exchange among the implementing institutions and the Flemish specialized universities is a major outcome of the FRIEND/Nile Project.

FRIEND/Nile Final Report IV

FRIEND/Nile Final Report

V

Table of Contents

Forward ___________________________________________________________________ I Project Summary___________________________________________________________III Table of Contents____________________________________________________________V List of Figures______________________________________________________________IX List of Tables ______________________________________________________________XI List of Acronyms and Abbreviations____________________________________________ XII Preface __________________________________________________________________ XV

Chapter 1 : Overview ______________________________________________________1 1.1) Introduction ____________________________________________________________ 1 1.2) Implemented Activities____________________________________________________ 1

1.2.1) Workshops and Meetings______________________________________________________1 1.2.1.1) Flood Frequency Analysis (FFA) Component_______________________________1 1.2.1.2) Rainfall-Runoff Modeling (RRM) Component _______________________________3 1.2.1.3) Drought and Low Flow Analysis (DLFA) Component ________________________5 1.2.1.4) Sediment Transport and Watershed Management (STWM) Component ___________6

1.2.2) Training Workshops and Technical Missions ______________________________________9 1.2.3) Project Governance_________________________________________________________10

1.2.3.1) Steering Committee Meetings __________________________________________10 1.2.3.2) Project Management Meetings _________________________________________12

1.2.4) Other Related Meetings______________________________________________________13 1.2.5) Second Phase Project Document Preparatory Meetings_____________________________14

1.3) Research Activities ______________________________________________________ 15 1.3.1) Data processing and analysis_________________________________________________16 1.3.2) Selection and Introduction of Suitable Models ____________________________________16 1.3.3) Application of the Selected Models_____________________________________________17

1.3.3.1) Flood Frequency Analysis Component ___________________________________17 1.3.3.2) Rainfall /Runoff Modeling Component ___________________________________18 1.3.3.3) Drought and Low Flow Analysis Component ______________________________18 1.3.3.4) Sediment Transport and Watershed Management Component _________________19

1.4) Reporting______________________________________________________________ 19 1.5) FRIEND/Nile Conference ________________________________________________ 20 1.6) The Way Ahead ________________________________________________________ 20

Chapter 2 : Flood Frequency Analysis Component _____________________________21 2.1) Introduction ___________________________________________________________ 21 2.2) Research Activities______________________________________________________ 22

2.2.1) Data Processing ___________________________________________________________22 2.2.2) Models Used ______________________________________________________________24

2.2.2.1) Floods Package (Model 1) ____________________________________________25 2.2.2.1.1) Applications of Model (2) on Selected Sites in the Nile Basin _________26 2.2.2.1.2) Regional Homogeneity and Regional Distribution__________________27 2.2.2.1.3) General Conclusion (Model 2)_________________________________27

2.2.2.2) Extreme Value Analysis Using Quantile-Quantile (Q-Q) Plots (Model 3) ________28 2.2.2.2.1) Applications of the Q-Q Approach to the Selected Sites in the Nile Basin 29

FRIEND/Nile Final Report VI

2.3) Regional Flood Frequency Analysis (RFFA) _________________________________37 2.3.1) Visual inspection of the FFDs _________________________________________________38 2.3.2) Correlation Analysis _______________________________________________________39 2.3.3) Analysis by the L-Moments method ____________________________________________42

2.3.3.1) L-Moments Ratio Diagram ____________________________________________42 2.3.3.2) Discordancy Measure, D (I) ___________________________________________43 2.3.3.3) Heterogeneity Test for Regions _________________________________________44 2.3.3.4) Goodness of Fit Test for Identify Parent Distribution ________________________45

2.4) Limit ations and Constraints ______________________________________________46 2.5) Conclusions ____________________________________________________________46 2.6) The Way Ahead_________________________________________________________46 2.7) References _____________________________________________________________47

Chapter 3 : Rainfall-Runoff Modeling Component _____________________________ 49 3.1) General Introduction_____________________________________________________49

3.1.1) The Rainfall-Runoff Modeling (RRM) Component__________________________________50 3.1.2) Objectives of the Rainfall-Runoff Modeling Component _____________________________51

3.2) Data Acquired for Rainfall-Runoff Modeling_________________________________51 3.2.1) Egypt ____________________________________________________________________51 3.2.2) Ethiopia __________________________________________________________________52 3.2.3) Kenya____________________________________________________________________52 3.2.4) Sudan ____________________________________________________________________53 3.2.5) Tanzania _________________________________________________________________54

3.3) Case Studies in Each Country _____________________________________________54 3.4) Models Used____________________________________________________________57

3.4.1) GFFS ___________________________________________________________________57 3.4.1.1) SLM______________________________________________________________57 3.4.1.2) LPM _____________________________________________________________57 3.4.1.3) LVGFM___________________________________________________________58 3.4.1.4) SMAR ____________________________________________________________59 3.4.1.5 ) Methods of Combining the Estimates of Different Models____________________60

3.4.2) HSPF ___________________________________________________________________62 3.4.3) WMS/HEC-1______________________________________________________________62 3.4.4) SWAT ___________________________________________________________________63 3.4.5) HMS_____________________________________________________________________65

3.5) Obtained Results ________________________________________________________66 3.5.1) GFFS Model Results ________________________________________________________66 3.5.2) WMS/HEC-1 Model Results___________________________________________________69 3.5.3) HSPF Model Results ________________________________________________________70 3.5.4) SWAT Model Results ________________________________________________________72 3.5.5) HMS Model Results _________________________________________________________75

3.6) Findings and Lessons Learned_____________________________________________76 3.6.1) Application of GFFS Models __________________________________________________76 3.6.2) Application of WMS/HEC-1 Model _____________________________________________77 3.6.3) Application of HSPF Model___________________________________________________77 3.6.4) Application of SWAT Model___________________________________________________78 3.6.5) Application of HMS Model____________________________________________________78

3.7) Limitations and Constraints_______________________________________________79 3.8) The Way Ahead_________________________________________________________79 3.9) Acknowledgement _______________________________________________________79 3.10) References ____________________________________________________________80


VII

Chapter 4 : Drought and Low-Flows Analyses Component_______________________81 4.1) Introduction ___________________________________________________________ 81 4.2) Data Requirements and Methods of Analyses ________________________________ 83 4.3) DLFAC Methodologies and Research Findings_______________________________ 85 4.4) Summaries of the DLFAC Research Activities and Methodologies_______________ 90

4.4.1) QDF Relationships for Low Flow Return Period Prediction _________________________90 4.4.1.1) Introduction _______________________________________________________90 4.4.1.2) Results ____________________________________________________________91 4.4.1.3) Relationships between Low-Flow Distribution Parameters and the Aggregation Period ________________________________________________________________________92 4.4.1.4) Conclusions________________________________________________________93

4.4.2) Low Flow Analysis Using Filter Generated Series for Lake Victoria Basin ______________94 4.4.2.1) Introduction________________________________________________________94 4.4.2.2) Results ____________________________________________________________95

4.4.3) Statistical Analysis of Dry Periods in Seasonal Rivers ______________________________95 4.4.3.1) Introduction________________________________________________________95 4.4.3.2) Study Cases ________________________________________________________96 4.4.3.3) Return Period Curves for Dry Spells _____________________________________96 4.4.3.4) Return Period Curves for Dry Period Aggregated Low-Flows _________________97 4.4.3.5) Conclusions________________________________________________________99

4.4.4) Analyses of Annual Droughts in Kenya Using an Objective Annual Rainfall Drought Index _99 4.4.4.1) Introduction________________________________________________________99 4.4.4.2) Data used__________________________________________________________99 4.4.4.3) Results and Discussions______________________________________________101

4.4.4.3.1) L-Moments Ratio Goodness-Fit-Tests__________________________102 4.4.4.3.2) Distribution of the Annual Droughts of Different Return Periods_____102

4.4.4.4) Conclusions_______________________________________________________103 4.4.5) Analysis of the Return Periods of Low Flow Hazards in Egypt and Sudan ______________104

4.4.5.1) Introduction_______________________________________________________104 4.4.5.2) Data Availability and Method of Analysis ________________________________105 4.4.5.3) Conclusions_______________________________________________________106

4.5) Achievements and Lessons Learned _______________________________________ 107 4.6) Limitations and Constraints _____________________________________________ 107 4.7) The Way Ahead _______________________________________________________ 108 4.8) References ____________________________________________________________ 108

Chapter 5 : Sediment Transport and Watershed Management Component _________109 5.1) Introduction __________________________________________________________ 109 5.2) Background___________________________________________________________ 109

5.2.1) Nile River Basin ___________________________________________________________110 5.2.2) River Nile Watershed _______________________________________________________111

5.3) STWMC Objectives ____________________________________________________ 114 5.3.1) General Objective _________________________________________________________114 5.3.2) Specific Objectives _________________________________________________________114

5.4) Data Acquisition _______________________________________________________ 114 5.4.1) Necessary Data for Sediment Transport Modeling ________________________________114 5.4.2) Case Study in Each Country _________________________________________________116

5.4.2.1) Kenya ___________________________________________________________116 5.4.2.2) Ethiopia__________________________________________________________117 5.4.2.3) Sudan____________________________________________________________118 5.4.2.4) Egypt ____________________________________________________________119

FRIEND/Nile Final Report VIII

5.4.2.5) Tanzania _________________________________________________________120 5.4.3) Summary of Available Data in Each Case Study__________________________________120

5.5) Methodology __________________________________________________________121 5.6) Sediment Transport Modeling Software ____________________________________122

5.6.1) Selection and Testing of Sediment Transport Modeling Software _____________________122 5.6.2) Calibration of SMS Model Software ___________________________________________123

5.6.2.1) Description of the SMS Software_______________________________________123 5.6.2.2) Basic Data Required by the SMS software _______________________________124

5.6.3) Problems Encountered in Modeling Task _______________________________________124 5.7) Case Studies ___________________________________________________________125

5.7.1) A Comparison between Two Different Transport Models to Predict Sediment Transport at the Simiyu River, Tanzania, as a Case Study ________________________________________125 5.7.2) Modeling of Sedimentation Process in Aswan High Dam Reservoir ___________________127 5.7.3) Nile River Sediment Modeling: Challenges and Opportunities _______________________127 5.7.4) Overview of Sediment Problems in Nile Basin ____________________________________128 5.7.5) Modeling Water and Sediment Fluxes in Steep River Channels: Case of Awash Basin _____129 5.7.6) Limitations of Hydro-dynamical Models with Limited Data Available Case Study: Sondu River Basin (Kenya)_____________________________________________________________132 5.7.7)Overview of Soil Erosion around Lake Victoria ___________________________________132 5.7.8)Effect of Upstream Structures on Delta Progress in Aswan High Dam Reservoir _________133

5. 8) Remarks on the Results _________________________________________________133 5..9) Limitations and Constraints _____________________________________________135 5.10)Conclusions ___________________________________________________________136 5.11)The Way Ahead _______________________________________________________136 5.12)Some Data and Results Listing ___________________________________________136

Appendix A. Management Team ________________________________________ 143

Appendix B. Research Teams ____________________________________________ 145

Appendix C. List of the papers published in the FRIEND/Nile Conference________ 151

Appendix D. List of Technical Reports 155


IX

List of Figures • Figure 2-1, • Figure 2-2, • Figure 2-3, • Figure 2-4, • Figure 2-5, • Figure 2-6, • Figure 2-7, • Figure 2-8, • Figure 2-9, • Figure 2-10, • Figure 2-11, • Figure 2-12, • Figure 2-13, • Figure 2-14, • Figure 2-15, • Figure 2-16, • Figure 2-17, • Figure 2-18, • Figure 2-19, • Figure 2-20, • Figure 2-21, • Figure 2-22, • Figure 2-23, • Figure 3-1, • Figure 3-2, • Figure 3-3, • Figure 3-4, • Figure 3-5, • Figure 3-6, • Figure 3-7, • Figure 3-8, • Figure 3-9, • Figure 3-10, • Figure 3-11, • Figure 3-12,

Three main regions of Lake Victoria ______________________________________ Flood frequency curves for the rivers of region 1, and the corresponding regional flood frequency curve.________________________________________________________________________ Flood frequency curves for the rivers of region 2, and the corresponding regional flood frequency curve. _________________________________________________________ Regional flood frequency curves for the main three regions; (region 1: N-E; region 2: S-E; region 3: river Kagera - west of Lake Victoria).___________________________________________ Distribution comparison, River Sobat (left) and River Pibor (right), MOM method (extreme value paper).______________________________________________________________________ Distribution comparison, River Yei (left) and River Jur (left), MOM method (extreme value paper).______________________________________________________________________ Different classes of distribution’s tail according to extreme value index (γ)._____________ Rivers and stations in North-Eastern and South-Eastern side of Lake Victoria.___________ Two examples to compare different EV-1 distributions in the Sobat Region; River Sobat at Hillet Doleib (left) and River Pibor (right).______________________________________________ Two examples to compare different EV-1 distributions in River Sudd Region; River Yei (left) and River Lol at Nyamlell (right).___________________________________________________ Exponential Q-Q plot for non-flooded and flooded events; River Pibor in Sobat region (left) and River Akkobo in Sudd region.___________________________________________________ Examples of the exponential Q-Q plot for EV1/Gumbel using MOM; Blue Nile at EL- Deim (left) and River Setitte at Hawata (right).___________________________________________ EV-1 and GEV distributions for the flooded and non-flooded segments of Blue Nile at Ed Deim station (left) and River Rahad at Heleiw station (right)._______________________________ Ev1 and GEV distribution plot for River Sondu (left) and River Nyando (right).____________ GEV distribution plot for River Nzoia (left) and River Awach (right)._______________________ Exponential Q-Q plot for EV1/Gumbel using MOM, ML, and PWM for stationNgono/Kyaka (left) and Moame/Mabuki (right)._________________________________________________ Comparison of EV1 and Extreme value distributions of Akaki (left), and flooding effect at Teji Rivers (right)._______________________________________________________________ Regional Flood Frequency Distribution (EV-1) for River Sobat and its Sub-Basins (left) and River in Sudd Region (right).___________________________________________________________ Regional data and regional frequency curves for the Blue Nile and Atbara River.___________ Regional data and regional frequency curves for the rivers in Awash basin in the Ethiopian plateau._____________________________________________________________________ Correlation of the MAF with both the Areas and the MAR (mm) for the Blue Nile and Atbara river._______________________________________________________________________ Correlation of the MAF with both the Areas and the MAR (mm) rivers in the North-astern side of Lake Victoria.________________________________________________________________ Moments ratio diagram presenting CS2 (left) and CS (right) versus kurtosis for the River Sobat region._________________________________________________________ The study basins around lake Victoria and Nile basin._________________________________ Schematic diagram of the Artificial Neural Network model.____________________________ Schematic diagram of SMAR Model._____________________________________________ HSPF conceptual hydrologic model.______________________________________________ Sub-basin command loop.______________________________________________________ Schematic diagram of HMS-SMA algorithm (HEC 2000).____________________________ Plots of observed and SMAR simulated discharges 1999. ________________________ Plots of observed and AR simulated discharges 1999.____________________________ Comparison between observed and simulated hydrographs at Wadi AL-Arbain using 1- method of unit hydrograph.______________________________________________________________ Comparison between observed and simulated hydrographs at Wadi AL-Arbain using method of losses._______________________________________________________________________ Comparison between observed and simulated hydrographs at Wadi Sudr using the method of losses._____________________________________________________________________ Final HSPF calibrated model run; simulated and observe stream flows for Simiyu Watershed at Road Bridge Station.__________________________________________________________

25 26 26 27 27 28 29 29 30 30 31 33 33 34 34 36 37 38 39 39 41 41 43 55 58 59 61 64 65 67 68 68 69 69 71

FRIEND/Nile Final Report X

• Figure 3-13, • Figure 3-14, • Figure 3-15, • Figure 3-16, • Figure 3-17, • Figure 3-18, • Figure 3-19, • Figure 4-1, • Figure 4-2, • Figure 4-3, • Figure 4-4, • Figure 4-5, • Figure 4-6, • Figure 4-7, • Figure 4-8, • Figure 4-9, • Figure 4-10, • Figure 4-11, • Figure 4-12, • Figure 4-13, • Figure 4-14, • Figure 4-15, • Figure 5-1, • Figure 5-2, • Figure 5-3, • Figure 5-4, • Figure 5-5, • Figure 5-6, • Figure 5-7, • Figure 5-8, • Figure 5-9, • Figure 5-10, • Figure 5-11, • Figure 5-12, • Figure 5-13, • Figure 5-14, • Figure 5-15, • Figure 5-16, • Figure 5-17, • Figure 5-18, • Figure 5-19, • Figure 5-20, • Figure 5-21, • Figure 5-22,

Validating the calibrated model of HSPF; validated and observe stream flows for Simiyu Watershed at Road Bridge Station._______________________________________________ Re-classed land use data for the SWAT simulations____________________________ Observed and estimated daily discharge 1972-1973 at Ndagalu_________________________ Annual rainfall and potential evaporation during calibration and validation periods.________ Observed and estimated annual daily discharge 1976-1983 at Ndagalu.__________________ Simulated and observed hydrographs during calibration; HMS results.__________________ Simulated and observed hydrographs during validation; HMS results.___________________ Return period curve for Eddeim 1 day low flows.___________________________________ Relationship between the distribution parameters β qt and the aggregation period D for Eddeim low flows.__________________________________________________________________ QDF plots for the Blue Nile low flows at Eddeim.__________________________________ QDF plot for the river Nzoia low flows at 1DD01 station.____________________________ Exponential Q-Q plot indicating a normal tail exponential distribution for dry spells and he Return period curve of dry spells at Kubur tion.________________________________________ Pareto Q-Q plot indicating a heavy tail distribution for the 1/Q dry period aggregated flows and the Return period curve of 1/Q dry period aggregated flows at Kubur station.________________ Pareto Q-Q plot indicating a heavy tail distribution for the 1/Q dry period aggregated flows and the Return period curve of 1/Q dry period aggregated flows at Hileiw station.________________ River Atbara average hydrograph.________________________________________________ Operation rules of the Khash el Girba Dam upstream River Atbara._____________________ Map of Kenya showing the location of the rainfall stations which were used in the study.___ Distribution of the Annual Drought Index in Comparison to the Distribution of the Annual Rainfall in two selected locations in Kenya._______________________________________ The sample L-moment and the GEV distribution L-moments for the Annual Drought Indices data in Kenya.___________________________________________________________________ Distribution of annual drought indices corresponding to the 50, 200 and 500 year GEV return periods.____________________________________________________________________ Location Map of the Selected Sites.______________________________________________ Natural Flow Series of Mean Annual Values at the Selected Stations.___________________ The Three Watersheds of the Nile River.___________________________________________ Schematic Diagram of the Nile River Natural Flows._________________________________ Hydrograph of the Nile River.___________________________________________________ Rivers in Sondu River Basin.___________________________________________________ Location map of the Aswash river study area._______________________________________ Locations of the X-sections in the Blue Nile river.___________________________________ Location of Aswan High Dam and Dongola station__________________________________ The Simiyu river with the stream network._________________________________________ A map showing bottom bed change after 144000 hours of simulation (Case study: Simiyu River)._____________________________________________________________________ Bed changes at different locations (Case study: Simiyu River)._________________________ Comparison of measured and predicted longitudinal profile for AHDR 2001 and 2003._____ Suspended Sediment Concentration in AHD Reservoir._______________________________ Comparison of rainfall, Discharge and Sediment Yield in the River Atbara (left) and the Blue Nile (right)._____________________________________________________________________ Sediment Volume and Content of Sennar Dam (left) and Roseires Dam (right).___________ Water Depth in the selected reach of Awash River.___________________________________ Average Velocity in selected reach of Awash River.__________________________________ A comparison between measured and simulated water depth in Awash River._____________ Results of Water velocity for the 5Km. Stretch ( Sondu River).________________________ Results of the Water depth at 5 km Stretch ( Sondu River).____________________________ Results of the water surface elevation ( Sondu River).________________________________ Results of bed change at t=0 ( Sondu River).________________________________ Results of bed change at t=1440 ( Sondu River).____________________________________

71 73 73 74 74 75 76 92 92 93 93 96 97 97 98 98 100 100 103 103 104 105 112 113 113 116 117 119 119 120 126 126 126 129 128 129 130 130 130 131 131 131 132 132


XI

List of Tables

• Table 1-1, Catchment characteristics of the researched catchments_______________________________18 • Table 1-2, Best model efficiency criteria results. _____________________________________________18 • Table 2-1, Data; Rivers, Stations, Locations and Flow record length of the FFAC.___________________23 • Table 2-2, Results for different parameter estimation methods for River Sobat and its sub-basins. _______31 • Table 2-3, Summary of distribution parameters o the EV-1 with MOM.____________________________32 • Table 2-4, Summary of distribution parameters for the rivers in N-E Side of Lake Victoria. ____________35 • Table 2-5, Summary of distribution parameters for the rivers in S-E Side of Lake Victoria._____________36 • Table 2-6, Overview of Catchment Characteristics in Blue Nile and Atbara River. ___________________40 • Table 2-7, Overview of catchment characteristics of the Rivers in Kenya. __________________________40 • Table 2-8, Summary results of the Discordancy Test for the Rivers of Sobat and Sudd Regions. _________44 • Table 2-9, Goodness of Fit Test (Z-Values) for River Sobat and Sudd regions. ______________________45 • Table 3-1, Timetable, activities and output of the rainfall-runoff research component. ________________49 • Table 3-2, Hydro-meteorological data: Ethiopia._____________________________________________52 • Table 3-3, hydro-meteorological data: Nzoia and Sondu catchments, Kenya. _______________________53 • Table 3-4, Hydro-meteorological data: Lake Victoria._________________________________________53 • Table 3-5, Data used for the GFFS modeling in the Victoria catchments___________________________54 • Table 3-6, The Physiographic characteristics of the Wadi Sudr and Al-Arbian of Sinai catchments. ______56 • Table 3-7 Nile catchments summarized characteristics. _______________________________________56 • Table 3-8, Model efficiencies in percentages for the simulation mode._____________________________66 • Table 3-9, Model efficiencies in percentages for the updating mode. ______________________________67 • Table 3-10, The WMS/HEC-1 parameters set up for the derived hydrographs of the storm of 22/3/1991 . ___70 • Table 3-11, Calibration Assessment Curve for Calibration and Validation of Stream Flows for Simiyu Watershed at Road Bridge Station. _______________________________________________71 • Table 3-12, Simiyu Land use classes matched with the SWAT land use classes._______________________72 • Table 3-13, Average long-term water balance 1970-1974 for SWAT model. _________________________73 • Table 4-1, Sample Inventory (in time-scales) of drought and low-flow problems in the Nile basin. _______83 • Table 4-2, Data Requirements per country. _________________________________________________85 • Table 4-3, details for the catchment chosen as case studies for Kenya. ____________________________86 • Table 4-4, details for the catchment chosen as a case study for Egypt._____________________________87 • Table 4-5, details for the catchment chosen as a case study for Sudan. ____________________________88 • Table 4-6, details for the catchment chosen as a case study for Tanzania. __________________________89 • Table 4-7, Calibration result of distribution parameters for Eddeim low flows. ______________________91 • Table 4-8, The AIC Estimates for the Log-GEV and Log-Normal Distributions for the selected Study- Stations. _____________________________________________________________101 • Table 5-1, Nile Basin: areas and rainfall by country. _________________________________________111 • Table 5-2, Summary of available data in each case study._____________________________________121 • Table 5-3, Daily Observation of Sediment Data Awash River Basin: Awash River at Hombole Station (Ethiopia case). _______________________________________________________137 • Table 5-4, Sediment flow data for Simiyu River outfall (Tanzania case). __________________________140 • Table 5-5 10 days-Mean Sediment Concentration for the Blue Nile at Different Locations (Sudan) case) _____________________________________________________________________140 • Table 5-6, Maximum Sediment Concentration during the flood season 2002 in different locations of the Blue Nile System (Sudan case).____________________________________________141 • Table 5-7, Suspended Sediment Concentration before AHD (1929-1955) (Egypt case). ______________141

FRIEND/Nile Final Report XII

List of Acronyms and Abbreviations

AHD: Aswan High Dam, AIC: Akaike Information Criterion, AM: Annual Maximum, AMHY: Alpine and Mediterranean HYdrology, ANN: Artificial Neural Network, AR: Auto-Regressive, CLS: Constrained Least Squares, CN: Curve Number, DEM: Digital Elevation Model, DLFA: Drought and Low Flow Analysis Component, DMCN: Drought Monitoring Centre in Nairobi, ESTC: Erosion and Sediment Transport Component, EV1: Extreme Value type 1, FEWS NET: Famine Early Warning System Network, FFA: Flood Frequency Analysis, FFCs: Flood Frequency Curves, FIGCC: Inter-Group Coordination Committee, GEV: Generalized Extreme Value, GFFS: Galway Flood Forecasting System, GIS: Geographic Information System, GLOG: The Generalized Logistic Distribution, GPAR: The Generalized Pareto Distribution, GPDs: Generalized Pareto Distributions, HEC-1: Hydrologic Engineering Center package 1 (Flood Hydrograph Package), HMS: Hydrologic Modeling System, HRU: Hydrologic Response Units, HSPF: Hydrological Simulation Program - Fortran, ICPACIGAD: Climate Prediction and Applications Center, IWRM: Integrated Water Resources Management, LOGN: The Three-parameter Lognormal Distribution, LPM: Linear Perturbation Model, LTF: The Linear Transfer Function, LVGF: Linear Varying Gain Factor model, MAF: Mean Annual Floods, MAR: Mean Annual Rainfall, MLM: Maximum Likelihood Method, MOM: Method of Moments, MSE: Mean Square Error, NBCBN-RE: Nile Basin Capacity Building Network for River Engineering, NBI: Nile Basin Initiative, NNM: The Neural Network Method, OLS: Ordinary Least Squares, POT: Peak over Threshold, PWM: Probability Weighted Moments, QDF: Flow (Q)-Duration-Frequency analysis, QQR: Quartile-Quartile plots, RFFA: Regional Flood Frequency Analysis, RFFCs: Regional Flood Frequency Curves, RMA2: A dynamic two-dimensional depth-averaged finite element hydrodynamic

model for computing water surface elevations and horizontal velocity components for subcritical, free-surface flow,

RMMC: Rainfall-Runoff Modeling Component,


XIII

SAM: The Simple Average Method, SCS: Soil Conservation Service, SED-2D: A dynamic, 2-dimensional finite element model for vertically-averaged

sediment transport. Noncohesive (sand) and cohesive (clay) sediments can be simulated, but not simultaneously,

SLM: Simple Linear Model, SMA: Soil Moisture Accounting, SMAR: Soil Moisture Accounting and Routing, SMS: Surface Water Modeling System, STWM: Sediment Transport and Watershed Management, SWAT Soil and Water Assessment Tool, UCI: User Control Input, UCWR: UNESCO Chair in Water Resources, USGS: United States Geological Survey, WAM: The Weighted Average Method, WM: Weighted Moments, WMS: Watershed Modeling System, WRRI: Water Resources Research Institute.

FRIEND/Nile Final Report XIV


XV

Preface

The FRIEND (Flow Regimes from International and Experimental Network Data) Project was

originally established by UNESCO in 1985 as part of the International Hydrological Programme (IHP).

In general, the Global FRIEND project aims at creating the necessary and sufficient knowledge in

addition to understanding the flow regimes on various scales based on regional data of experimental

basins and hydrological networks. Currently, there are ten FRIEND Projects covering the world.

The FRIEND/Nile project is a member of the global FRIEND family. It was initiated by UNESCO in

March 1996. It aims at creating better understandings and quantification of the river Nile system to

enhance the management of the Nile water resources and to improve the planning of water

resources projects in the Nile Basin countries.

Within the framework of the Flanders-UNESCO Science Fund-In-Trust cooperation, the Flemish

Government of Belgium supported the first phase of the FRIEND/Nile project for the period 2001-

2006 with an amount of US$ 929,700. The project has been officially launched in November 2001

and fully completed in early 2006.

Five Nile Basin countries have been actively contributing to the implementation of the research

activities of the project, namely: Egypt, Ethiopia, Kenya, Tanzania and Sudan. However, the project is

opened to all interested Nile basin countries. The Water Resources Research Institute is the overall

coordination center of the project.

During the first phase of the FRIEND/Nile project, a network of water resources experts in the Nile

basin and the Flemish community has been established. Four research teams have been defined

comprising researchers from the participating Nile Countries and Flemish experts. The technical

themes of FRIEND/Nile Project are:

1. Rainfall-Runoff Modeling, coordinated by the University of Dar Es Salaam of Tanzania;

2. Sediment Transport and Watershed Management, coordinated by the UNESCO-Chair in Water Resources of Sudan;

3. Flood Frequency Analysis, coordinated by the Water Resources Research Institute of Egypt; and

4. Drought and Low Flow Analysis, coordinated by the University of Nairobi of Kenya.

UNESCO Cairo Office has been successfully executing the FRIEND/Nile project in joint collaboration

with all stakeholders. Different activities have been implemented including:

• The organization of twenty two training workshops and seven technical and administrative meetings in Tanzania, Kenya, Ethiopia, Sudan, and Egypt;

• Data acquisition through the themes researchers;

• Equipment purchase for the project members in Tanzania, Kenya, Ethiopia, Sudan, and Egypt;

FRIEND/Nile Final Report XVI

• Securing proper needed software for the implementation of the research activities of the FRIEND/Nile components; and

• Development of a brochure and website for the project.

Over 200 experts from the Nile basin including project team members and a large number of

researchers in the participating countries in addition to the Flemish counterparts have participated in

the project activities. Female experts and young scientists were involved in the project activities.

The project has provided new methodologies, tools, technologies, and software as an effective

approach for enhancing the institutional and human resources capacity building in the field of the Nile

Basin water resource management.

Trust, confidence and better understandings have been built among the research teams. The

exchange of results among the different researchers has been a successful exercise that has been

achieved based on mutual trust and confidence among the research teams in the course of the

FRIEND/Nile project.

Furthermore, exchange of experience and hydrological knowledge between researchers and

scientists in the Nile countries and the Flemish community has been achieved during the first phase

of the FRIEND/Nile Project. Also, the project enhanced the research cooperation between the

implementing institutions in the Nile basin countries in general and with the Flemish universities in

particular. This is a good model for South-North and South-South cooperation that has been

envisaged as one of the main project objectives.

An international conference on the FRIEND/Nile Project was organized during the period 12-14

November 2005 in Sharm El Shiekh, Egypt to present the results of the research activities of the

project. Twenty-seven joint technical papers were presented by the research teams of the project

reflecting the remarkable achieved technical regional cooperation. About 120 participants from more

than 12 countries, comprising international and regional key-water experts, policy makers from the

Nile countries, FRIEND/Nile researchers, Flemish experts, stakeholders and representatives of the

ongoing Nile initiatives participated in the conference.

On the other side, additional technical Flemish contribution to the FRIEND/Nile project had

contributed to the successful implementation of the project, namely:

1. The secondment of Prof. Willy Bauwens to the FRIEND/Nile Project: Rainfall-Runoff Modeling component; and

2. Supporting candidates from Egypt and Sudan to participate in the Scientific and Technological Information Management in Universities and Libraries: an Active Training Environment.

Finally, it can be concluded that the first phase of the FRIEND/Nile project has successfully achieved the following results:

• improved institutional and human resources capacity building in the field of water resources of the Nile basin countries;


XVII

• fostered networking, regional cooperation and experience exchange among water experts and institutions in the Nile basin countries;

• enhanced knowledge and understanding of the hydrological processes of the River Nile;

• enhanced research cooperation among Nile Basin countries, through applied hydrological research in priority areas identified by the participating countries; and

• developed methodologies and promoted relevant applied hydrological research in the Nile basin.

It is important to ensure the sustainability of this project by developing a follow up mechanism that will

be built on the achieved results. A second phase of the project covering the period 2006-2010 is

being adopted.

FRIEND/Nile Final Report XVIII

FRIEND/Nile Final Report 1

Chapter

1

Overview

1.1) Introduction

The FRIEND/Nile project is one of the major UNESCO/FRIEND projects to strengthen and enhance

the research cooperation between Nile basin countries for a better understanding of hydrological

regimes of the Nile basin. The FRIEND/Nile Project is a Science Fund in Trust Project funded by the

Flemish Government of Belgium for a duration of 4 years starting November 2001, and aims also at

enhancing the capacity building and networking for the Nile countries. The project is executed by the

UNESCO Cairo Office. Four research themes are supported by the project:

1. Flood Frequency Analysis (FFA); Water Resources Research Institute (WRRI) of Egypt;

2. Rainfall-Runoff Modeling (RMM); coordinated by the; coordinated by University of Dar Es

Salaam of Tanzania;

3. Droughts and Low Flow Analysis; coordinated by University of Nairobi of Kenya; and

4. Sediment Transport and Watershed Management (STWM); coordinated by the UNESCO-

Chair in Water Resources of Sudan.

It was agreed that WRRI of Egypt to be the overall coordinator of the project. This chapter

summarizes and presents the overall implemented project activities of the FRIEND/Nile project during

the period 2001-2006. The main items of this chapter are: the implemented activities, the considered

problems and constraints, the way ahead and the priority issues to be raised for the next phase of the

project.

1.2) Implemented Activities

1.2.1) Workshops and Meetings

1.2.1.1) Flood Frequency Analysis (FFA) Component

In this section implemented activities of FFAC are listed as follows: FRIEND/Nile


• A workshop for Flood Frequency Analysis, and Drought and Low Flow Analysis was organized: in

Cairo, Egypt on 11 – 16 November 2001. Sixteen experts from the Nile Basin Countries and

Belgium participated in the workshop to identify priority research areas and to prepare a detailed

work plan and budget for this component for the project duration as per the direction of the project

management.

• The second Flood Frequency Analysis Workshop was held in Cairo, Egypt, in the period 1-3 April

2003. Twenty seven key experts participated in this workshop representing research team of the

Flood Frequency Analysis (FFA) component in Egypt, Sudan, Kenya, Tanzania, and Ethiopia, the

Flemish counterparts, and resource persons. The main aims of the meeting was to review the

progress in implementing the activities of the component and prepare the second year Work plan,

list of planned activities and their expected dates. Moreover, deliverables of each FFA theme

researcher were identified. Also, a group of free FFA software were compiled on a CD and

distributed to the FFA theme researchers. The workshop was covered by the media.

• The third workshop took place in Sharm El-Shiekh, Egypt in the period between 29th of November

to 2nd of December 2003. The research teams of the Flood Frequency Analysis (FFA) component

in Egypt, Sudan, Kenya, Tanzania, and the Flemish counterparts have participated in this

workshop. The FFA theme researchers presented their technical reports on regional analysis for

the different regions of the Nile Basin. The deliverables of each researcher were reviewed and

evaluated. A number of working group sessions were carried out where FFA experience has

been exchanged among the FFA researchers. Problems with data shortage and inconsistencies

in the approach presented by the different countries were identified and discussed. Procedures

for harmonizing the methodology of the regionalization analysis has been outlined and approved.

The third year work plan and activities was also identified.

• The fourth workshop was held in Borg El Arab, Egypt in the period 22-24 June 2004. The

Objectives of this workshop were to review the progress in implementing the FFA research

activities since the last FFA workshop and to adopt on a regional Flood Frequency Analysis, in

addition to prepare a detailed list of activities for each FFA theme researcher. Also, exchange

ideas and experiences among theme researchers was another of this workshop. Moreover, it

aimed at creating a platform of trust and confidence among the FFAC researchers in the Nile

Basin countries. Participants of this workshop were the Overall Coordinator, the Project Director;

the project manger, and the component theme researchers from Tanzania, Kenya, Sudan and

Egypt, besides the Flemish Counterpart. The implemented research work of the FFA by the

theme researchers was presented and discussed. Also, application of proposed new

methodologies for the regionalization analysis was discussed. Moreover, encountered problems

for conducting Regional Flood Frequency Analysis (RFFA) in the different countries were

reviewed and identified. In this workshop, a harmonized RFFA


• methodology was recommended and adopted by the FFA theme researchers in the different

countries. Finally, future works were planed.

• The fifth workshop took place in Nairobi, Kenya in the period 26th to 29th of November. About

eleven key experts participated in this workshop representing research teams of the Flood

Frequency Analysis (FFA) component in Egypt, Sudan, Kenya, Tanzania, and Ethiopia and the

Flemish counterpart. The implemented regional frequency analysis in these countries was

presented, reviewed and discussed. GIS visualization of the results for the whole Nile basin was

introduced. The USGS-DEM raw data of the whole area of the Nile Basin with resolution of

90m×90m was distributed to all FFA themes researchers. GIS manipulation of the DEM data was

guided by the Flemish counterpart to extract more physiographic parameters for the

enhancement of the regionalization analysis. Some statistical homogenous regions within the

Nile basin were defined. The workshop participants identified the framework and time schedule of

the FFA technical papers to be presented in the Final FRIEND/Nile International Conference as

an output of the FFA Component during the first phase of the FRIEND/Nile project. It was agreed

that a total of five FFA papers will be prepared for the November 2005 conference Future

research activities were defined for each of the research theme researcher.

• Finally, the sixth workshop was held in Khartoum, Sudan in the period 25-30 July 2005. Key

experts participated in this workshop representing the research teams of the FRIEND/Nile Project

in Egypt, Sudan, Kenya, Tanzania, and Ethiopia and the Flemish counterpart. The implemented

research activities of the FFA component in Kenya, Tanzania, Sudan, Ethiopia and Egypt were

presented and discussed. Improvement in the preparation of the technical papers to be

presented in the Final FRIEND/Nile International Conference was recognized in all countries. The

participants discussed and reviewed thoroughly all papers. The Flemish counterparts presented

their comments on each technical paper. Intensive working group sessions were conducted to

finalize the papers and to adopt the reviewers’ comments. The papers were totally finalized and

reviewed during the workshop. The workshop participants identified the framework and time

schedule of finalizing the rest of the FRIEND/Nile papers. Also, the outlines of the annual

progress report for the fourth year of the FFA component were discussed and reviewed by the

research team of the FFA component. Moreover, the outlines of the conference agenda and

schedule of activities were discussed and reviewed.

1.2.1.2) Rainfall-Runoff Modeling (RRM) Component

Several workshops were organized for the Rainfall-Runoff component. They are summarized as

follows:

• The first rainfall-runoff modeling workshop was held in Bagamoyo, Tanzania in the period 1-5

November, 2001. Twelve experts from the Nile Basin Countries and Belgium participated in the


workshop to identify priority research areas and to prepare a detailed work plan and budget for

this component for the duration of the project as per the direction of the project management.

• The second workshop was held in Alexandria, Egypt in the period 20-25 July, 2003. The

research team of the RRM in addition to Flemish counterpart, and resource persons attended this

concurrent training workshop. Two hydrological simulation models, namely: the Watershed

Modeling System (WMS) and Galway Flood Forecasting System (GFFS) were introduced to the

RRM research team focusing on the software modules/interfaces, automated watershed

delineation procedures, watersheds analysis and visualization tools and the different applications

of the software. The Hydro-meteorological data of Tanzania, Kenya, Ethiopia, Sudan, and Egypt

were compiled on a CD and distributed to the RRM theme researchers. The CD of the WMS and

training manuals were distributed to the RRM theme researchers. Also, the USGS - DEM data of

the whole area of Africa with resolution of 1km×1km was compiled on a CD and distributed to all

RRM theme researchers. Furthermore, the implemented research activities of the component

were discussed and presented. Additionally, the deliverables of each RRM theme researcher

during the second year of the project were defined based on the applications of the selected

software on the data of the case study areas in each country.

• The third workshop took place in Dar Es Salaam, Tanzania in the period 5 – 9 of January 2004.

All the rainfall–runoff modeling (RRM) theme researchers of the participating countries in addition

to the Flemish counterpart and representative of the USGS Famine Early Warning System

Network (FEWS NET) project attended this workshop. The implemented research activities of the

RRM component in Kenya, Tanzania, Sudan, Ethiopia and Egypt have been presented and

discussed. The encountered WMS and GFFS application problems and difficulties have been

reviewed and discussed. Solutions have been provided by the WMS and GFFS resource

persons. Intensive group working sessions were undertaken using real data of the selected

basins under the supervision of the resource persons. Various hydrologic models, incorporated in

both WMS and GFFS software, were also used. Technical progress pertaining to the application

of the WMS and GFFS software was achieved and noted. Moreover, the 3rd year (2004) RRM

work plan has been prepared. Finally, the future work and the next workshop, and dates for

submission of technical reports to the coordinating center were set.

• The fourth workshop was held in Addis Ababa, Ethiopia in the period 20-24 September, 2004. All

the rainfall–runoff Modeling (RRM) theme researchers of the participating countries in addition to

the Flemish counterpart and resource persons attended this workshop. The implemented

research activities of the RRM component in Kenya, Tanzania, Sudan, Ethiopia and Egypt have

been presented and discussed. The applications of the GFFS model on the case study areas in

Ethiopia, Kenya, Sudan and Tanzania, in addition to WMS/HEC-1 model on the case study areas

in Egypt have been finalized. The encountered WMS/HSPF application problems and difficulties

have also been reviewed and discussed where solutions have been provided by the WMS/HSPF

resource persons. Moreover, two hydrological simulation models, namely: WMS/HMS and SWAT


were generally introduced. WMS/HMS hands-on training was conducted to the RRM theme

researchers of Kenya, Ethiopia, Sudan and Egypt while SWAT hands-on training was conducted

to the RRM theme researchers of Tanzania and Sudan. Intensive working group sessions were

undertaken using real data of the selected basins. Technical progress pertaining to the

application of the WMS/HMS and HSPF software was achieved and noted. On the other hand,

the input data and parameters of the SWAT software were set up and prepared.

• The fifth workshop was held in Khartoum, Sudan on 25-30 July, 2005. The research team of the

RRM component representing Egypt, Sudan, Kenya, Tanzania, and Ethiopia participated in this

workshop. The implemented research activities of the RRM component in Kenya, Tanzania,

Sudan, Ethiopia and Egypt were presented and discussed. Improvement in preparation of the

technical papers to be presented in the Final FRIEND/Nile International Conference was

recognized in all countries. The participants discussed and reviewed thoroughly all papers. The

Flemish counterpart and the RRM coordinator gave some comments on each technical paper.

Intensive working group sessions were conducted to finalize the papers and to adopt the

reviewers’ comments. The papers were totally finalized and reviewed during the workshops. The

workshop participants identified the framework and time schedule of finalizing the rest of the

RRM papers. Also, the outlines of the annual progress report for the fourth year of the project was

discussed and reviewed by the research team. Finally, the participants discussed the structure

and outlines of the proposed second phase of the project.

1.2.1.3) Drought and Low Flow Analysis (DLFA) Component

Several workshops were held for the DLFA component. These include:

• The first Drought and Low Flow Analysis (DLFA) workshop was held in Nairobi, Kenya in the

period 25-28 August, 2003. The DLFA research teams in Kenya, Tanzania, and Egypt in addition

to the Flemish counterpart, the overall coordinator, and UNESCO representative have

participated in this workshop. The implemented activities of the DLFA component in Kenya,

Tanzania, and Egypt have been reviewed. The data requirements for the analysis of drought and

low flows in the region have also been defined for each country. Moreover, methodologies for

analyzing drought and low flows have been reviewed and approved to be adopted within the

component activities. Training on the applications of the Partial-Duration-Series method for

analyzing actual low flows data from the Nile basin was undertaken and future research activities

were defined for each theme researcher. Some future activities within specified timeframe were

proposed (e.g. identification of available data/each country, data collection and screening,

applications of adopted methodologies and submission of reports).

• The second workshop took place in Alexandria, Egypt in the period 18-21 of June, 2004. The

DLFA researchers of Kenya, Tanzania, Sudan and Egypt in addition to the Flemish counterpart,

the overall coordinator, and UNESCO representative have participated in this workshop. In this

workshop, the implemented activities of the DLFA component in the participating countries have


been reviewed. Also, methodologies for analyzing drought and low flows have been reviewed

and approved to be adopted within the component activities. Training and applications using the

Peak over Threshold (POT) method for analyzing actual low flows data from the Nile basin were

undertaken. Three models for analyzing low flow based on the application of the POT method as

developed by the Flemish counterpart were distributed to the DLFA theme researchers. A

number of group working sessions were undertaken during the workshop and obtained results

were presented and summarized at the end of the workshop. Some future activities (for example

identification of the available data/each country, data collection and screening, applications of the

adopted methodologies on more stations) were proposed. Future research activities were defined

for each of the research theme researcher.

• The third workshop was held in Nairobi, Kenya in the period 23-26 of November, 2004. All the

Drought and Low Flow Analysis (DLFA) theme researchers in the participating countries in

addition to the Flemish counterpart and resource persons attended this workshop. Advanced

training and applications using the Peak over Threshold (POT) method for analyzing actual low

flows data from the Nile basin were continued. Regionalization of the low flow analysis was

discussed and reviewed. Methodology for developing Discharge-Drought-Frequency (QDF)

relationships was also introduced. The QDF curves were developed using real data from the Nile

countries by the DLFA research team. The obtained results and outputs were presented and

summarized at the end of the workshop. Moreover, the workshop participants identified the

framework and time schedule of the preparation of four DLFA technical papers to be presented in

the Final FRIEND/Nile International Conference. Finally, future research activities were defined

for each of the research theme researchers.

• The fourth workshop took place in Khartoum, Sudan, in the period 23rd July to 30th of July, 2005,.

The research team of the DLFC component representing Egypt, Sudan, Kenya, Tanzania, and

Ethiopia, in addition to the Flemish Counterpart participated in this workshop. The implemented

research activities of the DLFA component in Kenya, Tanzania, Sudan, Ethiopia and Egypt were

presented and discussed. Also, improvement in the preparation of the technical papers to be

presented in the Final FRIEND/Nile International Conference was recognized in all countries. The

participants discussed and reviewed thoroughly all papers. The Flemish counterparts presented

their comments on each technical paper. Meanwhile, intensive working group sessions were

conducted to finalize the papers and to adopt the reviewers’ comments. The papers were totally

finalized and reviewed during the workshops. The workshop participants identified the framework

and time schedule of finalizing rest of the FRIEND/Nile papers. Also, outlines of the annual

progress report for the fourth year of each component were discussed and reviewed.

1.2.1.4) Sediment Transport and Watershed Management (STWM) Component

The following activities were organized for the Sediment Transport and Water Management (STWM)

research component:


• Consultancy mission and Launching Workshop of the Sediment Transport and Watershed

Management: Cairo, Egypt and Khartoum, Sudan in the period 16 – 23 November 2001. The

UNESCO Cairo Office implemented the consultancy mission of the Flemish sediment experts to

Egypt and Sudan. It was arranged for the consultants to meet various experts and institutions

working with Sediment Transport and Watershed Management in the Nile basin and identify

priority research areas. At the end of the consultancy mission, a workshop was held to identify

priority research areas and to prepare a detailed workplan and budget for this component for the

duration of the project in coordination with the project management.

• A meeting for focal persons of STWM was held in Khartoum, Sudan, on 22-24 December, 2002.

The STWM coordinator, the focal persons of the STWM component in the participating countries,

researchers of the UNESCO-Chair in Water resources of Sudan in addition to a number of

experts from Sudan attended this meeting. The main aim of the meeting was to review the

progress in implementing activities of the component during the first year of the project and to

prepare the second year work plan, a list of planned activities and a budget. Moreover, a criterion

for selecting sediment transport software for the research activities of the component has been

identified.

• The first Sediment Transport and Watershed Management (STWM) workshop was organized in

Dar Es Salaam, Tanzania, on 2–6 December 2003. All the STWM theme researchers of the

participating countries in addition to two Flemish counterparts attended this workshop. The

implemented research activities of the STWM component in Kenya, Tanzania, Sudan, Ethiopia

and Egypt have been presented and discussed. Encountered problems and difficulties using

SMS software have been reviewed and discussed. Also, solutions have been provided by the

SMS experienced-participants from Egypt and Sudan. SMS application experiences have been

exchanged through practical SMS training sessions conducted by the Egyptian participants. SMS

application (Hydro-dynamics runs) using data from the Nile countries has been carried out by the

STWM research team. Moreover, SED-2D hands-on training was conducted by the Egyptian

participants using the actual data of the Awash River in Ethiopia. Moreover, it was agreed that the

STWM team researchers would submit technical reports on the SMS application to their case

studies. It was also agreed to launch the Watershed Management component by June 2004. The

same case study areas in Ethiopia, Tanzania and Kenya are to be used. The catchment of the

Blue Nile and some wadis in Egypt will be also used. The acquired data of the RRM component

is proposed to be used. The Flemish counterparts promised to provide the STWM research team

a free catchment erosion model to be used in implementing the STWM research activities.

• The second workshop for Sediment Transport and Watershed Management (STWM) was held in

Alexandria, Egypt, on 19-24 June, 2004. About eighteen key experts participated in this

workshop representing the research team of the STWM component in Egypt, Sudan, Kenya,

Tanzania, and Ethiopia, the Flemish counterparts, resources persons and Egyptian stakeholders.


The main aims of the workshop were to review the progress in implementing research activities of

the component in addition to identify and discuss problems and difficulties facing application of

the SMS model. Hands-on training and experience exchange were carried out to close down

gaps in capacity building of STWM researchers in applying the SMS different modules. Solutions

have been provided by the SMS experienced-participants from Egypt and step by step

application instructions were provided using actual data of the Nile basin. SED-2D Model

applications were run and different levels of results were obtained. Catchment erosion modeling

methodologies and examples were presented by the STWM Flemish counterpart. It was agreed

to select one case study (the Blue Nile basin) to apply some selected methodologies. The

UNESCO Chair in Water Resource will take the responsibility of carrying out this pilot study.

• The third Sediment Transport and Watershed Management (STWM) Workshop, was held in

Nairobi, Kenya, in the period 26 – 29 of November, 2004,. The research team of the STWM

component in Egypt, Sudan, Kenya, Tanzania, and the Flemish counterpart has participated in

this workshop. The implemented research activities of the STWM component in the participating

countries have been presented and discussed. Improvement in the SMS application was

recognized in most countries. Moreover, intensive hands-on training on the SMS application was

conducted to the STWM research team of Kenya. The SMS experience gap in Kenya was

bridged. However, more data is still required to be collected. Flow and Sediment SMS dynamic

applications were implemented for all case study areas in the Nile. Also, catchment erosion

modeling methodologies and examples were presented by the STWM Flemish counterpart. The

USGS raw DEM data of the whole area of the Nile Basin with a resolution of 90m×90m was

distributed to all STWM themes researchers. The use of GIS tools to extract and to prepare

required parameters for catchment erosion models applications were introduced. The workshop

participants identified framework and time schedule of the STWM technical papers to be

presented in the Final FRIEND/Nile International Conference.

• The fourth STWM workshop took place in Khartoum, Sudan, on 25-30 July, 2005. Key experts

participated in this workshop representing the theme researchers of the STWM component and

the Flemish counterpart. The implemented research activities of the STWM component in Kenya,

Tanzania, Sudan, Ethiopia and Egypt were presented and discussed. The participants discussed

and reviewed thoroughly all technical papers to be presented in the Final FRIEND/Nile

International Conference. The Flemish counterparts presented their comments on each technical

paper. Moreover, intensive working group sessions were conducted to finalize the papers and to

adopt the reviewers’ comments. The papers were totally finalized and reviewed during the

workshop. The workshop participants identified framework and time schedule of finalizing rest of

the FRIEND/Nile papers. Also, outlines of the annual progress report for the fourth year were

discussed and reviewed.


1.2.2) Training Workshops and Technical Missions

The following training workshops were organized:

• Training Workshop on “Data Acquisition, Data Processing and Data Analysis”, Dar Es Salaam,

Tanzania; 19-26 May 2002: Based on the recommendations of the First Project Management

Meeting, the UNESCO Cairo Office and Dar Es Salaam University organized this training

workshop. All coordinators and focal persons of the project in addition to the Flemish counterparts

attended this workshop. Total number of participants was 30 persons. The aim of this training

activity was to strengthen capacity building of the researchers of the project in the field of data

handling, processing and analysis. Various data processing techniques and data requirements

for each research theme were discussed and presented by resource persons. During the

workshop, available data in each participating country in the project for each research component

was defined. outcomes of each component were specifically defined according to the overall

objectives of each component. Additionally, common pilot areas/catchments to be used in the

research activities of the four components were defined. Also during this activity, the data

acquisition issue has been raised by the coordinators of all research components as an essential

procedure to collect data. Moreover, some future activities during the first year of the project were

proposed for the different components.

• Statistical Hydrology Training Course for WRRI Local Staff, Cairo, Egypt; 14-19 December 2002:

A one-week training course was organized for local staff of the Water Resources Research

Institute (WRRI) as the coordination Center of the Flood Frequency Analysis component. This

training course addressed issues of statistical hydrology and flood frequency analysis with special

application on the Nile basin. About 15 engineers attended this training course.

Also, the following missions were organized:

• Technical mission of Dr. Patrick Willems to the Water Resources Research Institute (WRRI), 25-

30 October 2003; Cairo, Egypt: Dr. Patrick Willems, the Flemish counterpart of the Flood

Frequency Analysis Component (FFAC), conducted a consultancy mission to WRRI as the

coordination Center of the FFA component during the period 25-30 October 2003. He reviewed

the acquired data through the FFA theme researchers as well as the analysis and research

activities carried out by the FFA research team. Also, he provided technical advice to WRRI staff

to enhance the implementation of the FFA component research activities.

• Technical assistance of Dr. Khaled Hussein to the Water Resources Research Institute (WRRI),

June-August; Cairo, Egypt: Dr. Khalid Hussein (Faculty of Engineering - Cairo University)

provided a technical support and consultation assistance in the implementation of research

activities of the FFA component during June – August 2003. He also provided high-level training

and theoretical background support to the WRRI team. Moreover, Dr. Hussein provided WRRI a

high-level flood frequency analysis computer program, which was developed by him and was


curtailed, updated, and further developed to address needs of the FRIEND/Nile project. A

number of training sessions were conducted where the WRRI team was trained to perform

analysis using the provided computer program.

1.2.3) Project Governance

The project is managed through two levels:

1. The steering committee is the governing body of the project comprising representatives of

the participating countries as well as the thematic coordinators of the project, in addition to

UNESCO and donors,

2. The project management team consists of donors, UNESCO, and the overall coordinator of

the project.

1.2.3.1) Steering Committee Meetings

The steering committee has been formed in close consultation with the Nile Basin countries. Its

meetings were supported UNISCO since the launching of the FRIEND/NILE projects. Several

meetings of the steering committee took place in several countries of the basin. The main objectives

of these meetings are to supervise the implementation of the project activities, review and approve

the project overall policy and future actions and to evaluate the outcome of the Flemish FRIEND Nile

activities. The Steering Committee Meetings were organized annually as follows:

• Cairo, Egypt, 1997;

• Dar Es Salaam, Tanzania, 1998;

• Khartoum, Sudan, 1999;

• Cairo, Egypt, 2000;

• Cairo, Egypt, December 2001;

• Aswan, Egypt, January 2003;

• Mombassa, Kenya, February 2004;

• Addis Ababa, Ethiopia, February 2005; and

• Sharm El Sheikh, Egypt, November 2005.

In the following we briefly summarize activities of some of these meetings.

• The Fifth Steering Committee Meeting, Cairo, Egypt 13 December, 2001: The Fifth Steering

Committee Meeting of the project was held back to back with the First Project Management

Meeting. More than fourteen key experts representing funding agency, UNESCO, Nile basin

countries representatives, coordinators of the components and the overall coordinator of the

project attended this meeting. An overall policy for the project was determined in this


meeting. Based on the consultation and discussion, some conclusions are reached as: Link

of FRIEND/Nile Project to international and regional organizations such as Nile Basin

Initiative (NBI) is highly recommended to avoid duplication of research studies and to

optimize the limited available resources; Participation of other Nile Basin countries (Burundi,

Rwanda, and Congo) in the project is encouraged; Tailored training courses are of special

concern; Integration of FRIEND / Nile Basin project with other FRIEND projects all over the

world is of great importance.

• The Sixth Steering Committee Meeting, Aswan, Egypt; 8-10 January, 2003: Twenty three

participants attended this meeting representing funding agency, UNESCO, Nile basin

countries representatives, research themes coordinators, and other Nile networks. The

participants encouraged the enhancement of the cooperation among the implementing

institutes of the project and the timely implementation of the project activities. The role of the

themes researchers in the implementation of the project research activities was evaluated

and reviewed. Means of encouraging and motivating the theme researchers in the

implementation of the project research activities were discussed and reviewed.

• The seventh meeting, Kenya, Mombassa 12-13 February, 2004: The Seventh Steering

Committee Meeting of the project was held back to back with the Third Project Management

Meeting. Twenty one participants attended these meetings representing funding agency,

UNESCO, Nile basin countries representatives, research themes coordinators, and other

Nile networks. The implementation of the project activities according to the approved work

plans of the research themes was reviewed. Based on the discussion, some conclusions

were reached as: Strengthening linkage between NBI and FRIEND/Nile for the mutual

benefit of the Nile basin countries; Enhancing further collaboration through networking

among the research teams; National focal persons should disseminate the project outputs

and reports to all interested stakeholders in their respective countries.

• The Eighth Steering Committee Meeting Addis Ababa, Ethiopia, 23 – 24 February, 2005:

This meeting was attended by UNESCO, funding agency, Overall Coordinator of the project,

and the research themes coordinators. The main objectives of this meeting were to discuss

and evaluate the progress in the implementation of the different activities; to review and

approve the project document of the proposed Second Phase of the FRIEND/Nile document;

to focus on the proposed second phase overall policy, themes, modalities and framework; to

discuss the possible ways to ensure the sustainability of the project and to increase the

cooperation level among the implementing institutions; and to discuss executive procedures

for linking the project to other ongoing Nile basin projects and initiatives. It was agreed to

approach the Ministerial council and the NBI secretariat to link the activities of the

FRIEND/Nile Phase two activities to the NBI.


• The Ninth Steering Committee Meeting, Sharm El Sheikh, Egypt, 15 November, 2005: The

committee discussed the overall future policy of the project through the UNESCO-Flanders

cooperation. Furthermore, the committee evaluated the seriousness and commitment of the

participating countries in the implementation of the FRIEND/Nile activities of the first phase.

The involvement of other Nile countries in the project was thoroughly discussed and

reviewed. The participants highly encouraged the idea of consolidating the FRIEND/Nile

papers and publishing them in international journals. The funding possibilities of the second

phase of the FRIEND/Nile project were also discussed. The participants stressed on the

importance of securing funds of the second phase to ensure the sustainability of the project

activities.

1.2.3.2) Project Management Meetings

Several project meetings have taken place since the official start of the project. The main objectives of

these meetings were to review the implementation of the project activities and to approve the project

workplan and budget of each year based on the FUST project agreement and UNESCO regulations.

UNESCO and the Flemish donors' representatives and the overall coordinator attended these

meetings. They are listed below:

• First Project Management Meeting, Cairo, Egypt; 10 -12 December, 2001: The outcomes of the

launching workshops were reviewed and evaluated. Initially, it was decided that the project

would focus on three research components, namely: Rainfall - Runoff Modeling, Sediment

Transport and Watershed Management, and Flood Frequency Analysis. It was emphasized that

training has to be integrated to support the research theme activities, and training needs are

identified by the research teams. A joint training for various themes especially, in data acquisition,

data processing and analysis was recommended. Also, the meeting participants highly

encouraged the selection of common pilot areas and catchments to be used for the three

research themes, whenever possible.

• Second Project Management Meeting, Aswan, Egypt; 6-7 January 2003: During this meeting, it

was emphasized that any supported activity or equipment purchase should be directly linked to

well-defined results and deliverables from the various themes researchers’ side. It was agreed to

design in a high quality standard format a unified cover page for the reports of the implemented

activities of the three research themes.

• Third Project Management Meeting, Mombassa, Kenya; 10-11 February 2004: The Project

Management Team (PMT) approved funding the "Drought and Low flow Analysis" theme. It was

recommended to link this component’s activities to the Flood Frequency Analysis component

activities since the two components have the same Flemish counterpart. It was also decided to

integrate training within the research theme activities using the "Training of Trainers" modality.

Training needs should be identified by the research coordinators and involving qualified people

who will be working with a specific research activity in such training. Summary reports of


implemented training and technical workshops should be prepared and distributed to all Nile

basin countries and posted on the project website. The PMT approved the unified cover format

for the project reports. The next PMM will be held back to back with the 8th Steering Committee

meeting during the second half of January 2005.

• Fourth Project Management Meeting, Addis Ababa, Ethiopia,; 21-22 February 2005: The project

document of the proposed second phase of the FRIEND/Nile FIT project was reviewed and

approved. The PMT recommended submitting the document to the Flemish donors via UNESCO

HQ after the final approval of the 8th FRIEND/Nile Steering committee. Arrangements of

organizing the final international conference of the FRIEND/Nile in Sharm El Shiekh, Egypt; 12-15

November 2005 were discussed. The PMM stressed on importance of preparing quality technical

papers as one of important outputs of the project. The PMM agreed to secure the necessary

funds for the organization of this conference..

• Fifth Project Management Meeting, Sharm EL-Sheikh, Egypt; 14 November 2005: The

participants evaluated the achievements and outcomes of the project during the period 2001-

2006. The PMM team appreciated the successful implementation of all project activities. The

meeting participants acknowledged the successful preparation of 27 joint technical papers

reflecting the remarkable achieved technical regional cooperation and capacity building of the

research team of the project in the course of the FRIEND/Nile project. The PMT highly appraised

the technical efforts and contribution of the Flemish counterparts in the project research activities

and the technical papers preparation. The PMT approved proposed activities of the project for the

first few months of year 2006 till the funding of the second phase is operational., namely:

production of the final report of the project and organization of a brainstorming meeting to explore

the application of proposals to the European Union (EU) within the International Research Co-

operation 7th Research Framework Program (FP7)

1.2.4) Other Related Meetings

The following overall coordination meetings were organized:

• The Overall Coordinator Participation in the 4th FRIEND Inter-Group Coordination Committee

(FIGCC) Meeting, Cape Town, South Africa; 17-23 March 2002: The overall coordinator of the

project presented to the FIGCC recent development of the FRIEND/Nile project and the planned

activities of the project. He suggested an integration and collaboration framework to cooperate

with other global FRIEND projects all over the world.

• The Overall Coordinator and the Project Manager Participation in the 5th FRIEND Inter-Group

Co-ordination Committee (FIGCC) Meeting, Koblenz, Germany; 4-10 July 2004: The overall

coordinator and the project manager participated in the 5th FRIEND Inter-Group Coordinating

Committee (FIGCC) Meeting and the joint FRIEND/North Europe (NE) and FRIEND/Alpine and

Mediterranean (AMHY) Extremes Workshop to present the achieved progress in the


implementation of the FRIEND/Nile Project and to coordinate the FRIEND/Nile planned activities

with the Global FRIEND activities.

• The Overall Coordinator and the Project Manager Mission to Uganda for the establishment of FN

Research team, Kampala, Uganda; 15-19 June 2005: The overall coordinator and the project

manager undertook a mission to Uganda to identify the executive procedures to involve Uganda

in the FRIEND/Nile future research activities and to assist in the formation of a research team for

the project in Uganda. The FRIEND/Nile project objectives, research themes, framework, on

going research activities and the structure of the proposed second phase of the project were

presented to Her Excellency, the Minster of Water, Lands and Environment of Uganda and key

water governmental officials and professors in Mekerere University. Based on discussions and

deliberations with the national focal person of Uganda, a group of water experts from Uganda

from Mekerere University and Minster of Water, Lands and Environment was proposed to

participate in the FRIEND/Nile second phase in Uganda.

1.2.5) Second Phase Project Document Preparatory Meetings

To prepare necessary documents for the project second phase, several meetings were organized as

follows:

• FRIEND/Nile Second Phase Project Document Preparatory Meeting, Alexandria, Egypt; 18-20

October 2004: Seven participants attended this meeting representing the Flemish counterparts,

overall coordinator, themes coordinators and UNESCO representatives. It was stressed on the

importance of involving all interested Nile countries, especially Uganda, in the second phase

activities. The outline, framework, deliverables, overall work plan and budget of the second phase

were discussed and defined. The project document of the proposed second phase of the

FRIEND/Nile project was thoroughly discussed and reviewed. The following themes in the

second phase are:

1. Integrated Water Resources Management component which will be coordinated

by the Water Resources Research Institute in Egypt, (the Overall Coordination

Center). Main objectives of this component include development of

management scenario’s that will be investigated by other “technical” themes,

evaluation of results of the technical themes, to ascertain the coordination

between the different technical themes, and to facilitate access to data.

Members of this component will be senior experts in management issues and

coordinators of the themes.

2. Hydrologic Modeling component coordinated by the University of Dar ES

Salaam in Tanzania: Main objectives of this component are to consolidate

achievements of the first phase with special focus to develop rainfall-runoff

models for the available gauged catchments within the Nile Basin in view of the


analysis of integrated water resources management guidelines, to investigate

impacts of land use change or climatic change on the river flow, and to study

surface/groundwater interactions, if necessary, in view of management

problems.

3. Erosion and sediment transport modeling coordinated by the UNESCO-Chair in

Water Resources in Sudan: Main objectives of this component include

consolidating its achievements with extensions to catchment modeling of

erosion problems with special focus on understanding catchment erosion and

sedimentation processes within the Nile in view of the analysis of integrated

water resources management guidelines/scenarios, developing guidelines for

erosion problems and watershed management in the Nile Basin, enhanceing

regional research capacity on topics related to erosion modeling, and bringing

together professionals from the Nile Basin to exchange experience, ideas and to

foster common understanding and cooperation.

4. Stochastic Modeling Component coordinated by the University of Nairobi in

Kenya: Main objective of this component is to merge the Flood Frequency

Analysis theme (FFAC) and the Drought and Low Flow Analysis theme (DLFA)

of the first phase into this new component. Based on the tools and expertise

obtained during the 1st phase and within the view of integrated water resources

management guidelines or scenario’s, this component will focus on developing

regional design procedures for estimating flood magnitudes for a given

probability of exceedence at gauged and ungauged sites in the Nile basin, and

analyzing daily river flow data for estimation of low flow magnitudes – duration –

frequency relationships as well as drought analysis.

5. Eco-Hydrology Component coordinated by the University of Makere in Uganda:

Main objective of this component is to enhance the understanding of

ecohydrological processes/functions within the Nile River Basin and their

application in IWRM with special focus on establishing baseline information on

eco-hydrology issues and identifying the gaps on eco-hydrological issues in the

Nile basin in addition to applying eco-hydrological models as management tools

in IWRM.

1.3) Research Activities

The main target of research activities of all project research components (each component according

to its specific discipline) was to obtain output and results using such tools that can be applied to

improve design procedures of necessary future development projects in the Nile basin countries. The

following points summarize the research activities of all research components during the reported

period:


• Data processing and analysis.

• Selection and introduction of suitable models.

• Application of the selected models.

• Reporting (Semi-Annual and Annual progress reports).

• The way ahead.

Details of these activities are as explained in the following sections.

1.3.1) Data Processing and Analysis

Required data for the implementation of research activities of the four components were collected and

acquired through the various theme researchers in Egypt, Sudan, Kenya, Ethiopia and Tanzania.

Each country theme researcher worked with his data according to its availability in each country with

interested techniques and methodologies in that country. The data acquired for all research themes

become now almost available and well defined and known. Some case studies, based on the

availability of data, were selected in each country. Details of the acquired data and the selected case

studies in each country for each component were presented in the data processing report. Exchange

of research results has been achieved based on mutual trust and confidence developed among the

research teams in the course of this project.

1.3.2) Selection and Introduction of Suitable Models

Selection of a unified suitable model for each research theme was a very important task for all of the

project teams. After many investigations and discussions from all project teams and experts

(Coordinators, Focal researchers, Flemish Counterparts, and Resource persons), the following

models were recommended to be used:

• The Watershed Modeling System (WMS) for implementation of research activities of the RRM component;

• Galway Flood Forecasting System (GFFS), SWAT and HMS were introduced for implementation of research activities of the RRM component;

• The Surface Water Modeling System (SMS) for implementation of research activities of the STWM component;

• The Extreme Value Analysis (ECQ) Model for implementation of research activities of the two components of the FFAC and DLFAC.

These models were introduced with their training manuals for all theme researchers of the four

components through the different training workshops. Advanced training and application of these

models have been carried out using real data of the selected case study areas in the Nile Basin

during the training workshops of the four years of the project. It was also agreed to use any additional

suitable models being available to the thematic coordinators and researchers.


1.3.3) Application of the Selected Models

Training on application of the selected models WMS, SMS, GFFS, and ECQ was the main objective

of the training workshops held in Dar-Es-Salaam (19-26 May 2002) and Alexandria (20-25 July 2003).

In STWMC workshops, application of SMS using real data of the selected case studies has been

successfully done. SMS model is suitable for channel sedimentation only. It gives variation in

sediment load and bed variation. Reasonable results have been obtained. In RRMC workshops,

applications of WMS, HMS, SWAT and GFFS using real data of the selected case studies have been

made. WMS is suitable for a single storm, while GFFS is suitable for a continuous series of rainfall

storms. Both models require rainfall as an input and give the hydrograph and its characteristics as an

output. Results have been obtained and analyzed for selected case studies. The resource persons

guided the training and solved most of application problems. In FFAC and DLFAC workshops,

application of ECQ using real data of the selected case studies in the different countries has been

successfully carried out. A brief summary on the application of the selected models for the four

components is presented in the following sections. More technical details on the implemented

research activities and outputs of the research components of the project are provided in this

document.

1.3.3.1) Flood Frequency Analysis Component

A harmonized methodology of the RFFA has been conducted at the pre-defined areas by the

coordination center (WRRI) and theme researchers of Sudan, Tanzania, Kenya and Ethiopia. Based

on analysis of Q-Q plots, a normal tail is found for most of the rivers. Therefore, the selected statistical

distributions were evaluated as an exponential Q-Q plot for EV1/Gumbel using MOM, Ml, and PWM,

while the others were evaluated as a Pareto plot. The whole range of observations does not follow

one FF distribution at some sites. This might be due to influence of flooding along the river. Therefore,

it was suggested to calibrate a separate distribution for two sub-populations, one for the non-flood

part, and the other for the flood part. A comparison of the calibration results has been carried out for

the distribution parameters (for EV1/Gumbel and for GEV, and according to different parameter

estimation methods ML; MOM; and PWM).

A regionally calibrated relation between the Mean Annual Flood (MAF), and the catchment

characteristics (area, average slope and average annual rainfall) was established using multiple

linear regressions. This relation, together with the developed regional frequency curves, could be

used to estimate flood magnitudes with various return periods for un-gauged catchments at any

homogeneous region with the regions under consideration. Although most countries did not achieve

the final results and they recommended a follow up training on GIS tool. The latter can be used to

extract catchment physiographic characteristics such as land cover, slope and elevation which can

be used to improve the regionalization analysis.


1.3.3.2) Rainfall /Runoff Modeling Component

A single storm event approach was applied, using WMS and HMS, to three selected basins in Egypt.

These basins are Wadi al-Arbain, Wadi Gudierat and Wadi Sudr in Sinai. Many parameter estimation

techniques and objective functions were tried for hydrograph calculations. Modeling the river flows of

the Awash catchment in the Ethiopian Plateau using naturalized and regulated river flows was also

performed. The HEC-HMS model and GFFS software were applied for the Nzoia River and the other

selected basins in the North-Eastern side of Lake Victoria. Also, the SWAT model was applied and

calibrated in Simiyu catchment in the South-Eastern side of Lake Victoria, where good results were

obtained. Data of the Blue Nile and the Eddeim catchments were prepared and used in HMS, WMS,

and GFFS models where good results were obtained and presented. Table 1-1 shows catchment

characteristics of the studied catchments while Table 1-2 gives results for the best model efficiency

criteria in the participating countries in the FRIEND/Nile Project.

• Table 1-1, Catchment characteristics of the researched catchments

Country Region Topography Land-Use Climate

Egypt Sinai Mountainous Not defined Semi desert

Ethiopia Awash Hilly Not defined Wet

Kenya Nzoia, Nyando Hilly & Mild Grass/Woodland, Cultivated Wet / Dry

Sudan Eddeim Not defined Not defined Wet / Dry

Tanzania Simiyu Mild Grass/Woodland, Cultivated Wet / Dry

• Table 1-2, Best model efficiency criteria results.

Country Simulation

model

MOCT Parametric /

Updating mode

Remarks / Area

in km2

Egypt WMS - (HEC1) - (HEC1) Good fits

Ethiopia 72% (ANN) 80% (SAM,

WAM,NNM)

80% (SLM) 7,656

Kenya 67% (LPM,SMAR) - 97% (SLM,LPM) 3,450 / 12,676

Sudan 91% (LPM, LVGFM) 92% (WAM,NNM) 97% (LPM,LVGFM) 254,230

Tanzania 50% (LVGFM) 66% (NNM) - 5,320

1.3.3.3) Drought and Low Flow Analysis Component

The POT method to analyze all available data on river discharges or surface water levels using an

appropriate time-scale relevant for each country was used. A drought index for drought analysis was

developed and results were regionalized. The Q-Q plot of the ECQ software was applied to data from

all participating countries. Ten-day average flow discharges covering 24-40 years of at least 3 stations


in the catchment of River Sobat, in addition to 130 years of monthly flow discharge at Aswan on the

main steam of the River Nile, were used in the analysis. Annual rainfall data covering 28-30 years of

about 3 stations in the catchment of river Sobat were also used. Also, daily flow discharge with

monthly rainfall and evaporation in the Blue Nile and River Atbara catchments were used in the

analysis. Moreover, daily average flow discharge covering about 30 years of at least 5 stations in the

catchments of North-Eastern side of Lake Victoria, and daily rainfall data covering about 30 years of

about 150 stations in the catchment of Lake Victoria were used in the analysis. Finally, daily flow

discharge and daily rainfall data in the catchments of South-Eastern side of Lake Victoria located in

Tanzania were used in this analysis. Results of all case studies were obtained, discussed and

presented.

1.3.3.4) Sediment Transport and Watershed Management Component

SMS simulation of the long reach of Awash River does not result good outputs since there is a

limitation of RMA2 in steep slope condition. Hence, the hydrodynamic modeling for a segment of the

reach was done to improve the velocity magnitude, water depth and water surface elevation. In case

of Sondu basin, the river channel is long with a lot of meanderings and waterfalls, hence, it did give a

lot of complications during the calibration of the SMS. After some serious consultation, it was realized

that the SMS works better in river channels that do not have steep waterfalls. Therefore, Sondu basin

had to be divided in portions of about 5km stretches for ease of handling and better accuracy.

The two modules of the SMS model, namely, hydrodynamic (RMA2) and sediment module (SED2D)

have been applied to the Simiyu catchment. Obtained results were reasonable in terms of accuracy.

However, another independent station had to be used to check such accuracy. To run SMS-RMA2, it

was required that downstream water level should be above the highest bathymetry. This is only

applicable in mild flood plains and thus it is not applicable to hilly areas such as the Simiayu River.

SMS has also been applied to a selected study reach of the Blue Nile where problems facing

application of SMS were addressed. Sediment, flow profiles river morphology and expected changes

of sediment concentration required to use SMS model, have been used. Capacity building for the

technical staffs to use SMS model was also emphasized where the package was mastered.

The SMS was also applied to TOSHKA and AHDR (Aswan High Dam Reservoir) to study effects of

sediment on current and future backwater curve and to study the stability of sediment deposits.

Available data for input to SMS include banks and bed elevations in X,Y,Z coordinate, water surface

elevations, flow rate (discharge), velocities, channel and floodplain characteristic, wind velocity, water

temperature, and latitude. For future work, it was recommended to use SMS for small regions of 20

km length in Aswan High Dam Lake to estimate bed profile of these regions, to estimate water

surface profile and velocity distribution in 2-D, and to predict sediment deposition.

1.4) Reporting

Annual and Semi-Annual progress reports, one for each of the four components of the project, have

been prepared by the coordinators to present the implemented activities in each component during

every year of the project's period. All these reports are available on the website of the project.


1.5) FRIEND/Nile International Conference

An international conference entitled “International Conference of FRIEND/Nile, Towards a Better

Cooperation” was organized to present the obtained research outputs and the achievements of the

project. It was held in Sharm Al-Sheikh, Egypt, during 12-15 November 2005. The conference

covered the following topics:

• Hydrology of the Nile;

• Rainfall-Runoff Analysis;

• Extreme Events;

• Sediment Transport and Watershed Management; and

• Water Resources Management.

A Website was developed for the conference (http://www.friendnile.org). A first call for papers to

contribute to this conference was disseminated to water related agencies and institutions in Tanzania,

Kenya, Ethiopia, Sudan, and Egypt.

The research teams of the project identified the framework and time schedule of more than 27

technical papers which were presented in the FRIEND/Nile International Conference. The Scientific

Committee accepted 70 papers of which 27 papers belong to the implemented activities of the

project. The conference included 5 technical sessions, and poster session in addition to 4 keynote

lectures prepared by top-notch international experts. About 120 participants from than 12 countries,

comprising international and regional key-water experts, policy makers from the Nile countries,

FRIEND/Nile researchers, Flemish counterparts, stakeholders and representatives of the ongoing

Nile initiatives, attended and contributed to the deliberations of the conference.

1.6) The Way Ahead

The following are recommendations to be considered in preparation and implementation of the project

second phase:

• Communication among the thematic coordinators and the Flemish counterparts should be more enhanced.

• More attention should be given to contents and delivery time of the project progress reports.

• Continuous commitment of the research teams should be ensured to avoid change in the research team structure until end of the project to prevent any delay of work plans;

• Intensive and high level training on GIS applications may be necessary for focal persons of all components.

• A 2nd stage for the project is strongly recommended to complete and consolidate project results and outputs.


Chapter

2

Flood Frequency Analysis

2.1) Introduction

This chapter focuses on the implemented research activities of the Flood Frequency Analysis

Component (FFA) during the project period (2001–2006). Moreover, constraints and problems faced

during the implementation of the research activities are discussed. Final component results have also

been presented and discussed. Finally, few relevant research points are suggested for the second

project phase.

The Flood Frequency Analysis component coordinated by the Water Resources Research Institute

(WRRI) is one of the four research components of the FRIEND/Nile UNESCO-Flanders Science

Fund In Trust Project. Major objectives of the FFA component can be summarized as follows:

• Obtaining relationships between flood peaks and their corresponding return periods on both single and regional scales;

• Developing design procedures for flood estimation at gauged and ungauged catchments on regional basis;

• Producing regional flood frequency curves, and defining different hydrological regions of the basin.

This will help improving designs of hydraulic structures along or across streams, and planning flood

plain adjacent to a stream. It will also help designing storage works for irrigation, water supply, and

flood control. The analysis was coordinated by the WRRI of the National Water Research Center of

Egypt and assisted by the research focal persons from the Nile Basin countries and the resource

persons from Belgium. This report presents a summary of the analysis and discussion of results for

the selected areas of study.

Output of these research activities (i.e., Flood Frequency Curves for single sites and regional Flood

Frequency Curves for a homogenous region) has been obtained using two different packages. The

first package is a MATLAB code for Flood Frequency Analysis developed by Dr. Khalied Hussein,

Cairo University, Egypt. The second package is the one developed by Institute of Hydrology, UK.

These two packages are implementing similar flood frequency analysis techniques; both have at-site

and regional flood frequency analysis capabilities, probability plots, distribution comparison plots, site


comparison plots. The first model has more capabilities by giving regional homogeneity statistics,

goodness of fit tests, and moment ratio diagrams. The two models are capable of fitting different

statistical distributions which are commonly used in flood frequency analysis, using three different

parameter estimation methods, namely, the method of moments (MOM), the maximum likelihood

method (MLM), and the method of probability weighted moments (PWM). Another approach that is

not implemented in neither of the two packages is also used in this analysis. This approach is the

Extreme value analysis method, (Q–Q) Plot, developed by the Flemish Counterpart, Beirlant et al.

(1996) and Willems (1998). They have dealt with this problem by analyzing the tail of the distribution

through extreme value analysis and the quantile-quantile (Q-Q) plot. According to this analysis, the

shape of the tail is classified on the basis of the value of the extreme value index γ.

The data acquisition for the FFA has been carried out by each theme researcher of the participating

countries (i.e., Egypt; Sudan; Tanzania; Ethiopia and Kenya). This report gives also a brief description

and analysis of these data for the selected study areas in each country.

2.2) Research Activities

This section gives a summary of the analysis of the obtained data from the contributing countries. It

also highlights the main design procedures for determining the best-fit for flood frequency

distributions. Finally, it presents a procedure for fitting a Q-Q plot to the given data of the annual

maxima. Here, it should be mentioned that harmonized methodologies and procedures of at-site and

Regional Flood Frequency Analysis (RFFA) have been applied by all theme researchers of the

participating countries.

Analysis in this chapter consists of the following main points:

• Data Processing,

• Models construction for:

o Production of Flood Frequency Curves (FFCs);

o Production of Regional Flood Frequency Curves (RFFCs);

• Using GIS techniques for estimating the geometric characteristics;

• Applications of models developed to ungauged basins. FRIEND/Nile

2.2.1) Data Processing

The research activities listed above have been applied to the available daily flow data, rainfall, and

geometric characteristics obtained from the theme researchers of the participating countries (i.e.

Egypt; Sudan; Tanzania; Ethiopia and Kenya). According to the recommendations of the FFA

component workshop of (1st to 3rd of April, 2003 - Cairo, Egypt), every theme researcher was

requested to work with the available data in his country, and was asked to follow the pre-defined


harmonized approaches of the Flood Frequency Analysis (FFA). A copy of these data was submitted

to the Water Resources Research Institute (WRRI) – Egypt, the coordinator of the component, and a

contract of the data acquisition was signed by each theme researcher of the participating countries.

These data comprise the maximum daily annual flow data and their record lengths for years, and the

available total annual rainfall records. The data submitted by Egypt were for River Sobat and its two

main branches (River Baro and Pibor) which have two tributaries, namely, River Gila and Akkobo.

The data from Tanzania were given as a list of stream gauging stations for annual maximum flow

data, the flow record lengths, areas of the different catchments, and a list of rainfall stations located

within the Simiyu, Duma, Mara, Magogo and Kagera river basins. The data from Kenya were given as

a list of stream gauging stations for maximum daily flow data, the flow record lengths, areas of the

different catchments, and a list of rainfall stations. The data from Sudan were given as locations of the

measuring stations for the daily flow series, the years of data records, and regional data from

obtained from LANDSAT images for the Blue Nile at EL-Deim, Sennar and EL-Khartoum, river

Rahad, river Dinder; river Atbara, and the main river Nile at Malakal. Finally, the flow and catchment

characteristics data from Ethiopia was given for the Awash area which is not a part from the Nile

basin. Table 2-1 gives a summary of these data obtained from the participating countries.

• Table 2-1, Data; Rivers, Stations, Locations and Flow record length of the FFAC.

River Station Details location Record length

Blue Nile Ed Deim At border with Ethiopia 1964-1996

Blue Nile Sennar D/S Sennar Dam 1968-2001

Blue Nile Khartoum Before confluence with White Nile 1965-2002

Dinder River Hawata Before confluence with Blue Nile 1972-1998

Rahad River Gwasi Before confluence with Blue Nile 1972-1998

Atbara River Kubur On branch Setit 1972-1998

Atbara River Wad El Hileiw At upper Atbara 1972-1998

Awash 031012 At Melkakunture 1965-1999

Awash 031013 At Hombole 1968-2000

Berga 031001 Near Addis Alem 1975-2000

Holeta AW1002 Near Holeta 1975-2000

Awash Bello AW1020 At Bello 1986- 2000

Teji AW1003 Near Asgori 1975 -2000

Akaki 031004 At Akaki 1981 -1999

Sondu 1JG01 At Modjo 1969 -2000

Nyando 1GD04 upstream 1956-1995

Yala 1FG01 near outlet 1947-1993

Nzoia 1EE01 upstream 1963-1994

Migori 1KC03 Near outlet 1951-1985

Awach 1HA14 near outlet 1961-1988


River Station Details location Record length

Sondu 1GD04 - 1947-1991

Nyando 1GD04 - 1963-1989

Yala 1FG01 - 1947-1993

Nzoia 1EE01 - 1953-1994

Migori 1KC03 - 1951-1985

Mbogo 1GBO6 - 1951-1985

Ngono Muhutwe - 1971-1982

Ngono Balebe brg - 1970-1982

Ngono Kyaka rd brg - 1970-1982

Ruvuma Mwendo ferry - 1970-1982

Kagera Nyakanyasi - 1970-1978

Moame Mabuki brg - 1970-1982

Magogo Shinyanga rd - 1970-1982

Duma Sayaga - 1970-1982

Simiyu Road crossing - 1970-1978

Simiyu Ndagalu - 1970-1982

Grumet Mara rd crossing - 1970-1982

Magogo Mwanza brg - 1974-1982

2.2.2) Models Used

In this study, three models have been used to carry out the technical research activities. The first was

the FLOODS package developed by Institute of Hydrology, UK. The second model was the Flood

Frequency Analysis software developed by Cairo University, Egypt. The third model was the Extreme

value analysis software using the (Q–Q) Plot which is developed by the Flemish Counterpart. A brief

description of these models is given in sections to follow.


• Figure 2-1, Three main regions of Lake Victoria.

2.2.2.1) Floods Package (Model 1)

The FLOODS package has been developed by the Institute of Hydrology (IH) in UK. This package

can undertake a frequency analysis to determine the relationship between the size of an event and

the probability of that magnitude event being exceeded in the future. Therefore, a large number of

probability distributions that has been recommended for frequency analysis are included in the

FLOODS package. This package estimates the parameters of the various distributions by the method

of Probability-Weighted Moments (PWM) and method of Moments (MOM). Unfortunately, it can not

perform a test of goodness for selecting the best fit distributions. This model was used at the early

stage of the FFA component. It was applied to the annual maximum flow series from fifteen river

basins draining into Lake Victoria at the upper region of the Nile Basin. The size of the basins ranges

from 590 km2 to 11900 km2. Rainfall over these basins ranges from 1500 mm at N-E of the Lake to

800 mm at the S-E and West of the Lake. Figure 2-1 shows the main three regions of Lake Victoria.

As a result of this study, it was found that Extreme Value Type1 (EV1 or Gamble) Distribution and the

Generalized Extreme Value Distribution (GEV) are the most appropriate distribution for fitting the peak

discharge series at the different examined sites. Based on this analysis, tt is recommended using the

method of probability-weighted moments for estimating parameters of the fitted distributions. It can be

also concluded that the region of these rivers can be divided into three homogenous regions

according to the flood frequency curves (FFCs) of the gauged basins. Figure 2-2 and Figure 2-3 show

the FFCs and the corresponding regional flood frequency curves from the three regions by combining

all the dimensionless curves within each region using probability weighted moments (WM) method.


Figure 2-4 shows a comparison between the regional flood frequency curves of the main three main

regions of Lake Victoria.

Comparison of flood frequency curves, East L. Victoria Region

Sta

ndar

dise

d flo

od p

eak

Reduced variate-2 -1 0 1 2 3 4 5

0.00

0.35

0.70

1.05

1.40

1.75

2.10

2.45

2.80

3.15

3.50 1

2

3

4

5

6

Return period (years)

2 5 10 20 50 100

Fitting: GEV-PWM1 -- GUCHA2 -- MARA3 -- NYANDO4 -- NZOIA5 -- SONDU6 -- YALA

Regional flood frequency curves

Sta

ndar

dise

d flo

od p

eak


0.00

0.35

0.70

1.05

1.40

1.75

2.10

2.45

2.80

3.15

3.50

1


2 5 10 20 50 100

Fitting: GEV-PWM1 -- EAST VECTORIA LAKE

• Figure 2-2, Flood frequency curves for the rivers of region 1, and the corresponding regional flood frequency curve.

Comparison of flood frequency curves, South-East L. Victoria

Sta

ndar

dise

d flo

od p

eak


0.00

0.35

0.70

1.05

1.40

1.75

2.10

2.45

2.80

3.15

3.50

1

23

4

56

7


2 5 10 20 50 100

Fitting: GEV-PWM1 -- DUMA-22 -- MAG0G0-23 -- MARA-24 -- MBALAGITI-25 -- MORI-26 -- RAWANA-27 -- SIMIYU-2

Regional flood frequency curves, South-East L. Victoria

Sta

ndar

dise

d flo

od p

eak


0.00

0.35

0.70

1.05

1.40

1.75

2.10

2.45

2.80

3.15

3.50 1


2 5 10 20 50 100

Fitting: GEV-PWM1 -- SOUTH EAST LAKE VECTOR

• Figure 2-3, Flood frequency curves for the rivers of region 2, and the corresponding regional flood frequency curve.

2.2.2.1.1) Applications of Model (2) on Selected Sites in the Nile Basin

The model was applied at different sites in the Nile basin as given in Table 2-1. These sites are

namely the Blue Nile and River Atbara, River Sobat and its tributaries, the upper parts of some

selected rivers in the Sudd area, some selected rivers at the North-East and South-East sides of Lake

Victoria, and the Awash River in Ethiopia.

The data series of the selected sites have reasonable record lengths. Data statistics (e.g., mean,

standard deviations; skewness and kurtosis) were obtained for all selected sites. Such statistics give

a first indication about type and shape of the FF distributions. Goodness of Fit tests were used in

order to find out the most acceptable distributions at the selected sites. Visual inspection of probability

plots and the Moment Ratio diagrams were also used, and it was found that the most acceptable

distributions are the EV-1; GEV; and GPAR except for some stations which have errors or short

record length of the flow data. Figure 2-5 shows an example of the most acceptable flood frequency

distributions from some selected sites of the Sobat region. The figure shows results for River Sobat

(left) and River Pibor (right). Similarly, Figure 2-6 shows those results for River Yei (left) and River Jur

(right) of the Sudd region in the Nile Basin.


Regional flood frequency curves, Lake Victoria

Sta

ndar

dise

d flo

od p

eak


0.00

0.35

0.70

1.05

1.40

1.75

2.10

2.45

2.80

3.15

3.50

1

2

3


2 5 10 20 50 100

Fitting: GEV-PWM1 -- EAST VECTORIA LAKE2 -- KAKERA3 -- SOUTH EAST LAKE VECTOR

• Figure 2-4, Regional flood frequency curves for the main three regions; (region 1: N-E; region 2: S-E; region 3: river Kagera - west of Lake Victoria).

0

100

200

300

400

500

600

700

800

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500

GEV P-III LN3 GLOG EV1 GPAR EXPN Observed

0

200

400

600

800

1000

1200

1400

1600

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


R. SOBAT R. Pibor

0

100

200

300

400

500

600

700

800

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

200

400

600

800

1000

1200

1400

1600

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

100

200

300

400

500

600

700

800

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

100

200

300

400

500

600

700

800

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

200

400

600

800

1000

1200

1400

1600

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

200

400

600

800

1000

1200

1400

1600

Qua

ntile

0

200

400

600

800

1000

1200

1400

1600

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


R. SOBAT R. Pibor

• Figure 2-5, Distribution comparison, River Sobat (left) and River Pibor (right), MOM method (extreme value paper).

2.2.2.1.2) Regional Homogeneity and Regional Distribution

The regional homogeneity tests indicate possible heterogeneous regions, although no individual

station can be considered discordant. By inspecting the regional probability plots, it is clear that either

the EV-1; GEV, GPAR, or the P-III can be used as a regional distribution for the River Sobat region.

However, only the GPAR distribution passes the L-moments regional goodness of fit test. It was

concluded that the moment ratio diagram with the method of moment of parameter estimation gives a

reasonable results comparing to the probability weighted method.

2.2.2.1.3) General Conclusion (Model 2)

The EV-1, GEV and GPAR distributions are good candidates for both the at-site and regional flood frequency analyses. However, short records, on the other hand, shed some doubt on the validity of the analysis for some stations.


0

500

1000

1500

2000

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

100

200

300

400

500

600

700

800Q

uant

ile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


R. YeiR. Jur

0

500

1000

1500

2000

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

500

1000

1500

2000

Qua

ntile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

100

200

300

400

500

600

700

800Q

uant

ile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


0

100

200

300

400

500

600

700

800Q

uant

ile

P (%)T (yr)

11.01

101.11

502

805

9010

9520

9850

99100

99.5200

99.8500


R. YeiR. Jur

• Figure 2-6, Distribution comparison, River Yei (left) and River Jur (left), MOM method (extreme value paper).

2.2.2.2) Extreme Value Analysis Using Quantile-Quantile (Q-Q) Plots (Model 3)

In the preceding discussion, the conventional statistical theory was used to determine the appropriate

distribution of a flood series. Extreme flood events are highly dependent on the form of the right

portion of the underlying flood frequency distribution (the right tail) which is most difficult to estimate

from observed data since records are often of short lengths. Therefore there is a need for a separate

analysis to represent the form of a right tail of a modeled flood distribution. Hence, this will enable a

reliable estimation of floods with higher return periods.

More specifically, in extreme value analysis, the tail of a distribution describing the probability of

occurrence of extreme event was analyzed and modeled by a separate distribution (e.g. Gumbel,

Exponential, Generalized Pareto, Weibull, Pearson), and statistical tests were performed to find the

‘’best’’ distribution. A methodology has been developed to recognize the anomalies in tail behavior in

an easy and visual way by means of the so-called Q-Q plots (quantile-quantile plots). According to

this analysis, the shape of the tail is classified on the basis of the value of the extreme value index (γ).

The different classes of the distribution’s tail are shown in Figure 2-7.

The Q-Q plots methodology is implemented in a package called ECQ developed by the hydraulics

laboratory of K.U. Leuven, Belgium. It has been applied to the data of the selected areas by all theme

researchers of the participating countries to derive the best fit flood frequency distribution curves. For

each station the whole flood series was considered as one population and the tail of the distribution

was analyzed. A normal tail was found for most of the stations and consequently number of

candidate distributions was limited to the Gumbel EV1/Exponential class.


0

0 .0 0 5

0 .0 1

0 .0 1 5

0 .0 2

0 .0 2 5

0 .0 3

0 .0 3 5

0 .0 4

0 .0 4 5

0 .0 5

0 1 0 2 0 3 0 4 0 5 0 6 0

x

prob

abili

ty d

ensi

ty f X(

x)

E x tre m e v a lu e in d e x : p o s it iv e z e ro n e g a t iv e

γ > 0 Pareto class heavy tail γ = 0 Gumble/Exponential class normal tailγ < 0 Final right end point light tail

0

0 .0 0 5

0 .0 1

0 .0 1 5

0 .0 2

0 .0 2 5

0 .0 3

0 .0 3 5

0 .0 4

0 .0 4 5

0 .0 5

0 1 0 2 0 3 0 4 0 5 0 6 0

x

prob

abili

ty d

ensi

ty f X(

x)

E x tre m e v a lu e in d e x : p o s it iv e z e ro n e g a t iv e

γ > 0 Pareto class heavy tail γ = 0 Gumble/Exponential class normal tailγ < 0 Final right end point light tail

• Figure 2-7, Different classes of distribution’s tail according to extreme value index (γ).

• Figure 2-8, Rivers and stations in North-Eastern and South-Eastern side of Lake Victoria.

2.2.2.2.1) Applications of the Q-Q Approach to the Selected Sites in the Nile Basin

The FFA has been conducted by WRRI, Egypt, and the theme researchers of the participating

countries for the main regions in the Nile Basin as listed before in Table 2-1; see also Figure 2-8


a: River Sobat and its Sub-Basins and Upper Rivers of the Sudd Region:

Based on the analysis of the Q-Q plots for Rivers Sobat at Hillet Doleib, Pibor; Baro (at Gumbella and

U.S. Adura), Akkobo, Gila and upper Rivers of the Sudd Region, a normal tail was found. Therefore,

the selected distributions were evaluated in the exponential Q-Q plot for EV1/Gumbel and GEV using

MOM, ML, and PWM. Based on the previous tests, it was concluded that the Extreme Value–Type 1

(EV1) and the General Extreme Value (GEV) distributions are the most valid at-site distributions. A

comparison is made between the calibration results making use of the three methods (MOM, MLM

and PWM), and the differences were very small. Figure 2-9 shows the EV-1 distributions for River

Sobat at Hillet Doleib and River Pibor in the Sobat region. Also, Figure 2-10 shows the EV-1

distributions for River Yei and Lol at Nyamlell in the Sudd region. Comparison of the calibration results

for the distribution parameters (for EV1/Gumbel and for GEV, and according to different parameter

estimation methods) is given in Table 2-2.

0

200

400

600

800

1000

1200

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-ln(exceedance probability)

obse

rvat

ions

observations

extreme value distribution

optimal threshold

EV1/Gumbel, MOM

EV1/Gumbel, MOM0

50

100

150

200

250

300

350

400

450

500


obse

rvat

ions

observationsextreme value distributionoptimal thresholdEV1/Gumbel, MOMEV1/Gumbel, MLEV1/Gumbel, PWM

• Figure 2-9, A comparison of two EV-1 distributions in the Sobat Region; River Sobat at Hillet Doleib (left) and River Pibor (right).

0

50

100

150

200

250

300

350

400

450

500

0 0.5 1 1.5 2 2.5 3-ln(exceedance probability)

obse

rvat

ions

observations


optimal threshold

EV1/Gumbel, MOM

0

100

200

300

400

500

600

700

800

900

1000

0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)

obse

rvat

ions

observations


optimal threshold

EV1/Gumbel, MOM

• Figure 2-10, A comparison of two different EV-1 distributions in River Sudd Region; River Yei (left) and River Lol at Nyamlell (right).


0

100

200

300

400

500

600

700

800

900


obse

rvat

ions

observations


optimal threshold

censored POT values

extreme value distribution highest POT values

EV1/Gumbel, MOM

0

10

20

30

40

50

60

70

80

90

0 0.5 1 1.5 2 2.5 3 3.5-ln(exceedance probability)

obse

rvat

ions

observations


optimal threshold

censored POT values

• Figure 2-11, Exponential Q-Q plot for non-flooded and flooded events; River Pibor in Sobat region (left) and River Akkobo in Sudd region.

• Table 2-2, Results for different parameter estimation methods for River Sobat and its sub-basins.

Baro, Baro, Sobat, Distributions Pibor Akobo Gila

u.s. Andura Gumbeilla Hillet Doleib

β 69.97 9.91 9.96 86.25 291.42 85.88 MOM

xt 152.36 23.68 78.23 572.79 1059.42 705.37

β 78.16 10.48 12.25 86.19 461.89 105.34 ML

xt 148.81 23.386 77.5 570.32 1021.21 702.29

β 71.213 9.63 9.22 87.52 276.17 84.23

EV1

PWM xt 151.645 23.84 78.658 572.06 1067.78 706.33

γ -0.23 -0.115 -0.32 -0.143 -0.692 -0.1137

β 87.16 11.268 13.07 100.55 410.04 97.51 MOM

xt 158.97 24.06 79.71 577.14 1127.35 708.61

γ -0.283 -0.064 -0.362 -0.015 -0.764 -0.16

β 86.99 10.75 13.19 95.67 382.19 103.76 ML

xt 161.07 23.75 79.957 568.78 1187.76 711.13

γ -0.47 0.59 - -0.36 -0.852 -0.531

β 94.41 13.14 - 111.21 399.22 113.42

GEV

PWM

xt 170.09 27.004 - 588.82 1201.91 730.93

For some sites, the whole range of the flow observations does not follow an EV1/Gumbel distribution.

This might be explained by flooding influence along this river. Therefore, it is suggested to calibrate a

separate distribution for the two subpopulations (i.e., for the non-flooded observations and the flooded

observations) as shown in Figure 2-11, which shows an exponential


Q-Q plot for EV1/Gumbel for non-Flooded and Flooded Events for River Pibor of River Sobat Region

(left) and for River Akkobo of the Sudd Region (right). The physical reasons of the split-up between

the two subpopulations may be attributed to the flood plain of the ‘Machar marches’. River Sobat

reflects the combined influence of the Baro (low variability) and the Pibor (high variability). River Baro

shows less variability because peaks are overtopped due to spills into the ‘Machar marches’, while

Pibor, Akobo and Gila rivers have highly variable flows due to water supplied by the Ethiopian

tributaries spills from the Baro along the Mokwai swamps.

b: Blue Nile and River Atbara:

The Q-Q plots methodology was applied to the data of the Blue Nile and River Atbara (Eastern Nile)

using the ESQ software. For each station the whole flood series was considered as one population

and the tail of the distribution was analyzed. A normal tail was found for all stations and consequently

the number of the candidate distributions was limited to the Gumbel EV1/Exponential class. Figure 2-

12 shows examples of the Exponential Q-Q plot for EV1/Gumbel using the Method of Moment (MOM)

for the Blue Nile at Ed Deim station and river Setitte at Hawata. Table 2-3 gives a summary of the

parameters of the fitted EV1 distribution to the data of the Blue Nile at Ed Deim; Snnar; and

Khartoum, and River Atbara at Gwasi; Hawata; Wad Hileiw and Kubur stations. Also the results

derived on the basis of the LN3, LN2 and Pearson-III distributions were added. The differences

between the Q-Q plots of these rivers may be attributed to the catchment characteristics. Stations

Eddeim, Sennar, Gwasi and Hawata show less variability because peaks are overtopped due to spills

into flood plains since they are more downstream than other stations. Kubur and Wad El Hileiw have

highly variable flows due to the hilly areas of the catchments.

• Table 2-3, Summary of distribution parameters o the EV-1 with MOM.

Distributions Ed Deim Sennar Khartou

m

Gwasi Hawata Wad El

Hileiw

Kubur

RMSE 26.29 35.84 31.20 21.23 18.44 21.67 7.82 EV1 MOM/ML/PWM

Chi 37.63 82.69 79.16 5.93 9.49 9.49 2.62

RMSE 20.12 27.12 25.45 20.49 10.36 27.32 9.67 2LN MOM/ML/PWM

Chi 2.03 2.00 2.54 5.93 7.82 9.49 3.27

RMSE 8.83 21.12 21.84 17.83 9.85 16.34 6.78 3LN MOM/ML/PWM

Chi 0.30 0.30 1.79 5.52 7.82 7.82 2.42

RMSE 19.88 16.96 28.26 25.34 12.23 21.95 12.42 PIII MOM/ML/PWM

Chi 0.33 13.45 4.00 5.93 9.49 9.49 6.98


0

100200

300

400

500600

700

0 1 2 3 4-ln(exceedance probability)

obse

rvat

ions

observat ionsextreme value distribut ionoptimal thresholdextreme value distribut ion highest POT valuesLN3, momP-III, momEV1/Gumbel, momLN3, mml

0

200

400

600

800

1000

1200


obse

rvat

ions

observationsextreme value distributionoptimal thresholdP-III, momEV1/Gumbel, mom

0

200

400

600

800

1000

1200


obse

rvat

ions

observationsextreme value distributionoptimal thresholdP-III, momEV1/Gumbel, mom

• Figure 2-12, Exponential Q-Q plots for EV1/Gumbel using MOM for the Blue Nile at EL-Deim (left) and River Setitte at Hawata(right).

During the analysis, it was noted that data series have two parts where the upper part presents higher

discharge rates where observations bend down, and the lower part presents less discharge rates.

This can be attributed to the flooding influence at all the station for higher discharge rates. The

influence of flooding on the extreme value analysis was eliminated by censoring based on the

asymptotic properties of the extreme value distribution. The results of application of this approach to

the Blue Nile at EL-Deim and River Setit at Heleiw are shown in Figure 2-13.

0

200

400

600

800

1000

1200

1400

1600

1800


obse

rvat

ions

observationsextreme value distributionoptimal thresholdextreme value distribution highest POT valuesEV1/Gumbel, MOMGEV, MOM

0

5

10

15

20

25

30


obse

rvat

ions

observationsextreme value distributionoptimal thresholdcensored POT valuesextreme value distribution highest POT valuesEV1/Gumbel, MOMEV1/Gumbel, MOM

• Figure 2-13, EV-1 and GEV distributions for the flooded and non-flooded segments of Blue Nile at Ed Deim station (left) and River Rahad at Heleiw station (right).


0

50

100

150

200

250

300

350

0 0.5 1 1.5 2 2.5 3 3.5 4

-ln(exceedance probability)

Dis

char

ge [m

3/s]

observationsextreme value distributionoptimal thresholdEV1/Gumbel, MOMGEV, PWM

0

200

400

600

800

1000

1200

0 0.5 1 1.5 2 2.5 3 3.5 4

-ln(exceedance probability)

Dis

char

ge [m

3/s]

observationsextreme value distributionoptimal thresholdcensored POT valuesEV1/Gumbel, MOMEV1/Gumbel, MLGEV, PWM

• Figure 2-14, Ev1 and GEV distribution plot for River Sondu (left) and River Nyando (right).

0

1

2

3

4

5

6

7

8


ln(o

bser

vatio

ns)

observationsextreme value distributionoptimal thresholdGEV, PWMGEV, ML

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5


ln(o

bser

vatio

ns)

observationsextreme value distributionoptimal thresholdGEV, PWMGEV, ML

• Figure 2-15, GEV distribution plot for River Nzoia (left) and River Awach (right).

c- North-Eastern Side of Lake Victoria

FFA is performed for the previously mentioned stations in Table 2-1and shown in Figure 2-14. The

extreme value index was positive for all rivers except rivers Yala and Nzoia, which suggests that a

heavy tail distribution is preferred for these locations. The General Extreme Value distribution (GEV)

was therefore fitted for these stations. Rivers Yala and Nzoia, however, exhibit a normal tail behavior

in the upper points The Gumbel/Extreme Value type I (EV1) is therefore preferred for these two

locations. The optimal threshold, above which the weighted regression has to be performed in an

optimal estimation of γ, is the threshold that minimizes the mean squared error (MSE) of the

regression. It ranks for the six stations as 16, 16, 45, 25, 13 and 22 for Sondu, Nyando,Yala, Nzoia,

Migori and Awach, respectively. Above these ranks, the distributions were considered best fits.

However because of the effect of flooding in the case of Yala and Nzoia, it was necessary to fit

extreme value distribution to two split samples. A threshold rank of 13 was used to split the samples

in the case of Nzoia and 11 in the case of Yala to fit the upper tail points. and Figure 2-15 show the

results of this analysis for rivers Sondu,and Nyando, and Nzoia, and Awash, respectively. The whole

range of observations does not follow an EV1/Gumbel distribution for the case of River Yala. In River

Yala, for discharges higher than 23.5 m3/s, the observations bend down. This might be explained by

floods along this river and neighboring basins. Therefore, it is suggested to calibrate a separate


distribution for the two subpopulations one for the non-flooded observations and another for the

flooded observations. A comparison among the calibration results for the parameters of EV1/Gumbel

and GEV distributions using different parameter estimation methods is given in Table 2-4. Differences

between the Q-Q plots of these rivers may be attributed to the differences in the catchment

characteristics.

• Table 2-4, Summary of distribution parameters for the rivers in N-E Side of Lake Victoria.

Distributions &

Parameter estimation

Sondu

Nyando

Yala

Nzoia

Migori

Awach

β - - 27.49 112.022 - - MOM

ξτ - - 78.558 254.448 - -

β - - 28.647 111.24 - - ML

ξτ - - 78.183 254.448 - -

β - - 28.675 117.96-8 - -

EV1

P\WM ξτ - - 77.874 251.016 - -

γ 0.3821 0.3013 - - 0.3477 0.7266

β 99.433 36.06 - - 144.951 4.681 ML

ξτ 145.694 54.781 - - 197.835 7.594

γ 0.4019 0.2198 - - 0.2649 0.4041

β 96.898 38.599 - - 153.914 6.111

GEV

PWM

ξτ 143.43 55.617 - - 201.615 8.307

d- South-Eastern Side of Lake Victoria

FFA is conducted for the previously mentioned stations in Table 2-1 and shown in Figure 2-8. For the

south-eastern side of Lake Victoria and based on the analysis of the Q-Q plots, a normal tail is found

for stations in this area. The selected distribution is the EV1/Gumbel distribution. The EV1/Gumbel

distribution fitted using MOM, ML, and PWM is evaluated in the exponential Q-Q plot. The whole

range of observations follows an EV1/Gumbel distribution except for the single highest observed

value. Since the time series is quite short it cannot be suggested that the value is an outlier. Overall

EV1/PWM procedure gives the best fit the observed data. Figure 2-16 shows the results of this

analysis for stations Ngono/Kyaka (left) and Moame/Mabuki (right).


020

4060

80

100120

140160

180200

0 1 2 3-ln(exceedance probability)

obse

rvat

ions


0

10

20

30

40

50

60

70

80


obse

rvat

ions


0

10

20

30

40

50

60

70

80


obse

rvat

ions


• Figure 2-16, Exponential Q-Q plot for EV1/Gumbel using MOM, ML, and PWM for stations Ngono/Kyaka (left) and Moame/Mabuki (right).

A summary of distribution parameters of EV1/Gumbel obtained using different parameter estimation

methods for the rivers in south-east of Lake Victoria is given in Table 2-5.

• Table 2-5, Summary of distribution parameters for the rivers in S-E Side of Lake Victoria.

Moame/

Mabuki

Magogo/

Shinyanga

Simiyu/

Road

Brg

Kagera

Nyakanyasi

Distributions

Methods of

Parameter

estimation

Ngono/

Kalebe

Ngono/

Kyaka

Ruvuma/

Mwendo

β 57.74 80.48 196.29 27.64 30.67 217.21 276.09 MOM

ξτ 70.93 28.46 75.30 15.29 18.86 91.83 55.18

β 67.45 81.11 199.95 27.51 31.36 217.62 276.10 ML

ξτ 42.81 26.74 64.00 15.99 17.28 87.95 54.97

β 65.14 79.05 194.03 26.97 30.91 212.89 268.39

EV1

PWM ξτ 58.09 30.92 79.19 16.45 18.43 99.28 68.50

e- Upper Awash sub-basins (Ethiopian Plateau).

It is identified that the station on the Awash River near Bello is situated down-stream a flood plain

where flows spread in the plain rather than remain confined to the natural channel. This, in turn, led to

highly smoothened peaks and low magnitudes of maximum flows which did not match those obtained

from flood plains having similar drainage sizes, subject to similar climatic and physico-geographical

conditions, and receiving more or less the same magnitude of rainfall. Based on the analysis of the Q-

Q plots, a normal tail was found. Therefore, the selected distributions are evaluated in the


exponential Q-Q plot where the results showed that both EV1 and Extreme Value distributions are

reasonable representation of the tail. Figure 2-17shows a comparison between EV1 and Extreme

value distributions for Akaki River (left), and the flooding effect for the Teji River flood plain (right) to

select the best fit distributions for the two segments of observations (i.e. upper and lower parts).

-.

0

50

100

150

200

250

300

350

400


obse

rvat

ions

EV1/Gumbel, MOMEV1/

-.

0

50

100

150

200

250

300

350

400


obse

rvat

ions

EV1/Gumbel, MOMEV1/Gumbel, PWM

1/EV Gumbel, ML

observationsextreme value dist.optimal threshold

Gumbel, PWM

1/EV Gumbel, ML1/EV Gumbel, ML

observationsobservationsextreme value dist.extreme value dist.optimal thresholdoptimal threshold

0

50

100

150

200

250

300


obse

rvat

ions

observationsextreme value distributionoptimal thresholdcensored POT valuesEV1/Gumbel, MOM

• Figure 2-17, Comparison of EV1 and Extreme value distributions of Akaki (left), and flooding effect at Teji Rivers (right).

2.3) Regional Flood Frequency Analysis (RFFA)

A methodology for Regional Flood Frequency Analysis (RFFA) to be used in the coming phase for all

theme researchers has been defined. A digital elevation maps of the Nile basin with a resolution of 90

m is now available and being used for estimating the geometric characteristics of the selected basins.

A regionally calithbrated relation between the average daily maximum discharge and the catchment's

characteristics (area, average slope and average annual rainfall) was established using multiple linear

regressions. This relation, together with the developed regional frequency curves could be used to

estimate flood magnitudes with various return periods for un-gauged catchments in any region that

can be considered homogeneous with the regions under consideration. The Regional Flood

Frequency Analysis has been carried out by the following methods:

1. Visual inspection of the FFDs;

2. Correlation analysis;

3. Analysis by the L-Moments method:

3.1. L-Moments ratio diagram;

3.2. Discordancy measure, D(I);

3.3. Heterogenety measures, H(I);

3.4. Goodness of fit measures, Z value.


2.3.1) Visual inspection of the FFDs

The Annual Maximum (AM) flow data was normalized by the Mean Annual Floods (MAF) in order to

show the variability in the flow series. The Regional Flood Frequency Distributions (EV-1) and/or

(GEV) of the rivers were plotted and visualized in order to find out the best acceptable distribution of

the observations. The Regional Flood Frequency Distribution of River Sobat and Sudd regions are

shown in Figure 2-18.

0

0.5

1

1.5

2

2.5

3

3.5

4

1 10 100

Return period [years]

Dis

char

ge /

MA

F [-]

PiborAkoboGilaBaro, AnduraBaro, GumbellaSobat, outletRegional curve Baro-SobatRegional curve Pibor-Akoboriver flood curve Piborriver flood curve Akobo

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 10 100


Dis

char

ge /

MA

F [-]

LolTonjGelWokkoYeiWauRegional curve

• Figure 2-18, Regional Flood Frequency Distribution (EV-1) for River Sobat and its Sub-Basins (left) and River in Sudd Region (right).

It was found that the catchment characteristics explaining the differences in the flow series. River

Baro shows less variability because peaks are overtopped due to spills into the ‘Machar marches’

swamps. Pibor, Akobo and Gila rivers have highly variable flows due to the supplied water by the

Ethiopian tributaries spills from the Baro along the Mokwai. Hence, River Sobat reflects the combined

influence of the Baro (low variability) and the Pibor (high variability).

The discharge-return period (Q -T) relationships for all sites for the Blue Nile and Atbara River as

obtained from extreme value analysis were plotted together with the discharge being expressed in

dimensionless or standardized form (by dividing by the mean) and the results are shown Figure 2--19.

It is clearly observed that the Blue Nile region and River Atbara region are not homogeneous and this,

as has been mentioned before, is due to the varying hydrological phenomena responsible for

generating the flood events over the two regions. Consequently, a regional frequency curve has been

developed for each region. Similarly, Figure 2- 20 shows regional frequency visualization for the rivers

of the Awash basin in the Ethiopian plateau.


0

0.5

1

1.5

2

2.5

1 10 100

Return Period [year]

Disc

harg

e / M

AF

[-]

Ed DeimSennarKhartoumDinder at GwasiRahad at HawataRegional curve Blue NileRegional Frequency Curve for AtbaraWad El HeliewK b

• Figure 2-19, Regional data and regional frequency curves for the Blue Nile and Atbara River.

0

0.5

1

1.5

2

2.5

3

3.5

1 10 100


Dis

char

ge /

MA

F [-]

Berga, Addis AlemHoletaTeji, AsgoriAkakiAwash, Melka KuntureAwash, HomboleMojo river, Mojo VillegRegional curve Awashriver flood curve, Holetariver flood curve, Tejiriver flood curve, Awash Melka Kuntureriver flood curve, Awash Hombole

• Figure 2-20, Regional data and regional frequency curves for the rivers in Awash basin in the Ethiopian plateau.

2.3.2) Correlation Analysis

Correlation Analysis Includes:

a. Catchment Characteristics,

• MAF with Areas,

• MAF/Area with MA. rainfall,

• MAF with Lengths

b. Meteorological Measurements


Table 2-6 and Table 2-7 give rivers, stations, catchment areas, maximum daily flows and rainfall for

the Blue Nile, Atbara River, and the Rivers of the North-Eastern side of Lake Victoria. Figure 2-21

shows the correlation of the MAF with both the Areas and the MAR (mm) for the Blue Nile and Atbara

River, while Figure 2-22 shows it for the rivers in the North-Eastern side of Lake Victoria. The

regression equations are shown for each area with their correlation coefficient on the two figures.

• Table 2-6, Overview of Catchment Characteristics in Blue Nile and Atbara River.

River Station Catchmet Area

Sq. Km.

Av. Slope Av. Max. daily

disch. M m3/d

Av Max Daily

Rain mm

Blue Nile Ed Deim 179,486 0.0016 672 18.2

Rahad River Gwasi 35,600 0.0012 14 13.7

Atbara River Kubur 96,393 0.0050 213.09 22.4

Atbara River Wad El Hileiw 42,845 0.005 245.79 21.1

Blue Nile Khartoum 251,486 0.0015 638.43 17.85

Dinder River Hawata 37,000 0.0012 47 14.3

• Table 2-7, Overview of catchment characteristics of the Rivers in Kenya.

River Station Area [km2] MAR [mm]

Sondu 1JG01 3287 1064.47

Nyando 1GD04 2520 1425.32

Yala 1FG01 2388 1143.87

Nzoia 1EE01 11849 1679.8

Migori 1KC03 3046 1398.0

Awach 1HA14 104 1362.18


y = 2E-07x1.4529 , R2 = 0.8345

10

100

1000

100000.000 1000000.000 10000000.000

A [km2] * MAR [mm]

MA

F [M

illio

n m

3/da

y]

• Figure 2-21, Correlation of the MAF with both the Areas and the MAR (mm) for the Blue Nile and Atbara river.

y = 1.1061x + 2.1127R2 = 0.7555

3

4

5

6

7

8

9

0 1 2 3 4 5 6 7LN(AREA)

LN(M

AF)

• Figure 2-22, Correlation of the MAF with both areas and the MAR (mm) rivers in the North-Eastern side of Lake Victoria.


2.3.3) Analysis by the L-Moments method

L-moments are linear combinations of order statistics which are robust to outliers and unbiased for

small samples, making them suitable for flood frequency analysis, including identification of

distribution and parameter estimation. It is defined as linear combinations of probability weighted

moments (PWM). The L-moment of the frequency distribution can be used for the distribution

selection, and for deriving tests of homogeneity and discordance.

2.3.3.1) L-Moments Ratio Diagram

L-moments are linear combinations of order statistics which are robust to outliers and unbiased for

small samples, making them suitable for rainfall frequency analysis, including identification of

distribution and parameter estimation (Hosking, 1990; Hosking and Wallis, 1993). It is defined as the

linear combinations of probability weighted moments (PWM):

( )[ ]{ }TT xFE=β (1)

where F(x) is the cumulative distribution function of x. The first four L-moments expressed as linear

combinations of PWMs are:

01234

0123

012

01

123020,66

,2,

ββββλβββλ

ββλβλ

−+−=+−=

−==

(2)

The L-mean, λ1, is a measure of central tendency and the L-standard deviation, λ2, is a measure of

dispersion. The ratio, λ2/λ1 , is termed the L-coefficient of variation, τ2; the ratio, λ3/λ2 , is referred to L-

skewness, CS, the ratio, λ4/λ2 , is referred to L-kurtosis, CK. The L-moment of the frequency

distribution can be used for the distribution selection, and for deriving tests of homogeneity and

discordance.

It aims to form graphs of flood sites that approximately follow one distribution within a region. It can be

employed in the regional context to test if the data observed at different sites in a homogeneous

region can be assumed to arise from a common distribution, or, more generally, if one can assume

that certain distribution parameters related to kurtosis (Ck), skewness (Cs), or coefficient of variation

are constant within a region. For a given distribution, conventional moments can be expressed as

functions of the parameters of distributions. It follows that the higher order moments can be

expressed as functions of lower moments. The (Cs) and (Ck) of the L-Moment, and (Cs-Ck) moments

ratio diagrams for the popular probability distribution with the method of Probability Weighted Moment

(PWM) and the Method of Moment (MOM) of parameter estimation are shown in Figure 2-23 for the

River Sobat region.


2.3.3.2) Discordancy Measure, D (I)

If a single site does not appear to belong to the cloud of (Cs - Ck) points on the moment ratio

diagram, a test of discordance based on L-moments can be used to determine whether it should be

removed from the region. The test of discordance was applied by calculating the D-statistic D(I),

defined in terms of L-moments. If a site’s D-statistic exceeds 3.0, its data is considered to be

discordant from the rest of the regional data and two possibilities must be investigated; either there

may be an error in the data or the flow station may properly belong to another region or no region at

all. Table 2-8 gives summary results of the Discordancy Test for the River Sobat and its Sub-basins

and the Sudd regions.

• Figure 2-23, Moments ratio diagram presenting CS2 (left) and CS (right) versus kurtosis for the River Sobat region.

0 0.2 0.4 0.6 0.8 1 1.20

1

2

3

4

5

6

Beta1 (CS 2)

Bet

a2 (C

K)

SITESMEAN GEV GLOG LOGN P-IIIGPAR

R. Sobat

R.AkoboR.Gila

R. BaroR.Pipor

0 0.2 0.4 0.6 0.8 1 1.20

1

2

3

4

5

6

Beta1 (CS 2)

Bet

a2 (C

K)


R. Sobat

R.AkoboR.Gila

R. BaroR.Pipor

-1 -0.5 0 0.5 1 1.50

1

2

3

4

5

6

CS (Skewness)

CK

(Kur

tosi

s)


R.GilaR.Pipor R. Baro

R.Akobo

R. Sobat

-

-1 -0.5 0 0.5 1 1.50

1

2

3

4

5

6

CS (Skewness)

CK

(Kur

tosi

s)


R.GilaR.Pipor R. Baro

R.Akobo

R. Sobat

-


• Table 2-8, Summary results of the Discordancy Test for the Rivers of Sobat and Sudd Regions.

Region Rivers N L-SKEW L-CV L-KURT D(I)

River Sobat Sobat

Pibor

Akkobo

Gila

Baro

59

28

10

18

52

0.0773

0.2561

0.0761

0.2270

0.0974

-0.1292

-0.1021 -

1.2598 -

0.1579

-0.0430

0.1325

- 0.1236

- 0.2166

- 0.1190

- 0.0973

1.33

0.66

1.33

0.35

1.32

Sudd Lol

Tonj

Gel

Wokko

Yei

Jur

31

11

10

18

13

30

0.1768

0.1440

0.3221

0.3636

0.1646

0.1867

- 0.1952

- 0.8367

0.0065

0.2478

0.3249

0.0780

0.0289

-0.1997

-0.2561

0.0992

- 0.1482

- 0.0014

0.45

1.59

1.47

1.50

0.42

0.57

2.3.3.3) Heterogeneity Test for Regions

If the variability of the cloud is great, the possibility that they do not belong to a single population can

be tested by means of L-moment heterogeneity tests. The L-moment tests fit a four parameter Kappa

distribution to the regional data set, generate a series of a 500 equivalent regions, data by numerical

simulation and compare the variability of the L-statistics of the actual region to those of the simulated

series. Three heterogeneity statistics can be employed to test the variability of the different L-statistics:

H1 for L-Cv, H2 for the combination of L-Cv and L-Cs and H3 for the combination of the L-Ck and L-

Cs. The H-statistic indicates that the region is acceptable homogeneous when H < 1, possibly

heterogeneous when 1≤ H < 2, and definitely heterogeneous when H ≥ 2. A group of sites must

therefore have H < 2 to be considered as a possibly homogenous region. The results of the

HETEROGENEITY Measures for the selected flood measuring stations in the Sobat and Sudd

Regions are definitely heterogeneous since H >2.


• Table 2-9, Goodness of Fit Test (Z-Values) for River Sobat and Sudd regions.

Region Distributions L- KURTOSIS Z – VALUE

GLOG 0.190 9.02

GEV 0.113 5.91

GNOR 0.145 7.20

PIII 0.132 6.67

Sobat

GPAR 0.006 1.15 *

GLOG 0.186 6.84

GEV 0.110 4.58

GNOR 1.141 5.50

PIII 0.130 5.17

Sudd

GPAR 0.008 1.07*

( * : Z-statistic Values of the accepted distributions)

2.3.3.4) Goodness of Fit Test for Identify Parent Distribution

Once data within a region are homogenous and belongs to a single parametric distribution, a

goodness-of-fit criteria based on L-moments can be used to select one various unimodal distribution

(EV-1, GEV, GPAR, PIII, LN3, etc..) and to estimate its parameters. Flood frequencies within the

region were then determined based on the fitted regional distribution. The goodness-of-fit criteria for

each of various distributions were defined in terms of L-moments and termed the Z-statistic. It was

found that the all selected regional distributions are not accepted since the absolute values of Z-

statistic is greater than 1.64, except the GPAR since the value of Z is equal to 1.15 for River Sobat

region and 1.07 for the Sudd region, as given in Table 2-9. This means that no single regional

distribution can be used for all of these selected flood sites in the Nile Basin according to the

goodness-of-fit test.


2.4) Limitations and Constraints

Topographic, land use, soil type and digital elevation maps with a proper resolution are not available

for the study area(s). These maps are helpful to obtain the chatchment characteristics which are the

main parameters in the RFFA.

2.5) Conclusions

The following are the main findings and lesson learned through the FFA research component:

• Enhanced knowledge transfer and exchange among Flemish and Nile countries experts,

• Trust, confidence, mutual trust and understandings have been developed among the research team of the FFA component. These enable the exchange of data among the different countries, which is an important issue in the cooperation in the field of Flood Frequency Analysis,

• Enhanced methodologies and promoted relevant flood analysis practical research;

• Introduce new ideas for the application of the GIS procedures in the RFFA is recommended.

2.6) The Way Ahead

The following is a list of proposed future studies in the regionalization analysis as concluded during

the research work of the FFA component:

• Extraction of catchment characteristics:

o based on DEM: percentage in different slope classes (0-1, 1-2, 2-3, … degradation),

o based on land cover maps for percentage of urban, percentage of agricultural land and percentage of forest, (Africover data) at the web. of Africa Data Dissemination Service: (http://edcintl.cr.usgs.gov/add/)

o based on soil type map (percentage of sand, percentage of loam and percentage of clay) .

• Update analysis of MAF versus catchment characteristics;

• Mean annual rainfall (MAR) should also be re-defined and used in the context of the duration in which the flood peaks are observed;

• Comparison between Peak Over Threshold (POT) and annual maxima methods (for some selected stations);

• Estimation of Flow –Duration – Frequency (QDF) relationships (with different time scales);

• More advanced homogeneity tests;

• Uncertainty analysis on regional curves.


2.7) References

Beirlant J.et al. (1996), “Practical analysis of extreme values”, Leuven University Press, Leuven.

Willems P. (1998), “Hydrological applications of extreme value analysis”, in Hydrology in a changing environment, edited by H. Wheater and C. Kirby, John Wiley & Sons, Chichester, vol. III, 15-25.

Hosking, J.R.M., (1990), “L-Moments Analysis and estimation of distributions using linear combination of order statistics”, J.R. Stat. Ser. B, 52 (1), 105-124.

Hosking, J.R.M. and Wallis, J.R., (1993),’Some Statistics Useful in Regional Frequency Analysis, Water Resour. Res. 29 (2), 271-281.


Chapter

3

Rainfall-Runoff Modeling

3.1) General Introduction

This section is a summary of activities of the first phase (2001 to 2006) of the Rainfall-Runoff

Modeling Research Component under the FRIEND/NILE Project. Actual technical activities (mainly

on data, numerical models and applications) are presented in the main body of this chapter.

Participants of this research component are from Egypt, Ethiopia, Kenya, Sudan and Tanzania. The

coordinating center of this research component is at University of Dar es Salaam, Tanzania. Table

3-1 shows a timetable, a list of activities and output of this research component. These activities and

output will be discussed in more details later in this chapter.

• Table 3-1, Timetable, activities and output of the rainfall-runoff research component.

Year Activities Outputs

2002 1. Organize the research setting for the rainfall-runoff

component

2. Search for hydrological models to be used

3. Plan for training on model application by resource persons

1. Rainfall-runoff research component

2. Names of lumped and semi distributed

hydrological models

3. Training follow up and resource persons

2003 1. Acquire and processing of hydro-metrological data from each

country

2. Acquire GFFS and WMS models software

3. Consultation visits to the coordinating center by the Flemish

counterpart

4. Training workshop in Alexandria on the acquired models

5. GFFS, WMS-HEC1 and WMS-HSPF preliminary models

applications

1. Processed hydro-meteorological data sets

2. Hydrological models ready for training and

application

3. Compiled data for use in the training

workshop

4. Theme researchers trained on use of WMS

and GFFS models for application on their

countries data sets

5. Preliminary model results


Year Activities Outputs

2004 1. GFFS, WMS-HEC1, WMS-HSPF and HMS Models

applications

2. Follow up workshops in Dar es Salaam and Addis Abba

3. Consulting visit to the coordinating center by the Flemish

counterpart

4. Collect technical reports from each country

5. Review technical reports

6. Search topics for the next phase (2006-2009)

1. Model results

2. Presentations and discussions about model

applications results

3. Technical reports

4. Technical comments on the model

applications

5. Outline of research activities for the phase I

of research

2005 1. GFFS, WMS-HEC1, WMS-HSPF and HMS Models

applications

2. Collect technical reports from each country

3. Review technical reports

4. Present topics for the next phase (2006-2009)

5. Review papers for the 2005 conference

6. Follow up workshop in Khartoum

7. Prepare and hold the FRIEND NILE International conference

1. Model results

2. Technical reports

3. Technical comments on the model

applications

4. Details of research activities for the next

phase

5. Technical comments on the papers

6. Presentations, discussions and finalization

of conference papers on model applications

7. Review full papers and conference in

Sharm El Sheikh

3.1.1) The Rainfall-Runoff Modeling (RRM) Component

Theme researchers for the Rainfall-Runoff Modeling (RRM) component are from Egypt, Ethiopia,

Kenya, Sudan and Tanzania. The coordinating center of the component activities is the Water

Resources Engineering Department at University of Dar es Salaam, Tanzania. The coordinator for

the component is Prof. Felix W. Mtalo assisted by Dr. Deogratias M.M. Mulungu and other several

academic staff of the department. The RRM component has resource persons from Belgium, USA

and UK, who provide technical support and advice on the execution of the research activities.

The main activities in the first phase were acquiring equipment for the researchers and processing of

hydro-meteorological data; acquiring and building hydrological models, doing a literature review by

collecting related papers and reports, training and follow-up workshops, preparing proposals for the

next phase of the project, producing papers for the FRIEND/Nile conference, and preparing and

holding the Nile Friend conference in November 2005. Some bottlenecks emerged during execution

of research activities. Several problems, such as model software bugs, computer crashes, lack of

data from different countries, and delay in submission of reports, were among many other problems

that had been faced by the research team of this component. However, every effort was made to

rescue the situations and enable production of expected research outputs.


One of the most important achievements of the research activities in the first phase is a high level of

cooperation among the participating countries especially in the data sharing. The year 2006 marked

the end of the first phase of the project and based on the outputs and experiences of the first phase,

research activities for the next phase of the project were proposed.

3.1.2) Objectives of the Rainfall-Runoff Modeling Component

The main objective of the Rainfall-Runoff modeling component of the FRIEND/Nile Project is to apply

suitable rainfall runoff models on selected pilot catchments at least one per participating country in

order to:

• Achieve a reasonable degree of accuracy from the simulation point of view;

• Forecast flow at selected locations;

• Estimate flows at the ungauged catchments on regional basins;

• Investigate the impacts of the land use or climate change on the river flows.

3.2) Data Acquired for Rainfall-Runoff Modeling

The year 2002 marked the start of the research activities under the FRIEND/NILE acronym. For the

RRM component, setting of research activities were considered and mainly focused on hydrological

data collection ready for model application in the pilot catchments. The data collection took sometime

and in some cases it was a continuous activity throughout the first phase in order to perfect the model

results. The selected rainfall-runoff models were the Galway Flow Forecasting System (GFFS),

Watershed Modeling System (WMS) - Hydrologic Simulation Program FORTRAN (HSPF), Soil Water

Assessment Tool (SWAT), Watershed Modeling System (WMS) - Hydrologic Engineering Center

(HEC-1) and Hydrologic Engineering Center (HEC) – Hydrologic Modeling System (HMS). With

exception of GFFS that uses time series data only, other models in addition use satellite derived

spatial data such as Digital Elevation Model (DEM), Land use/cover and soil data to estimate runoff

and river flows. A list of data collected in each country and some used in the rainfall-runoff modeling

are presented in the following subsections.

3.2.1) Egypt

For Egypt, the collected data consisted of hourly storm rainfall and flow for three stations in Wadis Al-

Arbain, El-Guidierate and Sudr. Evaporation data for two years were initially collected. For detailed

model simulations, hydrologic and spatial data were provided and considered. The hydrologic data

includes stream flows data, which were measured by the flow measuring devices that were mounted

on the hydraulic structure at the main stream of each wadi. These hydraulic structures are sharp

crested weir of a 10 m width at wadi Sudr and a rectangular cross-section of 4.7 m width at wadi AL-

Arbain. The stream flows were measured as individual events according to the characteristics of the

arid regions. The rainfall was measured at the same times as the flow data. The measured rainfall

and runoff data for each basin were obtained and used for the simulation purposes.


For the spatial data, thematic coverage (soil type, vegetation cover, rainfall, evaporation, etc) were

compiled. The availability of this type of information is highly variable throughout the basin. In arid

regions like Sinai, the geology and soils are the most dominant factors in the estimation of the

resulting runoff. It is also known that in such regions, evaporation rate is very high and there is neither

land-use nor vegetation cover. The runoff from each Wadi was calculated on the basis of

synchronous observation of water levels, velocity and wetted cross section of the control section of

the measurement. Such observations are used to depict the relationship between the stage (levels)

and the flooding discharge that is known by the rating curve. Based on the discharge rating curves,

the runoff hydrographs together with the causative rainfall hyetographs are evaluated. Many different

hydrologic parameters were also determined such as rainfall intensity, lag-time, rainfall duration, loss

by infiltration and evaporation, peak flow and the runoff coefficient by the WMS software.

3.2.2) Ethiopia

For the Awash basin in Ethiopia, the used data in implementing research activities is as presented in

Table 3-2.

• Table 3-2, Hydro-meteorological data: Ethiopia.

Data type Details

Rainfall data 5 stations ranging from (1991-2001).

Climatic data: Temperature, Wind speed

humidity, sunshine hours Evaporation

Temperature – 1 station monthly data. Wind speed – 1 station monthly data.

Sunshine monthly hours for 1 station.

Relative Humidity: 1 station.

Evaporation – 3 stations for seasonal means

Discharge data Daily discharges – 4 stations.

1991-2001 – 2 stations

1991-1997 – 2 stations

Seasonal maximum and minimum.

3.2.3) Kenya

For Kenya, the theme researcher submitted data to the coordinating center as summarized in Table

3-3.


• Table 3-3, hydro-meteorological data: Nzoia and Sondu catchments, Kenya.

Data type Details

Rainfall data 25 rainfall stations in excel format (1990 – 2000) without location description. Cold

Cloud Cover (CCC) data for 89 stations at 1 x 1 degree resolution for 1990 –

1995.

Climatic data. Temperature,Evaporation. 16 Evaporation stations data

Discharge data 5 River flow/discharge data in Excel from 1990-2000.

5 Gauging stations with water level data with corresponding rating curves

Spatial data Topographical land use.

Soils for the whole of Kenya in Arc-View at 1 x 1 degree resolution.

The data was cleaned (screened) and made ready for application of GFFS. However, the researcher

emphasized the use of distributed models because of the complex hydrological phenomenon of the

catchments.

3.2.4) Sudan

In Sudan, the theme lead researcher changed from Dr. Ahmed Eldaw to Dr. K. Bashar. The data

received from Sudan included a water balance sheet on excel format for 4 stations, and monthly

rainfall data for two stations for the durations 1916 -1983 and 1931-1982. Initially, the models were

applied on the Blue Nile basin, but it was realized that the available data was not good for the models.

It was then decided after the Dar es Salaam workshop in January 2003 that Sudan should

concentrate on data from the Eddeim catchment of the Blue Nile basin.

• Table 3-4, Hydro-meteorological data: Lake Victoria.

Data type Details

Rainfall data Daily series for 132 stations with record durations of more than 20 years

maximum duration 76 years minimum 5 years.

Climatic data temperature, wind speed,

speed Relative Humidity, sunshine hours

and Radiation

13 stations with several lengths varying from 4 years to a maximum of 17

years. Most stations start 1970 and stop around 1984.

Discharge (flows) data 14 stations distributed within the catchments.

Spatial data -Topographical map (DEM) 1km x 1km.

-Land use map (shape files)

-Soil data – (shape files)

-Geological maps – without legend.


3.2.5) Tanzania

All the data from the Lake Victoria basin including Simiyu catchment was collected. The data was

processed to the format of the GFFS software and model calibration and validation was done.

Despite huge amount of efforts, the application of the WMS model on the Simiyu data was not

immediately possible due to model bug in the application of the WMS-HSPF software. The problem

was communicated to the resource persons and planned for discussions and solutions during the

follow up workshop. Also, work on WMS/HEC-1 used hypothetical data since real catchment data

such as hourly flow data was not available for calibration. Table 3-4 gives a summary of data

collected from the Tanzanian side of the Lake Victoria catchments. For Tanzania, all hydro-

meteorological data required for the research activities was acquired and made available at the

coordinating center (Table 3-5).

• Table 3-5, Data used for the GFFS modeling in the Victoria catchments

The data sets Basin Area

(km2)

Country Rainfall

Stations

Used No.

Years

Starting

Year

Calibration Verification

Nzoia 12,676 Kenya 12 10 1990 Jan 1990-Dec

1997

Jan 1998-July

1999

Sondu 3,450 Kenya 10 5 1970 Jan 1970-Dec

1972

Jan 1973-Dec

1974

Simiyu 5,320 Tanzania 5 9 1970 Jan 1970-

Dec 1975

Jan 1976-

Sept 1978

Blue Nile 176,572 Ethiopia/Sudan 10 7 1990 Jan 1990-Dec

1994

Jan 1995-Dec

1996

Awash 7,656 Ethiopia 9 12 1991 Jan 1991-

Dec 1999

Jan 2000-

Sept. 2002

3.3) Case Studies in Each Country

The geography of the Nile Basin is both distinct and varied. From the most remote source at the head

of the River Luvironzo near Lake Tanganyika and Lake Tana of Ethiopia, to its mouth on the

Mediterranean Sea downstream. The 6850 km long Nile is the world’s longest river, and flows from

south to north with a catchment basin covering approximately 10% of the African continent. The river

spreads across 10 countries with an area of 3 × 106 km2.

The countries participating in the Rainfall-Runoff modeling component of the FRIEND/NILE project

are Egypt, Ethiopia, Kenya, Sudan and Tanzania. Figure 3-1 presents the Nile basin countries and

the pilot catchments. The pilot catchments were: Sinai (El-Gudierate of Northeast Sinai, Al-Arbain of

South Sinai and Sudr of Southwest Sinai) catchments in Egypt, Awash in Ethiopia, Nzoia and Sondu


in Kenya, Wadi Abu Habil and later Addeim (bordering between Sudan and Ethiopia) in Sudan, and

Simiyu in Tanzania.

• Figure 3-1, The study basins around lake Victoria and Nile basin.

Active research activities started after the models training workshop in Alexandria, Egypt, after which

follow up workshops were organized to review the research progress following the training and work

on countries’ pilot catchments. This included finalization of data collection, revised model applications

and communication of bottlenecks in the execution of research.

The five catchments on the upstream of the Nile, which were used in the rainfall-runoff modeling

represent the main two sources of the Nile, namely the Ethiopian highlands and the equatorial lakes.

Two catchments were selected from the Ethiopian highlands, the Blue Nile (176,572 km2) and the

Awash (7,656 km2). In the equatorial lakes three basins were selected, Simiyu (5,320 km2) in

Tanzania, Sondu (3,450 km2) in Kenya and Nzoia (12,676 km2) in Kenya.

On the downstream, in the arid and semi-arid regions of Sinai in Egypt, Table 3-6 gives a summary of

the physiographic characteristics of the two selected catchments of the Wadi Sudr, South - East of

Sinai and Wadi AL-Arbain, South of Sinai in Egypt. The two catchments were selected to represent

the changes in topography, geology, and the climatic conditions in Sinai. Sudr catchment is located at

the South-Western side of Sinai and is one of the largest wadis in South – West of Sinai, which flows

westward, and discharges into the Gulf of Suez at Sudr town. It covers a total area of 560 km2 and a

drainage area of 360 km2 at the water level recorder of the flow measuring station. The wadi


originates in the hill slope of EL-Tih plateau. The Al-Arbain is located in the Southern part of Sinai at

the upper part of Wadi Feiran (1865 km2), which represents the mountainous area of EL Egma, and

EL Teih plateau at ST. Kathrien. It covers an area of 32 km2.

• Table 3-6, The Physiographic characteristics of the Wadi Sudr and Al-Arbian of Sinai catchments.

Physiographic characteristics Sudr catchment AL Arbain catchment

Location South-West of Sinai South Sinai

Origin Gable EL-Raha and Somar EL Egma and ELTeh Plateau

Outlet W. Sudr and Gulf of Suez Wadi Feiran and Gulf of Suez

Surface geology Limestone, Upper Cretaceous,

Cenomanian

Basement Granite

Area (km2) 360 32

Length (km) 76 7

There is a wide variation among the Nile catchments, which well represents the topography of the

Nile basin. Climatologically, the selected Nile catchments represent the whole scale from the arid to

the tropical zones where the mean annual rainfall varies from few hundreds of millimeters to some

thousands of millimeters.

The river catchments used for the research and the Nile basin at large are composed of varying

terrain, land cover and climate. A summary description of the pilot catchments is presented in Table

3-7, which summarizes the reported catchment characteristics. These features make the hydrology of

the basin interesting and challenging in addition to competing water uses among development

sectors and riparian countries. Therefore, there is a great need to increase the efforts to understand

the hydrology and for possible mitigation of adverse impacts on river flows and management

problems.

• Table 3-7, Nile catchments summarized characteristics.

Country Catchment Terrain Land cover Climate

Egypt Sinai Mountainous Bare land Semi desert

Ethiopia Awash Mountainous Not defined Wet

Kenya Nzoia, Nyando Hilly & Mild Grass/Woodland, Cultivated, Forest Wet, Wet / Dry

Sudan Eddeim Mountainous Not defined Wet / Dry

Tanzania Simiyu Mild Grass/Woodland, Cultivated Wet / Dry


3.4) Models Used

3.4.1) GFFS

The Galway River Flow Forecasting System (GFFS) is a software package developed at the

Department of Engineering Hydrology, National University of Galway Ireland, it comprises a suite of

models for simulation, updating and real-time forecasting applications. The degree of structural

complexity, associated parameter parsimony, and difficulty in objective function evaluation of these

models, varies considerably. The application of the GFFS (collection of systems and conceptual

models) software involved the models: Simple Linear Model (SLM), the seasonally based Linear

Perturbation Model (LPM), the wetness index based Linear Varying Gain Factor Model (LVGFM),

Artificial Neural Network (ANN) Model and the conceptual Soil Moisture Accounting and Routing

(SMAR). The LPM in non-parametric or parametric form, the LVGF model the ANN and the SMAR

model can be used to forecast (reproduce). In catchments that exhibit significant storage effects the

LPM and SMAR may perform better than the other models. In catchments with non-linear

transformation of rainfall to runoff the ANN may perform better. The model assumptions are described

below and data input for calibration and validation periods is presented in Table 3-5.

3.4.1.1) SLM

The SLM approach assumes a linear time-invariant relationship between the total rainfall and the total

discharge. For the non-parametric SLM, the input-output relationship for lumped, linear, time invariant

system expressed in terms of a series of pulses or mean values over successive short intervals T,

can be conveniently obtained from the response to unit pulse of duration T which is a convenient

expression of the operation of the system. The convolution summation relation expresses the discrete

linear input-output relationship in terms of sampled pulse response.

For the parametric modeling, the Gamma function model is used. Constraint to the shape and volume

of the estimated pulse response functions is obtained by parametric modeling where a solution is

sought within the constraint of an assumed model form. Based on prior knowledge of the system

behavior the response function is represented by a suitable mathematical equation involving only a

few parameters. The parameters must be estimated by optimization through a search in the space of

reasonable parameter values. For multiple input-single output system under the constraints of the

gamma function impulse response the parameters n and k (of the gamma function) and Gf (of the

SLM) must be found for each input.

3.4.1.2) LPM

LPM assumes that during a year in which the rainfall is identical to its seasonal expectation, the

corresponding discharge hydrograph is also identical to its seasonal expectation. However, in all other

years, when the rainfall and the discharge values depart from their respective seasonal expectations,

these departures series are assumed to be related by a linear time invariant system. Thus, the linear


perturbation model uses the information contained in the observed seasonal variation of the

hydrograph to reduce the dependence on linearity and increase the dependence on observed

seasonal behavior. The model therefore assumes the following: (1) if each input function for each day

of the year is equal to its expected value for that date, the output will also equal its expectation for that

date and (2) the perturbations or the departure from the date expected input values are linearly

related to the corresponding perturbations or departures from the date expected output values. The

relation of the departure (perturbation) series (input and output) of the LPM is also represented by the

convolution summation.

3.4.1.3) LVGFM

The LVGFM involves only the variation of the gain factor with the selected index of the prevailing

catchment wetness, but not the shape (i.e. the weights) of the response function. Using a time-varying

gain factor, the model output has the structure similar to SLM but in LVGFM the gain factor is linearly

related to an index of the soil moisture state, obtained from the outputs of the naïve SLM, which

operates as an auxiliary model.

SMAR

SLM

LMP

ANN

LVGFM

Output neuron

Inputs

Hidden layer

Input layer

SMAR

SLM

LMP

ANN

LVGFM

Output neuron

Inputs

Hidden layer

Input layer

• Figure 3-2, Schematic diagram of the Artificial Neural Network model.

The “multi-layer feed-forward network” type of artificial neural network was used (see Figure 3-2). It

consists of an input layer, an output layer and only one “hidden” layer located between the input and

the output layers. Each neuron of a particular layer has connection pathways to all the neurons in the

following adjacent layer, but none to those of its own layer or to those of the previous layer (if any).

Likewise, nodes in non-adjacent layers are unconnected. In the output layer, there is only one neuron,

for the single output. Because the neural network itself does not incorporate storage effects, storage

is implicitly accounted for by the use of the output series of the naive SLM. For a neuron either in the


hidden or in the output layer, each received input is transformed to its output by a mathematical

transfer function. The non-linear transfer function adopted for the neurons of the hidden and output

layers is the widely used logistic/sigmoid function, which has values bounded in the range [0,1]. The

concept of neurons and some constants can all be interpreted as parameters of the network

configuration.

Conversion to Potential Rate

( T )

Excess Rainfall x= R - T E×

Direct Runoff ( H )

Rainfall in Excess of Infiltration Capacity (Y )

Soil Moisture Storage in mms

( Z ) Evaporation

( C )

Moisture in Excess of Soil Capacity

(g)

Linear Routing Component

(n,nK)

Total estimated

Evaporation (E)

Rainfall(R)

Layer 1

Layer 2

Layer (Z/25)

oo

(1- ′H ) x ≤ Y

Ground water Component

A Linear Reservoir

(Kg)

r2 = (1- ′H ) x-Y if (1- ′H ) x > Y

Generated surface Runoff

T E×

r1= ′H x

r3

rg rs

• Figure 3-3, Schematic diagram of SMAR Model.

3.4.1.4) SMAR

The SMAR Model is a development of the ‘layers’ conceptual rainfall-runoff model with water balance

components. Using a number of empirical and assumed relations, which are considered to be at least

physically plausible, the non-linear water balance (i.e. soil moisture accounting) component ensures

satisfaction of the continuity equation, over each time-step. The routing component, on the other

hand, simulates the attenuation and the diffusive effects of the catchment by routing the various

generated runoff components through conservative linear time-invariant storage elements. For each

time-step, the combined output of the two routing elements adopted (i.e. one for generated ‘surface

runoff’ as input and the other for generated ‘groundwater runoff’ as input) becomes the simulated

discharge forecast.

The water balance component of SMAR operates as a vertical stack of horizontal soil layers. Each

layer can contain a certain amount of water at field capacity (see Figure 3-3). Evaporation occurs from

the top layer at a potential rate and from the second layer on exhaustion of the top layer at the


remaining potential rate multiplied by a parameter C whose value is less than unity. On exhaustion of

the second layer evaporation proceeds from the third layer at the remaining potential rate multiplied

by C2 and so on. Thus, a constant potential evaporation rate applied to the basin reduces the soil

moisture storage in a roughly exponential manner.

3.4.1.5) Methods of Combining the Estimates of Different Models

Instead of relying on one individual model or switching between models, an alternative approach is to

generate estimates / forecasts simultaneously from a number of different models and then combine

these forecasts in an optimum manner. This can be done in several ways such as:

The Simple Average Method (SAM):

This is the simplest method for combining the outputs of different individual models and finding the

average at each time step. It has been found that the SAM method can produce forecasts that are

better than those of the individual models and its accuracy depends mainly on the number of the

models involved and on the actual forecasting ability of the specific models included in the simple

average.

The Weighted Average Method (WAM):

The SAM method can be quite inefficient when some individual models selected for combination

consistently produce more accurate forecasts than others. In this case the use of a weighted average

method would be preferable and weights are attached to each model output and the equation can be

treated as a multiple linear regression model. Then the Ordinary Least Squares (OLS) estimate the

weights vectors. In the WAM, the sum of weights are normally constrained to unity and the OLS

solution of equation may not ensure the satisfaction of the constraints of equation therefore the

method of Constrained Least Squares (CLS) can be used to estimate the weights vector. Some

studies pointed out that the main disadvantage of the WAM is that it may suffer from multi-collinearity

problem, which results in unstable estimates of the weights reducing the advantages obtained from

combining the different models forecasts. The degree of multi-collinearity increases with the increase

in the forecasting ability of the individual models as well as when the forecasts of the individual

models used are very similar not necessarily being good.

The Neural Network Method (NNM):

The SAM and WAM methods are relatively simple methods of combining the forecasts. An alternative

method is the NNM, which can be used to test whether a more complex relationship such as a non-

linear function mapping of inputs into the network output, is needed for the combinations. The same

type of neural network, multi layer feed forward network is used and is very powerful in function

modeling.


The multi layer feed forward neural network used consists of an input layer, an output layer and only

one hidden layer between the input and output layers. A layer is usually a group of neurons having

same pattern of connection pathways to the other neurons of adjacent layers. Each neuron in a

particular layer has connection pathways to all the neurons in the next adjacent layer but not to those

of the same layer (see Figure 3-2). The number of neuron in the input layer is equal to the number of

elements in the external input array to the network. In this study the elements of the external array are

the forecast of selected models each of which is assigned to only one neuron. These inputs were

transformed to outputs using a transfer function. The outputs of these neurons in the input layer are

distributed through connection pathways to the neurons of the single hidden layer.

The hidden neurons have no direct connection with either the external input or output of the network.

Each neuron in the hidden layer receives its input through connection pathways from the neurons of

the input layer and transmits their output along the connection pathways to all the neurons of the

output layer. The output layer in turn accumulates the transmitted input and produces the network

output. The number of neurons in the output layer equals the number of outputs expected from the

network. A neuron in the hidden or output layer receives inputs and transforms it to output by a

mathematical transfer function. These network parameters are usually estimated by a procedure

referred to in neural networks as training analogous to the calibration procedure in hydrological

modeling. The transfer function is usually non-linear and the most widely used one is the logistic

function.

Precipitation

Interception Storage

Surface DetentionStorage

Infiltration

Inactive Groundwater

Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage

Active GroundwaterStorage

InterflowOutflow

GroundwaterOutflow

Simulated Stream flow

ET

ET

ET

ET ET

Precipitation



Infiltration


Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage


InterflowOutflow

GroundwaterOutflow


ET

ET

ET

ET ET

Precipitation



Infiltration


Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage


InterflowOutflow

GroundwaterOutflow


ET

ET

ET

ET ET

Precipitation



Infiltration


Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage


InterflowOutflow

GroundwaterOutflow


ET

ET

ET

ET ET

Precipitation



Infiltration


Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage


InterflowOutflow

GroundwaterOutflow


ET

ET

ET

ET ET

Precipitation



Infiltration


Overland Flow

InterflowStorage

Upper ZoneStorage

Percolation

Lower ZoneStorage


InterflowOutflow

GroundwaterOutflow


ET

ET

ET

ET ET

Figure 3-4, HSPF conceptual hydrologic model.


3.4.2) HSPF

The HSPF model is housed within the WMS. HSPF is a semi-distributed, continuous simulation

model that can perform a detailed simulation of the hydrology and water quality in a watershed.

Figure 3-4 shows a conceptual representation of the model. It is a versatile model that can simulate

watersheds varying greatly in size from parking lots to major watersheds. HSPF has a modular

structure; it has three main modules: PERLND, IMPLND and RCHRES. Pervious land segments over

which an appreciable amount of water infiltrates into the ground are modeled with PERLND module.

Impervious land segments, where infiltration is negligible, such as paved urban surfaces, are

simulated with IMPLND module. Processes occurring in water bodies like streams and lakes are

treated by RCHRES module.

Watershed Modeling System (WMS) is GIS-based pre/post processing software that supports many

hydrologic/hydraulic and water quality models widely used by water resources managers/engineers. It

provides a user-friendly interface for developing necessary input files for these models. It also

provides some graphics and animation capabilities, if applicable, to view the resultant output from

these models. The HSPF interface in WMS is used to help generate the necessary input file for

HSPF. WMS is used to analyze digital elevation and land use data as a preprocessor. A User Control

Input (UCI) file is generated and used to run the model. Calibration of the model is done to manually.

The calibrated model results are compared to the observed stream flows and statistical techniques

are used to verify the compliance.

3.4.3) WMS/HEC-1

The WMS/HEC-1 model is also housed within the WMS. The WMS provides tools for all phases of

watershed modeling including automated watershed and sub-basins delineation, geometric

parameter computation, hydraulic parameter computation (e.g. Curve Number (CN) and Lag-Time

(TL)) and result visualization. The digital terrain modeling functions of WMS were used to create

terrain models using Geographic Information Systems (GIS) data, and gridded DEMs. These data

were used to delineate watersheds, streams and sub-basins. These data are analyzed and used to

simulate the surface runoff storms using WMS/HEC-1 model. The model simulates runoff volumes

and hydrographs for rainfall storms. Different unit hydrograph methods, different loss estimation

methods and different methods of lag-time computation are available in WMS/HEC-1 and have been

analyzed. The range of CN can be obtained from standard tables according to the soil type and cover

of each basin. Since there are no definite calibration procedures in this software, the method of lag-

time (TL) computation was selected in this application according to the lag-time of each storm at each

basin and the CN in order to match the estimated hydrograph with the observed one with respect to

the volume, the peak, and the time to peak. The WMS/HEC-1 parameters were setup as follows:

1) Precipitation: A precipitation hyetograph is used as the input for all runoff calculations. The

precipitation – time distribution was entered with a time step of fifteen minutes. The total depth of the

rainfall was also given.


2) Loss method: One of several different loss methods can be chosen when generating synthetic

hydrographs. In this study, the SCS and Uniform methods were selected.

(3) Unit hydrograph method: One of several different unit hydrograph methods can be chosen when

generating synthetic hydrographs. The SCS dimensionless Unit Hydrograph, Snyder, and the Derived

Unit Hydrograph methods were used.

(4) Lag-Time: The SCS and the Riverside (mountains, foothills and valley) method of Lag-Time

computation have been analyzed. The Curve Number (CN) was obtained from standard tables

according to the soil type and cover of each basin.

3.4.4) SWAT

The SWAT model is a physically based input model and requires data such as weather variables, soil

properties, topography, vegetation and land management practices occurring in the catchment. The

model was developed for continuous simulation, as opposed to single event models. The physical

processes associated with water flow, sediment transport, crop growth, nutrient cycling, etc are

directly modeled by SWAT using the above-mentioned input data. Some of the advantages of the

model includes: modeling of ungauged catchments, prediction of relative impacts of scenarios

(alternative input data) such as changes in management practices, climate and vegetation on water

quality, quantity or other variables. SWAT also has a weather simulation model that generates daily

data for rainfall, solar radiation, relative humidity, wind speed and temperature from the average

monthly variables of these data. The occurrence of rain on a given day has a major impact on relative

humidity, temperature and solar radiation for the day. The weather generator first independently

generates rainfall/precipitation for the day. Maximum temperature, minimum temperature, solar

radiation and relative humidity are then generated based on the presence or absence of rain for the

day. Finally, wind speed is generated. This provides a useful tool to fill in missing daily data in the

observed records.

Moreover, the model performs dormancy calculations properly set for simulations in tropical areas.

Dormancy is a period in which a plant does not grow, awaiting for the necessary environmental

conditions such as temperature, moisture and nutrient availability. The term also refers to condition of

relative inactivity as applied to seeds, tubers and perennial plants during the winter. Since vegetation

is an important part of water movement in land surface processes, appropriate dormancy calculations

are important for estimating consumptive use of water by vegetation, which ultimately contributes to

good match of the water balance.


• Figure 3-5, Sub-basin command loop.

As a semi-distributed (subbasin set-up) model, SWAT is attractive for its computational efficiency as it

offers some compromise between the constraints imposed by the other model types such as lumped,

conceptual or fully distributed, physically based models. SWAT incorporates a kinematic storage

model for subsurface. A model operation in Hydrologic Response Units (HRUs) or sub-basin

command loop is shown in Figure 3-5

SWAT simulates the land phase of hydrologic cycle using the water balance equation, whose evapo-

transpiration components are estimated from three methods. SWAT uses the hourly and daily time

steps to calculate surface runoff. For hourly, the Green and Ampt equation is used and for the daily an

empirical SCS Curve Number (CN) method is used. The application in Simiyu catchment used daily

simulation and the water available for infiltration and subsequent percolation is obtained as the

difference between the rainfall and surface runoff.


GroundwaterLayer 2 storage

Layer 1 storageGroundwater

Percolation

Percolation

Percolation

Tensionzone

storage storagezone

Upper

Groundwater flow

Interflow

InfiltrationSoil profilestorage

Surfacedepressionstorage

Surface runoff

Precipitation Evapotranspiration

storageinterceptionCanopy

• Figure 3-6, Schematic diagram of HMS-SMA algorithm (HEC 2000).

3.4.5) HMS

Continuous hydrologic models account for the soil moisture balance in the catchment over a long-

term period. Various hydrologic physical processes such as: interception, surface depression storage,

infiltration, soil storage, percolation, and groundwater storage are considered in continuous

simulation. The Hydrologic Modeling System (HMS) model was used for continuous hydrologic

simulation of the Blue Nile. The model encompasses Soil Moisture Accounting (SMA) algorithm to

simulate the long-term relationship between rainfall, runoff, storage, evapotranspiration and soil

losses of the Blue Nile River / watershed. SMA algorithm counts on rainfall depths and

evapotranspiration rate as inputs to define rainfall, runoff, storage and losses relationships. There are


five storage zones as shown in Figure 3-6. For the simulation of water movement through the various

storage zones, the maximum capacity (maximum depth) of each storage zone, initial storage

condition in terms of percentage of the filled portion of each zone and the transfer rates such as the

maximum infiltration rate are required (Fleming and Neary, 2004). The SMA algorithm has a linear

structure and may be a source of error in simulating the rainfall-runoff process, which is a non-linear

process.

According to the SMA algorithm, evapotranspiration is assumed to take place only during dry periods

and in stages from the canopy interception storage then from surface depression storage and then

from the soil profile storage. Soil percolation will start only when the tension zone capacity is fulfilled.

The outflow from the “groundwater layer 2 storage” as percolation will be considered as a loss from

the system.

3.5) Obtained Results

3.5.1) GFFS Model Results

In GFFS models applications, systems and conceptual modeling techniques were applied to the Lake

Victoria catchments (Simiyu, Sondu and Nzoia), Awash and the Blue Nile catchment up to Eddeim of

the Ethiopian high lands. The models were applied in non-parametric and parametric forms.

Parameter optimization is carried out by ordinary least squares, Rosenbrock, Simplex and genetic

algorithm. The areal rainfall, which is the main input to these models, was estimated using arithmetic

mean method. The results of performances of the substantive models are shown in Table 3-8 and

Table 3-9, respectively, for the simulation and updating (parametric) modes.

• Table 3-8, Model efficiencies in percentages for the simulation mode.

Period Basin Method SLM LPM LVGFM SMAR ANN

Nzoia OLS 60.0 67.0 64.0 68.0 54.0

Sondu OLS 44.0 67.0 49.0 68.0 67.0

Simiyu OLS 32.3 39.4 49.9 46.5 52.7

Blue Nile OLS 77.8 92.1 91.2 90.5 91.8

Calibration

Awash OLS 52.0 72.0 53.0 72.0 51.0

Nzoia OLS 49.0 44.0 45.0 43.0 41.0

Sondu OLS 34.0 42.0 21.0 45.0 68.0

Simiyu OLS 24.8 31.6 41.8 31.0 40.9

Blue Nile OLS 76.0 91.1 89.0 89.2 90.7

Verification

Awash OLS 47.0 64.0 55.0 79.0 39.0


• Table 3-9, Model efficiencies in percentages for the updating mode.

MOCT Period Basin Method SLM LPM

SAM WAW NNM

Nzoia Parametric 96.0 97.0 98.0 96.0 97.0

Sondu Parametric 98.0 99.0 98.0 99.0 98.0

Simiyu Parametric 63.4 69.9 68.3 69.9 84.2

Blue Nile Parametric 98.2 98.6 89.9 92.4 92.4

Calibration

Awash Parametric 48.0 76.0 81.3 82.9 82.8

Nzoia Parametric 94.0 96.0 - - -

Sondu Parametric 89.0 90.0 - - -

Simiyu Parametric 73.6 71.6 73.1 71.6 60.8

Blue Nile Parametric 97.3 97.2 88.4 91.6 91.5

Verification

Awash Parametric 40.0 59.0 78.8 82.8 82.4

Examples of plots of observed and simulated discharges are shown in Figure 3-7 and Figure 3-8 for

the Upper Awash sub-basin in Ethiopia using SMAR model.

In Upper Awash sub-basin, the SMAR model performed better than all the models in all performance

criteria used in both simulation as well as verification mode. However, since the catchment is

heterogeneous land use/cover characteristics, it was recommended to divide the catchment to sub-

catchments and to model the sub-basin in a distributed way.

• Figure 3-7, Plots of observed and SMAR simulated discharges 1999.


• Figure 3-8, Plots of observed and AR simulated discharges 1999.

Comparison Between Different Simulated Hydrographs with the Obserevd (Wadi AL-Arbain Basin - South of Sinai) Storm of 22/3/1991

1- Method of Unit Hydrograph

0

1

2

3

4

5

6

9.5 10 10.1 10.2 10.3 10.4 10.5 10.6 10.7 10.8 10.9 11 11.5 12 12.5Time (hrs.)

Q m

^3/s

.

Observed Flow

Given Observed Unit Hydrograph

Snyder

SCS

• Figure 3-9, Comparison between observed and simulated hydrographs at Wadi AL-Arbain using 1-method of unit hydrograph.

In the model output-updating procedures, the Auto-Regressive (AR) updating procedure performed

best than the others in updating the model outputs specifically for the lead-time of one day. The

performance falls sharply as the lead-time increases. The Linear Transfer Function (LTF) followed by

the Non Linear Auto Regressive Exogenous input model using Neural Network (NARXM) gave very

good results. Therefore, it can be concluded that to forecast inflows, volumes and peak flows, to the

Koka reservoir for better reservoir management, the SMAR model coupled with AR updating

procedure is suitable to forecast reservoir inflows (volumes) for reservoir management and peak flows

for the operation of the spillway gates to avoid huge releases through the gates which might causes

floods down-stream.


Comparison between Observed and WMS Hydrographs (EL-Arbaien Basin - South Sinai)

Storm of 22/3/19911- Method of Losses

0

1

2

3

4

5

6

10 10.1

10.2

10.3

10.4

10.5

10.6

10.7

10.8

10.9 11 Time (hrs.)

Q

(m^3

/s.)

OBS.SCSUniform Loss

• Figure 3-10, Comparison between observed and simulated hydrographs at Wadi AL-Arbain using 1-method of losses.

Com parison betw een O bserved and W M S H ydrographs, Sudr basin - South S inai

1- M ethod of Losses

0

100

200

300

400

500

600

9.259.75

10.2510.75

11.2511.75

12.2512.75

13.2513.75

14.25

Tim e (hrs.)

Q

m^3

/s.

O bserved

W M S - SCS

W M S-Uniform Loss

• Figure 3-11, Comparison between observed and simulated hydrographs at Wadi Sudr using the method of losses.

3.5.2) WMS/HEC-1 Model Results

From the DEM of Wadi AL-Arbain, the sub-basins were delineated and the geometric data were

computed. The WMS/HEC-1 parameters (precipitation, loss method and the Lag-Time) were setup

and the outflow hydrographs for the given rainfall events were generated. The simulated hydrographs

are compared with the observed storm of 22/3/1991 as shown in Figure 3-9 and Figure 3-10. For

Wadi Sudr the results are as given in Figure 3-11.

The SCS unit hydrograph method and the Riverside County method of Lag-Time computation were

used. The range of the Curve Number (CN) was obtained from standard tables according to the soil


type and cover of each basin. Since there are no definite calibration procedures in this software, the

method of Lag-Time computation was selected according to the lag-time of each storm at each basin,

the initial abstraction and the CN were increased or decreased in order to match the derived

hydrograph with the observed one using the mean square error (MSE) with respect to the volume, the

peak, and the time to peak. Table 3-10 gives the WMS/HEC-1 parameters set up for the derived

hydrographs by WMS from the two selected basins.

• Table 3-10, The WMS/HEC-1 parameters set up for the derived hydrographs of the storm of 22/3/1991.

Loss Method

Wadi Initial Loss

(mm)

Curve

Number

CN

RIMA

Precipitation

Depth (mm)

Lag-Time

(hrs)

AL-Arbain 16.0 85 0.0 35.0 0.899

Sudr 14.2 79 0.0 34.5 2.555

3.5.3) HSPF Model Results

The initial model run indicated a need for calibration. Some model parameters are, thus, manipulated

to obtain a better fitting model. The calibrated model is, then, validated on a different data set. The

calibrated values were used to simulate the river discharge. Figure 3-12 shows the simulated and the

observed stream flows for Simiyu watershed at Road Bridge Station. Similarly, observed and

validated stream flows for Simiyu watershed at Road Bridge station is shown in Figure 3-13. It is

apparent in both figures that the two hydrographs demonstrate an overall relative agreement,

although the model tends to over-predict discharges especially in the winter. Simulated peak

discharges are reasonably consistent with measured ones. Validation period tends to show more

discrepancies than the calibration period. However, the overall simulated trend seems to be very well

matching the observed one for both calibration and validations periods.

To test if the simulated and the observed flow values are in agreement, linear regression is used and

was referred as calibration assessment curve. Observed flow values (response variable) are

regressed on the simulated values (explanatory variable). The main idea is to check if the slope of the

explanatory variable is significant and close to unity and that the intercept is not significant (or

significant and equal to zero). If this holds true, we could suggest that observed values are equal to

the simulated ones (as it will form one single line on which any point will have the same values on

both explanatory and response variables. The calibration assessment curve is done on both the

calibration and the validation portions of the model (see Table 3-11).


• Figure 3-12, Final HSPF calibrated model run; simulated and observe stream flows for Simiyu Watershed at Road Bridge Station.

• Figure 3-13, Validating the calibrated model of HSPF; validated and observe stream flows for Simiyu Watershed at Road Bridge Station.

• Table 3-11, Calibration Assessment Curve for Calibration and Validation of Stream Flows for Simiyu Watershed at Road Bridge Station.

Intercept Slope

Value 2.44 0.84

p-value (significance) 0.46 0.00

90% C.I. lower bound -3.25 0.72 Calibration

90% C.I. upper bound 8.13 0.96

Value 4.05 0.72

p-value (significance) 0.13 0.08

90% C.I. lower bound 2.4 0.62 Validation

90% C.I. upper bound 5.7 0.82


As shown in Table 3-11, the p-values for the intercept in both calibration and validation indicate that

we fail to reject the null hypothesis (i.e. the intercept is equal to zero). Thus, there is no evidence that

the intercept is significant. This is also supported by the 90% confidence interval that is containing

zero for the calibration period. On the other hand, the relatively small p-values for the slope, for both

calibration and validation, indicate that the null hypothesis can be rejected for the slope (i.e. the slope

is, then, not equal to zero). Thus, there is evidence that the slope is significant. Since, the value of the

slope is ranging between 0.72 and 0.96 for calibration and between 0.62 and 0.82 for validation in a

90% confidence interval; we could say that the slope of the calibration assessment curve is

considered to be close to 1.0 and accordingly the daily simulated flows can be considered equal to

the observed ones during the calibration period.

The calibration assessment curve assured, in agreement with what was noticed in Figure 3-13, that

the validation period does not show, at least relatively, the same agreement between observed and

simulated discharges as in the calibration period.

3.5.4) SWAT Model Results

The Simiyu catchment was divided into a single catchment that falls out at Ndagalu, the land use and

soil distributed dataset were imported and overlaid over the DEM. Since a dominant Hydrologic

Response Units (HRU) distribution was used, only a single HRU was created to represent (comprise)

a Grassland (RNGE, SWAT land use class:) land use - Loamy soil combination for the Simiyu

catchment at Ndagalu from the different land use classes shown in Table 3-12 and Figure 3-14. The

period 1970 – 1974 was used for model calibration and 1976 – 1983 was used for model validation.

• Table 3-12, Simiyu Land use classes matched with the SWAT land use classes.

USGS Land use Class SWAT Land use Class % Total catchment area Dominancy Rank

Grassland RNGE 82.89 1

Shrubland RNGB 15.34 2

Deciduous Broadleaf Forest FRSD 0.94 3

Dryland Cropland & Pasture PAST 0.45 4

Cropland / Grassland Mosaic AGRR 0.37 5


• Figure 3-14, Re-classed land use data for the SWAT simulations.

• Table 3-13, Average long-term water balance 1970-1974 for SWAT model.

Water variable Observed (mm) Estimated (mm)

Total Water Yield (WYLD) 63.45 67.89

Surface Runoff (SURQ) 37.24 39.03

Base flow (GWQ) 26.21 28.86

• Figure 3-15, Observed and estimated daily discharge 1972-1973 at Ndagalu.


• Figure 3-16, Annual rainfall and potential evaporation during calibration and validation periods.

Optimum parameters for the SWAT simulations were obtained by calibration. The model simulation

results during calibration are shown in Table 3-13, Figure 3-15, and Figure 3-16. The latter, the

annual rainfall and potential evaporation during the calibration period 1970-1974, shows a more or

less constant evaporation. This is an expected pattern as for semi arid areas, where the evaporation

rate (atmospheric driven) in Tanzania is about 1800+ mm/y (FAO, 1986) and where it does not vary

much throughout the year. During validation, the annual potential evaporation from 1976 was below

the threshold rate and also it can be speculated that the deviation might be due to weather-generated

data. For example, the year 1976 is entirely missing rainfall data and the year 1977 followed with the

large percentage (16.71%) of missing rainfall data in the period 1976-1983. Although the year 1983

had rainfall and other weather data intact, the missing temperature data could have impact on

simulated water fluxes. On average, the SWAT model therefore simulated the potential evaporation

well.

• Figure 3-17, Observed and estimated annual daily discharge 1976-1983 at Ndagalu.


Because of average poor daily fit during validation, annual data was used to explain the model

performance. Figure 3-17 shows the comparison on an annual basis, where a fair match was

observed. Poor matching was observed for the years 1977 and 1983. Part of the deviations might be

due to the fact that the weather-generator was used to produce some of the weather data. The flow

for year 1983 was high because of large underestimation of evaporation (see Figure 3-16).

3.5.5) HMS Model Results

The research evaluated the performance and potentiality of the HMS on the Blue Nile River.

Observed daily rainfall and runoff records for seven years (1990 -1996) and the Digital Elevation

Model (DEM) of the Blue Nile were used for calibration (5 years that include low, moderate and high

levels of flooding) and validation (2 years with one low and high flood year) of the model.

Figure 3-18 shows simulated flow by HMS compared to observed flow for the calibration period. The

model succeeded to produce a relatively similar hydrograph shape and the Nash-Sutcliffe (1970)

coefficient of efficiency was 0.78, which may satisfactorily judge similarity and consistency between

observed and estimated hydrograph shape.

0

2000

4000

6000

8000

10000

12000

1/1/19

90

6/30/1

990

12/27

/1990

6/25/1

991

12/22

/1991

6/19/1

992

12/16

/1992

6/14/1

993

12/11

/1993

6/9/19

94

12/6/

1994

6/4/19

95

Time

Flow

Dis

char

ge, m

3/se

c

Estimated Calibration

Observed Calibration

• Figure 3-18, Simulated and observed hydrographs during calibration; HMS results.

In the calibration stage, one set of parameters was applied to the whole period, which comprised

different levels of flooding for the years. Accordingly, the model performance differed from one year to

another. For example, the model extremely overestimated flows for year 1990, which represents the

low flooding level case. The model performance was fairly better for the moderate (years 1991 and

1992) and high (years 1993 and 1994) flooding level cases.


Figure 3-19 shows simulated and observed flow for the validation period. The simulated flow values

are higher than observed ones and the Nash-Sutcliffe (1970) coefficient of efficiency was 0.69, which

is relatively small value. Similar to the calibration period, the model overestimated flows for case of

low flood year.

0

2000

4000

6000

8000

10000

12000

6/15/1

994

12/12

/1994

6/10/1

995

12/7/

1995

6/4/19

96

12/1/

1996

5/30/1

997

11/26

/1997

Time

Flow

Dis

char

ge, m

3/se

c

Estimated Calibration

Observed Calibration

• Figure 3-19, Simulated and observed hydrographs during validation; HMS results.

Although, the model produced satisfactory results for the high flow period of year 1996, it did not

respond well to the considerable high rainfall occurred in the beginning of this year, which occurred

after dry period (i.e. high rainfall on dry condition). This may be tied to the effect of the linear structure

of the SMA algorithm in simulating the non-linear rainfall-runoff process.

3.6) Findings and Lessons Learned

3.6.1) Application of GFFS Models

The application of the GFFS (collection of systems and conceptual models) software proved to be

possible with variable efficiencies in the Nile River basin. The performance of the naïve SLM is clearly

inferior to that of all other models. For catchments, characterized by strong seasonality, the LPM

outperforms the LVGFM. For large catchments with such seasonality, the LPM performs is even

better than the SMAR model. For smaller catchments, however, the SMAR conceptual model

performs consistently better than the LPM. The ANN, although characterized by a large number of

weights (parameters), does not generally perform better than the simpler models. The SMAR model

variants, having either nine or ten parameters, fail to adequately simulate the hydrological behavior of

the large catchments.

Therefore, LPM in non-parametric or parametric form, the LVGF model, the ANN and the SMAR

model can be used to forecast (reproduce). In catchments that exhibit marked storage effects, for


example Sondu and Nzoia, LPM and SMAR performed better than the other models. In Simiyu River

it seems that the transformation cannot be done under the assumption of linearity and hence the ANN

performed better. These results showed that LPM was found to be the best candidate model that can

forecast the flows in the Nile basin under a wide range of conditions ranging from marked seasonality

to marked storage effects accounting for more than 90% of the initial variance.

3.6.2) Application of WMS/HEC-1 Model

In the arid catchments where rainfall is extremely episodic (Wadis in Sudan and Egypt) the event-

based HEC-HMS model seems to work well and it has predicted well runoff volume and peak flow.

However, there was high discrepancy between the predicted runoff volume and peak flows with the

measured values for Wadi Sudr.

Sensitivity analysis for the methods of Lag-Time (TL) computation and CN was carried out. It was

found that the short lag-time results in short time to peak and high peak flows and the high CN gives

high peak flow. The short Lag-Time results in earlier and higher peaks. The best TL (lag-time) was

estimated by the Riverside Mountainous Method as 0.899 hr for Wadi Al-Arbain and by the Denver

Method as 2.555 hr for Wadi Sudr, while the observed value was equal to 2.3 hr. The best value for

the CN was estimated for Wadi Al-Arbain as 85 and for Wadi Sudr as 79.

It is recommended to use more rainfall and runoff data in order to apply all the different methods of

the unit hydrographs or using the estimated basin’s unit hydrograph to obtain more accurate results. It

is also recommended to make a water balance for each storm in the selected basin to obtain the right

ratio of the initial losses as well as the excess rainfall.

The model can be useful in proposing the locations of the water resources projects since it evaluates

the amount of the flood volume in different locations throughout any proposed area of study. It may

also give the runoff hydrograph for any selected return period rainfall storm to provide (a) warnings

against floods in order to prevent loss of life and to minimize damages to property and livestock, (b)

proper management of water resources and flood preparedness.

3.6.3) Application of HSPF Model

Using any model to simulate long-term hydrological behavior of a given watershed is not an easy

task. Most, if not all, model input parameters vary significantly with seasons and

hydrological/meteorological conditions. Statistical comparisons for calibrated and validated model

showed no evidence of a great difference between the simulated and observed data. Hence, the

model can be used for future runoff predictions in the basin. On the other hand, continuous

enhancement efforts are required to improve the model predictive power. However, it was

recommended that the observed discharge data be further revised and verified. This conforms to the

results of previous research efforts in the watershed. Moreover, there are time series that are not

implemented to the model because of lack of data. Filling this gap will also enhance the model.


Although HSPF model results were realized after long and collective efforts, which were hindered by

model bugs and other limitations, some researchers were not able to apply the HSPF model

successfully due to some of the following reasons:

• The model data requirement is too heavy. Data on soils and river cross-sections is often not available in most of the catchments in the basin.

• The model is “Hard-wired” and even very simple modifications are impossible to implement.

• There had not been adequate opportunities to create a strong capacity in the use of the model.

• License limitations

Obviously, the “hard-wired” types of models, which are often accompanied by strict license limitations,

have little chance of success in the Nile basin due to the huge hydrological diversity. It is

recommended that emphasis should be invested in modified or tailor-made models, which are flexible

and can be adjusted for local conditions. This should indirectly provide better opportunities for

capacity building in catchment modeling in the Nile Basin.

3.6.4) Application of SWAT Model

The SWAT has so far proved to have a good potential in modeling the flows in some catchments of

the Lake Victoria basin. The comparison of the observed and estimated long-term average water

balance in the period 1970-1974 showed a good match. Therefore, for long term simulation the model

showed that the water balance matches well.

It was noted that in the model set-up, attention should be on the classification of land use and soil

type to match the SWAT’s classification and type respectively as these affects (are sensitive) very

much the runoff and river flow estimations. In a future study, it is recommended to use sub-basin

model set-up, more distributed HRU combinations and to avoid the use of the weather simulator for

model validation where the model had poor fit.

3.6.5) Application of HMS Model

The study recommended using more rainfall-runoff data, seasonal parameterization and modeling of

the Blue Nile watershed in sub-basins of smaller areas so as to improve the HMS model results

performance in a large-scale Blue Nile watershed (254,230 km2). In this case, development of model

parameterization methodology using geographic information systems is highly recommended. The

SMA algorithm has a linear structure and may be a source of error in simulating the rainfall-runoff

process, which is a non-linear process. Therefore, the seasonal or multi-parameter approach may

improve model. Although the HMS model has automatic calibration facility, manual calibration was

adopted to determine a practical range of the 12-parameter values preserving the hydrograph shape,

minimum error in peak discharges and volumes.


The obtained results were satisfactory taking into consideration the lumped time-invariant parameters

used in the calibration and verification of the model and the model accounted for more than 90% of

the initial variance.


Bottlenecks observed during the first phase were:

I. Lack of funds to support research assistants during the initial research undertaking.

II. Delay by the participating countries to contribute data to the coordinating center as well to other researchers.

III. HSPF model bugs, which resulted to waste of time that was invested for model familiarization and test runs.

IV. Technical reports by the focal persons’ submitted late to the coordinating center.

V. Delay in responding to mails. This may be either due to the fact that the focal persons have other obligations at their working place.

3.8) The Way Ahead

A concrete proposal of the future (phase2) research activities was prepared, presented and approved

in the year 2005. With regard to the activities of the second phase of the project, earlier, discussions

suggested new research areas such as water quality modeling and conflicts resolution. However,

after discussions it was noted that before we switch to the specific applications of the rainfall-runoff

models, it is better to research further on the same models and later on components that are

important in the runoff delivery processes. Detailed discussions were done and the proposed possible

areas of research were: Water management/administration, impacts of land use and climate change

on water resources. Also, discussions for sustainability of the project and for real application in the

region, linkage of the project to the Nile Basin Initiative (NBI) Nile-net and other networks such as Nile

Basin Capacity Building Network for River Engineering (NBCBN-RE) programs was considered

important.

3.9) Acknowledgement

The coordinating center wish to thank the contributions from the participating countries’ RRM theme

researchers in the component research, academic staffs at the coordinating center for executing the

activities at the coordinating center, the FRIEND/Nile coordinating office in the UNESCO Cairo office,

The Flemish government for the financial support and the FRIEND community worldwide for sharing

with us. Moreover, thanks to the SIDA/SAREC support to the staff of the Department of Water

Resources Engineering, University of Dar es Salaam. The latter enabled conducting a field study that

had pivotal contribution to some ground truth of the Simiyu River catchment data.


3.10) References

FAO (1986). African agriculture: The next 25 years. FAO, Rome, Italy.

Fleming, M. and Neary, V. (2004) “Continuous hydrologic modeling study with the hydrologic

modeling system.” ASCE Journal of Hydrologic Engineering 9(3), 175-183.

Nash, J.E. and Sutcliffe J.V. (1970) “River flow forecasting through conceptual models, Part 1, A

discussion of principles”, Journal of Hydrology 10: 282-290.


Chapter

4

Drought and Low-Flows Analyses

4.1) Introduction

The general objective of the Drought and Low-Flows Analysis (DLFA) Component of the FRIEND-

Nile Project was to analyze daily rainfall and river flow data within the Nile basin to obtain a better

understanding of both the spatial and temporal characteristics of the low-flows and droughts within

this basin. The analytical tools/techniques and software, which are available in KU Leuven for low-

flow analysis were utilized almost exclusively in the DLFA component. However, new techniques of

analysis and relevant methodologies for drought analyses were also developed.

The specific objectives of the DLFAC are:

• To develop a database (in University of Nairobi, the coordinating institution for DLFA component) for hydro-climatic information that can be used by the researchers in the component to investigate the characteristics of droughts and low-flows in the Nile Basin.

• To conduct workshops in the form of working sessions to gain more experience in Drought and Low-Flow analyses through guidance by the Flemish resource person to:

o Use the Peak-Over-Threshold (POT) method to analyze all available data on river discharges or surface water levels on the appropriate time scale relevant in the country.

o Use aggregation methods to investigate characteristics of the parameters of the extreme-value distributions over different time periods.

o Develop probability distributions for the identified drought indices such as run lengths, deficit volumes, etc., on appropriate time scales using all the available data.

The countries which participated fully in the DLFAC activities were: Egypt, Sudan, Tanzania, and

Kenya.

It is important to note that, this component was initially not funded outside the UNESCO support.

However, in January 2003, during the 6th Steering Committee Meeting which was held in Aswan,

Egypt, the members noted a good progress had been made in this component. Therefore, they

suggested that some funding should be extended to this research component during the 2003 year.

Main activities of the DLFAC commenced in 2003 with the first DLFAC workshop (25-28 August

2003, Nairobi, Kenya). The workshop was used to define the activities and way forward for the

component. This workshop defined its activities to cover two main areas:


I. Drought Analysis, and

II. Low-Flow Analysis.

In this workshop it was stressed through various presentations that, unlike the other types of

hydrological extremes, which often have specific definitions, it is more difficult to define drought. It was

stressed in this workshop that in this component, the most important questions related to drought and

low-flows are:

• What are the drought definitions?

• What is the extent of its severity over different locations?

• What is the effect on short and long duration?

• What are the statistical methods for drought analyses?

• What are the suitable models for simulating low flow and drought events?

• What are the methods of estimation of low flows at un-gauged locations?

The workshop participants underscored the importance of having an understanding to the answers of

these questions, particularly in the Nile basin. However, the difficulties encountered in the definition of

drought are mainly due to the fact that drought is conceived better from its impacts than from its

causes. Since impacts are both region-specific and user-specific, then a universal definition is difficult

to constitute. However, a simple definition of drought is “a prolonged and abnormally dry period when

there is no enough water for the user’s normal needs”. For the purposes of the DLFA C, the following

definition for drought was adopted: <<a “natural” event (hazard) resulting from a less than normal precipitation for an extended period of time>>.

It was also noted that a prolonged and abnormally dry period may not necessarily be a drought period

so long as the user’s water needs are adequately satisfied. Thus, the key factor in a drought is the

scarcity of water for a prolonged period of time. This scarcity may be due to either inadequate rainfall

(meteorological drought), or inadequate soil moisture (agricultural drought), or low levels of water in

the rivers, lakes, reservoirs and aquifers (hydrological drought). Implicit in this definition is the

requirement of a normal range of availability or variability of water supplies, within which the user’s

needs are not adversely affected. The normal range of variability is a concept, which develops from

past experiences covering a long period. This normal range of variability is region-specific and user-

specific. Drought can also be due to adverse human factors.

In the analyses of drought as a hazard, socio-economic and climatic factors are very important.

However, for drought as a natural phenomenon, which is imbedded within the climate system, only

climatic factors are important. The most commonly used climatic factors are:

• Rainfall,

• Evapotranspiration, and

• Temperature

Other implicit factors are:


• Soil moisture, and

• Water levels

Other methods involve the analysis of the time series of the lengths (often in the number of months) of

each of the separate drought durations within a study period. The time series of such lengths is often

referred to as the runs-time-series. On the other hand, Low flow analysis, is commonly done in the

frequency domain, much in the same way as is done for flood frequency analysis. However, care

must be exercised in order to preserve the definition of the return period for the low-flows, which may

occasionally contain zero values.

4.2) Data Requirements and Methods of Analyses

The first workshop of the DLFAC attempted to define not only data types requirements, but also the

methodologies that would be utilized for analysis. In an attempt to understand the types of data

requirements in the defined work plan, a list of different types of drought and low-flow problems in the

region was drawn. Such problems were given priority based on their severity in the region on

different time scales. The drought and low-flow issues and problems are summarized in Table 4-1.

Based on this list of drought and low-flow problems and the prioritization, see Table 4-1., a number of

indicator variables were defined. These were given as follows:

• Precipitation depths,

• Evaporation depths/air temperature,

• River discharges,

• Surface water levels,

• Groundwater levels,

• Soil moisture contents,

• Concentration of DO, pollutants and salts, and

• Water temperature

• Table 4-1, Sample Inventory (in time-scales) of drought and low-flow problems in the Nile basin.

Kenya Tanzania Egypt Sudan Ethiopia

Types of drought problems related to water based activities M M Y S M

Water quantity M M Y M M

River discharge limitation:

Water abstraction limitation for domestic drinking water supply M M Y S M

Water abstraction limitation for irrigation M M S S M

Water abstraction limitation for hydropower production W W M M W

Water abstraction limitation for other industry S S S S S


Kenya Tanzania Egypt Sudan Ethiopia

Groundwater table decrease:

Limited water supply by wells for agriculture S S M S S

Limited water supply by wells for domestic water supply S S M S S

Limited water supply by wells for industry N/A N/A N/A N/A N/A

Impact on the ecosystem Y Y S Y Y

Soil moisture decrease by low precipitation and high evaporation:

Damage to agricultural crops S S S S S

Surface water level decrease:

Limited water availability for industry M M S M M

Navigation: drought problem N/A N/A D S N/A

Fish spawn and fry hampered N/A N/A S N/A N/A

Limitation of recreational opportunities M M S N/A M

Impact on the ecosystem S S S S S

Water quality: high pollution levels, low DO concentrations, temperature, salt intrusion from the sea

Human health Y Y S Y Y

Aquatic damage: fish dying Y Y Y Y Y

Aquatic damage: hyacinth growth Y Y M Y Y

Bad smell from rivers M M N/A Y M

Pollution to the ecosystem Y Y Y Y Y

Legend: N/A = Not Applicable, D=Daily, W=Weekly or about 10days, M=Monthly, S=Seasonal, and Y=Annual

Precipitation depths and evaporation depths refer to a drought cause, while the others refer more to a

drought impact or drought problem. Precipitation depths and river discharges were considered the

primary indicator variables. The Drought and Low-Flow Component of FRIEND-Nile will therefore

mainly focus on these variables, and will try to derive statistical properties for these indicator variables.

This will be done at site in a first phase of the project; while in a second phase a regionalization

analysis will be conducted. The analysis will be done at the relevant time scales, as specified in Table

4-1.

On the basis of these discussions, the data requirements in each participating country were defined

and are summarized in Table 4-2. It was agreed that the project would assist in financing data

collection phase. The country representatives were therefore requested to study data requirements

carefully and prepare the data specifications for discussion and approval during the second DLFAC

workshop that was scheduled to be held in 2004.


• Table 4-2, Data Requirements per country.

Data type Kenya Tanzania Sudan Egypt

Precipitation depths

Evaporation depths/air temperature

River discharges

Surface water levels

Groundwater levels

Soil moisture contents

Concentration of DO, pollutants and salts

Water temperature Dai

ly

daily

Dai

ly an

d m

onth

ly

mon

thly

and

annu

al

4.3) DLFAC Methodologies and Research Findings

During the DLFAC first Workshop, it was emphasized that drought intensities/severities could be

analyzed in both temporal and spatial domains in terms of runs, persistence and probabilities or some

form of combination of the three. However, low-flows could be analyzed mainly in the frequency

domain which can be extended to the spatial domain through regional analyses. It was agreed that

the main activity of the DLFAC during the first phase of the project would focus on low-flow analysis.

In this analysis all the country research persons would use a common methodology. It was also

resolved that any research person who was interested to work with drought analysis using rainfall or

evaporation data would do that at individual level and report the progress during the subsequent

workshops.

During this workshop, Dr. Patrick Willems, the Flemish DLFAC resource person, discussed the

general methodological framework used in Europe for low flow analysis. Different steps in such

framework include problem definition, stakeholder analysis, definition of indicator variables and

criteria.

He also presented different methods for statistical analysis of the indicator variables (e.g. rainfall

depths, river discharges, surface water levels). For the discharge indicator variable he explained that

a time series of total rainfall-runoff discharges can be split into its sub flows (such as the overland

flow, the subsurface flow or interflow, and the groundwater flow or base flow) using a numerical digital

filter technique. Its physical interpretation is based on the linear reservoir-modeling concept.

He noted that the statistical analysis can be done either based on long-term time series of

measurements (for discharges, water levels, pollutant concentrations, etc.), or long-term simulation

results from mathematical models. For this purpose, an extreme value analysis is needed.


In extreme value analysis the tail of a distribution describing the probability of occurrence of extreme

events is analyzed and modeled by a separate distribution. The considered extremes might exist in

extreme rainfall intensities, storm volumes, water levels, discharges, water quality parameters, etc.

Examples in extreme value analysis were presented.

POT selection based on an independency criterion was also presented. This was based on two

methods. The first method considers two successive discharge peaks to be largely independent when

the smallest discharge in between the two peaks reaches almost the base flow value. The second

makes use of the ‘recession constant’ of overland flow and/or interflow (for separation of independent

quick flow events; used for POT selection in flood frequency analysis), or the recession constant of

base flow (for separation of independent slow flow events; used for the selection of independent

discharge minima or low flows). A practical session on sub flow filtering and POT selection and

extreme value analysis was also conducted.

The country representatives were, therefore, requested to practice using the discussed

methodologies as implemented in software handed by Dr. Willems. This was done using the data that

they already had collected in preparation for further discussions on the methodology during the

second DLFAC workshop t held in 2004.

The data requirements and acquisition opportunities were therefore extensively discussed during the

2nd DLFAC workshop which was held in Alexandria (Egypt) during the period 19-21 June 2004. As a

consequence to these discussions, a general agreement was reached regarding the type of data that

was going to be collected within each of the participating countries.

UNESCO-Cairo office prepared contracts for data collection soon after the end of the 2nd DLFAC

workshop in June 2004. Follow is a summary of these data for each participating country.

• Table 4-3, details for the catchment chosen as case studies for Kenya.

Site 1 Site 2 Site 3

River name Nzoia river Gucha-Migori river Nyando river

Station name At Webuye

Station id 1DA02

Series unit m3/s m3/s M3/s

Period Jan 1960 - Dec 1995 1/5/1969 – 30/3/1994 1969 - 1994

Time step daily Daily Daily


Kenya

A contract was prepared for Mr. Julius N. Kabubi, who is an employee of the Kenya Meteorological

Services, to complete the task in Kenya. He provided the following data to the DLFAC coordinator:

I. Daily average flow discharges covering about 30 years of five stations in the catchment of Lake Victoria in Kenya.

II. Daily rainfall data covering about 30 years of 150 stations in the catchment of Lake Victoria in Kenya.

A review of literature including the catchment characteristics of the studied rivers such as catchment

area, main stream slope and length, and means annual evaporation. Table 4-3 gives details for the

catchments that were chosen as a case study for Kenya.

Egypt:

A contract was prepared for Dr Ahmed Hassan Fahmi, an employee of the Water Resources

Research Institute of the National Water Research Center in Egypt, to carry over this task in Egypt. Dr

Hassan is also the DLFAC research focal person for Egypt. He provided the following data to the

DLFAC coordinator:

I. Ten-day average flow discharges covering 24-40 years of three stations in the catchment of River Sobat.

II. Monthly flow discharges covering 130 years of one station on the main stream of the River Nile at Aswan, Egypt.

III. Annual rainfall data covering 28-30 years of three stations in the catchment of River Sobat.

IV. A literature review and a summary of catchment characteristics of the studied rivers (e.g., catchment area, main stream slope and length, soil type, land use, means annual evaporation).

Table 4-4 shows details for the catchment that was chosen as a case study for Egypt.

• Table 4-4, details for the catchment chosen as a case study for Egypt.

Site 1

River name Nile River

Station name Aswan

Station id Up stream lake Nasser

Unit series Bm3/month

Period 1830 – 2000

Time step Monthly


• Table 4-5, details for the catchment chosen as a case study for Sudan.

Site 1

River name Blue Nile river

Station name Eddeim

Station id

Unit series Mm3/d

Period 1964 - 1996

Time step Daily

Sudan

Dr Muna M. Mirghan, who was an employee of the UNESCO Chair for Water Resources in Khartoum

– Sudan, was responsible for accomplishing the tasks related to this research component in Sudan.

She is also the DLFAC research focal person for Sudan. She provided the following data to the

DLFAC coordinator:

I. Daily flow discharges covering 31-44 years of six stations in the catchment of the river Nile in Sudan.

II. Daily Surface water levels covering 38 years of one station in the catchment of the river Nile in Sudan.

III. Monthly rainfall and evaporation data covering 20-30 years of nine stations in the catchment of the river Nile in Sudan.

IV. A review of literature including the catchment characteristics of the studied rivers such as catchment area, main stream slope and length, soil type, land use, and mean annual evaporation.

Table 4-5 shows details for the catchment that was chosen as a case study for Sudan.

Tanzania

A contract was prepared for Dr Raymond Mngodo, an employee of the Ministry of Water Resources

in Tanzania, to accomplish tasks related to this research component in Tanzania. Dr Mngodo is also

the DLFAC research focal person for Tanzania. He provided the following data and information to the

DLFAC coordinator:

I. Daily flow discharges covering 16-53 years of three stations in the Simiyu catchment of Lake Victoria.

II. Daily and monthly rainfall data covering 30-50 years of ten stations in the Simiyu catchment of Lake Victoria.

III. A review of literature including the catchment characteristics of the studied rivers such as catchment area, main stream slope and length, soil type, land use, and mean annual evaporation.

Table 4-6 shows details for the catchment that was chosen as a case study for Tanzania.


• Table 4-6, details for the catchment chosen as a case study for Tanzania.

Site 1

River name Simiyu River

Station name Ndagalu

Station id 5D1

Unit series m3/d

Period 1970 – 1996

Time step Daily

At the same time, UNESCO-Cairo office in collaboration with the DLFAC coordinator, made

arrangements for each of the active participants (Tanzania, Sudan, Egypt and Kenya) to receive a

computer, a flash memory stick(s) and a printer according to his/her specifications to assist him/her in

enhancing their work in the DLFAC. The coordinator is glad to report that all the research focal

persons in the DLFAC received the requested computer requirements at different times during the

year 2004.

Most of the time during the 2nd DLFAC Workshop was reserved for working sessions in which the

participants practiced the Peak-Over-Threshold (POT) and the frequency analysis and the frequency-

model identification software which were provided and guided by the Flemish DLFAC resource

person. At the end of the workshop, the participants showed a good understanding of the use of the

software and were able to present tentative results. It was agreed that the participants would continue

practicing with the software in order to be able to produce final results (after the acquisition of all the

data) during the 3rd DLFAC workshop that was scheduled later in the year 2004. The 3rd DLFAC

Workshop was held in Nairobi (Kenya) at the IGAD Climate Prediction and Applications Center

(ICPAC) during the period 23-26 November 2004. This workshop was dedicated mainly to finalize the

Low-Flow analysis and to prepare for the FRIEND-NILE (FN) Conference which was scheduled for

November 2005. The country presentations on the frequency analyses software applications showed

that the participants were fully acquainted with the software and that the results which had been

obtained so far were very useful and could therefore be synthesized into journal or the FN

Conference papers. It was agreed that these DLFAC results were to be summarized into several

Conference/journal papers.

The researchers eventually settled to prepare the following papers:

• A journal paper entitled “Analyses of Annual Droughts in Kenya Using an Objective Annual Rainfall Drought Index” to be coordinated by Kenya,

• A conference paper entitled “Low flow frequency Analysis Based on Flow Filtering and Independent Period Selection for some Selected catchments in the Lake Victoria basin” to be coordinated jointly by Kenya and Tanzania,

• Two conference papers to be coordinated by Sudan, namely:

I. QDF Relationships for Low Flow Return Period Prediction;


II. Statistical Analysis of Dry Periods in Seasonal Rivers; and

• A conference paper entitled “Analysis of the Return Periods of Low Flow Hazards in Egypt and Sudan” to be coordinated by Egypt.

The fourth and final workshop of the DLFAC was therefore scheduled for June/July 2005 to finalize

these conference papers. This 4th (final) DLFAC workshop was eventually held in Khartoum (Sudan)

at the UNESCO-Chair, during the period 27-30 July 2005. The main activity in this workshop was to

finalize the conference papers.

All produced articles, except the journal papers which have already been submitted to the Hydrologic

Sciences Journal for publication, were eventually presented in the FN Conference. The successful

conference was held in Sharm El-Sheik, Egypt, during the period 12-15 November 2005.

4.4) Summaries of the DLFAC Research Activities and Methodologies

In this section a summary of each of the DLFAC research activity is given.

4.4.1) QDF Relationships for Low Flow Return Period Prediction

4.4.1.1) Introduction

For sustainable development of the Nile Basin, an integrated water management approach

encompassing environmental considerations such as drought impacts is crucial. Drought has

extensively affected the region and left clear impacts on the planned development. This research

activity contributes to the assessment of drought impact on the Nile low flows using low flow –

duration – frequency (QDF) relationships and curves.

QDF modeling uses a multi-duration and multi-frequency description of observed low flows. Low flows

are analyzed at one key station on the Blue Nile in Sudan, and at a station on river Nzoia upstream of

Lake Victoria in Kenya.

Low-flow frequency analysis is performed by first selecting probability distributions to describe low

flow minima for different aggregation periods in the range from 1 day to 2 years. Relationships

between the parameters of these distributions and the aggregation periods are thereafter analyzed

and parameterized. QDF relations are then established, which can be used later as the basis for low

flow regionalization in the Nile Basin.


• Table 4-7, Calibration result of distribution parameters for Eddeim low flows.

Aggregation level [days] Threshold rank t β [MCM] qt [MCM]

1 16 0.075 0.18

10 15 0.064 0.15

30 14 0.058 0.13

120 12 0.030 0.094

180 10 0.011 0.070

240 7 0.011 0.044

365 3 0.0008 0.012

730 3 0.0005 0.011

4.4.1.2) Results

The distribution parameters (for Eddeim) resulting from the calibration are presented in Table 4-7. For

low flow minima, the return period T is to be calculated on the basis of the probability distribution

function F(q) and adjusted according to the equation below:

)exp(

)exp(

)(1][ 1

1

β−

β−

== −

−

q

q

tn

qFtnyearsT

t

(3)

where “exp” denotes the exponential function, β is the probability distribution parameter and t is the

rank of the minima which are considered below the threshold qt during n years. Figure 4-1shows the

result of the return period calculation of 1 day aggregation period. In this figure, low flow

extrapolations are made towards 1000 years.


0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.1 1 10 100 1000 10000Aggregation level [days]

Slo

pe ⎠

[MC

M]

slope

slope, fit

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18

0.2


Thre

shol

d va

lue

xt [M

CM

threshold value, inputthreshold value, calculated

threshold value, fit

0

2

4

6

8

10

12

14

16

18


Thre

shol

d ra

nk

threshold rankthreshold rank, fit

0

1

2

3

4

5

6

7

0.1 1 10 100 1000


Dis

char

ge [M

CM

]

empirical data

calibrated distribution

• Figure 4-1, Return period curve for Eddeim 1 day low flows.

4.4.1.3) Relationships between Low-Flow Distribution Parameters and the Aggregation Period

The calibration results of Table 4-7 are plotted in Figure 4-2 versus the aggregation period.

• Figure 4-2, Relationship between the distribution parameters β qt and the aggregation period D for Eddeim low flows.

In Figure 4-3, the q-D relationships are shown for return periods of 5, 10, 30, 100 and 1000 years. In

this figure, a comparison is also made with the empirical values for return periods of 5, 10 and 30

years. This shows that the empirical data appears less smooth due to the randomness in the

occurrence of extreme low flow conditions. Given this randomness, the empirical data are reasonably

well in agreement with the theoretical curves. Similar approach has been applied for the daily flow

data in the Nzoia River at 1DD01 station. The final QDF calibration results are shown in Figure 4-4.


1

10

100

0 50 100 150 200 250 300

Aggregation level [days]D

isch

arge

[MC

M]

5 years10 years30 yearsT=5T=10T=30T=100T=1000

• Figure 4-3, QDF plots for the Blue Nile low flows at Eddeim.

1

10

100

0 50 100 150 200 250 300

Aggregation level [days]

Dis

char

ge [c

umec

s]

5 years10 years30 yearsT=5T=10T=30T=100T=1000

• Figure 4-4, QDF plot for the river Nzoia low flows at 1DD01 station.

4.4.1.4) Conclusions

This research activity has provided first experimental work of its type to assess the drought impacts

on river flow along the Nile basin. This has been done for two selected flow stations, one in the arid

lower Nile region for the Blue Nile at Eddeim and one in the upper Nile region for Nzoia River in the

Lake Victoria sub-basin. The adopted methodology was developed by the DLFA Flemish counterpart

and fully discussed and explained in Willems, 2004a and 2004b.

The QDF relationships or curves can be used to estimate the availability of water for cumulative

volumes during specific time intervals (aggregation periods) and for different recurrence intervals or

return periods. For most water use applications (agricultural irrigation, domestic water supply,

hydropower supply, etc.) threshold levels are known below which the application runs out of water.

Depending on the storage capacity available (e.g. in reservoirs), the residence time of the water

varies from application to application. This requires water volumes to be analyzed for different time

intervals. For each application, given the threshold value and the time interval, the return period of

water shortage can be calculated.


Considering the Blue Nile system in Sudan, the QDF results produced in Figure 4-3 will be an

important tool for planning of the Roseris dam operation. They can be used to indicate the risk due to

not satisfying the required storage volume of the dam, under current conditions as well as potential

future conditions after changes of the dam operation.

4.4.2) Low Flow Analysis Using Filter Generated Series for Lake Victoria Basin


The Nzoia is one of the main Kenyan rivers, which drain into Lake Victoria. It is also perennial with a

total drainage area of about 12,800 Km2. The lower parts of the basins of this river, a few kilometers

just before it enters the lake is characterized by frequent episodes of floods, which often cover

expansive inhabited areas especially when the upper areas of the basin receive intense rainfall

amounts for significant period of time. These flood periods are often punctuated with long periods of

low flow or drought.

In water engineering applications, extremes are often analyzed for time series. Those extremes have

to be extracted first as peak-over-threshold values or annual maxima/minima from the time series in a

preliminary study. For the former, they can take the form of instantaneous, aggregated or averaged

values in fixed time duration or cumulative values in events such as storm volumes. As the extremes

in an extreme-value-analyses have to be independent. An “independency criterion” is used in the

extraction process. River flood applications consider consecutive peak floods as independent if the

intervening time exceeds a critical time and if an intervening discharge drops below a critical flow.

Such independency criterion influences the number of extremes and also the interpretation of the

return period of an extreme event. As a well-considered interpretation is needed in most applications,

the independency criterion is a subjective choice. It does often not totally agree with “statistical

independency”, which is meant in the theory. The subjective criterion, however, reaches very often a

large physical independency. This large independency then guarantees the existence of an extreme

value distribution or GPD distribution. When large return periods are estimated in the application, the

influence of the choice of the independency criterion becomes less important.

In this research activity, a new criterion ( Willems, 2004b), which identifies all independent low flow

values, is utilized. This criterion employs an objective method of “Peak-Over-Threshold” (POT)

method to first identify both the independent maximum flow data points in a given set of input flow

data. The POT maxima are subsequently used to locate the independent flow minima. The method

has been found to work successfully on perennial rivers such as the Nzoia River within the Lake

Victoria basin.

The extracted flow minima data is then analyzed for the tail behavior by fitting it to the three basic

Generalized Pareto Distributions (GPDs): the Exponential, Pareto and Weibull distributions. A

threshold value in the low flow series is identified as the value corresponding to the minimum mean

squared error (MSE) of each of the distributions. With this threshold, a technique based on the


regression in the quartile-quartile plots (QQR method) is used to discriminate between the

distributions after applying it on the transformed data (1/Q). All the three river gauging stations on the

Nzoia River exhibited a normal tail case and a high correlation in the Q-Q plots, which supports the

use of the exponential distribution for this catchment.

4.4.2.2) Results

The new criterion for isolating independent or nearly independent low flow series after locating the

independent high flow periods is undoubtedly an objective way of attaining a statistical requirement of

independency and randomness in time series analysis. The lengthening of the data series particularly

for the discharges is of great importance for extreme value forecasts. However, it was observed that,

for rivers with substantially high flows with well define climatic seasons, the number of independent

values of low flow tends to converge to that annual minimum series (number of years considered).

This was the case for the two down stream RGS for the catchment. However, for the independent

high flow values, the criterion is able to increase the number of high flow values.

It was observed that, the parameters corresponding to the base flow and interflow, as well as the

chosen length of independent period, plays a key role in apportioning the various sub-components of

the input series. More work is needed in improving the selection of this parameter probably from the

statistical characteristics of the input series. A goodness of fit criteria should also be included to test

the suitability of a distribution ounces isolated as the best by the Q-Q plots in the ECQ code.

For the Nzoia catchment, it was found that they produce normal tailed distributions for all the three

stations shown on Table 4-3. This indicates that, exponential distribution is more suited for low flow

studies for this catchment. However, more trials with other catchments within the basin need to be

carried out before any generalization can be made.

Standardized curves for the return period discharge were performed for the three stations, which

showed a good low flow fit by the flow at each RGS with the upstream catchment areas. The results

for these two stations are encouraging and perhaps mark the first steps in Regionalization studies.

Clearly shown from the Q-T forecasts for all the three river gauging stations, the Nzoia River is

dependable upon the low flow regulation and has some room for more water resources development

projects. It is more meaningful to consider the dry spells for non perennial rivers to get an idea on the

uncertainty of the length dry periods. This procedure of flow filtering undoubtedly paves the way of

being a very important tool for decomposition of a rainfall input series into its sub-components for use

in a Rainfall–Runoff modeling.

4.4.3) Statistical Analysis of Dry Periods in Seasonal Rivers


As drought impact indicators, discharges at two stations on the Nile system inside Sudan are

considered for analysis. The main objective of the current research activity is to infer low flow


frequency distributions using independent low flow discharges and quantile plots; as a first step to the

regionalization of low flow frequency estimates that ultimately contribute to sustainable management

of the Nile water resources.

The extraction of independent low flow discharges is done based on numerical techniques used for

baseflow filtering and by splitting such discharge series into independent low flow events. The latter is

done based on a pre-specified independency criterion. Low flow frequency distributions are obtained

at the selected stations and presented as return-period curves and low flow quantiles.

The fitting of extreme-value distributions to low flows was done based on methodologies used for

flood frequency analysis, after transforming the discharges Q by 1/Q. Being seasonal, low-flows at

the two selected stations contain zeros. The frequency of zero low flows therefore had to be taken

into account in the return period calculation of the low flows. Frequency analysis is alternatively also

applied to the dry spells as well as to aggregated dry period flow volumes.

0

50

100

150

200

250

300


Dry

Spe

ll [d

ays]

observations


optimal threshold

50

75

100

125

150

175

200

225

250

1 10 100


Dry

spe

lls (d

ays)

empirical data


• Figure 4-5, Exponential Q-Q plot indicating a normal tail exponential distribution for dry spells and he Return period curve of dry spells at Kubur station.

4.4.3.2) Study Cases

Two cases representing seasonal rivers typical to the arid Nile region in Sudan are considered. Daily

discharge series measured at Kubur and Hileiw stations at the outlet of two Nile sub-basins on

tributaries of River Atbara are analyzed for the purpose of this research activity. At Hileiw station, this

series is available during the period 1966-1992 (26 years). For Kubur station, this period equals 1966-

2002.

4.4.3.3) Return Period Curves for Dry Spells

Number of days of zero flow (dry spell) is calculated for each of the 36 years and is considered in the

frequency analysis. An exponential distribution is calibrated to the dry spells longer than 122 days. In

the exponential Q-Q plot, the exponential distribution indeed appears straight. The result shown in

Figure 4-5 indicates that at Kubur station a dry period of more than six months is taking place once

every 50 years.


-10

-9

-8

-7

-6

-5

-4

-3

-2

-1


ln(1

/ D

isch

arge

[MC

M])

observations


optimal threshold

censored POT values

0

500

1000

1500

2000

2500

3000

1 10 100


Dis

char

ge [M

CM

]

empirical data


• Figure 4-6, : Pareto Q-Q plot indicating a heavy tail distribution for the 1/Q dry period aggregated flows and the Return period curve of 1/Q dry period aggregated flows at Kubur station.

-11

-10

-9

-8

-7

-6

-5

-4

-3


ln(1

/ D

isch

arge

[MC

M])

observations


optimal threshold

0

1000

2000

3000

4000

5000

6000

1 10 100


Dis

char

ge [M

CM

]empirical data


• Figure 4-7, Pareto Q-Q plot indicating a heavy tail distribution for the 1/Q dry period aggregated flows and the Return period curve of 1/Q dry period aggregated flows at Hileiw station.

4.4.3.4) Return Period Curves for Dry Period Aggregated Low-Flows

Low flow aggregates were investigated as cumulative low flow volumes during the low flow period

(the 3rd decade of October and continues till the next floods − end of June). They follow a heavy tail

extreme value distribution for 1/Q; calibrated by regression in the Pareto Q-Q plot of Figure 4-6 and

Figure 4-7. The corresponding return period curves are also shown on these figures. Such frequency

curves can be used in water management, for instance to investigate available storage volumes. An

example of such application is worked out hereafter for the Khashm el Girba dam.


0

50

100

150

200

MAY 1

2

3

JUN

1

2

3

JUL

1

2

3

AUG 1

2

3

SEP 1

2

3

OCT 1

2

3

NOV 1

2

3

DEC 1

2

3

JAN

1

2

3

FEB 1

2

3

MAR 1

2

3

APR 1

2

3

Dis

char

ge (M

m3/

d)

0,0

0,5

1,0

1,5

2,0

2,5

3,0

3,5

4,0

Sedi

men

t Yie

ld (1

06 T/d

ay)

Kubur + Wad el Heleiw) Sediment Yield

• Figure 4-8, River Atbara average hydrograph.

Khashm El Girba ReservoirU/S Water Level (10-days average)

440

445

450

455

460

465

470

475

480

Jul I II III Aug I II III Sep I II III Oct I II III Nov I II III Dec I II III Jan I II III Feb I II III Mar I II III Apr I II III May I II III Jun I II III

Period (10-days)

Wat

er L

evel

(m)

Actual operation

• Figure 4-9, Operation rules of the Khash el Girba Dam upstream River Atbara.

The Khashm el Girba dam is a multipurpose dam important for the downstream development. Due to

the high sediment load in the river, the effective developed flow is that occurring after the flood peak

(Figure 4-8). The water year for Khashm el Girba dam starts on the 1st of July and ends at the end of

June of the next year, and has four operational periods, namely: the flood period, the filling period,

keeping filling till 473m level, and finally the emptying period (Figure 4-9). The 4th period, the emptying

period, depends on the incoming flow, and normally starts in the 3rd decade of October and continues

till the next floods. Therefore the incoming flow volume in this period is subject to low flow aggregate

analysis. Sufficient volume is required to irrigate 168,000 hectares with 1960 MCM annually, and at

the same time to allow for a storage level satisfying the turbine head for power generation, and to

allow for the water supply intake at the Showak pump station upstream of the reservoir. Assuming

reservoir storage of 550 MCM at level 473 m, a minimum inflow of 1410 MCM is required during the

low flow period to satisfy the current demand. According to the return period curves of the river

tributaries, this demand may not be satisfied once in 5 years. Given the planned upper Atbara

development plans, the existing demand may not be satisfied with this return period.



Low flows in the two branches of river Atbara are analyzed for frequency distribution. Being seasonal,

the annual minima at the two selected stations contain zeros. Frequency analysis is alternatively

applied to zero and non-zero low flows, to the length of the dry periods (the dry spells), as well as to

aggregated flow values during low flow periods.

Analysis of the dry spells proved that dry periods longer than 6 months occur on average with

recurrence intervals of 50 years. A more important indicator in water management is the volume

rather than time span of the low flow. For this reason, also aggregated dry period flow volumes have

been analyzed and return period curves have been prepared. It is shown that once in five years the

current demand from the river flow under study is not satisfied.

The reader has to take into account that these values may be subject to errors and uncertainties,(e.g.

due to the limited reliability of the rating curves applied on the basis of the daily flow series).

4.4.4) Analyses of Annual Droughts in Kenya Using an Objective Annual Rainfall

Drought Index


A drought hazard is difficult to define and, consequently, difficult to manage. Droughts are inevitable

and sometimes are essential regulators of climate-driven environments. Drought-hazards often

become disasters and enhance poverty and catalyze strong feedbacks with drought-vulnerabilities.

Inadequate rainfall plays a key role in the development of droughts. In this study, a meteorological

annual drought index is developed. Monthly rainfall data from 26 meteorological stations in Kenya is

used in the study. The performance of the General Extreme Value (GEV) distribution is tested against

that of the Normal distribution using the Akaike Information Criterion (AIC) and also the L-Moments

goodness-of-fit tests. The GEV gives a better fit to the log-transformed drought indices than the

Normal distribution. The model estimates show that strong droughts usually cover smaller areas in

the wet regions while weak droughts cover large areas in the dry regions.

4.4.4.2) Data used

The study utilizes monthly rainfall data for 26 meteorological stations in Kenya for the period 1950-

2002. This data was obtained from the Drought Monitoring Centre in Nairobi (DMCN). Within this

period, all except three stations had data for more than 30 years. The majority of the stations (19) had

data for more than 40 years. Figure 4-10 gives a map of Kenya showing the location of these rainfall

statins.


33 34 35 36 37 38 39 40 41 42

Longitude (degrees)

-5

-4

-3

-2

-1

0

1

2

3

4

5

6

L O D W A R

K A K A M E G AK ISU M U

K ISII N A R O K

E L D O R E T

M O Y A L E

M A K IN D U

V O I M A L IN D I

M O M B A SA

L A M U

N Y A H U R U R UG A R ISSA

M A R SA B IT

W A JIR

M A N D E R A

M E R U

E M B UN Y E R I

JK IAW IL SO ND A G O R E T T I

N A K U R UK E R IC H O

K IT A L E

ETHIO PIA

U G AN D A

SUD AN

K EN Y A

TANZ AN IA

Lake V ictoria

IndianO cean

Location (on a m ap of K enya) of the rainfall stations that w ere used in the study

• Figure 4-10, Map of Kenya showing the location of the rainfall stations which were used in the study.

Dagoretti Corner in Nairobi - Central parts of Kenya

0

200

400

600

800

1000

1200

1400

1600

1800

1955

1957

1959

1961

1963

1965

1967

1969

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

Years

Ann

ual r

ainf

all

0.00

0.50

1.00

1.50

2.00

2.50

3.00

Ann

ual D

roug

ht In

dice

s

Annual Rainfall (mm)Drought Indices

Lodwar in North-western Kenya

0

100

200

300

400

500

600

700

1950

1952

1954

1956

1958

1960

1962

1964

1966

1968

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002 Years

Ann

ual r

ainf

all

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

Ann

ual D

roug

ht In

dice

sAnnual Rainfall (mm)Drought Indices

• Figure 4-11, Distribution of the Annual Drought Index in Comparison to the Distribution of the Annual Rainfall in two selected locations in Keny

The annual drought index is defined herein as a power function of the absolute sum of the normalized

rainfall deficits which fall below a pre-specified threshold value for every month of the year. The

annual value of the power factor is given by the largest relative run-length of the deficit-months in a

given year, while the multiplier coefficient in the power function is given by the fraction of the deficit-

months in the year. Mathematically, the annual drought index Di for year i is expressed as:


NiZADiB

jjiii ,.....,2,1

12

1, =⎥⎦

⎤⎢⎣

⎡= ∑

=

(4)

where N represents the data length in years, the subscript j denotes the month of year i and jiZ , is

the magnitude of the rainfall deficit in month j of year i; Ai is the fraction of the deficit-months in year i,

and Bi is the largest relative run-length of the deficit-months in year i. In equation (4), Zi,j is defined

as:

⎪⎩

⎪⎨

⎧−

>=

otherwisePX

PXZ

j

jji

jij

ji

σ

%55,

%55,

,

0 (5)

where jiX , is the rainfall in month j of year i in the given station; %55, jiP is the annual monthly 55%

percentile of the rainfall which has an annual monthly standard deviation σj for the given month j in

year i. Large values of the drought index D indicate intensified annual drought conditions while low

values of D indicate mild or no annual drought conditions. Figure 4-11 shows the distribution of the

annual drought index in comparison to the distribution of the annual rainfall in two randomly selected

rainfall stations from the study cases.

4.4.4.3) Results and Discussions

Table 4-8 gives a summary of the Log-GEV and Log-Normal distributions for the selected study-

stations. The log-GEV scores best in 23 (bolded in the table) out of the total of 26 cases which

accounts for 88.5% of the log-GEV success. Interestingly, the Log-GEV performs almost as well as

the Lognormal(3) in the only three exceptional cases.

• Table 4-8, The AIC Estimates for the Log-GEV and Log-Normal Distributions for the selected Study-Stations.

STATION AIC-GEV AIC-NORMAL STATION AIC-GEV AIC-NORMAL

LODWAR 19.58171 14.10155 GARISSA -12.5477 -16.7133

KAKAMEGA -25.6632 -29.9772 MARSABIT -20.7338 -17.7445

KISUMU -24.003 -27.0423 WAJIR 2.413204 0.831579

KISII -19.7338 -29.6639 MANDERA 22.94262 21.16908

NAROK -28.5916 -29.4673 MERU -12.7294 -13.9342

ELDORET -22.9911 -25.6039 EMBU -23.4947 -28.7687

MOYALE -32.1101 -30.9692 NYERI -28.085 -31.9689

MAKINDU -12.0014 -14.2658 JKIA -13.3662 -17.4687

VOI -13.339 -14.706 WILSON -15.8346 -16.3397

MALINDI -21.1718 -23.4807 DAGORETTI -24.6417 -20.0923

MOMBASA -15.1236 -16.6499 NAKURU -25.9924 -26.0212


STATION AIC-GEV AIC-NORMAL STATION AIC-GEV AIC-NORMAL

LAMU -20.0719 -22.4144 KERICHO -19.7137 -22.5501

NYAHURURU -26.7439 -29.2918 KITALE -13.2372 -14.9948

4.4.4.3.1) L-Moments Ratio Goodness-Fit-Tests

Figure 4-12 shows the (τ3, τ4) scatter for the sample dimensionless annual drought indices, as well as

those for the fitted Lognormal(3) and Log-GEV distributions. The L-moments plot shows that the Log-

GEV distribution approximates the sample estimates better (R2=98%) than the Lognormal(3) which

has R2=94.5%.

On the basis of the good performance of the log-GEV in modeling the drought indices in Kenya, the

distribution is therefore used hereafter to estimate drought magnitudes for different return periods.

4.4.4.3.2) Distribution of the Annual Droughts of Different Return Periods

The Log-GEV distribution was used to estimate the drought magnitudes for return periods of 50, 200

and 500 years in the stations used in this study. These return periods were chosen arbitrarily. The

estimates Ty of the log-transformed drought indices for different return periods (T) were estimated

using equation:

μα ˆ11ln1ˆˆ

+⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

⎥⎦

⎤⎢⎣

⎡⎟⎠⎞

⎜⎝⎛ −−−=

k

T Tky (6)

Values of the untransformed drought estimates TyT eD ˆˆ = were estimated for the chosen return

periods in all the other rainfall stations.

The spatial distribution for the estimated magnitudes for each of the chosen return periods is shown

with a corresponding map on Figure 4-13. The patterns for the distribution of the drought estimates for

different return periods show that the drought intensities increase almost logarithmically with increase

in return period and are generally strongest in the wet regions central districts) of the country. The

spatial gradients of the estimates are also large in these regions, implying that strong droughts cover

smaller areas in these regions. However, the weakest annual droughts and gradients are observed in

the dry and semi-arid northern and eastern parts of the country. This may be attributed to the low

variation in the annual rainfall in the dry and semi-arid northern and eastern parts of Kenya and the

accountability of Di on the deficit in rainfall from its average values. The spatial gradients of the annual

droughts in these parts are near zero, implying that weak droughts of almost the same magnitude

cover large regions.


L-Moments for Annual Drought Indices

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

-0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7L-Skewness

L-Ku

rtosi

s

Sample L-Moments

Log-GEV (R^2=98%)

Log-normal(3) (R^2=94.5%)

• Figure 4-12, The sample L-moment and the GEV distribution L-moments for the Annual Drought Indices data in Kenya.

33 34 35 36 37 38 39 40 41 42-5

-4

-3

-2

-1

0

1

2

3

4

5

6

RETURN PERIOD=50 YEARS

33 34 35 36 37 38 39 40 41 42-5

-4

-3

-2

-1

0

1

2

3

4

5

6


33 34 35 36 37 38 39 40 41 42-5

-4

-3

-2

-1

0

1

2

3

4

5

6


• Figure 4-13, Distribution of annual drought indices corresponding to the 50, 200 and 500 year GEV return periods.


Droughts are common in Africa, although some countries are more vulnerable to drought hazards

than others. Unlike other types of hazards, droughts are more difficult to define and therefore difficult

to manage. They become disasters when the human induced environments become highly

vulnerable to the risks of the drought hazards. Drought disasters enhance poverty and act as drought-

vulnerability catalysts.

In Kenya, inadequate rainfall plays a key role in the development of drought. Studies of the

characteristics of drought in the country are often inhibited by the lack of practical drought indices. In

this study, a meteorological annual drought index is developed. Monthly rainfall data from 26

meteorological stations in Kenya is used in the study. Large values of the drought indices were found

to indicate intensified annual drought conditions while low values indicate mild or no annual drought

conditions. The developed index succeeded to reflect the variability in the drought severity and

temporal distribution in the analyzed period of the used stations. Results show that higher mean

annual values of the computed annual drought indices were prominent in the areas of higher mean

annual rainfall and vise versa. All the time-series of the annual drought indices showed no significant

serial dependence. The study compares the adequacy of the two commonly used 3-parameter


extreme-analysis distributions, namely, the GEV and the Normal probability distributions. The GEV

distribution fits the logarithmically transformed annual drought indices better that the Normal

distribution. The patterns for the distribution of the drought estimates for different return periods show

that drought intensities are strongest in the wet regions of the country while the weakest intensities

occur in the dry and semi-arid regions of the country.

Aswan Station

Dongola Station

• Figure 4-14, Location Map of the Selected Sites.

4.4.5) Analysis of the Return Periods of Low Flow Hazards in Egypt and Sudan


This research focuses on the flow downstream the river Nile during low flow periods using a

probabilistic approach to assess risks due to hydrologic droughts and low flows for Egypt and Sudan.

Discussion starts with the analysis of low-flow generating mechanisms operating in natural conditions

and the description of anthropogenic factors, which directly or indirectly affect low flows. From this

analysis, the river low flow discharge was identified as the main indicator variable. It also became

clear that this indicator variable has to be considered at different time scales (e.g. seasonal, annual).

Low flow extreme value distributions and low flow statistics were calculated at several gauged sites.

The low flow distributions and statistics were applied to thresholds for drought cause and drought

impact criteria in addition to the calculation of return periods for different low flow periods. This has

been done for the stream flow gauging stations at Aswan and Dongola (see Figure 4-14). The data

records comprise more than 130 years of monthly mean flow data at Aswan and about 15 years of

daily mean flow data at Dongola. The main outcome is an estimation of the return period or


0

25

50

75

100

125

150

1900-01 1920-21 1940-41 1960-61 1980-81 2000-1

Number of Time Steps (years)

Nat

ural

Flo

w (B

CM

)

Aswan Natural FlowDongola Natural Flow

recurrence interval in which the mean flow is lower than the water demand for Egypt and Sudan. The

time scales considered for the low flow extreme value analysis assume both the natural situation

without reservoir storage 1 year annual minima and 5 years residence time of the water. The latter

residence time reflects the potential combined effect of water storage in reservoirs within the countries

considered. It is concluded that Egypt and Sudan will run out of water with a return period of 40 years

in case the reservoirs have a combined mean residence time of 5 years, and with a return period of

10 years in case of no dams for water storage.

4.4.5.2) Data Availability and Method of Analysis

For the purpose of low flow frequency analysis, two methods of flow sampling are of relevance: the

use of annual series or the use of (nearly independent) low flow extremes extracted from the series

(e.g. partial duration series approach). The annual series can consist of the mean annual flow values,

the minimum monthly flow values, the monthly flow value from one selected month, or the mean

seasonal flow value for one selected season, etc.; in general one value per year. One of the aspects

in favor of this series is the reasonable assumption that the data values in the series are not serially

correlated, i.e., successive values are independent. This property is an important prerequisite for the

subsequent statistical treatment of data.

• Figure 4-15, Natural Flow Series of Mean Annual Values at the Selected Stations.


When minimum flow values are selected for the annual series, then this may have the disadvantage that the second or third, etc, lowest events in a particular year may be lower than the minimum event in another year and yet they are totally disregarded. In this study, mean monthly discharges were considered at Aswan and Dongola stations with appropriate record lengths for about 132 and 15 years, respectively.

Figure 4-15 shows the mean annual flow series derived from these mean monthly series from 1900

to 2001. It is shown that the estimated flow in the two stations has almost the same values. As a

result, to assess the low flow frequencies and related risks for Egypt and Sudan, the analysis will

primarily be based on Aswan station due to its long data series (132 years).

Annual series were calculated from the full mean monthly flow series and this for the annual mean

flows as well as for the mean monthly flows during the different months and the 4 climatic seasons of

the year. The monthly flow series were also aggregated (averaged) over different time lengths

(aggregation levels) using the Moving Average approach. This aggregation allows the availability of

flow to be investigated under the condition of water storage (in the reservoirs upstream of the dams),

which increases the residence time of the water and reduces the low flow risk. Along the downstream

Nile in Egypt and Sudan, different reservoirs are installed, the effect of which was eliminated in the

natural flow series. To calculate the real present low flow frequencies and risks, the storage effect

needs to be incorporated, requiring the low flow frequency analysis to be carried out for low flows

aggregated during time period lengths equal to the combined residence time of all reservoirs. For this

study, a comparison is made between the annual minima (no aggregation, representing the natural

situation), and the flow minima after use of an aggregation level of 5 years.


This research focused on the application of methods for low flow frequency estimation of stream-flow

in the downstream Nile area based on two stations in the Nile River. It is concluded that, Egypt will run

out of water once in approximately 10 years (as long term average or return period) if there is no

water storage at the downstream dams in Sudan including Aswan. This conclusion was derived from

the calibration of a low flow frequency distribution model, extrapolating the empirical distribution’s tail

towards more extreme conditions, based on the natural flow series at Aswan during the past 132

years. The other station at Dongola did not give additional empirical information because this series is

much shorter (only the last 15 years) and does appear to give similar flow values in comparison with

the Aswan data. The above-mentioned 10 years return period for water shortage was based on the

mean annual flow volumes, assuming that the water applications (irrigation, domestic water supply,

hydropower, etc) have a 1 year capacity to store water. For applications where this capacity is smaller

than few months, seasonal flows are of relevance. The study showed that approximately every 2

winters, every 3 springs, every 50 autumns and every 200 summers, Egypt and Sudan may run out of

water. Otherwise in case of existing water storage with 5 years residence time (mainly from dam

storage, and combined for all reservoirs in Sudan including Aswan in Egypt), on a long-term average,

There is a chance for water shortage in Egypt of approximately once in 40 years.. The reader has to

take into account that the 5 years residence time is an assumption made for this study to show the


effect of storage capacity on the return period results. The real residence time may differ from this

value, and needs to be considered for future research.

4.5) Achievements and Lessons Learned

Despite the late start of the DLFAC which was occasioned by some logistical complications in the

financial support and research guidance, the DLFAC progressed very well immediately after such

difficulties were resolved in 2002. Besides the specific achievements which have been highlighted in

various other sections of this report, the general achievements of the component were:

I. Development of a climate and hydrological database for some catchments of the Nile basin.

II. Development of a Nile-basin network of institutions and trained country focal persons for drought and Low-Flow research.

III. Achievement of good research outputs that create a pioneering basis for the future studies of both droughts and low-flows in the Nile Basin.

IV. Development of better understanding of the characteristics and the behavior of the droughts and low-flows (including flows with zero flows) in the Nile Basin, particularly in the frequency domain.

As the work of the DLFAC progressed over the few years of 1st Phase of the FN Project, it slowly

became very clear that working together had eased some of the political misunderstandings and

difficulties that had previously hampered progress in collaborative research in the Nile basin.


There were several problems which were faced in the DLFA Component. Initially, there were some

problems in acquiring DLFA data from some countries mainly due to limitations of trans-national data

exchange policies and in some cases due to the overwhelming cost of the relevant data from some

institutions. Although the problems of cost were partially solved through provision of funding for the

acquisition of the data, the issue of data particularly for regionalization still remains an issue to be

solved.

The lack of participation of researchers from Ethiopia and Eritrea was disturbing particularly in light of

the fact that these two countries had expressed interest to participate in the Component in various

FRIEND-Nile forums. There was no obvious reason that caused the countries not to participate in the

component. The participation of these countries would have enriched the results of low-flow analyses

which have been obtained so far.

Another constraint that was faced in the component was the lack of opportunities to train Masters

(and perhaps PhDs) students within the framework of the research activities of the component. A

general framework of involvement of such students would have ensured consistency of the research

efforts in each country, a high quality of research outcomes and also the achievement of the major

objectives of FRIEND-Nile in terms of capacity building and information transfer.


Another constraint that was faced within the component was the lack of adequate and reliable

communication facilities within the participating countries, particularly in the coordinating institution.

4.7) The Way Ahead

The structure of the research components of the 2nd Phase of the FN Project will not necessarily be

the same as those of the project during the 1st phase. However, there will be a significant presence of

consideration of the drought and low-flows research issues within the newly constituted Stochastic

Component in the 2nd Phase of the FN project. Thus, within the framework of the Stochastic

Component, it will be necessary to:

I. To develop a regional (Nile Basin) team of dedicated researchers for the Stochastic Component.

II. Complete the issues of drought and low-flow regionalization through seminars and working-session workshops.

III. To investigate how the drought and low-flow issues are destined to change in response to regional and global societal and environmental changes.

IV. To widen the research activities to cover as many countries of the Nile Basin as is be possible.

V. To strengthen the research and communications potential in the participating countries.

VI. To involve the MSc and PhD candidates within the component research activities.

4.8) References

Willems, P. (2004a), ‘ECQ: Hydrological extreme value analysis tool’, Reference Manual and User’s

Manual, Hydraulics Laboratory K. U. Leuven, Leuven, Belgium.

Willems, P. (2004b), Water Engineering Time Series PROcessing tool (WETSPRO) users’ manual,

Hydraulics Laboratory K. U. Leuven, Leuven, Belgium


Chapter

5

Sediment Transport and Watershed Management

5.1) Introduction

This chapter states the progress of the research activities of Sediment Transport and Watershed

Management Component (STWMC) within the FRIEND/Nile Project. So far only five Nile Basin

countries out of ten committed themselves to work together to achieve objectives of the project.

These five countries represent more than 90% of the Nile Basin area. However, they are looking

forward having the other five joining the program.

The first year of the project has been devoted to data acquisition and compilation. Moreover, available

literatures have been reviewed to furnish a base for the research work and data analysis. Although

the nomination of the Ethiopian, Kenyan and Tanzanian themes researchers were delayed until year

2002, a genuine progress from their side, in spite of many faced constraints, was spotted. This

chapter includes activities in the 1st phase of the project and it also discusses a work plan for the

STWMC for the second phase of the FRIEND/NILE project.

5.2) Background

Throughout geological times, natural processes of erosion, transport and deposition of sediment have

shaped the landscape in fundamental ways. Erosion often causes severe damage to agricultural land

by reducing the natural soil fertility and agronomic productivity. Besides, eroded soil is the largest

pollutant of surface waters in the world.

The largest source of sediment in the Nile Basin is located in the Ethiopian Highlands where 85% of

the Nile water comes from. The soil that is eroded from the Ethiopian highlands creates serious

problems in the operation, maintenance, and sustainability of irrigation canals and large reservoirs

constructed along the Nile (e.g., Roseries, Sennar, Girba and Aswan High Dam). Large amounts of

sediment deposition in such reservoirs strongly reduce their lifetime by reducing their water capacity.

Moreover, clear water released from such reservoirs often induces erosion of channel banks and bed

(e.g., erosion downstream of Aswan High Dam).


Furthermore, large sediment problems are also encountered in most catchments around the Victoria

Lake. The large quantity of sediment that is eroded from the highlands and is transported by rivers,

often deposits in flat areas around the Victoria Lake leading to several problems. The sediment

deposits reduce the flood carrying capacity of the stream channels, which often results in greater

flood damage to adjacent properties.

The sediment issue and their associated problems within the Nile Bain will be discussed in this

chapter taking into consideration its socio-economic and environmental impacts. Also, sources of

sediment and the transport media will be explained. Moreover, the sediment deposition in quantity

and distribution within the Nile Basin will be examined. Mitigation guidelines and sediment

management measures will be reported benefiting from the experience of the FRIEND/ Nile project in

this field.

The Sediment Transport and Watershed Management are very important to the Nile basin countries

when dealing with water resources. This is because the sediment related problems (erosion, transport

and deposition) create difficulties in managing watercourses within the Nile basin countries. For

example, soil erosion and associated land degradation in Ethiopia, Eritrea, Tanzania, and Kenya;

sediment deposition in reservoirs and irrigation schemes in Sudan; and the impact of the Aswan High

Dam (AHD) as a sediment trap reservoir on river behavior in Egypt are major problems. These

problems are growing yearly in response to the increasing population pressure on agricultural land as

well as the green cover. Rates of erosion, often well in excess of rates of soil formation, are a recipe

for disaster. There is a need for proper understanding of watershed management, soil loss and land

degradation, to formulate appropriate soil conservation strategies.

5.2.1) Nile River Basin

The Nile River extends for some 6700 km through much of the Northeastern Africa. The setting is

highly variable and ranges from tropical rain forest to desert and from mountainous relief to areas

which are below sea level. The Basin extends over many ranges and altitude and contains wide

variation in climate. The climate ranges from desert conditions in the north, to tropical climates in the

southern regions, and includes alpine extremes in mountain regions.

The main braches of the Nile River, the Atbara, the Blue Nile and the White Nile system form the

feature of the Nile Basin. The White Nile source is from the Equatorial Lake Plateau (Burundi,

Rwanda, Tanzania, Kenya, D.R. of Congo and Uganda), and the Blue Nile with its sources from the

Ethiopian Highlands. The sources are located in humid regions, with an average rainfall of over 1000

mm/year. The arid region starts in Sudan, the largest country in Africa, which can be divided into

three rainfall zones: the extreme south of the country where rainfall ranges from 1200 to 1500 mm/

year; the fertile clay-plains where 400 to 800 mm/year of rain falls annually; and the desert northern

third of the country where rainfall averages only 20 mm/year. Further north, in Egypt, precipitation falls

to less than 20 mm/year.


The area of the Nile Basin represents 10.3% of the total area of the African Continent and spreads

over 35% of the total area of ten countries. Table 5-1 shows the share and contribution of the 10

riparian countries to the Nile Basin.

• Table 5-1, Nile Basin: areas and rainfall by country.

Country Country

area (km2)

Country

Area

within the

basin

(km2)

As %

of total

basin

area

As % of

country

area

Av. annual rainfall in the basin

(mm)

Min. Max. Mean

Burundi 27 834 13 260 0.4 47.6 895 1 570 1 110

Rwanda 26 340 19 876 0.6 75.5 840 1 935 1 105

Tanzania 945 090 84 200 2.7 8.9 625 1 630 1 015

Kenya 580 370 46 229 1.5 8.0 505 1 790 1 260

Zaire 2344 860 22 143 0.7 0.9 875 1 915 1 245

Uganda 235 880 231 366 7.4 98.1 395 2 060 1 140

Ethiopia 1100 010 365 117 11.7 33.2 205 2 010 1 125

Eritrea 121 890 24 921 0.8 20.4 240 665 520

Sudan 2505 810 1 978 506 63.6 79.0 0 1 610 500

Egypt 1 001 450 326 751 10.5 32.6 0 120 15

For Nile basin

8889 534 3 112 369 100.0 35 0 2 060 615

5.2.2) River Nile Watershed

Nile River has three main distinct regions see Figure 5-1, from where it obtains its flow. These are

namely: the Equatorial Lake Plateau in the south, the Sudd (Bahr el Ghazal region in the center), and

the Ethiopian Highlands in the east). From the confluence of the Atbara River north to the

Mediterranean, see Figure 5-1, the Nile receives no effective inflow. The total estimated annual inflow

entering Lake Victoria from stream flow and rainfall is 118 billion m3 while the evaporation is estimated

to be 94.5 billion m3, leaving only 23.5 billion m3 to flow down the Victoria Lake. In the Sudd the loss

calculated as 33.9 billion m3, leaving 15 billion to flow the White Nile. In the High Aswan Dam

Reservoir, the losses are calculated to be 10.5 billion m3, while the losses within Sudan downstream

Malakal is estimated to be about 7.0 billion m3.


Southern Watershed

Eastern Main Watershed

Central Watershed

Southern Watershed

Eastern Main Watershed

Central Watershed

• Figure 5-1, The Three Watersheds of the Nile River.

On the hand, the remainder comes from the Ethiopian Highlands via the Sobat (13.5 billion m3), the

Blue Nile (54 billion m3) and Atbara River (12 billion m3). The White Nile is extremely well regulated

with relatively constant contribution to the Nile River because of several lakes and swamps within the

Sudd area. Flow from the Ethiopian Highlands is highly concentrated in the period from July to

October where 85% of the Nile River total flow occurs.

The Blue Nile is almost dry from January through June and Atbara is normally dry during that period

too. Figure 5-2 shows a schematic diagram of the Nile River natural flows, while Figure 5-3 shows

the hydrographs of the Nile system.


• Figure 5-2, Schematic Diagram of the Nile River Natural Flows.

1000 1000

• Figure 5-3, Hydrograph of the Nile River.


5.3) STWMC Objectives

5.3.1) General Objective

The main objective of STWMC is to build and sustain adequate capacity in sediment transport and

watershed management for the Nile Basin riparian countries in order to rationally develop and

properly manage their water resources

5.3.2) Specific Objectives

• To understand sediment processes in watershed, stream and reservoirs; and to select a robust sediment model applicable to the region.

• To develop guidelines for watershed management in the Nile Basin.

• To bring together professionals from the Nile Basin to exchange experience, ideas and to foster common understanding and cooperation.

• To enhance regional research capacity on topics related to sediment transport and watershed management.

• To raise level of awareness of policy makers, stakeholders and the public at large on various problems related to sediment transport and watershed management.

5.4) Data Acquisition

5.4.1) Necessary Data for Sediment Transport Modeling

The STWMC research team has identified the following Checklist of sediment data requirements;

however, not all of these data are available in the Nile Basin.

Catchment Level

• Drainage area,

• Elevation,

• Difference in elevation to highest point,

• Shape,

• Length from outlet to highest place in catchment,

• Detailed map,

• Environmental parameters, under the form of:

o GIS layers,

o If not available: percentages,

o If that is not available, at least: average for catchment.

• Land cover,

• Soil texture,

• Average length of agricultural plot,

• Soil organic matter content,


• Rainfall distribution,

• Slope gradient,

• Main geomorphic features, especially floodplains, marshes, alluvial fans, gullies,

• Total length, data on evolution…

For every sediment measuring station

• Drainage area to that place,

• Elevation,

• Difference in elevation to highest point,

• Shape,

• Length from station to highest place in catchment,

• Sediment transport data series (greatest resolution possible),

• Concentration,

• Absolute data (kg/s),

• Yearly totals (T/y),

• Available time series for:

o Rainfall,

o Rain intensity,

o River flow/Flood flow (Runoff),

o Spring flow.

• Details on measurement techniques.

Stream

• Water discharge with highest possible time resolution,

• Surface water profile,

• Slope of the river bed,

• Bed material mechanical analysis,

• Sediment concentration,

• Morphologic characteristics,

• Details of methodology and equipment used for sampling and analysis,

• Environmental parameters (e.g., Temperature, Aquatic weeds etc….)

Reservoir

• Area, Capacity versus Elevation,

• Evolution with respect to time,

• Operation rules of the reservoir,

• Discharge, (input and output),

• Bathymetric survey,

• Suspended sediment concentration, input, output and distribution along the reservoir, soil mechanical analysis (sediment properties),


• Velocity distribution along a reservoir,

• Water temperature with respect to depth.

• Figure 5-4, Rivers in Sondu River Basin.

5.4.2) Case Study in Each Country

5.4.2.1) Kenya

Study Area: Sondu River:

Sondu, also known as Miriu, is one of the major rivers in Nyanza Province; see Figure 5-. It rises to

the South east of Kericho at the elevation of about 3000 m in the Western Mau Forest. It enters

Victoria Lake at a point that is 22 km south-east of Kisumu Town at Nyakach Bay. Sondu Basin is

situated between Nyando Basin and Kuja-Migori Basin within the larger Lake Victoria Basin, it drains

to Kisii, Ondu Basin, and it exists within the elevation of between 1100m - 2000m above the sea level,

and altitude of between 35° 45’ E and 34° 45’ E, and longitude of between 0° 15’ S and 1° 00’ S; and

it drains a total basin area of approximately 5180 km2 with the population of approximately 314

persons per km2 living within the basin area according to the last census.. The Sondu River Basin

covers parts of Kericho, Kisumu and Rachuonyo Districts. Sondu Miriu flood plains form parts of Kano

Plains situated within the Lake Basin Region. The Kano plain covers location in Winam, Nyando and

lower Nyakach Divisions of Kisumu District. Within the plains area, there is an area of about 700 km2

that is normally affected by floods. The flood zone lies between the meridians 34° 45’ E and 34° 55’ E

and latitude 0° 15’ S. and 0° 21’ S. This covers a part of the sub-locations of the lower Kadiang’a and

West Koguta in West Nyakach location in Kisumu District. The two locations are known by local


people as the Sango area. The Sango area is normally affected by floods. Recently, flood hazards

has increased in this region due to various reasons including land use changes causing erosions and

changes in surface runoff in the basin to the lower flood plains. Hence huge amount of sediment is

transported, raising the level and consequently causing damaging floods.

Sondu River Basin is located in the western flank of the Gregory Rift (Kenya). The Sondu River Basin

covers mainly three areas: the highlands water catchment area, which spreads in Transmara,

Chepalungu forests of Kericho, and Narok Districts in the Rift Valley province. The streams starting

from these areas run through highlands up to Near Nyakach escarpment and join together at Ikonge

into the River Sondu, which eventually enters the flat land through the Odino falls. Lands, which

spread from the foot of the South Nyakach Escarpment to the shore of Lake Victoria, present parts of

the Kano plains.

• Figure 5-5, Location map of the Aswash river study area.

5.4.2.2) Ethiopia

Study Area: Awash River

The Awash basin is bordered by the catchment of the Wabi Shebelle River to the South, the

catchment of the Blue Nile to the west, the inland depressions of the Dankail desert to the north, and

Somalia to the east, see Figure 5-5. The Awash River originates from the high plateau some 150 km


west of Addis Ababa, at an altitude of about 3000m. It, then, flows eastwards through the Becho plain

areas and is joined by several small tributaries before entering the Koka reservoir, which is

considered as the downstream limit of the upper Awash basin. The basin extends as middle and

lower Awash with a total catchment area of 110,000 km2. The study area is above Koka reservoir that

encompasses an area including the capital Addis Ababa, other medium and minor towns, agricultural

and grazing land, and also swamps and flood plains entering the Koka reservoir at an elevation of

1500 m (a.m.s.l.).

Land use in the catchment area is mainly agricultural land used for rain fed crops and grazing lands.

There are some plantations scattered in the catchment. The rainfall pattern is bimodal with two rainy

seasons each year. The first short rainy season is from March to May, and the second main rainy

season starts in July and lasts till September. During the dry season, i.e. from October till February,

the prevailing winds are anticyclone winds, mainly blowing from the northeast. At other times of the

year, winds are variable in direction and strength, but are in general upper rain-bearing air currents

coming from the southwest.

5.4.2.3) Sudan

Study Area: Blue Nile River

The study area lies in the center of Sudan at the confluence of the White Nile and Blue Nile and

covers the reaches of the two Niles and Main Nile in the vicinity of Khartoum Center, see Figure 5-6

The demarcation starts from about 200 m upstream of Burri Bridge on the Blue Nile and ElIngaz

Bridge on the White Nile. It extends to about 200 m downstream Shambat Bridge on the Main Nile.

The area is recognized by the confluence of the Blue Nile with the White Nile and the presence of Tuti

Island. Thus the study area is influenced by the two rivers. The most important hydrometric aspects

that are believed to influence the characteristics of the river system within the study area are the

following:

• The seasonal pattern of the Blue Nile flows, which range from about 10 million cubic meters per day (Mm3/day) in the dry season to about 700 Mm3/day in flood periods. That result in large seasonal fluctuation of flow levels from about 373 to about 380 m (a.m.s.l.) (Sudan survey datum).

• Reasonably stable flows of the White Nile (Q = 73 Mm3/day +25%).

• Seasonal increase of the sediment load transported by the Blue Nile with the rising flood from a negligible value to remarkably high value.

• Regulation effects from Sennar and Jebel Aulia dams.

The Flows of the Blue Nile cause flood problems and surface drainage difficulties during the rainy

season. These flows are also responsible for sediment transport in the area, and thus, the Blue Nile is

believed to significantly influence any changes that may take place within the reach.

The area was surveyed in 20 main cross-sections denoted by X letter (land and bathymetric surveys),

see Figure 5-6 Additionally 13 auxiliary cross-sections were surveyed in order to fill the relatively wide

gaps between the main cross-sections (bathymetric survey only), denoted by A letter in Figure 5-6


• Figure 5-6, Locations of the X-sections in the Blue Nile river.

Dongola G.ST.

AHDR

Study Reach

Aswan High Dam

• Figure 5-7, Location of Aswan High Dam and Dongola station.

5.4.2.4) Egypt

Study Area: Aswan High Dam

Aswan High Dam (AHD) is a rock fill dam, closing the Nile River at a distance of 6.5 km upstream of

the old Aswan Dam, about 950 km south of Cairo as shown in Figure 5-7. The dam is 3600 m long

and has a width of 40 m at the top and 980 m at the bed level. The maximum height of the dam is 111

m above the river bed. Construction began on the Aswan High Dam in 1960. By 1964 the river was


blocked with a coffer dam, and the upstream reservoir began to fill. The construction of the Dam itself

was completed in 1970. The construction of AHD upstream of old Aswan Dam, made it possible to

have over-year water storage and thus create a reservoir upstream the dam. The length of AHD

reservoir is about 500 km at its maximum storage level, which is 182 m (a.m.s.l.), with an average

width of about 12 km and a surface area of 6540 km2. The storage capacity of the reservoir has a

volume of 162 km3 divided into three zones: dead storage capacity of 31.6 km3 between levels 85 m

and 147 m, live storage capacity of 90.7 km3 from level 147 m to 175 m, and flood protection capacity

of 39.7 km3 ranging between levels 175 m and 182 m, the maximum level of the reservoir. The reach

analyzed in this study was chosen from the kilometer 500 to the kilometer 350 upstream Aswan High

Dam with a total length of 150 km. Here, we focus on the mean bed channel which represents the

area with the most intensive sediment deposition.

• Figure 5-8, the Simiyu river with the stream network.

5.4.2.5) Tanzania

Simiyu River:

Figure 5- shows the watershed of river Simiyu with its stream network. The river is discharging into Lake Victoria with catchment area of 11,577 km2. Discharge data of the river are available since 1950. Also, sediment data for the last ten years are available although they are scanty. However, sediment monitoring has being continuing for the last two years to give more dense data. It is required to model processes taken place on catchment, stream and reservoir scales.

5.4.3) Summary of Available Data in Each Case Study

Table 5-2 shows a short summary of available data in each case study which were used in

applications of SMS models (also see section 5.12).


• Table 5-2, Summary of available data in each case study.

Country and Case Study Available Data

Kenya : Sondu River:

The river is discharging into the Lake Victoria.

The data is available since 1939 for discharge and for

sediment in 11 stations for 10-15 years

Ethiopia : Awash River

Catchment area 11,200 km2.

Data records are available since 1963.

There are 5 stations for sediment measurements

Sudan : Blue Nile River: The river is Originating from Ethiopia high plateau

Discharge records since 1912. Data for Sediment is

available in 4 monitoring stations for 5-10 years.

Egypt : Aswan High Dam Data is available discharge for more than 100 years.

Sediment monitoring stations (Dongola + Wad Halfa + Abu

Sumbel) is available before the AHD for the period (1957 –

1966) and after AHD from 1967 up to now but not complete.

Tanzania : Simiyu River The river is discharging into Lake Victoria with catchment

area of 11,577 km2

Data available for discharge since 1950.

Sediment data (scanty) for the last ten years, but sediment

monitoring has being continuing for the last two years

5.5) Methodology

• Setup clear objectives and specific aims.

• Setup action working plan for the research and select the study area (Blue Nile River system and its watershed in Sudan, Sondu River, Simiyu River in Tanzania, Awash in Ethiopia and Aswan High Dam Reservoir in Egypt).

• Data acquisition including (rainfall, maps, satellite imageries, water discharges, water levels, sediment data, … etc.) is collected.

• Selection of suitable model in the field of sediment transport (SMS Model is selected).

• Analyzing the data with the help of SMS Model and other means (Other software sediment models can be used for comparison).

• Results are to be discussed and reported.

• The study is to be documented including all the study cases in Egypt, Kenya, Ethiopia, Sudan and Tanzania in one report.

• The report will be presented to the FRIEND/NILE Steering Committee.


5.6) Sediment Transport Modeling Software

5.6.1) Selection and Testing of Sediment Transport Modeling Software

The selection of sediment transport model to be used by all researchers in the Nile Basin is not an

easy task. Despite the extensive research efforts, knowledge of erosion and sediment transport

models of sedimentation are diverse and varied in nature. Functionally, models can be classified into

five major categories: watershed, stream, reservoir, estuarine and coastal sedimentation models.

Many computer sedimentation models have been reviewed, but no single model is usable under all

conditions. Therefore, making proper selection and use of models that best fit particular

circumstances is always a critical and troublesome issue. However, STWMC Khartoum Meeting –

Dec. 2002, put forward criteria for selecting the proper software models for the sedimentation to be

used in the whole Nile Basin:

• Easiness of use (user friendly),

• Technical support and after sale service,

• Reliability (verified/tested),

• Documentation,

• Purpose of modeling to serve the specific purposes of the study,

• Data availability and quality,

• Flexibility (not a black box),

• Cost/efficiency.

First, an inventory was made of the sediment models in current use. Secondly, based on the above

criteria a group of researchers (from the UNESCO Chair in Water Resources (UCWR) and other

institutes inside Sudan and with consultation of several experts from outside Sudan), decided to go for

the SMS model as an effective sediment model. The RMA2 and SED2D models of SMS were used

by the STWMC research team to analyze the sedimentation problems within the Nile Basin countries

working in the FRIEND/Nile project. It is very important here to brief some of the findings about the

sediment models in general including the SMS model:

• Mathematically: almost all the models are simulating a boundary value problem.

• Hydraulically: all models include four major model components:

o Equation of motion for water.

o Continuity equations for water and sediment.

o One or more sediment transport functions.

o A relationship for channel resistance.

• Most models solve their boundary value problems and the related partial differential equations with the finite difference method (explicit technique).

• Most models have greatly simplified their flow problems by considering unsteady problems as steady ones (SMS uses both).

• Most models are uncoupled in the sense they solve the flow equations first and consider the sediment factors later.


• All models may have different sediment distribution assumptions.

• Only a few models, in limited ways, can model bank erosion, armouring effects of channel geometry and morphology changes (SMS considered all of them).

• Many models may produce significantly different results, even when run with the same set of inputs.

• Most of the existing stream sedimentation models are an amalgamation of three major modules that are water routing, sediment routing module and special function module.

• Most models used in water routing use one of the following methods:

o The finite difference method.

o The finite element method.

o The method of characteristics.

• Most models, at present, use sediment transport functions which are not applicable to wash load or unsteady flow conditions (SMS considers unsteady conditions).

5.6.2) Calibration of SMS Model Software

Obviously, erosion and sediment transport phenomena are significant progresses that affect the water

movement in a river system. These phenomena can either be benefit or detrimental to the utilization

of the water resources. Because of that, STWMC Khartoum meeting recommended testing and

selecting a suitable model that can simulate the erosion and sediment transport phenomena and

solve practical engineering problems.

According to these recommendations, successive studies have been carried out among several types

of models that can simulate the sediment transport problems, these models have been compared

depending on their types, governing equations, their applications, input data, capability of

visualization, their limitations, …etc. These comparisons lead to choosing the SMS model (Surface

Water Modeling System) as a useful tool to enhance understanding of the erosion and sediment

transport problems.

5.6.2.1) Description of the SMS Software

SMS is a pre- and post-processor for surface water modeling and analysis. It includes two-

dimensional finite element, two-dimensional finite difference, three-dimensional finite element and

one-dimensional backwater modeling tools. It consists of several water simulation models; each of

these models is designed to address a specific class of problem. These models can be classified as

follows:

• 2D Hydrodynamic Models including:

o RMA2 (computes hydrodynamic data such as water surface elevation and flow velocities for subcritical free-surface flow).

o FESWMS (supports both super and subcritical flow analyses).


o HIVEL-2D (analyzes flow fields, which have shocks such as hydraulic jumps and oblique standing waves).

• 2D Sediment Models, including:

o SED2D (calculates the bed sheer stress, sediment concentration, and bed changes using RMA2 outputs).

o SED2DH (same as above, but uses FESWMS outputs).

• 2D Contaminant Transport, including:

o RMA4 (uses for the transport of a contaminant, salinity intrusion, or tracking Dissolved Oxygen (DO) and Biochemical Oxygen Demand (BOD) in a system).

o Marine/Coastal Hydrodynamic.

o ADCIRC (uses for the computational domains encompassing the Deep Ocean, continental shelves, coastal seas).

o CGWAVE (near shore wave model).

o ST0WAVE (wave model).

SMS Model ranks high among other software packages when measured against the following criteria:

• User friendly and pre and post processing with GIS capabilities which allow dealing with a wide range of input data format.

• High numerical accuracy using 9 nodes rectangular elements.

5.6.2.2) Basic Data Required by the SMS software

Working with SMS for analyzing data for the Sediment Transport Watershed Management, the

thematic researchers of the component found out that more information than previously anticipated

were needed to run the model. In brief the basic data required to run the SMS model are as follows:

1. Digital map images (TIFF or JPEG images of topographic maps, or stereo aerial photos to construct a digital terrain model),

2. Bathymetry data (Land Surface and underwater ground Surface),

3. Digital soils/material data (polygons representing roughness of land and submerged areas),

4. Flow data (Flow rates, water surface elevations for key points), and

5. Sediment characteristics (soil properties of sediment in river).

Although the contracted data at the beginning were not covering all the above areas, information

provided by the Thematic Researchers indicated that all of them managed to run SMS even though

with few variable difficulties.

5.6.3) Problems Encountered in Modeling Task

During modeling of hydrodynamics and sediments of the case studies considered here, there are few

problems encountered. These include:

I. Scatter data and Interpolation Problems: SMS assigns data from scatter data set to finite

element network by using several interpolation techniques. There are only three methods of

interpolation in SMS. These are namely linear, inverse distance weight (IDW), and Natural

Neighbor (NN). The application of this tool gives false interpolation of the river morphology


and some times gives a zero reading because of the limitation of the linear method to

extrapolate beyond the convex hull of the scatter point sets. To overcome this problem we

are forced to enter the missing data manually. When using the IDW method the model gives

an error when choosing the local (use triangle topology) option and the model instantly

closes.

II. Boundary condition problems:

a. Rating curve problems: There is no direct way to assign rating curve directly in SMS, the only way is using the BRC (Boundary Rating Curve) card and trying to revise the value to the value required by the REV (revise the current time step) card. In our case it was difficult to apply the rating curve and the model failed to converge.

b. Water surface elevation problems: Because RMA2 is a wet model; it has to assign water surface elevations higher than any elevation node in the model. However, by using the REV card, the water surface elevation can be ramped down to the actual level. We can overcome these problems by using relatively high values of the eddy viscosity and marsh porosity, compared to the default model value.

c. Internal boundary conditions: There is no facility to split the flow in a particular ratio at the junction, so that the flow ratio in the confluence depends on the mesh size, which gives over or under estimation of the calculated water levels.

III. Model Divergence: The model some times had convergence problems, which leads the

model to stop before getting final results. This problem occurred when applying the following:

a. Ramping down the water surface elevation.

b. Applying the rating curve equation in the boundary conditions.

c. Changing the values of the marsh porosity.

d. Using the default viscosity value.

5.7) Case Studies

5.7.1) A Comparison between Two Different Transport Models to Predict Sediment

Transport at the Simiyu River, Tanzania, as a Case Study

First, a mesh was created for the 15 km long river section using the measured cross sections and

additional information from the digital elevation model of the area and the delineation of the river

reach. The model is able to extrapolate the bathymetry/cross sections at all other locations. The

interpolated model depths were compared with measured depth and they showed a good agreement.

This means that the model is using correct cross sections at all locations. The hydrodynamic model

was then run to simulate the stream flow under steady state condition and later under dynamic state

condition. The output of the hydrodynamic module with time series of discharge and sediment loads,

which correspond to a typical flood scenario during the flood event, was used for the simulation of a

dynamic state.


• Figure 5-9, A map showing bottom bed change after 144000 hours of simulation (Case study: Simiyu River).

0.25

0.50

0.75

1.00

1.25

0 50 100 150 200 250

Val

ue

Time

Upstream

Middle

Downstream

0.25

0.50

0.75

1.00

1.25

0 50 100 150 200 250

Val

ue

Time

Upstream

Middle

Downstream

• Figure 5-10, Bed changes at different locations (Case study: Simiyu River).

Comparison between Longtudinal Section for the High Aswan Dam Reservoir (HADR) at 2003

145.00

150.00

155.00

160.00

165.00

170.00

175.00

340360380400420440460480500Distance from High Dam (km)

Elev

atio

ns (m

)

2003- Measured

2003- Model

Flow Direction


145.00

150.00

155.00

160.00

165.00

170.00

175.00


Elev

atio

ns (m

)

2001- Measured

2001- Model

Flow Direction


145.00

150.00

155.00

160.00

165.00

170.00

175.00


Elev

atio

ns (m

)

2003- Measured

2003- Model

Flow Direction


145.00

150.00

155.00

160.00

165.00

170.00

175.00


Elev

atio

ns (m

)

2001- Measured

2001- Model

Flow Direction

• Figure 5-11, Comparison of measured and predicted longitudinal profile for AHDR 2001 and 2003.

Results of the sediment transport during a peak flood event have shown that shallow areas (e.g.,

flood plains) are subject to more deposition than the main river channel. Bed changes are more

pronounced in the upstream closer to the sediment input to the system; see Figure 5-9 and


Figure 5-10 Downstream shows less deposition indicating that a longer time is required for the

sediment to travel up to the downstream.

5.7.2) Modeling of Sedimentation Process in Aswan High Dam Reservoir

A comparison between the observed and modeled cross-sections indicates that there is a good

agreement between the modeled and the measured cross-sections, although some slight differences

are observed. In order to increase the accuracy of simulation, measurement locations of velocity in

the transverse direction should increase As comparing the modeled and measured longitudinal

sections of AHDR in 2001 and 2003, it was noted that the modeled bed level is higher than the

measured one in the whole inlet zone of the reservoir (i.e. from the kilometer 500 to the kilometer 400

upstream the dam) although there is a good agreement between modeled and measured longitudinal

profile in the rest of the reservoir (Figure 5-11). This is may be explained by the fact that part of the

sediment is probably trapped in the Sudanese reservoirs before entering AHDR, which is not

considered in the model. For the prediction of delta progress in the dam direction until year 2010, a

time series of seven successive years of high flood were simulated. The prediction indicates that the

bed level will rise along the whole reservoir by a value ranging between 3.5 and 1.5 m in the inlet

zone until the kilometer 370 upstream the dam, and by a value ranging between 0 and 1.5 m in the

rest of the length. These seven years of flood are predicted to be followed by five successive years of

low flood. The model predicts a level raise of the bed by 1.0 to 2.0 m in the inlet zone until the

kilometer 370 upstream the dam, and by 0 to 1.0 m in the rest of the length.

5.7.3) Nile River Sediment Modeling: Challenges and Opportunities

Although sediment models are based on well-known flow equations, most of them select sediment

functions without any good justification for such selection. Therefore, professional experience plays a

major role in this selection.

The use of SMS to simulate sediment process in the Nile Basin River and Awash River within the

FRIEND/ Nile Project has encountered some difficulties and problems. This could be attributed to the

different topography of the countries involved and the availability of suitable data.Application of SMS

model in the steep rivers (e.g., Kenyan, Tanzanian and Ethiopian case studies) showed difficulties in

obtaining straightforward results. Modeling such rivers requires high professional experience on their

nature.

This section recommends that the Nile Basin countries should give more attention to data collection in

the field of water resources management, specially the sediment data for better and efficient water

management.


5.7.4) Overview of Sediment Problems in Nile Basin

Several studies it was found that the sedimentation in the reservoir and the irrigation systems within

the Nile Basin has environmental and socio-economic impacts. Therefore, suitable sedimentation

management is a key for the sustainable water resources management.

In the Nile Basin, the total annual sediment load, that reaches Aswan High Dam, ranges between 140

and 160 million tons. Conversely, changes in human activities within the catchment can have

detrimental effects on both sediment quantity and quality. Sediment is socio-economic, environmental

and geomorphologic resources, as well as a tool of nature. However, changes in sediment quantity

and quality can have a significant impact on a range of social, economic and environmental systems.

Neglecting to manage sediment in a sustainable way, either by a back of adequate sediment

management strategies, or the cursory induction of sediment in generic policy and legislation, can

result in costs to both society and environment.

It is very important to evaluate environmental impacts involved in sediment management properly and

mitigate them as much as possible. It can be concluded that sedimentation rate in the last decade

(1990s) increased rapidly, indicating that huge and wide land degradation is occurring in the

catchment area of the Nile River system. Therefore, integrated sediment management is found to be

the best policy to minimize the adverse impacts of the sedimentation within the entire Nile River

Basin. Figure 5-12, Figure 5-13, and Figure 5-14 show more details about the results.

0 . 0 0

1 0 . 0 0

2 0 . 0 0

3 0 . 0 0

4 0 . 0 0

5 0 . 0 0

6 0 . 0 0

J A N F E B . M A R . A P R . M A Y J U N . J U L A U G . S E P . O C T . N O V . D E C .

M o n t h s

Sedi

men

t Con

cent

ratio

n

(mlg

/litr

e)

• Figure 5-12, Suspended Sediment Concentration in AHD Reservoir.


• Figure 5-13, Comparison of Rainfall, Discharge and Sediment Yield in the River Atbara (right) and the Blue Nile (left).

• Figure 5-14, Sediment Volume and Content of Sennar Dam (left) and Roseires Dam (right).

5.7.5) Modeling Water and Sediment Fluxes in Steep River Channels: Case of Awash

Basin

In this section, the results of the model runs for August 2000 are given in Figure 5-15, Figure 5- and

Figure 5-. They clearly indicate that the stream velocity decreases in the downstream direction. The

riverbed changes significantly during the flood season. Bed level changes are largest in the

downstream section, as the slope of the river channel decreases downstream. The Bed level

changes along the curved part of the river channel indicating that erosion occurs at the outer bend of

the river channel and that deposition is occurring at the inner bend.

0

100

200

300

400

500

600

700

800

900

1000

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

Jan

Feb

Mar

Apr

Rai

nfal

l / D

isch

arge

(Mm

3/d)

-0.5

0.5

1.5

2.5

3.5

4.5

5.5

6.5

Sedi

men

t Yei

ld (1

06 T

/ Day

)

Gonder + Makale Kubur + Wad El Heleiw Sediment Yield

0

500

1000

1500

2000

2500

May 1

Jun 1

Jul 1

Aug 1

Sep 1

Oct 1

Nov 1

Dec 1

Jan 1

Feb 1

Mar 1

Apr 1

Rai

nfal

l \ D

isch

arge

(Mm

3/d)

0

1000

2000

3000

4000

5000

6000

7000

8000

Sedi

men

t Con

cent

ratio

n (p

pm)

Rainfall (Mm3/d) Eddeim Discharge Sediment Concentration (ppm)


W ater Depth (m) at Time s tep 192hr

0.72

1.34

1.96

2.58

3.20

3.82

• Figure 5-15, Water Depth in the selected reach of Awash River.

W ater Velocity (m/s) at Time s tep 192hr

0.29

0.69

1.09

1.48

1.88

2.27

• Figure 5-16, Average Velocity in selected reach of Awash River.

Comaprision of Measured and Simulated Water Depth

3

3.2

3.4

3.6

3.8

4

4.2

4.4

4.6

4.8

8/11/04 8/12/04 8/13/04 8/14/04 8/15/04 8/16/04 8/17/04 8/18/04 8/19/04 8/20/04 8/21/04

Date in Days

Wat

er D

epth

(m)

Measured Water DepthSimulated Water Depth

• Figure 5-17, A comparison between measured and simulated water depth in Awash River.


55

1150velocity mag

0.01560

0.01650

0.01740

0.01830

0.01920

0.02010

0.02100

0.02190

0.02280

0.02370

0.02460

• Figure 5-18, Results of Water velocity for the 5Km. Stretch ( Sondu River).

water depth

4.0

5.1

6.2

7.3

8.4

9.5

10.6

11.7

12.8

13.9

15.0

• Figure 5-19, Results of the Water depth at 5 km Stretch ( Sondu River).

water surface elevation

1150.000002

1150.000014

1150.000026

1150.000038

1150.000050

1150.000062

1150.000074

1150.000086

1150.000098

1150.000110

1150.000122

• Figure 5-20, Results of the water surface elevation ( Sondu River).


550

1150.5

bed change : 0.000

0.009

0.027

0.045

0.063

0.081

0.099

0.117

0.135

0.153

0.171

0.189

• Figure 5-21, Results of bed change at t=0 ( Sondu River).

550

1150.5

bed change : 1440.000

0.009

0.027

0.045

0.063

0.081

0.099

0.117

0.135

0.153

0.171

0.189

• Figure 5-22, Results of bed change at t=1440 ( Sondu River).

5.7.6) Limitations of Hydro-dynamical Models with Limited Data Available Case

Study: Sondu River Basin (Kenya)

This section, using the little real and regenerated data available during research period, illustrates

results of a hydrodynamic model that was initially run under steady state condition. The sampled data

of April, 10th 1985 at 1JG 02 and 1JG 03 were chosen to assign the boundary conditions of the model

with water level of 1150 m (a.m.s.l), a discharge of 55 m3/sec and water depth of 340 cm. Results of

model runs are shown in Figure 5-18, Figure 5-19, Figure 5-20, Figure 5-21and Figure 5-22

5.7.7) Overview of Soil Erosion around Lake Victoria

There is a need to evaluate the economic impact of soil erosion and sediment transport in all

reservoirs already built or planned taking into account environmental implications. Regular and

uniform sediment monitoring should be improved. There is also a need for training and support of

large number of Kenyans in all aspects of water and soil management such as agriculture, forestry

and water resources conservation. It is also worth noting that:


• Soil and Water Management of the Nile watershed requires good understanding of:

o Hydrological processes in the basin (the WMS software can facilitate that),

o Rainfall – Runoff processes.

• The future sedimentation problems cannot be adequately assessed merely by measuring sediment in the river channel at the site planned for reservoir.

• Realistic evaluation must be based upon a thorough understanding of hydrologic and geomorphic processes as they relate to land use on the entire catchment.

• Identification of main processes of sediment production.

• Identification of driving factors of sediment production (e.g., land use, topography).

• The soil erosion modeling on the hill slopes which has been lacking should be given first priority.

5.7.8) Effect of Upstream Structures on Delta Progress in Aswan High Dam Reservoir

Aswan High Dam (AHD) is a rockfill dam. It closes the Nile at a distance of 6.5 km upstream of the old

Aswan Dam, about 950 km south of Cairo. The construction of the AHD upstream of the old Aswan

Dam made it possible to have an over year water storage and thus create a reservoir upstream the

dam. After construction and operation of AHD, about 9 -10 million tons of suspended sediment has

been depositing annually on the flood plains of the Nile. However, about 93% of the total average

annual suspended load of 124 million tons is carried out to the Mediterranean Sea. After the

construction of AHD, this value is used for estimation and prediction of sediment distribution in the

longitudinal and transverse directions in AHDR. Application of the SMS software in AHDR gives good

results for the estimated sediment load in the whole reservoir while it gives an over estimation in the

entrance reach only. There are some items that have not been taken into consideration in the model.

These items are mainly the decrease in sediment load per year with about 15% due to construction of

Roseires Dam on the Blue Nile River and Khashm EL-Gerba Dam on Atbara River. These are beside

the increase of water withdrawal with sediment by Sudan from 4 billion m3 to 14.5 billion m3 per year.

In addition, the construction of Merowe Dam in Sudan will decrease the sediment load that enters

AHDR. Based on the complete information of the future structures upstream AHDR, different

scenarios of water flow and sediment discharges until year 2017 will be examined using the SMS

software to study the development of delta deposits in AHDR.

5. 8) Remarks on the Results

One of the most important factors affecting the sediment modeling results is data availability. A

comprehensive set of data covering morphology and hydraulics is vital for any study area being

modeled. Furthermore, in modeling stream sedimentation problems, there are two types of

mathematical limitations (i.e., convergence and solution procedures).

The application of SMS software in the Nile Basin faced several constraints and problems, which are

briefly described below.


i) Model Scale:

Multiple scales of analysis (local, sub-regional, and regional scales) are needed for modeling

sediment transport in the Nile Basin. For example, at the local scale, erosion models can be used to

simulate the sediment yield at the catchment, including impacts of land management practices. At the

sub-regional scale, flow and sediment are transported through a system of channels. At present there

are several models to evaluate sediment transport and channel response using one, two and multi

dimensional sediment transport models. Regional scale analysis consists of the entire river basin

watershed and associated channels. At this scale, simulations include sediment and water routing

and channel response through the entire sediment region.

Analysis capabilities include channel stability and geomorphic response for all watershed channels,

distributaries, the main channel and receiving waters bodies (including reservoirs and estuaries).

Using these scales for sediment transport modeling in the Nile Basin gives flexibility in assessing the

sediment process throughout the basin.

ii) Data Availability

A basic impediment to successful sediment modeling is the lack of adequate input data. Model

calibration and verification require independent field data sets, preferably reflecting different field

conditions, for calibration and verification. The lack of proper sediment data in the Nile Basin has been

realized. The main reasons are lack of trained manpower, lack of laboratory facilities, lack of logistic

supports and unavailability of the facilities for maintaining sophisticated laboratory equipment and lack

of appropriate fund in the Nile Basin. The quality of the data is always affected by the condition of the

equipment and the methodology of data collection. In this respect, the authors of this chapter urge the

Nile Basin countries to give more attention to the collection of sediment data of good quality. This is

vital for the water resources management within the Nile Basin.

iii) Model Formulation

Similar to most of the sediment transport models, the SMS models incorporate certain simplifying

assumptions and approximations. Several difficulties are identified in model formulation, particularly

regarding the creation of a computational mesh. Such difficulties include:

• Finding an appropriate geo-referenced satellite image with a good resolution (for the base map) is difficult to obtain.

• Mesh creation requires a lot of experience and is time consuming.

• The eddy viscosity parameter is highly sensitive and affects the model convergence.

iv) Model Calibration

Regardless of the formulation chosen for the sediment model, model calibration is a key step in model

application. Adequate data regarding the flow and channel characteristics are of primary importance


for a successful model calibration. The SMS model was tested and calibrated using the data collected

from Tuti area where the two main tributaries meet at Khartoum City. The results obtained were fairly

acceptable, for more details see STWMC Second Annual Year Report, Khartoum, Sudan, 2003.

v) Model Application

The application of SMS model on steep rivers flow conditions (e.g. in Ethiopian, and Kenya cases

studies) require special professional experience and precaution regarding the concepts on which the

hydrodynamic and sediment transport models used in SMS are based on. On the other hand, large

numbers of cross-sections with short intervals are required to provide suitable results when SMS

models are implemented. Few problems related to model application can be listed as follows:

• Sudden falls in the longitudinal section (e.g. Awash River) complicate things and make the application of the model more difficult.

• Model convergence is one of the difficulties facing the model application.

• Intensive training on SMS application is required.

• The SMS model developers failed to provide technical support in time.

vi) Predictive Capability

The experience of SMS model application in the Nile River and Awash River showed difficulties in

proofing to be a truly predictive tool, particularly when dealing with river floods. The applications of

SMS model in Egypt and Sudan case studies raised some difficulties. For example, modeling of the

sediment transport in Aswan High Dam Reservoir showed that the predicted bed levels are higher

than the measured ones in the whole inlet zone, ElMoattassem et al. (2005).

Also, in the modeling of the sediment transport in the Blue Nile in Sudan, it is noted that both the

computed water level and bed level are higher than the measured ones for the entire selected reach,

for more details see STWMC Second Annual Year Report, Khartoum, Sudan, 2003. Regarding the

other case studies in Kenya, Tanzania and Ethiopia even more problems were faced during the

application of SMS model as a predictive tool.


There have been several constraints that have been faced by the STWMC Research Work; however,

the most important ones can be summarized as follows:

• The miscommunication with the Research Coordinator either through Emails, or Faxes ; or through the Focal Persons.

• Delay receiving the research seeds money, which has been decided in Dar Es Salaam Workshop.

• The difficult and slow procedure to obtain or purchase any equipment.


• Delay receiving equipment and computer purchased that had affected the data acquisition and compilation.

5.10) Conclusions

Most of available Sediment models, although all are based on well-known flow equations, select

sediment functions without clear justification for such selection. Therefore, professional experience

plays a major role in selecting a reasonable sediment function and, consequently, in modeling

sediment transport in water bodies.

The use of SMS to simulate sediment process in the Nile Basin River and Awash River within the

FRIEND/ Nile Project has encountered some difficulties and problems. This could be attributed to the

different topography of the countries involved and the suitable data availability.

Application of SMS model in the steep rivers as it can be seen from the Kenyan, Tanzanian and

Ethiopian case studies, showed difficulties in obtaining straightforward results. It requires high

professional experience on the nature of these rivers.

Modeling sediment is an effective tool in solving several water management problems; however, it

may be misleading if its results are not properly analyzed.

5.11) The Way Ahead

The first phase activities of The Sediment Transport and Watershed Management Component

(STWMC) will be reinforced and extended to include the modeling of erosion problems in the

catchment areas. It is therefore renamed to Erosion and Sediment Transport Component (ESTC) and

will be coordinated by Prof. Abdalla Abdelsalem Ahmed, UNESCO Chair in Water Resources, Sudan.

More countries from the Nile Basin will be asked to join this important research work. Based on the

pervious four years experience it is believed that the 2nd phase of the FRIEND/Nile Project will be

more successful.

The launching date of the project will be as soon as possible during the first half of 2006. The duration

of the project is four years. Immediate actions will be taken on the ground to start implementation of

the second phase.

5.12) Some Data and Results Listing

In this section some of the data and results obtained are listed.


• Table 5-3, Daily Observation of Sediment Data Awash River Basin: Awash River at Hombole Station (Ethiopia case).

Year Month Day G.H. Flow Daily

discharge

Sediment

concentration

Sediment

concentration

Daily

sediment

load

Sediment

loss

h (m) Q (m3/s) M3/d g/t mg/kg ppm g/m3 mg/ l t/d kg/km2 day

1986 12 19 0.22 3.51 303436.80 86.25 81.94 26.17 261.70

1987 7 22 1.64 43.16 3729110.40 636.00 604.20 2371.71 23717.10

1987 10 1 0.77 11.26 972950.40 253.08 240.43 246.23 2462.30

1988 3 3 0.16 2.88 249091.20 133.20 126.54 33.17 331.70

1988 5 27 0.18 3.28 283651.20 361.86 343.77 102.64 1026.40

1988 6 8 0.86 12.69 1096416.00 440.53 418.50 483.00 4830.00

1988 6 16 0.48 6.12 528768.00 821.34 780.27 321.20 3212.00

1988 6 16 0.43 5.48 473212.80 809.52 769.04 383.08 3830.80

1988 6 17 4.18 271.50 23457600.00 3006.99 2856.64 70536.77 705367.70

1988 6 17 0.53 6.82 589248.00 503.87 478.68 296.90 2969.00

1988 6 18 0.50 6.40 552960.00 717.51 681.63 396.74 3967.40

1988 6 20 0.64 8.52 736300.80 1453.19 1380.53 1069.98 10699.80

1988 6 20 0.65 8.69 750729.60 0.00 0.00 0.00

1988 6 21 0.52 6.68 577152.00 470.31 446.79 271.37 2713.70

1988 6 21 0.85 12.48 1078012.80 5528.45 5252.03 5959.74 59597.40

1988 6 26 0.98 15.43 1333152.00 15789.73 15000.24 21050.00 210500.00

1988 6 26 0.98 15.67 1353801.60 15678.57 14894.64 21225.67 212256.70

1988 7 8 3.96 239.44 20687529.60 16000.00 15200.00 331000.47 3310004.70

1988 7 10 2.96 124.22 10732694.40 24701.49 23466.42 265113.54 2651135.40

1988 7 11 2.96 124.22 10732694.40 24527.78 23301.39 263249.17 2632491.65

1988 7 12 2.74 105.02 9074073.60 9511.97 9036.37 86312.32 863123.20

1988 7 12 2.54 89.32 7717248.00 11680.53 11096.50 90141.55 901415.50

1988 7 12 2.12 61.42 5306947.20 10463.37 9940.20 55528.55 555285.50

1988 7 13 1.76 42.57 3677875.20 3913.04 3717.39 14391.67 143916.70

1988 7 13 2.54 89.32 7717248.00 11816.50 11225.68 91190.86 911908.60

1988 7 13 2.26 69.99 6046876.80 10254.88 9742.14 62010.00 620099.96

1988 7 14 1.70 39.84 3441830.40 3881.67 3687.58 13360.03 133600.30

1988 7 14 4.58 336.57 29079561.60 10763.88 10225.69 313008.91 3130089.10

1988 7 14 1.73 41.18 3558297.60 4452.02 4229.42 88480.40 884804.00

1988 7 15 1.52 32.34 2794089.60 16654.67 15821.94 46534.64 465346.40

1988 7 15 1.56 33.92 2930688.00 3795.82 3606.02 11124.85 111248.50

1988 7 16 1.56 33.92 2930601.60 4959.88 4711.89 14535.43 145354.30



discharge

Sediment

concentration

Sediment

concentration

Daily

sediment

load

Sediment

loss

1988 7 16 1.48 30.81 2661984.00 2243.34 2131.17 5971.74 59717.40

1988 7 17 2.00 54.64 4721155.20 3761.74 3573.65 17759.76 177597.56

1988 7 18 1.68 38.95 3365539.20 2595.26 2465.49 8734.43 87344.30

1988 7 18 1.66 38.08 3290371.20 2360.87 2242.83 7634.81 76348.06

1988 7 19 1.66 38.08 3290112.00 7563.11 7184.95 24885.44 248854.36

1988 7 19 1.48 30.81 2661638.40 4950.32 4702.80 12804.09 128040.90

1988 7 20 1.64 37.23 3216240.00 5756.38 5468.56 18513.91 185139.10

1988 7 20 1.58 34.73 3000412.80 248.45 236.03 745.45 7454.50

1988 7 20 1.53 32.73 2827872.00 2713.81 2578.12 11068.68 110686.80

1988 7 21 1.68 38.95 3365539.20 1857.35 1764.48 6250.99 62509.90

1988 7 21 1.58 34.73 3000412.80 2445.52 2323.24 7337.57 73375.70

1988 7 21 4.96 406.89 35155555.20 12602.02 11971.91 443030.84 4430308.35

1988 7 21 1.05 17.18 1484611.20 2071.43 1967.86 3075.27 30752.70

1988 7 22 1.46 30.06 2597011.20 4640.75 4408.71 24104.17 241041.70

1988 7 23 1.36 26.49 2288736.00 3383.60 3214.42 7744.17 77441.66

1988 7 24 1.90 49.37 4265913.60 7753.73 7366.04 33076.72 330767.20

1988 7 24 4.70 357.86 30918672.00 1523.80 1447.61 47113.87 471138.70

1988 7 25 1.96 52.49 4535481.60 2086.96 1982.61 28395.08 283950.80

1988 7 26 1.90 49.37 4265913.60 5947.76 5650.37 25370.96 253709.63

1988 7 26 1.94 51.44 4444416.00 4334.33 4117.61 19263.58 192635.83

1988 7 27 2.30 72.57 6270048.00 5098.90 4843.96 31970.37 319703.66

1988 7 28 1.66 38.08 3290371.20 4087.12 3882.76 13447.77 134477.70

1988 7 28 2.56 84.60 7309008.00 10197.50 9687.63 70533.10 705331.00

1988 7 29 2.28 71.27 6157728.00 7852.67 7460.04 48354.65 483546.46

1988 7 30 2.26 69.99 6046876.80 10682.58 10148.45 23896.78 238967.80

1988 7 31 3.28 155.93 13472438.40 10967.83 10419.44 147764.08 1477640.80

1988 7 31 2.20 66.23 5722185.60 4497.04 4272.19 25732.89 257328.90

1988 8 1 3.00 127.94 11053584.00 6569.84 6241.35 72621.24 726212.40

1988 8 1 3.46 175.84 15192230.40 7543.83 7166.63 114607.53 1146075.25

1988 8 1 3.80 217.72 18810576.00 4893.62 4648.94 92051.81 920518.10

1988 8 2 3.46 175.84 15192230.40 4109.31 3903.84 62429.58 624295.80

1988 8 3 2.90 118.78 10262678.40 4044.55 3842.32 41507.69 415076.85

1988 8 3 3.14 141.49 12224736.00 5187.34 4927.97 63413.90 634139.00

1988 8 4 3.54 185.18 15999379.20 10166.10 9657.79 162651.57 1626515.66

1988 8 4 5.54 531.25 45899827.20 8520.18 8094.17 391074.79 3910747.90

1988 8 6 3.76 212.49 18359049.60 6040.37 5738.35 110895.45 1108954.50



discharge

Sediment

concentration

Sediment

concentration

Daily

sediment

load

Sediment

loss

1988 8 6 3.77 213.79 18471456.00 17270.14 16406.63 319004.63 3190046.30

1988 8 6 3.78 215.09 18583948.80 5480.74 5206.70 101853.79 1018537.90

1988 8 7 2.99 127.00 10972800.00 2606.36 2476.04 28599.10 285991.03

1988 8 8 3.30 158.07 13656988.80 2681.98 2547.88 36627.70 366277.00

1988 8 14 1.56 33.92 2930601.60 4208.49 3998.07 12333.41 123334.10

1988 8 15 1.48 30.81 2661638.40 1990.23 1890.72 5297.27 52972.70

1988 8 18 4.18 271.50 23457600.00 3418.18 3247.27 80182.30 801823.00

1988 8 18 3.06 133.64 11546323.20 13333.33 12666.66 153950.94 1539509.40

1988 8 19 4.06 253.69 21919075.20 2761.04 2622.99 60513.74 605137.40

1988 8 19 1.53 32.73 2827785.60 2325.17 2208.91 6575.08 65750.80

1988 8 20 4.96 406.89 35155555.20 15527.27 14750.91 545869.80 5458698.00

1988 8 23 4.70 357.86 30918931.20 1252.20 1189.59 38716.53 387165.25

1988 8 24 3.98 242.25 20930140.80 85065.54 80812.26 1781450.83 17814508.30

1988 8 25 4.46 316.11 27311644.80 5746.62 5459.29 156949.64 1569496.40

1988 8 28 4.64 316.11 27311644.80 2816.72 2675.88 76929.12 769291.20

1988 8 28 3.52 182.18 15740697.60 2229.20 2117.74 35089.16 350891.60

1988 8 28 4.04 250.80 21669120.00 2579.62 2450.64 55890.09 558900.90

1988 8 30 3.52 182.18 15740697.60 2320.36 2204.34 36515.02 365150.20

1988 8 31 3.80 217.72 18810576.00 5866.45 5573.13 110351.30 1103513.00

1988 9 9 2.80 109.79 9485856.00 2782.75 2643.61 26396.70 263967.00

1989 2 1 0.30 3.64 314496.00 109.88 104.39 3.16 31.56

1989 4 25 1.33 23.90 2064873.60 12174.58 11565.85 25138.96 251389.60

1989 6 26 1.00 14.15 1222473.60 20941.35 19894.28 25600.25 256002.50

1989 7 19 2.26 57.94 5005929.60 2730.66 2594.13 13669.49 136694.90

1989 9 30 1.67 39.09 3377462.40 515.47 489.70 1740.98 17409.80

1989 12 8 0.28 4.45 384566.40 136.08 129.28 52.18 521.76

1990 2 21 0.44 6.25 539827.20 1277.84 1213.94 689.81 6898.10

1990 4 10 1.56 24.09 2081548.80 1483.96 1409.76 3088.93 30889.30

1990 5 27 0.26 2.21 190684.80 162.82 154.68 31.05 310.46

1999 5 14 0.13 2.89 249868.80 382.58

1999 7 10 2.58 78.29 6764083.20 18534.16

2002 10 5 0.70 11.05 954633.60 272.93



discharge

Sediment

concentration

Sediment

concentration

Daily

sediment

load

Sediment

loss

2003 4 6 2003 0.35 30240.00 349.91

• Table 5-4, Sediment flow data for Simiyu River outfall (Tanzania case).

Date Disharge Load (kg/day)7/12/2000 152.71 448.68/12/2000 122.775 180.339/12/2000 71.23 135.39

10/12/2000 97.04 2678.7711/12/2000 134.52 4399.1312/12/2000 139.4 4402.14

13/12/2000 91.02 86.51

• Table 5-5 10 – days-Mean Sediment Concentration for the Blue Nile at Different Locations (Sudan) case)

Mean Sediment Concentration (ppm)Month Period

El Deim Wad Alais Sennar

June II

III

1956 1172 -

July I

II

III

3361

3895

4335

2454

2724

3274

3200

4072

3612

August I

II

III

5660

3095

2948

2772

2859

2654

2790

2415

2154

Sept. I

II

III

3589

2305

1755

2588

1669

1028

1887

1500

1442

Oct. I

II

III

1294

591

317

990

946

-

900

-

-


• Table 5-6, Maximum Sediment Concentration during the flood season 2002 in different locations of the Blue Nile System (Sudan case).

Station Max. Concentration ppm Date

Roseires 21570 29/7/2002

Wad Elais 15044 6/8/2002

Sennar 12459 30/7/2002

Wad Medani 10106 16/7/2002

Gezira Canal 19472 31/7/2002

Managil Canal 21535 31/7/2002

• Table 5-7, Suspended Sediment Concentration before AHD (1929-1955) (Egypt case).

Months Suspended Sediment

Concentration (mg/l)

Weight of Sediment

(million Tons)

JAN 84.00 0.29

FEB. 60.00 0.15

MAR. 53.00 0.11

APR. 50.00 0.13

MAY 41.00 0.08

JUN. 44.00 0.09

JUL 278.00 1.81

AUG. 2820.00 56.22

SEP. 2497.00 56.44

OCT. 1034.00 15.54

NOV. 294.00 2.15

DEC. 121.00 0.53


Appendix

A

Management Team

COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE

Belgium Dr. Rudy Herman Senior Researcher Flanders Authority

Economy, Science and Innovation

Department, Ellips building

Koning Albert II-laan 35 bus

10 B- 1030 Brussel

Tel: 02/553 6001

Fax: 02/553 5981

E-mail: [email protected]

Egypt Dr. Radwan Al-

Weshah UNESCO Cairo Office

Garden City, Cairo 11541, Egypt ,

8 Abdel Rahman Fahmy St.,

Tel: 202/7945599, 7943036

Fax: 202/7945296

E-Mail: [email protected]

Egypt Prof. Mohamed

Abdel Motaleb


El-Qanater El-Khiria. P.O Box 13621,

Egypt.

Tel:202-2189437,2188787

Fax. No. 202-2184344


friend_np@ wrri.org.eg

Sudan Prof. Abdalla

Abdelsalam Ahmed

UNESCO Chair in Water Resources

Director

Water Resources, Sediment, Hyd.

Stru.

Tel:: +249-183-779599/ 786770/ 779540

Fax:: +249-9-12206586

+249-183-797758

E-Mail: [email protected],

Tanzania Prof. Felix Mtalo

Water Resources Engineering Dept.

P.O.Box35131. Dar es Salaam,

Tanzania

Tel:: 255-22-2410752,2410029

Fax:: 255-22-2410029


[email protected]

[email protected]

Kenya Prof F M Mutua Professor

University of Nairobi

P O Box 30197 00100 GPO

NAIROBI 00100

Tel:254 020 441045

Fax : 254 020 577373

Mobile : 254 722 835867




Egypt Dr. Abdel Aziz F. Zaki

UNESCO Cairo Office

Garden City, Cairo 11541, Egypt ,


Tel:: 202/7945599, 7943036

Fax:: 202/7945296



Appendix

B

Research Teams

• Table B- 1, Drought and Low Flow Analysis Research Team.


Egypt Prof. Ahmed

Hassan Fahmi

Water Resources Research Inst (WRRI), Head of

Hydrology Dept.

National Water Research Center Building, 3rd

Floor El-Qanater El-Khairiia, Qalyoubiia - Egypt

Tel:: +02 2188787

Mobile: +010 1537387

Fax: +02 2184344


Tanzania Dr. Raymond J.

Mngodo

Ministry of Water & Livestock Dev.

Principal Hydrologist I

Water Resources Division

P. O. Box 35066

Dar-Es-Salaam

Tel:; +255-22-2450838 Ext. 181

Mobile: +255-744-298330

Fax :: +255-22-2450005

E-mail: [email protected],

[email protected]

Sudan Dr. Muna M.

Mirghani

UNESCO Chair in Water Resources

Assistant Professor

P. O Box 1244 Khartoum

Tel:: +00 249 133 779599

Mobile: +00 249 9126 58768


Kenya Mr Julius

Njoroge Kabubi

Research Student

KMD/UoN

Dagoretti Corner

P O Box 30259 00100 GPO

NAIROBI

Tel: : 254 020 567880/9

Mobile :254 722 752228

Fax:: 254 020 576955


Kenya Prof F M Mutua Professor

University of Nairobi

P O Box 30197 00100 GPO

NAIROBI 00100

Tel:: 254 020 441045

Fax : 254 020 577373

Mobile : 254 722 835867




Belgium Prof. Willems

Patrick

Katholieke Universiteit Leuven – Hydraulics’

Laboratory

Postdoctoral Researcher & Lecturer

Kasteelpark Arenberg 40

B-3001 Heveriee (Levven)

BELGIUM

Tel:: +3216321658

Mobile: +320472993310

Fax:: +3216321989

E-mail:

[email protected]


• Table B- 2, Sediment Transport and Watershed Management Research Team.


Belgium Prof. Veerle Vanacker Katholieke Universiteit Leuven Department Of Geography 3 Place Louis Pasteur 1348 Louvain-la-Neuve Belgium

Tel: +32 10 478506 Fax: +32 10 472877 E-mail: [email protected]

Sudan Prof. Abdalla Abdelsalam

Ahmed

UNESCO Chair in Water

Resources

Director

Water Resources, Sediment, Hyd.

Stru.

Tel:: +249-183-779599/ 786770/

779540

Mobile: +249-9-12206586

Fax: +249-183-797758

E-Mail: [email protected],

[email protected]

Kenya Prof. Omyango Ogambo University of Nairobi

Focal Person

Tel:: +254-20-318262, +254-

72073422

Fax:: 254-20-245566


Egypt Prof. Mohamed El-Moatesem

El-Qotb

Professor in River Engineering

National Water Research Center

Ministry of Water Resources &

Irrigation

Tel:: +202-4466256

Mobile: +20-102623381

Fax::+202-4466256


Ethiopia Ms. Semunesh Golla Seyoum Hydrologist

Ministry of Water Resources

Addis Abeba, 1519, Ethiopia

Tel:: +2511610883, +2519123708

Fax: +251-9-611009

E-Mail:

[email protected]

Tanzania Prof. Mwanuzi Fredrick Focal Person

Sediment and Watershed

Management

University of Dar Es Salaam

P.O. Box 35131

Tel:: +255-22-2410029

Mobile: +255-741-292377

Fax: +255-22-2410029



• Table B- 3, Rainfall Runoff Modeling Research Team.


Belgium Prof. Willy Bauwens Vrije Universiteit Brussel Dpt. of Hydrology and Hydraulic Engineering Pleinlaan 2, 1050 Brussel, Belgium

Tel: 0032-2-6293038 Fax: 0032-2-6293022 E-mail: [email protected]

Sudan Dr. Kamaluddin

E.Bashar

Unesco Chair in Water

Resources

P.O. Box 1244, Khartoum,Sudan Tel: : 249-11-779599

Fax : 249-11-779604/797758


[email protected]

Kenya Prof. Francis Mutua

Department of meteorology,

P.O.Box 30197,Nairobi,kenya

Tel:: 254-2-449004 /577371

Fax:: 254-2-578343/ 577373

E-Mail:: [email protected]

EGYPT Dr.Mohamed Ali Sonbol

Water Resources Research Institute,

El-Qanater El-Khiria. P.O Box 13621, Egypt.

Tel:: 202-2189437,2188787

Fax.: 202-2184344



Ethiopia Mr. Deksyos Tarekejn Ministry of Water Resources

P.O Box 5673, Addis Ababa, Ethiopia.

Tel:: 251-1-610708

Fax:: 251-1-6110099


[email protected]

TANZANIA Prof. Felix Mtalo

Water Resources Engineering Dept.

P.O.Box35131. Dar es Salaam, Tanzania

Tel:: 255-22-2410752,2410029

Fax:: 255-22-2410029


[email protected]

[email protected]


• Table B- 4, Flood Frequency Analysis Research Team.


Sudan Prof. Gamal Mortada Abdo

Faculty of Eng. University of Khartoum Tel: : 249-11-771577

Mobile: 00294-12283976

E-Mail:

gabdo2000 @ yahoo.com

Kenya Dr. Alfred Opere

c/o Department of meteorology,

p.o.Box30197,Nairobi

Tel:: 254-2-449004Ex.2203

or 0722-858660

EGYPT Prof. Mohamed Abdel

Motaleb


El-Qanater El-Khiria. P.O Box 13621,

Egypt.

Tel:202-2189437,2188787

Fax. No. 202-2184344



Ethiopia Mr. Leuleseged Tadesse

P.O.Box: 5673, Addis Ababa (office)

P.O.Box: 31393, (private)

Tel: 251/1/625521, 611111ext 228

Fax : 251/1/ 630459/ 0885,611009


[email protected]

Tanzania Dr.Simon H. Mkhandi

Department of Water Resources

Engineering, University of Dar Es

Ssalaam, P.OBox 35131, Dar Es

Salaam.

Tel: 255-22-2410029

Fax: 255-22-2410029

E-Mail: mkhandi @ wrep.udsm.ac.tz


Appendix

C

List of the papers published in the FRIEND/Nile Conference The following is a list of keynote speeches and papers presented during the activities of the:

International Conference of UNESCO Flanders Fust Friend/Nile Project Towards A Better

Cooperation And The 5th Project Management Meeting And 9th Steering Committee Meeting,

12-15 November 2005.

Proceeding Page no.

CONFERENCE THEMES

Keynote Speeches

1 New Approaches and Perspectives in Flood Forecasting. By: Ezzio Todeini

22 Integrated water resources management: the importance of managing ecosystem goods and services. By: Patrick Meire,

Eric de Deckere, Jan Staes & Marleen Coenen

Hydrology of the Nile

23 Effect of Upstream Structures on Delta Deposit Progress in Aswan High Dam Reservoir By: Mohamed El Moatassem,

Tarek Abdel Aziz & Hossam El-Sersawy

34 Highlights on the Flooding Influence in the Flood Frequency Analysis in the Nile Basin By: Mohamed Sonbol, Gamal Abdo,

Patrick Willems & Mohamed Abdel Motaleb

47 Impact of Climate Change on the Hyrological Characteristics of Lake Nasser, By: Mamdouh Hassan

67 The Hydrological Interactions Between the White & Blue Niles at the Confluence Region By: Sohier Zaghloull, Mohamed

El Moatassem & Ahmed Rady

82 A User Friendly Forward and Inverse Modeling of Lake Nasser Reservoir By: Mohamed Megahed, Mahmoud Bakr,

Mohamed Abdel Motaleb & Mohamed El Fiki

95 Hydrodynamic Modeling of The Rosetta Branch in the Nile Delta By: Patrick Willems, Mona Radwan, Alaa El- Sadek &

Shaden Abdel Gawad

112 PHYSICO-Chemical Water Quality Modelling Of The Rosetta Branch in the Nile Delta By: Mona Radwan, Patrick Willems,

Alaa El- Sadek & Shaden Abdel Gawad

132 Promoting Water Eithiecs in the Nile Basin By: Magdy Hefny

Rainfall Runoff Analysis

159 Application of the Soil Water Assessment Tool (SWAT) in Simiyu River Catchement By: Deogratius Mulungu, Flix Mtalo

& Will Bauwens

171 Statistical and Trend Analysis of Rainfall and River Discharge: Yala River Basin, Kenya By: Githui Wairimu, Alfred Opere &

Willy Bauwens

182 A Precipitation Downscaling Model for the GCM Outputs Over the Nile Basin By: Mohamed Abdel Aty


201 Water Resources/Quality Modeling, Using Hydrological Simulation Program-Fortran (HSPF) and Watershed Modeling

System (WMS) By: Ahmed Salah & James Nelson 214 Rainfall-Runoff Modeling in Selected Catchments in the Lake Victoria Basins, By: Francis Mutua & Radwan Al- Weshah.

233 Appraisal Study to Select Suitable Rainfall-Runoff Model(s) for the Nile River Basin, By: Kamal eldin Bashar, Francis

Mutua, Deogratius Mulungu, Targi, Deksyos Tarekegn and Asaad Shamseldin

246 SMA Based Continuous Hydrologic Simulation of the Blue Nile, By: Kamal eldin Bashar, Abdel Aziz F. Zaki

256 Long Term Hydrologic Modeling for Simiyu Watershed, Tanzania Using Hydrologic Simulation Program-Fortran (HSPF)

By: Ahmed Salah, Deogratius Mulungu & Felix Mtalo

269 Analysis of Annual Rainfall Data in Jordan By: Ahmed Dahmsheh & Hafzullah Aksoy

276 Preparing Long –Term Watershed Simulations for the Nile River Basin, By: Christopher Smemoe & Lisa Adamson

293 Rainfall Runoff Modeling in Upper-Awash Sub-basin, By: Deksyos Tarekegn, Felix Mtalo & Radwan Al- Weshah.

306 Challenges of Modeling the Flows of the Nile River, By: Francis Mutua, Felix Mtalo & Willy Bauwens

Extreme Events

321 Low Flow Analysis Using Filter Generated Series for Lake Victoria Basin By: Julius Kabubi, Francis Mutua, Patrick Willems

& Raymond Mongodo.

336 Regional Flood Frequency Analysis for Northern Uganda Using the L-moment Approach By: Micheal Kizza, Henry Natale

& Albert Rugumayo

347 Flood Frequency Analysis of the Eastern Nile Rivers, By: Gamal Abdo, Mohamed Sonbol & Patrick Willems

360 At Site Flood Frequency Analysis for the Nile Equatorial Basins, By: Alfred Opere, Simon Makhandi & Patrick Willems

372 Comparison Between Annual Maximum and Peaks over Threshold Models for Flood Frequency Prediction By: Simon

Mkhandi, Alfred Opere & Patrick Willems

387 Analysis of the Return Periods of Low Flow Hazards in Egypt and Sudan By: Ahmed Hassan & Patrick Willems

399 QDF Relationships for Low-Flow Return Period Predication, By: Muna Mirghani, Patrick Willems & Julius Kabubi

407 Stastical Analysis of Dry Periods in Seasonal Rivers, By: Muna Mirghani & Patrick Willems

415 Regional Flood Frequency Analysis in the Nile Basin, By: Patrick Willems, Mohamed Sonbol, Gamal Abdo, Simon

Makhandi Alfred Opere, Leuleseged Taddesse, Mohamed Abdel Motaleb, Samir Farid, Abdel Aziz Zaki & Radwan Al-

Weshah.

432 Lake Nasser Flood Analysis By: Medhat Aziz & Sherine Ismail

453 River Nile Flood Forecasting Using Statistical Models By: Sherine Ismail

465 Nile River Different Flood Impacts By: Ahmed Moustafa, Mostafa Soliman, Medhat Aziz & Ehab Fatouh

477 At-Site and Regional Flood Frequency Analysis of the Upper Awash Sub-basin in the Ethiopian Plateau By: Leuleseged

Tadesse Mohamed Sonbol & Patrick Willems

492 Homogeneity Testing for Peak Flow in Catchments in the Equatorial Nile Basins By: Alfred Opere, Simon Mkhandi &

Patrick Willems

Sediment Transport and Watershed Management

505 Comparison of Two Different Transport Models to Predict SedimentTransport: Simiyu River, Tanzania, as a case study By:

Fredreick Mwanuzi & Verlee Vanacker

519 Modeling of Sedimentation Process in Aswan High Dam Reservo, By: Mohamed El- Moatassem, Tarek Abdel Aziz &

Hossam El-Sersawy.

534 Nile River Sediment Modeling: Challenges & Opportunities, By: Abdalla Abdelsalam , Hossam El-Sersawy, Verlee

Vanacker & Usama Ismail

545 Overview of Sediment Problems in Nile Basin, By: Abdalla Abdelsalam ,Verlee Vanacker, Usama Ismail & Radwan Al-

Weshah


557 Modelling Water and Sediment Fluxes in Steep River Channels, Case of Awash Basin By: Semunesh Golla, Hossam El-

Sersawy, Abdalla Abdelsalam &Verlee Vanacker

570 Limitations of Hydrodynamical Models with Limited Data Available Case Study: Sondu River Basin (Kenya) By: William

Ogembo & Benjamin Okellh

590 Sediment Deposition Control Towards Sustainability of Lake Nasser, By: Wael Khairy & Hussin El- Atfy

603 An Overview of Soil Erosion around Lake Victoria By: William Ogembo, Verlee Vanacker & Benjamin Okellh

Water Resources Management

612 Simulation of the High Aswan Dam using ResSim By: Mohamed Rami

629 Lake Nasser Flood and Drought Control Project (LNFDC) Utilization of Nile, Forecast System Capabilities & Foreseen

Climate Changes By: Mamdouh Antar

643 Flooding in the Rhine Catchment area – Past, Present and Future, By: Heribert Nacken

652 The River Nile:Cooperation Evolution and Lessons Learned By: Ahmed Bahaa

671 Watershed Modeling of Wadi Sudr and Wadi Al-Arbain in Sinai, Egypt, By: Mohamed Sonbol, Felix Mtalo, Medhat El-

Bihery & Mohamed Abdel Motaleb

683 Evaluation of Watershed Response Using GIS Based Hydrologic Model, By: Eman Hassan, Ali El- Bahrawy & Mohamed

Abdel Motaleb

695 Precopitation Recycling Over the Nile Basin By: Yasir Mohamed & H. Savenije

707 High Aswan Dam Reservoir Evaporation Losses By: Adel Makary & Nader Shafik

727 Integrated Water Resources Management in Nile Delta, Case Study: Mahmoudia and Meet Yazid By: Tarek Kotb,

Mohamed Mostafa, Yosry Khafagy & Gamal Fawzy

740 Impacts of Expoloration in Valley of the Kings on Flooding, By: Mohamed Saad, Hatem Abd El- Rahman, Sayed Ahmed &

Gamal Kotb

750 A Conceptual Model for Integrated Water Resources Management, By: Fathy Abdel Aziz, R. Abdel Azim & Ghoneim

Ghoneim

761 A Future Vision for Nile Basin Integrated-Water-Resources Management, Decision Support System By: Khaled Kheireldin

& Mohamed Abdel Motaleb

Poster

782 Feasibility of the Jonglei Canal Project after the Peace Treaty in Sudan: A Technical Perspective By: Sohier Zaghloull & Ali

El- Bahrawy

798 Artificial Groundwater Recharge as a potential Solution for Seawater Intrusion in El- Arish Area By: Gamal Kotb, Hatem

Mekhemer & Nadia El- Bahnasawy

808 Strengthening and Rehabilitation of Wadi- Al Asla Dike ,Jeddah Area, Kingdom of Saudi Arabia By: A. F. Khattab & Gamal

Kotb

821 Dynamic Properties and subsurface Structures of the North Western of the Nile Basin, Case study: Toshka Spillway

Control Barrage. By: Osama A. Raoof, Hatem Mekhemer, Ashraf El- Ashaal, Nadia El- Bahnasawy & Mohamed Abdel

Motaleb

835 Growth of Aquatic Weeds and Physico-Chemical Characteristics of Flowing Water in Khors Kalabsha , El Allaqi and

Toshka, Lake Nasser, Egypt, By: Magdy M. Hosny, Salwa Abou El- Ella & Mohamed Fawzy

854 Hydropower Generation in Egypt After the Operation of South Valley Project By: Nadia Abdel Salam, Mammdoh Abdel

Aziz, Medhat Aziz & Ahmed Zoubaa

868 Aquatic Weeds Management in Egyptian Channels By: Tarek El-Samman

881 Seepage Control Using Steel Fiber/Polymer Modified Concrete as Open Channel Liner By: Ashraf Ahmed

898 Multi-Purpose Mathematical Model By: Ahmed Negm


Appendix

D

List of Technical Reports

FRIEND/Nile Reports (2001 – 2006)

• For each FRIEND/Nile workshop, there is a technical report covering all implemented research activities and presentations during the workshop. This applies to the following workshops, namely:

1. Training Workshop on “Data Acquisition, Data Processing and Data Analysis”: Dar Es Salaam, Tanzania; 19-26 May 2002.

2. Sediment Transport Watershed Management Focal Persons Meeting, Khartoum, Khartoum, Sudan; 22-24 December 2002.

3. Flood Frequency Analysis Workshop, Cairo, Egypt; 1-3 April 2003.

4. The Rainfall-Runoff Modeling (RRM) and Sediment Transport and Watershed Management (STWM) Training Workshops, Alexandria, Egypt; 20-25 July 2003.

5. Drought and Low Flow Analysis (DLFA) Workshop, Nairobi, Kenya, 25-28 August 2003.

6. Flood Frequency Analysis Workshop, Sharm El-Shiekh, Egypt, 29 November–2 December 2003.

7. Sediment Transport and Watershed Management Workshop, Dar Es Salaam, Sudan, 2–6 December 2003.

8. Rainfall-Runoff Modeling (RRM) Theme Researchers Meeting, Dar Es Salaam, Tanzania, 5-9 January 2004.

9. Rainfall-Runoff Modeling (RRM) Theme Researchers Workshop, Addis Ababa, Ethiopia, 20-24 September 2004.

10. Drought and Low Flow Analysis (DLFA) Workshop, Nairobi, Kenya; 23-26 November 2004.

11. Flood Frequency Analysis (FFA) Workshop, Nairobi, Kenya; 26-29 November 2004.

12. Sediment Transport and Watershed Management (STWM) Workshop, Nairobi, Kenya; 26-29 November 2004.

13. FRIEND/Nile Workshops, Khartoum, Sudan; 25-30 July 2005.

• Proceeding of the International Conference for the FRIEND/Nile Project “Towards a Better Cooperation”, Sharm El Shiekh, Egypt; 12-14 November 2005.


• Progress report of the Components

A. Rainfall-Runoff Modeling Component 1. Annual progress report of year 2002. 2. Semi-Annual progress report of year 2003. 3. Annual progress report of year 2003. 4. Semi-Annual progress report of year 2004. 5. Annual progress report of year 2004. 6. Annual progress report of year 2005. 7. Completion report of the Rainfall-Runoff Modeling component.

B. Sediment Transport and Watershed Management Component

8. Semi-Annual progress report of year 2002. 9. Annual progress report of year 2002. 10. Semi-Annual progress report of year 2003. 11. Annual progress report of year 2003. 12. Semi-Annual progress report of year 2004. 13. Annual progress report of year 2004. 14. Annual progress report of year 2005. 15. Completion report of the Sediment Transport and Watershed Management component.

C. Flood Frequency Analysis Component 16. Semi-Annual progress report of year 2002. 17. Annual progress report of year 2002. 18. Semi-Annual progress report of year 2003. 19. Annual progress report of year 2003. 20. Semi-Annual progress report of year 2004. 21. Annual progress report of year 2004. 22. Annual progress report of year 2005. 23. Completion report of the Flood Frequency Analysis component.

D. Drought and Low Flow Analysis Component 24. Semi-Annual progress report of year 2004. 25. Annual progress report of year 2004. 26. Annual progress report of year 2005. 27. Completion report of the Drought and Low Flow Analysis component.


Contact UNESCO Rregional Office in Cairo / Water Program


Garden City, Cairo 11541, Egypt

Email: [email protected]

All reports and published materials of the project are available on the website: http://62.193.88.134/fn/

friend/nile - unesco · friend/nile final report i forward this is the final report of the capacity...

Documents