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.
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
Prof. Dr. Mohamed Abd-El Motaleb
FRIEND/Nile Overall Coordinator
Director
Water Resources Research Institute
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
FRIEND/Nile Final Report
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
FRIEND/Nile Final Report
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
FRIEND/Nile Final Report
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,
FRIEND/Nile Final Report
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
FRIEND/Nile Final Report
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;
FRIEND/Nile Final Report
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
FRIEND/Nile Final Report 2
• 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
FRIEND/Nile Final Report 3
• 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
FRIEND/Nile Final Report 4
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
FRIEND/Nile Final Report 5
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
FRIEND/Nile Final Report 6
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:
FRIEND/Nile Final Report 7
• 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.
FRIEND/Nile Final Report 8
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.
FRIEND/Nile Final Report 9
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
FRIEND/Nile Final Report 10
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
FRIEND/Nile Final Report 11
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.
FRIEND/Nile Final Report 12
• 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
FRIEND/Nile Final Report 13
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
FRIEND/Nile Final Report 14
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
FRIEND/Nile Final Report 15
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:
FRIEND/Nile Final Report 16
• 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.
FRIEND/Nile Final Report 17
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.
FRIEND/Nile Final Report 18
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
FRIEND/Nile Final Report 19
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.
FRIEND/Nile Final Report 20
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.
FRIEND/Nile Final Report 21
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
FRIEND/Nile Final Report 22
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
FRIEND/Nile Final Report 23
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
FRIEND/Nile Final Report 24
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.
FRIEND/Nile Final Report 25
• 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.
FRIEND/Nile Final Report 26
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
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
Return period (years)
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
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
23
4
56
7
Return period (years)
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
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
Return period (years)
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.
FRIEND/Nile Final Report 27
Regional flood frequency curves, Lake Victoria
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
Return period (years)
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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.
FRIEND/Nile Final Report 28
0
500
1000
1500
2000
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
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
0
500
1000
1500
2000
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
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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
GEV P-III LN3 GLOG EV1 GPAR EXPN Observed
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.
FRIEND/Nile Final Report 29
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
FRIEND/Nile Final Report 30
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
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5-ln(exceedance probability)
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
extreme value distribution
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
extreme value distribution
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).
FRIEND/Nile Final Report 31
0
100
200
300
400
500
600
700
800
900
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
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
extreme value distribution
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
FRIEND/Nile Final Report 32
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
FRIEND/Nile Final Report 33
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
0 1 2 3 4-ln(exceedance probability)
obse
rvat
ions
observationsextreme value distributionoptimal thresholdP-III, momEV1/Gumbel, mom
0
200
400
600
800
1000
1200
0 1 2 3 4-ln(exceedance probability)
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
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
obse
rvat
ions
observationsextreme value distributionoptimal thresholdextreme value distribution highest POT valuesEV1/Gumbel, MOMGEV, MOM
0
5
10
15
20
25
30
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
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).
FRIEND/Nile Final Report 34
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
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
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
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
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
FRIEND/Nile Final Report 35
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).
FRIEND/Nile Final Report 36
020
4060
80
100120
140160
180200
0 1 2 3-ln(exceedance probability)
obse
rvat
ions
observationsextreme value distributionoptimal thresholdEV1/Gumbel, MOMEV1/Gumbel, MLEV1/Gumbel, PWM
0
10
20
30
40
50
60
70
80
0 1 2 3-ln(exceedance probability)
obse
rvat
ions
observationsextreme value distributionoptimal thresholdEV1/Gumbel, MOMEV1/Gumbel, MLEV1/Gumbel, PWM
0
10
20
30
40
50
60
70
80
0 1 2 3-ln(exceedance probability)
obse
rvat
ions
observationsextreme value distributionoptimal thresholdEV1/Gumbel, MOMEV1/Gumbel, MLEV1/Gumbel, PWM
• 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
FRIEND/Nile Final Report 37
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
0 0.5 1 1.5 2 2.5 3 3.5-ln(exceedance probability)
obse
rvat
ions
EV1/Gumbel, MOMEV1/
-.
0
50
100
150
200
250
300
350
400
0 0.5 1 1.5 2 2.5 3 3.5-ln(exceedance probability)
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
0 0.5 1 1.5 2 2.5 3 3.5-ln(exceedance probability)
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.
FRIEND/Nile Final Report 38
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
Return period [years]
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.
FRIEND/Nile Final Report 39
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
Return period [years]
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
FRIEND/Nile Final Report 40
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
FRIEND/Nile Final Report 41
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.
FRIEND/Nile Final Report 42
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.
FRIEND/Nile Final Report 43
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)
SITESMEAN GEV GLOG LOGN P-IIIGPAR
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)
SITESMEAN GEV GLOG LOGN P-IIIGPAR
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)
SITESMEAN GEV GLOG LOGN P-IIIGPAR
R.GilaR.Pipor R. Baro
R.Akobo
R. Sobat
-
FRIEND/Nile Final Report 44
• 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.
FRIEND/Nile Final Report 45
• 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.
FRIEND/Nile Final Report 46
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.
FRIEND/Nile Final Report 47
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.
FRIEND/Nile Final Report 48
FRIEND/Nile Final Report 49
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
FRIEND/Nile Final Report 50
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.
FRIEND/Nile Final Report 51
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.
FRIEND/Nile Final Report 52
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.
FRIEND/Nile Final Report 53
• 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.
FRIEND/Nile Final Report 54
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
FRIEND/Nile Final Report 55
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
FRIEND/Nile Final Report 56
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
FRIEND/Nile Final Report 57
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
FRIEND/Nile Final Report 58
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
FRIEND/Nile Final Report 59
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
FRIEND/Nile Final Report 60
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.
FRIEND/Nile Final Report 61
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
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
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
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
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
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
Figure 3-4, HSPF conceptual hydrologic model.
FRIEND/Nile Final Report 62
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.
FRIEND/Nile Final Report 63
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.
FRIEND/Nile Final Report 64
• 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.
FRIEND/Nile Final Report 65
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
FRIEND/Nile Final Report 66
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
FRIEND/Nile Final Report 67
• 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.
FRIEND/Nile Final Report 68
• 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.
FRIEND/Nile Final Report 69
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
FRIEND/Nile Final Report 70
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).
FRIEND/Nile Final Report 71
• 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
FRIEND/Nile Final Report 72
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
FRIEND/Nile Final Report 73
• 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.
FRIEND/Nile Final Report 74
• 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.
FRIEND/Nile Final Report 75
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.
FRIEND/Nile Final Report 76
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
FRIEND/Nile Final Report 77
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.
FRIEND/Nile Final Report 78
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.
FRIEND/Nile Final Report 79
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.
3.7) Limitations and Constraints
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.
FRIEND/Nile Final Report 80
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.
FRIEND/Nile Final Report 81
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:
FRIEND/Nile Final Report 82
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:
FRIEND/Nile Final Report 83
• 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
FRIEND/Nile Final Report 84
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.
FRIEND/Nile Final Report 85
• 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.
FRIEND/Nile Final Report 86
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
FRIEND/Nile Final Report 87
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
FRIEND/Nile Final Report 88
• 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.
FRIEND/Nile Final Report 89
• 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;
FRIEND/Nile Final Report 90
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.
FRIEND/Nile Final Report 91
• 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.
FRIEND/Nile Final Report 92
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
0.1 1 10 100 1000 10000Aggregation level [days]
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
0.1 1 10 100 1000 10000Aggregation level [days]
Thre
shol
d ra
nk
threshold rankthreshold rank, fit
0
1
2
3
4
5
6
7
0.1 1 10 100 1000
Return period [years]
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.
FRIEND/Nile Final Report 93
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.
FRIEND/Nile Final Report 94
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
4.4.2.1) Introduction
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
FRIEND/Nile Final Report 95
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
4.4.3.1) Introduction
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
FRIEND/Nile Final Report 96
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
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
Dry
Spe
ll [d
ays]
observations
extreme value distribution
optimal threshold
50
75
100
125
150
175
200
225
250
1 10 100
Return period [years]
Dry
spe
lls (d
ays)
empirical data
calibrated distribution
• 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.
FRIEND/Nile Final Report 97
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
ln(1
/ D
isch
arge
[MC
M])
observations
extreme value distribution
optimal threshold
censored POT values
0
500
1000
1500
2000
2500
3000
1 10 100
Return period [years]
Dis
char
ge [M
CM
]
empirical data
calibrated distribution
• 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
0 0.5 1 1.5 2 2.5 3 3.5 4-ln(exceedance probability)
ln(1
/ D
isch
arge
[MC
M])
observations
extreme value distribution
optimal threshold
0
1000
2000
3000
4000
5000
6000
1 10 100
Return period [years]
Dis
char
ge [M
CM
]empirical data
calibrated distribution
• 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.
FRIEND/Nile Final Report 98
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.
FRIEND/Nile Final Report 99
4.4.3.5) Conclusions
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
4.4.4.1) Introduction
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.
FRIEND/Nile Final Report 100
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:
FRIEND/Nile Final Report 101
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
FRIEND/Nile Final Report 102
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.
FRIEND/Nile Final Report 103
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
RETURN PERIOD=200 YEARS
33 34 35 36 37 38 39 40 41 42-5
-4
-3
-2
-1
0
1
2
3
4
5
6
RETURN PERIOD=500 YEARS
• Figure 4-13, Distribution of annual drought indices corresponding to the 50, 200 and 500 year GEV return periods.
4.4.4.4) Conclusions
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
FRIEND/Nile Final Report 104
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
4.4.5.1) Introduction
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
FRIEND/Nile Final Report 105
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.
FRIEND/Nile Final Report 106
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.
4.4.5.3) Conclusions
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
FRIEND/Nile Final Report 107
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.
4.6) Limitations and Constraints
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.
FRIEND/Nile Final Report 108
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
FRIEND/Nile Final Report 109
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).
FRIEND/Nile Final Report 110
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.
FRIEND/Nile Final Report 111
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.
FRIEND/Nile Final Report 112
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.
FRIEND/Nile Final Report 113
• Figure 5-2, Schematic Diagram of the Nile River Natural Flows.
1000 1000
• Figure 5-3, Hydrograph of the Nile River.
FRIEND/Nile Final Report 114
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,
FRIEND/Nile Final Report 115
• 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),
FRIEND/Nile Final Report 116
• 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
FRIEND/Nile Final Report 117
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
FRIEND/Nile Final Report 118
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
FRIEND/Nile Final Report 119
• 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
FRIEND/Nile Final Report 120
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).
FRIEND/Nile Final Report 121
• 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.
FRIEND/Nile Final Report 122
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.
FRIEND/Nile Final Report 123
• 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).
FRIEND/Nile Final Report 124
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
FRIEND/Nile Final Report 125
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.
FRIEND/Nile Final Report 126
• 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
Comparison between Longtudinal Section for the High Aswan Dam Reservoir (HADR) at 2001
145.00
150.00
155.00
160.00
165.00
170.00
175.00
340360380400420440460480500Distance from High Dam (km)
Elev
atio
ns (m
)
2001- Measured
2001- Model
Flow Direction
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
Comparison between Longtudinal Section for the High Aswan Dam Reservoir (HADR) at 2001
145.00
150.00
155.00
160.00
165.00
170.00
175.00
340360380400420440460480500Distance from High Dam (km)
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
FRIEND/Nile Final Report 127
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.
FRIEND/Nile Final Report 128
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.
FRIEND/Nile Final Report 129
• 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)
FRIEND/Nile Final Report 130
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.
FRIEND/Nile Final Report 131
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).
FRIEND/Nile Final Report 132
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:
FRIEND/Nile Final Report 133
• 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.
FRIEND/Nile Final Report 134
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
FRIEND/Nile Final Report 135
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.
5.9) Limitations and Constraints
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.
FRIEND/Nile Final Report 136
• 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.
FRIEND/Nile Final Report 137
• 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
FRIEND/Nile Final Report 138
Year Month Day G.H. Flow Daily
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
FRIEND/Nile Final Report 139
Year Month Day G.H. Flow Daily
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
FRIEND/Nile Final Report 140
Year Month Day G.H. Flow Daily
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
-
-
FRIEND/Nile Final Report 141
• 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
FRIEND/Nile Final Report 150
FRIEND/Nile Final Report 143
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
Water Resources Research Institute
El-Qanater El-Khiria. P.O Box 13621,
Egypt.
Tel:202-2189437,2188787
Fax. No. 202-2184344
E-Mail: [email protected]
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
E-Mail: [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
E-mail: [email protected]
FRIEND/Nile Final Report 144
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
Egypt Dr. Abdel Aziz F. Zaki
UNESCO Cairo Office
Garden City, Cairo 11541, Egypt ,
8 Abdel Rahman Fahmy St.,
Tel:: 202/7945599, 7943036
Fax:: 202/7945296
E-Mail: [email protected]
FRIEND/Nile Final Report 145
Appendix
B
Research Teams
• Table B- 1, Drought and Low Flow Analysis Research Team.
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
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
E-mail: [email protected]
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],
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
E-mail: [email protected]
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
E-mail: [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
E-mail: [email protected]
FRIEND/Nile Final Report 146
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
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:
FRIEND/Nile Final Report 147
• Table B- 2, Sediment Transport and Watershed Management Research Team.
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
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],
Kenya Prof. Omyango Ogambo University of Nairobi
Focal Person
Tel:: +254-20-318262, +254-
72073422
Fax:: 254-20-245566
E-Mail: [email protected]
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
E-Mail: [email protected]
Ethiopia Ms. Semunesh Golla Seyoum Hydrologist
Ministry of Water Resources
Addis Abeba, 1519, Ethiopia
Tel:: +2511610883, +2519123708
Fax: +251-9-611009
E-Mail:
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
E-Mail: [email protected]
FRIEND/Nile Final Report 148
• Table B- 3, Rainfall Runoff Modeling Research Team.
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
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
E-Mail: [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
E-Mail: [email protected]
friend_np@ wrri.org.eg
Ethiopia Mr. Deksyos Tarekejn Ministry of Water Resources
P.O Box 5673, Addis Ababa, Ethiopia.
Tel:: 251-1-610708
Fax:: 251-1-6110099
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
E-Mail: [email protected]
FRIEND/Nile Final Report 149
• Table B- 4, Flood Frequency Analysis Research Team.
COUNTRY NAME ORGANIZATION TEL/FAX/E-MAIL/MOBILE
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
Water Resources Research Institute
El-Qanater El-Khiria. P.O Box 13621,
Egypt.
Tel:202-2189437,2188787
Fax. No. 202-2184344
E-Mail: [email protected]
friend_np@ wrri.org.eg
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
E-Mail: [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
FRIEND/Nile Final Report 150
FRIEND/Nile Final Report 151
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
FRIEND/Nile Final Report 152
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
FRIEND/Nile Final Report 153
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
FRIEND/Nile Final Report 154
FRIEND/Nile Final Report 155
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.
FRIEND/Nile Final Report 156
• 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.
FRIEND/Nile Final Report 157
Contact UNESCO Rregional Office in Cairo / Water Program
8 Abdel Rahman Fahmy St.,
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/