Walker, D 2018

Resilience of shallow groundwater resources and
their potential for use in small-scale irrigation: a

study in Ethiopia
David William Walker
School of Engineering
Newcastle University
A thesis presented for the degree of

Doctor of Philosophy
August 2018
i
ii
Abstract
Groundwater use for small-scale irrigation in sub-Saharan Africa is low, though is
expected to increase in the near future. There is currently limited understanding of
shallow groundwater resources, which are most likely to be exploited by poor rural
communities due to their accessibility. This PhD study aimed to determine the potential
for use of shallow groundwater for small-scale irrigation and the resilience of the
resources to increased abstraction, land-use change and climate variability.
Research was conducted principally at a study site in northwest Ethiopia with seasonal
rainfall and a predominance of rainfed agriculture. The shallow aquifer comprises a thin
weathered regolith above largely impermeable basalt. Hydrochemistry analyses
suggested little connection between the shallow aquifer and a deep fractured aquifer.
To fill gaps in formal hydrometeorological monitoring, a community-based monitoring

programme was initiated. Statistical comparisons confirmed that the datasets were of as
high or higher quality as those from formal networks, remote sensing and reanalyses.
A recharge assessment estimated annual recharge of 280-430 mm, confirming that a

sufficient renewable shallow groundwater resource is available for small-scale irrigation.
Four nested catchments were modelled using SHETRAN, a physically-based spatially-
distributed modelling program. The modelling identified the foot of hillslopes and narrow
valleys as showing the greatest potential for irrigated agriculture as groundwater in those
locations remained available and accessible for the longest periods. Potential future
scenarios were run in the SHETRAN models considering likely climate variability, land
use change and increasing abstraction. Around 35% of arable land in the modelled
catchments had shallow groundwater available throughout the dry season. During
simulated multi-year droughts, a significant percentage of arable land still had sufficient
groundwater available for irrigation of a second growing season. Conversion of pasture
and scrubland to cultivated land did not have a significant impact on water resources
while degradation of highlands to bareground had a positive impact. The severest impact
on water resources resulted from increased coverage of Eucalyptus. Notably, simulation
of increased abstraction and irrigation at smallholder levels had little impact on surface
and groundwater availability.
This study demonstrates the potential for greater exploitation of shallow groundwater for
small-scale irrigation by rural communities and the resilience of the resource to climate
variability, land use change and increasing abstraction.
i
For Robert and Kate, Jack and Mavis
Acknowledgements
Primarily I would like to thank Geoff Parkin and John Gowing for guiding me through
the PhD process and encouraging me to be ambitious while keeping me on the right path.
With their supervision, I have been able to move on from engineering consultancies and
into this much more gratifying world of research. I should also thank Geoff and John for
initially proposing the PhD project, winning the funding, and then selecting me to carry
it out.
I must acknowledge Newcastle University who funded this PhD through the Faculty of
Science, Agriculture and Engineering (SAgE) Doctoral Training Awards (DTA)
programme. I am also thankful to NERC/DfID who funded the initial AMGRAF catalyst
project (grant no. NE/L002019/1) under the UPGro programme that led to the PhD. I am
further grateful to Newcastle University for the award of the Harry Collinson Travel
Scholarship, to the Royal Geographical Society with IBG for the Dudley Stamp Memorial
Award, and to the International Association of Hydrogeologists for the John Day Bursary.
These grants enabled prolonged field visits, which were invaluable to the research project
and to my own development as a researcher. I am also indebted to the Rural Water Supply
Network (RWSN), especially Sean Furey, for funding attendance at their forum in Côte
d’Ivoire, which as well as freeing up project funds towards fieldwork, put me in touch
with a lot of important players in the field.
At Newcastle University I would like to thank numerous researchers: especially, Nathan

Forsythe for sharing data and wisdom and striving to incorporate me on many additional
projects; Steve Birkinshaw for tolerating my abundant SHETRAN queries and always
answering them without complaining; Russell Adams, Greg O’Donnell, and Stephen
Blenkinsop for their advice on many things relating to climate and hydrology, and; Liz
Lewis for initially showing me the ropes. I am also grateful to Peter Cunningham for
assistance in the fabrication of field equipment as well as David Race and Jane Davis for
providing hydrochemistry sampling equipment and for analysing the duplicate samples
that came back to Newcastle.
In Ethiopia, I am very grateful to International Water Management Institute (IWMI) for

support and guidance during field visits, especially to Petra Schmitter and Alemseged
Haile. At Addis Ababa University, I would like to thank Seifu Kebede for lending the
Radon-222 equipment and for conducting the stable isotope analysis. I am grateful to the
Ministry of Water Irrigation and Electricity (MoWIE), especially Surafel Mamo, for the
iii
provision of river flow records and lively discussion, and to the National Meteorological
Agency Bahir Dar office and several local observers for meteorological records and
information. From Bahir Dar University I most grateful to students Debebe Yiak, Abdu
Yimam and Melaku Tesema for their accompaniment in the field and to their supervisor
Seifu Tilahun for useful discussions and for lending the current meter to enable flow
gauging. In Dangila, I cannot thank enough Addisu Mulu and his colleagues at the
Woreda Office and to the Dangesheta Agricultural Office for sharing their time to be
involved in the project. I would also like to acknowledge the Dangila Town Water Supply
Office who granted me access to their deep boreholes. Finally in Dangila, I am very
grateful for the presence of Nara McCray and Norit Admasu who provided valuable
assistance with translation in the field and were a pleasure to spend time with. Perhaps
most importantly, I am forever indebted to the Dangila rural community, particularly in
Dangesheta kebele where I spent most of my time, especially to Assaye Molla, the other
citizen science observers and their families. The welcome and generosity of the people
throughout the woreda were endless, even if I’d eaten four times already that day and it
was only 9am.
In South Africa, I am particularly grateful to Nebo Jovanovic at the South African Council
for Scientific and Industrial Research (CSIR) for including me on the Limpopo project
and for providing office space, accompaniment in the field, and collaborations on papers.
Therefore, I must acknowledge the Water Research Commission (WRC) for funding this
CSIR project no. K5/2426. In Giyani, I am very grateful to Tebogo Lebea of the Limpopo
Department of Agriculture and Rural Development for sharing his time during field visits
and to the local farmers for providing useful information, especially Patrick Sekhula and
Adam Mabunda who permitted the use of their farms as focus farms where much of the
research was conducted. At Stellenbosch University I would like to thank Daniell du
Preez for assistance in the field and for sharing his geological knowledge and at the
University of the Witwatersrand, Johannesburg, I would like to thank Tamiru Abiye for
conducting the stable isotope analysis and for useful discussions. I am also grateful to the
Department of Water and Sanitation (DWS) for conducting the geophysical surveys. In
Pretoria, I am again indebted to International Water Management Institute for providing
me with an office and for inspirational discussions with Karen Villholth, Girma Ebrahim
and Barbara van Koppen.
iv
I would like to acknowledge the University of Birmingham Hydrogeology MSc team,
especially John Tellam, for inspiring me to desire a career in hydrogeology and for
providing me with the background theoretical knowledge to be able to pursue it.
I am grateful to the many researchers I have come across at the various conferences where
I have attended and presented for useful discussions about the project and for constructive
criticisms. Similarly, I am grateful to the anonymous reviewers of my journal submissions
for their suggestions and comments, and for helping me develop a thick skin.
Within the School of Engineering, I owe thanks to many fellow PhD students. Firstly,
anyone on adjacent desks, Chris Teasdale, Doug Richardson, Saad Almutairi, who have
to cope with my queries about the different colours on geological maps that I have
problems with. Then I also want to thank for their support in addition to the previous
three, Sam Mahaffey, Golnaz Jahanshahi, Amelie Ott and Maria-Valasia Peppa, as well
as Eleanor Starkey, Xiaodong Ming and David Pritchard for technical assistance, Mhari
Barnes and Dominic Carver for including me in their fieldwork during which I gained
valuable skills, and Maria Pregnolato for being an inspiration to us all on how a PhD
should be achieved.
Almost lastly, I am forever grateful to my mum and dad for putting up with my
meandering journey that led me to the PhD and then fully supporting me in every way as
it has progressed. Also, to my sister Joanna and to Claire for always being there and
keeping my feet on the ground.
Finally, I would like to thank Magdalena; officially for her invaluable assistance in
writing the python scripts to process the SHETRAN outputs, but mostly for inspiring me
to work that bit harder and be that bit brainier, and for providing all the distraction I need.
v
Related publications
The following publications contain work related to or derived from this thesis:
Walker, D., Jovanovic, N., Lebea, T., Bugan, R.D.H., Abiye, T.A., du Preez, D., Parkin,
G. and Gowing, J. (2018) 'Alluvial aquifer characterisation and resource
assessment of the Molototsi sand river, Limpopo, South Africa'. Under review.
Walker, D., Smigaj, M. and Jovanovic, N. (2018) 'Ephemeral sand river flow detection
from satellite optical remote sensing'. Under review.
Walker, D., Parkin, G., Schmitter, P., Gowing, J., Tilahun, S.A., Haile, A.T. and Yimam,
A.Y. (2017) 'Insights from a multi-method recharge estimation comparison study',
Groundwater. In-press.
Walker, D. (2016) 'Properties of shallow thin regolith aquifers in sub-Saharan Africa: a

case study from northwest Ethiopia', 7th RWSN Forum "Water for Everyone".
Abidjan, Côte d’Ivoire, Nov-Dec 2016. Available at:
https://rwsnforum7.files.wordpress.com/2016/12/full_paper_0061_submitter_01
66_walker_david-rev1.pdf
Walker, D., Forsythe, N., Parkin, G. and Gowing, J. (2016) 'Filling the observational void:
Scientific value and quantitative validation of hydrometeorological data from a
community-based monitoring programme', Journal of Hydrology, 538, pp. 713-
725.
Gowing, J., Parkin, G., Forsythe, N., Walker, D., Haile, A.T. and Alamirew, D. (2016)
'Shallow groundwater in sub-Saharan Africa: neglected opportunity for
sustainable intensification of small-scale agriculture?', Hydrol. Earth Syst. Sci.
Discuss., 2016, pp. 1-33.
vi
Contents
Abstract ............................................................................................................................. i
Acknowledgements.........................................................................................................iii
Related publications ....................................................................................................... vi
Contents ......................................................................................................................... vii
List of figures ................................................................................................................ xvi
List of tables ................................................................................................................. xxv
List of abbreviations and acronyms ........................................................................ xxvii
Chapter 1. Introduction ............................................................................................. 1
1.0 Chapter summary .................................................................................................. 1
1.1 Context .................................................................................................................. 1
1.2 Aims ...................................................................................................................... 5
1.3 Study site ............................................................................................................... 6
1.4 Methodology ......................................................................................................... 7
1.5 Thesis outline ........................................................................................................ 8
Chapter 2. Literature review................................................................................... 10
2.0 Chapter overview ................................................................................................ 10
2.1 The role of groundwater in Asia’s “Green Revolution” ..................................... 10
2.2 The potential for such a green revolution in sub-Saharan Africa ....................... 13
2.3 African hydrogeology ......................................................................................... 14
2.4 The African climate ............................................................................................ 16
2.5 Land use change .................................................................................................. 23
2.6 Research gap ....................................................................................................... 24
Chapter 3. Study site background .......................................................................... 29
3.1 Site selection ....................................................................................................... 29
vii
3.2 Country context ................................................................................................... 31
3.3 Site description.................................................................................................... 32
3.4 Agriculture .......................................................................................................... 36
3.5 Climate ................................................................................................................ 38
3.6 Geology ............................................................................................................... 39
3.7 Regolith hydrogeology........................................................................................ 42
Chapter 4. Field investigations and development of conceptual model .............. 45
4.1 Field visits ........................................................................................................... 45
4.2 Aims .................................................................................................................... 45
4.3 Study site observations........................................................................................ 45
4.4 Hydrology ........................................................................................................... 47
4.5 Dambos ............................................................................................................... 48
4.6 Geology ............................................................................................................... 51
4.7 Hydrogeology...................................................................................................... 53
4.8 Pumping tests ...................................................................................................... 55
4.9 Hydrochemistry................................................................................................... 56
4.10 Conceptual model ............................................................................................... 57
4.11 Conclusions ......................................................................................................... 60
Chapter 5. Filling the observational void: Scientific value and quantitative

validation of hydrometeorological data from a community-based monitoring
programme 61
5.1 Introduction ......................................................................................................... 62
5.2 Sub-Saharan Africa context ................................................................................ 63
5.3 Community-based monitoring ............................................................................ 65
5.4 Project context..................................................................................................... 67

5.4.1 AMGRAF research project ........................................................................................ 67
viii
5.4.2 Study area .................................................................................................................. 68
5.5 Data analysis methods ......................................................................................... 69

5.5.1 Sources of error ......................................................................................................... 69
5.5.2 Quality control ........................................................................................................... 70
5.5.3 Validation of hydrometeorological data .................................................................... 71
5.5.4 Behavioural differences in data ................................................................................. 73
5.6 Rainfall ................................................................................................................ 73

5.6.1 Formal ground observations ...................................................................................... 73
5.6.2 Community data ........................................................................................................ 74
5.6.3 Gridded datasets ........................................................................................................ 74
5.6.4 Performance of alternative rainfall sources ............................................................... 75
5.7 River flow ........................................................................................................... 77

5.7.1 Formal observations .................................................................................................. 77
5.7.2 Community data ........................................................................................................ 77
5.7.3 Performance of alternative river flow sources ........................................................... 78
5.8 Groundwater level data ....................................................................................... 80

5.8.1 Formal observations .................................................................................................. 80
5.8.2 Community data ........................................................................................................ 80
5.8.3 Performance of alternative groundwater level sources .............................................. 81
5.9 Discussion ........................................................................................................... 83

5.9.1 Qualitative and quantitative value of community-based monitoring observations ... 83
5.9.2 Recommendations for ensuring quality data production ........................................... 84
5.9.3 Wider application of community-based hydrometeorological monitoring................ 85
5.10 Conclusions ......................................................................................................... 86
Chapter 6. Insights from a multi-method recharge estimation comparison study

87
6.1 Introduction ......................................................................................................... 88
6.2 Groundwater recharge ......................................................................................... 90
6.3 Recharge estimation methods ............................................................................. 91
6.4 Study area ............................................................................................................ 91

6.4.1 General description .................................................................................................... 91
6.4.2 Climate ...................................................................................................................... 92
ix
6.4.3 Hydrogeology............................................................................................................. 93
6.5 Data used in the study ......................................................................................... 93
6.6 Recharge estimation methodologies ................................................................... 95

6.6.1 Empirical method ....................................................................................................... 95
6.6.2 Streamflow hydrograph analysis (three methods) ...................................................... 96
6.6.3 Soil moisture balance (SMB) ..................................................................................... 97
6.6.4 Basin water balance ................................................................................................... 97
6.6.5 Chloride mass balance (CMB) ................................................................................... 98
6.6.6 Water table fluctuation (WTF) ................................................................................... 98
6.6.7 Rainfall infiltration breakthrough (RIB) .................................................................... 99
6.6.8 Physically-based modelling ....................................................................................... 99
6.7 Recharge results ................................................................................................ 100
6.8 Sensitivity analyses ........................................................................................... 102
6.9 Discussion ......................................................................................................... 104

6.9.1 Reasons for the range in results ............................................................................... 104
6.9.2 Insights gained on the conceptual model ................................................................. 107
6.9.3 Recharge estimate for Dangila woreda .................................................................... 108
6.10 Conclusions ....................................................................................................... 109
Chapter 7. Modelling the shallow aquifer............................................................ 110
7.0 Chapter overview .............................................................................................. 110
7.1 Catchment modelling ........................................................................................ 110

7.1.1 Introduction .............................................................................................................. 110
7.1.2 Purpose of the modelling ......................................................................................... 111
7.1.3 Model types .............................................................................................................. 111
7.1.4 SHETRAN ............................................................................................................... 113
7.1.5 Methodology ............................................................................................................ 117
7.1.6 Similar modelling studies ......................................................................................... 117
7.2 Model construction ........................................................................................... 119

7.2.1 Modelled catchments and resolution ........................................................................ 119
7.2.2 Digital elevation model (DEM) and catchment masks ............................................ 120
7.2.3 Meteorology ............................................................................................................. 123
7.2.4 Soils and geology ..................................................................................................... 124
7.2.5 Land use land cover (LULC) ................................................................................... 127
7.3 Model calibration .............................................................................................. 128

x
7.3.1 Introduction ............................................................................................................. 128
7.3.2 Calibration data ....................................................................................................... 128
7.3.3 Calibration methodology ......................................................................................... 130
7.3.4 Calibration results .................................................................................................... 132
7.3.5 Optimum parameter values ...................................................................................... 135
7.4 Improved hydrogeological understanding from the calibration ........................ 137

7.4.1 Sensitivity to certain parameters.............................................................................. 137
7.4.2 Absence of a clay-rich low hydraulic conductivity layer ........................................ 137
7.4.3 Absence of fractured bedrock high hydraulic conductivity layer ............................ 138
7.4.4 Hydrogeological importance of the basal basalt bedrock ........................................ 138
7.4.5 Absence of leakage to a deeper aquifer ................................................................... 138
7.4.6 Evapotranspiration from groundwater ..................................................................... 139
7.5 Mapping the potential of shallow groundwater for irrigation use..................... 139
7.5.1 Methodology............................................................................................................ 139
7.5.2 Groundwater potential maps .................................................................................... 140
7.6 Discussion and conclusions .............................................................................. 145
Chapter 8. Resilience of shallow groundwater resources ................................... 149
8.1 Introduction ....................................................................................................... 149

8.1.1 Context .................................................................................................................... 149
8.1.2 Purpose of modelling ............................................................................................... 150
8.1.3 Methodology............................................................................................................ 150
8.1.4 Similar modelling studies ........................................................................................ 151
8.2 Identifying potential future scenarios ............................................................... 151

8.2.1 Climate variability ................................................................................................... 151
8.2.2 Climate change ........................................................................................................ 152
8.2.3 Land use land cover (LULC) change ...................................................................... 153
8.2.4 Groundwater abstraction.......................................................................................... 157
8.2.5 Khat production ....................................................................................................... 158
8.3 Historical drought analysis ................................................................................ 160

8.3.1 Purpose, definition and methodology ...................................................................... 160
8.3.2 Available hydrometeorological data ........................................................................ 161
8.3.3 SPI (standardised precipitation index) ..................................................................... 162
8.3.4 SPEI (standardised precipitation-evapotranspiration index) ................................... 163
8.3.5 River flow totals ...................................................................................................... 164
xi
8.3.6 Rainfed crop growing season length ........................................................................ 164
8.3.7 NDVI (normalized difference vegetation index) ..................................................... 165
8.3.8 LST (land surface temperatures) .............................................................................. 166
8.3.9 Annual rainfall totals ................................................................................................ 167
8.3.10 Annual potential evapotranspiration (PET) totals................................................ 168
8.3.11 SPI (standardised precipitation index) and SRA (standardised rainfall anomaly)
from other published studies ................................................................................................. 168
8.3.12 Local community perception ............................................................................... 169
8.3.13 Synthesis of results .............................................................................................. 170
8.3.14 Comparison between Dangila and Bahir Dar drought analyses .......................... 174
8.4 Generation of synthetic meteorological time series .......................................... 174

8.4.1 Purpose ..................................................................................................................... 174
8.4.2 Comparison between Dangila and Bahir Dar rainfall .............................................. 174
8.4.3 Methodology ............................................................................................................ 175
8.5 SHETRAN modelling ....................................................................................... 176

8.5.1 Initial/preceding conditions ...................................................................................... 176
8.5.2 Modelled catchments ............................................................................................... 178
8.6 Measuring potential impacts ............................................................................. 178

8.6.1 Additional irrigated growing seasons....................................................................... 178
8.6.2 Groundwater availability .......................................................................................... 179
8.6.3 Surface water availability ......................................................................................... 181
8.7 Climate variability............................................................................................. 181
8.8 Land use land cover (LULC) change ................................................................ 183

8.8.1 Simulated LULC scenarios ...................................................................................... 183
8.8.2 Results for simulated LULC scenarios .................................................................... 185
8.9 Increased groundwater abstraction.................................................................... 190

8.9.1 Simulated abstraction scenarios ............................................................................... 190
8.9.2 Results for simulated abstraction scenarios ............................................................. 191
8.10 Discussion and conclusions .............................................................................. 192
Chapter 9. Conclusions, including recommendations, transferability, and future

work 195
9.1 Conclusions ....................................................................................................... 195
9.2 Recommendations ............................................................................................. 199

xii
9.2.1 Potential for the productive use of shallow groundwater ........................................ 199
9.2.2 Abstraction strategy ................................................................................................. 200
9.2.3 Well design .............................................................................................................. 201
9.2.4 Floodplain cultivation .............................................................................................. 201
9.2.5 Conversion from Eucalyptus to Pine ....................................................................... 203
9.3 Transferability of results ................................................................................... 204

9.3.1 Transferability within Ethiopia ................................................................................ 204
9.3.2 Wider transferability within sub-Saharan Africa ..................................................... 205
9.4 Transferability of methodology ........................................................................ 206
9.5 Limitations of the study .................................................................................... 207
9.6 Future work ....................................................................................................... 213
References .................................................................................................................... 215
Appendix A: Properties of shallow thin regolith aquifers in sub-Saharan Africa:

a case study from northwest Ethiopia ....................................................................... 278
A-1 Introduction ....................................................................................................... 278
A-2 Testing Locations .............................................................................................. 279
A-3 Methodology ..................................................................................................... 280
A-4 Analysis ............................................................................................................. 281
A-5 Results ............................................................................................................... 282
A-6 Conclusions ....................................................................................................... 285
Sub-Appendix A1 – Plotted pumping test data ............................................................. 286
Appendix B. Hydrochemistry field investigations at Dangila woreda .............. 292
B-1 Introduction ....................................................................................................... 292
B-2 Purpose of sampling and in-situ testing ............................................................ 292
B-3 Sampling locations ............................................................................................ 293
B-4 Sampling methodology ..................................................................................... 294
B-5 Major ion hydrochemistry ................................................................................. 296

B-5-1 Analysis ............................................................................................................... 296
B-5-2 Quality assurance ................................................................................................ 296
B-5-3 Results ................................................................................................................. 298
xiii
B-6 Stable isotope .................................................................................................... 303
B-6-1 Analysis ............................................................................................................... 303
B-6-2 Quality assurance ................................................................................................. 303
B-6-3 Results ................................................................................................................. 303
B-7 In-situ testing..................................................................................................... 305

B-7-1 Methodology ........................................................................................................ 305
B-7-2 Results ................................................................................................................. 305
B-8 In-situ radon-222 testing ................................................................................... 306

B-8-1 Methodology ........................................................................................................ 306
B-8-2 Results ................................................................................................................. 307
B-9 Summary ........................................................................................................... 309
Appendix C. Filling the observational void: scientific value and quantitative

validation of hydrometeorological data from a community-based monitoring
programme – Supplementary material ..................................................................... 311
C-1 AMGRAF research project ............................................................................... 311
C-2 Dangesheta community-based monitoring network ......................................... 312
C-3 Formal ground observations – rainfall .............................................................. 313
C-4 Gridded datasets – rainfall ................................................................................ 315
C-5 River flow ......................................................................................................... 316
C-6 Formal observations – groundwater .................................................................. 317
Appendix D. Insights from a multi-method recharge estimation comparison

study – Supporting Information ................................................................................ 318
D-1 Non-selection of certain recharge estimation methods ..................................... 318
D-2 Recharge estimates for the study site from large-scale mapping and modelling
320
D-3 Field data not described in the main text [of Chapter 6] that was used to develop
the conceptual model and to parameterise models........................................................ 321
D-4 Empirical method .............................................................................................. 324
D-5 Streamflow hydrograph methods ...................................................................... 331
D-6 Soil moisture balance (SMB) ............................................................................ 334
xiv
D-7 Basin water balance .......................................................................................... 337
D-8 Chloride mass balance (CMB) .......................................................................... 338
D-9 Water table fluctuation (WTF) and rainfall infiltration breakthrough (RIB) .... 339
D-10 SHETRAN modelling ....................................................................................... 341
D-11 Comparison of recharge results from the three nested catchments ................... 344
D-12 Insights gained on the conceptual model in other recharge studies .................. 345
D-13 Annual recharge time series .............................................................................. 346
Appendix E. Bahir Dar drought analysis and rainfall comparison .................. 348
E-1 Bahir Dar drought analysis................................................................................ 348
E-2 Comparison between Dangila and Bahir Dar rainfall ....................................... 350
Appendix F. Water point surveys ........................................................................ 353
xv
List of figures
Figure 1-1. Comparison between South and East Asia, and Africa of percentage area
irrigated by groundwater (adapted from Siebert et al. (2010)). ............................ 2
Figure 1-2. Contrasting response and recovery behaviour of groundwater and surface
water to drought (from Robins and Fergusson (2014))......................................... 2
Figure 1-3. The aquifer productivity map across Africa, estimated depth to groundwater,
and population density map (adapted from MacDonald et al. (2011)). ................ 4
Figure 1-4. Study site location. ......................................................................................... 7
Figure 1-5. Methodology flow chart. ................................................................................ 8
Figure 2-1. The United Nations Sustainable Development Goals (SDGs) (UN, 2015). 12
Figure 2-2. The hydrogeological environments of SSA (from MacDonald and Davies
(2000))................................................................................................................. 14
Figure 2-3. Köppen-Geiger climate type map of Africa with description of climate
symbols and defining criteria (from Peel et al. (2007)). ..................................... 17
Figure 2-4. Mean annual precipitation map of Africa (from Kitamirike (2008)). .......... 19
Figure 2-5. Mean linear trends in annual temperature (oC per century) and annual
rainfall (% per century) over the period 1901-1995 (from Hulme et al. (2001)).
............................................................................................................................. 21
Figure 3-1. Ethiopia hydrogeological map (adapted from BGS (2018)). ....................... 30
Figure 3-2. Ethiopia precipitation map (adapted from BGS (2018)). ............................. 31
Figure 3-3. The link between annual rainfall and GDP growth in Ethiopia. Above
average rainfall since 2006 has led to consistent GDP growth of around 10%
(REACH, 2015). ................................................................................................. 32
Figure 3-4. Location of Dangila woreda field site showing woreda boundary (blue) and
position within Ethiopia (Geographical map of Ethiopia from Ezilon (2017)). . 33
Figure 3-5. Geography of Dangila woreda study site (image source: Google.Earth;
Imagery ©2017 DigitalGlobe), position within the Lake Tana Basin (map from
MoWR), and position within the Nile Basin (map from PRI (2017)). ............... 35
xvi
Figure 3-6. Monthly median, 10th and 90th percentile rainfall, and mean maximum and
minimum temperatures as measured (since 1987) by the NMA at the Dangila
weather station. ................................................................................................... 38
Figure 3-7. Generalised Cenozoic geology of the regions bordering the Main Ethiopian
Rift (MER) system (simplified from the Geological Map of Ethiopia). Circled
“A” is the location of Addis Ababa and “B” is the Dangila woreda study site.
From Prave et al. (2016). Note that there are some discrepancies between the
geological units, their ages and positions, on this map, the Figure 3-8 map, and
the descriptions within the text. This reflects the lack of agreement in the
published literature. ............................................................................................. 40
Figure 3-8. Geological map of the northern part of the Ethiopian plateau depicting the
distribution of the volcanic rocks, shield volcanoes and Strombolian cones (from
Ayalew (2011)). MER = Main Ethiopian Rift. ................................................... 41
Figure 3-9. Typical weathering profile developed upon crystalline rocks (from Acworth
(1987)) ................................................................................................................. 43
Figure 4-1. Typical wet season and dry season scenery in Dangila woreda of floodplain
pasture surrounded by low hills with small (<1 ha) individual plots of rainfed
agriculture. .......................................................................................................... 46
Figure 4-2. The flowing Amen River at the gauge location in Dangila town (left) and
the stagnant or dry Amen River (right) approximately 5 km downstream shortly
after reaching the extensive Kilti floodplain. ...................................................... 47
Figure 4-3. “Hanging” floodplain basins above the Brante river valley and the locations
of the community monitored raingauge and hand-dug wells, Dangesheta kebele.
Image source: Google.Earth; Imagery ©2015 DigitalGlobe). ............................ 51
Figure 4-4. Vesicular basalt boulders in the Brante riverbed, Dangesheta kebele (left).
As-dug weathered basalt beside a well under construction on the edge of the
Brante river floodplain. ....................................................................................... 51
Figure 4-5. Very deep (approximately 15 m) exposure of regolith in a gully on a large

slope, which leads to the Amen River, Kuandisha kebele (left). Massive and
fractured basalt bedrock exposed in the bed of the Gizani River, Sehara kebele.
Note the low (~1 m) river banks in this floodplain. ............................................ 52
xvii
Figure 4-6. Visible water table in the banks of the Amen River, Dangila kebele (left).
Massive basalt boulders visible within regolith in a cutting for an irrigation canal
near Giza. ............................................................................................................ 54
Figure 4-7. Small spring emerging from contact between gravelly and more solid
regolith (left). Large area of seepages emerging at contact between regolith and
massive basalt bedrock forming the riverbed. .................................................... 55
Figure 4-8. Developed spring in the centre of a floodplain, which is completely

submerged in the wet season (left). Spring emerging at the end of the wet season
around the edge of a floodplain........................................................................... 55
Figure 4-9. Cross sections showing the conceptual model during the wet and dry
seasons. The cross section represents a situation common in Dangesheta, such as
a west-east section from the floodplain containing MW2 and MW3, through
MW1 to the Brante River (Figure 4-3). Note the unsaturated zone flow and very
shallow groundwater flow at contradictory directions to the dominant regolith
aquifer groundwater flow during the wet season. Note the dominant regolith
aquifer groundwater flow direction disregarding the surface water flow divide
during the dry season. ......................................................................................... 59
Figure 4-10. Schematic diagram of a regional groundwater flow system (from van der
Heijde (1988)). .................................................................................................... 60
Figure 5-1. The global network of World Weather Watch stations colour-coded to show
reporting rates (from WMO, 2003)). .................................................................. 63
Figure 5-2. Location map of the study area in the Amhara region of Ethiopia. Map
shows formal rain gauges and river gauges near to Dangila woreda. Lake Tana is
visible at the top of the map. ............................................................................... 68
Figure 5-3. Locations of monitoring points (close to arrowhead in Figure 5-2). MW =

monitoring well, DSC = Dangesheta Service Cooperative, DAO = Dangesheta
Agricultural Office. (Image source: Google.Earth; Imagery ©2015
DigitalGlobe). ..................................................................................................... 69
Figure 5-4. Photographs of (left to right) the Kilti River gauge, the rain gauge, and
measuring groundwater level at monitoring well MW5. .................................... 69
Figure 5-5. Double mass check of rainfall for Dangila NMA with Dangesheta
community showing a good linear relationship indicating a consistent record.
xviii
Note that a good record is considered to be a straight line and not necessarily
x=y. ..................................................................................................................... 74
Figure 5-6. Variation with distance of Pearson correlation coefficient (PCC) between
daily rainfall from Dangila NMA rain gauge and other NMA rain gauges close
to Dangila woreda (a). Pearson correlation coefficient (PCC) between daily
rainfall from Dangila NMA rain gauge and alternative sources (b). .................. 76
Figure 5-7. Bias between daily rainfall from Dangila NMA rain gauge and alternative
sources (a). Note that bias is computed as a ratio and the bold line at 1.0
represents zero bias. Median monthly rainfall totals as percentage of annual total
for Dangila NMA rain gauge and gridded datasets (b). ...................................... 77
Figure 5-8. Complete time series to date of daily Kilti and Brante river stage, and
rainfall measurements from the community-based monitoring programme (2014-
2015). .................................................................................................................. 78
Figure 5-9. Unit runoff checks for river flow data from community-based monitoring
and formal sources. Gaps indicate insufficient flow measurements to calculate
monthly totals. Note that the 2014 formal measurements are considered by
MoWIE to be unreliable. ..................................................................................... 79
Figure 5-10. Pearson correlation coefficient (PCC) between daily rainfall from Dangila
NMA rain gauge and river flow measurements (daily totals) from formal and
community sources. Note that “All” includes incomplete months excluded from
other seasons explaining the contrast in relative PCC of, in particular community
monitored Kilti, data from individual seasons to “All”. ..................................... 80
Figure 5-11. Complete time series to date of groundwater level and rainfall
measurements from the community-based monitoring programme (2014-2015).
It is noted that wells MW1, MW3 and MW5 dry out in the dry season and their
minimum groundwater levels represent the basal level of the well. The gap in
the well MW4 time series is discussed in Section 5.8.3. .................................... 81
Figure 5-12. Pearson correlation coefficient (PCC) between community monitored

groundwater level data from monitoring wells MW1-5. .................................... 82
Figure 6-1. Location map of the study area. ................................................................... 92
xix
Figure 6-2. Monthly median, 10th and 90th percentile rainfall, and mean maximum and
minimum temperatures as measured (1987-2017) by the NMA at the Dangila
weather station. ................................................................................................... 92
Figure 6-3. Conceptual model of the study site. ............................................................. 93
Figure 6-4. Plot showing the relationship between annual rainfall and annual recharge in
Ethiopia based on 102 recharge estimates from 49 studies across the country. S =
standard error, R2 (adj.) = adjusted coefficient of determination. The Tigray,
Afar, Dire Dawa group has semi-arid climate and highly heterogeneous geology
ranging from Precambrian crystalline to Mesozoic sandstones and limestones to
Quaternary volcanics, generally overlain by leptosols with sparse and
herbaceous vegetation. Rift Valley and central Ethiopia have subtropical
highland and tropical savanna climate with Quaternary volcanic geology, highly
heterogeneous soils and rainfed cropland and mosaic forest and grassland. The
Lake Tana Basin has a tropical highland monsoon climate and Cenozoic
volcanic rocks overlain by luvisols or vertisols closer to the lake with mosaic
cropland/grassland/shrubland/forest (Tefera et al., 1996; Peel et al., 2007; Arino
et al., 2012; Jones et al., 2013). .......................................................................... 96
Figure 6-5. Median annual recharge estimates from all the techniques. The error bars
give the interannular recharge range. T-M = Thornthwaite-Mather method of
runoff or AET estimation. R-S = Remote sensing method of AET estimation.101
Figure 6-6. Comparison of the sensitivity of each recharge estimation method to ±10%
adjustment in measured input data and modelling parameters (left) and range of
uncertainty when the maximum likely deviations are applied (right). ............. 103
Figure 7-1. SHETRAN columns of finite difference cells showing modelled processes
and geological layers consistent with the four modelled catchments, including
lateral connections between layers of differing thickness. ............................... 115
Figure 7-2. The data layers incorporated into a SHETRAN model (after Lewis (2016)).
........................................................................................................................... 116
Figure 7-3. Location map of the four modelled catchments. ........................................ 119
Figure 7-4. SHETRAN grid, DEM and stream networks for the four modelled
catchments. Note the different horizontal and vertical scales of the catchments.
........................................................................................................................... 121
xx
Figure 7-5. Maps showing the distribution of the hydrogeological and LULC zones of
the four modelled catchments. .......................................................................... 122
Figure 7-6. Interannual variation in monthly PET at Dangila. ..................................... 123
Figure 7-7. Simulated groundwater hydrographs from the Brante model showing three
typical and representative hydrographs for each zone. Top: Representing Zone 1
floodplains and showing observed groundwater hydrograph for monitoring well
MW3. Middle: Representing Zone 2 inhabited and cultivated foothills. Bottom:
Representing Zone 3 hilly areas. It is restated that the monitoring wells all lie
within Zone 1 cells, therefore, observed hydrographs cannot be included in the
Zones 2 and 3 plots. .......................................................................................... 133
Figure 7-8. Simulated and observed river flow hydrographs........................................ 134
Figure 7-9. Simulated and observed cumulative monthly flow totals. ......................... 135
Figure 7-10. Photographs of hummocky tussocky floodplain (left) and rough surface left
by ox ploughing. ............................................................................................... 136
Figure 7-11. Maps showing The Brante catchment and SHETRAN model outputs of
water table depth at the peak of the wet and dry seasons and the transition
between the two seasons. .................................................................................. 139
Figure 7-12. Maps showing potential of shallow groundwater for irrigation use. ........ 142
Figure 7-13. Southern extremes of the Brante and Amen groundwater potential maps
overlain on Google.Earth. ................................................................................. 143
Figure 7-14. Central portion of the Brante and Amen groundwater potential maps
overlain on Google.Earth. ................................................................................. 143
Figure 7-15. Northern portions of the Kilti-Dangesheta, Amen and Brante groundwater
potential maps overlain on Google.Earth. A and B are the locations of the
photographs in Figure 7-16. .............................................................................. 144
Figure 7-16. Photographs taken during the dry season (January 2017) showing water-
stressed Eucalyptus in a low potential zone (left) and healthier Eucalyptus in a
high potential zone at the foot of a slope. See Figure 7-15 for the photograph
locations. ........................................................................................................... 144
Figure 7-17. All four catchments’ groundwater potential maps overlain on Google.Earth
(left) and the same view but showing only the Kilti-Durbete potential map. ... 145
xxi
Figure 8-1. LULC change around Dangila town: increased house-building. (Image
source: Google.Earth; Imagery ©2014 and 2016 DigitalGlobe). ..................... 155
Figure 8-2. LULC change around Dangila town: deforestation for conversion to house-
building and afforestation of cultivated land for Eucalyptus. The forest types
have been ground truthed. (Image source: Google.Earth; Imagery ©2005 and
2016 DigitalGlobe). .......................................................................................... 156
Figure 8-3. Khat production in Robit-Bata kebele........................................................ 159
Figure 8-4. SPI calculated for Dangila for spring (3-month, March-May) and summer
(5-month, June-October). Note the non-continuous year sequence. ................. 163
Figure 8-5. SPEI calculated for Dangila for spring (3-month, March-May) and summer
(5-month, June-October). Note the non-continuous year sequence, which is
different to the SPI plot in Figure 8-4. .............................................................. 163
Figure 8-6. Kilti and Amen Rivers’ percentage variation in unit runoff from the mean.
........................................................................................................................... 164
Figure 8-7. Growing season length at Dangila. Dashed lines indicate where onset or
cessation lies within missing data. .................................................................... 165
Figure 8-8. Counts of 16-day periods of spatially averaged absolute NDVI values
above/below thresholds to identify dry and wet years. High counts in the NDVI
dry years plot (left) indicate prevalent dry conditions and low counts in the
NDVI wet season length plot indicate short growing seasons. ......................... 166
Figure 8-9. Mean of LST daily difference per season. The greater the positive value the
lower the soil and canopy moisture content. ..................................................... 167
Figure 8-10. Median and percentile annual rainfall totals for Dangila from ground
observations, remote sensing and reanalysis products. ..................................... 167
Figure 8-11. Annual rainfall totals for Dangila from ground observations, remote
sensing and reanalysis products. ....................................................................... 168
Figure 8-12. Annual potential evapotranspiration totals from Dangila. ....................... 168
Figure 8-13. Community workshop in Dangesheta (left) and frost damage to banana
plant in January 2017. ....................................................................................... 169
Figure 8-14. Example plot of proportional coverage of failure (only rainfed agriculture
possible), success (second growing season possible) and high success (second
xxii
and third growing season possible; shallow groundwater available year-round)
per year for the Amen catchment historic time series. ...................................... 180
Figure 8-15. Amen catchment baseline plot of proportional coverage of failure (only
rainfed agriculture possible), success (second growing season possible) and high
success (second and third growing season possible; shallow groundwater
available year-round) per year for the 200-year simulation showing the potential
impact of climate variability on groundwater availability. ............................... 182
Figure 8-16. Brante catchment baseline plot of proportional coverage of failure (only
rainfed agriculture possible), success (second growing season possible) and high
success (second and third growing season possible; shallow groundwater
available year-round) per year for the 200-year simulation showing the potential
impact of climate variability on groundwater availability. ............................... 182
Figure 8-17. Cumulative frequency curves showing the impact of converting pasture to
arable land on shallow groundwater availability (the proportional coverage of
high success). .................................................................................................... 185
Figure 8-18. Cumulative frequency curves showing the impact of converting pasture to
arable land on surface water availability (the proportional of the year when flow
is <1 l/s). ............................................................................................................ 186
Figure 8-19. Cumulative frequency curves showing the impact of converting highlands
to arable land on shallow groundwater availability (the proportional coverage of
high success). .................................................................................................... 187
Figure 8-20. Cumulative frequency curves showing the impact of converting highlands
to arable land on surface water availability (the proportional of the year when
flow is <1 l/s). ................................................................................................... 187
Figure 8-21. Cumulative frequency curves showing the impact of degrading highlands
to bareground on shallow groundwater availability (the proportional coverage of
high success). .................................................................................................... 188
Figure 8-22. Cumulative frequency curves showing the impact of degrading highlands
to bareground on surface water availability (the proportional of the year when
flow is <1 l/s). ................................................................................................... 188
Figure 8-23. Cumulative frequency curves showing the impact of expanding Eucalyptus
plantations on arable land on shallow groundwater availability (the proportional
xxiii
coverage of high success). The two variations refer to alternative methods of
parameterising the Eucalyptus LULC category. ............................................... 189
Figure 8-24. Cumulative frequency curves showing the impact of expanding Eucalyptus
plantations on arable land on surface water availability (the proportional of the
year when flow is <1 l/s). The two variations refer to alternative methods of
parameterising Eucalyptus LULC category. ..................................................... 190
Figure 8-25. Cumulative frequency curves showing the impact of increasing abstraction
and irrigation on shallow groundwater availability (the proportional coverage of
high success). .................................................................................................... 192
Figure 8-26. Cumulative frequency curves showing the impact of increasing abstraction
and irrigation on surface water availability (the proportional of the year when
flow is <1 l/s). ................................................................................................... 192
Figure 9-1. Output of climate variability model simulations showing the proportion of
arable land that has shallow groundwater available in a particular year according
to that year’s severity of dryness....................................................................... 199
Figure 9-2. Map of Ethiopia showing distribution of main aquifer types (from Kebede
(2013))............................................................................................................... 204
Figure 9-3. map of Africa with estimated depth to groundwater (from Bonsor and
MacDonald (2011)) ........................................................................................... 206
xxiv
List of tables
Table 5-1. Pearson correlation coefficient (PCC) between community monitored river
stage and groundwater level from monitoring wells MW1-5 for the entire time
series.................................................................................................................... 83
Table 6-1. Hydrological zone, spatial scale, and data requirements of the applied
recharge estimation techniques. .......................................................................... 95
Table 6-2. Details of the three catchments modelled using SHETRAN (see Figure 6-1
for locations). .................................................................................................... 100
Table 6-3. Tabulated recharge results shown on the plot in Figure 6-5........................ 101
Table 6-4. Parameters and input data adjusted for the sensitivity and uncertainty
analysis. ............................................................................................................. 102
Table 6-5. Summary of methods and suggestions for lessening uncertainty in the
recharge results. It should be restated here that while the specific methods
usually compute the specified type of recharge, this is particular to the
conceptual model of the study site. ................................................................... 104
Table 7-1. Main processes represented in the water flow component of SHETRAN
(after Ewen et al. (2000)). ................................................................................. 114
Table 7-2. Flow equations for SHETRAN applicable in this study (after Ewen et al.
(2000)). .............................................................................................................. 114
Table 7-3. Information about the catchments modelled with SHETRAN. ................... 119
Table 7-4. The slope gradients used to delineate the three zones using ArcGIS analysis
of the SRTM DEM, and, based on field measurements, the mean well depths
used to determine the base of the regolith, and the regolith aquifer properties.125
Table 7-5. Details and statistics of the calibration and validation periods.................... 133
Table 7-6. Optimum calibrated parameter values for the SHETRAN models. ............ 136
Table 8-1. Identification of drought/wet years and comparison of drought analysis

methods for Dangila. X = insufficient data for analysis. v = very. Note the jump
from 1933 to 1956. For SPI, SPEI and SPI/SRA from other studies: v dry and v
wet are <-2 and >2 respectively, dry and wet are <-1.5 and >1.5 respectively.
For river flow: v high and v low are >±40% difference in unit runoff from the
xxv
mean, high and low are >±20% difference in unit runoff from the mean. For
growing season length: v long and v short are >±20% difference from mean
length in days, long and short are >±10% difference from mean length in days.
For NDVI dry year: dry is >3 16-day periods in the 10%ile and >4 in the 20%ile,
wet is <2 16-day periods in the 10%ile and <4 in the 20%ile. For NDVI wet
season length: long is >12 16-day periods in the 50%ile and >8 in the 75%ile,
short is <11 16-day periods in the 50%ile and <7 in the 75%ile. For LST wet
season daily difference: dry and wet are >±70% difference from the mean daily
difference. Annual rainfall: dry is rainfall total <10%ile, wet is >90%ile. ....... 171
Table 8-2. Approximate ranking of available years according to drought severity,

showing the eight worst (driest) and eight wettest years and the corresponding
annual rainfall total. .......................................................................................... 176
Table 8-3. Correlations between subsurface water storage from one year to the next for
the Amen and Kilti-Durbete catchments for 1998 to 2014 as computed by
SHETRAN. SSW = subsurface water storage. ................................................. 177
Table 8-4. Statistics of annual subsurface water storage in Amen and Kilti-Durbete
catchments for 1998 to 2014 as computed by SHETRAN. SSW = subsurface
water storage (spatially averaged mm), CV = coefficient of variation. ............ 178
Table 8-5. Statistics of proportional coverage of high success, success, and failure from
Amen and Brante baseline 200-year simulations. CV = coefficient of variation,
c10 = 10th percentile, c90 = 90th percentile. ...................................................... 183
Table 8-6. The potential future climate variability, LULC and abstraction scenarios
simulated with SHETRAN................................................................................ 183
xxvi
List of abbreviations and acronyms
ADSWE Amhara Design & Supervision Works Enterprise (consultancy

and laboratory in Bahir Dar, Ethiopia)
AET Actual evapotranspiration
ARBA Abay River Basin Authority
AMGRAF Adaptive management of shallow groundwater for small-scale
irrigation and poverty alleviation in sub-Saharan Africa
(Newcastle University research project)
ASTER GDEM Advanced Spaceborne Thermal Emission and Reflection
Radiometer Global Digital Elevation Model
BFI Baseflow index
BP Before present
CMB Chloride mass balance
DEM Digital elevation model
DFID Department for International Development (UK)
DHI Danish Hydrological Institute
EC Electrical conductivity
ENSO El Niño – Southern Oscillation
ERT Electrical resistivity tomography
ESRC Economic and Social Research Council (UK)
ET Evapotranspiration
Ethiopian ATA Ethiopian Agricultural Transformation Agency
FAO Food and Agriculture Organization of the United Nations
GCM General circulation model
GDP Gross domestic product
GPS Global positioning system
GSE Geological Survey of Ethiopia
GWI Groundwater irrigation
HDW Hand-dug well
IPCC Intergovernmental panel on climate change
ITCZ Inter tropical convergence zone
IUC Irrigation user community
IWMI International Water Management Institute
xxvii
K Hydraulic conductivity
LST Land surface temperatures
LULC Land use land cover
Ma Million years
MAP Median annual precipitation
m asl Metres above sea level
mbgl Metres below ground level
MC Soil moisture retention (root zone storage)
MDG Millennium Development Goal
MER Main Ethiopian Rift
METI Ministry of Economy, Trade, and Industry of Japan
MPI Multi-dimensional poverty index
MoWIE Ethiopian Ministry of Water, Irrigation and Electricity
MoWR Ethiopian Ministry of Water Resources
MoANR Ethiopian Ministry of Agriculture and Natural Resources
NASA United States National Aeronautics and Space Administration
NDVI Normalized difference vegetation index
NERC Natural Environment Research Council (UK)
NGO Non-Governmental Organisation
NMA Ethiopian National Meteorology Agency
NSE Nash and Sutcliffe efficiency
PET Potential Evapotranspiration
PCC Pearson correlation coefficient
QGIS An open source geographic information systems (GIS) application
RCM Regional climate model
RIB Rainfall infiltration breakthrough model
RMSE Root mean square error
R-S Remote sensing
RWSN Rural Water Supply Network
SDGs Sustainable Development Goals
SMB Soil moisture balance
SPEI Standardised precipitation-evapotranspiration index
SPI Standardised precipitation index
SRA Standardised rainfall anomaly
SRTM Shuttle Radar Topography Mission
xxviii
SSA Sub-Saharan Africa
Sy Specific yield
T Transmissivity
T-M Thornthwaite-Mather method
TEJ Upper-level tropical easterly jet
TRMM Tropical Rainfall Measuring Mission
UNSA Unconsolidated sedimentary aquifer
UPGro Unlocking the Potential of Groundwater for the Poor
(NERC/ESRC/DFID funded research programme)
USAID United States Agency for International Development (USA)
VSMOW Vienna Standard Mean Ocean Water
WETSPRO Water Engineering Time Series Processing Tool
WHAT Web GIS based Hydrograph Analysis Tool
WHO World Health Organisation
WHYMAP Worldwide hydrogeological mapping and assessment programme
WMO World Meteorological Organisation
WTF Water table fluctuation
WUA Water users association
WUC Water user committee
xxix
Chapter 1. Introduction
1.0 Chapter summary

This chapter provides the context of the PhD research, which will be explored in greater
detail in Chapter 2. The overall aims are presented along with the research questions that
this study attempts to answer. A general methodology states how the research was
conducted and how the chapters link together. Finally, an outline of the thesis explains
what can be found in each chapter.
1.1 Context
The availability of groundwater in Africa and its potential for agricultural use has been
increasingly reported in recent years with many authors predicting that a rapid expansion
in groundwater exploitation may be about to happen in sub-Saharan Africa (SSA)
(MacDonald et al., 2012a; Namara et al., 2013; Pavelic et al., 2013a; Villholth, 2013;
Baguma et al., 2017). Such a growth in the exploitation of groundwater resources in South
and East Asia since the 1970s promoted improved living standards and fostered economic
development (Calow et al., 2009a; Narayanamoorthy, 2010). MacDonald et al. (2012a)
estimate the total groundwater storage in Africa to be 0.66 million km3. Despite this
seemingly extremely high groundwater availability, over 90% of agriculture in Africa is
rainfed (Wani et al., 2009; McClain, 2013) and Siebert et al. (2010) report that only 3.3%
of arable land in SSA is irrigated, compared to 37% in Asia (Figure 1-1). Such figures
suggest there is ample water and space to expand areas under irrigation. Indeed, the
renewable groundwater resources per capita in some of the Asian countries that
experienced the “Green Revolution” stand at ~600 m3/a in China and ~400 m3/a in India,
while in SSA the available quantity is ~2400 m3/a (FAO (2003) cited in Giordano (2006)).
Considering the sustainability of the available resource, Frenken (2005) using FAO
AQUASTAT data, showed that the Sudano-Sahel region uses only 35% of its total
internally renewable water resource while coastal West Africa uses just 1.3%. The maps
in Figure 1-1 show a remarkable contrast between Asia and Africa; however, fifty years
ago, prior to the Green Revolution, the Asia map would have looked similar to how the
Africa map looks today.
1
Figure 1-1. Comparison between South and East Asia, and Africa of percentage area irrigated by
groundwater (adapted from Siebert et al. (2010)).
Small-scale irrigation, in particular from groundwater, is increasingly promoted by

governments, donors and NGOs as an important tool to alleviate poverty, improve food
security, boost rural employment and economic development, promote gender equality,
and mitigate against increasing climate variability (Kay, 2001; Ngigi and Denning, 2009;
Abric et al., 2011; Villholth, 2013). In the latter case, groundwater behaviour is significant
in that it responds slowly to drought, unlike the “flashy” response of surface water (Figure
1-2). Therefore, groundwater is considered able to buffer short-term climate impacts
(Calow et al., 1997; MacDonald et al., 2009).
Figure 1-2. Contrasting response and recovery behaviour of groundwater and surface water to drought
(from Robins and Fergusson (2014)).
Smallholders themselves, where they can find the necessary means and finances, prefer
groundwater irrigation in comparison to large-scale irrigation schemes, due to the
2
autonomy it provides and the enhanced livelihoods created by moving from subsistence
to market-oriented agriculture (Giordano et al., 2012; Dittoh et al., 2013). Groundwater
is further preferred because it is generally available at the point of use, can often be
developed quickly with low capital cost by individuals, and is available on-demand
(Foster and Shah, 2012). It is well documented that abundant groundwater abstracted via
deep formal systems can provide secure and perennial water availability. However,
deeper systems require large initial investment, high maintenance, and strict farmer
organisation to deal with water-sharing. Shallow informal systems, which tend to be
farmer-driven rather than in the control of the public sector, are spontaneous and
autonomous (Kay, 2001; Villholth, 2013). For example: in Tanzania, surveys by Sokile
and van Koppen (2004) revealed that more than 70% of water users prefer to settle water
disputes via informal channels, such as local water users associations (WUAs), rather
than relying on formal state-based institutions; and in Ghana, private smallholder
irrigation already employs 45 times more people and covers 25 times more area than
public irrigation schemes (Giordano et al., 2012).
The focus of this study concerns the utilisation of shallow groundwater resources for
small-scale irrigation. Notably, Figure 1-3 shows that low to moderate-yielding shallow
aquifers predominate in the more densely populated areas. There is little agreement in the
literature over how shallow is “shallow” when it comes to groundwater. For this research,
shallow groundwater is defined as <25 m. Although there are exceptions around the
world, 25 m depth is considered the maximum feasible limit of excavation of “hand-dug”
wells (Watt and Wood, 1977; Abbott, 2013). In addition, much of the existing small-scale
groundwater irrigation depends on a water table depth less than 5 m because of power
limits on water-lifting and because of available technology. Motorised pumps are much
less common than manual lifting methods in SSA – used in less than 20% of water-lifting
cases in various surveys conducted by Namara et al. (2013) – due to smallholder farmers’
lack of capital and ability to obtain credit. This means that groundwater irrigation is
restricted to shallow hand-dug wells for poorer farmers; depths of 50 m – the definition
of shallow by some authors – cannot be regarded as easily accessible for small-scale
irrigation. Because poor rural communities manually excavate wells, the predominance
of consolidated and crystalline bedrock restricts the potential locations for manual well
excavation. However, unconsolidated sediments cover approximately 25% of Africa
(Guiraud, 1988). What’s more, when river valley sediments and regolith above crystalline
basement or more recent volcanics are considered, the extent of the shallow geology with
3
potential for manually-excavated or manually-drilled wells becomes pervasive (a
conclusion supported by the work of Fussi (2011); Fussi et al. (2016)).
Figure 1-3. The aquifer productivity map across Africa, estimated depth to groundwater, and population
density map (adapted from MacDonald et al. (2011)).
The importance of these shallow materials as locally important aquifers is well reported.
Typically simultaneously reported is the dearth of observations of such groundwater
systems, in particular sustained time-series data (Martin and Van De Giesen, 2005;
Robins et al., 2006; Calow et al., 2009b; MacDonald et al., 2009; Taylor et al., 2009;
Ethiopian ATA, 2013). This scarcity of data leads to poor understanding of the resource.
What’s more, any attempts to model the hydrogeological system to improve
understanding would be problematic without monitoring data. Community-based
monitoring data was utilised for this research to circumvent the issue of the lack of formal
hydrometeorological time-series data. In addition to providing data in locations where
there are sparse formal alternatives, benefits to the local community include an increased
understanding of their resource and a sense of ownership and empowerment (Garduño
and Foster, 2010; Conrad and Hilchey, 2011; Buytaert et al., 2014).
It is not straightforward to define what is meant by “small-scale”, or “smallholder”,

irrigation. The World Bank (2003) defines smallholders as those having a low asset base
4
and less than 2 ha of cropped land. The FAO defines smallholders as farmers with
“limited resource endowments, relative to other farmers in the sector” (Dixon et al.,
2003). Essentially, “small-scale irrigation” is used to describe irrigation activities of small
spatial extent and/or involving a low number of irrigating farmers, although it is also
applied to irrigation schemes that have a high degree of local involvement at planning
and development stage, use locally available resources and technology, and have a local
impact (Adams and Carter, 1987). The latter such situation is frequently described as
“informal” to draw a distinction from centrally planned and bureaucratically controlled
schemes. While for many authors “informal” irrigation is synonymous with
“smallholder”, it is noted that many others draw a distinction between “smallholder” and
“small-scale” citing examples such as the large-scale irrigation projects in Egypt and
Sudan of 50+ ha that are farmed by hundreds of smallholders each occupying <1 ha (Kay,
2001). The only consensus on small farms seems to be the lack of a clear definition
(Nagayets, 2005). The working definition used here is from Adams and Carter (1987),
though in this study smallholder and small-scale irrigation are used interchangeably:
“…the management of the supply of water to crops or other economically useful plants,
which is organised and controlled by the landholder or groups of landholders; the extent
of such activities does not normally exceed 10 ha per farm family, and may be as little as
0.1 ha.” A study by Nagayets (2005) revealed that Ethiopia tops the list of African
countries with small farms, having almost 9.5 million small farms (<2 ha), comprising
87% of all farms within the country. Incidentally, the top two countries on the global list
are China and India with ~190 million and ~93 million small farms.
1.2 Aims
The aims of this PhD research were, firstly, to determine the potential for small-scale
irrigation and the resilience of shallow groundwater resources used by rural communities
at a representative study site in Ethiopia. Secondly, to develop transferable methodologies
for assessment of shallow groundwater resources throughout SSA. The following
questions were researched:
1. Do shallow aquifers have the requisite properties, in terms of hydraulic

conductivity, potential well yield, specific yield, aquifer geometry and
hydrochemistry, for productive groundwater use?
2. Due to the scarcity of time-series data, are community-based hydrometeorological
monitoring programmes able to produce useful, high quality data comparable to
formal data sources?
5
3. Can shallow groundwater be considered a renewable resource, and; which
recharge assessment methods provide the highest confidence in the calculated
recharge amounts when applied to these types of aquifers?
4. Are there identifiable zones that show the greatest potential for sustainable
intensification of agriculture through shallow groundwater irrigation?
5. How will climate variability, land use change and increased abstraction impact
shallow groundwater resources and surface water?
1.3 Study site

The principal research, including field investigations, was conducted for a field site in
Ethiopia (Figure 1-4). The field site was chosen to be representative of a much wider area.
The site is currently reliant on rainfed agriculture though has been identified by the
Ethiopian Agricultural Transformation Agency (ATA) as a location for potential future
agricultural expansion: Currently 90-95% of farmed land in SSA is rainfed (Wani et al.,
2009; McClain, 2013). The field site is at high elevation within highlands and experiences
high annual rainfall, which mostly falls during the five-month wet season; the Ethiopian
Highlands are laterally extensive comprising 50% of Ethiopia, consequently they are
known as the roof of Africa, and highland areas with >1000 mm/a seasonal rainfall exist
across Central and West Africa (Frenken, 1997). Cenozoic volcanics underlie the field
site: Around 40% of Ethiopia is underlain by Cenozoic volcanic rocks (Prave et al., 2016)
and similar geology can be all along the East African Rift. Ethiopia is classified as a “low-
income economy” by the World Bank (World Bank, 2017b), which translates on the
ground to poor access to agricultural equipment and markets and poor ability to cope with
climate or other stresses: 27 of SSA’s 48 countries are classified as low-income. The
study site is detailed in Chapters 3 and 4 while the transferability of the research is
discussed further in Chapter 9.
6
Dangila
study site
Ethiopia
Figure 1-4. Study site location.
1.4 Methodology
A graphic showing the step-by-step methodology is presented in Figure 1-5. Following
desk study analysis of data relating to the study sites, extensive field investigations were
conducted during three visits to Ethiopia. Time-series hydrometeorological data were
analysed from a community-based monitoring programme and formal sources for use in
modelling studies. Recharge assessments were conducted to estimate the renewable
shallow groundwater resource. Models were constructed combining data from field
investigations and hydrometeorological monitoring. At local scales, community
monitored networks allowed direct comparison between modelled hydrological responses
and field observations of groundwater levels and river flow. The modelling furthered the
development of the conceptual model, increasing understanding of the shallow
groundwater system, and enabling production of groundwater potential zone maps that
revealed the best locations for groundwater abstraction for productive use. Simulations
were run with potential future climate variability, land use and abstraction scenarios.
Modelling outputs were processed to assess the impact of these potential future scenarios
on the shallow groundwater resource and on surface water resources.
7
Figure 1-5. Methodology flow chart.
1.5 Thesis outline

Chapter 1 gives the context of the research and defines terms used throughout the thesis,
presents the aim and research questions, and introduces the methodology applied to
answer those questions.
Chapter 2 is a literature review that presents Asia’s “Green Revolution” as justification

for the study as well as providing background information on African hydrogeology and
the African climate. The chapter then reviews literature identifying the research gap this
PhD study aims to fill. It is noted here that this thesis does not have a single all-inclusive
literature review as each chapter and some sections within chapters have their own
relevant literature reviews.
Chapter 3 provides background information from desk study of the study site for which
the research was predominantly conducted.
Chapter 4 details the field investigations conducted at the study site that led to the
development of the described conceptual model.
8
Chapter 5 concerns community-based monitoring, detailing the programme setup in order
to gather hydrometeorological time-series data for the project. The chapter presents the
statistical comparison study conducted against data from formal sources in order to assess
the quality of the community data.
Chapter 6 discusses groundwater recharge assessments. The multiple recharge estimation

techniques applied are described and critically analysed.
Chapter 7 presents the modelling of the shallow aquifer. Model selection, construction
and calibration are described along with the resulting insights and update on the
conceptual model. Maps of shallow groundwater potential for irrigation generated from
the modelling are shown and discussed.
Chapter 8 presents the modelling of potential future climate, land use and abstraction
scenarios for assessment of impacts on the surface and shallow groundwater resources.
Chapter 9 draws together findings from all chapters and presents conclusions with
reference to the project aims. The chapter provides recommendations for stakeholders
and discusses the transferability of the results and the methodology, and presents
limitations of the study. Finally, suggestions are provided for future work.
9
Chapter 2. Literature review
2.0 Chapter overview

Firstly, information is provided on the “Green Revolution” that occurred in Asia in the
1960s and 70s explaining the potential for irrigation to have positive impacts on food
security, poverty alleviation, and gender equality, amongst other benefits. Justification
for this PhD research is then provided with literature concerning how such a green
revolution could occur in sub-Saharan Africa (SSA). This is followed by a background
on African hydrogeology and the African climate along with projections of future changes
to climate and land use. The chapter then details the research gap, namely: insufficient
understanding of shallow aquifers in sub-Saharan Africa, their potential for productive
use, and their resilience.
2.1 The role of groundwater in Asia’s “Green Revolution”

Growth in the exploitation of shallow groundwater resources in South and East Asia since
the 1970s has promoted living standards and fostered economic development (Calow et
al., 2009a; Narayanamoorthy, 2010). The impact on hunger and poverty from the
explosion in use of groundwater irrigation (GWI), combined with the spread of higher
yielding crops, fertilizers and modern pest control methods, has been such that it is known
as Asia’s “Green Revolution” (Evenson and Gollin, 2003; Hazell, 2009).
In Bangladesh, the area irrigated by groundwater increased from zero to 2.8 million ha
between 1980 and 2000. From the 1950s to 2000, the number of mechanised wells in
India increased from <100,000 to 19 million. In Bangladesh over the same time period
the increase in mechanised wells was <1000 to 1 million, while in Hubei Province, China,
mechanised wells increased from 730 to 840,000, and in Punjab, Pakistan, the number
increased from <1000 to 500,000. In India, agricultural productivity increased by 80%
per ha. Conversely, the number of people living below the poverty line from the 1970s to
2000 decreased from 250 million to 29 million in China, from 45% to 28% in Pakistan
and from 71% to 44% in Bangladesh. The latter figures are even more remarkable when
it is considered that global population doubled from three to six billion between 1960 and
2000, and in South and East Asia, national populations more than doubled (Palmer-Jones,
1992; Shah et al., 2003; Hussain et al., 2006; World Bank, 2017b).
The Green Revolution proved that the positive impacts from expansion of GWI can
include, at the local level (list adapted from Bhattarai et al. (2001)):
10
 Greater food security as communities are less reliant on there being consistent
rains
 Increased cropping intensity
 Crop diversification
 More wealth to enable purchase of fertilizers, pesticides and insecticides leading
to higher crop yield
 Year-round cropping opportunities
 Improved livelihoods
 Increased farm employment
 Less labour-intensive agriculture causing increased school attendance
 Improved gender equality because girls are no longer walking long distances for
water so can receive a better education and suffer less drudgery
 Increased wealth has a positive impact on off-farm activities and employment
such as house building and shops/markets
 The water can also be used for bathing, providing health benefits, and for
livestock.
At national level:
 Causes a decline in food prices that also helps the urban poor who still spend 50%
of their income on food
 Increases export revenue, for example coffee from Vietnam and rice from
Thailand.
The list above shows that expansion of GWI directly and indirectly targets many of the
United Nations Sustainable Development Goals (SDGs) (Figure 2-1). In addition to Goals
1-6, 8, 10, 11 and 15, Goal 13 “Climate Action” is also targeted, as GWI is a means to
mitigate against increasing climate uncertainty.
11
Figure 2-1. The United Nations Sustainable Development Goals (SDGs) (UN, 2015).
The impact of irrigation on farming income has been the focus of numerous studies.
Research by Bhatia (1991) in Bihar, India, revealed additional net income due to
improved irrigation access of Rs 2,511 (~£30) per ha. This translates to farm income in
irrigated areas of Bihar being 77% higher than income in unirrigated areas. The study
further states how only 32% of the cropped area of Bihar is irrigated while that value is
60% for the state of Haryana, also in north India. Consequently, the rural population living
below the poverty line stands at 51% in Bihar and 15% in Haryana. This shows how
improved access to irrigation contributes toward poverty alleviation and the improvement
of livelihoods in a region. A similar study by Ut et al. (2000) in Vietnam showed a 68%
increase in cropping intensity from irrigated over rainfed agriculture and a consequent
increase in income of US$188/ha. A broad study by Giordano et al. (2012) found that
having access to water during the dry season made a large difference to farmers’ incomes
and nutrition: In Madhya Pradesh, India, farmers’ incomes who began irrigating pulses
and wheat increased by more than 70%.
Bangladesh is analogous to the potential of SSA where shallow wells have proved easier
to manage and more cost-effective than deep wells which has encouraged investment in
shallow GWI. The outcome has been an increase in irrigated area and an improvement
from monopolistic water and food markets to oligopolistic markets (Akteruzzaman et al.,
1998). Hossain and Islam (2000) reported how women in rural areas have benefitted
particularly in Bangladesh, because prior to the increased access to safe water, they had
12
to carry water over long distances, with significant impact on their health and
productivity.
2.2 The potential for such a green revolution in sub-Saharan Africa

There are several studies available in the literature that suggest the green revolution is
both possible for SSA and has already begun (e.g. Inocencio et al. (2007); Foster et al.
(2008)). Giordano et al. (2012) provide examples from Tanzania where growing irrigated
vegetables contributes half of the dry season cash income of smallholders, and in Zambia,
35% more is earnt by smallholders who cultivate vegetables in the dry season with
irrigation than those who do not. A study in Ghana by Dittoh et al. (2013) showed the net
revenue from groundwater irrigation ($631.51/acre) was significantly greater than from a
reliance on rain ($129.46/acre) and that irrigators regard pumps as “saviours”, but the
major constraint is the financial means to buy them. Giordano et al. (2012) further state
how investing in motor pumps could benefit 185 million people in SSA and generate net
revenues up to US$ 22 billion per year. They estimate that 400,000 pumps were imported
into Ethiopia in the last decade, and that there are 20,000 motor pumps in Burkina Faso,
160,000 in Ghana, 70,000 in Tanzania, and 15,000 in Zambia. However, manual water-
lifting is still used by 70% of irrigators in Ethiopia, and from 84 to 91% of irrigators in
the other aforementioned countries.
A key difference with Asia is the generally lower yield of wells in SSA (low-productivity
basement aquifers vs productive alluvial aquifers) that restricts their cooperative use (an
important factor in the achievements of Bangladesh) because a single well typically has
only sufficient yield, generally less than 1 l/s, to irrigate a single smallholding
(MacDonald and Davies, 2000). However, this may not be entirely disadvantageous. The
low-yielding nature of typical SSA aquifers may provide, according to MacDonald et al.
(2012b), “a solution to the intractable problem of sharing common-pool natural resources
equitably (Hardin’s “Tragedy of the Commons”). In low permeability aquifer
environments, overpumping of an individual well is unable to exert much influence
beyond the hectare-sized plot on which most farming is conducted in sub-Saharan Africa.
As a result, the in situ hydrogeological conditions will restrict localised overexploitation
naturally.”
As of 2010, only 3.3% of arable land in sub-Saharan Africa was under irrigation, with a
groundwater demand of 2.178 km3/year (Siebert et al., 2010). Increasing the area of land
under irrigation will require increasing groundwater abstraction. Carter and Alkali (1996)
13
quote several studies from rural areas of West Africa with groundwater abstraction
estimates of 1-4 mm/a, increasing at 0.027 mm/a in line with population growth.
Expansion of agriculture is widely considered to be the only pathway to long-term and

pro-poor economic development in SSA, by stimulating growth in the wider economy
and absorbing excess labour through advances in the rural non-farm economy (Adelman,
1984; Collier and Dercon, 2014; Dawson et al., 2016).
2.3 African hydrogeology

In this context, an understanding of the hydrogeology is from the point of view of
understanding the potential groundwater resource. The availability of groundwater
resources is critically dependent on the geology, the degree of weathering and fracturing,
and recharge (whether historical or recent). According to MacDonald et al. (2008), the
SSA region can be divided into four hydrogeological provinces (Figure 2-2):
1. Crystalline basement, which occupies 40% of the land area.

2. Consolidated sedimentary rocks: 32%
3. Volcanic rocks: 6%
4. Unconsolidated sediments: 22%.
Figure 2-2. The hydrogeological environments of SSA (from MacDonald and Davies (2000)).
14
As can be seen in Figure 2-2, Precambrian crystalline basement is pervasive across SSA.
It comprises igneous and metamorphic rocks greater than 541 million years old. Where
unweathered and unfractured, the basement rocks contain negligible groundwater,
however, important aquifers can develop within fracture zones and the weathered mantle
(Wright and Burgess, 1992). Deep fractures within basement rocks are tectonically
controlled and can be important sources of groundwater, especially where sub-vertical
and below a thin or absent weathered zone. The weathered mantle, or regolith, consisting
of variously gravelly and clayey decomposing parent rock, can be 90 m thick in tropical
regions, though is more commonly in the range of 20-30 m (MacDonald et al., 2005).
Regolith hydrogeology is discussed more thoroughly in Chapter 3.
Large sedimentary basins are present within SSA that store vast quantities of
groundwater. However, in semi-arid and arid regions the groundwater is largely non-
renewable, or fossil, having been recharged in the past when rainfall was much greater;
any abstraction is essentially water mining (Döll, 2009). The best aquifers are found
within sandstone and limestone units; unfortunately, around 65% of consolidated
sedimentary rocks are low-permeability mudstones (Aplin et al., 1999). Where the
mudstones are hard, groundwater may be obtained from fracture zones. Similarly, older
sandstones may be cemented and it is again secondary (fracture) porosity that is targeted
for groundwater abstraction (MacDonald et al., 2005). Limestones generally make good
aquifers, though where fractures are enlarged through dissolution, i.e. karsts, giving high
groundwater flow and infiltration rates, there can be high vulnerability to contamination
and saline intrusion (Robins et al., 2007).
Volcanic rocks of SSA are mostly of Cenozoic age and formed during three major pulses
of late Eocene, mid-Miocene, and Plio-Pleistocene age related to the opening of the East
African Rift, though some volcanic rocks date from the Jurassic (Baker et al., 1972). The
groundwater potential of volcanic terrains is extremely variable, reflecting the complexity
of the geology. Volcanic sequences are often 100s m thick, consisting of interbedded
lavas and pyroclastic rocks. Massive lava flows can be impermeable, though extensive
jointing can allow groundwater infiltration and flow (MacDonald and Davies, 2000). The
junctions between lava flows often develop into important conduits as cracks, joints and
rubble develop when the base of the more recent lava flow rapidly cooled (Kehinde and
Loehnert, 1989). Ash layers are generally of low permeability but have high storage,
therefore, useful aquifer systems can form where ash is present in alternating layers with
fractured or rubbly lavas (MacDonald et al., 2005).
15
Relatively young (less than a few million years old) unconsolidated sediments range from
coarse gravels and sands to silts and clays, and from regionally extensive (e.g. coastal or
deltaic) deposits 100s m thick to thin strips of alluvium beside small rivers (MacDonald
et al., 2005). Large unconsolidated sedimentary aquifers (UNSAs) can be highly
productive where sand and gravel beds are continuous over 10s-100s km (Guiraud, 1988).
However, depending on the depositional environment, complex multi-layered aquifers
can form with sands and gravels interbedded with silt and clay aquitards; the sediments
varying laterally every few metres (MacDonald and Davies, 2000). Small UNSAs, less
than 100 m in width and with sediments less than 10 m in thickness, form locally
important aquifers, typically in valleys, deposited by modern rivers; groundwater is
usually close to the surface so pumping lifts are small (Owen, 1989; Carter and Alkali,
1996). Proximity to a river gives a reliable source of recharge though in semi-arid areas
where surface water is rare, i.e. sand rivers, recharge occurs during infrequent flood
events (Hussey, 2007; Love et al., 2007).
2.4 The African climate

The climate of Africa is best portrayed by the updated Köppen-Geiger climate type map
presented in Figure 2-3, based on global climate classification studies conducted by
Köppen a hundred years previous (Köppen and Geiger, 1930; Peel et al., 2007). The map
shows that only three of the main climate types are present in Africa and, of these, the
dominant climate type by land area is the arid B (57.2%), followed by tropical A (31.0%)
and temperate C (11.8%). The map is ground-truthed against 1436 precipitation and 331
temperature stations.
16
18oN
Equator
20oS
Figure 2-3. Köppen-Geiger climate type map of Africa with description of climate symbols and defining
criteria (from Peel et al. (2007)).
17
The North African coast with its temperate “Mediterranean” climate is not considered as
part of this study. Irrigated agriculture is unlikely until a latitude of approximately 18oN
is reached and the climate of this area is well described by Nicholson (1984): “The sub-
Saharan region immediately south of the Sahara Desert is characterized by low, highly
variable rainfall and a landscape that undergoes a marked and abrupt change between wet
and dry seasons within a year. Moving southwards, the rainfall increases along with the
length of the rainy season. Rainfall gradients are steep; as much as 100 mm per 100 km
in West Africa, passing from 100 mm in the northern region of the Sahelo-Saharan zone
to over 1600 mm in the Guinean zone. The duration of the rainy season also varies greatly,
ranging from 1-month in the desert margin to more than 8-months in the Guinean coastal
zone. Hence, the transition from desert to the humid tropics is abrupt.”
In the semiarid sub-Saharan zones, wet and dry seasons are controlled by the subtropical
high-pressure zone and the inter-tropical convergence zone (ITCZ). The ITCZ lies at the
convergence of the northeasterly and southeasterly trade winds; in Africa that relates to
the transition between the dry northeasterly harmattan winds of the Sahara and the moist
southwesterly monsoon flow originating from the tropical Atlantic. This convergence
moves northward during May to October bringing heavy cloud and intense rainfall. Rainy
season length at particular latitudes reflects the number of months that the ITCZ
dominates the local climatology (Nicholson, 1984).
Moving southwards to equatorial latitudes within the Congo Basin the rainy season has a
bimodal distribution peaking in April and again in October with short dry seasons running
from December to February and June to August (Washington et al., 2013). Further east
towards the East African coast is anomalously dry for its equatorial latitude primarily due
to the rain shadow effect of the Rwenzori Mountains and the Ethiopian Highlands.
As latitude increases in a southerly direction, a unimodal rainfall regime is again

established during opposing months to that seen in the northern hemisphere. Rains are
greatest during November to January decreasing south of approximately 20oS where the
Namib and Kalahari Deserts are located, although the east coast at this latitude does not
experience such aridity. The southern coast of the continent enjoys a temperate
“Mediterranean” climate.
Figure 2-4 shows the mean annual rainfall distribution across the continent as described
in the previous paragraphs.
18
10000
Figure 2-4. Mean annual precipitation map of Africa (from Kitamirike (2008)).
There are abundant publications on the climatic history of Africa, which describe
observed trends in precipitation and temperature. Hulme et al. (2001) investigated how
the African climate has changed over the last 100 years and report the following: “The
continent of Africa is warmer than it was 100 years ago. Warming through the 20th
century has been at the rate of about 0.5°C per century, with slightly larger warming in
the June–August (JJA) and September–November (SON) seasons than in December–
February (DJF) and March–May (MAM). The six warmest years in Africa have all
occurred since 1987, with 1998 being the warmest. This rate of warming is not dissimilar
to that experienced globally, and the periods of most rapid warming—the 1910s to 1930s
and the post-1970s—occur simultaneously in Africa and the rest of the world.” While
most of the continent is experiencing warming, large areas of cooling are noted, such as
along the coastal margins of Senegal/Mauritania and South Africa (of up to 1oC per
century), within Nigeria/Cameroon and in Somalia/Ethiopia/Sudan. In contrast, as can be
seen on Figure 2-5, warming of nearly 2°C per century is observed over the interior of
southern Africa and in the Mediterranean countries of northwest Africa.
19
A glance at African news on any particular day starkly reveals Africa’s notorious climate
variability. The following headlines all originated within the same 24-hour period on 29-
30th December 2016: “Namibia drought threatens food security” (The Namibian, 2016),
“DR Congo floods leave 50 dead in Boma” (BBC, 2016), “Uganda: Government to build
Shs4 trillion irrigation scheme in Pallisa …to address a persistent dry spell” (All Africa,
2016), “Sahara Snow Falls Once Again After 37 Years!” (Travelers Today, 2016),
“Drought mitigation package for Zim” (The Herald, 2016), and most tellingly though not
exclusively considering Africa; “Freaky weather the new normal” (The Straits Times,
2016). The inherent and often extreme temporal and spatial variability of the African
climate is well known (Nicholson, 1984; Cooper et al., 2008; Washington et al., 2013).
Interannual rainfall variability is large over most of Africa with multi-decadal variability
having been identified in many regions. Hulme et al. (2001) discuss three regions that
exhibit contrasting rainfall variability characteristics: “the Sahel displays large multi-
decadal variability with recent drying, East Africa a relatively stable regime with some
evidence of long-term wetting, and southeast Africa also a basically stable regime, but
with marked inter-decadal variability. There is no simple correlation between temperature
and rainfall in these three regions.” The pattern of rainfall trends presented in Figure 2-5
shows that parts of the eastern, and particularly the western, Sahel are drying by up to
25% per century. A moderate drying trend is also observed over a large part of southern
Africa. A wetting trend can be seen across much of the rest of the continent especially
over a wide zone south of the equator where wetting by over 10% per century is occurring.
20
Change in annual mean temperature Change in annual mean rainfall
Figure 2-5. Mean linear trends in annual temperature (oC per century) and annual rainfall (% per century)
over the period 1901-1995 (from Hulme et al. (2001)).
Looking further back into history by analysing varied records such as journals, diaries,
tree rings, animal distributions and ice core data, Nicholson (1981) presents historical
climatology for Africa. It is reported that droughts were widespread across Africa in
1820-1840, with higher than average rains during 1870-1895, followed by further
droughts in 1895-1920 being particularly severe in 1913/14. African lake levels (e.g. Lake
Chad, Malawi, Victoria, Tanganyika, etc.) were greatest around 1750 and again around
1850 with much lower levels in between, before 1750 and at all times since 1850. The
Sahel region has become increasingly arid since the 1600s. Africa was very arid during
the ice age peak 18000BP (before present) when the Sahara extended further south and
the Kalahari reached further north (similar to the 1830s and the 1968-73 drought years).
6000-5000BP was wetter with a populated Sahara and greater humidity in east and
southern Africa (like the end of the 1800s).
Sylla et al. (2013) explain how rainfall distribution is critical over Africa for applications
such as drought and flood forecasting, water resources management and agricultural
planning. Simulating and understanding the spatial and temporal precipitation variability
at the necessary timescale is a challenge, as it requires high quality observation data.
21
Datasets are available for Africa blending satellite and ground observations but they
suffer from likely uncertainty due to limitations in density and quality of available
stations.
Maystadt and Ecker (2014) state how a growing body of evidence shows a causal
relationship between extreme weather events and civil conflict incidence at the global
level. For example, they estimate that a one standard deviation increase in drought
intensity and length in Somalia raises the likelihood of conflict by 62%.
Buontempo et al. (2014) present an ensemble climate projection for Africa; a regional
climate model (RCM) for Africa, which improves over the GCM. The model runs from
1949 to 2100 and, they state, it captures annual temperature and rain cycles well but
slightly overestimates rainfall. For scenario A1b, rapid economic growth though with a
balance of fossil and non-fossil energy sources (Nakicenovic and Swart, 2000), the
predictions are:
 Temperature to increase everywhere in Africa by 3-4oC

 General increase in rain between 8oS and 8oN
 Generally becoming wetter in the east, in particular in the Congo basin
 General drying in the west, especially the Guineas
 Reduction in rainfall seasonality, i.e. less rain in the wet season and more rain in
the dry season.
Estimating the impact of climate change on water resources, Döll and Flörke (2005) used
two climate scenarios (A2 and B2, a divided world of self-reliant nations and a divided
but ecologically friendly world, respectively (Nakicenovic and Swart, 2000)) and two
GCMs, to present a global-scale estimation of diffuse groundwater recharge. Their global
scale maps of recharge quantities were validated against ground observations, though
these were mostly streamflow hydrograph assessments, few comparison studies were
from Africa, and those were predominantly in arid and semi-arid locations. The change
in annual recharge from 1961-90 to the 2050s was predicted to be an increase of around
100 mm between the equator and 8oN, as well as in the Horn of Africa and parts of East
Africa, otherwise an increase of about 30 mm in the east. While annual recharge was
predicted to decrease by 100 mm in Southwest Africa and the Mauritania coast area, and
generally decrease by around 30 mm in the west.
However, many authors cast doubt on the predictive skill of GCMs for data-poor regions
such as Africa (Hulme et al., 2001; Cooper et al., 2008; Taye et al., 2015). Bonsor et al.
22
(2010) further state that from a water resources point of view, “the exclusion of daily or
inter-annual climatic variability within GCMs is particularly important, as it is this very
short-term climatic variability which is thought to be highly important in simulating the
effect of intense rainfall events and the future frequency of droughts.” What’s more,
translating the predicted estimates of precipitation totals and intensity into more useful
variables, such as groundwater recharge, is extremely difficult (Bates et al., 2008). As
well as the direct impact of climate change on groundwater resources we also have to
consider the impact of changes in human behaviour (Taylor et al., 2013). For example,
sometimes recharge will decrease due to increased evapotranspiration caused by
warming, while sometimes recharge will increase due to increased irrigation as the
growing season lengthens.
2.5 Land use change

Future changes to land use / land cover (LULC) may be entirely anthropogenic or could
be climate related. Foster and Cherlet (2014) discuss the links between LULC and
groundwater noting how deforestation will increase recharge on flat ground but on
sloping ground there is a risk of soil erosion and eventual loss of recharge. This has been
observed during field visits elsewhere in Ethiopia, such as in Robit Bata kebele near Bahir
Dar and in Boloso Bombe woreda in SNNPR. In both cases, deforestation is said by local
officials to be due to increasing demand for wood for charcoal by a growing population,
in addition to overgrazing. Afforestation, not uncommon in Ethiopia due to increased
demand for eucalyptus for house-building and for charcoal, is likely to decrease recharge
as the evapotranspirative demand will rise (Jagger and Pender, 2003; Farley et al., 2005).
Conversion of native vegetation or pasture to cropped land would generally cause slight
increases in recharge, especially during fallow periods. Obviously under irrigated
conditions, especially flood irrigation, recharge significantly increases (Scanlon et al.,
2005).
Notable examples of the impact of land use change on groundwater can be found in West
Africa. Most of the rivers in the region (except in the Sahel) have seen a significant
decrease in discharge since the 1970s, due to a decrease in rainfall and consequent
lowering of the water table and reduction in baseflow contribution (Mahé and Paturel,
2009). Conversely, the much-discussed “Sahelian paradox” concerns how, despite
drought since the 1960s, groundwater levels in the Sahel and Niger River discharge have
been increasing. For example, the water table in southwest Niger rose continuously by 4
m from 1963-2007 despite a ~23% reduction in monsoonal rains from 1970-1998
23
(Favreau et al., 2009). This is due to vegetation clearance, mostly for firewood and
livestock fodder, in addition to expansion of cropped lands and overgrazing leading to an
increase in bare and crusted soils. Hortonian overland flow has increased forming
temporary endorheic ponds, which then infiltrate creating groundwater “mounds”. Leduc
et al. (2000) estimate a 150% increase in groundwater storage since the 70s. The recharge
rates measured in the Sahel are similar to those from similarly semi-arid regions of
Australia following land clearance (Favreau et al., 2002). Endorheic basins becoming
exorheic have increased the contributing basin area giving rise to more intense but shorter
annual floods creating flooding problems and shorter duration stream flow (Amogu et al.,
2010). Specifically, the Niger River now has a two-flood hydrograph: previously there
was a single flood from June-September rains in the Guinea Highlands, which is delayed
by the Inner Niger Delta so arrives downstream from November. Now there is an
additional local (red) flood caused by August and September monsoons (Descroix et al.,
2012).
Lambin and Ehrlich (1997) showed using continental-scale remotely sensed surface
temperature and vegetation indices that, for the period 1982-1991, climate variability is
responsible for most LULC changes. However, human-driven LULC changes were
observed to have a lower degree of reversibility and are, therefore, cumulative over time.
Concerning the Ethiopian Highlands in particular, Ali et al. (2011) showed that wetter
regions have seen a large shift since the 1970s with croplands replacing pasture. This is
less significant in drier regions as water scarcity prevents a large shift to cropping.
2.6 Research gap

It is a commonly expressed view that the hydrogeology of sub-Saharan Africa (SSA) is
under-studied and poorly understood (Robins et al., 2006; Calow et al., 2009b; Taylor et
al., 2009). In a review titled “Identifying the barriers and pathways forward for expanding
the use of groundwater for irrigation in Sub-Saharan Africa” by Pavelic et al. (2013b),
the first presented major obstacle is the inadequate knowledge of the aquifer systems.
This view is prevalent not just among researchers but is shared by the host governments
who could provide the most benefit to the SSA populace in the form of intervention
planning with better understanding of the hydrogeology. For example, Ethiopia’s
Ministry of Water Resources state: “Ethiopia’s hydrogeology is complex and at present
only partly understood” (MoWR, 2011), and from Uganda’s Ministry of Water and
Environment: “Because of the limited knowledge on the groundwater,... movement of
water across local, national and international boundaries are not known” (MoWE, 2009).
24
This research gap has been succinctly posed by Lapworth et al. (2013) discussing West
Africa; “Ideally, a thorough quantitative understanding of aquifer properties and recharge
mechanisms under a variety of climate, land use and geological environments is required
to confidently assess current groundwater availability, and forecast future availability
under different scenarios”.
Research remains limited into the impacts of climate on groundwater resources (Bovolo
et al., 2009; MacDonald et al., 2009; MacDonald et al., 2011). A significant disadvantage
is the poor availability of groundwater data, e.g. groundwater levels and withdrawals. As
a result, according to Taylor et al. (2013): “our ability to evaluate fully the responses of
ground water to climate variability and change, to estimate directly groundwater
replenishment, and to constrain models and satellite observations, is severely impaired.
There is, for example, a profound lack of knowledge regarding the quantity of
groundwater storage in most aquifers that may be sustainably used.”
The research gap is particularly apparent with regard to shallow groundwater, at least to
shallow groundwater in aquifers at less than 25 m depth. A number of studies have been
conducted on the resilience of African groundwater resources, particularly in the face of
predicted climate change. A study by MacDonald et al. (2011) concluded that
“…groundwater possesses a high resilience to climate change in Africa and should be
central to adaptation strategies. Increasing access to improved groundwater sources based
on handpumps is likely to be highly successful”. MacDonald et al reached similar
conclusions in a 2009 study. However, in both cases “shallow” groundwater resources
were considered those from boreholes of up to 50 m depth. Boreholes, especially at depths
beyond 25 m, are unfeasible for the small-scale groundwater considered as part of this
project, being beyond the technical and financial limit of poorer communities. As such,
many of the conclusions reached by the aforementioned studies are less appropriate for
poor rural communities. This view is shared by Lapworth et al. (2013) who state that
“although shallow groundwater sustains the vast majority of improved drinking-water
supplies in rural Africa, there is little information on how resilient this resource may be
to future changes in climate.”
Shallow groundwater resources are particularly sensitive to variations in recharge. Such

variations are highly likely with land use changes and increasing climate variability
(Carter and Parker, 2009; Taylor et al., 2013; Smerdon, 2017). General circulation model
(GCM) simulations offer a range of possible future climate scenarios for SSA from
increasing aridity to greater rainfall; both predicted for the western Sahel for example
25
(Hulme et al., 2001; Sheen et al., 2017). Most studies agree that an increase in extreme
events, such as intense storms, is likely. How this will translate to changes in effective
rainfall and the partitioning of this effective rainfall between different water resources
through altered patterns of surface run-off, soil moisture and groundwater recharge, is
unclear (BGR, 2008; Owor et al., 2009; Bonsor et al., 2010).
The few examples of published studies concerning shallow hydrogeological systems in

SSA are often quite specific and include a recent paper on permeability variations in
laterite soils in Nigeria by Bonsor et al. (2014), a method for estimating shallow
groundwater availability in small South African catchments by Ebrahim and Villholth
(2016), and many studies on groundwater quality in urban and peri-urban shallow
aquifers (e.g. Onwuka et al. (2004); Kulabako et al. (2007); Takem et al. (2015)). These
studies are in addition to several recharge studies to assess the sustainability of abstraction
of shallow groundwater, such as Edmunds et al. (1991) in NW Senegal and several from
Nigeria (e.g. Carter and Alkali (1996); Goes (1999); Akpan et al. (2013)). Sand rivers,
particularly in Zimbabwe and Botswana, are one shallow aquifer type that has received
more attention in the literature, though generally in the form of technical reports and it is
still typically quoted that greater research is called for to increase understanding of the
long-term sustainability of the resource (Owen, 1989; Davies et al., 1998; Hussey, 2007).
While these examples detail specific aspects of the shallow hydrogeology, there is a
general lack of transferrable shallow hydrogeological studies. This shortcoming was
identified by Taylor and Howard back in 1998 discussing the prevalent regolith shallow
aquifer systems: “… basic questions regarding both the geochemical evolution and the
hydrogeological nature of the regolith [in SSA] remain unsolved. Particular concerns are
the hydrogeological characteristics of the aquifer material, the hydraulic interaction of the
regolith with the underlying bedrock aquifer and the nature of groundwater recharge”;
their study still remains one of very few on shallow regolith aquifers. Another notable
example is the heavily studied 4.6 km2 Romwe catchment in southern Zimbabwe (e.g.
Macdonald et al. (1995); Bromley et al. (1999); Butterworth et al. (1999)). Groundwater
level monitoring, pumping tests, and hydrochemistry analysis from multiple boreholes,
piezometers and hand-dug wells revealed differences in aquifer properties and recharge
quantities in different zones of the regolith aquifer dependent on the nature of the
underlying crystalline basement. Zones with the greatest potential for well-siting were
identified considering geology and topography and a successful (for domestic use and
some small-scale irrigation) large diameter collector well was installed. It is noted that
26
none of the studies cited in this paragraph investigated the resilience of the shallow
groundwater resource beyond conducting a recharge assessment.
Reasons are occasionally suggested for the general lack of investigation of shallow
aquifers. They include: hydrogeological complexity in the case of Ethiopia (MoWR,
2011; Kebede, 2013), and investigations being limited to areas of highest population
density in the case of Ghana (Dapaah-Siakwan and Gyau-Boakye, 2000). Generally,
authors simply state that a better understanding of the shallow hydrogeology from the
point of view of potential agricultural use is a necessity (Giordano, 2006; Namara et al.,
2011; Evans et al., 2012; Pavelic et al., 2013a). A recent review of groundwater
conditions in fifteen SSA countries by Pavelic et al. (2012) concluded that “Quantitative
information on aquifer characteristics, groundwater recharge rates, flow regimes, quality
controls and use is still rather patchy”.
The aims of this PhD research clearly target a research gap that has been often identified
by others. That is: insufficient understanding of shallow aquifers in sub-Saharan Africa,
their potential for productive use, and their resilience. A study of smallholder shallow
groundwater irrigation development in Ghana by Namara et al. (2011), though stated by
the authors to be applicable throughout SSA, concludes that to get maximum benefit from
groundwater, the following are required:
1. “Better understanding of the nature and extent of the existing use of groundwater,
so that it is considered more in national planning and policy.
2. Better understanding of the hydrogeology, so that expansion can be profitably
planned.
3. Reducing some of the other identified constraints, including:
a. provision of land tenure security through innovative institutional
arrangements;
b. provision of decision support tools, such as easy to comprehend
groundwater maps for assessing the precise sitting of wells;
c. improving access to appropriate and affordable drilling technologies;
d. introducing tube-well technology, where applicable;
e. provision of research-based (or founded) extension advise on agronomic
practices (i.e. soil fertility management, crop protection, etc.) and water
management systems;
f. training farmers in safety precautions regarding the handling of agro-
chemicals;
27
g. improving the supply chain of complementary inputs (e.g. improved seeds,
fertilizer, herbicides, etc.); and
h. improving output marketing systems by, for example, organising farmers
using shallow groundwater irrigation into commodity value chains.”
From the above list, points 1, 2 and 3b are directly targeted through this PhD research
while points 3e, 3f and 3h were touched upon, though not always with prior intention in
the case of coming across pesticide application without the use of protective equipment.
28
Chapter 3. Study site background

Chapter 2 provided a background on the potential for a sustainable growth in the use of
shallow groundwater for irrigation by rural communities in sub-Saharan Africa. Whether
or not such an expansion is possible in sub-Saharan Africa is principally dependent on
the hydrogeology and the climate. Following this broad continental-scale study, Chapter
3 will zoom in to the regional and local-scale, providing background information on the
study site for which the research was conducted. Following a report on study site
selection, information is given from desk study, including analysis of received data,
concerning climate, agriculture, socio-economics, governance structures, geography and
geology.
3.1 Site selection

The study site is Dangila woreda in northwest Ethiopia. A woreda is the second smallest
administrative unit in Ethiopia, equivalent to a UK district. Dangila woreda was selected
at the onset of the AMGRAF project in September 2013, 12-months prior to
commencement of the PhD research. The field site was established in March 2014 with
the aid of a catalyst grant under the NERC-DFID-ESRC UPGro research programme. See
Appendix C for further details of UPGro and AMGRAF. Several areas of Ethiopia had
been identified by the Ethiopian ATA (Agricultural Transformation Agency) for an
intensification of agriculture, one of which being the southern portion of the Lake Tana
basin. Several woredas in the basin were considered based on their accessibility, the
dominant farming system, and their status within the Agricultural Growth Programme. In
collaboration with in-country partners, the Geological Survey of Ethiopia (GSE) and
International Water Management Institute (IWMI), Dangila woreda was selected as the
study site. In terms of geology, climate, and level of socio-economic and agricultural
development, Dangila woreda is considered representative of a wide area of upland
Ethiopia. In particular, the study site was chosen to represent an important type of shallow
aquifer. The distribution of this aquifer type can be seen in Figure 3-1 where volcanic
rocks cover a large proportion (~50%) of Ethiopia. The representativeness of the study
site climate can be seen in Figure 3-2 where high rainfall areas are widespread;
approximately 50% of Ethiopia receives >1000 mm/a rainfall and the country-wide
average is 817 mm/a (Fazzini et al., 2015).
29
Study site
Figure 3-1. Ethiopia hydrogeological map (adapted from BGS (2018)).
30
Study site
Figure 3-2. Ethiopia precipitation map (adapted from BGS (2018)).
Dangesheta kebele was chosen as a focus community within Dangila woreda to assess
the potential of the shallow groundwater resource to support increased irrigation use. A
kebele is the smallest administrative unit in Ethiopia, equivalent to a parish or ward.
Dangesheta is one of 27 rural kebeles within Dangila woreda. The selection of
Dangesheta kebele for hydrogeological study followed collaboration with GSE and
IWMI, and a field visit in September/October 2013 by the AMGRAF research team prior
to commencement of this PhD. The rural kebeles were ranked for intervention, according
to:
(i) Access to market, i.e. proximity to an all-weather road and distance to market:
both necessary for the adoption of groundwater irrigation
(ii) Experience in small-scale irrigation
(iii) Potential of shallow groundwater, i.e. evidence of existing shallow
groundwater use. Shallow groundwater is here defined as <25 m: that which
is accessible to poor rural communities using manual excavation methods.
3.2 Country context

Ethiopia is a large landlocked country in the Horn of Africa comprising mostly wet
highlands though with large arid areas in the east and north. It is the second most populous
country in Africa with a population of ~104 million in 2017 and a population growth rate
of ~2.5% per year (World Bank, 2017b). Ethiopia is considered a low-income country by
the World Bank with its per capita income of $590 being significantly lower than the
31
regional average. However, the country has experienced remarkable economic growth
and development in recent years, primarily attributed to two factors:
1. The absence of widespread drought; rainfall and gross domestic product (GDP)
have been shown to be strongly linked in recent decades (Figure 3-3) (Grey and
Sadoff, 2007).
2. A model of development that has driven investment in public infrastructure
(World Bank, 2015).
As a result, Ethiopia has achieved an annual economic growth rate of 10% per year and
has seen a reduction in its population living below the poverty line (income-based
poverty) from 44% to 30% since 2000 (REACH, 2015). Between 1990 and 2015, in
aiming to achieve Millennium Development Goal (MDG) #7 “To ensure environmental
sustainability”, Ethiopia ranked fifth in the world in increasing clean water access to its
rural population with an increase of 37.5 million people with safe access or from 3% to
49% (WHO, 2015). However, the multi-dimensional poverty index (MPI), that considers
health, education and living standards, still places 87.3% of Ethiopia’s population in
poverty (REACH, 2015).
Figure 3-3. The link between annual rainfall and GDP growth in Ethiopia. Above average rainfall since
2006 has led to consistent GDP growth of around 10% (REACH, 2015).
3.3 Site description

Dangila woreda lies approximately 70 km southwest of Bahir Dar, the capital city of the
Amhara Region, in northwest Ethiopia (Figure 3-4). The woreda has an area of
approximately 900 km2 with one significant population centre, Dangila town.
32
Lake Tana
Town/City Bahir Dar
Dangila
Addis Ababa
450 km
Figure 3-4. Location of Dangila woreda field site showing woreda boundary (blue) and position within
Ethiopia (Geographical map of Ethiopia from Ezilon (2017)).
Dangila woreda ranges in elevation from around 1600 m in the southwest to 2400 m in a
central hilly belt, dropping again in the east, which includes Dangila town, to around
2100 m. The northwestern border is formed by an escarpment, which falls over 700 m
towards the Benishangul-Gumuz Region. Much of Dangila woreda is formed of low hills
and expansive floodplains (Figure 3-5). West of the central hills drains to the Beles River,
33
while the east of the woreda drains via the Gilgel Abay River into Lake Tana. The Beles
and Lake Tana are both part of the Blue Nile, or Abay, river basin, the largest tributary
of the Nile that contributes 65% of the total Nile flow (Yates and Strzepek, 1998).
Dangila woreda boundary

River network
Dangila
town
34
Figure 3-5. Geography of Dangila woreda study site (image source: Google.Earth; Imagery ©2017
DigitalGlobe), position within the Lake Tana Basin (map from MoWR), and position within the Nile
Basin (map from PRI (2017)).
Dangila woreda has a population of around 175,000, of which 140,000 are rural (CSA,
2012). Most of the 35,000 urban population reside within Dangila town. Crop–livestock
mixed subsistence farming is the primary source of livelihood.
Studies from nearby areas of the Amhara Region typically show an increasing trend in
deforestation and an increasing trend in land converted to cultivation. For example, a
study by Zeleke and Hurni (2001) in Dembecha woreda, 80 km southeast of Dangila,
showed a 99% decrease in natural woodland since the 1950s and a 95% increase in
cultivated land. In the Koga watershed, 30 km east of Dangila, Yeshaneh et al. (2013)
report a 51% decrease in natural woodland over the same period and an increase in
agricultural land of only 5%, despite an increase in settlement size of 2,733%. This
35
increase in settlement size is a clear indication of the population pressures that the region
is experiencing.
3.4 Agriculture
Rainfed agriculture predominates with the main crops of teff (Eragrostis tef), maize,
barley and millet together making up 90% of the area coverage. The products are mostly
sold locally to local traders and brokers. The current product markets are regarded as
satisfactory and farmers feel that they received a fair price (Belay and Bewket, 2013).
Crop rotation and intercropping (less common) are practised. According to a survey by
Belay and Bewket (2013), conducted in three rural kebeles of Dangila woreda,
approximately 14% of cropland is irrigated, which equates to about 0.20 ha of irrigated
land per household. Despite this small size, irrigated land is seen as important in terms of
both cash income and household nutritional benefits. Irrigated crops are generally
vegetables, fruits and cash crops, e.g. onions, chilli peppers, coffee, rather than the
dominant cereals. In all cases in the survey, streams (86%) and springs (14%) provided
the irrigation water, with diversions constructed from locally available materials, such as
soil, tree branches, stones, and teff straw and chaff. Small groups of 5-10 households, who
share irrigation water from a common source, construct and maintain such systems.
Elected water user committees (WUCs) manage the rotational process of water sharing.
Notable in the study by Belay and Bewket (2013) is the lack of shallow groundwater use
from traditional hand-dug wells (HDWs). This is likely due to the high and hilly
topography, and consequent abundance of streams, of much of the three kebeles studied.
A recent and broader study by Abera (2017) of the Lake Tana Basin revealed that of the
Megech, Gumara-Rib and Gilgel Abay catchments, the latter of which includes the
eastern half of Dangila woreda, around 84% of cultivated land relies solely on rainfall
while 9% utilises flood recession and only 7% is irrigated. Similarly noted is that cereals
dominate the rainfed season comprising over 74% of cultivated land. Abera (2017) reports
“crop production and the rearing of livestock are closely integrated on the small farms,
with livestock utilizing crop residue and providing draft power for ploughing and
transport. Purchase of improved seed and use of chemical fertilizer is common, but per
hectare application is low. Most farms use traditional methods of cultivation, harvesting
and threshing. Only one crop is produced per year in rainfed areas. Crop losses are high
both in the field and during harvesting and storage. The combination of a reliance on
rainfed production, only one crop per year, small and degraded plots, a low use of
purchased inputs, traditional cultivation methods and high post-harvest losses means that
36
production is low and varies substantially from year-to-year. Most farm households in the
[Lake Tana] Sub-Basin are accordingly exceptionally poor. They market only a small
fraction of their total output, and often this is to meet urgent cash needs and requires the
repurchase of staples later in the year.” Whereas fertilizer application occurs for rainfed
crops, pesticides are more common for irrigated crops, though communities are not
adequately informed of the associated hazards. Therefore, farmers use such chemicals
without thorough understanding of the health and environmental impacts. Similarly,
farmers are apparently unaware of recommended seed rates for most crops. Too low or
high seed rate, combined with poor land preparation, reduces productivity (Abera, 2017).
In December 2013 as part of the AMGRAF project, Dr Elizabeth Oughton (Newcastle

University) and Dr Gebrehaweria Gebregziabher (IWMI), working with officers from the
Dangila Woreda Agricultural Office, conducted focus groups in two rural kebeles. The
investigations revealed that decisions over cropping, in both fields and backyards, are
overwhelmingly made by males of the household, even though females and children
provide labour for backyard agriculture. Secondly, there are a high proportion of female-
headed households in this region, as women that have been widowed, divorced or
abandoned are not permitted to remarry. Although women may retain legal ownership of
fields, they require male labour to farm them. Clearly, changes in the availability and
management of irrigation water could have very different effects on men and women
affecting relative poverty, livelihood and environment (Oughton and Gebregziabher,
2014).
At the onset of the AMGRAF project, Dr Jaime Amezaga investigated the governance
aspects of water use in the region: Ethiopia has a history of watershed management
initiatives dating back to the 1970s. The basic approach has shifted from top-down
planning to community-based approaches. There is now a supportive policy and legal
framework in the form of policies that facilitate decentralised and participatory
development, institutional arrangements that allow and encourage public agencies at all
levels to work together, and an approach to natural resources that reflects local legislation
and tenure practices. The institutional and legal framework designed by the Ministry of
Water Resources (MoWR) promotes farmer-managed small-scale irrigation through the
establishment of irrigation user communities (IUCs) under the national cooperative law
starting from 2002. Recently, 233 IUCs were reported to have been established in the
region, but their success is variable. Recently prepared (2010) draft regulations for
Irrigation Water Users Associations are currently under consideration (Amezaga, 2014).
37
3.5 Climate
According to the Köppen–Geiger climate classification system, this region of Ethiopia is
categorised as humid subtropical (Peel et al., 2007). There is little annual temperature
variation though high diurnal variation due to the elevation. A median annual daily
maximum temperature of 25 oC and minimum of 9 oC have been measured at the National
Meteorological Agency (NMA) weather station in Dangila. The median annual total
rainfall is 1541 mm, as measured (since 1987) at the Dangila NMA weather station, 91%
of which falls during May to October (Figure 3-6). The main June-September rains,
known in Ethiopia as kiremt, are principally controlled by the seasonal northward advance
of the inter tropical convergence zone (ITCZ), in addition to the upper-level tropical
easterly jet (TEJ) and convergence in the Red Sea region (Conway, 2000). Both the Choke
Mountains to the east and Lake Tana to the north affect the pattern of rainfall in the study
area. Most rain events are convective, have a duration shorter than 1-hour and often occur
in the late afternoon (Haile et al., 2009).
500 30
Rainfall, c90
Rainfall, median 25
400
Rainfall, c10
Temperature (oC)
20
Rainfall (mm)
300 Tmax
Tmin 15
200
10
100
5
0 0
J F M A M J J A S O N D
Figure 3-6. Monthly median, 10th and 90th percentile rainfall, and mean maximum and minimum
temperatures as measured (since 1987) by the NMA at the Dangila weather station.
The region experiences high interannual variability in rainfall, with historical annual
rainfall totals ranging from below 1000 mm to over 2000 mm. Interannual variations in
rainfall total are related to cyclone development in the southwest Indian Ocean;
specifically, a high frequency of tropical cyclones causes a delay in the onset of the rainy
season, due to a failure of the early short, or belg, rains, throughout the Ethiopian
Highlands (Shanko and Camberlin, 1998). Variation in kiremt rainfall total is related to
the El Niño–Southern Oscillation (ENSO) and TEJ; specifically, a strong ENSO or a poor
38
TEJ often results in drought, though ENSO and TEJ may temper the effects of each other
and the exact interrelationship remains uncertain (Seleshi and Zanke, 2004; Segele and
Lamb, 2005; Diriba and Anthony, 2007). Analysis of this climate variability can be found
in Chapter 8.
A single NMA weather station is present within Dangila woreda, situated in Dangila
town. Another weather station, and the NMA regional office, is located in Bahir Dar.
There are four further rain gauges in villages along the road between Dangila and Bahir
Dar, and three rain gauges in the hills to the south of Dangila towards Addis Ababa.
Further information and analysis of these monitoring sites is presented in Chapter 5.
3.6 Geology
The 1:2,000,000 scale Geological Map of Ethiopia by Tefera et al. (1996) states that the
geology of the area predominantly consists of Quaternary basalt and trachyte above
Eocene-Oligocene flood basalts and trachyte. There is much disagreement over the
thickness of the flood basalts in northwest Ethiopia, ranging from 250 m (Hautot et al.,
2006) to 1500 m (Pik et al., 1998) to ~4000 m (Hofmann et al., 1997), though the
generally accepted range is 500-3000 m (Mohr, 1983). There is further disagreement over
the age of the flood basalt formations that cover 25% of Ethiopia’s land surface (Figure
3-7), with some studies stating the entire vast volcanic plateau sequence was erupted in
an event lasting just a million years, approximately 30 million years ago (Hofmann et al.,
1997). These Ethiopian flood basalts are a classic example of mantle source continental
flood volcanism and the youngest global example of a major continental volcanic plateau
(Kieffer et al., 2004).
39
B
Figure 3-7. Generalised Cenozoic geology of the regions bordering the Main Ethiopian Rift (MER)
system (simplified from the Geological Map of Ethiopia). Circled “A” is the location of Addis Ababa and
“B” is the Dangila woreda study site. From Prave et al. (2016). Note that there are some discrepancies
between the geological units, their ages and positions, on this map, the Figure 3-8 map, and the
descriptions within the text. This reflects the lack of agreement in the published literature.
The deep geology at the study site was traditionally known as the Termaber flood basalts,
though is more recently termed the Upper Basalt sequence (Kebede, 2013). Another series
of volcanics are present in the area, though there is some uncertainty as to whether these
are present above the flood basalts at the field site: The Miocene-Pliocene shield
volcanics, erupted from large shield volcanoes (approximately 22 Ma) that today form
the Choke Mountains to the southeast of Lake Tana (Figure 3-8), contain more rhyolitic,
trachytic and ash layers than the flood basalts (Kieffer et al., 2004; Kebede, 2013).
Overlying this thick flood basalt and/or shield volcanic sequence across a large area south
of Lake Tana are more volcanic rocks of middle-Pleistocene to Holocene age, i.e. 10,000
40
to 1 million years old (Mohr, 1963). These thinly bedded and often scoriaceous basalts
and trachytes were erupted from relatively small and local Strombolian volcanoes
(Kieffer et al., 2004; Kebede, 2013), many of which are still visible, the locations are
shown on Figure 3-8. Largely impermeable dykes, sills and faults are present within the
region though these are most likely confined to the pre-Quaternary geology (Kebede,
2013) and, as such, should have little impact on the shallow aquifer.
Figure 3-8. Geological map of the northern part of the Ethiopian plateau depicting the distribution of the
volcanic rocks, shield volcanoes and Strombolian cones (from Ayalew (2011)). MER = Main Ethiopian
Rift.
41
Lake Tana itself had been believed to have been formed by the intersection of three graben
rift systems (Chorowicz et al., 1998). However, a recent study suggests that Lake Tana is
a collapsed caldera created during a super eruption around 30 million years ago (Prave et
al., 2016).
The basalts and trachytes found around Dangila woreda have had little time, less than
1 million years, for weathering and regolith formation in comparison to the majority of
the African continent. However, a review of various physical and chemical weathering
experiments of igneous rocks by Cawsey and Mellon (1983) indicates that significant
physical weathering may occur over relatively short periods of time (years), whereas
chemical weathering processes require much longer periods before the effects are readily
observed, though still just 100s to 1000s of years. Indeed, the Pleistocene-Holocene
volcanics south of Lake Tana have been observed to be more highly weathered than the
10-30 million years older flood basalts in the surrounding area (Poppe et al., 2013).
3.7 Regolith hydrogeology

Weathered materials that can be considered regolith overlie the Cenozoic volcanic rocks
of the study site. The generally accepted definition of “regolith” is the unconsolidated
heterogeneous material, including soil, which overlies the bedrock (Merrill, 1897).
Regolith consists of physically broken and, generally, chemically altered rocks (Scott and
Pain, 2009). While sometimes considered a synonym, “saprolite” refers to in-situ
weathered materials, whereas regolith may include transported materials (Taylor and
Eggleton, 2001). In this thesis, “regolith”, “weathered regolith” and “weathered mantle”
are used interchangeably.
Acworth (1987) describes the typical weathered profile, which forms above crystalline
(including volcanic) bedrock: Four weathering zones are present between the fresh rock
and the soil (Figure 3-9). Each zone is always present but may be so thin as to be
insignificant. The interface between zones is generally planar as it is related to the water
table but the interface between zone ‘d’ and bedrock can be highly irregular. Zones ‘a’
and ‘b’ are generally clay-rich and have low hydraulic conductivity. Zone ‘c’ acts as
storage and fractures in zone ‘d’ are the transmissive parts. If hydraulic conductivity is
high in zone ‘c’, then drainage will occur from zone ‘b’ above.
42
Figure 3-9. Typical weathering profile developed upon crystalline rocks (from Acworth (1987))
Acworth (1987) goes on to state that recharge is highest where zone ‘d’ directly underlies
stony soils whereas minimal recharge occurs where thick zones ‘a’ and ‘b’ exist.
However, there will be lateral flow between horizons. Hilly or plains areas have a higher
groundwater potential as weathering basins will have coalesced laterally as relief has been
lowered, with the result that an extensive zone ‘c’ saprolite aquifer has been produced
extending into the area previously occupied by the zone ‘d’ fractured material. Reduction
in relief also causes a reduction in groundwater flow as water table contours are a subdued
reflection of surface topography. In general, well sites should be chosen towards the
middle of the slope where the depth of weathering and saturated material is
maximized.
Jones (1985) reports that saprolite profiles tend to have similar characteristics over a wide
variety of rock types. This conclusion is extremely important for this PhD project. It
means that findings from shallow hydrogeological assessments conducted at the field site
may be transferrable across SSA, even to areas with different bedrock geology. Jones
(1985) states that geomorphological development by cyclic erosion has resulted in a
predictable distribution of the regolith aquifer and the analysis of data from over one
thousand wells shows it to be hydrogeologically uniform. This uniformity is traced to the
geomorphological development of the landscape; of lesser importance is the mineral
43
composition of the parent rock. However, some studies have identified differences in
regolith hydrogeological characteristics occurring over short distances related to local
variation in parent geology (e.g. Macdonald and Edmunds (2014)). Jones (1985)
description of a typical weathering profile above crystalline rocks in Africa is in
agreement with Acworth (1987) (Figure 3-9) and he further adds that the lower part of
the profile, the weathering front, if freely draining with active groundwater flow, will
preserve a relatively clay-free, gravel-like texture. The grain size of the saprolite
diminishes upwards until the clayey part of the saprolite is encountered and where the
groundwater flow is impeded. Furthermore, chemical weathering occurs without
significant volume change (indicated by undisturbed persistent quartz veins) though
dissolution removes mass causing structural unloading allowing deeper joints and fissures
to open, which allow deeper groundwater penetration. The transferability of the findings
from this study to other types of geology in SSA is discussed in Chapter 9.
44
Chapter 4. Field investigations and development of conceptual
model

Chapter 3 presented background information on the study site. The process provided
understanding of the climate, geology, agriculture, socio-economics and governance
issues prior to the field visits and enabled planning of fieldwork. Chapter 4 provides
descriptions and analysis of information from field investigations, essentially detailing
how the hydrogeological conceptual model was derived.
4.1 Field visits

Three field visits were conducted to Dangila woreda, in March/April 2015 at the end of
the dry season, in October/November 2015 at the end of the wet season, and in
January/February 2017 during the dry season. The first two visits were for approximately
a month with around 2-3 weeks in the field in Dangila and the surrounding area, the
remaining time being spent with local partners in Addis Ababa and Bahir Dar. The third
visit was for 2.5 weeks, principally spent in Boloso Bombe woreda in SNNPR Region in
southwest Ethiopia for a different research project, though with a few days in Dangila
woreda.
4.2 Aims
The aims of the field visit were to develop a conceptual model for use in subsequent
recharge and modelling studies. The field investigations conducted during this process
allowed the first research question from Chapter 1 to be evaluated, namely: Do shallow
aquifers have the requisite properties, in terms of hydraulic conductivity, potential well
yield, specific yield and aquifer geometry, for productive groundwater use?
4.3 Study site observations

Situated on the Addis Ababa to Bahir Dar highway, Dangila town is a popular transport
stop with more services, such as; markets, hotels and banks, than are seen in similar sized
towns in nearby woredas. Being well connected on the transport network, there is
potential for an increase in agricultural production due to good access to markets. The
majority of the woreda is devoted to agriculture. Seasonally inundated
floodplains/grasslands are utilised as pasture with mixed cropping and dwellings
occupying the adjacent slopes (Figure 4-1). Natural woodland is generally only found
around hilltop churches and along more steeply sloping riparian strips. The higher steeper
45
mountains often have thin soils and are covered with low scrub-like vegetation. Abera
(2017) notes that much of the steeper highland areas have been degraded through overuse
and erosion. Dangila town is by far the largest settlement in the woreda and, other than
Abadira town to the north, and Chara and Giza in the west, dwellings and the population
are quite scattered.
Figure 4-1. Typical wet season and dry season scenery in Dangila woreda of floodplain pasture
surrounded by low hills with small (<1 ha) individual plots of rainfed agriculture.
The flatter kebeles in the east of the woreda, e.g. Dangesheta, Zelesa, Zeguda and Workit,
that were most frequented for this PhD research exhibit very few stream diversions like
those observed by Belay and Bewket (2013) during their study of irrigation in the woreda.
However, most households have their own HDW for backyard irrigation. Such plots
generally occupy <0.25 ha and are more likely to be planted with fruits, vegetables, coffee
and khat (Catha edulis) than the main crops listed above. The typical observed backyard
irrigation system involves a rope and bucket and a watering can. Some smallholders have
the more efficient pulley and double bucket system while others have rope-and-washer
pumps often supplied by NGOs. Only one treadle pump was observed among the
hundreds of wells visited during the three field visits. Despite the advantages of being
less energy-intensive, locally-made and low cost (1300-3000 Birr or $60-150), broken-
down rope-and-washer pumps were frequently observed. Indeed, a study by MetaMeta
(2014) revealed that 35-50% of rope-and-washer pumps in the areas of Ethiopia they
surveyed were non-functional.
The main rainfed crops have a 5-8 month cropping cycle and are harvested in October
following the first substantial dry spell at the end of the rainy season. This dry period is
necessary to dry out the crop thus an extended wet season, according to the local
community, is a hindrance. The generally flat and seasonally inundated floodplains are
46
almost exclusively used for pasture for cows, sheep and goats, usually tended by local
children. Other observed floodplain activities were occasional observations of harvesting
wetland vegetation for animal fodder and small plant nurseries can sometimes be found
beside perennial reaches of streams during the dry season. While the floodplain marginal
areas are sought after for cropping to take advantage of residual moisture and shallow
water tables, cropping is largely non-existent within Dangila woreda on the floodplains
themselves.
4.4 Hydrology
During the dry season field visits, rivers were commonly observed to have dry reaches
between areas of flow. Similarly, flowing springs often form small streams, which later
dry up. Dry and low-flow reaches are commonly observed on large flat floodplains where
the rivers are losing water to the underlying sediments whereas upstream in narrower
steeper valleys there may be substantial flow. An example is the Amen River, which flows
through the centre of Dangila town but later dries up upon reaching the large floodplain
of the Kilti River (Figure 4-2).
Figure 4-2. The flowing Amen River at the gauge location in Dangila town (left) and the stagnant or dry
Amen River (right) approximately 5 km downstream shortly after reaching the extensive Kilti floodplain.
The floodplains become inundated during the wet season from spring discharge at their
edges and from pluvial flooding as opposed to overbank flow. These features appear akin
to “dambos”, which are discussed in the following section.
To increase understanding of the hydrology, gauge boards were installed in the Kilti and
Brante rivers that were monitored by the local community. Additionally, a manual
raingauge was installed and groundwater level was monitored in five hand-dug wells.
Further detail is provided of the community-based monitoring programme in Chapter 5.
47
4.5 Dambos
The definition of a “dambo” is under debate in earlier studies but is generally as described
in a review paper by von der Heyden (2004): “… shallow, seasonally waterlogged
depressions forming the headwaters of ephemeral and perennial streams in subtropical
and tropical Africa.” Dambo profiles are “primarily concave, with shallow slopes and
gradients of less than 6o (usually less than 2o). The size and shape of the dambo surface
in plan vary widely, with dambos ranging from several square kilometres of wide, oval
wetland to narrow, tortuous structures barely 100 m in length.” This definition matches
the pervasive floodplain wetland land form observed in Dangila woreda, as do the
characteristics, such as: (non-calcic) soils characteristics of 30-50% coarse sand, <10%
clays, low EC, and pH 5.3-6.5, and; a clay layer from in-situ weathering that forces soil
water to discharge at the level of the dambo at dambo verges.
Studies on dambos are predominantly from Malawi, South Africa, Zambia and
Zimbabwe, and though the term appears in papers concerning Ethiopia, specific dambo
studies are non-existent. Published literature on dambos generally ranges from the late
1980s to the early 2000s then dries up (attempted contact with authors has not revealed
why this occurred).
Uncertainty and conflicting hypotheses exist in the literature concerning the role dambos
play within catchments, relating to evapotranspiration (ET), dry season baseflow, and
attenuation of flood flow. The review paper by von der Heyden (2004) states:
1. Evapotranspiration from a dambo will likely be greater than from interfluves when
the latter is vegetated by short grasses and shrubs.
2. Baseflow and dry season flow augmentation is primarily a function of aquifer
groundwater discharge, with a secondary contribution from surface water storage
within a dambo. However, viewing the surface water as a separate entity from the
groundwater system is illogical as the surface water is an above-ground extension
of the groundwater table.
3. Storm flow is retarded and attenuated during the early wet season, through soil
infiltration and dambo filling. The extent of this retardation and attenuation is a
function of the soil characteristics. Following saturation of soils, the dambo effects
little influence over storm flow, with flashy responses to rainfall events noted.
A geochemical study on a dambo in the Zambia copperbelt with long term monitoring by
von der Heyden and New (2003) discusses the three issues in contention with the role of
48
dambos, agreeing with the points presented above: ET is increased but the main outlet of
a dambo is surface water flow; floods are retained at the onset of the wet season but the
dambo quickly fills and has no effect on downstream floods; dambos only contribute to
dry season surface flows until mid-dry season – later in the dry season dambos must be
fed by deeper aquifers. The study site has a two-aquifer system – the shallow regolith and
the deeper aquifer – akin to Dangila, and the hydrochemistry results for wetlands, deep
groundwater and surface water also match.
An earlier review paper by Bullock (1992) generally matches the conclusions drawn by
the von der Heyden (2004) review. Bullock states that ET is highest at dambo edges
though these areas make up only 10% of a dambo. The paper agrees that dambos are not
so significant for low-flow augmentation and references to this in many papers are from
misunderstandings of available literature.
McCartney and Neal (1999) present a case study from Zimbabwe with coarse to medium
loamy sand soils and a low-permeability clay lens. Low slope and low hydraulic
conductivity (K) mean that when saturated most flow is over the surface rather than
through the dambo. Deeper groundwater showed higher alkalinity indicative of
weathering, whereas shallow groundwater did not, confirming that the clay lens acts as a
barrier to upward flow. The study suggests pipe-flow is significant in natural pipes ~0.4
m below surface. During the wet season, 70% of storm flow is “new” water; 10 days
following a storm event, most water is “return” water having passed through soils first.
Contradictory to other studies, they state that most water loss is via ET, though they do
note that there is still much uncertainty.
A lengthy technical report by McFarlane (1989) describes the contrasting theories of

dambo formation: (i) fluviatile, with fining up sediments, or (ii) within irregularly
lowered land formed by differential leaching with colluvium fill. McFarlane (1989)’s
Malawi example argues against a fluviatile formation because some dambos are circular,
others cross watersheds, others are discrete (endorheic); exactly as observed in
Dangesheta (Figure 4-3). This is suggestive of formation due to irregular lowering of the
land caused by differential leaching. Further arguments against fluviatile formation
presented by McFarlane (1989) include:
 Dambos often contain smectite but it is not seen in interfluves because it is from
local weathering
49
 Some dambos have stepped forms which would have required multiple climate
change episodes
 There is a lack of sufficient energy in low gradient streams with little higher
catchment for formation of such features
 There is a lack of stratification of sediments with often vein quartz within clays.
Also described is the dambo-peripheral zone being characteristically sandy, which

“shows up well on the air photos as a light toned belt”; this can be seen on Google.Earth
satellite imagery around Dangila (Figure 4-3). At the Malawi study site, the dambo
peripheral belts are preferred localities for HDWs because the water table is near to the
surface; again, such a pattern can be observed in Dangila. Mechanisms for dambos
in-setting in the landscape include: 1) repeated dissolution and deposition (i.e. as seen in
African bauxite and laterite terrains), 2) physical subsidence as saprolite loses mechanical
strength, 3) upstream retreat of gully head and subsequent deposition on downstream
dambo. The lack of bauxite and laterite terrains makes mechanism 1 inapplicable at
Dangila. The geomorphology of the narrower, steeper dambos in hilly areas are
suggestive of mechanism 3, though mechanism 2 is likely to be dominant in Dangila
woreda. When discussing mechanism 2, McFarlane (1989) talks about HDWs in dambos
being notorious for collapsing; this may be another reason why the Dangila local
communities do not excavate wells in dambos and the only HDWs observed in dambos
are concrete-ring-lined MoWR installed handpumps. Many authors suggest grass slows
surface water flows off of dambos at the onset of the wet season, a delay of around a
month (McCartney and Neal, 1999), though McFarlane (1989) suggests it actually takes
a month for cracks to be closed by swelling clays before runoff is promoted. Interfluve
runoff would penetrate a dambo at the sandy periphery rather than over the surface; the
lack of surface deposits is evidence of this.
50
Largely endorheic
floodplain basins
MW4 Symbols indicate
surveyed wells
and springs
MW5
MW3
MW1 and
rain gauge N
MW2
500 m
Brante river
“Hanging” floodplains
with restricted outlets
Figure 4-3. “Hanging” floodplain basins above the Brante river valley and the locations of the community
monitored raingauge and hand-dug wells, Dangesheta kebele. Image source: Google.Earth; Imagery
©2015 DigitalGlobe).
4.6 Geology
Outcrops are visible in riverbeds, occasionally on steeper slopes and in a few man-made
excavations. Examples of the geology are more commonly visible as large boulders
within superficial materials, particularly in riverbeds and banks (Figure 4-4). The basalts
are variously massive, fractured and vesicular with variations occurring in short distances.
The more massive basalt often forms higher ground with valleys and floodplains
overlying more fractured and vesicular basalt, which is more easily weathered and eroded.
Figure 4-4. Vesicular basalt boulders in the Brante riverbed, Dangesheta kebele (left). As-dug weathered
basalt beside a well under construction on the edge of the Brante river floodplain.
51
Above the solid geology lies weathered basalt regolith, itself overlain by red clayey loam
soils. The red soils become more lithic and clayey with depth, grading into the regolith
usually with no obvious boundary. As-dug materials beside wells under construction
show the degree of weathering decreasing with depth. The regolith becomes greyer,
stronger, and must be chiselled as it deepens though it is still quite friable (Figure 4-4).
Local communities report there are rarely problems with well sidewall collapse. The most
friable and excavatable regolith is the result of weathering of low-density vesicular or
scoriaceous basalt. Often solid geology is reached abruptly and well excavation is halted.
The superficial materials underlying the floodplains are often browner in colour being
more organic-rich. Deep and wide desiccation cracks suggest a high clay content, though
the alluvial materials are occasionally very sandy and gravelly.
Depth to the top of the solid geology is variable. Wells are typically excavated until
further excavation becomes impossible, therefore, the location of rockhead can be
inferred from well depth. Over all the field visits, 80 wells were measured for estimation
of regolith thickness; more wells were visited but access for measurement, such as in the
case of wells fitted with handpumps, was not always possible. Rockhead was generally
found to be deeper in more steeply sloping areas and shallower in floodplains. For
example, in the north of Dangila woreda in Afafe Eyesus kebele, a hilly area with large
slopes, wells were found to be 14 to 17 m deep (Figure 4-5), whereas wells adjacent to
the floodplains in much flatter Dangesheta kebele are often just 3 to 4 m deep. Rivers
have often incised to the level of the rockhead where solid basalt forms the riverbed with
banks of only 1 to 3 m in height (Figure 4-5). Measurements and information from the
well and river surveys are presented in Appendix F.
Figure 4-5. Very deep (approximately 15 m) exposure of regolith in a gully on a large slope, which leads
to the Amen River, Kuandisha kebele (left). Massive and fractured basalt bedrock exposed in the bed of
the Gizani River, Sehara kebele. Note the low (~1 m) river banks in this floodplain.
52
In a study of gully formation and upland erosion by Tebebu et al. (2010) at a site south
of Lake Tana, approximately 50 km from Dangila woreda, they note that the area consists
of vertisols, which comprise only soils’ A-horizon then ‘C’ (no soil B-horizon) (see
Figure 3-9). The study further reports that the area is underlain by shallow highly
weathered and fractured basalt. The fractures are highly interconnected with limited clay
infillings. Surface exposures of weathered basalt (saprolite) can be found on the hilltops
and in mid-slope areas on the hillsides. A black clayey layer, regolith zone ‘b’, (note this
is different to the soil B-horizon as shown in Figure 3-9) is often present above the basalt
and a brown silty loam is common (zone ‘a’), as well as a compact stony friable layer
(zone ‘c’). These observations match those made during fieldwork for this PhD project in
both Dangila woreda and north of Bahir Dar at Robit-Bata kebele (when assisting a Bahir
Dar University student with his PhD fieldwork); these two locations being either side of
the study area of Tebebu et al. (2010). Such complementary observations indicate that
the “typical” regolith profile reported by Acworth (1987), Jones (1985), and others, is
applicable to this study site and the surrounding area.
4.7 Hydrogeology
In addition to season, topography appears to govern shallow groundwater availability.
The variations in geology are sufficiently subtle, particularly concerning the regolith,
which forms the shallow groundwater aquifer, to be less of a control on the hydrogeology
than geomorphology.
Near the end of the dry season in March/April within the floodplains where the solid
geology is at a depth of around 4 m, the water table lies at 2-4 mbgl (metres below ground
level). The water table can often be seen as a seepage face at this depth within riverbank
sections in floodplain sediments (Figure 4-6). However, on the larger and steeper slopes
where rockhead is around 15 m deep the water table is at a depth of 12-15 m. Thus, the
shallow aquifer is thicker on slopes giving deeper water tables and generally greater
saturated aquifer thickness though possibly with lower hydraulic conductivity (as
suggested by pumping tests). Superficial deposits are thinner in floodplains and the
material has higher hydraulic conductivity due to the visible sand and gravel content and
the possible presence of enhanced fracturing below the floodplains (refer to the earlier
discussion on dambos).
53
Figure 4-6. Visible water table in the banks of the Amen River, Dangila kebele (left). Massive basalt
boulders visible within regolith in a cutting for an irrigation canal near Giza.
It is noted that farmers often talk of a well excavation striking rock at a shallow depth and
being dry, then when the well is relocated a short distance away (~10 m) rock is struck at
greater depth and the well fills with water. Such a situation is commonly ascribed to
heterogeneous rockhead, however, the unsuccessful wells are perhaps more likely to be
due to the presence of large and massive basalt boulders lying higher in the weathered
profile as are often visible in riverbank sections (Figure 4-6).
Despite the shallow aquifer being considered the regolith above the solid geology, it is
likely that fractures in the upper layers of the bedrock are influential to the
hydrogeological regime. However, fissure flow is unlikely as any fractures are probably
filled with weathered material with similar properties to the overlying regolith. The
precise depth and degree of fracturing of the solid geology is very difficult to estimate
without subsurface investigations or geophysics. Heterogeneities within the weathered
basalt regolith, such as the clay content and the fractured or vesicular nature of the pre-
weathered rock, determine the productivity of a well, though this is similarly very difficult
to assess prior to excavation.
During the AMGRAF catalyst period in February and March 2014, 143 hand-dug wells
were surveyed by Demis Alamirew of GSE. A further 64 wells were surveyed by me
during field visits in March/April 2015 and October/November 2015. The surveys
included GPS location, depth and water level measurements, description of geology,
topography, land use, pump/lifting device and cover, in-situ measurement of water
temperature, pH and electrical conductivity, and discussions with local community over
the well’s use, seasonality and history. The information gathered during the water point
surveys is tabulated in Appendix F.
54
Water point surveys also included assessment of springs, many of which are used by the
local community, whether developed or not, to collect water for domestic and potable
use. Flow rates vary from over 20 l/s at Lunk in Sehara kebele, where water is piped to
tanks to supply the towns of Giza and Chara, to unmeasurably small seepages, though
often over a large area giving a combined high total flow rate and often forming streams
(Figure 4-7). Where springs and seepages emerge from gullies they commonly occur at
contacts between regolith and bedrock or gravelly regolith and more solid regolith (Figure
4-7). Springs and seepages are also very common around the edges of floodplains where
the water table from the surrounding slopes intercepts the ground surface (Figure 4-8).
Figure 4-7. Small spring emerging from contact between gravelly and more solid regolith (left). Large
area of seepages emerging at contact between regolith and massive basalt bedrock forming the riverbed.
Figure 4-8. Developed spring in the centre of a floodplain, which is completely submerged in the wet
season (left). Spring emerging at the end of the wet season around the edge of a floodplain.
4.8 Pumping tests

Eight pumping tests were conducted on shallow hand-dug wells in order to gain
information on the shallow aquifer properties. This particular aspect of the field
investigations was presented at the 7th RWSN (Rural Water Supply Network) Forum
55
"Water for Everyone" in Abidjan, Côte d’Ivoire, in November/December 2016. The
resulting peer-reviewed conference paper titled “Properties of shallow thin regolith
aquifers in sub-Saharan Africa: a case study from northwest Ethiopia” (Walker, 2016), is
presented as Appendix A. The paper essentially states that the drawdown and recovery
were analysed separately, applying the Moench (1985) and Barker and Herbert (1989)
methods respectively, providing consistent results, confirming suitability of methods.
Hydraulic conductivity estimates ranged from 0.2 to 6.4 m/d (mean = 2.3 m/d, median =
1.6 m/d) in the dry season and ranged from 2.8 to 22.3 m/d (mean = 9.7 m/d, median =
6.5 m/d) in the wet season when the water table was higher. This difference indicates the
importance of excavating wells as deeply as possible to increase the likelihood of
intercepting more transmissive layers. Specific yield estimations have a wider range
(0.00001 to 0.32) and are more uncertain though the mean of 0.09 (median of 0.08) is
reasonable. Estimates of well yield average 0.5 l/s though this increases to >1 l/s in the
wet season; giving optimism that small-scale abstraction and irrigation is achievable.
4.9 Hydrochemistry
A water sampling and in-situ testing programme was undertaken with the aims of:
identifying water types, assessing aquifer connectivity, groundwater aging, assessing
recharge mechanism, analysing the consistency of the hydrochemistry, and to identify
losing and gaining reaches of surface water. The water sampling and in-situ testing
programme undertaken during the field visits is described in Appendix B and details of
the sampling and testing locations can be found in Appendix F. In summary, 49 samples
of shallow and deep groundwater, surface water and rainwater were sampled during the
first two field visits; many of these being repeat samples from the same locations.
Laboratory analysis involved measurement of major ions, some trace elements, and stable
isotopes oxygen-18 and deuterium (δ18O and δ2H). In-situ testing involved measurement
of pH, electrical conductivity (EC), total dissolved solids (TDS), temperature, and radon-
222 concentration.
The shallow groundwater is consistent in chemistry both spatially and temporally.

Residence time is low, indicated by low EC and ionic concentrations, suggesting that the
resource could be vulnerable to drought. Surface water and shallow groundwater belong
to the “bicarbonate calcium” type typical of recent recharge. The deep groundwater is of
“bicarbonate sodium” type indicative of higher mineralisation due to longer residence
time and greater distance of flow. The shallow groundwater samples from the wet season
are very similar in chemistry to surface water samples indicating a high degree of and
56
rapid interconnectivity. This was expected in the wet season from the observed very
shallow water table. There is no hydrochemical evidence to suggest mixing between the
shallow and deep groundwaters; they belong to clearly different water types. What’s
more, Radon-222 measurements showed the opposite of what would be expected if
surface water and shallow groundwater were being drawndown by abstraction from the
222
deep boreholes: Rn concentrations would be lower in the vicinity of the abstracting
boreholes as groundwater discharge would be prevented but the reverse was measured.
Radon-222 measurements did suggest that the large floodplains in river valleys are areas
of groundwater discharge from the shallow regolith aquifer, whereas the narrower valleys
with basalt riverbeds are not discharge areas. However, it should be noted that the faster,
more turbulent flow through rocky reaches would have a degassing effect on the river
water thus reducing radon-222 concentrations (Cook et al., 2003).
Other interesting findings resulted from comparing individual samples collected during
the same visit. Wells reported by the community to have good year-round supply often
showed greater stable isotope enrichment and higher ionic concentrations than would be
expected from their topographic position close to a flow divide. Often, across the flow
divide, there was a dambo. The higher enrichment (through evaporation when inundated)
and higher concentrations (due to the longer residence time) suggest the sampled water
originated in the dambo, which provides continuous groundwater supply through the dry
season with groundwater flow paths contradicting surface water flow paths.
Considering only hydrochemistry and not microbial content, the analyses indicate that the
shallow groundwater tested is suitable for both irrigation and domestic use: Ethiopia is
known for having problems with fluorosis caused by excess fluorine in groundwater
(though the issue is typically confined to the Rift Valley and deep boreholes, (Tekle-
Haimanot et al., 2006)) but here F- levels were below the WHO recommended maximum;
nitrate was suspected to be a possible contaminant due to the proximity of pit latrines and
wells but NO3- levels were also below the WHO limit, and; sodium adsorption ratios
(SAR) were well below the acceptable limit for irrigation water.
4.10 Conceptual model

The shallow and deep aquifers have little interconnectivity while the shallow aquifer and
surface water are in connection particularly during the wet season when large expanses
are inundated. Recharge is rapid and groundwater residence time is low.
57
The clay content of the floodplain sediments means that early rains may quickly create a
low-permeability layer at shallow depth, which may exacerbate flooding as surface water
would pass overland into the river systems and local recharge would be restricted.
However, the constricted floodplain outlets restrict discharge, thus creating wetlands that
increase infiltration through coarser lenses or slowly through the low-K layer. If there is
a surface water / groundwater disconnection, it probably only occurs at the beginning of
the wet season and during dry season rains when the water table is deeper. During the wet
season, because the shallow aquifer under floodplains is only a few metres thick,
sufficient water would enter this aquifer laterally, even if not vertically, to create a single
connected surface and groundwater body.
The floodplains fully saturate and flood during the wet season and discharge in directions
not necessarily matching surface water flow paths. Groundwater flows laterally from
floodplain basins and probably also flows into and through regolith-filled fractures in the
upper portion of the basalt bedrock. Numerous surveyed wells were said to have good
perennial supplies despite appearing to be close to watershed boundaries. Further
investigation often reveals up-gradient floodplains that may lie across a surface watershed
divide though the solid bedrock topography promotes groundwater flow in the direction
of the well. Stable isotope results of groundwater from wells in such locations show
evaporation at recharge consistent with infiltration below a floodplain wetland. Published
literature on regolith and saprolite hydrogeology supports this hypothesis where shallow
low-K layers direct infiltration perpendicular to surface contours whereas groundwater
flow predominantly occurs in the deeper higher-K layers immediately above the solid
bedrock and may be in a different direction, as shown in Figure 4-9. The literature
indicates that regolith is often thickest on slopes and this has been confirmed by surveys
of well depth. The greater saturated thickness of aquifer in these areas leads to greater
well yield, as there is more likelihood of intercepting more transmissive layers.
58
Wet season
Inundated floodplain Hand-dug well
Water table Spring Productive borehole
Groundwater flow
Little to no
groundwater River
exchange between
regolith and basalt
Largely impermeable
bedrock with
groundwater only in
fractured and
scoriaceous layers
Dry season
Dry well
Soil Productive borehole
Regolith
Fractured basalt Productive
well
Dry
Basalt river
Fractured basalt
Figure 4-9. Cross sections showing the conceptual model during the wet and dry seasons. The cross
section represents a situation common in Dangesheta, such as a west-east section from the floodplain
containing MW2 and MW3, through MW1 to the Brante River (Figure 4-3). Note the unsaturated zone
flow and very shallow groundwater flow in contradictory directions to the dominant regolith aquifer
groundwater flow during the wet season. Note the dominant regolith aquifer groundwater flow direction
disregarding the surface water flow divide during the dry season.
The situation of the wet season in the conceptual model in Figure 4-9 of seemingly having
groundwater flow in two different directions at the same place is an example of shallow-
flow in the upper soil and regolith layers after storm events but with the
main groundwater flow in the opposite direction, which would be exacerbated by a low-
K layer higher up within the regolith as much of the literature describes. Such a situation
was first described by Toth (1963) where local groundwater flow systems are at odds with
regional flow systems and diagrams based on his paper are found in most hydrogeology
textbooks (Figure 4-10).
59
Figure 4-10. Schematic diagram of a regional groundwater flow system (from van der Heijde (1988)).
4.11 Conclusions
In addition to allowing development of the conceptual model described in the previous
section, the field investigations indicated that the shallow aquifer has suitable properties
for an increase in productive groundwater use. While hydraulic conductivity and
consequently well yield are quite low, systems of multiple or large diameter wells
providing storage of water could be employed. Aquifer thickness and specific yield are
also quite low but not too low to prevent productive shallow groundwater use.
60
Chapter 5. Filling the observational void: Scientific value
and quantitative validation of hydrometeorological
data from a community-based monitoring
programme

Chapter 4 described the field investigations that led to development of a hydrogeological
conceptual model. To enable resource assessment, time series data was also required; the
acquisition and analysis of such data is the basis of Chapter 5. This chapter shows how
community-based hydrometeorological monitoring programmes can provide reliable
high-quality measurements comparable to formal observations, thus answering research
question 2 from Chapter 1. Time series of daily rainfall, river stage and groundwater
levels obtained by a local community in Dangila woreda, have passed accepted quality
control standards and have been statistically validated against formal sources. In a region
of low-density and declining formal hydrometeorological monitoring networks, a
situation shared by much of the developing world, community-based monitoring can fill
the observational void providing improved spatial and temporal characterisation of
rainfall, river flow and groundwater levels. Such time series data are invaluable in water
resource assessment and management, particularly where, as shown here, gridded rainfall
datasets provide gross under or over estimations of rainfall and where groundwater level
data are non-existent. Discussions with the local community during workshops held at
the setup of the monitoring programme and since have demonstrated that the community
have become engaged in the project and have benefited from a greater hydrological
knowledge and sense of ownership of their resources. This increased understanding and
empowerment is at the relevant scale required for effective community-based
participatory management of shallow groundwater and river catchments.
This aspect of the research was published in Journal of Hydrology in 2016:
community-based monitoring programme', Journal of Hydrology, 538, pp. 713-725.
https://doi.org/10.1016/j.jhydrol.2016.04.062
The co-authors provided support in planning and final editing of the paper; all the
analysis, writing the paper and preparing figures was conducted by David Walker.
61
The text and figures comprising the published manuscript are provided here with little
alteration. The study area section has largely been deleted to avoid repetition with
Chapters 3 and 4 while the supplementary material is provided in full as Appendix C.
5.1 Introduction
Continuous time series of rainfall, river flow and groundwater level vary in their
availability. For many areas of, particularly the developing, world, such data is patchy or
non-existent. Unfortunately, the areas of greatest data scarcity typically coincide with
areas that suffer the greatest impacts from adverse hydrological conditions where more
data could be used to better assess the current situation and to forecast future scenarios
allowing for better mitigation and adaptation strategies. The importance of quantitative
information on the rainfall, which controls spatially and temporally variable water
resources, and of measurements of the surface/groundwater resources themselves is not
in doubt (Washington et al., 2006; Conway et al., 2009; Bonsor and MacDonald, 2011).
Satellite and reanalysis rainfall products are often promoted as the solution to low-density
gauge networks, however, the greatest accuracy of such products is achieved in areas with
abundant ground observation data to aid calibration (Fekete et al., 2004; Dinku et al.,
2008; Symeonakis et al., 2009). What’s more, the necessary spatial averaging means
spatial resolution is commonly insufficient for smaller than regional scale hydrological
and hydrogeological studies. Datasets at the relevant scale to inform local resource
management strategies are increasingly being obtained by local communities providing a
low-cost and highly useful source of hydrometeorological time series data where they
would be otherwise unavailable (Liu et al., 2008a; Gomani et al., 2010). The numerous
additional benefits of such community-based monitoring programmes include the
engagement and empowerment of local communities in their own water resources
(Conrad and Hilchey, 2011; Buytaert et al., 2014). A recent editorial in Nature discussing
the rise of “citizen science” in various fields states that data quality is the prime concern
of critics (Nature, 2015). The majority of the literature presenting community-based
monitoring programmes has sought to detail the benefits brought to the community
though few (if any) papers have attempted to quantitatively validate the collected data in
a statistical manner akin to the abundant literature validating remote sensing products
against ground observations. It will be determined here whether community-based
monitoring can provide data that can be satisfactorily validated against formal sources to
provide improved spatial and temporal resolution, and whether it can supply reliable
hydrogeological data where there are no formal alternatives. As formal monitoring
62
networks continue to decline in many parts of the world, we determine if community-
based monitoring programmes can be a viable complement.
5.2 Sub-Saharan Africa context

Rain gauge distribution across sub-Saharan Africa (SSA) is sparse, particularly in
comparison with Europe, North America and South Asia. There are 1152 World
Meteorological Organization (WMO) World Weather Watch stations in Africa at an
average station density of just one per 26,000 km2, 8 times lower than the WMO
minimum recommended level (Washington et al., 2006). Figure 5-1 shows the network
of WMO stations clearly indicating the sparsity of stations in Africa and their uneven
distribution resulting in substantial areas going unmonitored. Within SSA, rain gauge
densities are highest in coastal West and Southern Africa, and the East Africa Highlands
of Kenya and Uganda, whereas areas of greater aridity are underrepresented.
Furthermore, it is widely reported that rain gauge networks in SSA are in decline as
weather services make cut backs (Nicholson, 2001; Washington et al., 2004; Maidment
et al., 2014). Willmott et al. (1994) report a peak in African rain gauge density occurring
in the 1950s and a sharp decline after 1970. South Africa has generally been commended
for its relative abundance of rain gauges although Pegram and Bardossy (2013) report
that even South African rain gauge records are dying off; after mid-2000 they found that
out of the 279 gauges in the 5 regions only 180 survived until 2008. A more extreme
example is Angola which had over 500 meteorological stations as a Portuguese colony
which were all but destroyed during four decades of civil war until a government
rebuilding programme had increased the number to eight by 2007 (Cain, 2015).
Figure 5-1. The global network of World Weather Watch stations colour-coded to show reporting rates
(from WMO, 2003)).
63
River flow monitoring networks in SSA are unfortunately experiencing a similar decline
to meteorological monitoring networks. Monitoring stations globally have been
decreasing in number over the last few decades. Tourian et al. (2013) note that among the
8424 identified gauging stations in the Global Runoff Data Center (GRDC) database only
40% of stations provide discharge data after 2003. Many of these monitoring stations
going offline were located in SSA. The requirement to reverse the trend of decreasing
hydrological monitoring is a widely held view (Kundzewicz, 1997; Owor et al., 2009;
Taylor et al., 2009).
Even so, surface water is densely monitored in comparison with groundwater. There is
general agreement that a better understanding of the shallow hydrogeology of SSA from
the point of view of potential agricultural use is a necessity (Giordano, 2006; Namara et
al., 2011; Evans et al., 2012; Pavelic et al., 2013a). Lapworth et al. (2013) state the issue
succinctly; “Ideally, a thorough quantitative understanding of aquifer properties and
recharge mechanisms under a variety of climate, land use and geological environments is
required to confidently assess current groundwater availability, and forecast future
availability under different scenarios”. A recent review of groundwater conditions in 15
SSA countries (Pavelic et al., 2012) concluded that: “Quantitative information on aquifer
characteristics, groundwater recharge rates, flow regimes, quality controls and use is still
rather patchy”.
Invariably simultaneously reported alongside comments on the need for greater

understanding of SSA hydrogeology is the dearth of observations of groundwater
systems, in particular sustained time series data (Martin and Van De Giesen, 2005; Calow
et al., 2009b; MacDonald et al., 2009; Taylor et al., 2009; Ethiopian ATA, 2013). The
situation with groundwater data is different to the aforementioned decreasing
meteorological and hydrological time series data because there have never been many
monitoring systems in place. For example; considering the hydrogeology atlas of the
SADC region (the Southern African Development Community which includes fifteen
member states south of and inclusive of the Democratic Republic of Congo and
Tanzania), Robins et al. (2006) report that only six of the member states (Lesotho,
Mauritius, Namibia, South Africa, Swaziland and Zimbabwe) have formal monitoring
networks involving water level and some type of water quality measurements. In the
remaining countries, sporadic measurement occurs though in an ad hoc fashion with little
or no data reaching the national groundwater authority. This issue is not restricted to
southern Africa as Martin and Van De Giesen (2005) report that the only data on shallow
64
aquifers in Ghana and Burkina Faso is the total number of wells in a region while even
production figures for small formalised piped groundwater supplies are not recorded.
Dapaah-Siakwan and Gyau-Boakye (2000) who conducted broad-scale hydrogeological
research in this region of West Africa chose to ignore shallow aquifers altogether because:
“Even though many hand-dug wells have been constructed in various hydrogeologic
formations (a total of about 60,000 as of March 1998; Ministry of Works and Housing,
1998), these were not taken into consideration in the analyses for this paper due to the
dearth of data from these sources.” The limited groundwater data available in SSA is
almost exclusively from deep abstraction boreholes, however, shallow groundwater is the
resource that is accessible and exploited by the majority of rural communities via
hand-dug wells.
5.3 Community-based monitoring

It is increasingly advocated that community involvement should be strongly supported by
the scientific community to improve links between science and local level planning policy
(Ridder and Pahl-Wostl, 2005). While there are an increasing number of published works
on stakeholder participation in environmental decision-making, there are few concerning
a participatory approach in quantitative environmental monitoring. The potential benefits
of community-based monitoring are listed by Conrad and Hilchey (2011), compiled from
an extensive literature review across a variety of fields, and include:
 Increasing environmental democracy (sharing of information).

 Scientific literacy (Broader community/public education).
 Social capital (volunteer engagement, agency connection, leadership building,
problem-solving and identification of resources).
 Citizen inclusion in local issues.
 Data provided at no cost to government.
 Ecosystems being monitored that otherwise would not.
 Government desire to be more inclusive is met.
 Support/drive proactive changes to policy and legislation.
 Can provide an early warning/detection system.
Published studies of data collection from non-specialists, often termed “citizen science”,
commonly involve the collection of “snapshots” of, for example; wildlife, soil type, or
plants (Roy et al., 2012; Vianna et al., 2014; Rossiter et al., 2015). Monitoring of bird
populations in programmes such as eBird (Sullivan et al., 2009), where several million
65
species/date/location records are added monthly from around the world and believed to
be the largest citizen science project in existence (Hochachka et al., 2012), are the only
known studies where time series data is collected by non-specialists though the data are
not necessarily gathered at regular times from the same locations. These momentary
observations are less useful for most hydrological applications where complete time
series of transient data are required. The theory and practice of citizen science in
hydrology and water resources management has emerged mainly through experiences in
developed countries in response to growing environmental activism. To date its scope is
limited and there are only a few published examples within the hydrology and water
resources literature of successfully implemented community-based monitoring
programmes:
The APWELL project, instigated in the 1990s, developed participatory monitoring

including 230 rain gauges and 2100 observation wells across 370 villages in the most
drought-prone region of Andhra Pradesh, India. The project provided farmers with the
necessary knowledge, data and skills to understand and manage their groundwater
resource. The outcome was more efficient groundwater use, increased crop yield, and
poverty reduction (Garduño et al., 2009; Garduño and Foster, 2010).
Gomani et al. (2010) detail an “integrated participatory approach” in setting up a

monitoring network in a large (2780 km2) catchment in Tanzania as part of a project with
an overall aim of assessing climate change impacts and land use options. The approach
aimed to assimilate local and expert knowledge with some voluntary monitoring by the
community including weather, river flow and groundwater measurements.
A smaller scale community-based monitoring programme in South Africa with the overall
objective of watershed management for the increase of food production and improving
rural livelihoods is detailed by Kongo et al. (2010). This monitoring network was
extremely equipment intensive and involved monitoring weather, river flow, deep and
shallow groundwater, sediment load, overland flow, soil moisture and crop transpiration.
It is claimed that the participatory aspect led to an appreciation of the research, which
sustained the goodwill of the community to safeguard the instruments and structures
comprising the network. It is stressed that there is always a process to be followed when
engaging stakeholders that needs to be based on trust, honesty and friendship.
Buytaert et al. (2014) present case studies detailing the benefits of community
involvement in hydrological issues from Peru; identifying the hydrological impacts of
66
land use change on ecosystems in remote upland areas beyond the range of formal
monitoring networks, from Ethiopia; engaging farmers to rehabilitate gullies following
soil erosion caused by poorly implemented land management practices, from Nepal;
where communities have taken the lead in water sharing arrangements in an arid region,
and from Kyrgyzstan; where water users associations (WUAs) are being set up who are
installing monitoring schemes to replace those which died out at the end of the Soviet
period.
The few other published case studies of water resource community-based monitoring
programmes generally concern monitoring of water quality for various applications. They
include; water quality monitoring in rural Mexico for public health where no professional
assessments exist (Burgos et al., 2013); for monitoring river sediment load and nutrient
contamination to assess the impact of soil erosion in a remote area of Mindanao, the
Philippines (Deutsch et al., 2005), and; biological measurements (faecal coliform levels
and macroinvertebrate indices) for protection of aquatic habitats in Georgia, USA
(Conners et al., 2001).
This paper presents a case study of a community-based monitoring programme in

Ethiopia and aims to show that community measured hydrometeorological data can pass
strict published quality control procedures. Such data can be validated against formal
sources proving that the data is reliable, of high quality, and can offer improved spatial
and temporal resolution over formal ground observation and gridded datasets. To our
knowledge, there are no other published examples of attempts to rigorously validate data
from community-based monitoring programmes.
5.4 Project context
5.4.1 AMGRAF research project

The AMGRAF (Adaptive Management of shallow GRoundwater for small-scale
irrigation and poverty alleviation in sub-Saharan AFrica) research project commenced in
2013 with the overarching aim of establishing whether development of shallow
groundwater resources for small-scale irrigation (and other purposes) can be used
sustainably to alleviate poverty in SSA. The first field site selected was Dangila woreda
in northwest Ethiopia; an area identified by the Ethiopian ATA (Agricultural
Transformation Agency) for an increase in irrigated agriculture. Further information on
the AMGRAF research project can be found in Appendix C.
67
5.4.2 Study area
The Dangila woreda study area was described in Chapters 3 and 4; the location is shown
again in Figure 5-2 with nearby hydrometeorological monitoring stations.
Figure 5-2. Location map of the study area in the Amhara region of Ethiopia. Map shows formal rain
gauges and river gauges near to Dangila woreda. Lake Tana is visible at the top of the map.
The community-based monitoring programme was initiated in February 2014. The

community were consulted and involved in siting the rain and river gauges and identifying
the wells to be monitored (Figure 5-3 and Figure 5-4). Hydrologically suitable areas were
identified, i.e. narrow channels and valleys for the river gauges where river stage
fluctuations would be most pronounced and open areas with no overhead obstructions for
the rain gauge. Certain locations were excluded for being too open where the community
expressed concern over the security of the equipment. Ultimately, the rain gauge was
situated within the smallholding of the community member who would monitor the
gauge. The monitored wells were chosen to provide a transect from close to the river and
floodplain up towards a watershed boundary that would include successful wells with
perennial supply and also unreliable seasonal wells. Another influence on monitoring well
selection was the route that could be taken by the community member who would
measure well level, which leads in a broad circle from his house to his place of work
(Figure 5-3).
68
Brante river gauge N DAO
MW4
Kilti river gauge
Brante river
MW1-5 and
rain gauge
DSC
Area of
MW5
focus MW3
Kilti river
Brante river
Rain gauge MW1
N
MW2
1 km 250 m
Dangila
Figure 5-3. Locations of monitoring points (close to arrowhead in Figure 5-2). MW = monitoring well,
DSC = Dangesheta Service Cooperative, DAO = Dangesheta Agricultural Office. (Image source:
Google.Earth; Imagery ©2015 DigitalGlobe).
The five monitoring wells are manually dipped every two days with a dip meter and the
rain gauge is measured daily at 9am by reading the level of the internal graduated
cylinder. The river gauges are monitored daily at 6am and 6pm by reading the river stage
from the permanently installed gauge boards. Hard copy records of measurements are
provided by community monitors on a monthly basis to the Dangila woreda office, where
they are transferred to an Excel spreadsheet and forwarded to the research team. Further
information on the monitoring network can be found in Appendix C.
Figure 5-4. Photographs of (left to right) the Kilti River gauge, the rain gauge, and measuring
groundwater level at monitoring well MW5.
5.5 Data analysis methods
5.5.1 Sources of error

Potential errors in rainfall measurements can broadly be divided into two categories;
sampling error and observational error. Sampling error results from spatial and temporal
variability of rainfall. Sampling error increases with increased rainfall and decreases with
increased gauge density and duration of rainfall event (Huff, 1970). Therefore, warmer
regions where convective storms of high-intensity and short duration are common will
see the greatest errors, particularly where rain gauge density is low (the Ethiopian
69
Highlands fit this category). Observational error can be due to inaccurate measurements
on individual days arising through observer errors, either during measurement or
transcription. Detecting such errors is problematic because the skewed distribution of
daily rainfall quantities signifies that in all but the most extreme cases a suspect
measurement has a considerable likelihood of being correct (New et al., 2001).
Measurement biases arise through gauge undercatch caused mostly by wind turbulence
around a gauge though splash and evaporation can also have an effect (Legates and
Willmott, 1990; Peterson et al., 1998; New et al., 2001).
The sources of error presented above are similarly applicable to river stage and
groundwater level measurements. Biases can arise from taking measurements relative to
poorly chosen reference points or due to equipment maintenance issues. Other
observational errors which may be more likely to result from measurements by non-
professionals include family or work commitments necessitating adjustments to
observation time or a temporary change in observer.
5.5.2 Quality control

The quality control procedures of WMO, as presented in their “Guide to climatological
practices”, have been followed in order to verify whether a reported data value is
representative of what was intended to be measured and has not been contaminated by
unrelated factors (WMO, 2011). Checks recommended by WMO comprise:
 Format tests, e.g. impossible dates or words in numeric fields, typically caused by
transcription errors.
 Completeness tests, e.g. missing data, which may or may not be important; a
missing daily rainfall total during a wet period could have a significant effect on
the monthly rainfall total whereas a single missing groundwater level
measurement would not be crucial.
 Consistency tests, further divided into four types of check:
o internal consistency checks, e.g. do maximum measurements exceed
minimum or is wind direction between 0o and 360o (such tests are less
applicable for this community data);
o temporal consistency checks, where the amount of change with prior and
subsequent values is not greater than might be expected for the given time
interval;
70
o spatial consistency checks, comparing observations with what would be
expected based on observations from neighbouring locations;
o summarisation consistency checks, e.g. do annual rainfall totals equal the
sum of monthly and daily totals (this is less applicable for the community
data where only daily measurements are received).
 Tolerance tests, which set upper and lower limits on possible values with recourse
to historical values or via spatial interpolation methods.
Similar to temporal and spatial consistency checks, care must be taken with tolerance
tests to avoid excluding correct and particularly informative extreme values, such as
happened with the Boscastle flood of 2004 in Cornwall, UK and the Great Storm of 1987
in southeast England when seemingly anomalous measurements could have improved
forecasts to provide more warning of what became disastrous weather events (Woodroffe,
1988; Golding et al., 2005).
Considering the community data received in this case study, the initial screening
procedure would reveal any gross errors, which may simply be typographical errors
revealed by format and consistency tests, or extended gaps in the measurements revealed
by completeness tests. Errors were revealed by this visual inspection including; received
spreadsheets often had a mixture of English and Amharic characters which were not
recognised by all computers, commas were often used in place of decimal points or spaces
were present either side of decimal points, and there were occasional errors in the
conversion from the Ethiopian to the Gregorian/Western calendar. Such errors were
simple to rectify.
An additional quality control procedure is the double mass check, which involves plotting
the cumulative data of one station against the cumulative data of another nearby station.
If the data records are consistent, a straight line is obtained. Data from stream flow gauges
can be compared with data for other flow gauges in the same general area, and, similarly,
data for rainfall gauges can be compared. Where an inconsistency is observed, such as a
break in the slope of the line, an investigation into the cause should be performed.
Relocation of weather stations and dam constructions are examples of causes of such
breaks in slope in rainfall and river flow data respectively (O’Donnell, 2012).
5.5.3 Validation of hydrometeorological data

There is much published literature which aims to validate alternative sources of rainfall
measurements against ground observations from formal institutions (Robinson et al.,
71
2000; Nicholson et al., 2003; Wolff et al., 2005; Ebert et al., 2007). The validation
methodologies used are similar and consist of statistical comparisons typically evaluating
correlation coefficient, error and bias. The alternative rainfall sources comprise satellite
and reanalysis products. Specific examples covering Ethiopia include validation of
different gridded rainfall datasets by Dinku et al. (2007) and Dinku et al. (2008).
Published literature concerning validation of river flow and groundwater level data
generally compares modelling simulations to observations (Beven, 1993; Refsgaard and
Knudsen, 1996; Motovilov et al., 1999). No examples have been found in the literature
of validation of data from community-based monitoring.
For this study, the community and formal data were compared using the Pearson
correlation coefficient (PCC) and bias. PCC is the typical standard (including in all the
studies cited in the previous paragraph) used to validate data from an alternative source
against a formal source: a negative or low value indicating poor performance and
questionable validity. However, because PCC simply measures the strength of the linear
relationship between the datasets, a high PCC would result from a match in the structure
of the data even if absolute values varied significantly. Therefore, bias is also computed
to determine whether variation is systematic and could therefore be reduced with bias
correction, or is due to random error. High seasonal variation between absolute
measurements mean bias is a more useful descriptive indicator than other methods of
calculating error such as mean error and RMS error. Gridded datasets have been evaluated
using the same methodology in order to compare their performance with that of the
community data.
𝑵 ∑ 𝑪. 𝑭 − (∑ 𝑪). (∑ 𝑭)
𝑷𝑪𝑪 =
(5-1)
√(𝑵 ∑ 𝑪𝟐 − (∑ 𝑪)𝟐 ). (𝑵 ∑ 𝑭𝟐 − (∑ 𝑭)𝟐 )
∑𝑪 (5-2)
𝑩𝒊𝒂𝒔 =
∑𝑭
Where 𝐶 = community monitored data or gridded data set, 𝐹 = formal ground
observation data, and 𝑁 = number of data pairs.
The seasonality of the climate in this region means high correlations would be expected
during the long dry season when little to no rainfall occurs and surface/groundwater levels
are relatively static. Therefore, statistical comparisons were separately conducted for the
wet season onset (May-June), wet season peak (July-August), wet season retreat
72
(September-October), and the dry season (November-April), as well as for the full time
series.
5.5.4 Behavioural differences in data

It is important to note that the formal and the community monitoring locations are not
immediately adjacent and, as such, near-perfect correlations and zero bias are not
expected. Variations in groundwater and river levels and in rainfall due to geographic
position provide insights into local hydrogeology, hydrology and meteorology and the
lower PCC derived from such variations does not call for rejection of data as long as the
quality control procedures are passed. What’s more, seemingly extreme values should not
always cause the rejection of data during the quality control process but should be
investigated properly. Local knowledge gained through field visits combined with
anecdotal evidence from contacts in the area means extreme observations highlighted for
rejection during tolerance tests may be correctly incorporated and are highly valued.
5.6 Rainfall
5.6.1 Formal ground observations

Rainfall data in Ethiopia is collected by the NMA. The density of rain gauges is low, as
can be seen in Figure 5-2, with only one rain gauge within 900 km2 Dangila woreda and
only an additional eight within a surrounding area of 5000 km2. All the rain gauges
outside of the woreda, particularly those to the south, lie at significantly different altitudes
to Dangila woreda. In addition, the rain gauges to the northeast of the woreda lie along a
straight line; the Dangila to Bahir Dar highway, which leads to unconfident extrapolation
of rainfall data either side of the highway via methods such as Kriging or Thiessen
polygons. Rainfall data has been collected for the nine rain gauges shown in Figure 5-2
with the available datasets varying in length. Dangila is the closest NMA rain gauge to
Dangesheta at 5.7 km distant to the south and at approximately the same altitude (~11 m
difference). The Dangila rainfall record is the third longest (since 1987) but more
importantly is the most complete while all other rain gauges have significant data gaps,
often for a year or more. For these reasons of proximity and completeness, the Dangila
rainfall record is used to evaluate the performance of the alternative rainfall sources (see
Appendix C for further information substantiating the use of Dangila data for validation
purposes).
73
5.6.2 Community data
At the time of writing, 18-months of data were available; March 2014 to October 2015,
which span two wet seasons. The wet season is pronounced with approximately 85% of
rainfall recorded between May and October. However, the wet season of 2014 was
atypical in that it started earlier, ended later, and had a less pronounced peak in July and
August compared to historical records from the NMA for all nearby rain gauges. A double
mass check conducted for rainfall from the community-based monitoring programme
against Dangila NMA confirms a reliable record (Figure 5-5); based on double mass
checks it appears more reliable than records from most of the alternative formal rain
gauges.
Dangesheta community cumulative rainfall (mm)
4000
3000
2000
1000
0
0 500 1000 1500 2000 2500 3000 3500 4000
Dangila NMA cumulative rainfall (mm)
Figure 5-5. Double mass check of rainfall for Dangila NMA with Dangesheta community showing a good
linear relationship indicating a consistent record. Note that a good record is considered to be a straight
line and not necessarily x=y.
Closer to the community rain gauge than the NMA formal rain gauge is an electronic
automatic weather station, which is 960 m to the north beside the Dangesheta Agricultural
Office (DAO on the map in Figure 5-3) and at approximately the same altitude (~14 m
difference). Installed by Bahir Dar University in March 2015, the electronic weather
station incorporates a tipping bucket rain gauge, though unfortunately it stopped
recording during the peak of the wet season leading to limited data with which to conduct
comparisons.
5.6.3 Gridded datasets

The gridded remote sensing and reanalysis rainfall datasets that have been considered are
TRMM, ERA-Interim, NASA MERRA, JRA-55 and NCEP (see Appendix C for further
information). The spatial resolution of these gridded datasets varies from 0.25o x 0.25o,
or ~28 x 28 km, (TRMM) to 1.25o x 1.25o, or ~140 x 140 km, (JRA-55) though this
74
coarsest dataset provides the longest time series; since 1958. Such large grid squares over
this region of Ethiopia necessarily comprise large altitudinal ranges, often of several
thousand metres, and where multiple NMA rain gauges are present within a grid square
the observed variations in rainfall totals can be very high.
5.6.4 Performance of alternative rainfall sources

Spatial consistency testing conducted as part of the quality control procedure involved
plotting daily rainfall totals from Dangila NMA, Dangesheta community, and Dangesheta
electronic rain gauges. The plots were very similar but with a slight shift in the peaks. It
was immediately apparent that there had been an error in conversion from the Ethiopian
to the Gregorian/Western calendar and when the community rainfall time series was
shifted by a day the peaks matched. Further investigation of rainfall data from the
electronic rain gauge revealed that daily totals were summed from a 24-hour period
spanning midnight to midnight. When the totals were recalculated for a 9 am to 9 am
period, as per the formal and community measurements, the timing of peaks from all three
datasets were in agreement. Values for the tolerance tests could be taken from the
extensive formal rainfall datasets from the nine nearby rain gauges, which were also used
for consistency testing. All community rainfall data passed quality control testing.
Before correlating daily rainfall from the community gauge with the formal source, it was
necessary to determine what PCC could be considered good performance. By correlating
rainfall from the other nearby NMA rain gauges with that from Dangila, variations in
PCC would show the degree of spatial and temporal variation in rainfall. The PCC was
calculated using as long a time series as available for each rain gauge; the results are
presented in Figure 5-6a. As would be expected, the PCC increases as distance from the
Dangila rain gauge decreases because error due to spatial and temporal variation lessens.
The trendline is projected to the distance of the community rain gauge (5.7 km) and it can
thus be stated that a PCC below this line (less than approximately 0.68) likely includes a
degree of observational error.
Because the community data is available only from March 2014, the same period was
used to evaluate the relative performance of the gridded datasets. JRA-55 and NCEP are
excluded because the data were not available for this evaluation period. Where multiple
seasons have occurred, i.e. wet season peak in 2014 and 2015, the mean PCC is taken;
nowhere was it necessary to take means of markedly different values (Figure 5-6b).
75
0.80 0.90
a. 0.70 b.0.80 Community data TRMM ERA-Interim NASA-MERRA
0.60 0.70
0.68
0.60
0.50
0.50
PCC
0.40
0.40
PCC
0.30 0.30
0.20 0.20
0.10
0.10
0.00
0.00 Dry Onset Peak Retreat All
0 10 20 30 40 50 60 70 -0.10
Distance from Dangila (km) -0.20
Figure 5-6. Variation with distance of Pearson correlation coefficient (PCC) between daily rainfall from
Dangila NMA rain gauge and other NMA rain gauges close to Dangila woreda (a). Pearson correlation
coefficient (PCC) between daily rainfall from Dangila NMA rain gauge and alternative sources (b).
Immediately apparent from Figure 5-6b is that the community data outperforms the
gridded datasets for all seasons. Localised short-lived storm events leading to high spatial
and temporal variability are proposed for the reason behind the poor correlation of all
alternative sources of rainfall data during the wet season peak. When all the data is
considered (the far right of the graph), the PCC of 0.73 for the community data is greater
than the value predicted in Figure 5-6a and the discrepancies with the formal dataset can
therefore be considered sampling rather than observational error. Because they are just
900 m apart, it would be expected that the community rain gauge and the electronic rain
gauge would correlate better than the dataset pairs presented in Figure 5-6b; indeed the
calculated PCC is 0.84.
Analysis of bias is presented in Figure 5-7a. Again, the community data shows the least
bias and, importantly, the greatest consistency, suggesting that the bias is due to
systematic error. This error could be due to undercatch as the community rain gauge is
close to a small tree, which may provide some sheltering. However, when compared to
the nearby electronic rain gauge which is in an open position like the Dangila NMA rain
gauge, bias is just 1.05, suggesting that the bias of ground observations in Figure 5-7a is
primarily due to spatial variability.
76
4.5 35%
a. b. TRMM 1998-
4.0 Community data 2015
30%
ERA-Interim
3.5 TRMM 1979-2015
ERA-Interim 25% NASA-MERRA
% of annual total
3.0
NASA-MERRA 1979-2015
2.5 20% NCEP 1979-2009
bias
2.0 15% JRA-55 1979-
2013
1.5 Dangila Ground
10%
1.0 Obs. 1987-2015
5%
0.5
0.0 0%
Dry Onset Peak Retreat All J F M A M J J A S O N D
Figure 5-7. Bias between daily rainfall from Dangila NMA rain gauge and alternative sources (a). Note
that bias is computed as a ratio and the bold line at 1.0 represents zero bias. Median monthly rainfall
totals as percentage of annual total for Dangila NMA rain gauge and gridded datasets (b).
Figure 5-6 and Figure 5-7a suggest that the gridded datasets perform poorly for this
location, particularly in comparison to the community-based monitoring. To test whether
the gridded datasets always perform poorly or solely for the period of overlap with the
community data, the full available time series were analysed and monthly totals are
considered in order to smooth out extreme events that reduce the PCC during daily rainfall
analysis. When median monthly totals are normalised to annual total (Figure 5-7b) the
performance of the gridded datasets is improved. However, capturing the wet season peak
still appears to be problematic which could have serious consequences for water resource
assessment if these datasets were to be relied on in place of ground observations.
5.7 River flow
5.7.1 Formal observations

River flow data in Ethiopia is collected by the Ministry of Water, Irrigation and Electricity
(MoWIE). It can be seen in Figure 5-2 that two river gauges lie within Dangila woreda
though the most useful for this project are named “Amen @ Dangila”, which is upstream
of the community Kilti gauge, and “Kilti Nr Durbete”, which is downstream of the
community Kilti and Brante gauges and situated outside the woreda. Measurement of
river stage at these locations is taken from depth gauge boards and the available time
series spans 1988 to 2014 though with some significant gaps in the data lasting from
months to years.

The two MoWIE monitored river gauges within Dangila woreda are located on ephemeral
streams and it appears that either measurement does not always take place or monitoring
records have not yet been completely digitised. A continuous time series of river stage
measurements is therefore only available from the community-based monitoring
77
programme. Following a decision taken by the community themselves, measurements
take place twice a day as opposed to the daily formal river monitoring. In addition, with
no external prompting, the community members who conduct the monitoring regularly
add notes to their river stage records noting if a flood peak passed at a particular time and
at what level. Such information is not available from formal sources.
The full time series of rainfall and river stage measurements collected by the Dangesheta
community are presented in Figure 5-8. It can be seen that the rivers are very flashy with
sharp peaks in river stage quickly following rainfall events.
3.5 100
Daily rainfall Kilti Brante 90
3
80
2.5 70
River stage (m)
Rainfall (mm)
2 60
50
1.5 40
1 30
20
0.5
10
0 0
Figure 5-8. Complete time series to date of daily Kilti and Brante river stage, and rainfall measurements
from the community-based monitoring programme (2014-2015).
5.7.3 Performance of alternative river flow sources

A complete twice-daily record of river stage is held which is straightforward to cross-
examine between rivers and with rainfall to determine if all peaks and troughs pass
consistency tests. Suitable values for tolerance testing were derived from anecdotal and
physical evidence obtained during field visits to the monitoring sites; such as the Kilti
River’s maximum peak in October 2014, which damaged the river gauge. Quality control
procedures were passed for all of the community monitored river data.
Unfortunately, there is only a very short period of overlap between the formal and the
community river flow data. Therefore, correlations with formal sources are not
considered the principal method of validating the river flow data. However, correlating
the overlapping data between formal (Kilti and Amen) and community (Kilti and Brante)
daily totals gives 0.52-0.58, similar to the correlation between the two formal river flow
sources for their complete daily records, PCC = 0.58.
78
A unit runoff check involves dividing the (monthly) runoff by the catchment area in order
to determine the runoff as a depth. This is compared for consistency with values obtained
from nearby hydrologically similar catchments. This check is particularly useful in
identifying abrupt changes in river flows resulting from river basin management activities
(O’Donnell, 2012). Unit runoff checks were conducted for the Kilti and Brante flow
measurements from the community-based monitoring programme and for the formal flow
measurements for the Kilti and Amen (Figure 5-9). The differences in unit runoff depths
from the formal sources are increasingly significant from 1997 to 2001 and 2007 to 2010.
This may be due to a period of unreliability of the rating curve and ongoing revision
efforts which was the explanation given by MoWIE for considering the 2014 data to be
unreliable (S. Mamo, personal communication, 10 December 2015). Thus, no conclusions
should be drawn from the poor match with the community data during the 6-month
overlap in 2014 (Figure 5-9). It can be seen on the unit runoff check that there is a
reasonable match between the two community monitored rivers, at least as good a match
as has typically been seen between the two formally monitored rivers in previous years.
400 200
350 NHS Kilti NHS Amen
Monthly runoff (mm)
Community-monitored Kilti Community-monitored Brante 160

300
250 120
200
150 80
100
40
50
0 0
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2014
2015
Figure 5-9. Unit runoff checks for river flow data from community-based monitoring and formal sources.
Gaps indicate insufficient flow measurements to calculate monthly totals. Note that the 2014 formal
measurements are considered by MoWIE to be unreliable.
When river flows (daily totals) are correlated against rainfall from the NMA Dangila rain
gauge, the PCCs are lower for all seasons and for all gauges (Figure 5-10) than was
achieved when validating rainfall and groundwater data. The low values reflect the
geography and hydrogeology of the catchments where peak floods have been observed to
occur with a very short time lag after the onset of a rainfall event. Very heavy overnight
storms were experienced during fieldwork though when the rivers were visited early the
following morning the river stage had already dropped from the level still visible on the
banks to the level observed the previous day. Because rainfall measurements are
cumulative and river stage measurements are momentary, monitoring would have to be
undertaken at a much higher frequency in order to achieve better correlations with
79
rainfall. However, the PCCs in Figure 5-10 are similar for both the formal and the
community measurements particularly when all seasons are considered.
0.70
NHS Kilti
0.60 NHS Amen
Community monitored Kilti
0.50
Community monitored Brante
0.40
PCC
0.30
0.20
0.10
0.00
Dry Onset Peak Retreat All
-0.10
Figure 5-10. Pearson correlation coefficient (PCC) between daily rainfall from Dangila NMA rain gauge
and river flow measurements (daily totals) from formal and community sources. Note that “All” includes
incomplete months excluded from other seasons explaining the contrast in relative PCC of, in particular
community monitored Kilti, data from individual seasons to “All”.
5.8 Groundwater level data
5.8.1 Formal observations

The community monitored groundwater level data is the only means of assessing water
table depth and recession anywhere within Dangila woreda. Extremely limited data on
boreholes and groundwater are available from formal sources (see Appendix C).

It would be expected, given that the monitoring wells are in close proximity (maximum
separation of 970 m), that groundwater levels from different wells would follow a similar
pattern of seasonal variation. Peaks in water level during dry spells or plateaus spanning
numerous rainfall events would suggest unreliable data. It is immediately obvious from
Figure 5-11 that the patterns in water level response are consistent, quality control
procedures have been passed (with a single exception discussed below) and the validity
of the data is confirmed when statistical comparisons are conducted between wells and
with river stage.
Minor abstraction occurs from the monitoring wells at the level of a few buckets (20-50
litres) per day for domestic use. If abstraction took place immediately prior to
measurement then a drawdown of a few centimetres may be incorporated into the
observation. Such discrepancies between aquifer and well groundwater level are likely
within the expected measurement error. Furthermore, measurement took place as much
as possible early in the morning prior to well abstraction.
80
Daily rainfall MW1 MW2 MW3 MW4 MW5
0 100
1 90
2 80
Well water level (mbgl)

3 70
Rainfall (mm)
4 60
5 50
6 40
7 30
8 20
9 10
10 0
Figure 5-11. Complete time series to date of groundwater level and rainfall measurements from the
community-based monitoring programme (2014-2015). It is noted that wells MW1, MW3 and MW5 dry
out in the dry season and their minimum groundwater levels represent the basal level of the well. The gap
in the well MW4 time series is discussed in Section 5.8.3.
Differences in amplification of water level responses to particular rainfall events have

physical reasons: either due to features of the well itself, e.g. MW1 and MW5 peak the
most significantly as they are open to direct precipitation and overland flow, or due to
aquifer characteristics, e.g. MW4 declines the most gradually, proposed to be due to a
lack of high-transmissivity layers, such as fractured bedrock within the shallow weathered
regolith aquifer, which are present in other well bores (during workshop discussions the
local community spoke of not striking rock when excavating MW4 unlike in other wells,
particularly MW1 and MW5 where a rapid decline in water level is observed at the onset
of the dry season). Analysis of the differences in well responses and discussions during
community workshops have been invaluable in gaining a greater understanding of the
shallow hydrogeology in the area.
5.8.3 Performance of alternative groundwater level sources

The quality control procedures had to be most carefully applied to the groundwater level
data. Completeness tests showed occasional gaps of two days rather than the expected
measurements every other day with some gaps of three days and one exceptional gap of
eight days. It is noted that these larger gaps occur during the dry season when there is
little groundwater level fluctuation and there are just as many measurements at a higher
than required frequency on consecutive days. No groundwater level dataset was excluded
for reasons of completeness. Consistency tests often highlighted errors where large
“steps” in the data were present from one month of measurements to the next. Further
investigations typically revealed that a spreadsheet had been labelled incorrectly and
81
when the data was switched to the correct well the consistency test was passed. One such
step in the data which failed according to spatial consistency (neighbouring wells do not
show such a large drop at that time) and temporal consistency (such a large overnight
drop has no physical explanation) has yet to be resolved and the excluded month can be
seen in well MW4 in Figure 5-11. Other than this single month of data for one particular
well and following some corrective reorganisation of datasets, the groundwater level data
passes quality control procedures.
The groundwater level data cannot be validated against formal sources as no such data
exists. Figure 5-12 shows the Pearson correlation coefficient between water level
responses of different monitoring wells. Bias is inapplicable because the response of each
well is expected to vary in absolute value; such variations are due to differing well and
water table depths, variations in aquifer properties and differences in position on the
groundwater flow path. Accordingly, precise agreement, i.e. correlations of 1, would not
be expected. Indeed, it is the subtle differences in groundwater level response that are
aiding understanding of the shallow hydrogeology of the area. Analysis was conducted
for individual seasons and for the full time series.
Dry Onset Peak Retreat All

1.00
0.90
0.80
0.70
PCC
0.60
0.50 MW1 v MW2 MW1 v MW3 MW1 v MW4

MW1 v MW5 MW2 v MW3 MW2 v MW4
0.40 MW2 v MW5 MW3 v MW4 MW3 v MW5
MW4 v MW5
0.30
Figure 5-12. Pearson correlation coefficient (PCC) between community monitored groundwater level data
from monitoring wells MW1-5.
Figure 5-12 shows that there is very good correlation between monitoring wells; the mean
PCC between all wells for the full time series is 0.86. The raw data was investigated
where the PCC is below 0.75 and in all cases a physical reason was apparent such as
comparisons between wells during a period where one was predominantly dry (e.g. MW1
for long periods).
To further validate the groundwater level data, correlations were conducted with river
stage from the two nearby community-monitored gauges. River depth is being compared
to depth to groundwater thus when river stage is high it would be expected that depth to
82
the water table would be low and a perfect correlation would yield -1. However, the flashy
response of the rivers to rainfall events and the lag until groundwater responds means it
is unlikely that very close to -1 would be obtained but the results should still be high in
the negative. The results of the correlations are presented in Table 5-1 and show highly
satisfactory correlations with an average of -0.73.
Table 5-1. Pearson correlation coefficient (PCC) between community monitored river stage and
groundwater level from monitoring wells MW1-5 for the entire time series.
Brante river stage vs; Kilti river stage vs;

Well: MW1 MW2 MW3 MW4 MW5 MW1 MW2 MW3 MW4 MW5
PCC: -0.75 -0.80 -0.76 -0.63 -0.83 -0.70 -0.76 -0.73 -0.64 -0.74
5.9 Discussion
5.9.1 Qualitative and quantitative value of community-based monitoring

observations
The qualitative value of the community data is in contributing to the conceptual
understanding of the shallow groundwater system. Conceptual understanding has only
been possible with a combination of fieldwork and analysis of differences in well and
river responses using data from the community-based monitoring programme. Slow
declines in groundwater levels following rainfall events can indicate high storativity of
the aquifer and significant river baseflow throughout the dry season can indicate an
aquifer with the potential for exploitation.
The community data has quantitative value through providing complete time series
spanning numerous seasons. For the purpose of understanding the shallow
hydrogeological system to enable simulation of the impacts of increased abstraction, land
use change, and climate variability; physically-based numerical models are being
constructed using SHETRAN (Ewen et al., 2000). Construction and calibration of these
necessarily transient models at scales useful for local management of water resources is
only possible with the time series of river flow and groundwater level gathered by the
community. Alongside traditional methods such as chloride mass balance (CMB),
recharge assessments for the Dangesheta area are being conducted using the RIB model
(Xu and van Tonder, 2001; Sun et al., 2013) and water table fluctuation method; neither
of which would be possible without the time series of groundwater level. The close
agreement of the community-gathered and nearest formal rainfall dataset gives
83
confidence that the formal rainfall dataset can be used in the models to extend the time
series prior to the commencement of the community-based monitoring programme.
Consistency of anomaly patterns as evidenced by the PCC between community gathered
and gridded rainfall datasets enables selection of the most appropriate gridded dataset for
infilling gaps in formal ground observation rainfall totals, which occurred historically.
A key value of community-based monitoring programmes is the engagement of the local

community, which, as the wider research programme progresses, will hopefully evolve
to active management of their resources. The value to the local community has been
expressed as a feeling of partnership in the project rather than constantly being subjects
of research. Questions posed by the Dangesheta community during recent workshops
involving the dissemination of findings demonstrated a level of engagement and an
increase in hydrological knowledge that was not observed during workshops at the project
onset. Proffered reasons for differences between recession curves for groundwater levels
from various wells, e.g. zones of aquifer with greater storage properties, have been
incorporated into conceptual models. The community also speak of a sense of pride that
their community are participating in the programme that may have implications beyond
Dangesheta.
This research has shown that high quality hydrometeorological data for various
applications can be collected by non-specialists from local communities. The data can
reliably supplement that from formal sources or provide time series where no formal
alternatives are available.
5.9.2 Recommendations for ensuring quality data production

The potential for community-based monitoring programmes to infill gaps in sparse,
declining or non-existent formal monitoring networks is clear. However, there are
numerous critical factors for ensuring quality data production. The early involvement of
the local community is important to instil a sense of ownership of the equipment and the
project. Assistance in site selection for monitoring points is an ideal way to engage the
community early and was achieved in this case via the focus groups and participatory
mapping workshops. Variations in well level responses indicate the monitoring wells
were successfully identified to provide information on aquifer zones with varying
potential for exploitation. Selection of the community members to be involved in the
programme is particularly crucial. The completeness of these community datasets and
their success in passing the WMO quality control standards indicates selection of
monitoring personnel was successful in this case. Known and respected community
84
members who live or work in close proximity to the monitoring points should be selected,
if willing to participate, to ensure security of the equipment and to demonstrate to the
community, simply by their involvement, that the programme has value. We are aware of
community-based monitoring programmes in other areas of Ethiopia that have suffered
issues such as vandalism of equipment (Zemadim et al., 2014) and falsification of data
(D.L. Yiak, personal communication, 5 April 2015). In these cases, monitoring or in-situ
downloading of data has been conducted by outside (unknown to the community) people
or a casually selected community member who may have been purely interested in the
financial incentive. Notably, these examples were more equipment intensive and offered
higher financial incentives than the Dangesheta case. Vandalism or data falsification have
not been encountered during this study further confirming the value of community
participation in site selection and nomination of community members to undertake the
monitoring. The importance of feedback has been reported to us concerning Dangesheta
and other examples of community-based monitoring programmes in Ethiopia: this could
be delivered through visits and support as well as workshops presenting the collected
data, eliciting from the community what the data reveals, explaining what the data is being
used for, and giving the community the opportunity to ask questions, provide their own
explanations for patterns in the data, and give suggestions for improving the community-
based monitoring programme. The continued performance of the community-based
monitoring programme in generating high-quality observations is evidence of the value
of the workshops.
5.9.3 Wider application of community-based hydrometeorological monitoring

It has been shown that community-based monitoring can be used to provide improved
spatial density of measurements in areas of sparse and/or declining formal monitoring
networks. In addition, where there exist relatively high densities of formal
hydrometeorological monitoring points, community-based monitoring still has much to
offer.
Gridded datasets are a viable alternative source of rainfall data in many regions though it
has been shown here that over complex terrain with large differences in altitude gross
over and under estimations of rainfall totals are possible, especially where grid size is
large. Community-based monitoring can provide data of sufficient quality to add to the
ground observation datasets used to calibrate and validate these gridded datasets.
85
5.10 Conclusions
The research shows that high-quality daily rainfall totals, sub-daily river stage and daily
to sub-weekly groundwater level measurements are achievable by an astutely
implemented and managed community-based monitoring programme. Formal rain gauge
networks in many regions of the world are inadequately dense to provide confident
interpolation of rainfall quantities. Gridded datasets with their necessarily low resolutions
often cannot achieve good agreement with ground observations particularly in areas of
spatial heterogeneity of intense convective precipitation and particularly when sub-
monthly rainfall totals are required. Formal river monitoring networks are also
insufficient with few available datasets for use in modelling catchments at less than the
regional scale. Furthermore, formal river monitoring networks, along with formal rain
gauge networks, are in decline as national institutions embark on cost-cutting practices;
an issue which is particularly severe in less economically developed countries. In sub-
Saharan Africa, groundwater level monitoring networks are essentially non-existent when
it comes to shallow groundwater – the resource used by the majority of poor rural
communities. It has been shown that community-based monitoring can provide high
quality data to help fill these observational voids. Data screening for quality control
indicates reliable and consistent measurements, as good as formal monitoring, can be
obtained by local communities. Community-based monitoring can improve spatial and
temporal characterisation of rainfall, river flow and groundwater level, reducing the
uncertainty of using extrapolated/interpolated values from formal and gridded datasets or
from modelling simulations. Statistical comparisons of the community-based monitoring
data against formal sources and against other data simultaneously gathered by the local
community validate their quality for use in further study. Our research has shown that
benefits to the community include a greater understanding of their local hydrology and
hydrogeology, a sense of ownership of their water resources, and a sense of being a
research partner as opposed to a subject. Such increased hydrological understanding in
sub-Saharan Africa provides the basis for communities to manage their own resources,
which could increase food security by reducing reliance on rainfed agriculture.
It is noted here that the community-gathered data from Dangesheta beyond the period
covered in this chapter (collected after publication of the associated paper) passed through
the same quality control procedures. This longer time series of consistently high-quality
data was utilised in subsequent chapters.
86
Chapter 6. Insights from a multi-method recharge estimation
comparison study

Chapter 5 showed that high quality hydrometeorological time series can be obtained from
community-based monitoring programmes, in addition to informing the conceptual
model. The community gathered data was therefore approved for use in resource
evaluations. Chapter 6 concerns the recharge assessment conducted for Dangila woreda
that aimed to answer research question number 3 from Chapter 1: Can shallow
groundwater be considered a renewable resource, and; which recharge assessment
methods provide the highest confidence in the calculated recharge amounts when applied
to these types of aquifers?
Although most recharge estimation studies apply multiple methods to identify the
possible range in recharge values, many do not distinguish clearly enough between
inherent uncertainty of the methods and other factors affecting the results. We
investigated the additional value that can be gained from multi-method recharge studies
through insights into hydrogeological understanding, in addition to characterising
uncertainty. Nine separate groundwater recharge estimation methods, with a total of 17
variations, were applied. These gave a wide range of recharge values from 45 to
814 mm/a. Critical assessment indicated that the results depended on what the recharge
represents (actual, potential, minimum recharge or change in aquifer storage), and spatial
and temporal scales, as well as uncertainties from application of each method. Important
insights into the hydrogeological system were gained from this detailed analysis, which
also confirmed that the range of values for actual recharge was reduced to around 280-
430 mm/a. This chapter demonstrates that even when assumptions behind methods are
violated, as they often are to some degree especially when data are limited, valuable
insights into the hydrogeological system can be gained from application of multiple
methods.
This aspect of the research was published in Groundwater journal in 2018:
Groundwater (in press).
87
The co-authors provided support in planning and final editing of the paper, or were
involved in data collection; all the analysis, writing the paper and preparing figures was
conducted by David Walker.
The paper is provided here with little alteration; the study area section has been largely
reduced to avoid repetition with Chapters 3 and 4, though some climatic and geological
information is repeated where particularly important for the recharge study. The
supporting information is provided in full as Appendix D. It is noted that one of the
recharge estimation methods applied utilised SHETRAN modelling, which is not
described in detail until Chapter 7. Brief information on the SHETRAN modelling is
provided in this chapter and in Appendix D.
6.1 Introduction
Estimates of groundwater recharge allow quantification of renewable groundwater
resources and can be used to indicate aquifer vulnerability to contamination or drought,
assess groundwater contribution to streams (baseflow) and wetlands, and identify the
implications of changes to land use, land cover or climate (Misstear, 2000; de Vries and
Simmers, 2002; Healy, 2010). Several notable reviews published over the past decades
discuss various methodologies of estimating groundwater recharge (Simmers, 1988;
Lerner et al., 1990; Scanlon et al., 2002; Healy, 2010). It is well known, and stated by
these reviews, that groundwater recharge estimates often vary between methods due to
the uncertainties inherent with each method, the different spatiotemporal scales at which
they operate, and the type of recharge they represent. It is normally recommended,
therefore, that multiple methods be used. However, recharge estimation methods are often
chosen in practice according to data availability even though the method may not be the
most suitable for the particular climate or hydrogeological conceptual model. Often, the
violation of a method’s assumptions may only become apparent when the recharge result
is compared to results from different methods. In addition, some recharge estimation
studies do not make a clear distinction between the reasons why the recharge results
differ, whether it is due to genuine uncertainties in data and methods, unsatisfied
assumptions, different spatiotemporal scales, or if the method is actually computing a
different type of recharge. However, recognising these distinctions in multi-method
recharge estimation comparison studies can help to provide useful insights into the
hydrogeological system.
A recharge assessment was conducted at a study site in northwest Ethiopia (Dangila

woreda, a local administrative district), in the context of an investigation into the
88
resilience of shallow groundwater resources used for irrigation by rural communities.
Following recommended approaches, e.g. Scanlon et al. (2002) and Healy (2010), several
techniques were initially applied, and it was found that they gave a wide range of recharge
estimates. This is commonly reported in the literature, e.g. Berehanu et al. (2017) and
Afrifa et al. (2017), although it is less common for studies to report investigation of the
reasons for the range of values. Some previous studies, typically using at most three to
five recharge estimation techniques, have considered the basis for differing recharge
estimates in more detail, and concluded that the range of recharge estimates contains
useful information to inform further understanding of the conceptual model (e.g. Coes et
al. (2007); King et al. (2017)). For our study, there were sufficient data of suitable quality
to apply a larger number of recharge estimation methodologies at a single site, so a wider
investigation was made to assess which of the most commonly applied recharge
estimation methods could help to provide insights and increase understanding of the
hydrogeological system. Nine different recharge estimation techniques were applied, with
a total of 17 variations, including variants of methods and variations in how input data
were derived. The methods are presented here in order of increasing data requirement and
complexity: an empirical method, streamflow hydrograph methods (three variations), soil
moisture balance (two variations), basin water balance (three variations), chloride mass
balance, water table fluctuation (two variations), rainfall infiltration breakthrough, and
physically-based modelling. The ninth method is large-scale mapping and modelling
(three variations) from which recharge values have been obtained for comparison from
published studies.
The three aims of this paper are to:
1. Demonstrate quantitatively the range of recharge results that can be

calculated from as many methods as feasible for the study site, and analyse
the underlying reasons for the different recharge values
2. Assess the utility of applying multiple methods in order to gain insights on
the hydrogeological system
3. Provide a recharge estimate with uncertainty for Dangila woreda.
The study highlights and analyses the general problem of interpretation of variability in
recharge estimates obtained from different methods. It is noted that all methods were
applied even if assumptions may not be fully complied with, since this is a factor relevant
to uncertainty in recharge estimation in many published studies. It is not uncommon for
recharge results to be reported without explicit statement of assumptions and limitations
89
or the type of recharge being computed (Sophocleous, 1985; Wood, 1999; Halford and
Mayer, 2000). It may only be through identifying significant discrepancies between
recharge results from different methods that violation of a method’s assumptions are
realised and the hydrogeological conceptual model can be amended and better
understood. Additionally, this study provides a useful recharge estimate for a shallow
aquifer in northwest Ethiopia. Published recharge estimation studies from sub-Saharan
Africa are not great in number, not well geographically distributed, and many are grey
literature (Bonsor and MacDonald, 2010; Wang et al., 2010; Pavelic et al., 2012; Chung
et al., 2016). The majority of studies are concentrated in arid and semi-arid regions due
to water scarcity in these areas. However, many regions of apparent high rainfall also
experience water scarcity during the dry season (Rijsberman, 2006) and when sub-
Saharan Africa’s variable climate unpredictably delivers low wet season rains (Van
Koppen, 2003; Bonsor et al., 2010).
6.2 Groundwater recharge

Lerner et al. (1990) provide the classical definition of recharge: “the downward flow of
water reaching the water table, forming an addition to the groundwater reservoir”. This
defines “actual recharge” and is referred to as such by many authors, e.g. Scanlon et al.
(2002); Misstear et al. (2007); Healy (2010). According to Rushton (1997), the term
“actual recharge” is used to distinguish it from potential or minimum recharge. Potential
recharge is water passing downward through the unsaturated zone that could potentially
contribute to the aquifer. Potential recharge is the term used by many authors for recharge
computed from unsaturated zone methods as this infiltrated water may be subject to losses
(e.g. root zone uptake, interflow then surface discharge) before contributing to the aquifer
(e.g. Simmers (1988); Rushton (1997); Healy (2010)). Minimum recharge refers to
groundwater discharge to rivers or springs, when the two quantities are considered to be
in balance. It is termed minimum recharge because other losses (e.g. evaporation from the
saturated zone, seepage to deeper aquifers) may have occurred since the water was
recharged (e.g. Szilagyi et al. (2003); Vegter and Pitman (2003); Risser et al. (2005)).
In humid regions characterised by shallow water tables and gaining rivers, diffuse (or
direct) recharge dominates. In arid regions characterised by deep water tables and losing
rivers, recharge is usually focussed (or indirect) along river corridors with rates generally
limited by water availability at the surface (Allison, 1988; Scanlon et al., 2002). The
factors that influence the amount and type of recharge (diffuse or focussed) include:
precipitation (volume, intensity, duration); topography (slope, above ground storage);
90
vegetation (cropping pattern, rooting depth) and evapotranspiration; soil and subsoil
types; flow mechanisms in the unsaturated zone (uniform or preferential); bedrock
geology; available groundwater storage; presence of influent rivers, and; presence of karst
features (Misstear, 2000).
6.3 Recharge estimation methods

Various techniques are available for estimating recharge, the selection of which is not
straightforward (Lerner et al., 1990; Scanlon et al., 2002). Each technique has different
assumptions as well as limitations. Therefore, it is recommended to use multiple methods
to reduce uncertainty and to improve conceptual understanding of recharge at a study site
(de Vries and Simmers, 2002; Healy and Cook, 2002). Generally, selection of a technique
is dependent on data availability, which is often lacking in many regions. Such data
scarcity can lead to the selection of a less suitable recharge estimation method as well as
no additional methods to corroborate the findings. Rather than data driving the
methodology used, the user should select methodologies depending on the desired
spatiotemporal resolution. This is easier for primary data collection but less obvious when
dependent on secondary data sources. Then the user must determine what the recharge
result represents, according to the fundamental theory of the method applied and the
satisfaction of the assumptions.
6.4 Study area
6.4.1 General description

The study site is Dangila woreda within the Amhara Region of northwest Ethiopia, 70 km
southwest of Bahir Dar on the Addis Ababa to Bahir Dar road (Figure 6-1).
91
Figure 6-1. Location map of the study area.
6.4.2 Climate
The climate of the region is moist subtropical with little annual temperature variation
though high diurnal variation. The median annual total rainfall is 1541 mm, as measured
(1987-2017) at the National Meteorological Agency (NMA) weather station in Dangila
town, 91% of which falls during May to October (Figure 6-2). Both the mountains to the
east and Lake Tana to the north affect the pattern of rainfall in the study area. Most rain
events have a duration shorter than 1-hour and often occur in the late afternoon (Haile et
al., 2009).
500 30
Rainfall, c90
Rainfall, median 25
400
Rainfall, c10
Temperature (oC)
20
Rainfall (mm)
300 Tmax
Tmin 15
200
10
100
5
0 0
J F M A M J J A S O N D
Figure 6-2. Monthly median, 10th and 90th percentile rainfall, and mean maximum and minimum
temperatures as measured (1987-2017) by the NMA at the Dangila weather station.
92
6.4.3 Hydrogeology
Bedrock geology consists of Cenozoic basalt and trachyte (Tefera et al., 1996) that are
variously massive, fractured and vesicular. Above the bedrock lies weathered basalt
regolith, itself overlain by red clayey loam soils (nitisol). The superficial materials of the
floodplains are occasionally very sandy and gravelly though deep and wide desiccation
cracks suggest a high clay content (vertisol).
Diffuse (direct) recharge dominates across the study site (Figure 6-3) with quantities
likely to vary according to local position. Upslope areas will receive less recharge due to
higher runoff and interflow gradients whereas overland flow, interflow and groundwater
flow collect in the topographic lows. The large floodplains, which are prevalent in the
landscape, become waterlogged in the wet season from direct rainfall and spring
discharge (rather than from overbank flooding).
Precipitation
Hand-dug well
. . . . . . . . . . .
. .. . . . . .
.. . . . . .. . . .. . . . . . . ..
. .. . .. . . .. . .. .
. Recharge . . . . .. .
. . . . . . . . . . . . . . .
Evapotranspiration .. . . . . . . . . . .
.. . . . .Infiltration . . . .
. . . . . . . . . . . . . . .. .
. . .
. . .. .. .. . . . . . . . . . . . .. .. .
.
. .. . . . .. . . . . . . . . . .
. . . . .. . . . . . . .. . . . . . . . .. . . . . Regolith . . . . .. . . . .
.. . . . . . . . . . .. . . . . . .
. . . .. .. . Spring . . .. . . . . . . . . . . . .
. .. . . . . . Wetland . . . . v
. . . . .
. .. . . .. . . v
. . . . . . . . . River
. .. . . . . . . . . . . . . . . . . . . . v
. . . . . . .. . . . . .. . v v Basalt bedrock v v
. . . .
. . . . . .. . . . . . . v
v v v v v v v
Seepage? v
Figure 6-3. Conceptual model of the study site.
6.5 Data used in the study

In this study, nine frequently used methods were applied using data from all possible
hydrological zones. Additional methods were explored and rejected for various reasons,
as discussed in Appendix D. The data requirements for the various methods applied are
shown in Table 6-1. Meteorological data was measured by the NMA weather station in
Dangila town: the only formal weather station in the district. River flow data in Ethiopia
is collected by the Ministry of Water, Irrigation, and Electricity (MoWIE): Amen and
Kilti river flow data were utilised, the latter catchment forming a large portion of Dangila
district, even though the gauging station lies outside the district (Figure 6-1). The
available time series date from 1988 (Amen) and 1997 (Kilti) to late 2014, though there
are occasional gaps in the data. In addition to these formal data sources,
hydrometeorological time series are available from a community-based monitoring
programme at Dangesheta village from March 2014 to January 2017. River stage in the
Brante river was measured twice-daily, rainfall was measured daily in a manual
93
raingauge, and groundwater levels were measured bi-daily in five wells since March 2014
and daily in 25 wells since February 2015. The hand-dug wells have an average diameter
of 1 m with depths ranging from 3-21 m. Rainfall and river stage from the community-
based monitoring have been validated against formal sources confirming the quality of
the data (Walker et al., 2016). The Amen (37.0 km2) and Brante (65.5 km2) are sub-
catchments of the Kilti (631.7 km2). The catchment-scale recharge assessment methods
were applied to all three catchments. Thirty-one shallow groundwater samples were
collected for chloride analysis, from locations distributed throughout the study site, in
March/April 2015 and October/November 2015. Rain could only be sampled during the
second field visit because it did not rain during the four weeks of the earlier dry season
visit, nor during a third visit in January 2017. Three samples were collected from two
sites and occurred whenever rainfall was sufficient to enable direct sampling. All samples
were filtered upon collection and, to prevent evaporation, the nalgene bottles were
completely filled and kept in a refrigerator prior to laboratory analysis by Dionex ion
chromatography. Additional data used in development of the conceptual model and
required to parameterise models resulted from three periods of fieldwork, which included
pumping tests on hand-dug wells (Walker, 2016), geological surveys, hydrochemistry and
stable isotope sampling, radon-222 measurements, water point surveys, and workshops
with the local community (further information is provided in Appendix D). Proportions
of different land use land cover (LULC) types were taken from ADSWE (2015).
Data from three large-scale mapping and modelling recharge studies were also assessed.
The global-scale WHYMAP (WHYMAP, 2016) by BGR (the German Federal Institute
for Geosciences and Natural Resources) and UNESCO gave recharge values of 20-
100 mm/a for the study site, the continental-scale map by Altchenko and Villholth (2015)
gave 100-300 mm/a, and a national-scale map by Ayenew et al. (2008b) gave 250-400
mm/a. Further information on the large-scale mapping and modelling can be found in
Appendix D.
94
Table 6-1. Hydrological zone, spatial scale, and data requirements of the applied recharge estimation
techniques.
Land use/land cover surveys or maps
Digital elevation model (DEM)

Groundwater level time series
Geological surveys or maps

Spatial scale of the method
Meteorological time series

Meteorological averages
Groundwater sampling
River flow time series
Vegetation properties
Access to literature
Hydrological zone
Aquifer properties
Rainfall sampling
Soil properties
 
Empirical A D  a

Stream hydrograph SW C 

SMB UZ D  
 
Water balance A C  
  
CMB SZ D     
WTF SZ L  
  
RIB SZ L   
        
Phys. based modelling A C       
Large-scale mapping A R
b
A = all zones, SW = surface water, UZ = unsaturated zone, SZ = saturated zone, R = regional (1000s km2),
D = district (woreda), C = catchment (10s-100s km2), L = local (100s m2).
a
Access to literature only required if developing a new empirical equation.
b
Assuming consideration of published studies as opposed to developing new large-scale maps.
6.6 Recharge estimation methodologies
6.6.1 Empirical method

In an attempt to establish a rainfall-recharge relationship for Ethiopia, a thorough and
systematic literature search was conducted. Appendix D provides detailed information on
the literature search and a map of the study site locations, which were distributed around
Ethiopia (Figure D1). Forty-nine quantitative studies were located that provided 102
annual recharge estimates to plot against annual rainfall (Figure 6-4). A quadratic
trendline, reflecting an increase in recharge disproportionate to increasing precipitation,
achieved the best R2 and standard error. The resulting relationship is presented as Eqn. 6-
1. Separating the data into the geographic (and consequently, climatic and geological)
regions as shown in Figure 6-4 and fitting linear trendlines gave similar recharge values
as the trendlines plot close to the quadratic regression line. Additional analysis of site-
specific, rather than regional, rainfall intensity, topography, soils and vegetation is
beyond the scope of this study. The regression line is not extended to rainfall below
95
500 mm/a as this is considered the lower limit of applicability of Eqn. 6-1. Where rainfall
is below 500 mm/a, the relationship with recharge is more complex (Bonsor and
MacDonald, 2010) and there were insufficient studies from which a relationship could be
established.
𝑹 = 𝟏𝟑𝟔. 𝟔 − 𝟎. 𝟑𝟎𝟎𝟓𝑷 + 𝟎. 𝟎𝟎𝟎𝟐𝟕𝟏𝑷𝟐 (6-1)
Where R is recharge and P is annual precipitation.
800
Tigray, Afar and
700
Annual recharge (mm)
Dire Dawa
600 Rift Valley and
central Ethiopia
500 Lake Tana Basin
400 (NW Highlands)
300 Other
R2 (adj.) = 0.587
200 S = 81.12
100
0
0 500 1000 1500 2000 2500
Annual rainfall (mm)
Figure 6-4. Plot showing the relationship between annual rainfall and annual recharge in Ethiopia based
on 102 recharge estimates from 49 studies across the country. S = standard error, R2 (adj.) = adjusted
coefficient of determination. The Tigray, Afar, Dire Dawa group has semi-arid climate and highly
heterogeneous geology ranging from Precambrian crystalline to Mesozoic sandstones and limestones to
Quaternary volcanics, generally overlain by leptosols with sparse and herbaceous vegetation. Rift Valley
and central Ethiopia have subtropical highland and tropical savanna climate with Quaternary volcanic
geology, highly heterogeneous soils and rainfed cropland and mosaic forest and grassland. The Lake Tana
Basin has a tropical highland monsoon climate and Cenozoic volcanic rocks overlain by luvisols or
vertisols closer to the lake with mosaic cropland/grassland/shrubland/forest (Tefera et al., 1996; Peel et
al., 2007; Arino et al., 2012; Jones et al., 2013).
6.6.2 Streamflow hydrograph analysis (three methods)

Recharge estimation using streamflow hydrograph methods typically involves separating
the baseflow component (Figure D2 in Appendix D) and idealising that precipitation
entering the aquifer as recharge must be balanced by groundwater discharge into rivers
that forms baseflow. However, there are several ways in which groundwater can be
depleted without contributing towards baseflow, including abstractions, leakage to deeper
aquifers, and evapotranspiration from the saturated zone. Without quantifying these
fluxes, equating baseflow to recharge will lead to underestimation of recharge. It is
important to remind, therefore, that quantifying baseflow is an estimate of groundwater
discharge and provides a minimum estimate of recharge (Szilagyi et al., 2003; Vegter and
96
Pitman, 2003; Risser et al., 2005). Three streamflow hydrograph methods were used in
this study, the baseflow recession method presented by (Meyboom, 1961), and two digital
recursive filter tools, the Web GIS based Hydrograph Analysis Tool (WHAT) (Lim et al.,
2005) and WETSPRO (Willems, 2009). Details of the application of these methods are
presented in Appendix D.
6.6.3 Soil moisture balance (SMB)

The Thornthwaite-Mather (1955, 1957) (T-M) method is essentially a water balance of
the root zone performing monthly book-keeping of precipitation, evapotranspiration and
soil moisture. Deep infiltration below the root zone occurs only when field capacity is
exceeded (see Steenhuis and Van Der Molen (1986)). The direct runoff component is
dealt with by applying a runoff factor or by subtracting a portion of soil moisture surplus;
both methods were applied here. Details of the parameterisation and the tabulated
calculations can be found in Appendix D.
A key assumption of unsaturated zone methods, such as the SMB method, is that the soil
moisture surplus will infiltrate to the water table. However, this water may flow laterally
through the unsaturated zone as interflow without necessarily recharging the aquifer
(Misstear, 2000; Hendrickx and Flury, 2001). Hence, Simmers (1988), Rushton (1997),
Healy (2010) and others refer to the recharge computed from unsaturated zone methods
as potential recharge.
6.6.4 Basin water balance

The water balance, or water budget, simplifies the full water balance equation by
neglecting Qin, A, Qout and S in
𝑷 + 𝑸𝒊𝒏 = 𝑹𝑶 + 𝑨𝑬𝑻 + 𝑹 + 𝑨 + 𝑸𝒐𝒖𝒕 + ∆𝑺 (6-2)
Where P is precipitation, Qin is groundwater flow into the basin, RO is runoff (i.e.
overland flow and interflow out of the basin), AET is actual evapotranspiration (from the
unsaturated and saturated zones and from surface water), R is recharge, A is abstraction,
Qout is groundwater flow out of the basin, and S is the change in storage. The
assumptions are that S is balanced over long time-periods (this appears valid from
groundwater level records), Qin and Qout are negligible as these are headwater catchments
with thin aquifers and rivers founded on bedrock (hence no groundwater flow beneath the
gauge), and abstraction is negligible due to sparse wells with manual-lifting technology.
AET is not straightforward to estimate and was calculated with three methods for
comparison: (1) The T-M method; (2) Application of Turc’s formula (Turc, 1954), and;
97
(3) A value estimated by Allam et al. (2016) for this region of the Tana Basin by
combining remote sensing and river flow records. The average AET values were 789, 831
and 931 mm/a, respectively. See Appendix D for details of the AET and runoff
estimations. Accurate quantification of all the fluxes is always troublesome though is
required in order to leave an accurate residual that is equated to actual recharge (Scanlon
et al., 2002).
6.6.5 Chloride mass balance (CMB)

The CMB method requires mean annual precipitation, chloride concentration of that
precipitation and chloride concentration of the groundwater, is independent of whether
recharge is diffuse or focussed, and integrates recharge rates both spatially across a region
and temporally over long time-periods. The method has several assumptions (Bazuhair
and Wood, 1996):
 All chloride within groundwater originates from precipitation, i.e. there are no
alternative chloride sources such as evaporites or pollution.
 Chloride is conservative in the system (this is generally the case as chloride is not
adsorbed, is unlikely to form salts, and has rare biochemical interaction).
 Recycling of chloride does not occur within the basin area.
 The chloride concentration in runoff is equal to that in precipitation.
 Evaporation of groundwater does not occur upgradient of groundwater sampling
points.
The basic equation applicable for evaluation of recharge using the CMB is
(𝑷𝒆𝒇𝒇 )(𝑪𝒍𝒘𝒂𝒑 ) (6-3)

𝑹=
𝑪𝒍𝒈𝒘
Where R is annual recharge, Peff is average annual effective precipitation (rainfall minus
direct runoff), Clwap is the weight-average chloride concentration in precipitation
including dry deposition, and Clgw is the average chloride concentration in groundwater.
Clgw averaged 2.10 mg/l with a standard deviation of 1.33 mg/l and Clwap averaged 0.68
mg/l (standard deviation = 0.32 mg/l). Details of the parameterisation can be found in
Appendix D.
6.6.6 Water table fluctuation (WTF)

In the WTF method, the upward movement of groundwater level with respect to time is
an indication of recharge and the downward movement indicates recession of
groundwater; no assumptions are made regarding recharge mechanism (see Healy and
98
Cook (2002) for details). Groundwater recharge R is calculated for a particular well by
multiplying the change in water level of two successive groundwater level readings by
the specific yield Sy of the aquifer:
𝑹 = 𝑺𝒚 ∗ ∆𝒉⁄∆𝒕 (6-4)
Where h is water level and t is time. To correctly estimate h, it is necessary to extrapolate
the antecedent recession curve to the point below the peak, i.e. the point that the
groundwater level curve would have reached without precipitation (Figure D5). This
extrapolation was conducted manually on each of the 30 well hydrographs, following the
graphical method described by Delin et al. (2006). For comparison, another approach was
followed that involves calculating the water level rise from one day to the next with a
negative rise, i.e. a fall in groundwater level, counting as zero. This method would be
expected to underestimate recharge because groundwater recession with the absence of
recharge is not considered (e.g. Delin et al. (2006); Choi et al. (2007); Varni et al. (2013)).
Sy of 0.08 was used, obtained from 11 pumping and recovery tests in the area (Walker,
2016).
6.6.7 Rainfall infiltration breakthrough (RIB)

The RIB method is a model for groundwater recharge estimation developed by Xu and
Beekman (2003) based on the cumulative rainfall departure (CRD) method (Wenzel,
1936). The conditions at the field site fit well the requirements detailed by (Sun et al.,
2013): “… the RIB model is best suited for shallow unconfined aquifers with relatively
low transmissivity”. The model considers not only rainfall from a single event but the
series of preceding events that influence breakthrough of water at the water table (for
details see Xu and van Tonder (2001) and Sun et al. (2013)). Time series of rainfall are
required, plus groundwater level and aquifer Sy. The RIB method utilised data from the
30 community-monitored wells and raingauge in addition to Sy of 0.08 (Walker, 2016).
Further details can be found in Appendix D. As with the WTF method, there is the
possibility of accounting for groundwater level rise from lateral flows in recharge
estimation.
6.6.8 Physically-based modelling

SHETRAN (Système Hydrologique Européen TRANsport) is a physically-based
spatially distributed finite difference modelling system for coupled surface and
subsurface water flow in river basins and is openly available at
http://research.ncl.ac.uk/shetran/. SHETRAN is well established in the literature, having
99
been applied to a variety of situations (e.g. Birkinshaw and Ewen (2000b); Bathurst et al.
(2011a); Starkey et al. (2017)), however, it has not previously been used to quantify
recharge. Model setup requires a DEM, catchment mask, geological, soil, vegetation and
LULC information. Further details of SHETRAN, including how recharge is computed
within the model and how the models were parameterised, can be found in Appendix D.
Three nested catchments were modelled, details of which are in Table 6-2. The calibration
procedure involved adjusting geological layer thicknesses, aquifer properties, channel
characteristics, Strickler overland flow roughness coefficient, and evapotranspiration
characteristics until satisfactory matches with observed groundwater level and river
discharge data were achieved. The nested nature of the catchments meant a final matching
set of optimum parameters was selected to satisfy the calibration requirements of all
catchments. Table 6-2 shows calibration statistics for a calibration period; subsequent
simulations during a validation period were deemed acceptable (see Appendix D).
Table 6-2. Details of the three catchments modelled using SHETRAN (see Figure 6-1 for locations).
Catchment Area (km2) Resolution (m) Run length Calibration NSE RMSE
Amen 37 100x100 17 years (Jan 98 to Sep 14) River flow 0.79 0.19 m3/s
Kilti 632 500x500 18 years (Apr 97 to Oct 14) River flow 0.78 1.47 m3/s
Brante 66 100x100 3 years (Mar 14 to Jan 17) GW levels 0.69 2.01 m
NSE = Nash-Sutcliffe efficiency. RMSE = root mean square error.
6.7 Recharge results

Recharge estimates from the various methods show high variability: 45-814 mm/a or
3-53%MAP (median annual precipitation) for the median annual recharge (Figure 6-5
and Table 6-3). The WHYMAP and Meyboom methods were rejected for this study with
full reasoning provided in Appendix D.
100
1000
60%
Annual recharge (%MAP)

800

600 40%
400
20%
200
0 0%
Empirical
National map
WHAT
SMB (T-M)
SMB (runoff factor)
CMB
RIB
SHETRAN
WHYMAP
Continental map
Meyboom
WETSPRO
Water balance (R-S)

Water balance (Turc's)
Water balance (T-M)
WTF (graphical)
WTF (simple)
Figure 6-5. Median annual recharge estimates from all the techniques. The error bars give the interannular
recharge range. T-M = Thornthwaite-Mather method of runoff or AET estimation. R-S = Remote sensing
method of AET estimation.
Table 6-3. Tabulated recharge results shown on the plot in Figure 6-5.

Technique Minimum Median Maximum
WHYMAP 20 60 100
Continental map 100 200 300
National map 250 325 400
Empirical 317
Meyboom 25 45 98
WHAT 146 190 295
WETSPRO 121 176 283
SMB (T-M) 168 374 471
SMB (runoff factor) 253 547 840
Water balance (Turc’s) 392 493 802
Water balance (T-M) 434 535 844
Water balance (R-S) 292 393 702
CMB 427
WTF (graphical) 666 814 872
WTF (simple) 682 752 793
RIB 283 451 466
SHETRAN 260 279 317
101
6.8 Sensitivity analyses
Measured input data and modelling parameters were individually adjusted by ±10% to
assess sensitivity. For some methods, only measured input data could be adjusted, e.g.
rainfall or groundwater level fluctuation. For other methods, it was possible to adjust
modelling parameters determined during additional investigations, calibration or by
“expert opinion”, e.g. the recession constant for WHAT and WETSPRO analysis.
Additionally, to suggest the range of uncertainty, recharge was computed by each method
using the likely maximum deviation in parameter values. Table 6-4 details the parameter
adjustment and Figure 6-6 shows the sensitivity and uncertainty for each method.
Table 6-4. Parameters and input data adjusted for the sensitivity and uncertainty analysis.
Method Parameters/input data individually Maximum likely deviation of
adjusted by ±10%. Most sensitive parameters/input data giving the uncertainty
parameter underlined. range.
Empirical Annual average rainfall 95% prediction interval from the rainfall-recharge
relationship curve
WHAT River flow, BFImax, recession constant Derived BFImax and maximum/minimum recession
constant that still gave an acceptable baseflow
separation
WETSPRO River flow, recession constant, w Maximum/minimum recession constant and w that still
gave an acceptable baseflow separation
SMB (T-M) Rainfall, PET, MC, LULC proportions, % Combined adjustment by ±10% of % surplus to runoff,
surplus to runoff MC and LULC proportions*
SMB (runoff factor) Rainfall, PET, MC, LULC proportions, Combined adjustment by ±10% of runoff factor, MC
runoff factor and LULC proportions*
Water balance Rainfall, temperature (for AET), runoff Combined adjustment by ±10% of rainfall, temperature
(Turc’s) (for AET) and runoff*
Water balance (T-M) Rainfall, AET, runoff Combined adjustment by ±10% of rainfall, AET and
runoff*
Water balance (R-S) Rainfall, AET, runoff Combined adjustment by ±10% of rainfall, AET and
runoff*
CMB Annual average rainfall, Clgw, Clwap Measured range of Clwap (0.38-1.12 mg/l)
WTF (graphical) Water level fluctuation, Sy Measured range of Sy (0.05-0.3)
WTF (simple) Water level fluctuation, Sy Measured range of Sy (0.05-0.3)
RIB Water level fluctuation, rainfall, Sy Measured range of Sy (0.05-0.3)
102
SHETRAN (phys. Rainfall, PET, Strickler coefficient, Sy, Combined adjustment of layer thicknesses and AE/PE
based modelling) hydraulic conductivity, layer thicknesses, ratio by ±10%, and Sy and hydraulic conductivity
AE/PE ratio within measured range that still gave an acceptable
calibration
BFImax = maximum value of long-term ratio of baseflow to total streamflow. w = portion contributing
directly to runoff. PET = Potential evapotranspiration. MC = root zone storage. AE/PE = actual to
potential evaporation ratio. See methodological descriptions in Appendix D for more information on these
parameters.
* The range in parameter/input data was uncertain, i.e. there was no constraining measured range nor
calibration targets.
Maximum absolute change in recharge result when:

most sensitive parameter/input data adjusted by +10% parameters/input data adjusted by likely maximum range
0% 10% 20% 30% 40% 0% 50% 100% 150% 200% 250% 300%
Empirical
WHAT
WETSPRO
SMB (T-M)
SMB (runoff factor)
Water balance (Turc's)
Water balance (T-M)
Water balance (R-S)
CMB
WTF (graphical)
WTF (simple)
RIB
SHETRAN
Figure 6-6. Comparison of the sensitivity of each recharge estimation method to ±10% adjustment in
measured input data and modelling parameters (left) and range of uncertainty when the maximum likely
deviations are applied (right).
The left plot in Figure 6-6 highlights the varying sensitivity of the methods. For example,
it shows the water balance methods’ high sensitivity to rainfall input and, essentially,
lower sensitivity to any single parameter when the number of parameters increases (e.g.
SMB and SHETRAN). The right plot in Figure 6-6 highlights the varying range of
uncertainty in recharge result from different methods, which is dependent on the degree
of uncertainty in the input parameters. For example, while the WTF and RIB methods
show low sensitivity to a 10% variation in parameters, the recharge result has high
uncertainty because the measured range in Sy was high; Sy is commonly uncertain due to
the difficulties in accurate measurement and the heterogeneous nature of many aquifers.
Uncertainty reduces with those methods that involve calibration, e.g. WHAT and
SHETRAN, as the maximum possible deviation in parameter values decreases.
103
Additionally, when there is high uncertainty in input data, Figure 6-6 suggests which
methods may be better selected.
6.9 Discussion
6.9.1 Reasons for the range in results

The range of recharge results presented in Figure 6-5 does not necessarily mean that some
results are incorrect, as they need to be considered in the context of their spatiotemporal
scale, what they represent, and the limitations of each method. A recharge value that is
comparatively high or low can provide insights on the conceptual model, especially if
previously the conceptual model expected the method to provide an actual recharge
estimate, and insights on uncertainty. A summary of the methods is provided in Table
6-5.
Table 6-5. Summary of methods and suggestions for lessening uncertainty in the recharge results. It
should be restated here that while the specific methods usually compute the specified type of recharge,
this is particular to the conceptual model of the study site.
Method Type of recharge Under/over Uncertaintyb How to lessen uncertaintyc
computed estimatesa
WHYMAP Actual Under Rejected because its scale is inappropriate for this study resulting
in gross underestimation of recharge
Continental map Actual Under High Use other methods
National map Actual Applicable High Use other methods
Empirical Actual Applicable High Increase number of recharge studies considered

with greater geological, soils, vegetation and
climate detail
Meyboom Minimum Under Rejected due to problems of application on the study site
hydrographs resulting in gross underestimation of recharge
WHAT Minimum Under Low Utilise longer river flow time series and
additional series from nested catchments
WETSPRO Minimum Under Low-medium As above
SMB (T-M) Potential Applicable Medium Increase rainfall and PET measurement density,
utilise higher resolution soil and vegetation
mapping and use a daily computation time step
SMB (runoff factor) Potential Over Low-medium As above
Water balance (Turc’s) Actual Over Medium-high Increase rainfall and PET measurement density,
utilise higher resolution vegetation mapping for
better AET estimation, and use a daily
computation time step
Water balance (T-M) Actual Over Medium As above
Water balance (R-S) Actual Applicable Low-medium As above
CMB Actual Applicable Medium Increase rainfall chloride sampling frequency
104
WTF (graphical) Change in storage Over High Obtain more Sy estimates, utilise piezometers
that are not biased towards good groundwater
supply
WTF (simple) Change in storage Over High As above
RIB Change in storage Over Medium-high As above
SHETRAN (phys. Actual Applicable Low Increase rainfall and PET measurement density,
based modelling) obtain more Sy and hydraulic conductivity
estimates, and aquifer geometry measurements
(e.g. with geophysics), utilise more river flow
and groundwater level records for calibration
a
In comparison to the estimated actual recharge range for the study site of 280-430 mm/a.
b
This relates to the sensitivity and uncertainty ranges in Figure 6-6 and the robustness of the method.
c
The suggestions present a best-case scenario should time and budget allow.
As previously stated, unsaturated zone methods may overestimate recharge, explaining
why the SMB methods applied here show high recharge values, i.e. they are calculating
potential recharge. The other uncertainty relates to which method to choose to determine
the amount of runoff; application of a runoff factor based on measured river flows has
lower uncertainty.
The streamflow hydrograph methods provide the lowest recharge estimates, supporting
their classification as computing minimum recharge. While the Meyboom method was
rejected (see Appendix D), the similarity of the WHAT and WETSPRO recharge results
provides confidence in their minimum recharge estimates.
Considering the WTF and RIB methods, the suggestion by Healy and Cook (2002) that
monitoring wells should be positioned in a “representative” location is reasonable for
purposely installed piezometers, but hand-dug wells will naturally be excavated where
generations of experience indicate has good potential for groundwater abstraction, i.e.
there is a bias towards areas that receive lateral in addition to vertical recharge. It is
unsurprising then that the WTF methods give the highest recharge estimates of all
methods as they are actually computing the change in aquifer storage on a much smaller
scale (10s of metres) than the other methods. The RIB method utilised the same
groundwater level datasets and Sy, though is constrained by the incorporation of a rainfall
time series thus giving lower recharge estimates.
The empirical method is simple, but is built upon a substantial quantity of work by the
authors of the studies used in the development of the method. However, confidence in the
recharge result is low, due to several factors:
105
 Confidence in the quality of the published studies: The generation of the rainfall-
recharge relationship considered recharge estimates from all identified studies,
even though 56% used only a single recharge estimation method and there was
often uncertainty if the conceptual model meant applicability of assumptions or
the type of recharge computed.
 Confidence in the transferability of the results: Figure D1 in Appendix D shows
that the geographical distribution of the studies is biased to the Lake Tana Basin,
Tigray, Dire Dawa, and around Addis Ababa. These four regions have specific
rainfall intensity, evapotranspiration, hydrogeological and topographic
characteristics that control the recharge rate.
 Confidence in the appropriateness of using annual rainfall total: Considering only
the annual total rainfall fails to recognise the importance of rainfall intensity and
distribution throughout the year. For example, a unimodal and a bimodal rainfall
pattern would give different recharge rates even with the same annual total rainfall
(Kingston and Taylor, 2010).
The water balance methods should give actual recharge if the other fluxes are accurately
quantified. While we may have a degree of confidence in values used for runoff and
precipitation, AET is difficult to estimate, as the range in AET estimates from the three
applied methods attests. The relatively high recharge estimates from the water balance
methods are likely to be a symptom of underestimation of AET and greater uncertainty
comes with decreasing robustness of AET estimation.
There is some uncertainty in the CMB recharge result due to the small number of rainfall
chloride measurements and the assumption that chloride is not introduced into
groundwater from any other source but precipitation. This assumption is valid at the study
site regarding pollution and evaporites, which are not present, however,
evapotranspiration from the saturated zone or from seepages that re-infiltrate may cause
an increase in the chloride concentration of groundwater. The discrepancy in recharge
result of the CMB method may be because it averages over a longer period and larger
area than the other applied methods.
SHETRAN modelling computes the change in aquifer storage for each cell, which
becomes actual recharge when integrated over the catchment area as adjacent lateral
inflows cancel. There is high confidence in these recharge estimates due to: substantial
locally derived data was used to set up and calibrate the models as opposed to relying on
just a few, potentially highly localised, input datasets or relying on averages; interannual
106
variations in recharge totals correlate well between catchments with r = 0.81, and;
recharge estimates are not sensitive to adjustments in individual parameter values. The
spatially distributed nature of the model means that spatial variations in recharge due to
lateral groundwater flow can be observed and understood, rather than providing
misleading recharge estimations. Similarly, interannual variations in storage can be
observed and measured rather than assumed to be negligible. However, this robustness of
method depends on quantity and quality of data available for model setup, calibration and
validation in addition to requiring a skilled operator with the necessary time available.
Exploring the models’ mass balances indicated why the SHETRAN recharge estimates
are lower than those from other methods: recharge is reduced because, unlike other
methods here presented, SHETRAN computes canopy and open water evaporation, both
of which are significant at this site.
The map presented by Ayenew et al. (2008b) was produced only at Ethiopian
national-scale and incorporates more local studies and experience than is possible with
global or continental-scale maps. Therefore, assuming that those local studies were
conducted robustly, the national map gives a recharge estimate for which we have greater
confidence.
6.9.2 Insights gained on the conceptual model

The obvious insights gained from the multi-method comparison was that not all methods
were computing actual recharge or the results would be more similar (given similar
spatiotemporal scale). Therefore, some assumptions must have been unsatisfied, which,
rather than invalidating a method altogether, meant that the method was computing
potential or minimum recharge or change in aquifer storage. Insights gained on the
conceptual model mostly concern the amount and type of evapotranspiration, and the
spatial variability of groundwater availability. High recharge values from the SMB
methods indicate that all infiltration, which unsaturated zone methods are actually
measuring, does not form recharge and there is likely to be interflow followed by
discharge and/or evapotranspiration. The streamflow hydrograph methods’ lowest
recharge estimates indicate that groundwater is depleted prior to contributing to baseflow.
Evapotranspirative losses from the saturated zone must be significant, which was thought
likely given the shallow wet season water tables and spring/seepage-fed inundated
floodplains. High recharge values from the water balance methods are also suggestive
that evapotranspiration may have been underestimated. Further evidence for this is the
lower recharge estimate from SHETRAN that is due to its comprehensive simulation of
107
canopy and open water evaporation and transpiration from the unsaturated and saturated
zones resulting in greater total evapotranspiration losses. The high recharge values from
the water table fluctuation methods, and high variability between wells, demonstrate the
spatial variability in groundwater availability. The results show that groundwater flow,
interflow and storage in certain areas can provide high potential for abstraction. Examples
of other studies where fewer methods were applied and useful insights were gained are
included in Appendix D.
6.9.3 Recharge estimate for Dangila woreda

Considering which types of recharge and spatiotemporal scales are relevant to this study,
we restate the purpose as being to determine the resilience of shallow groundwater
resources used for irrigation by rural communities in the Dangila area of Ethiopia;
estimates of long-term annual actual recharge at multiple catchment-scales are therefore
of primary interest. Although spatial assessments of aquifer storage change for small-
scale shallow aquifers, particularly at the seasonal-scale, are also of significant interest to
identify areas with the greatest potential for groundwater abstraction.
Considering the different types of recharge (see Table 6-5), while the median recharge
values from all of the methods used range from 45 to 814 mm/a, we expect that the long
term actual recharge averaged over the general study area lies somewhere between the
minimum and potential values of 176 and 547 mm/a, given by the lowest streamflow
hydrograph and highest SMB methods, respectively. The range of median values given
by all actual recharge methods is 279-535 mm/a.
With regard to spatial scales, the methods based on groundwater level time series are
highly localised and dependent on lateral inflows and other local factors, with values of
recharge for individual wells from the RIB and the WTF methods ranging from under
100 to over 1600 mm/a. At the catchment-scale, recharge values for the three catchments
for each method used were generally consistent (see Appendix D), indicating some spatial
consistency at this scale.
Having separated out and considered results by different types of recharge and spatial
scales, determination of reliable actual recharge estimates for the general area around
Dangila requires consideration of the confidence given to each relevant method. This can
be based on factors discussed earlier, including: temporal representativeness of time-
series data; spatial representativeness of data; errors and uncertainties in input data;
sensitivity of models to parameter values and input data; whether assumptions of methods
108
are met. We have greatest confidence in the water balance method using the higher AET
rate, the CMB method, and the SHETRAN modelling. Thus, we identify a reliable
recharge range for the Dangila area of 280-430 mm/a, which is consistent with the range
from the national map (Ayenew et al., 2008b).
6.10 Conclusions
Nine methods, with 17 variations, of groundwater recharge estimation were applied for a
shallow aquifer in Ethiopia resulting in a wide range of median annual recharge values
from 45 to 814 mm. This research shows that application of a range of methods may give
a broad range of recharge values, but that it may not be necessary to discard results that
appear to be outliers as these provide useful information. Consideration must be given to
exactly what the “recharge” value represents: potential, minimum, or actual recharge, or
change in aquifer storage. It is clear from the results presented that some methods
providing estimates of potential recharge or storage change are likely to deliver
overestimates of actual recharge while others that represent minimum recharge will
deliver underestimates of actual recharge. Considering each method’s spatiotemporal
scale and uncertainty, we conclude that the most reliable recharge estimates for actual
recharge in the general Dangila area are in the range 280 to 430 mm/a.
Insights gained from the multi-method comparison study, including in particular

assessment of results from methods where the usual assumptions were not strictly valid,
enabled the hydrogeological system be better understood. Firstly, by indicating that
evapotranspiration is significant from a) the saturated zone, and b) the unsaturated zone
following infiltration past the root zone due to interflow and seepage. Secondly, by
revealing the spatial variation of the change in aquifer storage, which locally can be
significantly higher than actual recharge estimates, giving further insight and confidence
that areas could be identified with high potential for abstraction for small-scale irrigation.
Even though our recharge range is comparable to the national map results, we now have
much higher confidence in the results and better understanding of our catchments and
aquifers from our analyses.
This study has demonstrated for an extensive range of commonly used recharge methods
applied at a single site that, in addition to quantifying uncertainty of recharge estimations,
results from multi-method comparisons should be clearly interpreted in relation to the
types of recharge and spatiotemporal scale they represent, but can also provide additional
benefits through improved hydrogeological understanding.
109
Chapter 7. Modelling the shallow aquifer

The recharge assessments in Chapter 6 showed that the Dangila woreda shallow aquifer
has the potential to support development of small-scale irrigation. How the resource
availability varies spatially and temporally is the subject of Chapter 7. This chapter
presents information on the many catchment-modelling options and explains why
SHETRAN was chosen and how the program works. Details are provided on model
construction and parameterisation followed by a description and results of the calibration
process. The modelling aims were to simulate observed historical conditions to improve
understanding of the hydrogeological system; this improved understanding is described.
The resulting maps of shallow groundwater potential for irrigation are shown and
discussed.
7.1 Catchment modelling
7.1.1 Introduction
The conceptual model of the study site, presented in Chapter 4 and developed though
field investigations and observations, illustrates our understanding of the shallow aquifer
system. However, the static nature of the conceptual model means it alone is insufficient
to enable us to inform water resource management strategies. Due to the standard issue
of paucity of observational data, we cannot confidently make predictions for areas or
time-periods where data is unavailable. While Chapter 6 gave an estimation of recharge
– the renewable portion of aquifer storage – that is also insufficient to inform water
resource management strategies. Bredehoeft (2002) stated quite clearly, following on
from works by Theis (1940), Brown (1963) and Sophocleous (2000): “The idea that
knowing recharge is important in determining the size of a sustainable groundwater
development is a myth… The important entity in determining how a groundwater system
reaches a new equilibrium is capture. How capture occurs in an aquifer system is a
dynamic process. For this reason, hydrologists are occupied in studying aquifer dynamics.
The principal tool for these investigations is the groundwater model.” To understand the
dynamic nature of a hydrogeological system it must be modelled numerically, which
allows the conceptual model to be tested and consequently updated. Spatial variation of
fluxes and storage can be analysed. What’s more, potential changes to conditions at the
study site can be simulated for assessment of impacts (see Chapter 8).
110
7.1.2 Purpose of the modelling
The initial aim of the modelling was to increase hydrogeological understanding of the
shallow groundwater system to aid in development of the conceptual model. The
following research questions were posed:
a. Are the models satisfactorily reflecting reality in terms of river flows and
groundwater levels?
b. Are the parameter values appropriate considering field investigations and
published literature?
c. What effect does incorporating various hydrogeological features presented in key
regolith and dambo literature, but not necessarily identified in the field, have on
the simulations?
d. Is groundwater availability recharge controlled or storage controlled?
Once these questions were answered satisfactorily and the models were considered to be
simulating current conditions well, the objective was then to answer research question 4
from Chapter 1: Are there easily identifiable zones that show the greatest potential for
sustainable intensification of agriculture through shallow groundwater irrigation?
7.1.3 Model types

A model is by definition a simplified representation of a real-world system or process
(Fetter, 2001; Wagener, 2003). A conceptual model, such as that described for the Dangila
study site (presented in Chapter 4), aims to describe the hydrogeological/hydrological
system with a much reduced complexity reflecting a qualitative understanding of how the
system works. On the other hand, a numerical model applies equations to link the
quantitative inputs and outputs of a system in order to emulate observations enabling us
to better understand the spatial and temporal variation (Refsgaard and Abbott, 1990).
Numerical hydrogeological and hydrological models have a broad range of applications:
from large-scale studies of impacts of climate change on groundwater resources, e.g.
Goderniaux et al. (2009); Jackson et al. (2011); Ali et al. (2012); to small-scale flood
frequency estimation, e.g. Prudhomme et al. (2003); Blazkova and Beven (2009); Calver
et al. (2009); to multi-scale studies of aquifer contamination (e.g. Conan et al. (2003);
Harvey et al. (2006); Karatzas (2017). In these cases and many others, numerical
modelling is a useful tool to aid decision-making concerning water resource management,
adaptation planning, risk mitigation, hydraulic structure design, and remediation strategy,
amongst other applications.
111
Hydrological models are generally classified in three main groups: (1) empirical “black
box”, (2) lumped conceptual or “grey box”, and (3) physically-based distributed or “white
box” (Refsgaard and Knudsen, 1996; Devia et al., 2015). Empirical black box models are
based entirely on mathematical relationships between observed data and hydrological
inputs/outputs (Refsgaard, 1996), examples include data-driven approaches and neural
network models (e.g. Sudheer et al. (2002); Wu and Chau (2010); Kan et al. (2015)). The
next class of models are more physically-meaningful with mathematical functions
describing fluxes between different storages (Todini, 2007) such as SIMHYD (Peel et al.,
2000) or TOPMODEL (Beven et al., 1984). Physically-based models are the most
complex of the numerical hydrological model types, being the most extreme in their
detailed representation of physical processes and in the number of parameters that must
be evaluated (Parkin et al., 1996). Physically-based models have been in use for
catchment hydrology for almost 50 years, following the blueprint proposed by Freeze and
Harlan (1969). Owing to their distributed nature and implication of sound physical
reasoning, physically-based models have often been considered to be particularly
applicable to modelling changes to catchments (e.g. climate or land use) and spatial
variation of inputs and outputs (e.g. Beven and O'Connell (1982); Abbott et al. (1986a);
Bathurst (2011)). For these reasons, a physically-based model was selected for use in this
study.
Probably the most commonly used groundwater model is the MODFLOW program from
the U.S. Geological Survey, used in various guises for over 40 years by academics and
consultants (McDonald and Harbaugh, 1988; Refsgaard et al., 2010; Hughes et al., 2017).
MODFLOW was not selected for this study for several reasons. Firstly, it only simulates
flow in the saturated zone and is not recommended for use when there are important
surface water processes such as floods to consider (Cushman and Tartakovsky, 2016).
Consequently, the U.S. Geological Survey developed GSFLOW by combining
MODFLOW and their precipitation runoff modelling system (PRMS) to better simulate
surface/groundwater interactions. Secondly, evapotranspiration packages (e.g. ETS and
EVT) allow limited vegetation specific adjustment, essentially only evapotranspiration
extinction depth and the rate of evapotranspiration per hydraulic head (i.e. from the
saturated zone). Given the climate of the study site with its long dry season and the low
storage volume of the aquifer due to its thinness, it is important from a water resource
assessment perspective to have accurate representation of evapotranspiration.
Parameterisation of rooting depth, leaf area index, and proportional vegetation cover are
desirable, as well as simulation of evapotranspiration rates according to soil moisture
112
levels and pressure, particularly when the land use / land cover (LULC) change
simulations are expected to most strongly impact on the surface and groundwater regimes
through changes to evapotranspiration. Thirdly, time series data available to calibrate a
model are sparse, in spatial and temporal extent. A physically-based model allows
investigation of the conceptual model and spatial responses of the system, constrained by
sensitivity analyses and the available observational data. Fourthly, the graphical user
interfaces on which MODFLOW is run are expensive. An important criterion of this PhD
project is that all the shallow groundwater investigations should be able to be repeated in
different regions of sub-Saharan Africa by local researchers; hence, freely available
software was prioritised.
7.1.4 SHETRAN
SHETRAN is a physically-based spatially distributed finite difference modelling system
for modelling coupled surface and subsurface water flow in river basins (Ewen et al.,
2000). The main advantages of SHETRAN over alternative physically-based spatially
distributed river basin modelling systems are its comprehensive nature and capabilities
for modelling subsurface flow and transport. The subsurface is treated as a variably
saturated heterogeneous porous medium, and fully three-dimensional flow and transport
can be simulated for combinations of confined, unconfined, and perched systems. The
‘‘unsaturated zone’’ is modelled as an integral part of the subsurface, and subsurface flow
and transport are coupled directly to surface flow and transport (Ewen et al., 2000).
SHETRAN is freely available at http://research.ncl.ac.uk/shetran/.
SHETRAN originated from SHE (Système Hydrologique Européen), which was

developed in the 1980s by the British Institute of Hydrology, Danish Hydraulic Institute
(DHI), and SOGREAH, France (see Abbott et al. (1986a)). The program has since
evolved in two directions, as MIKE-SHE, further developed by DHI on a commercial
trajectory, and as SHETRAN, developed at Newcastle University (Refsgaard et al.,
2010). The improvements over the original SHE program were in the incorporation of
fully three-dimensional subsurface components in additional to solute and sediment
transport modules (Ewen et al., 2000). The main physical processes represented in the
water flow component of SHETRAN (the sediment and solute transport components were
not used for this study) are shown in Table 7-1. These physical processes are represented
by physical, mostly partial differential, equations that are listed in Table 7-2.
113
Table 7-1. Main processes represented in the water flow component of SHETRAN (after Ewen et al.
(2000)).
Water flow component Processes

Surface water flow on the ground Canopy interception of rainfall
surface and in stream channels; Evaporation and transpiration
soil-water and groundwater flow Infiltration to subsurface
in unsaturated and saturated Surface runoff (overland, overbank, and in channels)
zones, including systems of Snowpack development and snowmelt
confined, unconfined and perched Storage and 3D flow in variably saturated subsurface
aquifers Combinations of confined, unconfined, and perched
aquifers
Transfers between subsurface and river water
Groundwater seepage discharge
Well abstraction
River augmentation and abstraction
Irrigation
Table 7-2. Flow equations for SHETRAN applicable in this study (after Ewen et al. (2000)).
Process Equation Reference

Subsurface flow Variably saturated flow equation (3D) Parkin (1996)
Overland flow Saint-Venant equations, diffusion approximation Abbott et al. (1986b)
(2D)
Channel flow Saint-Venant equations, diffusion approximation Abbott et al. (1986b)
(flow in a network of 1D channels)
Canopy interception Rutter equation Abbott et al. (1986b)
and drip
Evaporation Fraction of potential evaporation derived from Abbott et al. (1986b);
Penman-Monteith equation Allen et al. (1998)
A SHETRAN model is divided into a grid with each cell comprising a column of finite
difference cells (Figure 7-1). Where geological layer thicknesses vary between adjacent
columns, the number of cells may differ; to minimise computational difficulties, yet give
a very flexible system, each cell in each column exchanges water with a maximum of two
cells in each adjacent column (Figure 7-1) (Ewen et al., 2000). The columns, along with
channel links, form the main computational structures in SHETRAN. Channel links run
along the edges of grid elements (as shown for the four modelled catchments in Figure
7-4) and are assigned widths and cross-sections. Stream-aquifer interaction occurs
114
through a channel’s bed, which is handled in the same way as flows at the ground surface,
and through channel sides, where a time-varying lateral head boundary condition is
prescribed (Parkin, 1996).
Surface water
Soil Lateral flows
Unsaturated
zone
Water table
Regolith
Vertical flows
Saturated
zone
Basalt
Figure 7-1. SHETRAN columns of finite difference cells showing modelled processes and geological
layers consistent with the four modelled catchments, including lateral connections between layers of
differing thickness.
SHETRAN has a high data requirement; the datasets required for parameterisation are
listed below:
 Meteorological time series

 Mask delineating the catchment
 Topography including presence of lakes
 Size and location of columns, river links and finite-difference cells
 Soil/rock types and depths
 Porosity and specific storage of soils/rocks
 Hydraulic conductivity for soils/rocks
 Land use/vegetation
 Canopy drainage parameters and storage capacities
 Ground cover fractions
 Vegetation root density over depth
 Human-controlled:
o channel flow diversions and discharges
115
o Rates of abstraction and artificial recharge
The datasets are combined as layers for the construction of a SHETRAN model (Figure
7-2).
Figure 7-2. The data layers incorporated into a SHETRAN model (after Lewis (2016)).
SHETRAN is proven in the literature, having been used for a variety of catchment
modelling studies. Examples include modelling nitrate transport in a catchment in
southwest England (Birkinshaw and Ewen, 2000a), modelling landslide sediment yield
in the Spanish Pyrenees (Bathurst et al., 2006), and estimating runoff and flood peaks in
a catchment in Maharashtra, India (Naseela et al., 2015). Chapter 6 included a novel use
of SHETRAN for recharge assessment, for which new code was required for the
SHETRAN program (Walker et al., 2018). The use of SHETRAN detailed in this Chapter
is similar to previous uses available in the literature that have assessed impacts of climate
variability and LULC change. For example, Lukey et al. (2000) used SHETRAN to
estimate the impact of reforestation on a heavily eroded “badlands” environment in
southeast France showing how runoff and sediment yield would decrease. Bathurst et al.,
(2011a and 2011b) modelled four Latin American catchments with SHETRAN to show
the degree of forest cover required for different-sized catchments to affect peak discharge
116
from extreme rainfall events. Schmidt et al. (2008) modelled various LULC scenarios,
adjusting the area of pasture and native vegetation, for a catchment in New Zealand to
assess changes to sediment yield. Two independent studies for southern Iberia used
downscaled global circulation model (GCM) and regional climate model (RCM) outputs
with SHETRAN to simulate potential runoff reduction due to climate change (Mourato
et al., 2014; Guerreiro et al., 2017). Notably, of all the mentioned studies, only that by
Guerreiro et al. (2017) had a significant consideration of hydrogeology, the other studies
essentially only considered shallow soil water in their subsurface component. While
SHETRAN has been used to assess the impacts of groundwater abstraction on river flows
(Parkin et al., 2007), it has not (knowingly) previously been used to assess the potential
for groundwater abstraction nor the impacts of the onset of groundwater abstraction on
surface and groundwater resources nor the impacts of climate variability and LULC
change on those resources. Similarly, SHETRAN has had limited application in Africa
with only a single study found in the literature that assessed soil erosion at a catchment
in Burkina Faso (Op de Hipt et al., 2017). In addition, a conference paper was identified
online of uncertain origin and date detailing a study from Uganda where it was shown
that land use change to increased agricultural land has a had a greater impact on flood
occurrence and surface water availability than a slight upward trend in precipitation
(Bernard et al., ?). Again, neither of these African studies had a significant groundwater
component.
7.1.5 Methodology
Catchment models were built using SHETRAN based on the conceptual model and data
from field investigations and open source remote sensing products. Calibration of
uncertain parameters against groundwater levels and river flow led to further
development of the conceptual model. Once the models were considered to be
satisfactorily representing the natural system, post-processing of spatially distributed
SHETRAN outputs could be used to create maps of varying potential for shallow
groundwater abstraction for irrigation.
7.1.6 Similar modelling studies

Modelling studies from the Lake Tana Basin are not uncommon in the literature, with a
predominance in the use of the SWAT (Soil and Water Assessment Tool) catchment
modelling system. Frequent studies involve predicting flows and sediment yields of the
rivers that feed Lake Tana, often for assessment of land degradation (e.g. (Setegn et al.,
2008; Easton et al., 2010; Setegn et al., 2010; Addis et al., 2016)). SWAT is highly used
117
being freely available, applicable on geographic information system (GIS) interfaces on
which much of the required data is also freely available, and is straightforward to set up
and calibrate. However, van Griensven et al. (2012) are critical of SWAT’s application
in the Upper Nile Basin countries due to reported use of unrealistic parameter values, a
lack of reporting of parameter values making critical evaluation impossible, a general
lack of attention to vegetation parameters, and comparison of SWAT applications at the
same study site by different research teams and/or model versions giving very different
results. Less common from the Tana Basin are modelling studies with explicit
consideration of groundwater. Chebud and Melesse (2009) used MODFLOW to estimate
the groundwater contribution of the Gumera watershed to Lake Tana. Even though
Kebede et al. (2006) and Kebede et al. (2011), using a water balance and chemical
isotopes, also showed groundwater contribution to Lake Tana, most modelling studies
consider the lake and groundwater as separate entities in order to simplify the modelling
by including an empirical groundwater component (e.g. Wale et al. (2009); Dargahi and
Setegn (2011)).
Further afield in a small (2 km2) sub-basin of the Zenako-Argaka basin Tigray, with a
shallow aquifer similar to Dangila (vertisol and colluvium above trap basalts),
Walraevens et al. (2009) used a combination of a runoff model and a soil moisture balance
model to estimate recharge and a MODFLOW groundwater flow model calibrated against
occasional groundwater level measurements from six piezometers. This study increased
hydrogeological understanding and gave insights into recharge and discharge
mechanisms of the aquifer to improve the effectiveness of the implemented water
conservation measures. Also in Tigray, in the 5260 km2 Geba Basin, Gebreyohannes et
al. (2013) used the spatially distributed water balance model WetSpass to produce maps
of long-term average runoff, evapotranspiration and recharge. The maps showed where a
groundwater could be abstracted during the wet season to supplement rainfed agriculture
when rains are poor. However, the safe yield was simply calculated as 25% of
groundwater recharge for the particular location; groundwater flow was not modelled and
no aquifer parameters (transmissivity or specific yield) were incorporated into the model.
This study appears to be the first in the region that explicitly models shallow groundwater,
coupled with surface water, in order to assess the potential of the shallow aquifer resource
for productive use.
118
7.2 Model construction
7.2.1 Modelled catchments and resolution

The four modelled catchments are shown in Figure 7-3. Table 7-3 provides information
on the catchments. The decision to model four nested catchments of different sizes was
to assess if optimum parameters achieved through calibration could satisfactorily be
applied to all the models giving confidence in the optimum parameter uniqueness, to
assess if there was any scale dependency in results, and to assess the required resolution
for mapping of groundwater abstraction potential and impacts of changes. The smaller
models (Brante and Amen) have a resolution of 100 m while the smaller Kilti model,
“Kilti-Dangesheta”, has a resolution of 200 m, and the larger “Kilti-Durbete” has a
resolution of 500 m (Figure 7-4). As high a resolution as possible was chosen within the
limits of the expected simulation run time, and considering the DEM had a resolution of
~90 m.
Figure 7-3. Location map of the four modelled catchments.
119
Table 7-3. Information about the catchments modelled with SHETRAN.
Name Referred to as: Gauge Monitored Area Model Sub-

by: resolution catchment
of:
Kilti Kilti-Durbete, Kilti@Durbete MoWIE 632 km2 500 m Gilgel

larger Kilti Abay
model
Amen Amen Amen@Dangila MoWIE 37 km2 100 m Smaller

Kilti model
Kilti Kilti- Dangesheta Dangesheta 165 km2 200 m Larger

Dangesheta, community community Kilti model
smaller Kilti
model, Kilti-sub
Brante Brante Dangesheta Dangesheta 66 km2 100 m Larger

community community Kilti model
7.2.2 Digital elevation model (DEM) and catchment masks

The original SHETRAN model constructed by Dr G. Parkin during the one-year
AMGRAF project under a NERC catalyst grant used an ASTER GDEM with a 25 m
resolution from The Ministry of Economy, Trade, and Industry (METI) of Japan and the
United States National Aeronautics and Space Administration (NASA). It was
determined during the first field visit in March/April 2015 that this DEM was inaccurate.
The stream network is created within SHETRAN according to DEM elevations and many
of the created streams were not observed on the ground or streams that run parallel had
been merged by SHETRAN. This ground-truthing led to the selection of an alternative
DEM from NASA’s Shuttle Radar Topography Mission (SRTM) which, while of a lower
resolution (3 arc-seconds, or approximately 90 m), proved to give the best representation
of the topography of the area of focus. The DEM and minimum DEM were resampled
from the SRTM DEM using QGIS, an open source geographic information system
application, to the required resolution for each model. The catchment boundaries, or
catchment masks, were delineated and resampled using tools within the QGIS GRASS-
GIS toolbox. Lakes layers were unnecessary as no lakes are found within the catchments.
120
Figure 7-4. SHETRAN grid, DEM and stream networks for the four modelled catchments. Note the different horizontal and vertical scales of the catchments.
121
Outlet
Outlet Kilti-Dangesheta Kilti-Durbete
Amen Brante
Outlet
Outlet
Zone 1 – Flat
floodplain pasture
Zone 2 – Rolling hills

and cultivated land
Zone 3 – Uplands with

scrub-like vegetation
0 1 2 3 km 0 1 2 3 km
0 5 10 15 km
100 m grid squares 200 m grid squares 1 km grid squares
Figure 7-5. Maps showing the distribution of the hydrogeological and LULC zones of the four modelled catchments.
122
7.2.3 Meteorology
SHETRAN models require time series of precipitation and potential evapotranspiration
(PET). Daily precipitation was available from the Ethiopian National Meteorological
Agency (NMA) for the weather station within Dangila town from 1st January 1994 to 31st
October 2015. The Dangila town weather station is the only formal monitoring station
within Dangila woreda and any of the modelled catchments. Daily precipitation is also
available from the Dangesheta community-based monitoring programme from 10th March
2014 to 8th January 2017. The quality of the data from the community-based monitoring
programme was confirmed by statistical comparisons with data from formal sources as
described in Chapter 5 and in Walker et al. (2016). PET was calculated using the
standardised FAO-56 Penman-Monteith reference evapotranspiration (Allen et al., 1998),
which requires maximum and minimum daily temperatures, wind speed, sunshine hours
and relative humidity. These parameters were available from the NMA for the Dangila
weather station at a monthly time-step from January 1985 to December 2006 and daily
from 10th January 2010 to 31st October 2015. Attempts to obtain the missing (2007-2009)
PET data and obtain precipitation and PET data since November 2015 proved
unsuccessful; despite visits to the NMA regional office in Bahir Dar and promises made
by their staff. In order to fill the gaps in the PET data, it was determined that using average
monthly values was satisfactory: The interannual variation in monthly PET totals is low,
as can be seen in Figure 7-6; the coefficient of variation for interannual monthly values
ranges from only 3.4% (September) to 6.8% (June). The single meteorological monitoring
station operating at any one time within the catchments means the precipitation and PET
layer for SHETRAN was homogenous across the whole catchment.
6
Monthly average PET (mm/day)
2
1985 1986 1987 1988 1989 1990 1991 1992
1 1993 1994 1995 1996 1997 1998 1999 2000
2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011 2012 2013 2014 2015
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 7-6. Interannual variation in monthly PET at Dangila.
123
7.2.4 Soils and geology
Hydrogeological zones
Chapter 4 provided details of the geological investigations. Essentially, loamy soils

overlie regolith that overlies very low permeability basalt. The soils and geology layer
for input into SHETRAN relied on field investigations as soil and geological mapping of
the area is not available at sufficiently high resolution. The observed variation in well
depths, differences in monitored well responses, variation in well pumping/recovery test
results, geological and soil observations, and discussions with local communities led to
the definition of three hydrogeological zones. It is noted that each zone is a land type and
not a specific location, i.e. there are multiple locations designated as each zone.
1. The first zone comprises the widespread and often expansive floodplains. These
features are flat or of very low gradient and provide year-round pasture for
livestock. Desiccation cracked surfaces during the dry season indicate a clay-rich
composition though sandy and gravelly patches are observed, particularly closer
to flowing channels. The floodplains are generally waterlogged during the wet
season, more from spring discharges at their edges and pluvial flooding rather
than from streams bursting their banks. The regolith of zone 1 is considered to
have a lower hydraulic conductivity than other zones due to the presence of the
clays (this was also reported by McCartney and Neal (1999) following slug testing
at dambos in Zimbabwe). Riverbank sections were observed to be homogenous,
therefore, an additional soil layer above the regolith is not incorporated. The basalt
bedrock of zone 1 may have a higher hydraulic conductivity than other zones
because an area of preferential leaching leading to the formation of the dambo
must have greater fracturing or a higher content of more permeable materials,
however, such fracturing may be filled with clays from the enhanced weathering
once again reducing the hydraulic conductivity.
2. The second and most abundant zone comprises all areas that are neither
floodplains (zone 1) nor hillslopes (zone 3). The zone is characterised by low
rolling foothills with shallow slopes comprising the highest population density
and greatest agricultural intensity. Expanses of zone 2 may be split by zone 1
floodplains or such floodplains may sit or “hang” within higher areas of zone 2
thus being both up and down gradient of zone 2. Numerous small sub-basins are
formed within the low hills of zone 2 generally draining to a zone 1 feature.
Hydraulic conductivity of the zone 2 regolith has been determined from well
124
pumping and recovery tests and is considered higher than in zone 1 due to a lower
clay content and less settlement, which would close pores and fractures. The basalt
bedrock of zone 2 may be of lower hydraulic conductivity than zone 1 because
differential leaching has not occurred in these areas.
3. The third zone comprises the upland areas. This zone is characterised by higher
relief and ephemeral streams, often deeply incised into thicker regolith. While
agriculture may be present, more common is scrub-like vegetation. Dwellings are
fewer as population density is lower. Wells are deeper and generally poor
providers of water during the dry season. Zone 3 typically forms the perimeter of
the catchments. Well pumping and recovery tests were not conducted in areas that
fit the zone 3 definition, however, discussions with local communities suggests
that regolith hydraulic conductivity is less in zone 3 as wells recover slower than
elsewhere. The basalt bedrock is considered similar to that of zone 2.
Delineation of the zones was conducted by analysis of ground-truthed Google.Earth

imagery and slope analysis using QGIS. The zone 1 floodplains are simple to visually
identify on Google.Earth though ground-truthing was conducted during the field visits to
confirm the ease of identification and delineation. A sample area of ground-truthed
floodplains were digitised by hand by drawing polygons on Google.Earth then importing
into QGIS. Using the “slope” tool within the terrain analysis toolbox and applying it to
the DEM, the gradient thresholds were varied until the delineated zones matched the
digitised floodplains. The same methodology was used to define the zone 3 hillslope
areas. The gradients selected to delineate the three zones are presented in Table 7-4. The
distribution of the zones is shown in Figure 7-5.
Table 7-4. The slope gradients used to delineate the three zones using ArcGIS analysis of the SRTM
DEM, and, based on field measurements, the mean well depths used to determine the base of the regolith,
and the regolith aquifer properties.
Zone Gradient Mean well depth (m) Mean hydraulic Mean specific yield
conductivity (m/d) (-)
1 0 – 1.4o 6.64 7.5 0.09
2 1.4 – 6 o
7.59 2.1 0.09
3 > 6o 12.00 1.1 0.09
125
Layer thicknesses
As described above, the models were constructed with three layers: soil, regolith and
basalt bedrock. Soil layer thickness was varied during modelling from 0 to 2.5 m in
accordance with sections that could occasionally be observed in riverbanks and well
bores. The thickness of the regolith layer was determined from well surveys. Wells are
typically excavated until manual excavation is no longer possible due to the presence of
strong rock, therefore, well depth is inferred to be the base of the regolith layer, which is
the top of the basalt bedrock. Demis Alamirew of GSE surveyed 143 hand-dug wells
during the AMGRAF catalyst period in February and March 2014 and I surveyed 64 wells
during the field visits in March/April 2015 and October/November 2015. Where wells
had been re-visited only the most recent measurements of depth were considered for
inference of the regolith thickness. Access to the well bore below hand pumps and, with
some exceptions, rope-and-washer pumps was not possible, therefore, recorded
measurements were often provided word-of-mouth. Such word-of-mouth measurements
have not been accepted for analysis of regolith depth because when such measurements
could be tested, large discrepancies were common. Typically, this is due to the traditional
local measuring unit of the cubit simply being doubled to provide the depth in metres or
being translated directly into metres rather than cubits. Such measurements are also reliant
on good memory as the particular well may have been excavated and sealed many years
previously. Following exclusions, 80 wells could be used for estimating regolith
thickness. The average well depths per zone are shown in Table 7-4; these depths are
analogous to the top of the basalt bedrock.
The underlying basalt is 500-3000 m thick (Mohr, 1983). Only the upper part of this
sequence was incorporated in the models; the part which is considered to interact with the
shallow aquifer where fracturing and weathering are enhanced. Basalt layer thickness was
varied during modelling from 0 to 10 m with zero corresponding to a completely
impermeable bedrock and 10 m being the likely maximum limit of hydrogeological
interaction with the regolith aquifer. As the depth of fracturing of the basalt bedrock was
not observable, 10 m was selected based on discussion with local expert opinion (Demis
Alamirew, GSE hydrogeologist, personal communication, 17th March 2015); ultimately
this value was reduced during the modelling process.
126
Aquifer properties
Pumping tests were undertaken during the first and second field visits as described in
Chapter 3 and in Walker (2016). Hydraulic conductivity estimates averaged 2.29 m/d in
the dry season and 9.65 m/d in the wet season when the water-table was higher. This
difference indicates vertical heterogeneity of the regolith with more transmissive layers
occurring at shallower depth that are only intercepted when the water column is higher.
Specific yield estimations have a wider range and are more uncertain with an average of
0.09.
The tested wells were categorised into the three hydrogeological zones, according to slope
gradient, land use and observable geology, then the mean hydraulic conductivity,
considering both dry and wet season tests, was determined. The specified hydraulic
conductivity and storage values for each zone are presented in Table 7-4. It is noted that
most of the wells lie in zone 1 from a point of view of slope gradient though they are
situated immediately on the edge of floodplains rather than within them. The calculated
hydraulic conductivity for these zone 1 wells is higher than for zone 2 and 3 wells even
though the floodplain regolith is considered to be clay-rich and of lower permeability. It
has been reported that the floodplain or dambo edges are typically sandy and quite
permeable (McFarlane, 1989) which may explain these high values from the pumping
tests. The hydraulic conductivity values are not considered representative of zone 1 as a
whole and were reduced for the modelling. The same specific yield value was used for
regolith in all zones due to the greater uncertainty in its estimation.
7.2.5 Land use land cover (LULC)

LULC was divided into three zones matching the hydrogeological zones described
previously. Zone 1 is characterised as grassland as the floodplains are almost always and
entirely utilised as pasture. Zone 2 is categorised as arable because the majority comprises
land devoted to rainfed agriculture. The crops planted are 87% cereals (39% maize, 35%
teff and 13% millet), with the rest being pulses, oilseeds, sugarcane, potatoes, vegetables,
fruits, onions, garlic, and tomatoes (Belay and Bewket, 2013), the latter few generally
occupying backyard plots. In addition to these crops within zone 2, the vegetation actually
ranges from grassland to Eucalyptus plantations with some areas of bare ground near
areas of habitation. Zone 3 with its higher gradients is categorised as shrub due to the
characteristic scrub-like vegetation, though in reality the hillslopes also contain lesser
amounts of crops, grassland and forest, with dense native forest occurring around
churches. How SHETRAN converts the potential evapotranspiration time series into
127
actual evapotranspiration is specific for each of these zones. Each zone within SHETRAN
differs in rooting depth and root density at different depths, ground coverage at maximum
seasonal extent, canopy storage capacity, leaf area index, and AET/PET at particular soil
moisture tensions. These parameter values were estimated from ground observations in
order to generalise vegetation and crop types followed by consultation of the key
instructional texts for calculating water demand; FAO24 (Doorenbos and Pruitt, 1975)
and FAO56 (Allen et al., 1998), and other published studies providing detail of particular
vegetation types, particularly, Canadell et al. (1996); Dardanelli et al. (1997); Cain (1998)
and; Fan et al. (2016).
7.3 Model calibration
7.3.1 Introduction
Theoretically, all parameters could have been measured in the field, though in reality, this
is rarely possible due to cost, time and experimental constraints, as well as problems of
scaling (Beven et al., 1980). Therefore, some calibration of parameters is required to
provide confidence that the catchment model is representing reality. Calibration of
physically-based models can be complex and expensive/time-consuming due to
sophisticated model structures, computation requirements and the large number of
parameters (Blasone et al., 2007; Zhang et al., 2013). Manual calibration requires
rigorous and purposeful adjustment of parameter values, is extremely time-consuming,
tedious and can be subjective (Refsgaard, 1997). However, this time is well spent as, in
addition to achieving satisfactory calibration statistics, hydrogeological understanding is
greatly increased and the conceptual model can be updated and validated.
7.3.2 Calibration data

Datasets from two sources were used to quantitatively calibrate the models:
1. The Dangesheta community-based monitoring programme

a. Groundwater levels from five wells
b. River stage for the Brante and Kilti rivers, converted to river flow
2. Formal data from the Ministry of Water, Irrigation and Electricity (MoWIE)
a. River flow records for the Kilti and Amen rivers
The community-based monitoring data was discussed in Chapter 5 and Walker et al.
(2016). The time series used span 10th March 2014 to 8th January 2017. River flow records
are available from MoWIE for the Kilti and Amen rivers from 17th April 1997 to
128
4th October 2014 and 26th April 1988 (though are continuous only from 1998) to 27th
September 2014 respectively.
In addition to the quantitative calibration of modelled to observed groundwater level time

series, a qualitative groundwater calibration took place in other areas of the catchments.
The groundwater data used were occasional groundwater level measurements taken
around the catchments on multiple field visits, and anecdotal or observational evidence
of areas prone to flooding in the wet season or of dry wells in the dry season.
An issue with calibrating modelled against observed discharge is that the Brante and
Amen rivers were observed to have ephemeral reaches. During dry season field visits
when there was flow at the gauge sites, reaches both upstream and downstream of the
gauges were observed to be dry. Generally, the rivers appeared to be losing through the
flat floodplains where the rivers were dry and gaining through the narrow basalt riverbed
flowing reaches. Incidentally, radon-222 measurements during the wet season field visit
suggested that this pattern is reversed in the wet season with groundwater discharge in
the floodplains (see Chapter 4).
Concerning calibrating modelled to observed groundwater levels, the individual well

responses are often controlled by features unique to those wells. For example, rapid rises
in water level of monitoring well MW5 are due to overland flow directly entering the well
via termite and rat holes (such information was proffered during community workshops)
and MW3 has no lid meaning direct precipitation will influence the groundwater
hydrograph. Some features that are hydrogeologically significant, such as MW4
maintaining its groundwater level longer into the dry season due to its position directly at
the foot of a slope and recharge area, are not represented at the resolution of the models;
even at the 100 m resolution of the finest models. Given that the monitoring wells often
dry out in the dry season, it is known that the water table is below the well base at their
locations but the precise water table depth is uncertain. In addition, the wells are subject
to abstraction for domestic and agricultural use (water for livestock and backyard
irrigation). Abstracted volumes may be low in total and have minimal effect on the overall
well response but the timing of abstraction could be immediately prior to measuring, i.e.
first thing in the morning, therefore some measurements may not be showing natural
aquifer response to recharge and discharge. What’s more, the monitoring wells are
clustered in a small area of around 0.5 km2, though their positions were selected to be
representative of the wider area.
129
7.3.3 Calibration methodology
The five community monitored wells are within the Brante catchment that is within the
larger Kilti catchment. However, only the Brante model is of sufficiently fine resolution
that all the wells lie in different cells. Therefore, the Brante model was chiefly calibrated
against groundwater level data, as well as against the community river flow data. At the
same time, the Amen and Kilti models were calibrated against their longer period river
flow records. A set of optimum parameters was selected iteratively that satisfied the
groundwater level calibration of the Brante and the river flow calibration of the Amen
and Kilti rivers. This was while also considering the river flow of the Brante and smaller
Kilti model and the qualitative groundwater level observations. The search for matching
optimum parameters to suit all four modelled catchments is justified due to the nested and
overlapping nature of the catchments in addition to the observed similarities in soils,
geology and LULC.
The calibration parameters are those identified from literature to be sensitive in

SHETRAN modelling (e.g. Bathurst et al. (2004); Bathurst et al. (2011a); Starkey et al.
(2017)) in addition to parameters considered uncertain for this study site:
 The Strickler coefficient is the inverse of the Manning roughness coefficient and
controls the speed of runoff. It is essentially a friction factor representing raindrop
impact, flow channelization, obstacles such as rocks and vegetation, surface
frictional drag, and erosion and transport of sediment (Engman, 1986). The
Strickler coefficient value was initially selected based on literature review of
modelling studies in similar environments. The adjusted range was from 0.5 to
10.0 m-1/3/s.
 The AE/PE ratio controls the amount of evaporative loss. The values were
adjusted with consideration of the SHETRAN mass balance output in order to
achieve satisfactory long-term discharge totals. The ratios reduce with depth and
were adjusted between 0.1 and 1.0.
 Hydraulic conductivity, specific yield and specific storage values were adjusted
within the range estimated from the pumping tests and based on hydrogeological
experience and textbook values in the case of soils and basalt (e.g. Fredlund et al.
(1993); Fetter (2001)). The adjusted ranges for hydraulic conductivity were: soil
10-100 m/d, regolith 0.2-22.3 m/d, and basalt 1 x 10-5 to 1 x 10-2 m/d. for specific
yield the adjusted ranges were: soil 0.05-0.4 and regolith 0.03-0.32. Specific
storage of basalt was adjusted between 1 x 10-3 to 1 x 10-6 m-1.
130
 Soil and geological layer thicknesses were adjusted within the range observed in
the field: soil 0-3 m, regolith 3-15 m and basalt 0-10 m.
Calibration and validation periods were selected to give “typical” ranges of hydrological
conditions and ran from the end of a wet season recession to the same point one, two or
three years later. In addition to visual comparison of plotted observed and simulated data,
the following performance indicators were utilised: Nash-Sutcliffe Efficiency (NSE)
coefficient (Nash and Sutcliffe, 1970), where a value greater than 0.5 is considered
acceptable (Moriasi et al., 2007), and root mean square error (RMSE), with units
matching the compared data thus the value should be as low as possible. NSE is very
sensitive to peak flows (Krause et al., 2005), therefore, given the flashy nature of the
rivers with short-lived and relatively extremely high peaks, NSE (and RMSE) were
calculated on baseflow following hydrograph separation. For the Brante model, NSE and
RMSE were calculated on groundwater levels in the five monitoring wells, excluding the
periods when the wells were dry as during this time the piezometric surface was at an
unknown level below the well base. A validation period was run to confirm that the
calibrated parameters still produced a satisfactory simulation for independent input
datasets. However, it is inappropriate to calibrate the models purely against absolute
values of the observed river flow and groundwater level time series using standard
statistical techniques such as NSE and RMSE because high skill will not necessarily mean
the model is representative of the natural system and vice versa. The purpose of the
modelling is to assess availability of groundwater for productive use outside of the wet
season, therefore, a realistic representation of groundwater seasonal response is deemed
the most important process to simulate, rather than precise and entire groundwater level
time series at five specific points. Simulated river flow was also assessed by comparing
cumulative monthly flow totals using the full model run rather than calibration and
validation periods). Monthly rather than daily flow totals are used for calibration in order
to negate the aforementioned effects of temporarily dry reaches and flashy peaks.
Achieving good daily matches between observed and simulated flows was not
particularly important in this study; rather, understanding of the long-term water balance
was critical. Comparing simulated to observed river flow purely using NSE proved an
unsatisfactory calibration methodology as a good NSE could be achieved even when
annual flow totals or cumulative monthly flow totals were mismatched.
131
7.3.4 Calibration results
The groundwater level calibration of the Brante model gave acceptable NSE values,
however, RMSE was quite large (Table 7-5). This is suspected to be due to the varying
well depths indicating heterogeneous aquifer thickness while the cells in which the wells
lie have uniform aquifer thickness. Consequently, the model was simulating greater
groundwater fluctuation where the aquifer is specified thicker than the wells indicate is
the reality and vice versa (Figure 7-7). Correct simulation of the groundwater behaviour
was the main calibration criterion for the model. Sample groundwater hydrographs are
shown in Figure 7-7 that are representative of the three zones: Zone 1 floodplains areas
known to have shallow water tables, flood during the wet season though often dry out in
the dry season, Zone 2 cultivated and inhabited interfluve areas with groundwater
response somewhere between Zones 1 and 3, and Zone 3 hilly areas observed to have
deep water tables and known anecdotally to suffer water scarcity in the dry season. These
observed and anecdotal responses can be seen in the simulated groundwater hydrographs,
therefore, calibration was deemed satisfactory. The difference in the observed and
simulated hydrograph for monitoring well MW3 in Figure 7-7 is due to the model output
being the groundwater fluctuation of a 100 m x 100 m cell whereas the observed
hydrograph is at a point with unique soil and geology parameters, particularly regarding
layer thicknesses and elevations. What’s more, the particular well is located just beyond
the edge of a floodplain whereas the simulated output represents the floodplain itself. It
is common at the study site for backyard agriculture plots of <0.1 ha to have multiple
wells all with different depths and water levels, therefore, it was unsurprising that a better
match could not be achieved between the simulated and observed groundwater
hydrographs. The further semi-quantitative groundwater level calibration was also
considered satisfactory for the Amen, Kilti and Kilti-sub models as SHETRAN correctly
simulated Zone 1 floodplain areas (shallow water tables and wet season floods), and zone
3 hilly areas (deep water tables and dry season water scarcity).
132
Table 7-5. Details and statistics of the calibration and validation periods.
Catchment Calibration period No. of days NSE RMSE

Brante* Mar 2014 to Mar 2015 (year 1) 365 0.69 2.01 m
Amen Apr 1999 to Apr 2001 (years 2-3) 731 0.79 0.19 m3/s
Kilti Apr 1998 to Apr 2000 (years 2-3) 731 0.78 1.47 m3/s
Kilti-sub Mar 2015 to Mar 2016 (year 2) 365 0.64 0.27 m3/s
Catchment Validation period No. of days NSE RMSE

Brante* Mar 2015 to Mar 2016 (year 2) 365 0.53 2.08 m
Amen Mar 2010 to Mar 2013 (years 13-15) 1096 0.75 0.13 m3/s
Kilti Apr 2004 to Apr 2007 (years 8-10) 1096 0.67 2.30 m3/s
Kilti-sub Jan 2016 to Jan 2017 (year 3) 365 0.08 1.17 m3/s
* Calibration is against groundwater levels rather than river flow.
0
Groundwater level (m bgl)
2
Zone 1
8 MW3 well base

Observed MW3
10 Simulated MW3
12 Simulated
Simulated
14
0
Groundwater level (mbgl)
2 Zone 2
4
10 Simulated
Simulated
12
Simulated
14
0
Groundwater level (mbgl)
2 Zone 3
4
10 Simulated
12 Simulated
Simulated
14
2014 2015 2016
Figure 7-7. Simulated groundwater hydrographs from the Brante model showing three typical and
representative hydrographs for each zone. Top: Representing Zone 1 floodplains and showing observed
groundwater hydrograph for monitoring well MW3. Middle: Representing Zone 2 inhabited and
cultivated foothills. Bottom: Representing Zone 3 hilly areas. It is restated that the monitoring wells all lie
within Zone 1 cells, therefore, observed hydrographs cannot be included in the Zones 2 and 3 plots.
133
The observed and simulated river flow hydrographs match well for the Amen and Kilti-
Durbete models (Figure 7-8) and both have good NSE and RMSE values for the
calibration and validation periods (Table 7-5). More importantly, the simulated
cumulative flow totals match well with observed (Figure 7-9). Using the optimum
calibrated parameter values in the Kilti-Dangesheta model gave good calibration statistics
for the calibration period but not for the validation period. However, the cumulative flow
totals comparison is good, as it is for the Brante. Therefore, the calibration was considered
satisfactory and confidence was given in the uniqueness of the optimum parameter values
as a satisfactory calibration was achieved for all the nested catchments utilising the same
values.
Figure 7-8. Simulated and observed river flow hydrographs.
134
Figure 7-9. Simulated and observed cumulative monthly flow totals.
7.3.5 Optimum parameter values

The optimum parameter values are considered those that gave satisfactory calibration of
both groundwater levels in the short duration (3-year) Brante model and river flows in the
longer duration (17 and 18-year) Amen and Kilti-Durbete models. Further validation
occurred when satisfactory calibration statistics and simulated vs observed plots were
achieved from running the set of optimum parameter values in the Kilti-Dangesheta
model. Calibration occurred for all models concurrently; therefore, while slight
adjustments to parameter values could give slightly better individual calibrations, the
differences were minor. The optimum parameter values are presented in Table 7-6. The
aquifer properties remain similar to the observed and measured values, and to literature
values in the case of basalt (e.g. Fetter (2001)). Strickler coefficients are at the low end
of literature values (e.g. Engman (1986)) reflecting the hummocky tussocky floodplains,
the rough surfaces produced by ox ploughing in small plots that are not conducive to
forming rills, and the often dense underbrush of the shrub (Figure 7-8). The relatively
high AE/PE for the arable zone reflects the high proportion of tall crops such as maize
and sugar cane that have high water demand, i.e. PET would be greater than the grass
135
reference value calculated using the Penman-Monteith FAO56 method (Doorenbos and
Pruitt, 1975; Allen et al., 1998).
Figure 7-10. Photographs of hummocky tussocky floodplain (left) and rough surface left by ox ploughing.
Table 7-6. Optimum calibrated parameter values for the SHETRAN models.
Zone Layer / vegetation Parameter Optimum value

1 Regolith Hydraulic conductivity 0.5 m/d
Floodplains Specific yield 0.08
Depth 6.6 mbgl
Basalt Hydraulic conductivity 0.0003 m/d
Specific storage 0.001 m-1
Depth 8.6 mbgl
Grassland Strickler coefficient 1.0 m-1/3/s
AE/PE 1.0 (at soil moisture tension of -0.1 m)
0.85 (-1.0 m)
0.65 (-10.0 m)
0.45 (-20 m)
0.25 (-50 m)
2 Soil Hydraulic conductivity 20 m/d
Cultivated Specific yield 0.1
and Depth 0.5 mbgl
populated Regolith Hydraulic conductivity 0.25 m/d
areas Specific yield 0.08
Depth 11.0 mbgl
Depth 13.0 mbgl
Arable Strickler coefficient 1.5 m-1/3/s
0.9 (-1.0 m)
0.8 (-10.0 m)
0.6 (-20 m)
0.4 (-50 m)
136
3 Soil Hydraulic conductivity 20 m/d
Upland areas Specific yield 0.1
Depth 0.25 mbgl
Regolith Hydraulic conductivity 0.5 m/d
Specific yield 0.08
Depth 12 mbgl
Depth 12.5 mbgl
Shrub Strickler coefficient 1.0 m-1/3/s
1.0 (-1.0 m)
0.8 (-10.0 m)
0.6 (-20 m)
0.4 (-50 m)
7.4 Improved hydrogeological understanding from the calibration
7.4.1 Sensitivity to certain parameters

Adjustment of geological layering and of parameter values during calibration not only
informed and confirmed the conceptual model but also was essentially a sensitivity
analysis. The parameter to which the models were most sensitive was layer thickness with
a few metres adjustment to soil, regolith or basalt depth leading to large changes in
groundwater and river flow hydrographs. Therefore, the likely heterogeneous thickness
of the soil and regolith layers, as indicated by variations in well depth, has significant
impacts on local groundwater flow and river level. It would be extremely difficult to
simulate these very local impacts accurately; it would require very high resolution
modelling following widespread geophysical investigations to three-dimensionally map
the layers.
7.4.2 Absence of a clay-rich low hydraulic conductivity layer

The key regolith literature (Jones (1985) or Acworth (1987)) place a clay-rich low-K layer
within the regolith profile below the soil (zones ‘a’ and ‘b’ of Figure 3-9). This layer was
simulated in the model runs, placed between the soil and regolith at thicknesses of 0.5-
1.0 m with hydraulic conductivity of 0.001 to 0.01 m/d. However, the model repeatedly
crashed due to excessive river discharges and drying out of aquifer cells, i.e. rainfall could
not sufficiently infiltrate. This suggests that such a low-K layer is absent at this depth, or
at least there are preferential flow pathways to the aquifer. Observed storm peaks in
discharge are not at such high levels that suggest infiltration does not occur during intense
137
rainfall events. The regolith has a high clay content throughout and the absence of a layer
of particularly high clay and consequent very low hydraulic conductivity may be due to
the relatively young age of the bedrock (Benvenuti et al., 2002).
7.4.3 Absence of fractured bedrock high hydraulic conductivity layer

The key regolith literature also describe a fractured bedrock high-K layer at the base of
the regolith profile (zone ‘d’ of Figure 3-9). Inclusion or exclusion of fractured basalt as
a thin (<1 m) high hydraulic conductivity (20-75 m/d) layer between the regolith and
basalt does not have a significant impact on the model results. When the fractured layer
was ascribed a hydraulic conductivity >40 m/d, the model would only run when the basal
basalt hydraulic conductivity was significantly reduced, and it was this adjustment to the
basal basalt that had a greater impact on the simulation. The relatively young age of the
basalt bedrock may mean that significant fracturing has not had time to occur.
7.4.4 Hydrogeological importance of the basal basalt bedrock

The greatest impact on simulated river flows and groundwater levels resulted from
adjustment of basal basalt hydraulic conductivity. Decreasing the hydraulic conductivity
increases the wet season baseflow significantly while decreasing the specific storage and
porosity decreases recession length. Prior to the modelling, it was suspected that the basalt
bedrock was impermeable, therefore, no basalt was simulated in the model and the base
of the regolith formed the base of the model. However, the simulations gave
unsatisfactory river flow and groundwater levels. To achieve a better simulation, some
permeability had to be applied to a basalt basal layer. The thickness of this layer has less
of an influence than the hydraulic conductivity. When the layer was simulated to be thick
(>10 m) the model would not run, even when transmissivity was equal to a thinner layer,
suggesting that underlying the regolith of the catchments is a thin, slightly weathered and
fissured, basalt layer above impermeable basalt bedrock.
7.4.5 Absence of leakage to a deeper aquifer

The conceptual model was of a shallow thin perched aquifer with possible leakage to a
deeper aquifer. That deeper aquifer is likely to comprise fractured and scoriaceous zones
within the deeper basalts and trachytes (Kebede, 2013). Leakage to a deeper aquifer was
simulated by allowing gravity drainage through the base of the model. This was initially
applied only below the zone 1 floodplains because the dambo literature suggests more
fractured bedrock may be present there (McFarlane, 1989). Allowing gravity drainage
had little effect on the simulations until the zone 1 basalt hydraulic conductivity was
increased to allow greater groundwater flow out of the model. It was clear from the
138
hydrographs that the adjustment of the basalt hydraulic conductivity had a greater effect
on the simulation than leakage as groundwater flowed laterally into streams more rapidly.
Therefore, the modelling suggests that leakage to a deeper aquifer is not significant at
these scales; a conclusion supported by field evidence described in Chapter 4.
7.4.6 Evapotranspiration from groundwater

The heavy wet season rains lead to very shallow water tables across much of the study
site for 4-6 months of the year. This was most apparent at the seasonally inundated
floodplains that were generally still too wet to cross on foot during the second field visit
in October 2015 a month or so after significant rains had ceased. Springs and seepages
remain active for much of the year and the shallow water tables support phreatophytes.
Consequently, it was suspected that the evapotranspiration loss from the saturated zone
would be significant. This was confirmed by SHETRAN modelling during the recharge
assessment method comparison presented in the previous Chapter.
7.5 Mapping the potential of shallow groundwater for irrigation use
7.5.1 Methodology
Once the models were considered to be simulating natural conditions well, analysis could
be conducted on the model outputs. Because SHETRAN is fully spatially distributed,
variations in groundwater level across the catchments could reveal areas that have the
best potential for exploitation of the shallow groundwater resource for irrigation (Figure
7-11). Such areas could be identified by slow groundwater recessions at the end of the
wet season and an abstractable saturated thickness of aquifer throughout the dry season.
Figure 7-11. Maps showing The Brante catchment and SHETRAN model outputs of water table depth at
the peak of the wet and dry seasons and the transition between the two seasons.
139
Post-processing was conducted using a Python script that analysed the groundwater level
in every cell during every day of a simulation. The script counted the number of days per
hydrological year that groundwater was available for abstraction in each cell. This was
defined as a groundwater level >0.5 m above the base of the regolith, i.e. the groundwater
level was not in the basal basalt layer below the base of wells and a sufficient water
column was present to allow pumping by mechanical or motorised means. The annual
counts of groundwater availability were averaged for the full model runs and groundwater
potential zones were assigned as follows:
High: <10 days per year when groundwater is unavailable – Irrigation is possible
year-round.
Medium: 10-100 days per year of groundwater unavailability – Irrigation is

possible for a second growing season in addition to the main rainfed growing
season.
Low: 100-200 days per year when groundwater is unavailable – Irrigation is

unlikely to be possible outside of the wet season.
Very low: >200 days per year when groundwater is unavailable – Irrigation
outside of wet season impossible.
It may seem that the criteria are strict as very low potential areas may still have available
groundwater for 165 days of the year. However, this period of groundwater availability
would obviously coincide with the wet season when no irrigation is required due to the
rains.
7.5.2 Groundwater potential maps

The maps of potential for groundwater abstraction are presented in Figure 7-10. The
patterns of zoning make most sense when the maps are overlain on Google.Earth and the
viewing angle, or tilt, and transparency are adjusted to enable simultaneous viewing of
topography and land use.
Figure 7-13, Figure 7-14 and Figure 7-15 show examples of the commonly observed
relationships between the groundwater potential zoning, topography and land use. The
high potential areas are found at the foot of hillslopes and in narrow valleys. Medium
potential zones typically surround areas of high potential or are themselves surrounded
by areas of low potential. Low-lying land and floodplains are typically areas of low
potential and small hills in their midst are often medium potential. Very low potential
140
zones are mountainous areas and ridges, typically around the catchment boundaries.
Commonly, the low potential areas clearly appear lighter in colour and drier on
Google.Earth while medium and higher potential areas appear darker. These observations
were ground-truthed during the January 2017 field visit. At that time of year, only
Eucalyptus is cultivated and plantations were noticeably less water-stressed in the areas
of high potential than those in low potential zones (Figure 7-16).
141
Kilti-Durbete
Kilti-Dangesheta
Brante
Amen
2 km
2 km
2 km
5 km
Figure 7-12. Maps showing potential of shallow groundwater for irrigation use.
142
Figure 7-13. Southern extremes of the Brante and Amen groundwater potential maps overlain on
Google.Earth.
Figure 7-14. Central portion of the Brante and Amen groundwater potential maps overlain on
Google.Earth.
143
Figure 7-15. Northern portions of the Kilti-Dangesheta, Amen and Brante groundwater potential maps
overlain on Google.Earth. A and B are the locations of the photographs in Figure 7-16.
Figure 7-16. Photographs taken during the dry season (January 2017) showing water-stressed Eucalyptus
in a low potential zone (left) and healthier Eucalyptus in a high potential zone at the foot of a slope. See
Figure 7-15 for the photograph locations.
Production of the groundwater potential maps revealed that the 500 m resolution of the
Kilti-Durbete model was too coarse to correctly locate the different potential zones. There
was agreement in potential zone correspondence to topography and land use for the other
three models and general agreement in the overlapping portions of the Kilti-Dangesheta
and Amen models. However, the Kilti-Durbete model could not identify the high potential
zone narrow valleys and thin strips at the foot of slopes, nor the very low potential ridges
144
around catchment boundaries. Figure 7-17 compares the potential maps produced from
the different catchment models illustrating that the Kilti-Durbete model gives contrasting
results of high potential in low-lying and floodplain areas. Slight variations in topography
that have been revealed the controlling factor on groundwater availability cannot be
picked up at coarse resolution. Therefore, no confidence could be given to the Kilti-
Durbete potential map for advising on most suitable well locations for productive use.
The 200 m resolution Kilti-Dangesheta potential map appears to perform well, though the
100 m resolution Amen and Brante maps are preferable for making intervention
recommendations given the small-scale agriculture of typically <1 ha plot size that
dominates the area. These maps are of use should a government ministry, private investor,
NGO, or other organisation wish to implement irrigation in the area; resources could be
focussed in the zones with highest potential. The groundwater potential maps were
provided to relevant stakeholders during the January 2017 field visit; including the
Dangesheta kebele office, the Dangila woreda office, the Abay River Basin Authority
(ARBA) in Bahir Dar and the Ministry of Agriculture and Natural Resources (MoANR)
in Addis Ababa. The response was positive and a follow up research project is underway
with these partners with workshops planned for May 2018 where the maps will be further
presented and described.
Figure 7-17. All four catchments’ groundwater potential maps overlain on Google.Earth (left) and the
same view but showing only the Kilti-Durbete potential map.
7.6 Discussion and conclusions

SHETRAN was chosen to model the shallow aquifer due to its comprehensive nature,
ability to model coupled surface and subsurface, saturated and unsaturated components,
and because the spatially distributed outputs would aid in understanding the behaviour of
the hydrogeological system in all parts of the catchments.
145
Four catchments were modelled at different resolutions, from 100 m to 500 m, and their
nested nature meant one set of parameter values should satisfy calibration for all the
models. Calibration was chiefly against groundwater levels for the Brante model and river
flows for the Kilti-Durbete, Amen and Kilti-Dangesheta models, though with qualitative
calibration of groundwater levels for all models.
The initial aim of the modelling, to increase hydrogeological understanding of the shallow
groundwater system, was certainly achieved. Referring back to the specific research
questions posed in Section 7.1.2: Are the models satisfactorily reflecting reality in terms
of river flows and groundwater levels?
Calibration was satisfactory in terms of the long-term behaviour and water balance.
However, short-term (daily) response was not necessarily picked up and groundwater
level calibration was problematic due to the point locations of the observational time
series vs cell-wide simulated outputs that could not incorporate monitoring well
specificities such as: very local elevation and topography, geological heterogeneities (soil
and regolith layer thicknesses, hydraulic conductivities and specific yields), and direct
water ingress via precipitation and overland flow. This is a generic problem in spatial
models representing point measurements. The models did produce the general range of
observed responses and were considered valid for the scenarios analysis presented in
Chapter 8.
Again from 7.1.2: Are the parameter values appropriate considering field investigations and
published literature?
The calibration procedure and sensitivity testing adjusted parameters within either
measured ranges or, where impossible to measure, within published ranges. The optimum
values were similar to those measured and were deemed appropriate. Model outputs are
sensitive to geological layer thicknesses. The few exposures available in well bores and
riverbanks, and the depths to bedrock inferred from well depth measurements, suggest
layer thicknesses are heterogeneous even within short distances. Therefore, the layer
thicknesses are ascribed in the models with uncertainty, which further explains the
problematic groundwater level calibration given the sensitivity to this parameter.
From 7.1.2: What effect does incorporating various hydrogeological features presented in
key regolith and dambo literature, but not necessarily identified in the field, have on the
simulations?
146
The calibration and sensitivity testing process allowed update and confirmation of the
conceptual model, revealing a lack of a clay-rich low permeability layer beneath the
floodplains and lack of a high permeability fractured layer at the top of the basalt.
However, there is evidence supporting a permeable horizon, although not with very high
permeability that may be associated with fractured basalts; seepage through this layer to
a deeper aquifer is minimal. Likelihood of the presence of such features was determined
by their improvement or worsening impact on, in terms of comparisons between
simulated and observed, groundwater levels and river flows. The use of SHETRAN to
provide recharge estimates as described in Chapter 6 confirmed that evapotranspiration
from the unsaturated zone is significant.
From 7.1.2: Is groundwater availability recharge controlled or storage controlled?
Assessment of the spatiotemporal variability of recharge and storage indicated that

shallow groundwater availability is storage rather than recharge controlled. This was
unsurprising given the high recharge estimates presented in Chapter 6 and the thinness of
the aquifer as described in Chapter 4.
The final objective, to identify areas that show the greatest potential for the onset of
irrigation using shallow groundwater, was achieved with production of the groundwater
potential maps. Maps of the potential to utilise shallow groundwater for irrigation were
produced by processing the SHETRAN water table depth outputs using Python, which
were then overlain on Google.Earth. Areas described as being of high potential were
where groundwater was on average available year-round, allowing just 10 days of
unavailability. Such zones were typically found at the bases of hillslopes and in narrow
valleys, corresponding to around 17% of the study site. Medium potential zones had 10-
100 days of unavailability and were found surrounding areas of high potential or formed
“islands” of higher relief areas within low-lying floodplains (corresponding to
approximately 31% of the study site). Low potential areas (100-200 days of groundwater
unavailability) comprised much of the low-lying land including the floodplains (~46% of
the study site) while very low potential zones (>200 days of unavailability) were located
in mountainous areas and ridges (~6% of the study site). These maps could be utilised for
intervention planning and were provided during the third field visit in January 2017 to
local, regional and national stakeholders, who showed interest and requested follow up
discussions and possible workshops (planned for May 2018 as part of a subsequent
research project).
147
While a complete SHETRAN modelling study would not be feasible to evaluate the
abstraction and irrigation potential for every new site, the topographic and geological
identifiers of high potential zones can be located during desk study analysis of
topographic and geological maps followed by ground truthing, e.g. geological
observations, well depth measurements and dry season well productivity assessments.
148
Chapter 8. Resilience of shallow groundwater
resources

Chapter 7 provided justification for the use of SHETRAN and described the model
construction, parameterisation and calibration processes. The models could next be used
to simulate potential future scenarios, which is the basis of Chapter 8. This chapter details
the likely future scenarios that may affect surface and shallow groundwater resources. A
multimethod drought analysis is presented revealing the historical climate variability in
the area. SHETRAN modelling of the potential future climate, land use and abstraction
scenarios is described and results are presented of the impacts on surface and groundwater
availability.
8.1 Introduction
8.1.1 Context
Groundwater is often considered a potential saviour for sub-Saharan Africa’s
unfortunately prevalent issues of food security, poverty and vulnerability to climate
change (Carter and Bevan, 2008; Calow et al., 2009b; Taylor et al., 2009; MacDonald et
al., 2012a). Rural communities, being poorer and more vulnerable, have the most to gain
from increased groundwater use, and these communities are most likely, or only have the
capacity, to exploit shallow groundwater. Reports are common of water table decline in
parts of the world like India, China and Pakistan where a rapid growth in abstraction for
small-scale irrigation, not only increased food security and alleviated poverty, but has
caused a worryingly drastic depletion of groundwater resources (Konikow and Kendy,
2005; Aeschbach-Hertig and Gleeson, 2012; Reshmidevi and Nagesh Kumar, 2014). It is
also well known qualitatively and empirically the likely effects of climate and land use
land cover (LULC) change on water resources, e.g. increased storm intensity and
urbanisation and/or cultivation encroaching on natural vegetation, all of which would
increase rapid runoff and decrease recharge and baseflow (Meyer and Turner, 1992;
Arnell, 2004; Liu et al., 2008b). However, few (if any) studies have been conducted at
community scales in sub-Saharan Africa to determine what will happen quantitatively to
the shallow groundwater resource if smallholders actually start abstracting it, and in the
face of climate variability and LULC change.
149
8.1.2 Purpose of modelling
Chapter 7 showed how the modelling aims were achieved, in terms of increasing
hydrogeological understanding of the shallow groundwater system, developing the
conceptual model, and identifying areas with the greatest potential for sustainable
intensification of agriculture through irrigation. The objective for the second phase of the
modelling was to answer research question 5 from Chapter 1: How will climate
variability, LULC change and increased abstraction impact the shallow groundwater
resource and surface water?
The groundwater potential maps give satisfactory guidance under current conditions in
terms of the recent climate, current land use and likely slow initiation of shallow
groundwater abstraction for irrigation. But going forward, climate variability is predicted
to become more extreme (Bates et al., 2008; Taye et al., 2015) and population pressures
will likely lead to reductions in natural vegetation cover (Hurni et al., 2005; Amsalu et
al., 2007) and increased groundwater abstraction (Awulachew et al., 2007; Evans et al.,
2012). It is important to understand how these changes will impact the surface and
groundwater resources. Further modelling of potential future scenarios was conducted to
determine the resilience of the shallow groundwater resource.
8.1.3 Methodology
The second part of the modelling study involved simulating the potential impacts of future
scenarios on surface and groundwater resources. As discussed in Chapter 2, the likely
causal changes are climate, land use, and abstraction. The initial step was to assess
existing climate variability by conducting a drought analysis of historical data. A
literature review was also conducted of trends identified in highland Ethiopia
meteorology and projected climate change. The aims of these climate variability
assessments were the derivation of synthetic rainfall and potential evapotranspiration
(PET) time series for incorporation into the SHETRAN models presented in Chapter 7.
Ground observations, remote sensing analysis and literature review of LULC change
trends determined LULC scenarios for incorporation into the SHETRAN models by
adjusting the original land cover/vegetation layer. Groundwater abstraction and irrigation
were simulated by SHETRAN at calculated rates in distributed locations. The simulations
were run for 200 years in order to observe a full range of impacts on surface and
groundwater. Variations in shallow groundwater and surface water availability were
analysed to assess the impacts of the potential future scenarios.
150
8.1.4 Similar modelling studies
Many studies have investigated hydrological response to climate and LULC change in
Ethiopia. Legesse et al. (2003), for the Lake Ziway area of central Ethiopia, used the
physically based distributed Precipitation Runoff Modelling System (PRMS) to simulate
predicted changes in rainfall, temperature and transition between cultivated/grazing land
and native forest. Despite problems of data scarcity and calibration of peak river flows,
suspected to be due to errors from extrapolation of ratings when flow was above the
gauge, their simulations saw changes of 20-30% in stream discharge. Koch et al. (2012)
conducted a study using the SWAT model, which is very widely used in Ethiopian studies
(see Chapter 7), evaluating LULC change against hydrological response in the northwest
Highlands of Ethiopia, approximately 120 km from Dangila. Poor input data (gaps in
hydrometeorological time series and a coarse DEM) meant the model could only achieve
an acceptable calibration with monthly river flows (Nash and Sutcliffe efficiency (NSE)
daily = 0.24, NSE monthly = 0.71). The authors also stated that SWAT struggled with the
strong seasonality, i.e. floods in the wet season and dry streams with only subsurface flow
in the dry season. Cultivated land had increased in proportional coverage from 53 to 70%,
largely at the expense of grasslands, shrubs and bushes and simulations showed higher
peak runoffs and decreased low flows, but it is stated that the results are unreliable and
further modelling should be conducted. These studies, and similar for various catchments
around Ethiopia giving similar results, have no explicit simulation of groundwater flow
nor its response to climate and LULC change. While there are groundwater modelling
studies evaluating formal abstraction on deep aquifers in Ethiopia (e.g. Ayenew et al.
(2008a); Asrie and Sebhat (2016); Kerebih and Keshari (2017)), no examples of
modelling studies have been found from anywhere in Africa simulating the effects of
potential increases or onset of shallow groundwater abstraction.
8.2 Identifying potential future scenarios
8.2.1 Climate variability

Taye et al. (2015) state that when an ENSO (El Niño Southern Oscillation) event is
followed by La Niña in the same year, there is a 67% chance of an extreme flood occurring
in the Upper Blue Nile region. Conversely, they found that 83% of ENSO events starting
in April–June cause droughts in the Upper Blue Nile region. Seleshi and Zanke (2004)
found only links between ENSO events with summer droughts. Seleshi and Zanke (2004)
demonstrated no trends in total annual rainfall, seasonal rainfall, or the number of rainy
days per year in the Ethiopia Highlands. Meze-Hausken (2004) similarly detected no
151
trends in rainfall data for the same region; however, interestingly, farmers' perceptions
indicate progressively unfavourable conditions over the previous 20-30 years. While
trends have not been identified in rainfall in the Upper Blue Nile Basin, the considerable
interannual rainfall and river flow variation is well reported, e.g. Conway (2000); Hurni
et al. (2005); Kebede et al. (2006). For example, annual rainfall totals measured at
weather stations in the area of the study site range from a minimum of 1185 mm/a to a
maximum of 2009 mm/a at Dangila (1987-2015) and from 1059 to 2043 mm/a at
Meshenti (1987-2013). The annual flow totals of the Kilti measured at the Durbete gauge
ranges from a minimum of 206 Mm3/a to a maximum of 396 Mm3/a (1997-2014), though
this range is an underestimate as there is missing data during very wet periods due to the
gauge being overtopped or damaged. The local climate variability was analysed further
and is discussed in later sections.
8.2.2 Climate change

Buontempo et al. (2014), using a regional climate model (RCM), predicted for this region
of Ethiopia by 2100: an increase in rainfall of 2 mm/d from September to May (the dry
season) and a decrease of 2 mm/d in June-August (wet season), in other words a decrease
in rainfall seasonality, while the IPCC (Intergovernmental panel on climate change)
predict an intensification of extreme rainfall events for Ethiopia as a whole (Bates et al.,
2008). Laprise et al. (2013), using combinations of GCMs and RCMs, project no change
in this region in January-March precipitation up to 2100, whereas for July-September
precipitation the various models give projections from -1 to +1 mm/d. Clearly, there is
significant uncertainty in climate change projections. This arises from underlying
uncertainties in the emissions scenarios used to force the climate models and in the
assumptions made of physical processes of climatic systems within GCMs (Bonsor et al.,
2010). Hulme et al. (2001) state that while GCMs simulate changes to African climate
due to increased greenhouse gas concentrations, two very important climate drivers either
are represented poorly in the case of the influence of ENSO or are not represented at all
in the case of LULC change. Bonsor et al. (2010) agree that the predictive capability of
GCMs in data-poor regions like Africa has significant limitations. They give the example
of the Sahel where some GCMs predict significant drying and others predict progressive
wetting. These uncertainties are evident at the smaller regional scale as Taye et al. (2015)
report various rainfall projections for the Upper Blue Nile Basin from studies using
GCMs, some of which predict a wetter climate after the 2050s while others predict drying.
Similarly, Mengistu and Sorteberg (2012) in modelling changes in streamflow in the Blue
Nile due to climate change could not give confident predictions because the 47
152
temperature and precipitation scenarios they applied from 19 GCMs disagreed in both the
strength and the direction of future precipitation changes.
There is general agreement that temperatures in Africa will rise with climate change (e.g.
New et al. (2001); Mitchell et al. (2004); Buontempo et al. (2014)). According to Hulme
et al. (2001), based on the median of seven GCM experiments, temperatures around the
continent will rise by around 1-2 oC by the 2080s with the B1-low scenario, or around
4-7 oC by the 2080s with the A2-high scenario. While these temperature increases will
certainly increase evapotranspiration, the uncertainties regarding changes to rainfall
described in the previous paragraph mean the effects of temperature on soil moisture and
water bodies are more problematic to forecast.
Initial ideas for the future climate change scenarios involved downscaling multiple GCM
and RCM outputs for the study site, or a simpler method of bias-correction of GCM/RCM
outputs for an observed period then applied to projected data. However, as described
above, the uncertainties in climate change prediction for the region and consequent
scepticism in the projections meant it was decided not to incorporate a climate change
scenario. What’s more, the observed climate variability shown in a later section is much
greater than the projected daily average increases/decreases due to climate change.
Simulating the observed extremes of climate variability is much less speculative than
utilising GCM projections for which there is little confidence.
8.2.3 Land use land cover (LULC) change

From 1950 to 2000, the population of the Ethiopian highlands is estimated to have
quadrupled, from approximately 16 million to 65 million; of which about 26 million
reside in the Blue Nile Basin (Hurni et al., 2005). Such population increases lead to
intensification of land use, including shortening and eventual abandonment of fallow
periods, expansion of cultivated land into grazing land, and deforestation (Lambin et al.,
2001). A literature review by Taye et al. (2015) reported that generally seen in the
Ethiopian Highlands is a decrease of low flows and an increase of high flows, i.e. trending
towards more hydrological extremes. The proposed cause is land use change: a decrease
in natural vegetation and an increase in cropped land, overgrazing, and Eucalyptus
planting. The result of these changes being greater transpiration and, consequently,
reduced baseflow. Contrastingly, a comparison of a series of photographs by Nyssen et
al. (2009) taken in the Rift Valley in 1868 by the British military expedition to Abyssinia
and in 2008 show improved land management and vegetation cover, albeit often from
Eucalyptus plantations, despite a ten-fold increase in population density, all of which
153
would decrease rapid run off. Past and future changes to LULC may be entirely
anthropogenic or could be climate related. Either way, the trend towards less forest and
increased coverage of cultivated land is widely reported, e.g. Abate (1994); Zeleke and
Hurni (2001); Amsalu et al. (2007).
Land degradation could be a threat at the Dangila study site as soil erosion has been
observed to be occurring during field visits elsewhere in Ethiopia, such as in Robit Bata
kebele near Bahir Dar and in Boloso Bombe woreda in SNNPR (Southern Nations,
Nationalities and Peoples’ Region). In both cases, according to local officials,
deforestation is due to increasing demand for firewood and charcoal production by a
growing population, which is often followed by overgrazing. The rate of population
growth in Dangila woreda is increasing; the total population increased from 149,000 to
160,000 to 175,000 between the 1994, 2007 and 2012 censuses, respectively (CSA, 1994;
2008; 2012). Even between field visits, the first in March 2015, the second in October
2015, and the third in January 2017, expansion of house-building was observed at the
edges of Dangila town on formerly cultivated land. This development can be observed
using the “time slider” on Google.Earth (Figure 8-1), in addition to both deforestation of
native woodland for house-building and afforestation of cultivated land for Eucalyptus
(Figure 8-2). Analysis of Landsat imagery by Jaleta et al. (2016) for the Meja watershed
(~200 km south of Dangila) revealed that Eucalyptus coverage had expanded from <1 to
15% between 1976 and 2015, and farmers intended to plant more. The community around
Dangila regularly speak of how the growing number of Eucalyptus plantations is
negatively affecting water level in their hand-dug wells. The Eucalyptus is not only
utilised locally for house building and charcoal, but, according to the Dangesheta kebele
office agronomist and Abiyu et al. (2016), is transported to Tigray and exported to Sudan.
Similar Google.Earth time slider assessment of the rural kebeles sees a growing number
of houses, though they are still sparse, increases in Eucalyptus plantations in cultivated
areas, and little conversion of pasture or shrubland to cultivated land. This latter
observation is suggestive that the hydrological/agricultural regime is stable, therefore,
encroachment onto the floodplain pasture to plant crops is considered too risky due to
likely flood damage (this has been stated by the local community) and that highland areas
are cultivated as much as they can be according to the gradient.
154
Figure 8-1. LULC change around Dangila town: increased house-building. (Image source: Google.Earth;
Imagery ©2014 and 2016 DigitalGlobe).
155
Figure 8-2. LULC change around Dangila town: deforestation for conversion to house-building and
afforestation of cultivated land for Eucalyptus. The forest types have been ground truthed. (Image source:
Google.Earth; Imagery ©2005 and 2016 DigitalGlobe).
There are several studies concerning the eco-hydrological impact of Eucalyptus in the
Lake Tana Basin. Soil analysis and farmer surveys by Chanie et al. (2013) at Koga, 30 km
northwest of the study site, revealed nutrient depletion and springs drying up following
Eucalyptus planting. Growers were aware of these impacts but insisted on continued
Eucalyptus planting due to the cash income generated. Simple plot experiments by
Alebachew et al. (2015) at Mecha, 20 km northwest of the study site, showed large yield
and biomass reduction (up to fifteen-fold) in maize and finger millet grown adjacent to
156
Eucalyptus plantations. The effects were not statistically significant beyond 20 m from
tree stands. Enku et al. (2017a) monitored hourly groundwater level fluctuations in
Eucalyptus plantations at Fogera and calculated dry season evapotranspiration of
2300 mm, almost double the reference evapotranspiration, compared to 900 mm without
Eucalyptus. The potentially negative impact of Eucalyptus led to its planting on farmland
being banned in the Tigray Region in 1999 (Hagos et al., 1999). However, many authors
claim its planting should be encouraged in marginal areas due to its general utility and
potential to alleviate poverty among growers, even stating it could dry saturated lands for
food production (Pohjonen and Pukkala, 1990; Jagger and Pender, 2003; Zegeye, 2010).
Looking at the impact of Eucalyptus across all Ethiopia, there are many more studies
available though most are conference proceedings and master’s theses and their impact
assessments were generally conducted with community questionnaires. As they all state,
further research is required on the impact of Eucalyptus on water resources.
8.2.4 Groundwater abstraction

Groundwater abstraction is currently minimal, only for domestic use and for backyard
irrigation predominantly utilising rope and bucket technology. Should abstraction
increase, the quantities and rates would depend on:
(i) Irrigation water requirement – controlled by irrigation method, crop type, growing
season length and climate
(ii) Well yield – controlled by aquifer properties
(iii)Water-lifting technology – controlled by availability, capital costs and running
costs
(iv) Area under irrigation – controlled by population size, access to water-lifting
technology, market demand and access, and land available
(i) A typical irrigation requirement would be 1 l/s/ha, pumping continuously 24 hours per
day, based on the crop water demand of typical crops in the climate of the Ethiopian
Highlands (Brouwer et al., 1992). The growing season length varies from up to 6-months
for the main rainfed crops, to ~100 days for the short-cycle crops that would be most
likely to be planted should a second irrigated growing season be possible (FAO, 2017).
(ii) The pumping tests described in Chapter 4 and Appendix A suggest that abstraction at
1 l/s/ha may not always be achievable. Although, the ease of well digging reported by the
community and confirmed by the small backyard plots that often have several wells,
means the required yield per hectare could be achieved by abstracting from multiple wells
(one at a time abstracting from storage, allowing other wells to refill). (iii) The general
157
desire among farmers in Ethiopia is for motorised pumps, the barriers being capital costs
and running costs (Evans et al., 2012). The shallow water tables and very short distance
from well to crop mean simple motorised pumps could be utilised and fuel costs would
be minimised. However, it must be borne in mind that at this altitude of ~2000 m,
centrifugal/suction pumps will have a lift of only around 2-3 m and are likely to be
unsuitable. (iv) Dangila has its own busy market and is around an hour from the important
city of Bahir Dar on a good road; therefore, distance/time to market is not a restricting
factor for agricultural expansion here. What’s more, rainfed land lies fallow for much of
the year and would be available for dry season irrigation. Population growth is occurring,
as described previously. An initiative offering water-lifting technologies directly or
through credit would likely have to occur to kick-start irrigation in the area.
The literature includes many studies on the impacts of water use on the Ethiopian Rift
Valley lakes (e.g. Legesse and Ayenew (2006); Ayenew (2007)), and modelling studies
on the impacts potential hydropower and irrigation schemes could have on Lake Tana
(e.g. Alemayehu et al. (2010); McCartney et al. (2010)). However, no studies have been
identified from Ethiopia assessing or simulating the likely increase of abstraction and
irrigation nor impacts on shallow groundwater resources.
8.2.5 Khat production

Field visits to Robit-Bata kebele, approximately 8 km northeast of Bahir Dar, while
assisting a Bahir Dar University project, revealed that since irrigation infrastructure had
become available, much of the (observed) agricultural land had been given over to khat
farming. This trend has been reported elsewhere in Ethiopia where the preference with
irrigation is to opt for a cash crop such as khat (Meshesha et al., 2014). The infrastructure
consists of tanks raised on Eucalyptus scaffolds or on topographic highs above hand-dug
wells. The tanks are filled using rope-and-pulley double-bucket systems with irrigation
then conducted manually with hoses (Figure 8-3). The leaves of the khat bush (Catha
edulis Forsk) are a mild stimulant and have been chewed in parts of East Africa and
Southern Arabia for centuries, reducing feelings of fatigue and hunger (Kalix and
Braenden, 1985). Khat is chosen for its high profitability, its return during the same year
as planting, and because it can be harvested, thus provide an income, year-round (Dube,
2014). Khat accounts for 13.4% of Ethiopia’s export earnings (third after coffee and oil
seed), Ethiopia being the primary global producer (Megerssa et al., 2014). Studies from
Jimma zone in southwest Ethiopia have shown that khat chewing causes social problems
such as family disintegration, health issues, lack of education when consumed by school
158
children, and increasing use of khat by 18-30 age group has negatively affected labour
productivity (Dube, 2014; Megerssa et al., 2014). Some studies claim khat is grown at
the expense of vegetables thus increasing food insecurity (Gebissa, 2010; Gezon, 2012).
However, other studies claim the potential export earnings (unlike the regular fluctuations
in the internationally controlled coffee price) and the fact that khat can be intercropped
with vegetables and grown on marginal (steep or poor soil) agricultural land, mean its
planting should be encouraged (Hailu, 2005; Klein et al., 2009). Studies explicitly
researching the impact of khat growing on water resources have not been identified,
though its year-round water demand is reported to have had negative impacts on lake and
groundwater levels elsewhere in Ethiopia where exploited for irrigation, causing wells to
require deepening and conflict between water users (Lemma, 2011; Meshesha et al.,
2014).
Figure 8-3. Khat production in Robit-Bata kebele.
Khat has only been observed in backyard plots within Dangila woreda for home
consumption and consumption generally does not appear to be prevalent in Dangila
compared to Bahir Dar; the local population has confirmed this anecdotally. The travel
time to a major population centre from the rural kebeles of Dangila woreda may restrict
the popularity of khat growing, unlike for Robit-Bata where the highest value freshly
picked leaves can be at market in Bahir Dar very quickly. What’s more, though popular
in Bahir Dar, especially with the young student population, khat production and
consumption is prevalent mainly in the Islamic east of the country; namely Oromia and
Harar Regional States (Lemessa, 2001; Gebissa, 2008). Therefore, an increase in khat
production was not considered a potential future scenario for simulation.
159
8.3 Historical drought analysis
8.3.1 Purpose, definition and methodology

A drought analysis was conducted using methods that would also identify particularly
wet periods. The purpose of the drought analysis was to assess the climate variability of
the study site. Analysis of all types of drought, rather than focussing on rainfall quantities
alone, would increase understanding of the likely impacts of climate variability.
Consideration of extreme wet periods, in addition to droughts, is also important due to
stakeholders reported (anecdotally) and observed negative impacts of high rainfall and
floods. A comparative drought analysis of a nearby area would confirm that the study site
meteorological time series spans sufficient climate variability extremes experienced in
the region. These drought analyses would enable the generation of future climate
variability scenarios for simulation in the models.
According to the European Drought Centre (EDC, 2016): “Drought is a sustained and
regionally extensive occurrence of below average natural water availability. Drought
affects all components of the water cycle from a deficit in soil moisture, through reduced
groundwater recharge and levels, and to low streamflows or dried up rivers…Drought
should not be confused with aridity, which is a permanent feature of a dry climate. Neither
with water scarcity which implies a long-term imbalance of available water resources and
demands.” Sheffield et al. (2013) consider meteorological drought, agricultural drought
(soil moisture), ecological drought, and hydrological drought (streamflow). Clearly,
assessment of annual rainfall totals alone is insufficient in determining the presence and
strength of a drought or wet period. The methods used here to evaluate dryness and
wetness were:
 SPI (standardised precipitation index)

 SPEI (standardised precipitation-evapotranspiration index)
 Annual river flow totals (unit runoff)
 Growing season length – 2 methods: Segele and Lamb (2005) and Stern et al.
(2006)
 NDVI (normalized difference vegetation index)
 LST (land surface temperatures)
 Annual rainfall from ground observation, remote sensing and reanalysis products
 Annual potential evapotranspiration (PET) from ground observation
 SPI and SRA (standardised rainfall anomaly) from other published studies
160
 Local community perception
The analyses were principally conducted for Dangila, with application of some methods
at Bahir Dar to determine if years not represented within the Dangila hydrometeorological
datasets were extreme drought or extreme wet years and, as such, should be incorporated
into climate variability projections for Dangila. Note that all annual computations
consider the hydrological year: 1st March to 28th/29th February. The following sections
describe the methods and show results while a synthesis and comparison of results from
all methods is provided last.
8.3.2 Available hydrometeorological data

In addition to the 22 years (1994-2015) of daily observations from the Dangila NMA
weather station used for SHETRAN modelling, monthly rainfall totals are available from
1987, though with abundant missing months of data prior to 1993. Historical monthly
rainfall records were obtained from the Global Historical Climatology Network (GHCN)
via the NOAA (National Oceanic and Atmospheric Administration) website (NOAA,
2017). Two periods were available for Dangila: 1955-1969 (with many missing months)
and an almost complete record from 1922-1934.
Bahir Dar weather data is commonly analysed in the literature due to its long available
time series. It was obtained for this study for comparison with the Dangila data, especially
to assess climate variability during periods when Dangila data is unavailable, i.e. the
infamous early 1980s droughts. The available daily data from the Bahir Dar NMA
weather station is from 2007 to 2015, and monthly data from 2002, though there are long
(12+ months) missing periods. Daily rainfall and average temperature data has been
sourced from the NOAA website for 1961-2000.
The other local raingauge observations (see Chapter 5) were not utilised due to their short
and discontinuous records. The NOAA website does not have data for these raingauges.
River flow records were available from MoWIE for the Kilti and Amen rivers from April
1997 to October 2014 and April 1988 (though are continuous only from 1998) to
September 2014 respectively.
Remote sensing and reanalysis rainfall products were downloaded from online open
sources for the grid squares containing the Dangila and Bahir Dar weather stations
respectively. The following products and periods were utilised: TRMM (1998-2014),
ERA-Interim (1979-2014), NASA-MERRA (1979-2014), NCEP (1979-2008), and JRA-
161
55 (1979-2013); see Chapter 4 for more detail on these remote sensing and reanalysis
products.
Normalised difference vegetation index (NDVI) and land surface temperature (LST) were
obtained from MODIS (moderate-resolution imaging spectroradiometer) on-board the
Terra and Aqua satellites. The data availability is dictated by the operational periods of
the satellites: 2001 (NDVI) and 2003 (LST) to 2013. The data products are available at a
16-day timestep and a resolution of 250 m that were spatially averaged across Dangila
woreda.
8.3.3 SPI (standardised precipitation index)

SPI is a normalised index of precipitation deficit or excess at specified time-scales, first
presented by McKee et al. (1993). It is the method recommended by the World
Meteorological Organisation (WMO) for monitoring drought severity (WMO, 2012)
because when using a long time-step it is stated that it identifies agricultural and
hydrological drought. A probability density function is determined to describe the
observations, then, once the distribution is established, the cumulative probability of an
observed precipitation amount is computed and the inverse normal (Gaussian) function is
applied to the probability resulting in the SPI (Guttman, 1998). Open source code is
available to run SPI analyses using R requiring monthly rainfall totals for a suggested
minimum of 30-years. A drought event is defined as a period in which the SPI is
continuously negative and reaches a value of -1.0 or less. The drought begins when the
SPI first falls below zero and ends with the next positive value (McKee et al., 1993). The
gradations are 0 to -0.99 = mild drought, -1 to -1.49 = moderate drought, -1.5 to -1.99 =
severe drought, ≤-2 = extreme drought. The same gradations are applied in the positive
for wet periods.
When considering the impact of drought on agriculture, it is most applicable to analyse

SPI for particular seasons. Poor spring rains would delay planting or prevent crops from
developing whereas poor summer rains would damage crops and shorten the growing
season. Obviously when poor spring rains are followed by poor summer rains, the impact
is exacerbated. Spring in this region is defined as March to May (13% of long-term mean
annual rainfall) in accordance with the onset of some rain; late November to February
experiences almost zero rainfall. Summer is considered June to October (84% of long-
term mean annual rainfall) when the onset of the wet season is clearly visible in late May-
early June rainfall totals, peaking in July and August, and dropping off in late October.
Thus, for spring the SPI is analysed at a 3-month time-step and calculated for May (the
162
end of the time-step) while for summer the SPI is analysed at a 5-month time-step and
analysed for October. SPI was calculated with all available years combined in a single
series (deleting the gaps) for the most robust probability density function estimate (Figure
8-4). SPI indicates extreme droughts in Dangila for spring 2003 and summer 1995. While
there are a few severe droughts and wet periods, mostly the SPI oscillates between mild
and moderate, wet and drought.
Figure 8-4. SPI calculated for Dangila for spring (3-month, March-May) and summer (5-month, June-
October). Note the non-continuous year sequence.
8.3.4 SPEI (standardised precipitation-evapotranspiration index)

SPEI, developed by Vicente-Serrano et al. (2010), builds on SPI by incorporating
evapotranspiration. It is mathematically similar to SPI though computes a climatic water
balance with the addition of temperature data. Like SPI, open source code is available to
run SPEI analyses using R requiring monthly rainfall totals. The drought and wet period
gradations match those for SPI. Analysis was similarly conducted for spring and summer
seasons with all available years combined in a single series (Figure 8-5), though the lack
of historical temperature data restricted the time series length. SPEI is applied with less
confidence than SPI due to the shorter time series. SPEI indicates extreme drought in
Dangila for summer 1995, otherwise, there are a few severe droughts and wet periods but
mostly SPEI oscillates like SPI between mild and moderate, wet and drought.
Figure 8-5. SPEI calculated for Dangila for spring (3-month, March-May) and summer (5-month, June-
October). Note the non-continuous year sequence, which is different to the SPI plot in Figure 8-4.
163
8.3.5 River flow totals
To directly assess hydrological drought, river flows were converted to unit runoff (total
annual flow divided by basin area) and the percentage variation from the mean was
computed. Figure 8-6 shows that there is high interannual variation in river flow, though
with consistently lower flow in both rivers from 2002-2005. The Amen is a small
(37 km2) catchment nested within the much larger (632 km2) Kilti catchment, therefore,
it is intriguing that in 2001, 2007, 2010 and 2012, the two rivers show divergence from
mean unit runoff in opposite directions. It could be explained by the small size of the
Amen catchment meaning localised intense storms would significantly affect the flow
records whereas the effects would be diluted within the large catchment.
Figure 8-6. Kilti and Amen Rivers’ percentage variation in unit runoff from the mean.
8.3.6 Rainfed crop growing season length

Segele and Lamb (2005), in investigating wet season variability as a cause of drought and
famine in Ethiopia, proposed a method for identifying growing season length. Assessment
is conducted on daily rainfall totals with thresholds of onset, cessation and dry-spell
determined by reported crop requirements and research of Ethiopian meteorology and
agriculture. Onset is defined as the first day of the year’s first wet-spell of at least three
days totalling 20 mm or more rainfall, provided there were no sequences of eight or more
dry (<0.1 mm) days in the subsequent 30 days. Cessation is the first day of a dry-spell
(<0.1 mm/d) of at least 20 days. A similar method of measuring growing season length,
proposed by Stern et al. (2006), was also applied. Onset is defined as the first occasion
with more than 20 mm of rainfall in a 2-day period after 1st April and no dry spell of 10
days or more within the following 30 days. Cessation is the first day after 1st September
that the water balance drops to zero. The daily water balance is computed with effective
rainfall (rainfall minus PET) and soil capacity. The soil capacity value was specified as
164
150 mm/m based on the FAO-UNESCO world soils database and mapping (see Batjes
(1997) and Nachtergaele et al. (2010)). Figure 8-7 shows that the interannual variability
in growing season length is significant, ranging from 140-237 days with Segele and Lamb
(2005) and 156-252 days with Stern et al. (2006). The typical range is April/May to
October/November with the few cessations in December mostly estimated by the Stern et
al. (2006) method, which generally predicts a later cessation. The Segele and Lamb
(2005) cessations match observed and anecdotal evidence better, which is understandable
as the method was tuned specifically for Ethiopian climate and agriculture.
Figure 8-7. Growing season length at Dangila. Dashed lines indicate where onset or cessation lies within
missing data.
8.3.7 NDVI (normalized difference vegetation index)

The NDVI is derived from the red/near-infrared reflectance ratio from the amounts of
near-infrared and red light reflected by vegetation and captured by the satellite sensor
(Pettorelli et al., 2005). The index is based on the fact that chlorophyll absorbs red light
whereas the mesophyll leaf structure scatters near-infrared (Myneni et al., 1997).
Therefore, NDVI values range from −1 to +1, where negative values equate to vegetation
absence. NDVI may be able to identify droughts by revealing lengthy periods of low
greenness or, conversely, periods of high greenness could indicate the length of the wet
season (Peters et al., 2002). Simple methods were derived to conduct the analyses. NDVI
data at 250 m spatial and 16-day temporal resolution was downloaded from the NASA-
MODIS website (https://modis.gsfc.nasa.gov/data/dataprod/mod13.php). The absolute
NDVI values per pixel were then averaged across Dangila woreda for each time step. To
reveal possible droughts the number of 16-day periods with NDVI in the lowest 10th
percentile and lowest 20th-percentile were counted for each year. To estimate wet season
length the number of 16-day periods with NDVI above the 50th percentile and 75th
percentile were counted for each year. Figure 8-8 shows the high interannual variability
165
in greenness and little correspondence between the two plots (which is not unexpected as
the left plot assesses the dry season and the right plot assesses the wet season).
Figure 8-8. Counts of 16-day periods of spatially averaged absolute NDVI values above/below thresholds
to identify dry and wet years. High counts in the NDVI dry years plot (left) indicate prevalent dry
conditions and low counts in the NDVI wet season length plot indicate short growing seasons.
8.3.8 LST (land surface temperatures)

The data used is the difference in LST between the MODIS Terra satellite measurement
at 13:30 and MODIS Aqua satellite measurement at 10:30, spatially averaged across the
woreda, again on a 16-day cycle. A high positive difference when LST is greater at 13:30
means low soil (or canopy) moisture; this would be expected in the dry season. A high
negative difference means LST is greater at 10:30, therefore, there is evaporative cooling
at 13:30; this would be expected in the wet season. Daily difference in LST may be able
to identify droughts by revealing long periods of high positive difference, which equate
to low soil or canopy moisture. Figure 8-9 shows there is little variation in mean LST
daily difference during the dry season and spring; however, there is greater interannual
variation during the wet season with clearly wetter and drier years.
166
Figure 8-9. Mean of LST daily difference per season. The greater the positive value the lower the soil and
canopy moisture content.
8.3.9 Annual rainfall totals

To assess purely meteorological drought and to make interesting comparison to other
drought analysis methods, annual rainfall totals from ground observations, remote sensing
and reanalysis were compared. Apparent from Figure 8-10 and Figure 8-11 is the
interannual variability in rainfall total and the overestimation in totals from most of the
reanalysis products. The Pearson correlation coefficients between annual totals from
remote sensing and reanalysis against ground observations are <0.60 for all the products.
Figure 8-10. Median and percentile annual rainfall totals for Dangila from ground observations, remote
sensing and reanalysis products.
167
4000
Dangesheta community obs. Dangila ground obs. Dangila
3500 TRMM NCEP
JRA-55 ERA-Interim
3000
Annual rainfall (mm)
NASA-MERRA
2500
2000
1500
1000
500
Figure 8-11. Annual rainfall totals for Dangila from ground observations, remote sensing and reanalysis
products.
8.3.10 Annual potential evapotranspiration (PET) totals

PET was calculated using the Penman-Monteith FAO56 method (Allen et al., 1998) with
meteorological data from the Dangila NMA weather station. It can be seen in Figure 8-12
that there is little interannual variability in PET totals; the coefficient of variation is just
3%.
Figure 8-12. Annual potential evapotranspiration totals from Dangila.
8.3.11 SPI (standardised precipitation index) and SRA (standardised rainfall

anomaly) from other published studies
Regional and nearby drought assessments appear in the literature and these were
compared with the Dangila analyses. SPI analysis of the northwest highlands of Ethiopia
was conducted by Viste et al. (2013) for spring and summer periods 1972-2010. Bewket
and Conway (2007) applied the SRA method (annual rainfall minus mean annual rainfall
all divided by standard deviation of annual rainfall) to several raingauge records of which
the 1962-2003 Bahir Dar assessment was considered here. The Bahir Dar SRA
assessment by Ayalew et al. (2012) was also considered as the different time series, 1979-
2008, gives different results to the previous study.
168
8.3.12 Local community perception
Rather than being conducted systematically, whenever the opportunity arose during field
visits, such as at community workshops (Figure 8-13), questions were put to the local
community regarding historical droughts and wet periods. Given that the great majority
of the agriculture is rainfed and domestic water comes from shallow wells; local farmers
and communities would be uniquely aware of all types of drought. Common responses
were that Dangila woreda does not suffer the type of droughts for which Ethiopia is
renowned, though the early 80s droughts were felt, and that the climate is becoming less
predictable. Indeed, there was a frost for the first time in living memory a few nights prior
to the January 2017 field visit, which caused visible damage to banana plants (Figure
8-13). It was commonly stated that 2014 was wet and 2015 was dry, these being in recent
memory. Memory triggers were given to the community for earlier suspected wet and dry
years with mixed results. A number of people remembered that 2012 was a poor growing
season because it was quite recent and due to the memory trigger of the death of Prime
Minister Meles Zenawi and the succession of Hailemariam Desalegn. The year before
millennium of the Ethiopian calendar, 2006 in the Gregorian calendar, was remembered
as being very wet in Dangila and causing flooding across Ethiopia. Earlier memory
triggers included elections, the war with Eritrea, football world cups and the Olympics
but there was little agreement in dry/wet years, perhaps complicated by Ethiopia’s unique
calendar and usually having to converse through a translator. Furthermore, Meze-
Hausken (2004) explains how statistical rainfall anomalies do not always create
memorable droughts as more depends on the preceding environmental, socioeconomic
and cultural conditions (preceding soil moisture, crop prices, increased population
densities due to migration, low soil fertility, low seed availability, poor crop choice, low
availability of aid, etc.) that impact people’s preparedness.
Figure 8-13. Community workshop in Dangesheta (left) and frost damage to banana plant in January
2017.
169
8.3.13 Synthesis of results
The different spatial and temporal scales of the drought analysis methods mean they
cannot be compared quantitatively to rank years in terms of dryness. However, a
qualitative comparison can be conducted using Table 8-1. There were many interesting
findings from this drought analysis:
 There is general agreement between methods; often enough to justify their

applicability for drought/wet period analysis in this area.
 Prior to the early-1990s, there are insufficient data to apply multiple methods to
confidently identify drought/wet years.
 Drought years are identified as: 1995, 2002-2003, 2009, 2012
 Wet years are identified as: 1999-2000, 2006, 2013-2014
 The requirement for using more complex drought analysis methods rather than
relying only on annual rainfall totals is illustrated by 2012, which had above
average annual rainfall, but the timing of the rainfall meant it was one of the worst
years, in terms of drought, on record.
 The reanalysis products perform particularly poorly in the early-80s where the
famous drought years are recorded as being especially wet.
170
Table 8-1. Identification of drought/wet years and comparison of drought analysis methods for Dangila. X = insufficient data for analysis. v = very. Note the jump from 1933
to 1956. For SPI, SPEI and SPI/SRA from other studies: v dry and v wet are <-2 and >2 respectively, dry and wet are <-1.5 and >1.5 respectively. For river flow: v high and
v low are >±40% difference in unit runoff from the mean, high and low are >±20% difference in unit runoff from the mean. For growing season length: v long and v short are
>±20% difference from mean length in days, long and short are >±10% difference from mean length in days. For NDVI dry year: dry is >3 16-day periods in the 10%ile and
>4 in the 20%ile, wet is <2 16-day periods in the 10%ile and <4 in the 20%ile. For NDVI wet season length: long is >12 16-day periods in the 50%ile and >8 in the 75%ile,
short is <11 16-day periods in the 50%ile and <7 in the 75%ile. For LST wet season daily difference: dry and wet are >±70% difference from the mean daily difference.
Annual rainfall: dry is rainfall total <10%ile, wet is >90%ile.
Year SPI SPEI River flow Growing NDVI LST Annual rainfall Annual Viste et al. (2013) Bewket and Ayalew et Local
season length PET NW Highlands * Conway al. (2012) community
spring summer spring summer Kilti Amen Segele Stern dry wet wet season spring summer (2007) Bahir perception
Bahir Dar Dar
NASA-MERRA
and et al. year season daily SPI SPI
ERA-Interim
Lamb (2006) length difference SRA SRA
ground obs.
(2005)
TRMM
JRA-55
NCEP
1922 X X X X X X X X X X X X X X X X X X X
1925 X X X X X X X X X dry X X X X X X X X X X
1927 dry X X X X X X X X X X X X X X X X X X X
1928 dry X X X X X X X X X X X X X X X X X X X
1930 dry X X X X X X X X X dry X X X X X X X X X X
1931 X X X X X X X X X dry X X X X X X X X X X
1960 wet X X X X X X X X X X X X X X X X X X X
1962 X X X X X X X X X X X X X X X X X X
171
1964 X X X X X X X X X X X X X X X X X wet X
1965 X X X X X X X X X X X X X X X X X dry X
1967 wet X X X X X X X X X X X X X X X X X X
1969 X X X X X X X X X dry X X X X X X X X X
1970 X X X X X X X X X X X X X X X X X X X X X
1971 X X X X X X X X X X X X X X X X X X X X wet X
1972 X X X X X X X X X X X X X X X X X X dry X
1973 X X X X X X X X X X X X X X X X X X v wet X
1974 X X X X X X X X X X X X X X X X X X v wet X
1975 X X X X X X X X X X X X X X X X X X dry X
1979 X X X X X X X X X X X X X wet wet X
1980 X X X X X X X X X X X X X wet wet wet X dry dry
1981 X X X X X X X X X X X X X wet wet X
1982 X X X X X X X X X X X X X wet wet wet X v dry v dry v dry v dry
1983 X X X X X X X X X X X X X wet wet wet X
1984 X X X X X X X X X X X X X wet X dry dry v dry
1985 X X X X X X X X X X X X X wet wet wet dry
1986 X X X X X X X X X X X X X wet wet v dry
1987 X X X X X X X X X X X X X wet wet wet v dry dry
1988 X X X X X X X X X X X X X
1989 X X X X X X X X X wet wet
1990 X X X X X X X X X X X X dry dry
1991 X X X X X X X X X X X X dry dry
1992 X X X X X X X X v dry
1993 X X X X X X X X wet
1994 X X long X X X X dry dry
1995 v dry v dry X X short short X X X dry X dry dry dry dry dry
1996 wet X v high long long X X X X
1997 X X long v long X X X X wet v dry dry
1998 X high short X X X wet
172
1999 wet wet v high v long long X X X wet wet
2000 wet wet high v high long X X X wet wet
2001 low v high long X dry
2002 v low dry X dry dry dry
2003 v dry dry v low v short v short dry dry dry v dry v wet
2004 low v short short wet X
2005 v low X
2006 v high v long v long wet wet wet X v wet wet
2007 X X high low wet wet X
2008 wet X X v high wet dry X
2009 dry X X v low short short dry X dry dry v dry X X
2010 X X short short dry wet dry X dry dry X X
2011 X X dry X X X X X
2012 dry X X v short dry short dry X X X X X dry
2013 v wet X X v high wet X wet X X X X
2014 v wet X X v long X X X wet X X wet X X X X wet
2015 X X X X X X X X X X X X X X X X dry
* Viste et al. (2013) present only droughts, not wet periods.
173
8.3.14 Comparison between Dangila and Bahir Dar drought analyses
A similar drought analysis was conducted for Bahir Dar, details of which are provided in
Appendix E. The analysis indicated that Bahir Dar generally experienced the same
drought and wet periods as Dangila, that there are particularly dry and wet periods that
are not represented in the available Dangila time series, and that reanalysis does not pick
up the early-80s droughts, misrepresenting those years as being particularly wet.
8.4 Generation of synthetic meteorological time series
8.4.1 Purpose
The SHETRAN models presented in Chapter 7 were run for as long as possible according
to the length of the river flow or groundwater level calibration time series or the length
of the daily rainfall and PET time series, whichever was shorter. To fully assess the
potential effects of climate variability on the shallow groundwater and surface water
resources, it was necessary to run the SHETRAN models for longer periods than as
presented in Chapter 7. Very dry years have been identified from the drought analysis but
because the time series simulated in the models were quite short, extreme years rarely
occur consecutively. Whereas, a longer time series is more likely to contain extended
droughts and different dry/wet year sequences. Therefore, a long time series of daily
rainfall and PET needs to be generated. Unfortunately, the pre-1994 Dangila rainfall data,
from the NMA or from GHCN, are monthly totals, which are inappropriate for the daily
simulation required in this study. However, a longer time series of daily rainfall data is
available for Bahir Dar and drought analysis showed that the climate is similar. A long
time series could be generated for incorporation into SHETRAN, combining Dangila and
Bahir Dar daily meteorological data if the datasets are comparable.
8.4.2 Comparison between Dangila and Bahir Dar rainfall

Bahir Dar lies 70 km northeast of Dangila, at 300 m lower altitude and adjacent to Lake
Tana, therefore has a slightly different climate. For that reason, it was necessary to test
whether the Bahir Dar data could be used to inform the longer period climate at Dangila,
i.e. are the coincident portions of the datasets sufficiently similar. Details of the statistical
analyses are provided in Appendix E. Essentially, satisfactory correlation testing of
rainfall totals and wet days justified the use of Bahir Dar rainfall years in the Dangila time
series used for the climate variability future scenario SHETRAN simulations. Further
statistical comparisons indicated that climatic extremes, both wet and dry, were not
captured within the 22-year Dangila rainfall dataset used in the SHETRAN modelling;
confirming what was discovered during the drought analysis. Notably, the Bahir Dar
174
dataset includes the infamous Ethiopian droughts of the early-80s and the very wet early-
70s.
8.4.3 Methodology
One option for generating a synthetic daily rainfall time series is the use of a Markov
chain model or other weather generator. The models generate new time series using the
persistence and periodicity of observed historical data to estimate transitional
probabilities between dry and wet conditions (Haan et al., 1976). Unfortunately, the
available Dangila daily rainfall time series is short and, we now know from the Bahir Dar
analysis, does not include the climatic extremes that would lead to a satisfactory
probability distribution function for generating the full range of likely climate variability.
What’s more, long periods of zero rainfall (at Dangila that is typically from mid-
November to early March) limit the applicability and effectiveness of rainfall generators
(Kilsby et al., 2007).
Therefore, the existing years of Dangila daily rainfall would be randomly arranged into a
longer sequence for simulation with SHETRAN. However, the previous section and the
drought analyses confirmed that additional extreme years should be added to the 22-year
Dangila dataset to simulate the full range of climate variability experienced in Dangila
but not represented in its meteorological record. Scaling the existing Dangila data
according to the rainfall range identified at Bahir Dar would be subjective as rules would
have to be generated governing the scaling factor for rainfall above and below certain
thresholds and whether to create or eliminate zero rainfall days. Instead, the extreme years
from the Bahir Dar time series were simply added unchanged to the sequence of Dangila
daily rainfall years.
To decide which additional years to add, the analysed years were approximately ranked
according to drought severity. Only 25 years at Dangila had sufficient datasets to enable
confident ranking. The 25 ranked years were divided into three eight-year periods to
represent the dry, normal (nine years) and wet conditions. The highest and lowest eight
years in the drought ranking are presented in Table 8-2 along with the eight driest and
wettest years from Bahir Dar. The ranking is approximate because rarely were enough
data available to apply more than a few of the possible analysis methods per year. Any
years in either column of Table 8-2 not represented in the Dangila time series were
incorporated, i.e. 1964, 1965, 1971, 1973, 1974, 1980 and 1982 (note that only monthly
rainfall records for 1964 and 1965 are available at Dangila thus the daily Bahir Dar
records were added).
175
Table 8-2. Approximate ranking of available years according to drought severity, showing the eight worst
(driest) and eight wettest years and the corresponding annual rainfall total.
Extreme drought Extreme wet

Year Dangila Bahir Dar Year Dangila Bahir Dar
(mm/a) (mm/a) (mm/a) (mm/a)
1982 ? 890 1973 ? 2035
1980 ? 1114 1974 ? 1966
1995 1186 1182 2014 2005 1712
2003 1375 ? 1999 1965 1460
2009 1451 ? 2006 1867 1652
2012 1640 ? 1971 ? 1859
2002 1351 ? 2000 1889 ?
1965* 1639 1089 1964* 1420 1783
* Selected primarily due to the Bahir Dar annual rainfall total.
PET time series for the additional rainfall years were not available. However, they may
be of questionable applicability at Dangila with its higher altitude and lower average
temperature. Therefore, existing years of Dangila PET time series were selected and
matched to rainfall total to coincide with the added rainfall years. Such practice was
robust as Figure 8-12 and analysis in Chapter 7 revealed the small interannual variability
in PET with a very low coefficient of variation (3.4-6.8%) between interannual monthly
totals.
The meteorological time series were randomly arranged using a Python script into a
200-year sequence based on the 29 existing years (22 from Dangila and 7 from Bahir
Dar). A 200-year sequence was considered long enough to allow generation of multi-year
drought and wet periods while not being too computationally time-consuming.
8.5 SHETRAN modelling
8.5.1 Initial/preceding conditions

Given that river discharge reduces to almost zero and groundwater drops to similar levels
at the end of each dry season, and that groundwater levels reach the surface in many areas
at the peak of the wet season, it was expected that there would be little “memory” in the
hydrological system and only a single year run-in period would be necessary. To test this
hypothesis, the subsurface water storage (including the saturated and unsaturated zones)
outputs from the SHETRAN models were divided by hydrological year and analysed.
The correlations in Table 8-3 show that the conditions of the previous year have no
statistically significant influence on the conditions of the following year, with the
176
exception of the correlation between the mean subsurface water storage and the previous
year’s minimum. Importantly, the minimum subsurface water storage, which controls the
abstractable quantity of groundwater during the dry season, is independent of whether the
previous year was wet or dry. Table 8-4 shows that subsurface water storage varies little
year on year with very small standard deviation, coefficient of variation, and absolute
range. Therefore, rainfall is sufficient during the wet season for the shallow aquifer to
essentially fill up every year. This is evident from observational and anecdotal evidence,
as well as from SHETRAN outputs, of the large expanses of inundated floodplain during
the wet season. The small range in absolute subsurface water storage shows that there is
little interannual variation and that the system is storage-controlled rather than recharge-
controlled regarding groundwater availability. This is a significant conclusion in itself
though understanding the spatial distribution of groundwater availability and how it
varies necessitated the more detailed modelling.
Table 8-3. Correlations between subsurface water storage from one year to the next for the Amen and
Kilti-Durbete catchments for 1998 to 2014 as computed by SHETRAN. SSW = subsurface water storage.
SSW min and previous SSW max and previous SSW mean and previous
year’s: year’s: year’s:
mean min. max. mean min. max. mean min. max.
Amen
Pearson 0.40 0.09 0.40 -0.10 0.56 0.34 0.38 0.82 0.52
P-value 0.13 0.73 0.13 0.71 0.02 0.19 0.15 0 0.04
Kilti-Durbete
Pearson 0.34 0.02 0.43 0.01 0.64 0.40 0.36 0.81 0.55
P-value 0.20 0.93 0.10 0.96 0.01 0.13 0.17 0 0.03
177
Table 8-4. Statistics of annual subsurface water storage in Amen and Kilti-Durbete catchments for 1998
to 2014 as computed by SHETRAN. SSW = subsurface water storage (spatially averaged mm), CV =
coefficient of variation.
Amen Kilti-Durbete
SSW SSW SSW max. SSW mean SSW min. SSW max.
mean min.
Mean 3042 2805 3237 3212 2959 3417
St. deviation 26.1 37.2 12.0 35.1 40.6 11.6
CV 0.9% 1.3% 0.4% 1.1% 1.4% 0.3%
Minimum 3006 2750 3207 3165 2895 3386
10th percentile 3011 2758 3224 3173 2906 3402
Median 3040 2803 3238 3215 2950 3419
th
90 percentile 3075 2862 3251 3241 3018 3428
Maximum 3087 2869 3252 3310 3029 3429
This analysis confirmed that a single year spin up period was sufficient. The models were
run for 201 years and the first year was discarded. The first year spin up period added was
2011 because this year falls in the centre of the drought analysis ranking; 2011 is ranked
15 of 29 and has close to mean and median annual total rainfall.
8.5.2 Modelled catchments

The potential future scenario simulations were run on the higher resolution (100 m) Amen
and Brante catchment SHETRAN models described in Chapter 7. The Kilti-Durbete
model was excluded as the groundwater potential mapping exercise suggested that the
500 m resolution was insufficient to capture groundwater level subtleties to enable
community-scale resource assessment. The Kilti-Dangesheta model was also excluded
from future scenario simulation, partly due to its lower (200 m) resolution and partly due
to computational restraints (Amen and Brante 201-year simulations were run on eight
powerful and fast Dell Blade servers and still took 4-7 days to run).
8.6 Measuring potential impacts
8.6.1 Additional irrigated growing seasons

Cultivation in Dangila woreda is predominantly rainfed. The aim of using shallow
groundwater for irrigation would be to enable a second growing season after the main
rainfed growing season. Therefore, increasing food security, employment and wealth, as
well as having a host of additional benefits described elsewhere such as improved
nutrition and gender equality. The most likely onset of a second growing season would
178
be straight after harvest of the main rainfed crop, taking advantage of residual soil
moisture and shallow water tables following the wet season. Harvest currently takes place
in October, which could be immediately followed by land preparation then seed sowing.
A potential second growing season is therefore simplified to begin on 1st November, and
be of 3-month duration. Traditional short-cycle crops in Ethiopia are wheat (104-170 day
cropping cycle), barley (91-174 days) and teff (78-123 days) (Meze-Hausken, 2004;
Reynolds, 2008) (cropping cycle lengths from FAO (2017)). However, research has
shown that small-scale farmers prefer to grow high-value vegetables and cash crops when
irrigated agriculture becomes an option (Kloos, 1991; Rockström et al., 2002; Emana et
al., 2015). Therefore, more likely, and more suitable for the prescribed growing season
length of 3-months, short-cycle crops include: okra (75-90 days), chilli peppers (82-97
days), chickpea (78-135 days) and tomato (75-130 days). These crops have all been
observed locally under backyard irrigation and are grown under rudimentary irrigation
elsewhere in Ethiopia (Wiersinga and de Jager, 2009); tomatoes being preferably grown
outside of the rainy season in the Lake Tana Basin otherwise they are vulnerable to pests
and disease (Abera, 2017).
8.6.2 Groundwater availability

Whether or not a second irrigated growing season is possible depends on groundwater
availability throughout that growing season. The length of groundwater availability will
vary by location and for each location it will vary interannually. To assess this variability
of shallow groundwater availability, the SHETRAN output of spatiotemporal distribution
of water table depth was processed using Python. The most useful measure of impact on
the shallow groundwater resource from the potential future scenarios would be a measure
of how much the proportion of the catchment where a second growing season is possible
varies spatially and interannually. The processing and analysis methodology was as
follows:
1. The groundwater level accessibility threshold was assigned as per Chapter 7:

Groundwater is considered accessible when the level is >0.5 m above the base of
the regolith, i.e. the water table is not in the basal basalt layer below the base of
wells and a sufficient water column is present to allow pumping by mechanical or
motorised means.
2. For each cell, the days from 1st November onwards when groundwater level was
above threshold were counted and the cell was assigned a category:
179
o Failure: groundwater is partially or unavailable for the 3-month period,
therefore, a second growing season after the main rainfed growing season
using shallow groundwater for irrigation is impossible.
o Success: groundwater is available for the whole 3-month period, therefore,

a second growing season is possible.
o High success: groundwater is available year-round, therefore, a second,

and possibly a third, growing season are possible.
3. The proportional spatial coverage of “Failure”, “Success” and “High success” was
determined for each simulated hydrological year and plotted.
4. In addition to assessing how climate variability affects the proportional areas, the
initial run was considered the baseline against which the land use and abstraction
scenarios were compared.
5. For quantitative comparison of the potential future scenarios, cumulative

frequency curves were plotted of the proportional coverage of “High success” for
each land use and abstraction scenario. These plots show the degree to which
groundwater availability is affected by the different scenarios.
An example of a success/failure plot is given in Figure 8-14 based on the historic Amen
catchment time series rather than a new 201-year simulation. The format of this plot was
developed with consideration of the paper by Forni et al. (2016) that discusses
visualisations to condense model results into meaningful formats for decision makers.
The aim was to combine spatiotemporal, multivariate, multi-scenario information into an
easily explainable and understandable format for local stakeholders.
Figure 8-14. Example plot of proportional coverage of failure (only rainfed agriculture possible), success
(second growing season possible) and high success (second and third growing season possible; shallow
groundwater available year-round) per year for the Amen catchment historic time series.
180
Apparent from Figure 8-14 is that a substantial proportion of the catchment always
achieves “High success” whereas “Success” is always a small proportion and the majority
of the arable land has potential for only rainfed agriculture. This realisation led to the
cumulative frequency curves being plotted for proportional coverage of “High success”
as described above. Also apparent by the high proportion of failure are the drought years
of 2002-2003, 2009 and 2012 as revealed in the Drought Analysis section.
8.6.3 Surface water availability

The impact on surface water resources were also considered in terms of availability. The
processing and analysis methodology was as follows:
 A low flow threshold of 1 l/s was set as the lower limit of productive flow.
 The SHETRAN outputs of daily discharge were assessed by counting the number
of days per year with flow <1 l/s.
As with groundwater availability, for quantitative comparison of the potential future

scenarios, cumulative frequency curves were plotted of the percentage of each year that
the low flow threshold was exceeded (flow <1 l/s). These plots show the degree to which
surface water availability is affected by the different scenarios. It should be noted that
SHETRAN provides this information on groundwater and surface water availability for
conversion into spatial plots if required.
8.7 Climate variability

The 200-year simulations applying the synthetic rainfall and PET time series indicated
that interannual variability in shallow groundwater availability would be high. This can
be seen visually in Figure 8-15 and Figure 8-16 and is confirmed by the statistics in Table
8-5. On average, a little over 30% of the catchments could be cultivated under irrigation
all year. The average proportional coverage of high success is almost identical for both
catchments, though with greater interannual variability in the Amen catchment as shown
by the high coefficient of variation of ~36%. However, even in the worst drought years,
7 and 10% of the Amen and Brante catchments have groundwater available for year-
round irrigation. The worst cluster of drought years can be seen during a four-year period
from years 6-9 in Figure 8-15 and Figure 8-16, although, the proportion of the arable land
that could sustain a second and third growing season only once drops below 20% during
this period for the Amen catchment and only twice drops below 25% for the Brante. The
coverage of success is lower for the Brante catchment though is generally very low for
both. This is significant and indicates that either year-round irrigation is feasible or only
181
rainfed agriculture is possible for most areas. Therefore, irrigated cultivation need not be
restricted to short-cycle crops, but the main long-cycle crops usually grown in the wet
season may be grown under irrigation.
Figure 8-15. Amen catchment baseline plot of proportional coverage of failure (only rainfed agriculture
possible), success (second growing season possible) and high success (second and third growing season
possible; shallow groundwater available year-round) per year for the 200-year simulation showing the
potential impact of climate variability on groundwater availability.
Figure 8-16. Brante catchment baseline plot of proportional coverage of failure (only rainfed agriculture
possible), success (second growing season possible) and high success (second and third growing season
possible; shallow groundwater available year-round) per year for the 200-year simulation showing the
potential impact of climate variability on groundwater availability.
182
Table 8-5. Statistics of proportional coverage of high success, success, and failure from Amen and Brante
baseline 200-year simulations. CV = coefficient of variation, c10 = 10th percentile, c90 = 90th percentile.
Amen Brante
High success Success Failure High success Success Failure
Mean 31.4% 2.8% 65.8% 31.7% 0.6% 67.6%
St.dev 11.2% 2.4% 10.5% 7.9% 0.8% 7.6%
CV 35.9% 84.3% 15.9% 24.9% 126.4% 11.3%
Min. 6.9% 0% 30.4% 10.3% 0% 46.8%
c10 14.5% 0.3% 52.7% 19.9% 0% 59.7%
Median 33.9% 2.1% 63.8% 34.2% 0.4% 65.5%
c90 43.6% 6.4% 79.4% 39.6% 1.5% 78.4%
Max. 68.4% 12.8% 91.5% 53.0% 6.6% 89.3%
The simulations presented here formed the baseline for comparison when applying the
potential future LULC and abstraction scenarios.
8.8 Land use land cover (LULC) change
8.8.1 Simulated LULC scenarios

The LULC change scenarios were selected based on observations and on literature review
of changes occurring in the region as described earlier caused by a growing population.
Essentially, a sensitivity analysis approach was used since detailed LULC spatial
information was not available, e.g. from processed satellite images. LULC change
simulations involved adjusting the proportional coverage of LULC classes. The coverage
of a particular class was increased by 10, 20, 50 and 100% over another class:
1. Pasture was progressively converted to arable land to simulate encroachment of

cultivation onto floodplain or other grasslands.
2. Highland areas were progressively converted to arable land to simulate increasing
cultivation on areas currently considered marginal due to slope or naturally
vegetated areas that are being cleared.
3. Highland areas were progressively converted to bareground to simulate excessive
firewood collection and overgrazing.
4. Arable areas were progressively converted to Eucalyptus to simulate expansion
of plantations.
The full range of LULC scenarios simulated is shown in Table 8-6.
183
Table 8-6. The potential future climate variability, LULC and abstraction scenarios simulated with
SHETRAN.
LULC scenario LULC Abstraction Abstraction Abstraction area

change rate period
Baseline Existing - - - -
1–4 Pasture converted 10% - - -
to arable 20%
50%
100%
5–8 Highlands 10% - - -
converted to arable 20%
50%
100%
9 – 12 Highlands 10% - - -
converted to 20%
bareground 50%
100%
13 – 20 Arable land 10% - - -
converted to 20%
Eucalyptus (two 50%
variations) 100%
21 – 23 Existing - 1 l/s/ha 3 months 10% of arable area
50% of arable area
100% of arable
area
The particular cells that had their LULC converted were randomly selected within their
land class with a Python script. The two additional LULC classes that were not run in the
historical models discussed in Chapter 7 were bareground and Eucalyptus. Bareground
was simulated with a higher Strickler coefficient of 2.0 m-1/3/s, a low AET/PET ratio and
proportional vegetation coverage reduced to zero. Eucalyptus was simulated in two ways:
1. With a low Strickler coefficient of 0.5 m-1/3/s, high AET/PET ratios, increased
rooting depth and root density at depth, high leaf area index (4.0) and increased
canopy storage capacity; the parameter values were based on numerous studies
including some that reported field investigations from the Lake Tana Basin, e.g.
Moroni et al. (2003); White et al. (2010); Chanie et al. (2013).
2. With parameters as for variation 1 and with increased PET for the cells containing
Eucalyptus. The PET time series was calculated using the FAO56 Penman-
184
Monteith method although in trees PET is higher because of the lower
aerodynamic resistance compared to the reference crop short grass. A multiplier
of 1.5 was applied to the PET time series for areas of Eucalyptus to simulate the
additional expected evapotranspirative loss. This multiplier is applied with some
uncertainty as published studies on the topic include: Enku et al. (2017a) who
measured evapotranspiration of Eucalyptus to be double the reference PET in the
Lake Tana Basin; Sharma (1984) measured Eucalyptus evapotranspiration up to
three times reference PET in the wet season of Australia though it was generally
equal in the dry season; Roohi and Webb (2016) quote a value for Eucalyptus of
15-20% greater than PET, and; Hutley et al. (2000) measured evapotranspiration
of Eucalyptus of around a third of PET (the latter two studies are also from
Australia). The average variation in additional evapotranspiration of Eucalyptus
above PET measured in these studies led to the selection of the 1.5x PET
multiplier.
8.8.2 Results for simulated LULC scenarios

Firstly, considering the conversion of pasture to arable land, Figure 8-17 and Figure 8-18
show that the impacts on surface and groundwater availability are negative and, though
quite small for the Amen catchment, are significant for the Brante catchment. The
negative impact is due to the greater evapotranspiration losses of the simulated crops over
grasses. The reason for the discrepancy between catchments is the higher proportion of
pasture in the Brante catchment, 47% compared to 34% of the Amen, which when
converted has a correspondingly greater impact on the overall catchment water balance.
Figure 8-17. Cumulative frequency curves showing the impact of converting pasture to arable land on
shallow groundwater availability (the proportional coverage of high success).
185
Figure 8-18. Cumulative frequency curves showing the impact of converting pasture to arable land on
surface water availability (the proportional of the year when flow is <1 l/s).
Secondly, considering the conversion of highlands to arable land, almost no impacts are
seen on surface and groundwater availability in either catchment (Figure 8-19 and Figure
8-20). The principal reason is the small proportional coverage of the highland LULC
category, 17% for the Amen and 5% for the Brante catchments. In addition, the water
demand of the scrub-like vegetation is only slightly greater than the common tall crops
such as maize and sugarcane, hence little difference in the overall water balance, and
therefore only slight positive impacts on surface and groundwater availability were
expected and seen.
Potentially significant for the scenarios converting pasture or highlands to arable land,
though not specifically simulated in the models, is the impact of tillage on infiltration.
While there is research that shows that the roughness resulting from ploughing required
by some crops increases infiltration (e.g. Lipiec et al. (2006); de Almeida et al. (2018))
other research suggests tillage decreases infiltration (e.g. Abid and Lal (2009); Tuzzin de
Moraes et al. (2016)).
186
Figure 8-19. Cumulative frequency curves showing the impact of converting highlands to arable land on
Figure 8-20. Cumulative frequency curves showing the impact of converting highlands to arable land on
Thirdly, regarding the degradation of highlands to bareground, significant positive

impacts are seen on surface and groundwater availability in both catchments (Figure 8-21
and Figure 8-22). There is an increase in flashiness of the river levels but more significant
is the reduction in evapotranspirative losses causing an increase in recharge maintaining
groundwater levels and surface water flows for longer into the dry season. However,
importantly, the modelling does not simulate hillslope erosion commonly observed
elsewhere in Ethiopia. The positive hydrological impacts may be temporary until gully
formation and soil erosion lower water tables.
187
Figure 8-21. Cumulative frequency curves showing the impact of degrading highlands to bareground on
Figure 8-22. Cumulative frequency curves showing the impact of degrading highlands to bareground on
Finally, Figure 8-23 and Figure 8-24 show that the expansion of Eucalyptus plantations
on arable land had the greatest impacts of all the LULC scenarios. The increase in
evapotranspirative losses significantly reduces surface and groundwater availability. The
negative impacts are greater for the Amen catchment even though the proportional
coverage of arable land is similar for both catchments, 50% (Amen) and 48% (Brante).
The second variation in Eucalyptus parameterisation shows greater impacts, explained by
the increased PET time series applied to Eucalyptus cells. The uncertainty over the correct
simulation of the increased evapotranspiration from Eucalyptus means the most likely
hydrological impacts are somewhere between the results of the two alternative methods.
188
Figure 8-23. Cumulative frequency curves showing the impact of expanding Eucalyptus plantations on
arable land on shallow groundwater availability (the proportional coverage of high success). The two
variations refer to alternative methods of parameterising the Eucalyptus LULC category.
189
Figure 8-24. Cumulative frequency curves showing the impact of expanding Eucalyptus plantations on
arable land on surface water availability (the proportional of the year when flow is <1 l/s). The two
variations refer to alternative methods of parameterising Eucalyptus LULC category.
8.9 Increased groundwater abstraction
8.9.1 Simulated abstraction scenarios

The future abstraction scenarios are based on literature review of the “Green Revolution”
in South and East Asia where cheap pumps and energy subsidies kick-started the
explosion in small-scale irrigation. Therefore, this future scenario carries the assumption
that pumps, fuel and irrigation infrastructure are going to become available at low cost
and/or a government ministry, NGO or private investor provides, directly or via credit,
the necessary equipment. As defined previously, the most likely future abstraction rate is
1 l/s/ha and the most likely pumping period would be a second growing season
immediately after harvest of the main rainfed crop for 3 months (1st November to 31st
January). The most likely short-cycle crop types, as discussed in Section 8.6.1, would be
high-value vegetables and cash crops such as okra, chilli peppers and tomatoes. The areas
190
subjected to abstraction and irrigation were 10%, 50% and 100% of the area of arable
land cover; this translated as 177, 887 and 1773 cells for the Amen model and 300, 1500
and 3000 cells for the Brante model. This arbitrary distribution of abstraction and
irrigation was specified based on the Asia examples where small-scale irrigation
expanded autonomously. The resolution of the models meant that one cell was one hectare
in size, therefore, to simulate a well abstracting at 1 l/s/ha, an abstraction of 1 l/s was
applied to the selected cells. Irrigation occurred in the same cell as the groundwater was
abstracted and at the same time, i.e. there was no storage for later irrigation. Irrigation
was simulated as additional rainfall input specifically onto cells with abstracting wells.
Wells were simulated, as observed, to be of the same depth as the regolith shallow aquifer.
The range of abstraction scenarios is listed in Table 8-6. It is noted that the baseline case
involves no abstraction as current levels of abstraction from the shallow aquifer are for
domestic use and as such are low and sparsely distributed.
8.9.2 Results for simulated abstraction scenarios

The results from the abstraction simulations showed little impact on surface and shallow
groundwater resources from abstracting and irrigating (Figure 8-25 and Figure 8-26).
Shallow groundwater availability reduced as the area under abstraction and irrigation
increased. However, surface water availability increased (at the expense of groundwater
availability) as some excess irrigation water not evapotranspired by crops augmented dry
season low flows. Return flow of irrigation water to river channels is well documented,
often constituting a large proportion of flow (Blodgett et al., 1992; Smakhtin, 2001).
What’s more, the simulations abstracted and irrigated at a constant rate regardless of
occasional dry season rainfall, which would also contribute to return flows. The Amen
and Brante catchments showed similar responses with the 10% arable area under
irrigation simulation giving results very similar to the baseline, especially concerning
shallow groundwater availability. The slight positive impact on surface water availability
and negative impact on shallow groundwater availability showed that flow paths from
fields to rivers must be rapid and are likely via surface runoff and unsaturated zone flow.
It should be noted that the increase in surface water availability is measured at the outlet.
It is possible that certain reaches may see slight decreases in surface water availability,
though, this would be less significant given the method of abstraction (from hand-dug
wells) compared to stream diversions for irrigation.
191
100% 100%
Amen Brante
90% 90%
80% 80%
70% 70%
Frequency
60%
Frequency
60%
50% Baseline 50%
40% 40%
Irrigating 10%
30% arable land 30%
20% Irrigating 50%

20%
arable land
10% Irrigating 100% 10%
arable land
0% 0%
Catchment coverage of high success Catchment coverage of high success
Figure 8-25. Cumulative frequency curves showing the impact of increasing abstraction and irrigation on
100% 100%
90% 90%
80% 80%
70% 70%
60% 60%
Frequency
Frequency
50% 50%
Baseline
40% 40%
Irrigating 10%
30% arable land 30%

arable land
arable land
0% 0%
Proportion of year when flow is below threshold Proportion of year when flow is below threshold
Figure 8-26. Cumulative frequency curves showing the impact of increasing abstraction and irrigation on
8.10 Discussion and conclusions

Even though rainfall, and therefore recharge, has high interannual variability, the shallow
aquifer geometry means that availability of shallow groundwater is mostly storage- rather
than recharge-controlled. This means that even in drought years, recharge is sufficient to
substantially refill the thin regolith aquifer.
Climate variability will strongly affect surface and groundwater availability with high
interannual variation in proportional coverage of areas able to have an additional growing
season after the main rainfed season. However, even during the most extreme and
prolonged droughts, a substantial proportion of the catchment will still have available and
accessible shallow groundwater for a second and possibly a third growing season.
Examples of shallow groundwater resilience to climate variability include northern Ghana
192
where shallow groundwater irrigation has led to poverty alleviation (Laube et al., 2008).
A particularly bad drought in 2006 led to many farmers losing crops due to water
shortages, though many others simply dug their wells deeper and still found groundwater
at <8 m.
Increased abstraction for irrigation resulted in small impacts on surface and groundwater
availability. This finding is significant and positive for farmers who gain access to pumps,
to the local community who use groundwater for domestic supply, to downstream water
users, and to ecosystems reliant on the same water resources. The finding somewhat
contradicts the traditional view that considers excessive withdrawal followed by aquifer
and surface water depletion to be a constraint on the use of shallow groundwater for small-
scale irrigation (e.g. Adams (1993); Giordano and Villholth (2007); Ngigi and Denning
(2009)). Abric et al. (2011) extol the benefits of small-scale irrigation in West Africa then
go on to describe how overdrafting of groundwater leads to continuous lowering of the
water table and increased exploitation costs. However, their studied region of West Africa
(Niger, Burkina Faso, Mali and Nigeria) receives much less rainfall and recharge than the
Dangila study site and the shallow groundwater exploitation examples are somewhat
large-scale, e.g. the Fadama Development Project in Nigeria (Inocencio et al., 2007).
Similarly, the Haromaya watershed in eastern Ethiopia is an oft-cited example of
excessive abstraction causing surface water bodies to dry up (e.g. Alemayehu et al.
(2007); Werner et al. (2013)). However, this region of Ethiopia is also semi-arid, low
transmissivity constrains excessive abstraction from the shallow aquifer (Tadesse et al.,
2010a) and reductions in lake volumes are most likely due to LULC change and land
degradation (Gebere et al., 2016). The slight decreases in shallow groundwater
availability simulated for the Dangila study site mean some local management of
groundwater use may be required to negate conflicts between adjacent (upgradient-
downgradient) water users. However, the slight increases in surface water availability due
to irrigation return flows mean wetlands and downstream surface water users would not
be negatively impacted by small-scale irrigation.
LULC change could have the greatest impact on water resources. However, it is noted
that the 100% change scenario would be extremely unlikely; it was simulated mainly to
see the direction the impacts on surface and groundwater resources could take. Wet
season flooding prevents significant agriculture, of the type currently practised, on most
of the floodplain pasture, high slope gradients would prevent all of the highland areas
from being cultivated and the basic human need for food crops surely limits the land
193
allocation that can be made to Eucalyptus planting thus preventing it from taking over
completely. In addition, the World Bank funded Sustainable Land Management Project
is active in Ethiopia with a large budget and wide geographical reach aimed at preventing
and restoring degraded land (World Bank, 2017a), therefore, it is hoped a 100%
conversion of highlands to bareground would be avoided. The 50% change scenarios are
less far-fetched though they may require a change to farming practices such as terracing
of hillslopes and retreat cultivation of floodplains as seen on the edges of Lake Tana. The
most likely and most impactful change is the increased planting of Eucalyptus. While
there is uncertainty over which variation most accurately simulates its impact, if the first
variation is considered the lower end of the impact range, then the negative effects on
surface and groundwater availability are still pronounced. This finding is in agreement
with the many observational studies in the literature that report decreased river flows and
reduced groundwater levels following Eucalyptus planting, e.g. Van Lill et al. (1980);
Sikka et al. (2003); Rodriguez Suarez et al. (2014) due to greater wet season interception
and greater dry season evapotranspiration. The additional impact only briefly mentioned
is the yield losses from crops grown in proximity to Eucalyptus plantations, which would
obviously be more pronounced if the number of plantations increases.
It has been shown that the shallow groundwater resource at Dangila woreda is resilient to
climate variability, LULC change and increasing abstraction. There exists the potential
for a sustainable intensification in irrigated agriculture. Whereas the maps presented in
Chapter 7 showed the areas that, on average, have greatest potential for small-scale
irrigation, the model outputs from this chapter reveal how the potential areas spatially
vary under different climatic conditions and with future variations to land use and
abstraction. This more detailed analysis enables greater targeting of areas for
development of small-scale irrigation and provides more confidence that such
development is sustainable.
194
Chapter 9. Conclusions, including recommendations,
transferability, and future work

Chapter 8 showed, through simulation of potential future scenarios with SHETRAN, that
the shallow groundwater resource is resilient to climate variability, land use change, and
increasing abstraction. This chapter will draw together findings from all the previous
chapters presenting the conclusions with reference to the aim and research questions in
Chapter 1. Recommendations are provided for stakeholders, and the transferability of the
results and of the methodology is discussed. The limitations of the study are described
before the final section gives suggestions for future work.
9.1 Conclusions
To conclude this thesis, we will revisit the aims and research questions presented in
Chapter 1. The overall aims of this PhD research were:
Firstly, to determine the potential for small-scale irrigation and the resilience of
shallow groundwater resources used by rural communities at a representative
study site in Ethiopia.
The research demonstrated that shallow groundwater resources do have the potential for
exploitation for small-scale irrigation and the most suitable areas were identified. The
research also showed the resilience of the resource in the face of likely climate variability,
and only the most extreme land use changes had severe negative impacts on the resource.
Critically, the resource is resilient to small-scale abstraction for irrigation.
Secondly, to develop transferable methodologies for assessment of shallow

groundwater resources throughout SSA.
The research demonstrated that citizen science for the collection of hydrometeorological
time series is an effective measure in areas of data scarcity, benefitting both researchers
and local stakeholders. Field hydrogeological investigative techniques were developed
and local partners, such as Bahir Dar University students, were trained in their
implementation for shallow groundwater assessment. The extensive recharge assessment
revealed which methods are most appropriate for recharge estimation of shallow aquifers
with wet, though seasonal, climates. The modelling indicated areas with highest potential
for shallow groundwater exploitation and suggested simpler means for identification of
195
such areas without requiring costly and complex modelling, i.e. through field
hydrogeological and geomorphological assessment and through local community
discussions.
The specific research questions in Chapter 1 were:
1. Do shallow aquifers have the requisite properties, in terms of hydraulic

conductivity, potential well yield, specific yield, aquifer geometry and
hydrochemistry, for productive groundwater use?
The field investigations conducted in Dangila woreda described in Chapter 4 showed that
while hydraulic conductivity and consequently well yield, and specific yield of this
shallow aquifer are quite low, they are not too low for small-scale irrigation; especially,
if abstraction commenced at the immediate cessation of the wet season when well yield
is high. Specifically, dry season average K was 1.6 m/d while the wet season average was
6.5 m/d; the greater saturated thickness led to interception of higher K layers thus
increasing average K. These results gave dry season average yield of 0.2 l/s and wet
season average of 1.7 l/s, while the median specific yield was 0.09. Multiple wells and/or
wells of larger diameter would likely have to be excavated to increase well bore storage,
as described in the recommendations section of this chapter. The aquifer is also quite thin
in many areas (<5 m), which restricts the shallow groundwater volumes that can be stored.
However, the modelling in Chapter 7 revealed areas where shallow groundwater is
available year-round and when abstraction was modelled in Chapter 8, the resource was
not depleted. Hydrochemistry analysis described in Chapter 4 indicated that the shallow
groundwater is suitable for irrigation. What’s more, contrasting major ion and stable
isotope hydrochemistry of the shallow and deep groundwaters suggest that there is little
to no connection between the shallow regolith aquifer and the deep fractured aquifer.
Radon-222 measurements support this conclusion, as shallow aquifer leakage was not
identified. The stable isotope analysis also enabled development of a conceptual model
where shallow groundwater flow does not necessarily match surface water flow directions
from the small sub-catchment “dambos”. In summary, the research showed that shallow
aquifers could have the requisite properties for productive use.
2. Due to the scarcity of time-series data, are community-based hydrometeorological

monitoring programmes able to produce useful, high quality data comparable to
formal data sources?
196
As explained in Chapter 5, the satisfactory application of accepted quality control
procedures and statistical comparisons with formal data sources indicated that the
community-monitored hydrometeorological data from Dangila is of high quality.
Furthermore, the community data often outperforms some formal and remote sensing or
reanalysis data; especially concerning groundwater where there is no formal equivalent.
The research showed that such citizen science data can infill the gaps in sparse and
deteriorating formal monitoring networks. This hydrometeorological data was invaluable
for the initial understanding of the hydrological system and the development of the
conceptual model. The data were then used in the recharge assessments to assess if a
renewable shallow groundwater resource was present. Finally, the time series data
enabled modelling of the shallow groundwater as simulated river flows and groundwater
levels were calibrated against the observed data.
3. Can shallow groundwater be considered a renewable resource, and; which

recharge assessment methods provide the highest confidence in the calculated
recharge amounts when applied to these types of aquifers?
Chapter 6 reported that recharge to the shallow aquifer was sufficient for small-scale
irrigation to consider the shallow aquifer a renewable resource, i.e. groundwater would
not be “mined”: The methods computing actual recharge gave a catchment-wide recharge
estimation of 280-430 mm/a. A simplistic calculation can be run with this recharge
estimate, albeit ignoring other factors relevant to irrigable area: If we specify a nominal
irrigation demand of 1000 mm and take the median recharge value from the actual
recharge range, a little over a third of the study site could be irrigated without exceeding
the recharge rate. The recharge assessment also revealed the range of results that can be
computed for a single site, from 45 to 814 mm/a, when multiple recharge estimation
methods are applied. The discrepancy in recharge results can provide insights on the
hydrogeological system allowing update of the conceptual model and informing the type
of recharge computed by the different methods. Concerning recharge method selection,
it is crucial to be sure of the spatiotemporal scale of the method, to use high quality input
data, to have developed but be willing to further develop the conceptual model, and to
know what type of recharge is being computed. In this study, the conceptual model meant
streamflow hydrograph and soil moisture balance methods were computing minimum and
potential recharge, respectively, rather than actual. If these methods are to be applied to
shallow aquifers elsewhere, consideration of the conceptual model and application of
additional methods is recommended. For identifying areas of shallow aquifers that may
197
be the most productive for irrigation, water table fluctuation methods that actually
calculate change in aquifer storage can reveal areas of highest potential; certain wells
analysed in this study showed “recharge” of over 1600 mm (incorporating lateral
groundwater flow).
4. Are there easily identifiable zones that show the greatest potential for sustainable
intensification of agriculture through shallow groundwater irrigation?
The modelling described in Chapter 7 showed that the base of hillslopes and narrow
valleys, corresponding to approximately 17% of the study area, have the greatest potential
for small-scale irrigation as the shallow groundwater resource remains available for the
longest period in these areas. Areas with the least potential, where shallow groundwater
was available for short periods, were high on steeper hills especially near ridges.
Additionally, low potential areas were those where the aquifer is thinnest and easily
drained such as the large flat floodplains. At shallow aquifers elsewhere, these locations
should be easily identifiable from topographic maps and DEMs though well surveys
would still be required for appreciation of aquifer thickness variations.
5. How will climate variability, land use change and increased abstraction impact
shallow groundwater resources and surface water?
Chapter 8 revealed that, on average, around 35% of the modelled catchments’ arable land
had available shallow groundwater year-round (Figure 9-1), allowing cultivation with
irrigation, additional to the main rainfed growing season. Modelling of climate variability
indicated that even during prolonged droughts, significant proportions of the catchments
would still have the potential for a second crop planted immediately following the main
rainfed crop harvest; Figure 9-1 shows that in years in the driest 5th percentile, around
15% of the arable area has shallow groundwater available all year.
198
40%
Proportion of arable land that has

35%
shallow groundwater available

throughout the dry season
30%
Amen
Brante
25%
20%
15%
10%
5%
0%
5th %ile 10th %ile 50th %ile
Severity of dryness
Figure 9-1. Output of climate variability model simulations showing the proportion of arable land that has
shallow groundwater available in a particular year according to that year’s severity of dryness.
Simulating progressive land use change had varying effects on shallow groundwater and
surface water resources. Conversion of pasture (the floodplain grasslands) to arable land
had a slight negative impact on shallow groundwater availability but very little effect on
surface water. Converting highlands to cultivation had almost no impact on water
resources whereas highlands degrading to bareground had a positive impact due to
decreased evapotranspiration. Though it is noted that the likely related impacts of soil
erosion and gully formation were not simulated and these would increase flashiness of
the streams and probably decrease shallow groundwater availability through decreased
infiltration and recharge. The land use change scenario that had the greatest (negative)
impact was increasing the area of eucalyptus plantations.
Increasing abstraction and irrigation at likely smallholder levels (rate of 1 l/s/ha) had
small impacts on shallow groundwater and surface water availability. Simulation of a 3-
month irrigated growing season immediately following the main rainfed growing season
led to slight reductions in shallow groundwater availability. However, rapid return flows
of excess irrigation water led to slight increases in surface water availability.
9.2 Recommendations
9.2.1 Potential for the productive use of shallow groundwater

This research indicated that there is available shallow groundwater for an expansion of
small-scale irrigated agriculture at the study site (Chapter 7). What’s more, the shallow
groundwater resource can be considered renewable (Chapter 6) and is resilient to climate
199
variability, land use change and increasing abstraction based on the simulated future
scenarios (Chapter 8).
The recommendation is for a second growing season to commence immediately after the
main growing season, taking advantage of residual soil moisture and shallow water tables
following the wet season. This would require immediate land preparation and planting
following harvest of the main rainfed crop. This is of course with the assumption that
pumps and irrigation infrastructure will become available. Cooperation between farmers
regarding mobilisation and sharing of the infrastructure should not be problematic as land
is currently cultivated cooperatively. There is some risk to planting a second long-cycle
crop (e.g. grain), even in areas classed as “high potential”, as the modelling in Chapter 8
showed that in the worst drought years, shallow groundwater available year-round is
restricted to around 15% of the arable area of the catchments (Figure 9-1). Planting of a
short-cycle crop carries less risk of insufficient irrigation water and is the more likely
eventuality as research shows that small-scale farmers prefer to grow high-value
vegetables and cash crops when irrigated agriculture becomes an option (Kloos, 1991;
Rockström et al., 2002; Emana et al., 2015). While it has traditionally been feared that
groundwater use for irrigation will have unacceptable impacts on domestic water supply,
wetlands and other groundwater-dependent ecosystems (Adams, 1993; Giordano and
Villholth, 2007; MacDonald et al., 2009), this research showed that the impacts from
small-scale irrigation on surface and groundwater resources would be minor (Chapter 8).
9.2.2 Abstraction strategy

Pumping tests showed that potential well yields might not achieve the desired 1 l/s for
irrigation of a hectare (Chapter 4). However, wells are not difficult to excavate, as
reported by the local community and indicated by the small (<0.25 ha) backyard plots
that commonly have multiple wells. Therefore, to expand irrigation onto the fields
currently only utilised for rainfed agriculture, a high density of wells may have to be
excavated. Therefore, pumping equipment will have to be mobile, or many pumps will
be required, in order to switch to abstracting from a different well as each is depleted. In
the meantime and overnight, the non-abstracting wells will refill naturally. The
recommendation is for electric submersible pumps. Solar power would be preferable,
though, the local community have spoken of the desire for small generators that could
also be used at home: During workshops, the local community questioned having
generators in fields to power pumps when they do not have light in their houses. The
generally shallow water table means simple and low cost motorised vacuum or centrifugal
200
pumps would be sufficient, however, the altitude of ~2000 m means suction lift is
restricted to around 2 m. Therefore, for such pumps to be practicable, shelves would need
to be excavated at depths in the wells for pump placement, as described by Carter and
Alkali (1996).
9.2.3 Well design

Large diameter wells would be required if shelves for pumping infrastructure were to be
excavated, and larger diameter is also a recommendation from a point of view of storage.
The reason for the multiple well recommendation is that the low transmissivity would
lead to a well being pumped dry before it could naturally refill. Larger diameter wells
would increase the well bore storage meaning less wells would be required and the
pumping equipment could be moved less frequently. This is not a new idea to Ethiopia as
large diameter wells are utilised for these same reasons in Tigray, where they are often
stone-lined to prevent collapse (Woldearegay and Van Steenbergen, 2015).
9.2.4 Floodplain cultivation

Modelling the future land use change of conversion from pasture to cultivation indicated
that this scenario would not have a significantly negative impact on surface and
groundwater resources (Chapter 8). Therefore, could this scenario be encouraged? The
generally flat and seasonally inundated floodplains are almost exclusively used as pasture
for cows, sheep and goats, typically tended by local children. Other infrequently observed
floodplain activities are harvesting of wetland vegetation for animal fodder and small
plant nurseries found beside perennial reaches of streams during the dry season. While
the floodplain marginal areas are sought after for cropping to take advantage of residual
moisture and shallow water tables, cropping is largely non-existent within Dangila
woreda on the floodplains themselves. Elsewhere in the Amhara Region, wetlands have
been put to use for rice growing. A study by Tefera (2017) from Fogera Plain adjacent to
Lake Tana reported that participation in the rice industry increased from 30 households
in two kebeles with an area coverage of 6 ha in 1993/94 to 34,249 households in 24
kebeles with an area of 20,230 ha by 2014/15. Fogera district now produces 60% of
Ethiopia’s rice. Surveys and interviews revealed that ~55% of producers rely on rain and
~45% supplement rainfall with irrigation water from rivers and springs. Multiple
cropping cycles per year are practiced utilising residual moisture in seasonally
waterlogged areas. In addition to local consumption or market sale, the grain straw was
used for house construction, animal feed, as a fuel source, and as a raw material for some
manufacturing processes. Floods and related water-borne diseases were serious problems
201
on the plains around Lake Tana (Sewmehon, 2012), however, with the introduction of
rice as a crop, these water-rich ecosystems have developed from an environmental
problem to an economic and lifestyle opportunity (Tefera, 2017). Rice production has had
such a positive impact that the crop is known locally as “white gold” (MoWIE, 2015;
AgroBIG, 2016). Tefera (2017) states that there was initial resistance to rice production
due to a perception that the grain causes infertility in humans. This was apparently
resolved through collaborative and “aggressive” promotion by the local agricultural
research centre. It seems feasible for such a transformation of wetlands to occur in
Dangila woreda. As in Fogera, the local population generally speak of the problems
caused by the seasonally inundated floodplains, in particular the extended journey times
as the wetlands can often not be crossed during the wet season, however, in Fogera the
seasonal inundation is now considered a blessing rather than a curse (Tefera, 2017).
Similar complaints, including of flooded or waterlogged fields, were recorded in the
wider Gilgel Abay catchment by Abera (2017). Whereas Fogera plain occupies an
expansive lakeshore position, an area of over 500 km2, there are many examples of much
smaller scale agricultural exploitation of seasonally flooded land. A well-developed
system of rice cultivation exists in small seasonal valley bottom swamps in Sierra Leone,
which are also cultivated with such crops as wheat, tomatoes and onions during the dry
season to take advantage of the better soil fertility compared to upland soils (Richards,
1985; Dries, 1991). Small-scale ‘garden’ cultivation of rice with dry season crops at the
margins, based on indigenous water management techniques, takes place in dambos of
Zimbabwe, Zambia and Malawi (Dixon, 2003) that are geomorphologically similar to the
features seen in Dangila. As long ago as 1998, Rwanda already had 165,000 hectares of
wetlands, 50% of which are cultivated, and the Tanzania Ministry of Agriculture were
developing policies for utilisation of its ~850,000 ha of wetlands (Inocencio et al., 2003).
However, before promoting agricultural development of Dangila’s wetlands, it must be
stated that there is currently little awareness of the status of Ethiopia’s wetlands
ecosystems (Abebe and Geheb, 2003). These ecosystems support both aquatic and
terrestrial biodiversity, e.g. around 25% of Ethiopia’s bird species are wetland-dependent
(Wondefrash, 2003) and some wetland plants are used locally for medicinal purposes
(Hailu, 2003). In addition, the hydrological controls of the floodplains in terms of
groundwater recharge and flood control is uncertain (Abebe and Geheb, 2003). Both
Dugan (1988) and Roggeri (2013), however, argue that the conversion of wetlands to a
different ecosystem should be considered where a cost benefit analysis proves that
conversion is the most effective means of contributing to the social and economic needs
202
of the human population. The current use of the floodplains is likely in relative harmony
with the ecosystem: Fynn et al. (2015) discuss how even overgrazing does not impact
such ecosystems because they are so waterlogged in the peak wet season that animals can
only access the periphery allowing vegetation to fully recover.
9.2.5 Conversion from Eucalyptus to Pine

The impact on surface and groundwater availability caused by increased Eucalyptus
planting was the greatest of all proposed land use change scenarios modelled (Chapter 8).
Eucalyptus planting is reported to be on the increase in the Lake Tana Basin even though
farmers are aware of its negative impacts on groundwater levels, springs and streamflow,
and on adjacent crop yield (Chanie et al., 2013; Alebachew et al., 2015; Jaleta et al.,
2016). The demand (locally, nationally to Tigray, and internationally to Sudan) for wood
for house-building and charcoal production is unlikely to diminish in the short-term due
to the high population growth rate and loss of native forest that could otherwise provide
timber and firewood (Abiyu et al., 2016). Pine could be an alternative cash crop that is
not as water intensive. Eucalyptus is popular because it is so highly productive with
growth rates of >35 m3/ha/a and because of its short rotation length of 6-8 years (Albaugh
et al., 2013). However, in many parts of the world, pine is favoured, even though the
growth rate is less at 25-27 m3/ha/a and the rotation period is longer at 15-25 years, as it
is known to be less water intensive (Maier et al., 2017). Paired catchment studies are
abundant in the literature comparing the effects of pine with that of Eucalyptus. Scott and
Smith (1997) working at several sites in South Africa recorded streams drying up eight
years after Eucalyptus afforestation of a catchment whereas in afforested pine catchments
streams took 12 years to cease flowing. Lima et al. (1990) measured more than 200 mm
greater evapotranspirative losses from Eucalyptus than pine and significant reductions in
soil moisture at a site in Brazil. A global analysis of 26 catchment data sets with 504
observations, including annual runoff and low flow by Farley et al. (2005) showed that
Eucalyptus afforestation has the greatest impact on water yield reducing runoff by 75%
(±10%), compared with a 40% (±3%) average decrease with pines and similar decreases
of low flows. Research by Calder et al. (1993) at multiple sites in India showed how soil
erosion was greater in Eucalyptus plantations compared to pine and soil moisture was
greatly depleted, the latter conclusion also reported by many other studies, e.g. Lima et
al. (1990); Zavala et al. (2009); Santos et al. (2016). Pinus canariensis, the Canary Island
pine, may be a specific alternative for Ethiopia as the species is tolerant to climate
variability, especially drought (Dr Juan Suarez, Forest Research, Roslin, personal
communication, 18 December 2017 and Wieser et al. (2016)). The species has the lowest
203
canopy storage capacity of conifers (though it is still slightly greater than Eucalyptus)
meaning more water is transferred to the soil surface (McPherson et al., 2017), and has
been successfully forested for decades in South Africa (Richardson and Higgins, 2000).
If the shallow groundwater resource is increasingly exploited, planting should consider
the optimum tree cover that results in greatest recharge; above this the trees use too much
water and below this there is increased overland flow (Ilstedt et al., 2016).
9.3 Transferability of results
9.3.1 Transferability within Ethiopia

The transferability of this PhD research to other areas of Ethiopia can be identified from
the map in Figure 9-2 of the main aquifer types within Ethiopia and their distribution.
Aquifers termed “shallow” and “very shallow” have a wide distribution, approximately
50% of the country (Kebede, 2013). What’s more, it is likely that the deeper volcanic
aquifers have superficial geology very similar to Dangila woreda. Considering Ethiopia’s
climate, all but the east and north has similarly high rainfall and seasonality to Dangila.
Fieldwork elsewhere in Ethiopia, and travelling to those field sites, showed similarities
in geomorphology with Dangila, i.e. expansive floodplain wetlands and low hills.
Legend
Rhyolite Ridges – Aquicludes
Deep Mesozoic Sediment Aquifers
Very Deep Mesozoic Sediment Aquifers
Very Deep Volcanic Aquifers
Deep Volcanic Aquifers
Shallow Volcanic Aquifers
Very Shallow Alluvio-Lacustrine Aquifers
Shallow Alluvio-Lacustrine Aquifers
Shallow Basement Aquifers
Very Shallow Basement Aquifers
Study site
Figure 9-2. Map of Ethiopia showing distribution of main aquifer types (from Kebede (2013)).
204
9.3.2 Wider transferability within sub-Saharan Africa
The field site was selected to be representative of shallow aquifers at a broader scale than
northwest Ethiopia. It would be expected that the findings from this research in terms of
areas of high and low potential and the resilience of the resource would be applicable in
areas of similar climate, geology and geomorphology. The map in Figure 9-3 shows areas
of shallow (<25 m) water tables abound in SSA. Considering rainfall, a large swathe of
the African continent between 10o North and 20o South receives rainfall of 800-3200
mm/a (Figure 2-4) with a similar seasonality to Dangila (Guiraud, 1988; Frenken, 1997).
Within that rainfall zone and within the shallow water tables zone, young (Cenozoic to
Quaternary) volcanic rocks are present all along the East African Rift from Eritrea to
Malawi and in unconnected areas such as western Cameroon/eastern Nigeria and
southwest Sudan (MacDonald et al., 2011). In addition, key texts on regolith state that
hydrogeological properties are similar independent of parent rock age and mineralogy
(Jones, 1985). The abundance of crystalline rock in Africa (Figure 2-2) suggests shallow
weathered regolith aquifers overlying very low to zero permeability bedrock, as at the
study site, may be common, though the transferability to these areas requires further
research. Similar geomorphology to Dangila may be expected in the previously
mentioned regions due to the similar climate and geology (Grove, 1986; Burke and
Gunnell, 2008).
205
o
10 N
o
20 S
Figure 9-3. map of Africa with estimated depth to groundwater (from Bonsor and MacDonald (2011))
9.4 Transferability of methodology

The methodology of this research is confidently transferable. While a thorough review of
background information and previous work is always recommended (Chapter 3), not all
the field investigations applied here (Chapter 4) would have to be carried out during a
shallow groundwater resource assessment. Surveys of wells (measuring the depth,
groundwater level and determining the seasonality) are invaluable for assessment of
aquifer geometry, groundwater recession rates and areas of highest/lowest potential for
increased abstraction. Such surveys, including questioning local stakeholders, enable
qualitative determination of aquifer properties. If modelling is required, or accurate
estimation of potential well yield, pumping tests must be applied to quantitatively
determine aquifer properties, even though with hand-dug wells and manual water-lifting
the method is quite rough-and-ready. Hydrochemistry and stable isotope analysis are
useful in revealing aquifer connectivity and recharge mechanisms enabling development
of the conceptual model.
206
Citizen science for the gathering of hydrometeorological time series (Chapter 5) is being
successfully applied as part of a REACH accelerated project, “Water security risk science:
local knowledge for participatory resource management”, (REACH, 2018) in Boloso
Bombe woreda, SNNPR, based on, and improving, the Dangila programme. Other
successful citizen science programmes with which potential collaborations have been
discussed include Sierra Leone where community-based monitoring continued
throughout the 2014-2016 Ebola virus epidemic when formal monitoring halted (Prof.
Richard Carter and Peter Dumble, personal communication, 13 September 2016) and
Burkina Faso where market gardeners plot groundwater levels in order to decide which
crops to grow that season, or which alternative activities to pursue in the case of low
levels, based on groundwater availability (Djibril Barry, WaterAid, personal
communication, 8 December 2016). The latter example shows that, ultimately,
communities conducting the monitoring can independently utilise the data for resource
management with no external influence.
The techniques applied during the recharge assessment (Chapter 6) are all transferable,
though again, not all must be applied in shallow groundwater assessment studies. It is
recommended, however, to apply as many methods as the available quality data allows
as then more can be learnt about the hydrogeological system from the discrepancies in
the results and greater confidence can be had over the final actual recharge range.
The degree of parameterisation and time taken for calibration and simulation means
SHETRAN modelling would not be feasible for shallow groundwater resource
assessment at every study site (Chapter 7). However, as shown here, if time and data
permit, SHETRAN can be used to show areas of greatest potential for productive
groundwater use and determine the resilience of the resource in the face of various future
scenarios (Chapter 8). The modelling results indicate that areas of groundwater potential
could feasibly be identified from topographic and geological mapping
9.5 Limitations of the study

This discussion will consider the chapters separately. The limitation of the methodology
being applied at a single study site is the subject of the discussion in the transferability
section of this chapter. It should be noted that many of the limitations presented below
may be considered opportunities for future work.
207
Limitations revealed by the literature review presented in Chapter 2
Literature review revealed that the widespread use of groundwater for irrigation in South
and East Asia suggested such growth could occur in SSA. However, as stated by
MacDonald et al. (2009), the rapid expansion in groundwater use experienced in parts of
Asia was only made possible through provision of cheap energy, credit and market
integration, which catalysed private investment in productive aquifers, such as the deep
sedimentary aquifer in Gujarat and the extensive basaltic aquifers of the Deccan (Kulkarni
et al., 2000). Policies in Asia in the 1960s supported intensification of small-scale
agriculture through massive public investment and price guarantees to raise smallholders’
incomes (Birner and Resnick, 2010). African governments today are dealing with greater
conditionality structures from structural adjustment policies set up in the 1980-90s and
African smallholders must deal with lower and more volatile prices (Dorward et al.,
2004). Additional factors compound the situation: Africa has traditionally seen a greater
diversity in crops, i.e. the green revolution in Asia was dominated by just two crops, rice
and wheat, and in Mexico by just maize and wheat (Mellor, 2014); SSA is more
vulnerable to climate change (Collier et al., 2008), and; physical infrastructure is far
inferior to Asia, leading to high input prices, notably fertiliser, and low output prices
(Rashid et al., 2013; Mellor, 2014).
At local-scale, there are numerous examples from SSA where the impact of smallholder
irrigation has not lived up to expectations. Comas et al. (2012) report a situation in
Mauritania where farmers resorted back to traditional rainfed and flood recession (beside
the Senegal River) cropping because the returns were exceeded by start-up loan
repayments and running costs. Considering drip irrigation systems introduced by NGOs,
Friedlander et al. (2013) conducted surveys in Ethiopia, Malawi, Senegal and Zambia and
found that 36% of farmers had abandoned the systems. The authors admit there is bias in
their sampling technique towards successful adoption and present findings by Kulecho
and Weatherhead (2005) from Kenya where 78% of farmers had abandoned drip
irrigation systems within two years and by Belder et al. (2007) from Zimbabwe where
68% of farmers had abandoned the systems within 1-2 years. Reasons for abandonment
included water shortage, effort of dismantling, storing and reassembling the system
according to season, and lack of funds for replacements following damage due to
vandalism and by animals (particularly common with subsistence farmers). The final
negative example comes from Fogera, east of Lake Tana in Ethiopia and approximately
110 km northeast of the Dangila study site, where pumps can be rented cheaply and used
208
autonomously, unlike river diversions and dams that are controlled by water user
committees (WUCs). Dessalegn and Merrey (2015) report that, seeing as there is no driver
for cooperation, over-abstraction has led to water shortages with some rivers running dry
in the dry season and a necessary shift from water intensive crops such as onions to grain.
It is clear from these few examples that uptake of small-scale groundwater irrigation alone
will not solve all problems of food insecurity and poverty. The important issues of
availability of credit, ensuring equality for women and other disadvantaged groups, strong
governance, and fluctuating crop prices are beyond the scope of this study. However, the
fundamental issues of shallow groundwater availability and resilience were investigated
and proved optimistic. The final example from Fogera in the previous paragraph bears
out the requirement for such investigations.
Chapter 4. Field investigations and development of conceptual model
The field investigations described in Chapter 4 that led to development of the conceptual
model were limited by the lack of exposures of the underlying geology. Basalt bedrock
was frequently visible in streambeds (Figure 4-5) though exposures of the weathered
regolith that constitutes the shallow aquifer were uncommon. Riverbank sections were
present but the floodplains where these sections existed may have a different composition
to regolith elsewhere. Borehole logs are unavailable, therefore, characterisation of the
regolith relied on inspections of the few wells under construction with adjacent as-dug
materials (Figure 4-4), community discussions recalling well excavations, and the
pumping tests conducted on hand-dug wells. Record keeping of well excavations with
simple logging (ease of excavation, colour and texture of material, depth of excavation)
would greatly aid future hydrogeological investigations.
In addition to this limitation of thorough understanding of aquifer properties,

understanding of aquifer geometry also had limitations. Estimates of aquifer thickness
were gained through extensive well depth and riverbank depth measurements. Such
measurements were applicable where wells were excavated to basalt bedrock and the river
had incised to basalt bedrock. In the case of wells, communities revealed that wells were
typically excavated until further excavation became impossible, i.e. the materials were
too strong to excavate, which was inferred as the base of the regolith aquifer. However,
logic and experience elsewhere suggest that wells excavated in areas with perennial water
supply would have been excavated to a point below the water table when water inflow
made further excavation problematic, dangerous, and unnecessary. This may or may not
209
coincide with the base of the aquifer. Local communities also spoke of wells being dry
though when dug less than 10 m away they hit water. As shown in Figure 4-6, this may
be due to residual basalt boulders within the regolith though it is likely to also be due to
a heterogenous surface of the basalt/regolith boundary, which would be more of a zone
than a boundary. Therefore, well depth measurements may not be ideal for estimating
aquifer thickness, however, it was the only available option. Geophysical investigations
could better estimate aquifer thickness. Two-, or ideally three-, dimensional electrical
resistivity tomography (ERT) surveys should differentiate the low resistance saturated
clay-rich regolith and the high resistance basalt bedrock. Although, resistivity results
obtained from GSE for surveys conducted in the Dangila area prior to commencement of
this PhD were inconclusive. GSE, partners in the AMGRAF catalyst project, could not
be persuaded to provide the raw geophysical data nor conduct further surveys.
Chapter 5. Filling the observational void: Scientific value and quantitative validation of
hydrometeorological data from a community-based monitoring programme
The limitation of the community-based monitoring described in Chapter 5 concerns the

groundwater level measurements and their consideration in conceptual and numerical
model development. Only five wells were monitored, and they were within a few 100 m
of each other. Ideally, for more confident extrapolation of water table response, many
more wells could be monitored across the catchment representing different topographical
and geological locations.
Chapter 6. Insights from a multi-method recharge estimation comparison study
A conclusion of Chapter 6 is that there is sufficient renewable shallow groundwater for a

sustainable intensification of small-scale irrigation. However, the study did not consider
potential future climate impacts on recharge. While studies applying GCMs and RCMs
generally predict increased rainfall totals for Ethiopia (e.g. Mitchell et al. (2004); IPCC
(2007)), this may not necessarily lead to increased recharge for two reasons: Firstly, these
same studies generally predict increasing rainfall intensity that could reduce recharge due
to enhanced runoff; secondly, the studies generally project rainfall increasing in the dry
season and lessening in the wet season (notwithstanding the increases in wet season
rainfall intensity) and a higher proportion of dry season rainfall would likely be lost to
evapotranspiration (Carter and Parker, 2009). This limitation of Chapter 6 is not
considered significant due to the relatively high recharge totals computed, i.e. the shallow
210
aquifer is not a marginally sustainable resource where lack of consideration of climate
change impacts would be a serious limitation.
Chapter 7. Modelling the shallow aquifer
The limitations of the modelling presented in Chapter 7 begin with the previously
discussed limitations on thorough understanding of aquifer properties and geometry.
Three hydrogeological categories were applied with particular properties and layer
thicknesses. This categorisation was selected for simplicity due to a lack of data for
assigning more categories. It is likely that the aquifer has some heterogeneity in properties
and in thickness that was not represented in the models. A similar limitation is the
assignment of the three LULC categories. Existing LULC mapping was not identified at
sufficiently high resolution for incorporation into the models (e.g. Setegn et al. (2011);
ADSWE (2015)) and other maps shared by local partner IWMI of uncertain origin).
Generating new LULC maps based on remote sensing was beyond the scope of this PhD.
The assignment of LULC categories based on field observations and slope analysis was
validated with comparisons with Google.Earth, however, there are areas within the
catchments that are not correctly represented, i.e. cultivated though flat areas and sloping
floodplains comprising pasture. Utilisation of MODIS or Sentinel imagery could provide
improved LULC mapping, though, as shown in Figures 8-1 and 8-2, LULC is dynamic at
the study site and a static map may not be representative for long. In addition to the
assignment of LULC categories, the vegetation properties applied within SHETRAN
were necessarily averaged for each category. The “Arable land” category comprises
vegetation with a range of properties (e.g. from onions to Eucalyptus), especially in areas
close to settlements where plots may be <0.1 ha and are often intercropped. Even high
resolution remote sensing imagery with extensive ground truthing may struggle to
provide LULC mapping across the catchments for which realistic vegetation properties
could be confidently assigned.
Concerning the time series inputs to the models described in Chapter 7, the scarcity of
rainfall and PET time series within the modelled catchments meant extrapolating rainfall
and PET measurements from a single point. This was not a significant limitation for the
smaller and more homogenous high resolution Brante and Amen models that were used
for the subsequent future scenarios analysis. However, the larger two Kilti models have
greater ranges of elevation and topography meaning the blanket rainfall and PET inputs
may have less accuracy in some areas of the catchments. Additionally, the
aforementioned localised groundwater level measurements used for calibration were a
211
limitation of the modelling; ideally, groundwater level time series spread across the
catchments would have been utilised.
A limitation on the conclusions of Chapter 7, that is the maps showing the potential for
an increase of small-scale irrigation, is the non-consideration of land ownership,
governance and the finance for and availability of irrigation equipment. These issues must
be addressed alongside issues of water resource availability to render the maps
practicable.
Chapter 8. Resilience of shallow groundwater resources
The fact that the future scenarios described in Chapter 8 do not consider climate change
may be considered a limitation. Even though it was shown that observed climate
variability was much greater than climate change projections, this variability could be
expected to become more extreme but by how much is not known (Smerdon, 2017).
The quite short daily rainfall time series available for Dangila and Bahir Dar were a
limitation as they meant that a synthetic long period time series had to be developed for
future scenario simulation. If longer time series would have been available, then the return
period of agricultural and hydrological droughts could have been estimated.
It is likely that the LULC change scenarios will occur in tandem though this was not
simulated. However, this is considered only a minor limitation as the combined effects of
LULC changes can be estimated from the results, i.e. conversion of pasture to cultivated
land combined with increased Eucalyptus planting would significantly negatively affect
water resources. Similarly, the non-simulation of simultaneous LULC change and
increased abstraction is not considered a limitation due to the lack of significant impacts
from the abstraction scenarios. Detailed analysis of historic aerial or satellite imagery
could be used to estimate the rate of different LULC changes and the Asia example could
be further analysed to assess the rate of irrigation uptake. However, simulation of these
combined changes would still be uncertain going forward as the rates and type of LULC
change and irrigation uptake will vary according to various factors such as crop and fuel
prices, government policy, climate, population growth, etc.
A further limitation in the future scenarios modelling was exposed by the uncertainty over
accurate simulation of the cropping cycle, i.e. different crop growth stages are not
represented. This could be reduced with the implementation of a crop model within the
SHETRAN program that would provide greater confidence in the conclusion that small-
212
scale abstraction and irrigation will not negatively impact shallow groundwater and
surface water resources.
Summary of limitations
It should be noted that most of these limitations are omnipresent with hydrogeological
and hydrological modelling: aquifer properties and dimensions are never certain,
vegetation properties are always variable, time series data for model set up and calibration
are always fewer than desirable, climate change effects are uncertain, and socioeconomic
factors typically complicate conclusions. Some of these limitations could be reduced with
further work that was beyond the scope or financial means of this PhD. Other limitations
would be difficult to resolve and must be considered when making recommendations
based on this study.
9.6 Future work

Future work would involve testing the research findings at other field sites in Ethiopia
and elsewhere. The topographic criteria identified in Dangila for areas of high potential
for abstracting groundwater and the criteria for low potential areas could be tested
thoroughly with a modelling assessment as carried out during this research though it may
be enough to conduct water point surveys during a field hydrogeological assessment and
query local stakeholders. In addition to testing the findings in areas of similar young
volcanic geology, it would be enlightening to conduct comparative research in areas of
similar climate but crystalline bedrock to determine if regolith hydrogeology really is
independent of bedrock age and type. If this is the case, then the transferability of this
PhD research widens substantially. To further target the research gap concerning shallow
aquifers, in addition to research of regolith aquifers above crystalline basement, other
common shallow aquifer types could be targeted for research such as unconsolidated
coastal and river valley aquifers, and coastal limestone aquifers. These studies should be
conducted at community scale; the scale at which understanding the potential and
resilience of the groundwater resource could have a direct benefit to poor rural
communities.
As climate varies, land use changes and abstraction increases, the community
hydrometeorological monitoring data can confirm or otherwise the resilience of the
shallow groundwater and surface water resources, validating the modelling. It is
important to continue this monitoring with regular feedback of results to the local
community. Research in East Africa by Comte et al. (2016) indicates that local
213
stakeholders are critical of researchers who commonly do not share project findings. The
study also found that communities want to be actively involved in water resource
management. Therefore, the hope is that the Dangesheta community become independent
with the monitoring and utilisation of the monitoring data as occurred in the Burkina Faso
example described in the transferability section of this chapter.
214
References
Abate, S. (1994) 'Land use dynamics, soil degradation and potential for sustainable use
in Metu area, Illubabor region, Ethiopia', in Geographica Bernensia, African
Studies Series A 13. University of Bern, Switzerland.
Abbott, M.B., Bathurst, J.C., Cunge, J.A., O'Connell, P.E. and Rasmussen, J. (1986a) 'An
introduction to the European Hydrological System—Systeme Hydrologique
Europeen,“SHE”, 1: History and philosophy of a physically-based, distributed
modelling system', Journal of hydrology, 87(1-2), pp. 45-59.
Abbott, M.B., Bathurst, J.C., Cunge, J.A., O'Connell, P.E. and Rasmussen, J. (1986b) 'An
introduction to the European Hydrological System—Systeme Hydrologique
Europeen,“SHE”, 2: Structure of a physically-based, distributed modelling system',
Journal of hydrology, 87(1-2), pp. 61-77.
Abbott, S.P. (2013) Hand dug wells: choice of technology and construction manual.
World Bank Group, Washington DC, USA, 59pp.
Abdella, A. (2011) Sustainable groundwater development and management for irrigation

in Raya and Kobo Valleys, northern Ethiopia. MSc thesis, Department of Earth
Sciences, Addis Ababa University, Ethiopia.
Abebe, Y.D. and Geheb, K.E. (2003) 'Wetlands of Ethiopia. Proceedings of a seminar on
the resources and status of Ethiopia's wetlands', IUCN Wetlands and Water
Resources Programme, Gland, Switzerland, 136pp.
Abera, M. (2017) 'Agriculture in the Lake Tana Sub-basin of Ethiopia', in Social and
Ecological System Dynamics. Springer, New York, USA, pp. 375-397.
Abid, M. and Lal, R. (2009) 'Tillage and drainage impact on soil quality: II. Tensile
strength of aggregates, moisture retention and water infiltration', Soil and Tillage
Research, 103(2), pp. 364-372.
Abiy, A.Z., Demissie, S.S., MacAlister, C., Dessu, S.B. and Melesse, A.M. (2016)
'Groundwater Recharge and Contribution to the Tana Sub-basin, Upper Blue Nile
Basin, Ethiopia. ', in Landscape Dynamics, Soils and Hydrological Processes in
Varied Climates. Springer, pp. 463-481.
215
Abiyu, A., Teketay, D., Gratzer, G. and Shete, M. (2016) 'Tree planting by smallholder
farmers in the upper catchment of Lake Tana Watershed, Northwest Ethiopia',
Small-scale Forestry, 15(2), pp. 199-212.
Abric, S., Sonou, M., Augegard, B., Onimus, F., Durlin, D., Soumaila, A. and Gadelle, F.
(2011) Lessons learned in the development of smallholder private irrigation for
high-value crops in West Africa. World Bank Group, Washington DC, USA, 59pp.
Acworth, R.I. (1987) 'The development of crystalline basement aquifers in a tropical

environment', Quarterly Journal of Engineering Geology and Hydrogeology, 20(4),
pp. 265-272.
Adams, W.M. (1993) 'Indigenous use of wetlands and sustainable development in West
Africa', Geographical journal, pp. 209-218.
Adams, W.M. and Carter, R.C. (1987) 'Small-scale irrigation in sub-Saharan Africa',
Progress in Physical Geography, 11(1), pp. 1-27.
Addis, H.K., Strohmeier, S., Ziadat, F., Melaku, N.D. and Klik, A. (2016) 'Modeling
streamflow and sediment using SWAT in Ethiopian Highlands', International
Journal of Agricultural and Biological Engineering, 9(5), pp. 51-66.
Addisu, D. (2012) Hydrogeochemical and isotope hydrology in investigating

groundwater recharge and flow processes, Lower Afar, Eastern Ethiopia. MSc
thesis, Department of Earth Sciences, Addis Ababa University, Ethiopia.
Adelman, I. (1984) 'Beyond export-led growth', World development, 12(9), pp. 937-949.
Adler, R., Braun, S., Stocker, E. and Marius, J. (2007) 'Tropical Rainfall Measuring
Mission, TRMM, Senior Review Proposal', Laboratory for Atmospheres, NASA
Goddard Space Flight Centre, Greenbelt, MD, USA.
ADSWE (2015) Technical Report: Land Use Land Cover and Change Detection,
Amhara National Regional State, Bureau of Environmental Protection, Land
Administration and Use (BoEPLAU). . Tana Sub Basin Integrated Land Use
Planning and Environmental Impact Study Project, Amhara Design & Supervision
Works Enterprise (ADSWE), Bahir Dar, Ethiopia, 71pp.
216
Aeschbach-Hertig, W. and Gleeson, T. (2012) 'Regional strategies for the accelerating
global problem of groundwater depletion', Nature Geoscience, 5(12), pp. 853-861.
Afrifa, G.Y., Sakyi, P.A. and Chegbeleh, L.P. (2017) 'Estimation of groundwater recharge
in sedimentary rock aquifer systems in the Oti basin of Gushiegu District, Northern
Ghana', Journal of African Earth Sciences, 131(Supplement C), pp. 272-283.
AgroBIG (2016) Programme for Agribusiness induced growth in the Amhara region:
Rice Value Chain Analysis. Bahir Dar, Ethiopia, 36pp.
Akpan, A.E., Ugbaja, A.N. and George, N.J. (2013) 'Integrated geophysical, geochemical
and hydrogeological investigation of shallow groundwater resources in parts of the
Ikom-Mamfe Embayment and the adjoining areas in Cross River State, Nigeria',
Environmental earth sciences, 70(3), pp. 1435-1456.
Akteruzzaman, M., Kondo, T., Osanami, F. and Doi, T. (1998) 'Structure of the
groundwater market in Bangladesh: a case study of tubewell irrigation investment',
International Journal of Water Resources Development, 14(1), pp. 55-65.
Albaugh, J.M., Dye, P.J. and King, J.S. (2013) 'Eucalyptus and water use in South Africa',
International Journal of Forestry Research, 2013, pp. 1-13.
Alebachew, M., Amare, T. and Wendie, M. (2015) 'Investigation of the Effects of

Eucalyptus camaldulensis on Performance of Neighbouring Crop Productivity in
Western Amhara, Ethiopia', Open Access Library Journal, 2(03), p. 1.
Alemayehu, T., Furi, W. and Legesse, D. (2007) 'Impact of water overexploitation on

highland lakes of eastern Ethiopia', Environmental geology, 52(1), pp. 147-154.
Alemayehu, T., McCartney, M. and Kebede, S. (2010) 'The water resource implications
of planned development in the Lake Tana catchment, Ethiopia', Ecohydrology &
Hydrobiology, 10(2), pp. 211-221.
Alene, Y. (2006) GIS and remote sensing assisted water balance computation of the Geba
basin, Tigray, Ethiopia. MSc thesis, Water Resources Engineering, KULeuven -
VUB, Belgium.
Ali, H., Descheemaeker, K., Steenhuis, T.S. and Pandey, S. (2011) 'Comparison of land
use and land cover changes, drivers and impacts for a moisture-sufficient and
217
drought-prone region in the Ethiopian highlands', Experimental Agriculture, 47, pp.
71-83.
Ali Jr, K.A. (2006) 'Groundwater resources assessment in hard rock terrain using
conventional remote sensing and GIS approach in Southern peripheral part of
Ethiopia.', GSA (Geological Society of America) Philadelphia Annual Meeting, 22-
25 October 2006. Philadelphia, USA.
Ali, R., McFarlane, D., Varma, S., Dawes, W., Emelyanova, I., Hodgson, G. and Charles,
S. (2012) 'Potential climate change impacts on groundwater resources of south-
western Australia', Journal of Hydrology, 475(Supplement C), pp. 456-472.
All Africa (2016) Uganda: Government to Build Shs4 Trillion Irrigation Scheme in
Pallisa. Available at: http://allafrica.com/stories/201612300007.html (Accessed: 7
January 2017).
Allam, M.M., Jain Figueroa, A., McLaughlin, D.B. and Eltahir, E.A.B. (2016) 'Estimation
of evaporation over the upper Blue Nile basin by combining observations from
satellites and river flow gauges', Water Resources Research, 52(2), pp. 644-659.
Allen, R.G., Pereira, L.S., Raes, D. and Smith, M. (1998) 'Crop evapotranspiration -
Guidelines for computing crop water requirements (FAO Irrigation and Drainage
Paper 56)', Food and Agriculture Organisation of the United Nations, Rome, Italy.
Allison, G.B. (1988) 'A review of some of the physical, chemical and isotopic techniques
available for estimating groundwater recharge', in Simmers, I. (ed.) Estimation of
natural groundwater recharge, Springer, Netherlands: 49-72. Springer,
Netherlands: 49-72.
Altchenko, Y. and Villholth, K.G. (2015) 'Mapping irrigation potential from renewable
groundwater in Africa - a quantitative hydrological approach', Hydrology and Earth
System Sciences, 19(2), pp. 1055-1067.
Amezaga, J. (2014) Preliminary assessment of policies, regulations and institutions

required for management of shallow groundwater at local community level in
Dangila woreda, Ethiopia. Available at: https://research.ncl.ac.uk/amgraf/outputs/
(Accessed: 5 June 2017).
218
Amogu, O., Descroix, L., Yéro, K.S., Le Breton, E., Mamadou, I., Ali, A., Vischel, T.,
Bader, J.-C., Moussa, I.B. and Gautier, E. (2010) 'Increasing river flows in the
Sahel?', Water, 2(2), pp. 170-199.
Amsalu, A., Stroosnijder, L. and de Graaff, J. (2007) 'Long-term dynamics in land

resource use and the driving forces in the Beressa watershed, highlands of Ethiopia',
Journal of Environmental management, 83(4), pp. 448-459.
Andualem, E. (2008) Evaluation of Ground Water Resources Potential of the Teji River
Cathment South West Shoa Zone, Oromia Region. MSc thesis, School of Earth
Aplin, A.C., Fleet, A.J. and Macquaker, J.H.S. (1999) 'Muds and mudstones: Physical
and fluid-flow properties', Geological Society, London, Special Publications,
158(1), pp. 1-8.
Arino, O., Ramos Perez, J.J., Kalogirou, V., Bontemps, S., Defourny, P. and Van Bogaert,
E. (2012) 'Global Land Cover Map for 2009 (GlobCover 2009)'. PANGAEA ©
European Space Agency (ESA) & Université catholique de Louvain (UCL).
Available at: https://doi.org/10.1594/PANGAEA.787668.
Arnell, N.W. (2004) 'Climate change and global water resources: SRES emissions and
socio-economic scenarios', Global environmental change, 14(1), pp. 31-52.
Asmerom, G.H. (2008) Groundwater contribution and recharge estimation in the Upper
Blue Nile flows, Ethiopia. MSc thesis, ITC, Enschede, The Netherlands, 133pp.
Asrie, N.A. and Sebhat, M.Y. (2016) 'Numerical groundwater flow modeling of the
northern river catchment of the Lake Tana, Upper Blue Basin, Ethiopia', Journal of
Agriculture and Environment for International Development, 110(1), pp. 5-26.
Awulachew, S.B., Yilma, A.D., Loulseged, M., Loiskandl, W., Ayana, M. and Alamirew,
T. (2007) Water resources and irrigation development in Ethiopia. Working Paper
123. International Water Management Institute (IWMI), Colombo, Sri Lanka, 78pp.
Ayalew, D. (2011) 'The relations between felsic and mafic volcanic rocks in continental
flood basalts of Ethiopia: implication for the thermal weakening of the crust',
Geological Society, London, Special Publications, 357(1), pp. 253-264.
219
Ayalew, D., Tesfaye, K., Mamo, G., Yitaferu, B. and Bayu, W. (2012) 'Variability of
rainfall and its current trend in Amhara region, Ethiopia', Afr J Agric Res, 7(10), pp.
1475-1486.
Ayalew, H. (2010) Geological framework for groundwater occurence in Lake Tana

basin, northwestern Ethiopia, Amhara region. MSc thesis, Department of Earth
Aychluhim, D. (2006) Integrated water resources potential investigation of the Weybo

River Catchment Welayita-Hadiya Zones, Southern Ethiopia. MSc thesis,
Department of Earth Sciences, Addis Ababa University, Ethiopia.
Ayenew, T. (2007) 'Water management problems in the Ethiopian rift: Challenges for
development', Journal of African Earth Sciences, 48(2), pp. 222-236.
Ayenew, T. (2008) 'Hydrological system analysis and groundwater recharge estimation

using semi-distributed models and river discharge in the Meki River Basin', SINET:
Ethiopian Journal of Science, 31(1), pp. 29-42.
Ayenew, T., Demlie, M. and Wohnlich, S. (2008a) 'Application of numerical modeling

for groundwater flow system analysis in the Akaki catchment, central Ethiopia',
Mathematical geosciences, 40(8), pp. 887-906.
Ayenew, T., Demlie, M. and Wohnlich, S. (2008b) 'Hydrogeological framework and

occurrence of groundwater in the Ethiopian aquifers', Journal of African Earth
Sciences, 52(3), pp. 97-113.
Ayenew, T., GebreEgziabher, M., Kebede, S. and Mamo, S. (2013) 'Integrated

assessment of hydrogeology and water quality for groundwater-based irrigation
development in the Raya Valley, northern Ethiopia', Water International, 38(4), pp.
480-492.
Ayenew, T. and Tilahun, N. (2008) 'Assessment of lake–groundwater interactions and

anthropogenic stresses, using numerical groundwater flow model, for a Rift lake
catchment in central Ethiopia', Lakes & Reservoirs: Research & Management,
13(4), pp. 325-343.
220
Azagegn, T., Asrat, A., Ayenew, T. and Kebede, S. (2015) 'Litho-structural control on
interbasin groundwater transfer in central Ethiopia', Journal of African Earth
Sciences, 101, pp. 383-395.
Baguma, A., Bizoza, A., Carter, R., Cavill, S., Foster, S., Foster, T., Jobbins, G., Hope,
R., Katuva, J. and Koehler, J. (2017) 'Groundwater and poverty in sub Saharan
Africa a short investigation highlighting outstanding knowledge gaps', UPGro
working paper, Skat Foundation, St. Gallen, Switzerland.
Baker, B.H., Mohr, P.A. and Williams, L.A.J. (1972) 'Geology of the eastern rift system
of Africa', Geological Society of America Special Papers, 136, pp. 1-68.
Bakundukize, C., Van Camp, M. and Walraevens, K. (2011) 'Estimation of groundwater

recharge in Bugesera region (Burundi) using soil moisture budget approach',
Geologica belgica, 14(1-2), pp. 85-102.
Barker, J.A. and Herbert, R. (1989) 'Nomograms for the analysis of recovery tests on
large-diameter wells', Quarterly Journal of Engineering Geology, 22(2), pp. 151-
158.
Bates, B., Kundzewicz, Z., Wu, S. and Palutikof, J. (2008) 'IPCC: Climate change and
water', IPCC Working Group II, Technical Paper of the Intergovernmental Panel
on Climate Change, IPCC Secretariat, Geneva, Switzerland, 210pp.
Bathurst, J.C. (2011) 'Predicting impacts of land use and climate change on erosion and
sediment yield in river basins using SHETRAN', in Morgan, R.P.C. and Nearing,
M. (eds.) Handbook of Erosion Modelling. John Wiley & Sons, New Jersey, USA,
pp. 263-288.
Bathurst, J.C., Birkinshaw, S.J., Cisneros, F., Fallas, J., Iroume, A., Iturraspe, R., Novillo,
M.G., Urciuolo, A., Alvarado, A., Coello, C., Huber, A., Miranda, M., Ramirez, M.
and Sarandon, R. (2011a) 'Forest impact on floods due to extreme rainfall and
snowmelt in four Latin American environments 2: Model analysis', Journal of
Hydrology, 400(3-4), pp. 292-304.
Bathurst, J.C., Burton, A., Clarke, B.G. and Gallart, F. (2006) 'Application of the
SHETRAN basin‐scale, landslide sediment yield model to the Llobregat basin,
Spanish Pyrenees', Hydrological Processes, 20(14), pp. 3119-3138.
221
Bathurst, J.C., Ewen, J., Parkin, G., O'Connell, P.E. and Cooper, J.D. (2004) 'Validation
of catchment models for predicting land-use and climate change impacts. 3. Blind
validation for internal and outlet responses', Journal of Hydrology, 287(1), pp. 74-
94.
Bathurst, J.C., Iroumé, A., Cisneros, F., Fallas, J., Iturraspe, R., Novillo, M.G., Urciuolo,
A., De Bièvre, B., Borges, V.G. and Coello, C. (2011b) 'Forest impact on floods
due to extreme rainfall and snowmelt in four Latin American environments 1: Field
data analysis', Journal of Hydrology, 400(3), pp. 281-291.
Batjes, N.H. (1997) 'A world dataset of derived soil properties by FAO–UNESCO soil
unit for global modelling', Soil use and management, 13(1), pp. 9-16.
Bazuhair, A.S. and Wood, W.W. (1996) 'Chloride mass-balance method for estimating
ground water recharge in arid areas: examples from western Saudi Arabia', Journal
of Hydrology, 186(1), pp. 153-159.
BBC (2016) DR Congo floods leave 50 dead in Boma. Available at:

http://www.bbc.co.uk/news/world-africa-38458404 (Accessed: 7 January 2017).
BCEOM (1998) Abbay Basin Master Plan Phase 2 Sectoral Studies, Part 3,
Hydrogeology, Main Report. BCEOM French Engineering Consultants: Ministry
of Water Resources (MoWR), Addis Ababa, Ethiopia.
BCEOM (2005) Harar water supply project groundwater resources. BCEOM French
Engineering Consultants: Ministry of Water Resources (MoWR), Addis Ababa,
Ethiopia.
Belay, E.A. (2009) Growing lake with growing problems: integrated hydrogeological
investigation on Lake Beseka, Ethiopia. PhD thesis, The Faculty of Mathematics
and Natural Sciences, University of Bonn, Germany.
Belay, M. and Bewket, W. (2013) 'Traditional irrigation and water management practices
in highland Ethiopia: Case study in Dangila woreda', Irrigation and Drainage,
62(4), pp. 435-448.
Belder, P., Rohrbach, D., Twomlow, S.J. and Senzanje, A. (2007) 'Can drip irrigation
improve the livelihoods of smallholders? Lessons learned from Zimbabwe', Global
222
Theme on Agroecosystems Report no. 33, Monograph. International Crops
Research Institute for the Semi-Arid Tropics, Bulawayo, Zimbabwe, 33pp.
Benvenuti, M., Carnicelli, S., Belluomini, G., Dainelli, N., Di Grazia, S., Ferrari, G.A.,
Iasio, C., Sagri, M., Ventra, D., Atnafu, B. and Kebede, S. (2002) 'The Ziway–Shala
lake basin (main Ethiopian rift, Ethiopia): a revision of basin evolution with special
reference to the Late Quaternary', Journal of African Earth Sciences, 35(2), pp. 247-
269.
Berehanu, B., Azagegn, T., Ayenew, T. and Masetti, M. (2017) 'Inter-Basin Groundwater
Transfer and Multiple Approach Recharge Estimation of the Upper Awash Aquifer
System', Journal of Geoscience and Environment Protection, 5(03), p. 76.
Berhail, S., Ouerdachi, L. and Keblouti, M. (2015) 'Combination of two evolutionary

methods for groundwater recharges estimation in semi-arid regions, Northeastern
Algeria', International Journal of Water, 9(3), pp. 286-301.
Bernard, B., Vincent, K., Waema, T.M. and Macopiyo, L. (?) 'Streamflow responses to
changes in land use and climate in a tropical catchment: Malaba River Catchment,
Eastern Uganda', Unknown, available at
http://www.iwra.org/congress/resource/3164380.pdf.
Bertin, C. and Bourg, A.C.M. (1994) 'Radon-222 and Chloride as Natural Tracers of the
Infiltration of River Water into an Alluvial Aquifer in Which There Is Significant
River/Groundwater Mixing', Environmental Science & Technology, 28(5), pp. 794-
798.
Beven, K. (1993) 'Prophecy, reality and uncertainty in distributed hydrological

modelling', Advances in water resources, 16(1), pp. 41-51.
Beven, K.J., Kirkby, M.J., Schofield, N. and Tagg, A.F. (1984) 'Testing a physically-
based flood forecasting model (TOPMODEL) for three UK catchments', Journal of
Hydrology, 69(1-4), pp. 119-143.
Beven, K.J. and O'Connell, P.E. (1982) 'On the role of physically-based distributed
modelling in hydrology', Inst. Hydrol. Rep. No. 81. Wallingford, UK.
223
Beven, K.J., Warren, R. and Zaoui, J. (1980) 'SHE: towards a methodology for
physically-based distributed forecasting in hydrology', IAHS Publ, 129, pp. 133-
137.
Bewket, W. and Conway, D. (2007) 'A note on the temporal and spatial variability of
rainfall in the drought‐prone Amhara region of Ethiopia', International Journal of
Climatology, 27(11), pp. 1467-1477.
BGR (2008) Groundwater and Climate Change: Challenges and Possibilities. Policy
Advice Groundwater - Resources and Management, German Federal Ministry for
Economic Cooperation and Development, Hanover, Germany, 20pp.
BGS (2018) Africa Groundwater Atlas: Hydrogeology of Ethiopia. Available at:

http://earthwise.bgs.ac.uk/index.php/Hydrogeology_of_Ethiopia (Accessed: 14
February 2018).
Bhatia, R. (1991) 'Irrigation financing and cost recovery policy in India: case studies from
Bihar and Haryana', in Meinzen-Dick, R.S. and Svendsen, M. (eds.) Future
directions for Indian irrigation: research and policy issues. International Food
Policy Research Institute, Washington DC, USA, pp. 168-213.
Bhattarai, M., Sakthivadivel, R. and Hussain, I. (2001) Irrigation impacts on income

inequality and poverty alleviation: Policy issues and options for improved
management of irrigation systems. Working Paper 39, International Water
Management Institute (IWMI), Colombo, Sri Lanka, 37pp.
Birkinshaw, S.J. and Ewen, J. (2000a) 'Modelling nitrate transport in the Slapton Wood
catchment using SHETRAN', Journal of hydrology, 230(1), pp. 18-33.
Birkinshaw, S.J. and Ewen, J. (2000b) 'Nitrogen transformation component for

SHETRAN catchment nitrate transport modelling', Journal of Hydrology, 230(1-
2), pp. 1-17.
Birner, R. and Resnick, D. (2010) 'The political economy of policies for smallholder
agriculture', World Development, 38(10), pp. 1442-1452.
Blasone, R.S., Madsen, H. and Rosbjerg, D. (2007) 'Parameter estimation in distributed

hydrological modelling: comparison of global and local optimisation techniques',
Hydrology Research, 38(4-5), pp. 451-476.
224
Blazkova, S. and Beven, K. (2009) 'A limits of acceptability approach to model evaluation
and uncertainty estimation in flood frequency estimation by continuous simulation:
Skalka catchment, Czech Republic', Water Resources Research, 45(12).
Blodgett, J.C., Walters, J.R. and Borchers, J.W. (1992) Streamflow gains and losses and
selected flow characteristics of Cottonwood Creek, north-central California, 1982-
85. US Department of the Interior, US Geological Survey.
Bonsor, H.C. and MacDonald, A.M. (2010) 'Groundwater and climate change in Africa:
review of recharge studies', British Geological Survey Internal Report, IR/10/075,
p. 30pp.
Bonsor, H.C. and MacDonald, A.M. (2011) An initial estimate of depth to groundwater
across Africa. British Geological Survey Open Report OR/11/067, Wallingford,
UK, 26pp.
Bonsor, H.C., MacDonald, A.M. and Calow, R.C. (2010) 'Potential impact of climate
change on improved and unimproved water supplies in Africa', RSC Issues in
Environmental Science and Technology, (31), pp. 25-50.
Bonsor, H.C., MacDonald, A.M. and Davies, J. (2014) 'Evidence for extreme variations
in the permeability of laterite from a detailed analysis of well behaviour in Nigeria',
Hydrological Processes, 28(10), pp. 3563-3573.
Bovolo, C.I., Parkin, G. and Sophocleous, M. (2009) 'Groundwater resources, climate and
vulnerability', Environmental Research Letters, 4(3), p. 4.
Bredehoeft, J.D. (2002) 'The water budget myth revisited: why hydrogeologists model',
Groundwater, 40(4), pp. 340-345.
Bromley, J., Butterworth, J.A., Macdonald, D.M.J., Lovell, C.J., Mharapara, I. and
Batchelor, C.H. (1999) 'Hydrological processes and water resources management
in a dryland environment I: An introduction to the Romwe Catchment Study in
Southern Zimbabwe', Hydrology and Earth System Sciences, 3(3), pp. 322-332.
Brouwer, C., Hoevenaars, J.P.M., Van Bosch, B.E., Hatcho, N. and Heibloem, M. (1992)
'Irrigation Water Management Training Manual no. 6-Scheme Irrigation Water
Needs and Supply', Food and Agriculture Organisation of the United Nations,
Rome, Italy.
225
Brown, R.H. (1963) 'The cone of depression and the area of diversion around a
discharging well in an infinite strip aquifer subject to uniform recharge', US
Geological Survey Water-Supply Paper C, 1545, pp. C69-C85.
Bullock, A. (1992) 'Dambo hydrology in southern Africa—review and reassessment',

Journal of Hydrology, 134(1), pp. 373-396.
Buontempo, C., Mathison, C., Jones, R., Williams, K., Wang, C. and McSweeney, C.
(2014) 'An ensemble climate projection for Africa', Climate Dynamics, pp. 1-22.
Burgos, A., Páez, R., Carmona, E. and Rivas, H. (2013) 'A systems approach to modeling
Community-Based Environmental Monitoring: a case of participatory water quality
monitoring in rural Mexico', Environmental Monitoring and Assessment, 185(12),
pp. 10297-10316.
Burke, K. and Gunnell, Y. (2008) The African erosion surface: a continental-scale

synthesis of geomorphology, tectonics, and environmental change over the past 180
million years. Geological Society of America, Colorado, USA, 201pp.
Burton, A. and Bathurst, J.C. (1998) 'Physically based modelling of shallow landslide
sediment yield at a catchment scale', Environmental Geology, 35(2-3), pp. 89-99.
Butler, M.J. and Verhagen, B.T. (2001) 'Isotope studies of a thick unsaturated zone in a
semi-arid area of Southern Africa', Isotope Based Assessment of Groundwater
Renewal in Water Scarce Regions, IAEA-Tecdoc-1246, pp. 45-70.
Butterworth, J., Macdonald, D.M.J., Bromley, J., Simmonds, L.P., Lovelll, C.J. and
Mugabe, F. (1999) 'Hydrological processes and water resources management in a
dryland environment III: Groundwater recharge and recession in a shallow
weathered aquifer', Hydrology and Earth System Sciences, 3(3), pp. 345-352.
Buytaert, W., Zulkafli, Z., Grainger, S., Acosta, L., Bastiaensen, J., De Bièvre, B., Bhusal,
J., Chanie, T., Clark, J., Dewulf, A., Foggin, M., Hannah, D.M., Hergarten, C.,
Isaeva, A., Karpouzoglou, T., Pandey, B., Paudel, D., Sharma, K., Steenhuis, T.,
Tilahun, S., Van Hecken, G. and Zhumanova, M. (2014) 'Citizen science in
hydrology and water resources: opportunities for knowledge generation, ecosystem
service management, and sustainable development', Frontiers in Earth Science, 2.
226
Cain, A. (2015) 'Water resource management under changing climate in Angola's coastal
settlements', Presented during Special Session 10 at the World Water Congress XV
(May 2015). Edinburgh, UK.
Cain, J.D. (1998) Modelling evaporation from plant canopies. Institute of Hydrology
Report no. 132, Wallingford, UK, 60pp.
Calder, I.R., Hall, R.L. and Prasanna, K.T. (1993) 'Hydrological impact of Eucalyptus
plantation in India', Journal of Hydrology, 150(2), pp. 635-648.
Calow, R.C., Howarth, S.E. and Wang, J.X. (2009a) 'Irrigation Development and Water
Rights Reform in China', International Journal of Water Resources Development,
25(2), pp. 227-248.
Calow, R.C., MacDonald, A.M., Nicol, A.L. and Robins, N.S. (2009b) 'Ground Water
Security and Drought in Africa: Linking Availability, Access, and Demand',
Ground Water, 48(2), pp. 246-256.
Calow, R.C., Robins, N.S., MacDonald, A.M., MacDonald, D.M.J., Gibbs, B.R., Orpen,
W.R.G., Mtembezeka, P., Andrews, A.J. and Appiah, S.O. (1997) 'Groundwater
management in drought-prone areas of Africa', International Journal of Water
Resources Development, 13(2), pp. 241-262.
Calver, A., Stewart, E. and Goodsell, G. (2009) 'Comparative analysis of statistical and
catchment modelling approaches to river flood frequency estimation', Journal of
flood risk management, 2(1), pp. 24-31.
Canadell, J., Jackson, R.B., Ehleringer, J.B., Mooney, H.A., Sala, O.E. and Schulze, E.D.
(1996) 'Maximum rooting depth of vegetation types at the global scale', Oecologia,
108(4), pp. 583-595.
Carter, R.C. and Alkali, A.G. (1996) 'Shallow groundwater in the northeast arid zone of
Nigeria', Quarterly Journal of Engineering Geology and Hydrogeology, 29(4), pp.
341-355.
Carter, R.C. and Bevan, J.E. (2008) 'Groundwater development for poverty alleviation in
sub-Saharan Africa', Applied groundwater studies in Africa, 13, pp. 25-42.
227
Carter, R.C. and Parker, A. (2009) 'Climate change, population trends and groundwater
in Africa', Hydrological Sciences Journal, 54(4), pp. 676-689.
Cawsey, D.C. and Mellon, P. (1983) 'A review of experimental weathering of basic
igneous rocks', Geological Society, London, Special Publications, 11(1), pp. 19-24.
CCME (2011) Protocols manual for water quality sampling in Canada. Canadian
Council of Ministers of the Environment, Winnipeg, Canada.
Chanie, T., Collick, A.S., Adgo, E., Lehmann, C.J. and Steenhuis, T.S. (2013) 'Eco-
hydrological impacts of Eucalyptus in the semi humid Ethiopian Highlands: the
Lake Tana Plain', Journal of Hydrology and Hydromechanics, 61(1), pp. 21-29b.
Chapman, T. (1999) 'A comparison of algorithms for stream flow recession and baseflow
separation', Hydrological Processes, 13(5), pp. 701-714.
Chebud, Y.A. and Melesse, A.M. (2009) 'Numerical modeling of the groundwater flow
system of the Gumera sub‐basin in Lake Tana basin, Ethiopia', Hydrological
processes, 23(26), pp. 3694-3704.
Chilton, P.J. and Smith-Carington, A.K. (1984) 'Characteristics of the weathered

basement aquifer in Malawi in relation to rural water supplies', in Challenges in
African Hydrology and Water Resources (Proc. Harare Symposium July 1984).
IAHS Publ. no. 144, pp. 57-72.
Chishugi, J.B. and Alemaw, B.F. (2009) World Environmental and Water Resources
Congress 2009: Great Rivers.
Choi, I.-H., Woo, N.-C., Kim, S.-J., Moon, S.-K. and Kim, J. (2007) 'Estimation of the
groundwater recharge rate during a rainy season at a headwater catchment in
Gwangneung, Korea', Korean Journal of Agricultural and Forest Meteorology,
9(2), pp. 75-87.
Chorowicz, J., Collet, B., Bonavia, F.F., Mohr, P., Parrot, J.F. and Korme, T. (1998) 'The
Tana basin, Ethiopia: intra-plateau uplift, rifting and subsidence', Tectonophysics,
295(3), pp. 351-367.
228
Chung, I.M., Sophocleous, M.A., Mitiku, D.B. and Kim, N.W. (2016) 'Estimating
groundwater recharge in the humid and semi-arid African regions: review',
Geosciences Journal, 20(5), pp. 731-744.
Coes, A.L., Spruill, T.B. and Thomasson, M.J. (2007) 'Multiple-method estimation of
recharge rates at diverse locations in the North Carolina Coastal Plain, USA',
Hydrogeology Journal, 15(4), pp. 773-788.
Collier, P., Conway, G. and Venables, T. (2008) 'Climate change and Africa', Oxford
Review of Economic Policy, 24(2), pp. 337-353.
Collier, P. and Dercon, S. (2014) 'African Agriculture in 50Years: Smallholders in a

Rapidly Changing World?', World development, 63, pp. 92-101.
Comas, J., Connor, D., Isselmou, M.E.M., Mateos, L. and Gómez-Macpherson, H. (2012)
'Why has small-scale irrigation not responded to expectations with traditional
subsistence farmers along the Senegal River in Mauritania?', Agricultural Systems,
110(0), pp. 152-161.
Comte, J.-C., Cassidy, R., Obando, J., Robins, N., Ibrahim, K., Melchioly, S., Mjemah,
I., Shauri, H., Bourhane, A., Mohamed, I., Noe, C., Mwega, B., Makokha, M., Join,
J.-L., Banton, O. and Davies, J. (2016) 'Challenges in groundwater resource
management in coastal aquifers of East Africa: Investigations and lessons learnt in
the Comoros Islands, Kenya and Tanzania', Journal of Hydrology: Regional
Studies, 5, pp. 179-199.
Conan, C., Bouraoui, F., Turpin, N., de Marsily, G. and Bidoglio, G. (2003) 'Modeling
flow and nitrate fate at catchment scale in Brittany (France)', Journal of
Environmental Quality, 32(6), pp. 2026-2032.
Conners, D.E., Eggert, S., Keyes, J. and Merrill, M.D. (2001) 'Community-based water
quality monitoring by the Upper Oconee Watershed network', Proceedings of the
2001 Georgia Water Resources Conference (March 2001), University of Georgia,
Athens, Georgia, USA.
Conrad, C.C. and Hilchey, K.G. (2011) 'A review of citizen science and community-based
environmental monitoring: issues and opportunities', Environmental Monitoring
and Assessment, 176(1-4), pp. 273-291.
229
Conway, D. (2000) 'The Climate and Hydrology of the Upper Blue Nile River', The
Geographical Journal, 166(1), pp. 49-62.
Conway, D., Persechino, A., Ardoin-Bardin, S., Hamandawana, H., Dieulin, C. and
Mahé, G. (2009) 'Rainfall and water resources variability in sub-Saharan Africa
during the twentieth century', Journal of Hydrometeorology, 10(1), pp. 41-59.
Cook, P.G., Favreau, G., Dighton, J.C. and Tickell, S. (2003) 'Determining natural
groundwater influx to a tropical river using radon, chlorofluorocarbons and ionic
environmental tracers', Journal of Hydrology, 277(1–2), pp. 74-88.
Cooper, P.J.M., Dimes, J., Rao, K.P.C., Shapiro, B., Shiferaw, B. and Twomlow, S.
(2008) 'Coping better with current climatic variability in the rain-fed farming
systems of sub-Saharan Africa: An essential first step in adapting to future climate
change?', Agriculture, Ecosystems & Environment, 126(1–2), pp. 24-35.
Crosbie, R.S., Jolly, I.D., Leaney, F.W. and Petheram, C. (2010) 'Can the dataset of field
based recharge estimates in Australia be used to predict recharge in data-poor
areas?', Hydrology and Earth System Sciences, 14(10), p. 2023.
CSA (1994) 'Summary and statistical report of the 1994 population and housing census –
population size by age and sex', Central Statistics Agency, Federal Democratic
Republic of Ethiopia Population Census Commission, Addis Ababa.
Cushman, J.H. and Tartakovsky, D.M. (2016) The Handbook of Groundwater

Engineering. CRC Press, Florida, USA, 1073pp.
Dapaah-Siakwan, S. and Gyau-Boakye, P. (2000) 'Hydrogeologic framework and

borehole yields in Ghana', Hydrogeology Journal, 8(4), pp. 405-416.
230
Dardanelli, J.L., Bachmeier, O.A., Sereno, R. and Gil, R. (1997) 'Rooting depth and soil
water extraction patterns of different crops in a silty loam Haplustoll', Field Crops
Research, 54(1), pp. 29-38.
Dargahi, B. and Setegn, S.G. (2011) 'Combined 3D hydrodynamic and watershed

modelling of Lake Tana, Ethiopia', Journal of Hydrology, 398(1), pp. 44-64.
Davies, J., Rastall, P. and Herbert, R. (1998) 'Final Report on application of Collector
Well Systems to Sand Rivers Pilot Project. BGS Report WD/98/2C.', British
Geological Survey, Wallingford, UK.
Dawson, N., Martin, A. and Sikor, T. (2016) 'Green Revolution in Sub-Saharan Africa:
Implications of Imposed Innovation for the Wellbeing of Rural Smallholders',
World Development, 78(Supplement C), pp. 204-218.
de Almeida, W.S., Panachuki, E., de Oliveira, P.T.S., da Silva Menezes, R., Sobrinho,
T.A. and de Carvalho, D.F. (2018) 'Effect of soil tillage and vegetal cover on soil
water infiltration', Soil and Tillage Research, 175, pp. 130-138.
de Vries, J.J. and Simmers, I. (2002) 'Groundwater recharge: an overview of processes

and challenges', Hydrogeology Journal, 10(1), pp. 5-17.
Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P., Kobayashi, S., Andrae,
U., Balmaseda, M.A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A.C.M., van
de Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer,
A.J., Haimberger, L., Healy, S.B., Hersbach, H., Hólm, E.V., Isaksen, L., Kållberg,
P., Köhler, M., Matricardi, M., McNally, A.P., Monge-Sanz, B.M., Morcrette, J.J.,
Park, B.K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.N. and Vitart, F.
(2011) 'The ERA-Interim reanalysis: configuration and performance of the data
assimilation system', Quarterly Journal of the Royal Meteorological Society,
137(656), pp. 553-597.
Delin, G.N., Healy, R.W., Lorenz, D.L. and Nimmo, J.R. (2006) 'Comparison of local- to
regional-scale estimates of ground-water recharge in Minnesota, USA', Journal of
Hydrology, 334(1–2), pp. 231-249.
Demlie, M. (2015) 'Assessment and estimation of groundwater recharge for a catchment

located in highland tropical climate in central Ethiopia using catchment soil–water
231
balance (SWB) and chloride mass balance (CMB) techniques', Environmental
Earth Sciences, 74(2), pp. 1137-1150.
Demlie, M., Wohnlich, S., Gizaw, B. and Stichler, W. (2007) 'Groundwater recharge in
the Akaki catchment, central Ethiopia: evidence from environmental isotopes
(δ18O, δ2H and 3H) and chloride mass balance', Hydrological processes, 21(6), pp.
807-818.
Dennis, R. (2017) 'Regional recharge estimation by applying a genetic algorithm to

various groundwater balance methods.', Presented at the 44th IAH (International
Association of Hydrogeologists) Congress, 25th-29th September 2017, Dubrovnik,
Croatia.
Descroix, L., Genthon, P., Amogu, O., Rajot, J.-L., Sighomnou, D. and Vauclin, M.
(2012) 'Change in Sahelian Rivers hydrograph: The case of recent red floods of the
Niger River in the Niamey region', Global and Planetary Change, 98–99(0), pp.
18-30.
Dessalegn, M. and Merrey, D.J. (2015) 'Motor pump revolution in Ethiopia: Promises at
a Crossroads', Water Alternatives, 8(2).
Deutsch, W.G., Busby, A.L., Orprecio, J.L., Bago-Labis, J.P. and Cequina, E.Y. (2005)
'Community-based hydrological and water quality assessments in Mindanao,
Philippines', Forests, Water and People in the Humid Tropics. Cambridge
University Press, Cambridge, pp. 134-149.
Devia, G.K., Ganasri, B.P. and Dwarakish, G.S. (2015) 'A Review on Hydrological
Models', Aquatic Procedia, 4(Supplement C), pp. 1001-1007.
Dimova, N.T., Burnett, W.C., Chanton, J.P. and Corbett, J.E. (2013) 'Application of
radon-222 to investigate groundwater discharge into small shallow lakes', Journal
of Hydrology, 486, pp. 112-122.
Dinku, T., Ceccato, P., Grover‐Kopec, E., Lemma, M., Connor, S.J. and Ropelewski, C.F.
(2007) 'Validation of satellite rainfall products over East Africa's complex
topography', International Journal of Remote Sensing, 28(7), pp. 1503-1526.
232
Dinku, T., Chidzambwa, S., Ceccato, P., Connor, S.J. and Ropelewski, C.F. (2008)
'Validation of high‐resolution satellite rainfall products over complex terrain',
International Journal of Remote Sensing, 29(14), pp. 4097-4110.
Diriba, K. and Anthony, G.B. (2007) 'Predictability of June–September Rainfall in

Ethiopia', Monthly Weather Review, 135(2), pp. 628-650.
Dittoh, S., Awuni, J.A. and Akuriba, M.A. (2013) 'Small pumps and the poor: a field
survey in the Upper East Region of Ghana', Water International, 38(4), pp. 449-
464.
Dixon, A. (2003) Indigenous Management of Wetlands: Experiences in Ethiopia. Ashgate

Publishing, Farnham, UK, 258pp.
Dixon, J., Taniguchi, K. and Wattenbach, H. (2003) 'Approaches to assessing the impact
of globalization on African smallholders: Household and village economy
modeling', Proceedings of a working session on Globalization and the African
Smallholder Study. FAO (Agricultural Support Systems Division [AGS] and
Agricultural and Development Economics Division [ESA]) and the World Bank.
Rome: Food and Agriculture Organization of the United Nations.
Döll, P. (2009) 'Vulnerability to the impact of climate change on renewable groundwater

resources: a global-scale assessment', Environmental Research Letters, 4(3), p.
035006.
Döll, P. and Fiedler, K. (2007) 'Global-scale modeling of groundwater recharge',

Hydrology and Earth System Sciences Discussions, 4(6), pp. 4069-4124.
Döll, P. and Flörke, M. (2005) 'Global-scale estimation of diffuse groundwater recharge:

model tuning to local data for semi-arid and arid regions and assessment of climate
change impact', Frankfurt Hydrology Paper 03, Institute of Physical Geography,
Frankfurt University, Frankfurt am Main, Germany, p. 21pp.
Doorenbos, J. and Pruitt, W.O. (1975) 'Guidelines for predicting crop water requirements
(FAO Irrigation and Drainage Paper 24)', Food and Agriculture Organisation of the
United Nations, Rome, Italy.
Dorward, A., Kydd, J., Morrison, J. and Urey, I. (2004) 'A policy agenda for pro-poor
agricultural growth', World development, 32(1), pp. 73-89.
233
Dries, I. (1991) 'Development of wetlands in Sierra Leone: farmers' rationality opposed
to government policy', Landscape and urban planning, 20(1-3), pp. 223-229.
Dube, D.K. (2014) 'Why a shift from Coffee to Chat? A Study of the Kersa Woreda in
Jimma Zone of South Western Ethiopia', Research on Humanities and Social
Sciences, 4(19), pp. 176-182.
Dugan, P.J. (1988) 'Research priorities in wetland science; wetland conservation and
agricultural development', Conservation and Development: The Sustainable Use of
Wetland Resources. Proceedings of the Third International Wetlands Conference,
Rennes, France.
Easton, Z.M., Fuka, D.R., White, E.D., Collick, A.S., Biruk Ashagre, B., McCartney, M.,
Awulachew, S.B., Ahmed, A.A. and Steenhuis, T.S. (2010) 'A multi basin SWAT
model analysis of runoff and sedimentation in the Blue Nile, Ethiopia', Hydrology
and earth system sciences, 14(10), pp. 1827-1841.
Ebert, E.E., Janowiak, J.E. and Kidd, C. (2007) 'Comparison of near-real-time

precipitation estimates from satellite observations and numerical models', Bulletin
of the American Meteorological Society, 88(1), pp. 47-64.
Ebrahim, G.Y. and Villholth, K.G. (2016) 'Estimating shallow groundwater availability
in small catchments using streamflow recession and instream flow requirements of
rivers in South Africa', Journal of Hydrology, 541, pp. 754-765.
Eckhardt, K. (2005) 'How to construct recursive digital filters for baseflow separation',
Hydrological processes, 19(2), pp. 507-515.
Eckhardt, K. (2008) 'A comparison of baseflow indices, which were calculated with seven
different baseflow separation methods', Journal of Hydrology, 352(1), pp. 168-173.
EDC (2016) European Drought Centre. Available at: http://www.geo.uio.no/edc/

(Accessed: 20 April 2016).
Edmunds, W.M., Gaye, C.B. and Fontes, J.C. (1991) 'A record of climatic and
environmental change contained in interstitial waters from the unsaturated zone of
northern Senegal', Isotope Techniques in Water Resources Development, pp. 533-
549.
234
Emana, B., Afari-Sefa, V., Dinssa, F.F., Ayana, A., Balemi, F. and Temesgen, M. (2015)
'Characterization and assessment of vegetable production and marketing systems in
the Humid Tropics of Ethiopia', Quarterly Journal of International Agriculture,
54(2), pp. 163-187.
Engman, E.T. (1986) 'Roughness coefficients for routing surface runoff', Journal of
Irrigation and Drainage Engineering, 112(1), pp. 39-53.
Enku, T., Melesse, A., Ayana, E., Tilahun, S., Abate, M. and Steenhuis, T. (2017a)
Response of Groundwater table to Eucalyptus Plantations in a Tropical Monsoon
Climate, Lake Tana Basin, Ethiopia. 19th EGU General Assembly, EGU2017,
proceedings from the conference held 23-28 April, 2017 in Vienna, Austria.,
p.4652.
Enku, T., Melesse, A.M., Ayana, E.K., Tilahun, S.A., Abate, M. and Steenhuis, T.S.
(2017b) 'Groundwater Evaporation and Recharge for a Floodplain in a Sub‐humid
Monsoon Climate in Ethiopia', Land Degradation & Development, 28(6), pp. 1831-
1841.
Ethiopian ATA (2013) 'Realizing the Potential of Household Irrigation in Ethiopia',

Working Strategy Document Ethiopian Agricultural Transformation Agency, Addis
Ababa, Ethiopia, 94pp.
Evans, A.E.V., Giordano, M. and Clayton, T. (2012) 'Investing in agricultural water

management to benefit smallholder farmers in Ethiopia. AgWater Solutions Project
country synthesis report', International Water Management Institute (IWMI), IWMI
Working Paper 152, Colombo, Sri Lanka, 35pp.
Evenson, R.E. and Gollin, D. (2003) 'Assessing the impact of the Green Revolution, 1960
to 2000', Science, 300(5620), pp. 758-762.
Ewen, J., Parkin, G. and O'Connell, P.E. (2000) 'Shetran: distributed river basin flow and
transport modeling system', Journal of Hydrologic Engineering, 5(3), pp. 250-258.
Ezilon (2017) The Regional Directory and Information. Available at:

http://www.ezilon.com/ (Accessed: 14 June 2017).
Fan, J., McConkey, B., Wang, H. and Janzen, H. (2016) 'Root distribution by depth for
temperate agricultural crops', Field Crops Research, 189, pp. 68-74.
235
FAO (2017) 'Food and Agriculture Organization of the United Nations - Crop Calendar:
http://www.fao.org/agriculture/seed/cropcalendar/welcome.do', Accessed 28
August 17.
Farley, K.A., Jobbágy, E.G. and Jackson, R.B. (2005) 'Effects of afforestation on water
yield: a global synthesis with implications for policy', Global change biology,
11(10), pp. 1565-1576.
Favreau, G., Cappelaere, B., Massuel, S., Leblanc, M., Boucher, M., Boulain, N. and
Leduc, C. (2009) 'Land clearing, climate variability, and water resources increase
in semiarid southwest Niger: A review', Water Resources Research, 45, p. 18.
Favreau, G., Leduc, C., Marlin, C., Dray, M., Taupin, J.D., Massault, M., La Salle, C.L.
and Babic, M. (2002) 'Estimate of recharge of a rising water table in semiarid Niger
from H-3 and C-14 modeling', Ground Water, 40(2), pp. 144-151.
Fazzini, M., Bisci, C. and Billi, P. (2015) 'The Climate of Ethiopia', in Billi, P. (ed.)
Landscapes and Landforms of Ethiopia. Springer, Dordrecht, The Netherlands, pp.
65-87.
Fekete, B.M., Vörösmarty, C.J., Roads, J.O. and Willmott, C.J. (2004) 'Uncertainties in
precipitation and their impacts on runoff estimates', Journal of Climate, 17(2), pp.
294-304.
Fetter, C.W. (2001) Applied Hydrogeology. Prentice Hall, Upper Saddle River, New
Jersey, USA.
Forni, L.G., Galaitsi, S.E., Mehta, V.K., Escobar, M.I., Purkey, D.R., Depsky, N.J. and
Lima, N.A. (2016) 'Exploring scientific information for policy making under deep
uncertainty', Environmental Modelling & Software, 86, pp. 232-247.
Foster, S. and Cherlet, J. (2014) The links between land use and groundwater.
Perspectives Paper: Global Water Partnership, Stockholm, Sweden, 20pp.
Foster, S. and Shah, T. (2012) 'Groundwater Resources and Irrigated Agriculture',

Perspectives Paper, Global Water Partneship, Stockholm, Sweden, 20pp.
Foster, S.S.D., Tuinhof, A. and Garduño, H. (2008) 'Groundwater in Sub-Saharan Africa

- A strategic overview of developmental issues', in Adelana, S.M.A. and
236
MacDonald, A.M. (eds.) Applied groundwater research in Africa. IAH Selected
Papers on Hydrogeology Volume 13, CRC Press, Leiden, The Netherlands, pp. 9-
21.
Fredlund, D.G., Rahardjo, H. and Rahardjo, H. (1993) Soil mechanics for unsaturated
soils. John Wiley & Sons.
Freeze, R.A. and Harlan, R.L. (1969) 'Blueprint for a physically-based, digitally-
simulated hydrologic response model', Journal of Hydrology, 9(3), pp. 237-258.
Frenken, K. (1997) Irrigation potential in Africa: A basin approach (FAO Food and
Water Bulletin 4). Food and Agriculture Organisation of the United Nations, Rome,
Italy, 178pp.
Frenken, K. (2005) Irrigation in Africa in figures: AQUASTAT Survey, 2005. FAO (Food
& Agriculture Organisation of the United Nations), Rome, Italy, 29pp.
Friedlander, L., Tal, A. and Lazarovitch, N. (2013) 'Technical considerations affecting

adoption of drip irrigation in sub-Saharan Africa', Agricultural Water Management,
126(Supplement C), pp. 125-132.
Furi, W., Razack, M., Haile, T., Abiye, T.A. and Legesse, D. (2011) 'The hydrogeology
of Adama-Wonji basin and assessment of groundwater level changes in Wonji
wetland, Main Ethiopian Rift: results from 2D tomography and electrical sounding
methods', Environmental Earth Sciences, 62(6), pp. 1323-1335.
Fussi, F. (2011) 'Mapping of suitable zones for manual drilling as a possible solution to
increase access to drinking water in Africa', 6th RWSN Forum Kampala, Uganda,
Nov-Dec 2011.
Fussi, F., Fumagalli, L., Hamidou Kane, C. and Bonomi, T. (2016) 'Mapping of suitable
zones for manual drilling. An overview of the method and the application as
decision tools', 7th RWSN Forum "Water for Everyone". Abidjan, Côte d’Ivoire,
Nov-Dec 2016.
Fynn, R.W.S., Murray-Hudson, M., Dhliwayo, M. and Scholte, P. (2015) 'African

wetlands and their seasonal use by wild and domestic herbivores', Wetlands Ecology
and Management, 23(4), pp. 559-581.
237
Galloway, D.L. and Hoffmann, J. (2007) 'The application of satellite differential SAR
interferometry-derived ground displacements in hydrogeology', Hydrogeology
Journal, 15(1), pp. 133-154.
Galloway, D.L., Hudnut, K.W., Ingebritsen, S.E., Phillips, S.P., Peltzer, G., Rogez, F. and
Rosen, P.A. (1998) 'Detection of aquifer system compaction and land subsidence
using interferometric synthetic aperture radar, Antelope Valley, Mojave Desert,
California', Water Resources Research, 34(10), pp. 2573-2585.
Garduño, H. and Foster, S. (2010) 'Sustainable groundwater irrigation approaches to

reconciling demand with resources', GW-MATE Strategic Overview Series, World
Bank Group, Washington DC, USA, 40pp.
Garduño, H., Foster, S., Raj, P. and van Steenbergen, F. (2009) 'Addressing groundwater
depletion through community-based management actions in the weathered Granitic
basement aquifer of drought-prone Andhra Pradesh–India', World Bank GW-MATE
Case Profile Collection, 19.
Gebere, S.B., Alamirew, T., Merkel, B.J. and Melesse, A.M. (2016) 'Land use and Land
cover change impact on groundwater recharge: the Case of Lake Haramaya
Watershed, Ethiopia', in Landscape Dynamics, Soils and Hydrological Processes
in Varied Climates. Springer, pp. 93-110.
Gebissa, E. (2008) 'Scourge of life or an economic lifeline? Public discourses on khat

(Catha edulis) in Ethiopia', Substance Use & Misuse, 43(6), pp. 784-802.
Gebissa, E. (2010) 'Khat in the Horn of Africa: Historical perspectives and current trends',
Journal of Ethnopharmacology, 132(3), pp. 607-614.
Gebregziabher, G. and Haile, A.T. (2013) 'Field report for the project: Adaptive
Management of Groundwater Resources for Small-scale Irrigation in Sub-Saharan
Africa (AMGRAF)', Unpublished.
Gebremeskel, G. (2015) Estimation of groundwater recharge and potentials under

changing climate in Werii Watershed, Tekeze River Basin. MSc thesis, School of
Natural Resource and Environmental Engineering, Haramaya University, Ethiopia.
Gebreyohannes, T., De Smedt, F., Walraevens, K., Gebresilassie, S., Hussien, A., Hagos,
M., Amare, K., Deckers, J. and Gebrehiwot, K. (2013) 'Application of a spatially
238
distributed water balance model for assessing surface water and groundwater
resources in the Geba basin, Tigray, Ethiopia', Journal of Hydrology, 499, pp. 110-
123.
Gezon, L.L. (2012) 'Drug Crops and Food Security: The Effects of Khat on Lives and
Livelihoods in Northern Madagascar', Culture, Agriculture, Food and
Environment, 34(2), pp. 124-135.
Gibb and Seureca (1996) The town of Harar water supply, Hydrogeological study interim
working paper. Sir Alexander Gibb and Seureca Consulting Engineers: Ministry of
Water Resources (MoWR), Addis Ababa, Ethiopia.
Giordano, M. (2006) 'Agricultural groundwater use and rural livelihoods in sub-Saharan

Africa: A first-cut assessment', Hydrogeology Journal, 14(3), pp. 310-318.
Giordano, M., de Fraiture, C., Weight, E. and van der Bliek, J. (2012) Water for wealth
and food security. Synthesis report of the AgWater Solutions Project. Colombo, Sri
Lanka: International Water Management Institute (IWMI). 50pp.
Giordano, M. and Villholth, K.G. (2007) The agricultural groundwater revolution:

opportunities and threats to development. CABI, Wallingford, UK, 95pp.
Girmay, E., Ayenew, T., Kebede, S., Alene, M., Wohnlich, S. and Wisotzky, F. (2015)
'Conceptual groundwater flow model of the Mekelle Paleozoic–Mesozoic
sedimentary outlier and surroundings (northern Ethiopia) using environmental
isotopes and dissolved ions', Hydrogeology Journal, 23(4), pp. 649-672.
Goderniaux, P., Brouyère, S., Fowler, H.J., Blenkinsop, S., Therrien, R., Orban, P. and
Dassargues, A. (2009) 'Large scale surface–subsurface hydrological model to assess
climate change impacts on groundwater reserves', Journal of Hydrology, 373(1),
pp. 122-138.
Goes, B.J.M. (1999) 'Estimate of shallow groundwater recharge in the Hadejia–Nguru

Wetlands, semi-arid northeastern Nigeria', Hydrogeology Journal, 7(3), pp. 294-
304.
Golding, B., Clark, P. and May, B. (2005) 'The Boscastle flood: Meteorological analysis
of the conditions leading to flooding on 16 August 2004', Weather, 60(8), pp. 230-
235.
239
Gomani, M.C., Dietrich, O., Lischeid, G., Mahoo, H., Mahay, F., Mbilinyi, B. and
Sarmett, J. (2010) 'Establishment of a hydrological monitoring network in a tropical
African catchment: An integrated participatory approach', Physics and Chemistry
of the Earth, Parts A/B/C, 35(13–14), pp. 648-656.
Gooddy, D.C., Darling, W.G., MacDonald, A.M. and Morris, B.L. (2012) 'The
practicalities of using CFCs and SF6 for groundwater dating and tracing', Applied
Geochemistry, 27(9), pp. 1688-1697.
Gov. WA (2009) Surface water sampling methods and analysis - technical appendices.
Government of Western Australia Department of Water, Perth, Australia.
Gowing, J., Parkin, G., Forsythe, N., Walker, D., Haile, A.T. and Alamirew, D. (2016)
'Shallow groundwater in sub-Saharan Africa: neglected opportunity for sustainable
intensification of small-scale agriculture?', Hydrol. Earth Syst. Sci. Discuss., 2016,
pp. 1-33.
Greitzer, Y. (1970) Stratigraphy, hydrogeology and Jurassic ammonites of the Harar and
Diredawa area, Ethiopia. PhD thesis, Hebrew University, Jerusalem, Israel.
Grey, D. and Sadoff, C.W. (2007) 'Sink or swim? Water security for growth and
development', Water policy, 9(6), pp. 545-571.
Grove, A.T. (1986) 'Geomorphology of the African rift system', Geological Society,
London, Special Publications, 25(1), pp. 9-16.
Guerreiro, S.B., Birkinshaw, S., Kilsby, C., Fowler, H.J. and Lewis, E. (2017) 'Dry getting
drier – The future of transnational river basins in Iberia', Journal of Hydrology:
Regional Studies, 12(Supplement C), pp. 238-252.
Guiraud, R. (1988) 'L'hydrogéologie de l'Afrique', Journal of African Earth Sciences (and

the Middle East), 7(3), pp. 519-543.
Guttman, N.B. (1998) 'Comparing the Palmer drought index and the standardized
precipitation index', JAWRA Journal of the American Water Resources Association,
34(1), pp. 113-121.
Haan, C.T., Allen, D.M. and Street, J.O. (1976) 'A Markov chain model of daily rainfall',
Water Resources Research, 12(3), pp. 443-449.
240
Habtamu, G.D. (2009) Groundwater Potential Investigation of upper wabe river
catchment, south eastern central Ethiopia. MSc thesis, Department of Earth
Hagos, F., Pender, J. and Gebreselassie, N. (1999) 'Land degradation in the highlands of
Tigray and strategies for sustainable land management', Socioeconomic and Policy
Research Working Paper No. 25, Livestock Policy Analysis Project, International
Livestock Research Institute, Addis Ababa, Ethiopia.
Hagos, M.A. (2010) Groundwater Flow Modelling assisted by GIS and RS Techniques
(Raya Valley-Ethiopia). MSc thesis, ITC, Enschede, The Netherlands.
Haile, A.T., Rientjes, T., Gieske, A. and Gebremichael, M. (2009) 'Rainfall variability
over mountainous and adjacent lake areas: the case of Lake Tana basin at the source
of the Blue Nile River', Journal of Applied Meteorology and Climatology, 48(8),
pp. 1696-1717.
Hailu, A. (2003) 'Wetlands research in south-western Ethiopia: the experience of the

Ethiopian Wetlands Research Programme', in Abebe, Y.D. and Geheb, K. (eds.)
Wetlands of Ethiopia. IUCN Wetlands and Water Resources Programme, Gland,
Switzerland.
Hailu, D. (2005) 'Supporting a nation: khat farming and livelihoods in Ethiopia', Drugs
and Alcohol Today, 5(3), pp. 22-24.
Halford, K.J. and Mayer, G.C. (2000) 'Problems Associated with Estimating Ground
Water Discharge and Recharge from Stream-Discharge Records', Ground Water,
38(3), pp. 331-342.
Harvey, C.F., Ashfaque, K.N., Yu, W., Badruzzaman, A.B.M., Ali, M.A., Oates, P.M.,
Michael, H.A., Neumann, R.B., Beckie, R., Islam, S. and Ahmed, M.F. (2006)
'Groundwater dynamics and arsenic contamination in Bangladesh', Chemical
Geology, 228(1), pp. 112-136.
Hautot, S., Whaler, K., Gebru, W. and Desissa, M. (2006) 'The structure of a Mesozoic
basin beneath the Lake Tana area, Ethiopia, revealed by magnetotelluric imaging',
Journal of African Earth Sciences, 44(3), pp. 331-338.
241
Hazell, P.B.R. (2009) The Asian green revolution. IFPRI Discussion Paper 00911:
International Food Policy Research Institute, Washington DC, USA, 20pp.
Healy, R.W. (2010) Estimating groundwater recharge. Cambridge University Press,

Cambridge, UK, 256pp.
Healy, R.W. and Cook, P.G. (2002) 'Using groundwater levels to estimate recharge',
Hydrogeology journal, 10(1), pp. 91-109.
Hendrickx, J.M.H. and Flury, M. (2001) 'Uniform and preferential flow mechanisms in
the vadose zone', Conceptual models of flow and transport in the fractured vadose
zone, National Research Council, National Academy Press, Washington DC, USA,
pp. 149-187.
Hochachka, W.M., Fink, D., Hutchinson, R.A., Sheldon, D., Wong, W.K. and Kelling, S.
(2012) 'Data-intensive science applied to broad-scale citizen science', Trends in
Ecology & Evolution, 27(2), pp. 130-137.
Hoehn, E. and Von Gunten, H.R. (1989) 'Radon in groundwater: A tool to assess
infiltration from surface waters to aquifers', Water Resources Research, 25(8), pp.
1795-1803.
Hofmann, C., Courtillot, V., Feraud, G. and Rochette, P. (1997) 'Timing of the Ethiopian
flood basalt event and implications for plume birth and global change', Nature,
389(6653), p. 838.
Hossain, G.M. and Islam, M. (2000) 'Water Resources Management in Bangladesh', in

Building Partnerships. ASCE, Virginia, USA, pp. 1-10.
Huang, T., Pang, Z., Liu, J., Yin, L. and Edmunds, W.M. (2017) 'Groundwater recharge
in an arid grassland as indicated by soil chloride profile and multiple tracers',
Huff, F.A. (1970) 'Sampling errors in measurement of mean precipitation', Journal of

Applied Meteorology, 9(1), pp. 35-44.
Hughes, J.D., Langevin, C.D. and Banta, E.R. (2017) Documentation for the MODFLOW
6 framework (6-A57). Reston, VA: Survey, U.S.G. [Online]. Available at:
http://pubs.er.usgs.gov/publication/tm6A57.
242
Hulme, M., Doherty, R., Ngara, T., New, M. and Lister, D. (2001) 'African climate
change: 1900-2100', Climate Research, 17(2), pp. 145-168.
Hurni, H., Tato, K. and Zeleke, G. (2005) 'The implications of changes in population,
land use, and land management for surface runoff in the upper Nile basin area of
Ethiopia', Mountain Research and Development, 25(2), pp. 147-154.
Hussain, M., Hussain, Z. and Ashfaq, M. (2006) 'Impact of small scale irrigation schemes
on poverty alleviation in marginal areas of Punjab, Pakistan', International Journal
of Finance and Economics, 6(6).
Hussen, E. (2006) Water resource potential evaluation of Berga River Catchment, West
Shewa Zone, Oromiya Regional State. MSc thesis, Department of Earth Sciences,
Addis Ababa University, Ethiopia.
Hussey, S.W. (2007) Water from sand rivers. Guidelines for abstraction. Water,
Engineering and Development Centre (WEDC) Loughborough University of
Technology, UK, 194pp.
Hutley, L.B., O'Grady, A.P. and Eamus, D. (2000) 'Evapotranspiration from Eucalypt
open-forest savanna of Northern Australia', Functional Ecology, 14(2), pp. 183-
194.
IAEA (-) Sampling procedures for isotope hydrology. International Atomic Energy
Agency, Vienna, Austria.
Ilstedt, U., Tobella, A.B., Bazie, H.R., Bayala, J., Verbeeten, E., Nyberg, G., Sanou, J.,
Benegas, L., Murdiyarso, D., Laudon, H., Sheil, D. and Malmer, A. (2016)
'Intermediate tree cover can maximize groundwater recharge in the seasonally dry
tropics', Scientific Reports, 6, p. 12.
Inocencio, A., Kikuchi, M., Tonosaki, M., Maruyama, A., Merrey, D., Sally, H. and de
Jong, I. (2007) Costs and performance of irrigation projects: A comparison of sub-
Saharan Africa and other developing regions. International Water Management
Institute (IWMI), Research Report no. 109, Colombo, Sri Lanka, 86pp.
Inocencio, A., Sally, H. and Merrey, D.J. (2003) Innovative approaches to agricultural
water use for improving food security in sub-Saharan Africa. Working Paper 55,
International Water Management Institute (IWMI), Colombo, Sri Lanka, 25pp.
243
IOH (1980) Low Flow Studies. Institute of Hydrology, Wallingford, UK.
IPCC (2007) 'Climate Change 2007: Impacts, Adaptation and Vulnerability', in Parry,
M.L., Canziani, O.F., Palutikof, J.P., Hanson, C.E. and van der Linden, P.J. (eds.)
Contribution of working group II to the fourth assessment report of the
Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press,
Cambridge, UK.
Jackson, C.R., Meister, R. and Prudhomme, C. (2011) 'Modelling the effects of climate
change and its uncertainty on UK Chalk groundwater resources from an ensemble
of global climate model projections', Journal of Hydrology, 399(1), pp. 12-28.
Jagger, P. and Pender, J. (2003) 'The role of trees for sustainable management of less-
favored lands: the case of eucalyptus in Ethiopia', Forest Policy and Economics,
5(1), pp. 83-95.
Jaleta, D., Mbilinyi, B., Mahoo, H. and Lemenih, M. (2016) 'Evaluation of land use/land
cover changes and Eucalyptus expansion in Meja watershed, Ethiopia', J Geo
Environ Earth Sci Int, 7(3), pp. 1-12.
Jones, A., Breuning-Madsen, H., Brossard, M., Dampha, A., Deckers, J., Dewitte, O.,
Gallali, T., Hallett, S., Jones, R. and Kilasara, M. (2013) Soil atlas of Africa.
European Commission, Publications Office of the European Union, Luxembourg.
Jones, M.J. (1985) 'The weathered zone aquifers of the basement complex areas of
Africa', Quarterly Journal of Engineering Geology and Hydrogeology, 18(1), pp.
35-46.
Kahsay, G.H. (2008) Groundwater resource assessment through distributed steady-state

flow modeling, Aynalem wellfield (Mekele, Ethiopia). MSc thesis, ITC, Enschede,
The Netherlands.
Kalix, P. and Braenden, O. (1985) 'Pharmacological aspects of the chewing of khat

leaves', Pharmacological reviews, 37(2), pp. 149-164.
Kan, G., Yao, C., Li, Q., Li, Z., Yu, Z., Liu, Z., Ding, L., He, X. and Liang, K. (2015)
'Improving event-based rainfall-runoff simulation using an ensemble artificial
neural network based hybrid data-driven model', Stochastic environmental research
and risk assessment, 29(5), pp. 1345-1370.
244
Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S.-K., Hnilo, J.J., Fiorino, M. and
Potter, G.L. (2002) 'Ncep-doe amip-ii reanalysis (r-2)', Bulletin of the American
Meteorological Society, 83(11), pp. 1631-1643.
Karatzas, G.P. (2017) 'Developments on modeling of groundwater flow and contaminant

transport', Water Resources Management, pp. 1-10.
Kay, M. (2001) Smallholder irrigation technology: prospects for sub-Saharan Africa.

FAO (Food & Agriculture Organisation of the United Nations), Rome, Italy, 45pp.
Kebede, S. (2013) Groundwater in Ethiopia: Features, numbers and opportunities.

Springer Verlag, Berlin, Germany, 284pp.
Kebede, S., Admasu, G. and Travi, Y. (2011) 'Estimating ungauged catchment flows from
Lake Tana floodplains, Ethiopia: an isotope hydrological approach', Isotopes in
Environmental and Health studies, 47(1), pp. 71-86.
Kebede, S., Travi, Y., Alemayehu, T. and Ayenew, T. (2005) 'Groundwater recharge,
circulation and geochemical evolution in the source region of the Blue Nile River,
Ethiopia', Applied Geochemistry, 20(9), pp. 1658-1676.
Kebede, S., Travi, Y., Alemayehu, T. and Marc, V. (2006) 'Water balance of Lake Tana
and its sensitivity to fluctuations in rainfall, Blue Nile basin, Ethiopia', Journal of
hydrology, 316(1), pp. 233-247.
Kehinde, M.O. and Loehnert, E.P. (1989) 'Review of African groundwater resources',
Journal of African Earth Sciences (and the Middle East), 9(1), pp. 179-185.
Kerebih, M.S. and Keshari, A.K. (2017) 'GIS-Coupled Numerical Modeling for
Sustainable Groundwater Development: Case Study of Aynalem Well Field,
Ethiopia', Journal of Hydrologic Engineering, 22(4), p. 05017001.
Keywood, M.D., Chivas, A.R., Fifield, L.K., Cresswell, R.G. and Ayers, G.P. (1997) 'The
accession of chloride to the western half of the Australian continent', Soil Research,
35(5), pp. 1177-1190.
Kieffer, B., Arndt, N., Lapierre, H., Bastien, F., Bosch, D., Pecher, A., Yirgu, G., Ayalew,
D., Weis, D. and Jerram, D.A. (2004) 'Flood and shield basalts from Ethiopia:
magmas from the African superswell', Journal of Petrology, 45(4), pp. 793-834.
245
Kilsby, C.G., Jones, P.D., Burton, A., Ford, A.C., Fowler, H.J., Harpham, C., James, P.,
Smith, A. and Wilby, R.L. (2007) 'A daily weather generator for use in climate
change studies', Environmental Modelling & Software, 22(12), pp. 1705-1719.
King, A.C., Raiber, M., Cox, M.E. and Cendón, D.I. (2017) 'Comparison of groundwater
recharge estimation techniques in an alluvial aquifer system with an
intermittent/ephemeral stream (Queensland, Australia)', Hydrogeology Journal,
25(6), pp. 1759-1777.
Kingston, D.G. and Taylor, R.G. (2010) 'Sources of uncertainty in climate change impacts
on river discharge and groundwater in a headwater catchment of the Upper Nile
Basin, Uganda', Hydrology and Earth System Sciences, 14(7), pp. 1297-1308.
Kitamirike, J. (2008) Training workshop on water resources law and practice in Uganda.
Kampala, Uganda. Accessed (21-9-15) from.
http://www.nemaug.org/padelia_meas_Project/reports4.html.
Klein, A., Beckerleg, S. and Hailu, D. (2009) 'Regulating khat—Dilemmas and

opportunities for the international drug control system', International Journal of
Drug Policy, 20(6), pp. 509-513.
Kloos, H. (1991) 'Peasant irrigation development and food production in Ethiopia',

Geographical Journal, 157, pp. 295-306.
Kloos, H. and Haimanot, R.T. (1999) 'Distribution of fluoride and fluorosis in Ethiopia
and prospects for control', Tropical Medicine & International Health, 4(5), pp. 355-
364.
Kobayashi, S., Ota, Y., Harada, Y., Ebita, A., Moriya, M., Onoda, H., Onogi, K.,
Kamahori, H., Kobayashi, C. and Endo, H. (2014) 'The JRA-55 Reanalysis: General
specifications and basic characteristics', J. Meteor. Soc. Japan, submitted.
Koch, F.J., van Griensven, A., Uhlenbrook, S., Tekleab, S. and Teferi, E. (2012) 'The
Effects of land use change on hydrological responses in the choke mountain range
(Ethiopia) - a new approach addressing land use dynamics in the model SWAT',
Proceedings of 2012 international congress on environmental modeling and
software managing resources of a limited planet, sixth biennial meeting, Leipzig,
Germany. Citeseer, pp. 1-5.
246
Kongo, V.M., Kosgei, J.R., Jewitt, G.P.W. and Lorentz, S.A. (2010) 'Establishment of a
catchment monitoring network through a participatory approach in a rural
community in South Africa', Hydrol. Earth Syst. Sci., 14(12), pp. 2507-2525.
Konikow, L.F. and Kendy, E. (2005) 'Groundwater depletion: A global problem',

Köppen, W. and Geiger, R. (1930) Handbuch der klimatologie. Gebrüder Borntraeger

Berlin, Germany.
Krause, P., Boyle, D.P. and Bäse, F. (2005) 'Comparison of different efficiency criteria
for hydrological model assessment', Advances in Geosciences, 5, pp. 89-97.
Kulabako, N.R., Nalubega, M. and Thunvik, R. (2007) 'Study of the impact of land use
and hydrogeological settings on the shallow groundwater quality in a peri-urban
area of Kampala, Uganda', Science of the Total Environment, 381(1), pp. 180-199.
Kulecho, I.K. and Weatherhead, E.K. (2005) 'Reasons for smallholder farmers
discontinuing with low-cost micro-irrigation: A case study from Kenya', Irrigation
and Drainage Systems, 19(2), pp. 179-188.
Kulkarni, H., Deolankar, S.B., Lalwani, A., Joseph, B. and Pawar, S. (2000)
'Hydrogeological framework of the Deccan basalt groundwater systems, west-
central India', Hydrogeology Journal, 8(4), pp. 368-378.
Kumai, H. and Mitamura, M. (2004) 'Application of stream hydrograph separation

method to estimate the recharge in the northern part of the Bandung Basin, west
Java, Indonesia', Journal of Groundwater Hydrology, 46(3), pp. 213-225.
Kundzewicz, Z.W. (1997) 'Water resources for sustainable development', Hydrological

Sciences Journal, 42(4), pp. 467-480.
Lambin, E.F. and Ehrlich, D. (1997) 'Land-cover changes in sub-saharan Africa (1982–
1991): Application of a change index based on remotely sensed surface temperature
and vegetation indices at a continental scale', Remote Sensing of Environment,
61(2), pp. 181-200.
Lambin, E.F., Turner, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes,
O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J.,
247
Leemans, R., Li, X., Moran, E.F., Mortimore, M., Ramakrishnan, P.S., Richards,
J.F., Skånes, H., Steffen, W., Stone, G.D., Svedin, U., Veldkamp, T.A., Vogel, C.
and Xu, J. (2001) 'The causes of land-use and land-cover change: moving beyond
the myths', Global Environmental Change, 11(4), pp. 261-269.
Laprise, R., Hernández-Díaz, L., Tete, K., Sushama, L., Šeparović, L., Martynov, A.,
Winger, K. and Valin, M. (2013) 'Climate projections over CORDEX Africa
domain using the fifth-generation Canadian Regional Climate Model (CRCM5)',
Climate Dynamics, 41(11-12), pp. 3219-3246.
Lapworth, D.J., MacDonald, A.M., Tijani, M.N., Darling, W.G., Gooddy, D.C., Bonsor,
H.C. and Araguas-Araguas, L.J. (2013) 'Residence times of shallow groundwater
in West Africa: implications for hydrogeology and resilience to future changes in
climate', Hydrogeology Journal, 21(3), pp. 673-686.
Laube, W., Awo, M. and Schraven, B. (2008) Erratic Rains and Erratic Markets:
Environmental change, economic globalisation and the expansion of shallow
groundwater irrigation in West Africa. ZEF Working Paper Series, University of
Bonn, Germany.
Leduc, C., Favreau, G. and Schroeter, P. (2000) 'Long-term rise in a sahelian water-table:
the Continental Terminal in South-West Niger', Journal of Hydrology, 243(1-2),
pp. 43-54.
Legates, D.R. and Willmott, C.J. (1990) 'Mean seasonal and spatial variability in gauge‐
corrected, global precipitation', International Journal of Climatology, 10(2), pp.
111-127.
Legesse, D. and Ayenew, T. (2006) 'Effect of improper water and land resource utilization
on the central Main Ethiopian Rift lakes', Quaternary international, 148(1), pp. 8-
18.
Legesse, D., Vallet-Coulomb, C. and Gasse, F. (2003) 'Hydrological response of a

catchment to climate and land use changes in Tropical Africa: case study South
Central Ethiopia', Journal of Hydrology, 275(1), pp. 67-85.
Lemessa, D. (2001) 'Khat (Catha edulis): botany, distribution, cultivation, usage and
economics in Ethiopia', Addis Ababa: UNDP - Emergencies Unit for Ethiopia.
248
Lemma, B. (2011) 'The impact of climate change and population increase on Lakes
Haramaya and Hora-Kilole, Ethiopia (1986–2006)', Impacts of climate change and
population on tropical aquatic resources, p. 9.
Lerner, D., Issar, A. and Simmers, I. (1990) Groundwater Recharge: A Guide to

Understanding and Estimating Natural Recharge. International Contributions to
Hydrogeology. (8 vols). Heise, Germany, 345pp.
Lewis, E. (2016) A robust multi-purpose hydrological model for Great Britain. PhD
thesis, School of Civil Engineering and Geosciences, Newcastle University, UK.
Lim, K.J., Engel, B.A., Tang, Z., Choi, J., Kim, K.S., Muthukrishnan, S. and Tripathy, D.
(2005) 'Automated web GIS based hydrograph analysis tool, WHAT', JAWRA
Journal of the American Water Resources Association, 41(6), pp. 1407-1416.
Lima, W.d.P., Zakia, M.J.B., Libardi, P.L. and de Souza Filho, A.P. (1990) 'Comparative
evapotranspiration of eucalyptus, pine and natural “cerrado” vegetation measure by
the soil water balance method”', IPEF International, Piracicaba, 1, pp. 5-11.
Lipiec, J., Kuś, J., Słowińska-Jurkiewicz, A. and Nosalewicz, A. (2006) 'Soil porosity and
water infiltration as influenced by tillage methods', Soil and Tillage Research,
89(2), pp. 210-220.
Liu, B.M., Abebe, Y., McHugh, O.V., Collick, A.S., Gebrekidan, B. and Steenhuis, T.S.
(2008a) 'Overcoming limited information through participatory watershed
management: Case study in Amhara, Ethiopia', Physics and Chemistry of the Earth,
Parts A/B/C, 33(1–2), pp. 13-21.
Liu, M., Tian, H., Chen, G., Ren, W., Zhang, C. and Liu, J. (2008b) 'Effects of Land‐Use
and Land‐Cover Change on Evapotranspiration and Water Yield in China During
1900‐2000', JAWRA Journal of the American Water Resources Association, 44(5),
pp. 1193-1207.
Love, D., Hamer, W.d., Owen, R., Booij, M., Uhlenbrook, S., Hoekstra, A. and Zaag, P.
(2007) 'Case studies of groundwater: Surface water interactions and scale
relationships in small alluvial aquifers', 8th WATERNET/WARFSA/GWP-SA
Symposium, 31 October - 2 November 2007, Lusaka, Zambia, pp. 6.
249
Lukey, B.T., Sheffield, J., Bathurst, J.C., Hiley, R.A. and Mathys, N. (2000) 'Test of the
SHETRAN technology for modelling the impact of reforestation on badlands runoff
and sediment yield at Draix, France', Journal of Hydrology, 235(1), pp. 44-62.
Lʹvovich, M.I. (1979) World water resources and their future. American Geophysical
Union, Washington DC, USA, 264pp.
MacDonald, A., Davies, J., Calow, R. and Chilton, J. (2005) Developing groundwater: a
guide for rural water supply. ITDG publishing, Bradford, UK, 358pp.
MacDonald, A.M., Bonsor, H.C., Calow, R.C., Taylor, R.G., Lapworth, D.J., Maurice,
L., Tucker, J. and O Dochartaigh, B.E. (2011) Groundwater resilience to climate
change in Africa. British Geological Survey Open Report OR/11/031, Wallingford,
UK, 25pp.
MacDonald, A.M., Bonsor, H.C., Dochartaigh, B.E.O. and Taylor, R.G. (2012a)
'Quantitative maps of groundwater resources in Africa', Environmental Research
Letters, 7(2), p. 7.
MacDonald, A.M., Calow, R.C., MacDonald, D.M.J., Darling, W.G. and Dochartaigh,
B.E.O. (2009) 'What impact will climate change have on rural groundwater supplies
in Africa?', Hydrological Sciences Journal-Journal Des Sciences Hydrologiques,
54(4), pp. 690-703.
MacDonald, A.M. and Davies, J. (2000) 'A brief review of groundwater for rural water
supply in sub-Saharan Africa', British Geological Survey Technical Report
WC/00/33.
MacDonald, A.M., Davies, J. and Calow, R.C. (2008) 'African hydrogeology and rural
water supply', Applied Groundwater Studies in Africa. IAH Selected Papers in
Hydrogeology, 13, pp. 127-148.
MacDonald, A.M., Taylor, R.G. and Bonsor, H.C. (2012b) 'Is there sufficient water to
support the intensification of agriculture from ‘land grabs’?', Handbook of Land
and Water Grabs in Africa: Foreign direct investment and food and water security,
p. 376.
250
Macdonald, D.M.J. and Edmunds, W.M. (2014) 'Estimation of groundwater recharge in
weathered basement aquifers, southern Zimbabwe; a geochemical approach',
Applied Geochemistry, 42, pp. 86-100.
Macdonald, D.M.J., Thompson, D.M. and Herbert, R. (1995) 'Sustainability of yield from
wells and boreholes in crystalline basement aquifers', British Geological Survey
Technical Report WC/95/50, Keyworth, UK.
Mahé, G. and Paturel, J.-E. (2009) '1896–2006 Sahelian annual rainfall variability and
runoff increase of Sahelian Rivers', Comptes Rendus Geoscience, 341(7), pp. 538-
546.
Maidment, R.I., Grimes, D., Allan, R.P., Tarnavsky, E., Stringer, M., Hewison, T.,
Roebeling, R. and Black, E. (2014) 'The 30 year TAMSAT African Rainfall
Climatology And Time series (TARCAT) data set', Journal of Geophysical
Research: Atmospheres, 119(18), pp. 10,619-10,644.
Maier, C.A., Albaugh, T.J., Cook, R.I., Hall, K., McInnis, D., Johnsen, K.H., Johnson, J.,
Rubilar, R.A. and Vose, J.M. (2017) 'Comparative water use in short-rotation
Eucalyptus benthamii and Pinus taeda trees in the Southern United States', Forest
Ecology and Management, 397, pp. 126-138.
Martin, N. and Van De Giesen, N. (2005) 'Spatial distribution of groundwater production

and development potential in the Volta River basin of Ghana and Burkina Faso',
Water International, 30(2), pp. 239-249.
Mau, D.P. and Winter, T.C. (1997) 'Estimating ground‐water recharge from streamflow
hydrographs for a small mountain watershed in a temperate humid climate, New
Hampshire, USA', Groundwater, 35(2), pp. 291-304.
Maystadt, J.-F. and Ecker, O. (2014) 'Extreme Weather and Civil War: Does Drought
Fuel Conflict in Somalia through Livestock Price Shocks?', American Journal of
Agricultural Economics, 96(4), pp. 1157-1182.
McCartney, M., Alemayehu, T., Shiferaw, A. and Awulachew, S. (2010) Evaluation of

current and future water resources development in the Lake Tana Basin, Ethiopia.
Working Paper No. 134, International Water Management Institute (IWMI),
Colombo, Sri Lanka, 44pp.
251
McCartney, M.P. and Neal, C. (1999) 'Water flow pathways and the water balance within
a headwater catchment containing a dambo: inferences drawn from hydrochemical
investigations', Hydrology and Earth System Sciences, 3(4), pp. 581-591.
McClain, M.E. (2013) 'Balancing Water Resources Development and Environmental

Sustainability in Africa: A Review of Recent Research Findings and Applications',
Ambio, 42(5), pp. 549-565.
McDonald, M.G. and Harbaugh, A.W. (1988) 'A modular three-dimensional finite-
difference ground-water flow model', US Geol Surv, Techniques of water
resources investigations Book 6, Chap A1, Washington DC, USA, 586pp.
McFarlane, M.J. (1989) 'Dambos–their characteristics and geomorphological evolution

in parts of Malawi and Zimbabwe, with particular reference to their role in the
hydrogeological regime of surviving areas of African surface', Proceedings of the
groundwater exploration and development in crystalline basement aquifers
workshop, Commonwealth Science Council Technical Paper, 273, pp. 254-302.
McKee, T.B., Doesken, N.J. and Kleist, J. (1993) Proceedings of the 8th Conference on
Applied Climatology. American Meteorological Society Boston, MA, USA.
McPherson, E.G., Xiao, Q., van Doorn, N.S., Peper, P.J. and Teach, E. (2017) 'Surface
Storage of Rainfall in Tree Crowns: Not All Trees Are Equal', Arborist News, June,
pp. 30-33.
Mechal, A., Wagner, T. and Birk, S. (2015) 'Recharge variability and sensitivity to
climate: The example of Gidabo River Basin, Main Ethiopian Rift', Journal of
Hydrology: Regional Studies, 4, pp. 644-660.
Megerssa, B., Esayas, A. and Mohamed, A. (2014) 'Socio-Economic Impact of Khat in

Mana District, Jimma Zone, South Western Ethiopia', Discourse Journal of
Agriculture and Food Sciences, 2(2), pp. 21-32.
Mellor, J.W. (2014) 'High rural population density Africa – What are the growth
requirements and who participates?', Food Policy, 48(Supplement C), pp. 66-75.
Mengistu, D.T. and Sorteberg, A. (2012) 'Sensitivity of SWAT simulated streamflow to

climatic changes within the Eastern Nile River basin', Hydrology and Earth System
Sciences, 16(2), pp. 391-407.
252
Merrill, G.P. (1897) A treatise on rocks: Rock-weathering and soils. Macmillan
Company, London, UK, 484pp.
Meshesha, D.T., Tsunekawa, A., Tsubo, M., Ali, S.A. and Haregeweyn, N. (2014) 'Land-
use change and its socio-environmental impact in Eastern Ethiopia’s highland',
Regional environmental change, 14(2), pp. 757-768.
MetaMeta (2014) Rope Pump Improvement Ethiopia. Final Report: Ministry of Water,
Irrigation and Energy (MOWIE), The Netherlands, 98pp.
Meyboom, P. (1961) 'Estimating ground-water recharge from stream hydrographs',

Journal of Geophysical Research, 66(4), pp. 1203-1214.
Meyer, W.B. and Turner, B.L. (1992) 'Human population growth and global land-
use/cover change', Annual review of ecology and systematics, 23(1), pp. 39-61.
Meze-Hausken, E. (2004) 'Contrasting climate variability and meteorological drought

with perceived drought and climate change in northern Ethiopia', Clim Res, 27, pp.
19-31.
Misstear, B., Banks, D. and Clark, L. (2007) Water wells and boreholes. John Wiley &
Sons, Chichester, UK, 498pp.
Misstear, B.D.R. (2000) Proc. National Hydrology Seminar on River Basin Management
(Irish National Committees of the International Hydrology Programme and the
International Committee for Irrigation and Drainage), Tullamore.
Misstear, B.D.R., Brown, L. and Johnston, P.M. (2009) 'Estimation of groundwater

recharge in a major sand and gravel aquifer in Ireland using multiple approaches',
Mitchell, T.D., Carter, T.R., Jones, P.D., Hulme, M. and New, M. (2004) 'A
comprehensive set of high-resolution grids of monthly climate for Europe and the
globe: the observed record (1901–2000) and 16 scenarios (2001–2100)', Tyndall
centre for climate change research working paper, 55(0), p. 25.
Moench, A.F. (1985) 'Transient flow to a large-diameter well in an aquifer with storative
semiconfining layers', Water Resources Research, 21(8), pp. 1121-1131.
Mohr, P. (1983) 'Ethiopian flood basalt province', Nature, 303(5918), pp. 577-584.
253
Mohr, P.A. (1963) The geology of Ethiopia. University College of Addis Ababa Press,
Addis Ababa, Ethiopia, 7pp.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D. and Veith,
T.L. (2007) 'Model evaluation guidelines for systematic quantification of accuracy
in watershed simulations', Transactions of the ASABE, 50(3), pp. 885-900.
Moroni, M.T., Worledge, D. and Beadle, C.L. (2003) 'Root distribution of Eucalyptus
nitens and E. globulus in irrigated and droughted soil', Forest Ecology and
Management, 177(1), pp. 399-407.
Motovilov, Y.G., Gottschalk, L., Engeland, K. and Rodhe, A. (1999) 'Validation of a

distributed hydrological model against spatial observations', Agricultural and
Forest Meteorology, 98, pp. 257-277.
Mourato, S., Moreira, M. and Corte-Real, J. (2014) 'Water availability in southern

Portugal for different climate change scenarios subjected to bias correction',
Journal of Urban and Environmental Engineering, 8(1).
MoWE (2009) Strategic Sector Investment Plan for the Water and Sanitation Sector in
Uganda.Government of Uganda Ministry of Water and Environment, Kampala,
Uganda, 310pp.
MoWIE (2015) Ethiopian Nile Irrigation and Drainage Project. Final Resettlement
Action Plan for RIBB Irrigation and Drainage Project: Phase I. The Federal
Democratic Republic of Ethiopia Ministry of Water, Irrigation and Energy, Addis
Ababa, Ethiopia, 91pp.
MoWR (2011) Ethiopia: Strategic Framework for Managed Groundwater Development.

Ethiopian Ministry of Water Resources, Addis Ababa, 60pp.
Myneni, R.B., Ramakrishna, R., Nemani, R. and Running, S.W. (1997) 'Estimation of
global leaf area index and absorbed PAR using radiative transfer models', IEEE
Transactions on Geoscience and remote sensing, 35(6), pp. 1380-1393.
Nachtergaele, F., van Velthuizen, H., Verelst, L., Batjes, N.H., Dijkshoorn, K., van
Engelen, V.W.P., Fischer, G., Jones, A. and Montanarela, L. (2010) Proceedings of
the 19th World Congress of Soil Science, Soil Solutions for a Changing World,
Brisbane, Australia, 1-6 August 2010.
254
Nagayets, O. (2005) Small farms: current status and key trends. Prepared for the Future
of Small Farms Research Workshop. Wyne College, June 26-29, 2005. The 2020
Vision for Food. Agriculture, and the Environment, Information Brief. International
Food Policy Research Institute (IFPRI). Washington, DC, USA, 365-367.
Nakicenovic, N. and Swart, R. (2000) 'Special report on emissions scenarios', Cambridge

University Press, Cambridge, UK, 612pp.
Namara, R.E., Awuni, J.A., Barry, B., Giordano, M., Hope, L., Owusu, E.S. and Forkuor,
G. (2011) 'Smallholder shallow groundwater irrigation development in the upper
east region of Ghana', International Water Management Institute (IWMI), IWMI
Working Paper 143, Colombo, Sri Lanka, 35pp.
Namara, R.E., Gebregziabher, G., Giordano, M. and De Fraiture, C. (2013) 'Small pumps
and poor farmers in Sub-Saharan Africa: an assessment of current extent of use and
poverty outreach', Water International, 38(6), pp. 827-839.
Narayanamoorthy, A. (2010) 'India's groundwater irrigation boom: can it be sustained?',

Water Policy, 12(4), pp. 543-563.
NASA (2016) 'Amazing GRACE https://science.nasa.gov/science-news/science-at-

nasa/2001/ast30oct_1', Accessed 21 December 16.
Naseela, E.K., Dodamani, B.M. and Chandran, C. (2015) 'Estimation of Runoff Using
NRCS-CN Method and SHETRAN Model', Int. Adv. Res. J. Sci. Eng. Technol, 2,
pp. 23-28.
Nash, J.E. and Sutcliffe, J.V. (1970) 'River flow forecasting through conceptual models
part I - A discussion of principles', Journal of Hydrology, 10(3), pp. 282-290.
Nature (2015) 'Rise of the citizen scientist', [Editorial 18-8-15].
Nedaw, D. (2010) 'Water Balance and Groundwater Quality of Koraro Area, Tigray,
Northern Ethiopia', Momona Ethiopian Journal of Science, 2(2), pp. 110-127.
Netsanet, K. (2007) Groundwater Resources Evaluation and Management in Dugda

Woreda, Central Rift Valley, Ethiopia. MSc thesis, Department of Earth Sciences,
Addis Ababa University, Ethiopia.
255
New, M., Todd, M., Hulme, M. and Jones, P. (2001) 'Precipitation measurements and
trends in the twentieth century', International Journal of Climatology, 21(15), pp.
1889-1922.
Ngigi, S.N. and Denning, G. (2009) Climate change adaptation strategies: water
resources management options for smallholder farming systems in sub-Saharan
Africa. The MDG Centre and The Earth Institute at Columbia University, New
York, USA, 189pp.
Nicholson, S. (1984) 'Appendix B: The Climatology of Sub-Saharan Africa', in

Environmental Change in the West African Sahel. National Research Council
(U.S.), Washington DC, USA.
Nicholson, S.E. (1981) 'The historical climatology of Africa', Climate and History, pp.
249-270.
Nicholson, S.E. (2001) 'Climatic and environmental change in Africa during the last two
centuries', Climate Research, 17(2), pp. 123-144.
Nicholson, S.E., Some, B., McCollum, J., Nelkin, E., Klotter, D., Berte, Y., Diallo, B.M.,
Gaye, I., Kpabeba, G., Ndiaye, O., Noukpozounkou, J.N., Tanu, M.M., Thiam, A.,
Toure, A.A. and Traore, A.K. (2003) 'Validation of TRMM and Other Rainfall
Estimates with a High-Density Gauge Dataset for West Africa. Part II: Validation
of TRMM Rainfall Products', Journal of Applied Meteorology, 42(10), pp. 1355-
1368.
NOAA (2017) 'National Centers for Environmental Information, National Oceanic

Atmospheric Administration https://www.ncdc.noaa.gov/', Accessed 24 August 17.
Nuramo, D. (2016) Temporal changes in Groundwater Recharge in the Upper Awash

Basin with particular emphasis to Becho and Koka areas, Central Ethiopia. MSc
thesis, Department of Earth Sciences, Addis Ababa University, Ethiopia.
Nyssen, J., Haile, M., Naudts, J., Munro, N., Poesen, J., Moeyersons, J., Frankl, A.,
Deckers, J. and Pankhurst, R. (2009) 'Desertification? Northern Ethiopia re-
photographed after 140 years', Science of The Total Environment, 407(8), pp. 2749-
2755.
256
O’Donnell, G.M. (2012) Technical Note On Data Quality Control, Infilling and Record
Extension (Unpublished). NBI Water Resources Planning and Management Project,
Nile Basin Decision Support System (DSS).
Olaniyan, I.O., Agunwamba, J.C. and Ademiluyi, J.O. (2010) 'Assessment of aquifer
characteristics in relation to rural water supply in part of Northern Nigeria',
Researcher, 2(3), pp. 22-27.
Olson, G. (2012) Soils and the environment: A guide to soil surveys and their
applications. Springer Science & Business Media, Berlin, Germany.
Onwuka, O.S., Uma, K.O. and Ezeigbo, H.I. (2004) 'Potability of shallow groundwater
in Enugu town, southeastern Nigeria', Global Journal of Environmental Sciences,
3(1), pp. 33-39.
Op de Hipt, F., Diekkrüger, B., Steup, G., Yira, Y., Hoffmann, T. and Rode, M. (2017)
'Applying SHETRAN in a Tropical West African Catchment (Dano, Burkina
Faso)—Calibration, Validation, Uncertainty Assessment', Water, 9(2), p. 101.
Oughton, E. and Gebregziabher, G. (2014) Livelihoods and the institutional context:

potential opportunities and constraints associated with community management of
shallow groundwater irrigation. Available at:
https://research.ncl.ac.uk/amgraf/outputs/ (Accessed: 5 June 2017).
Owen, R.J.S. (1989) 'The use of shallow alluvial aquifers for small scale irrigation with
reference to Zimbabwe', Final report of ODA Project R4239, University of
Zimbabwe and Southampton University, UK, 121pp.
Owor, M., Taylor, R.G., Tindimugaya, C. and Mwesigwa, D. (2009) 'Rainfall intensity
and groundwater recharge: empirical evidence from the Upper Nile Basin',
Environmental Research Letters, 4(3), p. 6.
Oyarzún, R., Jofré, E., Maturana, H., Oyarzún, J. and Aguirre, E. (2014) 'Use of 222radon
as a simple tool for surface water–groundwater connectivity assessment: a case
study in the arid Limarí basin, north-central Chile', Water and Environment Journal,
28(3), pp. 418-422.
257
Palmer-Jones, R.W. (1992) 'Sustaining serendipity? Groundwater irrigation, growth of
agricultural production, and poverty in Bangladesh', Economic and Political
Weekly, pp. A128-A140.
Papadopulos, I.S. and Cooper, H.H. (1967) 'Drawdown in a well of large diameter', Water
Resources Research, 3(1), pp. 241-244.
Parkin, G. (1996) 'A three-dimensional variably-saturated subsurface modelling system

for river basins', PhD thesis, School of Civil Engineering and Geosciences,
University of Newcastle, UK.
Parkin, G., Birkinshaw, S.J., Younger, P.L., Rao, Z. and Kirk, S. (2007) 'A numerical
modelling and neural network approach to estimate the impact of groundwater
abstractions on river flows', Journal of Hydrology, 339(1), pp. 15-28.
Parkin, G., Odonnell, G., Ewen, J., Bathurst, J.C., Oconnell, P.E. and Lavabre, J. (1996)
'Validation of catchment models for predicting land-use and climate change impacts
.2. Case study for a Mediterranean catchment', Journal of Hydrology, 175(1-4), pp.
595-613.
Pavelic, P., Giordano, M., Keraita, B., Ramesh, V. and Rao, T. (2012) 'Groundwater
availability and use in Sub-Saharan Africa: a review of 15 countries', International
Water Management Institute (IWMI), Colombo, Sri Lanka, 274pp.
Pavelic, P., Villholth, K.G., Shu, Y.Q., Rebelo, L.M. and Smakhtin, V. (2013a)
'Smallholder groundwater irrigation in Sub-Saharan Africa: country-level estimates
of development potential', Water International, 38(4), pp. 392-407.
Pavelic, P., Villholth, K.G. and Verma, S. (2013b) 'Identifying the barriers and pathways
forward for expanding the use of groundwater for irrigation in Sub-Saharan Africa',
Water International, 38(4), pp. 363-368.
Peel, M., H. S. Chiew, F., Western, A. and McMahon, T. (2000) Extension of Unimpaired
Monthly Streamflow Data and Regionalisation of Parameter Values to Estimate
Streamflow in Ungauged Catchments. National Land and Water Resources Audit,
Canberra, Australia, 37pp.
258
Peel, M.C., Finlayson, B.L. and McMahon, T.A. (2007) 'Updated world map of the
Köppen-Geiger climate classification', Hydrology and Earth System Sciences
Discussions Discussions, 4(2), pp. 439-473.
Pegram, G. and Bardossy, A. (2013) 'Downscaling Regional Circulation Model rainfall

to gauge sites using recorrelation and circulation pattern dependent quantile-
quantile transforms for quantifying climate change', Journal of Hydrology, 504, pp.
142-159.
Peters, A.J., Walter-Shea, E.A., Ji, L., Vina, A., Hayes, M. and Svoboda, M.D. (2002)
'Drought monitoring with NDVI-based standardized vegetation index',
Photogrammetric engineering and remote sensing, 68(1), pp. 71-75.
Peterson, T.C., Easterling, D.R., Karl, T.R., Groisman, P., Nicholls, N., Plummer, N.,
Torok, S., Auer, I., Boehm, R., Gullett, D., Vincent, L., Heino, R., Tuomenvirta,
H., Mestre, O., Szentimrey, T., Salinger, J., Forland, E.J., Hanssen-Bauer, I.,
Alexandersson, H., Jones, P. and Parker, D. (1998) 'Homogeneity adjustments of in
situ atmospheric climate data: A review', International Journal of Climatology,
18(13), pp. 1493-1517.
Pettorelli, N., Vik, J.O., Mysterud, A., Gaillard, J.-M., Tucker, C.J. and Stenseth, N.C.
(2005) 'Using the satellite-derived NDVI to assess ecological responses to
environmental change', Trends in Ecology & Evolution, 20(9), pp. 503-510.
Pik, R., Deniel, C., Coulon, C., Yirgu, G., Hofmann, C. and Ayalew, D. (1998) 'The
northwestern Ethiopian Plateau flood basalts. Classification and spatial distribution
of magma types', Journal of Volcanology and Geothermal Research, 81(1-2), pp.
91-111.
Pohjonen, V. and Pukkala, T. (1990) 'Eucalyptus globulus in Ethiopian forestry', Forest

Ecology and Management, 36(1), pp. 19-31.
Poppe, L., Frankl, A., Poesen, J., Admasu, T., Dessie, M., Adgo, E., Deckers, J. and
Nyssen, J. (2013) 'Geomorphology of the Lake Tana basin, Ethiopia', Journal of
Maps, 9(3), pp. 431-437.
Prave, A.R., Bates, C.R., Donaldson, C.H., Toland, H., Condon, D.J., Mark, D. and Raub,
T.D. (2016) 'Geology and geochronology of the Tana Basin, Ethiopia: LIP
259
volcanism, super eruptions and Eocene–Oligocene environmental change', Earth
and Planetary Science Letters, 443(Supplement C), pp. 1-8.
PRI (2017) Public Radio International Available at: https://cdn1.pri.org/ (Accessed: 14

June 2017).
Prudhomme, C., Jakob, D. and Svensson, C. (2003) 'Uncertainty and climate change
impact on the flood regime of small UK catchments', Journal of hydrology, 277(1),
pp. 1-23.
Rashid, S., Tefera, N., Minot, N. and Ayele, G. (2013) 'Can modern input use be promoted
without subsidies? An analysis of fertilizer in Ethiopia', Agricultural Economics,
44(6), pp. 595-611.
REACH (2015) 'Country Diagnostic Report, Ethiopia. REACH Working Paper 2'.
Oxford, UK: REACH, University of Oxford, p. 28 pp. Available at:
http://reachwater.org.uk/resource/country-diagnostic-report-ethiopia/.
REACH (2018) REACH: Improving water security for the poor. Available at:
https://reachwater.org.uk/ (Accessed: 1 February 2018).
Refsgaard, J.C. (1996) 'Terminology, Modelling Protocol And Classification of

Hydrological Model Codes', in Abbott, M.B. and Refsgaard, J.C. (eds.) Distributed
Hydrological Modelling. Dordrecht: Springer Netherlands, pp. 17-39.
Refsgaard, J.C. (1997) 'Parameterisation, calibration and validation of distributed

hydrological models', Journal of hydrology, 198(1), pp. 69-97.
Refsgaard, J.C. and Abbott, M.B. (1990) 'The role of distributed hydrological modelling
in water resources management', in Distributed hydrological modelling. Springer,
pp. 1-16.
Refsgaard, J.C. and Knudsen, J. (1996) 'Operational validation and intercomparison of

different types of hydrological models', Water Resources Research, 32(7), pp.
2189-2202.
Refsgaard, J.C., Storm, B. and Clausen, T. (2010) 'Système Hydrologique Europeén

(SHE): review and perspectives after 30 years development in distributed
physically-based hydrological modelling', Hydrology Research, 41(5), pp. 355-377.
260
Reshmidevi, T.V. and Nagesh Kumar, D. (2014) 'Modelling the impact of extensive
irrigation on the groundwater resources', Hydrological Processes, 28(3), pp. 628-
639.
Reynolds, C. (2008) 'Ethiopia 2008 crop assessment travel report', USDA Foreign
Agriculture Service. Available at http://www.pecad.fas.usda.gov.
Reys, A. (2016) Ground Water Potential Evaluation and Use Trends in Upper Awash
Basin: with Special Emphasis to Koka-Becho area. MSc thesis, Department of
Earth Sciences, Addis Ababa University, Ethiopia.
Richards, P. (1985) Indigenous agricultural revolution: ecology and food production in

West Africa. Westview Press, Colorado, USA.
Richardson, D.M. and Higgins, S.I. (2000) 'Pines as invaders in the southern hemisphere',
Ecology and biogeography of Pinus, p. 450.
Ridder, D. and Pahl-Wostl, C. (2005) 'Participatory Integrated Assessment in local level

planning', Regional Environmental Change, 5(4), pp. 188-196.
Rienecker, M.M., Suarez, M.J., Gelaro, R., Todling, R., Bacmeister, J., Liu, E.,
Bosilovich, M.G., Schubert, S.D., Takacs, L., Kim, G.-K., Bloom, S., Chen, J.,
Collins, D., Conaty, A., da Silva, A., Gu, W., Joiner, J., Koster, R.D., Lucchesi, R.,
Molod, A., Owens, T., Pawson, S., Pegion, P., Redder, C.R., Reichle, R., Robertson,
F.R., Ruddick, A.G., Sienkiewicz, M. and Woollen, J. (2011) 'MERRA: NASA’s
Modern-Era Retrospective Analysis for Research and Applications', Journal of
Climate, 24(14), pp. 3624-3648.
Rijsberman, F.R. (2006) 'Water scarcity: fact or fiction?', Agricultural water

management, 80(1), pp. 5-22.
Risser, D.W., Gburek, W.J. and Folmar, G.J. (2005) Comparison of methods for
estimating ground-water recharge and base flow at a small watershed underlain by
fractured bedrock in the eastern United States. US Department of the Interior, US
Geological Survey.
Robins, N.S., Chilton, P.J. and Cobbing, J.E. (2007) 'Adapting existing experience with
aquifer vulnerability and groundwater protection for Africa', Journal of African
Earth Sciences, 47(1), pp. 30-38.
261
Robins, N.S., Davies, J., Farr, J.L. and Calow, R.C. (2006) 'The changing role of
hydrogeology in semi-arid southern and eastern Africa', Hydrogeology Journal,
14(8), pp. 1483-1492.
Robins, N.S. and Fergusson, J. (2014) 'Groundwater scarcity and conflict – managing
hotspots', Earth Perspectives, 1(1), p. 6.
Robinson, M., Kulie, M.S., Silberstein, D.S., Marks, D.A., Wolff, D.B., Amatai, E.,
Ferrier, B.S., Fisher, B.L. and Wang, J. (2000) 'Evolving improvements to TRMM
ground validation rainfall estimates', Physics and Chemistry of the Earth, Part B:
Hydrology, Oceans and Atmosphere, 25(10–12), pp. 971-976.
Rockström, J., Barron, J. and Fox, P. (2002) 'Rainwater management for increased
productivity among small-holder farmers in drought prone environments', Physics
and Chemistry of the Earth, Parts A/B/C, 27(11), pp. 949-959.
Rodriguez Suarez, J.A., Diaz‐Fierros, F., Perez, R. and Soto, B. (2014) 'Assessing the
influence of afforestation with Eucalyptus globulus on hydrological response from
a small catchment in northwestern Spain using the HBV hydrological model',
Hydrological processes, 28(22), pp. 5561-5572.
Roggeri, H. (2013) Tropical freshwater wetlands: a guide to current knowledge and

sustainable management. Springer Science & Business Media, Berlin, Germany.
Roohi, R. and Webb, J.A. (2016) 'Thermal and visible remote sensing for estimation of
evapotranspiration of rainfed agrosystems and its impact on groundwater in SE
Australia', Thermosense: Thermal Infrared Applications XXXVIII. International
Society for Optics and Photonics, p. 98610A.
Rossiter, D.G., Liu, J., Carlisle, S. and Zhu, A.X. (2015) 'Can citizen science assist digital
soil mapping?', Geoderma, 259–260, pp. 71-80.
Roy, H.E., Pocock, M.J.O., Preston, C.D., Roy, D.B., Savage, J., Tweddle, J.C. and
Robinson, L.D. (2012) 'Understanding citizen science and environmental
monitoring', Final report on behalf of UK Environmental Observation Framework.
NERC Centre for Ecology and Hydrology and Natural History Museum.
262
Rozanski, K., Araguas-Araguas, L. and Gonfiantini, R. (1996) 'Isotope patterns of
precipitation in the East African region', The Limnology, climatology and
paleoclimatology of the East African Lakes, pp. 79-94.
Rushton, K. (1997) 'Recharge from permanent water bodies', Recharge of phreatic

aquifers in (semi) arid areas. AA Balkema, Rotterdam, pp. 215-255.
Santos, J.M., Verheijen, F.G.A., Tavares Wahren, F., Wahren, A., Feger, K.-H., Bernard-
Jannin, L., Rial-Rivas, M.E., Keizer, J.J. and Nunes, J.P. (2016) 'Soil Water
Repellency Dynamics in Pine and Eucalypt Plantations in Portugal – A High-
resolution Time Series', Land Degradation & Development, 27(5), pp. 1334-1343.
Scanlon, B.R., Healy, R.W. and Cook, P.G. (2002) 'Choosing appropriate techniques for
quantifying groundwater recharge', Hydrogeology Journal, 10(1), pp. 18-39.
Scanlon, B.R., Reedy, R.C., Stonestrom, D.A., Prudic, D.E. and Dennehy, K.F. (2005)
'Impact of land use and land cover change on groundwater recharge and quality in
the southwestern US', Global Change Biology, 11(10), pp. 1577-1593.
Schmidt, J., Elliott, S. and McKergow, L. (2008) 'Land-use impacts on catchment erosion
for the Waitetuna catchment, New Zealand', IAHS publication, 325, p. 453.
Scott, D.F. and Smith, R.E. (1997) 'Preliminary empirical models to predict reductions in
total and low flows resulting from afforestation', Water SA, 23(2), pp. 135-140.
Scott, K. and Pain, C. (2009) Regolith science. CSIRO Publishing, Australia, 461pp.
Segele, Z.T. and Lamb, P.J. (2005) 'Characterization and variability of Kiremt rainy
season over Ethiopia', Meteorology and Atmospheric Physics, 89(1-4), pp. 153-180.
Seleshi, Y. and Zanke, U. (2004) 'Recent changes in rainfall and rainy days in Ethiopia',
International journal of climatology, 24(8), pp. 973-983.
Setegn, S.G., Rayner, D., Melesse, A.M., Dargahi, B., Srinivasan, R. and Wörman, A.
(2011) 'Climate Change Impact on Agricultural Water Resources Variability in the
Northern Highlands of Ethiopia', in Melesse, A.M. (ed.) Nile River Basin:
Hydrology, Climate and Water Use. Springer, Dordrecht, The Netherlands, pp. 241-
265.
263
Setegn, S.G., Srinivasan, R. and Dargahi, B. (2008) 'Hydrological modelling in the Lake
Tana Basin, Ethiopia using SWAT model', The Open Hydrology Journal, 2(1).
Setegn, S.G., Srinivasan, R., Melesse, A.M. and Dargahi, B. (2010) 'SWAT model
application and prediction uncertainty analysis in the Lake Tana Basin, Ethiopia',
Sewmehon, D. (2012) 'Water and rural livelihoods in the crop-livestock system of

Amhara Region, Ethiopia: Multiple use system (MUS) approach for water
productivity improvement', Proceedings of the second national workshop on
challenges and opportunities of water resources management in Tana Basin, Upper
Blue Nile Basin, Ethiopia, 26–27 Mar 2012. Blue Nile Water Institute - Bahir Dar
University (BNWI-BDU), Bahir Dar, Ethiopia, pp 251–257.
Shah, T., Roy, A.D., Qureshi, A.S. and Wang, J.X. (2003) 'Sustaining Asia's groundwater
boom: An overview of issues and evidence', Natural Resources Forum, 27(2), pp.
130-141.
Shanko, D. and Camberlin, P. (1998) 'The effects of the Southwest Indian Ocean tropical
cyclones on Ethiopian drought', International Journal of Climatology, 18(12), pp.
1373-1388.
Sharma, M.L. (1984) 'Evapotranspiration from a Eucalyptus community', Agricultural

Water Management, 8(1-3), pp. 41-56.
Sharp, J.M. (2014) Fractured Rock Hydrogeology. CRC Press, Boca Raton, USA.
Shawul, A.A., Alamirew, T. and Dinka, M.O. (2013) 'Calibration and validation of
SWAT model and estimation of water balance components of Shaya mountainous
watershed, Southeastern Ethiopia', Hydrology and Earth System Sciences
Discussions, 10(11), pp. 13955-13978.
Sheen, K.L., Smith, D.M., Dunstone, N.J., Eade, R., Rowell, D.P. and Vellinga, M.
(2017) 'Skilful prediction of Sahel summer rainfall on inter-annual and multi-year
timescales', Nature Communications, 8, p. ncomms14966.
Sheffield, J., Wood, E.F., Chaney, N., Guan, K., Sadri, S., Yuan, X., Olang, L., Amani,
A., Ali, A., Demuth, S. and Ogallo, L. (2013) 'A Drought Monitoring and
264
Forecasting System for Sub-Sahara African Water Resources and Food Security',
Bulletin of the American Meteorological Society, 95(6), pp. 861-882.
Shimelis, A., Megerssa, O. and Fantahun, A. (2014) 'Estimation of Groundwater

Recharge Using Water Balance Model Coupled with Base flow Separation in
Bulbul River Catchment of Gilgel-Gibe River Basin, Ethiopia', Asian Journal Of
Applied Science And Engineering, 3(2), pp. 235-243.
Siebert, S., Burke, J., Faures, J.M., Frenken, K., Hoogeveen, J., Doll, P. and Portmann,
F.T. (2010) 'Groundwater use for irrigation - a global inventory', Hydrology and
Earth System Sciences, 14(10), pp. 1863-1880.
Sikka, A.K., Samra, J.S., Sharda, V.N., Samraj, P. and Lakshmanan, V. (2003) 'Low flow
and high flow responses to converting natural grassland into bluegum (Eucalyptus
globulus) in Nilgiris watersheds of South India', Journal of Hydrology, 270(1), pp.
12-26.
Simmers, I. (1988) Estimation of natural groundwater recharge. NATO ASI Series.

Series C. Mathematical and physical sciences ; vol. 222, D Reidel Publishing
Company, Dordrecht, Holland, 514pp.
Sintayehu, L.G. (2009) Integrated Hydrogeological Investigation Of Upper Bilate River

Catchment, South Western escarpment of Main Ethiopian Rift. MSc thesis,
Smakhtin, V.U. (2001) 'Low flow hydrology: a review', Journal of hydrology, 240(3), pp.
147-186.
Smerdon, B.D. (2017) 'A synopsis of climate change effects on groundwater recharge',
Journal of Hydrology, 555, pp. 125-128.
Sokile, C.S. and van Koppen, B. (2004) 'Local water rights and local water user entities:
the unsung heroines of water resource management in Tanzania', Physics and
Chemistry of the Earth, 29(15-18), pp. 1349-1356.
Sophocleous, M. (1985) 'The Role of Specific Yield in Ground‐Water Recharge

Estimations: A Numerical Study', Groundwater, 23(1), pp. 52-58.
265
Sophocleous, M. (2000) 'From safe yield to sustainable development of water resources—
the Kansas experience', Journal of Hydrology, 235(1), pp. 27-43.
Starkey, E., Parkin, G., Birkinshaw, S., Large, A., Quinn, P. and Gibson, C. (2017)
'Demonstrating the value of community-based (‘citizen science’) observations for
catchment modelling and characterisation', Journal of Hydrology, 548(Supplement
C), pp. 801-817.
Steenhuis, T.S. and Van Der Molen, W.H. (1986) 'The Thornthwaite-Mather procedure
as a simple engineering method to predict recharge', Journal of Hydrology, 84(3),
pp. 221-229.
Stern, R., Rijks, D., Dale, I. and Knock, J. (2006) 'INSTAT climatic guide', Statistical
Services Centre, University of Reading, Reading, UK.
Subyani, A.M. (2004) 'Use of chloride-mass balance and environmental isotopes for
evaluation of groundwater recharge in the alluvial aquifer, Wadi Tharad, western
Saudi Arabia', Environmental Geology, 46(6-7), pp. 741-749.
Sudheer, K.P., Gosain, A.K. and Ramasastri, K.S. (2002) 'A data‐driven algorithm for
constructing artificial neural network rainfall‐runoff models', Hydrological
processes, 16(6), pp. 1325-1330.
Sullivan, B.L., Wood, C.L., Iliff, M.J., Bonney, R.E., Fink, D. and Kelling, S. (2009)
'eBird: A citizen-based bird observation network in the biological sciences',
Biological Conservation, 142(10), pp. 2282-2292.
Sun, X., Xu, Y., Jovanovic, N.Z., Kapangaziwiri, E., Brendonck, L. and Bugan, R.D.H.
(2013) 'Application of the rainfall infiltration breakthrough (RIB) model for
groundwater recharge estimation in west coastal South Africa', Water SA, 39(2), pp.
221-230.
Sylla, M.B., Giorgi, F., Coppola, E. and Mariotti, L. (2013) 'Uncertainties in daily rainfall
over Africa: assessment of gridded observation products and evaluation of a
regional climate model simulation', International Journal of Climatology, 33(7),
pp. 1805-1817.
266
Symeonakis, E., Bonifaçio, R. and Drake, N. (2009) 'A comparison of rainfall estimation
techniques for sub-Saharan Africa', International Journal of Applied Earth
Observation and Geoinformation, 11(1), pp. 15-26.
Szilagyi, J., Harvey, F.E. and Ayers, J.F. (2003) 'Regional estimation of base recharge to
ground water using water balance and a base‐flow index', Groundwater, 41(4), pp.
504-513.
Tadesse, N., Bheemalingeswara, K. and Abdulaziz, M. (2010a) 'Hydrogeological

investigation and groundwater potential assessment in Haromaya watershed,
Eastern Ethiopia', Momona Ethiopian Journal of Science, 2(1).
Tadesse, N., Tadios, S. and Tesfaye, M. (2010b) 'The Water Balance of May Nugus
Catchment, Tigray, Northern Ethiopia', International Journal of Earth Sciences and
Engineering, 3(5), pp. 609-625.
Takem, G.E., Kuitcha, D., Ako, A.A., Mafany, G.T., Takounjou-Fouepe, A., Ndjama, J.,
Ntchancho, R., Ateba, B.H., Chandrasekharam, D. and Ayonghe, S.N. (2015)
'Acidification of shallow groundwater in the unconfined sandy aquifer of the city
of Douala, Cameroon, Western Africa: implications for groundwater quality and
use', Environmental Earth Sciences, 74(9), pp. 6831-6846.
Takounjou, A.F., Ngoupayou, J.R.N., Riotte, J., Takem, G.E., Mafany, G., Marechal, J.C.
and Ekodeck, G.E. (2011) 'Estimation of groundwater recharge of shallow aquifer
on humid environment in Yaounde, Cameroon using hybrid water-fluctuation and
hydrochemistry methods', Environmental Earth Sciences, 64(1), pp. 107-118.
Tapley, B.D., Bettadpur, S., Ries, J.C., Thompson, P.F. and Watkins, M.M. (2004)
'GRACE measurements of mass variability in the Earth system', Science,
305(5683), pp. 503-505.
Taye, M.T., Willems, P. and Block, P. (2015) 'Implications of climate change on

hydrological extremes in the Blue Nile basin: A review', Journal of Hydrology:
Regional Studies, 4, Part B, pp. 280-293.
Taylor, G. and Eggleton, R.A. (2001) Regolith geology and geomorphology. John Wiley
& Sons, New Jersey, USA, 379pp.
267
Taylor, R., Tindimugaya, C., Barker, J., Macdonald, D. and Kulabako, R. (2010)
'Convergent Radial Tracing of Viral and Solute Transport in Gneiss Saprolite',
Ground Water, 48(2), pp. 284-294.
Taylor, R.G. and Howard, K.W.F. (1998) 'The dynamics of groundwater flow in the
regolith of Uganda', International Contributions to Hydrogeology, 18, pp. 97-114.
Taylor, R.G., Koussis, A.D. and Tindimugaya, C. (2009) 'Groundwater and climate in
Africa—a review', Hydrological Sciences Journal, 54(4), pp. 655-664.
Taylor, R.G., Scanlon, B., Doll, P., Rodell, M., van Beek, R., Wada, Y., Longuevergne,
L., Leblanc, M., Famiglietti, J.S., Edmunds, M., Konikow, L., Green, T.R., Chen,
J., Taniguchi, M., Bierkens, M.F.P., MacDonald, A., Fan, Y., Maxwell, R.M.,
Yechieli, Y., Gurdak, J.J., Allen, D.M., Shamsudduha, M., Hiscock, K., Yeh, P.J.F.,
Holman, I. and Treidel, H. (2013) 'Ground water and climate change', Nature Clim.
Change, 3(4), pp. 322-329.
Tebebu, T.Y., Abiy, A.Z., Zegeye, A.D., Dahlke, H.E., Easton, Z.M., Tilahun, S.A.,
Collick, A.S., Kidnau, S., Moges, S. and Dadgari, F. (2010) 'Surface and subsurface
flow effect on permanent gully formation and upland erosion near Lake Tana in the
northern highlands of Ethiopia', Hydrology and Earth System Sciences, 14(11), pp.
2207-2217.
Tefera, B. (2017) 'Water-Induced Shift of Farming Systems and Value Addition in Lake
Tana Sub-basin: The Case of Rice Production and Marketing in Fogera District,
Northwestern Ethiopia', in Social and Ecological System Dynamics. Springer, pp.
545-562.
Tefera, M., Chernet, T., Haro, W., Teshome, N. and Woldie, K. (1996) Geological map
of Ethiopia. Geological Survey of Ethiopia, Addis Ababa.
Tekle-Haimanot, R., Melaku, Z., Kloos, H., Reimann, C., Fantaye, W., Zerihun, L. and
Bjorvatn, K. (2006) 'The geographic distribution of fluoride in surface and
groundwater in Ethiopia with an emphasis on the Rift Valley', Science of The Total
Environment, 367(1), pp. 182-190.
Teklebirhan, A., Dessie, N. and Tesfamichael, G. (2012) 'Groundwater recharge,

evapotranspiration and surface runoff estimation using WetSpass modeling method
268
in Illala catchment, northern Ethiopia', Momona Ethiopian Journal of Science, 4(2),
pp. 96-110.
Tesfagiorgis, K., Gebreyohannes, T., De Smedt, F., Moeyersons, J., Hagos, M., Nyssen,
J. and Deckers, J. (2011) 'Evaluation of groundwater resources in the Geba basin,
Ethiopia', Bulletin of Engineering Geology and the Environment, 70(3), pp. 461-
466.
Tesfaye, T.G. (2010) Groundwater potential evaluation based on integrated GIS and
remote sensing techniques in Bilate River Catchment: South Rift Valley of Ethiopia.
MSc thesis, Department of Earth Sciences, Addis Ababa University, Ethiopia.
The Herald (2016) Drought mitigation package for Zim. Available at:
http://www.herald.co.zw/drought-mitigation-package-for-zim/ (Accessed: 7
January 2017).
The Namibian (2016) Namibia drought threatens food security – Mutorwa. Available at:
http://www.namibian.com.na/49575/read/Namibia-drought-threatens-food-
security---Mutorwa (Accessed: 7 January 2017).
The Straits Times (2016) Freaky weather the new normal. Available at:
http://www.straitstimes.com/world/freaky-weather-the-new-normal (Accessed: 7
January 2017).
Theis, C.V. (1940) 'The source of water derived from wells', Civil Engineering, 10(5),
pp. 277-280.
Thiem, G. (1906) Hydrologische Methoden: Dissertation zur Erlangung der Wurde eines.
JM Gebhardt.
Thornthwaite, C.W. and Mather, J.R. (1955) 'The Water Balance', Publications in
climatology, 8(1), pp. 1-104, Drexel Institute of Climatology, Centerton, New
Jersey.
Thornthwaite, C.W. and Mather, J.R. (1957) 'Instructions and tables for computing
potential evapotranspiration and the water balance', Publications in Climatology,
X(3), p. 311.
269
Tilahun, K. and Merkel, B.J. (2009) 'Estimation of groundwater recharge using a GIS-
based distributed water balance model in Dire Dawa, Ethiopia', Hydrogeology
Journal, 17(6), pp. 1443-1457.
Todini, E. (2007) 'Hydrological catchment modelling: past, present and future',

Hydrology and Earth System Sciences, 11(1), pp. 468-482.
Toth, J. (1963) 'A theoretical analysis of groundwater flow in small drainage basins',
Journal of geophysical research, 68(16), pp. 4795-4812.
Tourian, M.J., Sneeuw, N. and Bardossy, A. (2013) 'A quantile function approach to
discharge estimation from satellite altimetry (ENVISAT)', Water Resources
Research, 49(7), pp. 4174-4186.
Travelers Today (2016) Sahara Snow Falls Once Again After 37 Years! Available at:
http://www.travelerstoday.com/articles/32053/20161230/sahara-snow-falls-once-
again-after-37-years-here-are-everything-you-should-know-about-it-video.htm
(Accessed: 7 January 2017).
Turc, L. (1954) 'Le bilan d'eau des sols: Relations entre les precipitations, l'evaporation
et l'ecoulement', Annals Agronomiques, (5), pp. 491-595.
Tuzzin de Moraes, M., Debiasi, H., Carlesso, R., Cezar Franchini, J., Rodrigues da Silva,
V. and Bonini da Luz, F. (2016) 'Soil physical quality on tillage and cropping
systems after two decades in the subtropical region of Brazil', Soil and Tillage
Research, 155, pp. 351-362.
UN (2015) United Nations sustainable development goals. Available at:

http://www.un.org/sustainabledevelopment/sustainable-development-goals/
(Accessed: 7 January 2017).
Ut, T.T., Hossain, M. and Janaiah, A. (2000) 'Modern farm technology and infrastructure
in Vietnam: Impact on income distribution and poverty', Economic and Political
Weekly, pp. 4638-4643.
Vallet-Coulomb, C., Legesse, D., Gasse, F., Travi, Y. and Chernet, T. (2001) 'Lake
evaporation estimates in tropical Africa (Lake Ziway, Ethiopia)', Journal of
Hydrology, 245(1–4), pp. 1-18.
270
van Beek, L.P.H., Wada, Y. and Bierkens, M.F.P. (2011) 'Global monthly water stress:
1. Water balance and water availability', Water Resources Research, 47, p. 25.
van der Heijde, P.K.M. (1988) 'Spatial and temporal scales in groundwater modelling', in
Rosswall, T., Woodmansee, R.G. and Risser, P.G. (eds.) Scales and Global Change.
John Wiley & Sons, New Jersey, USA pp. 195-223.
van Griensven, A., Ndomba, P., Yalew, S. and Kilonzo, F. (2012) 'Critical review of
SWAT applications in the upper Nile basin countries', Hydrology and Earth System
Sciences, 16(9), p. 3371.
Van Koppen, B. (2003) 'Water reform in Sub-Saharan Africa: what is the difference?',
Physics and Chemistry of the Earth, Parts A/B/C, 28(20–27), pp. 1047-1053.
Van Lill, W.S., Kruger, F.J. and Van Wyk, D.B. (1980) 'The effect of afforestation with
Eucalyptus grandis Hill ex Maiden and Pinus patula Schlecht. et Cham. on
streamflow from experimental catchments at Mokobulaan, Transvaal', Journal of
Hydrology, 48(1), pp. 107-118.
Vandecasteele, I., Nyssen, J., Clymans, W., Moeyersons, J., Martens, K., Van Camp, M.,
Gebreyohannes, T., Desmedt, F., Deckers, J. and Walraevens, K. (2011)
'Hydrogeology and groundwater flow in a basalt-capped Mesozoic sedimentary
series of the Ethiopian highlands', Hydrogeology journal, 19(3), pp. 641-650.
Varni, M., Comas, R., Weinzettel, P. and Dietrich, S. (2013) 'Application of the water
table fluctuation method to characterize groundwater recharge in the Pampa plain,
Argentina', Hydrological Sciences Journal, 58(7), pp. 1445-1455.
Vegter, J.R. and Pitman, W.V. (2003) Groundwater Recharge Estimation in Southern
Africa. UNESCO IHP Series, Paris, France.
Vianna, G.M.S., Meekan, M.G., Bornovski, T.H. and Meeuwig, J.J. (2014) 'Acoustic
Telemetry Validates a Citizen Science Approach for Monitoring Sharks on Coral
Reefs', PLoS ONE, 9(4), p. e95565.
Vicente-Serrano, S., M., Beguería, S. and López-Moreno, J., I. (2010) 'A Multiscalar
Drought Index Sensitive to Global Warming: The Standardized Precipitation
Evapotranspiration Index', Journal of Climate, 23(7), pp. 1696-1718.
271
Villholth, K.G. (2013) 'Groundwater irrigation for smallholders in Sub-Saharan Africa -
a synthesis of current knowledge to guide sustainable outcomes', Water
International, 38(4), pp. 369-391.
Viste, E., Korecha, D. and Sorteberg, A. (2013) 'Recent drought and precipitation
tendencies in Ethiopia', Theoretical and Applied Climatology, 112(3-4), pp. 535-
551.
von der Heyden, C.J. (2004) 'The hydrology and hydrogeology of dambos: a review',
Progress in Physical Geography, 28(4), pp. 544-564.
von der Heyden, C.J. and New, M.G. (2003) 'The role of a dambo in the hydrology of a
catchment and the river network downstream', Hydrology and Earth System
Sciences, 7(3), pp. 339-357.
Wagener, T. (2003) 'Evaluation of catchment models', Hydrological processes, 17(16),

pp. 3375-3378.
Wale, A., Rientjes, T.H.M., Gieske, A.S.M. and Getachew, H.A. (2009) 'Ungauged
catchment contributions to Lake Tana's water balance', Hydrological processes,
23(26), pp. 3682-3693.
Walker, D. (2016) 'Properties of shallow thin regolith aquifers in sub-Saharan Africa: a

case study from northwest Ethiopia', 7th RWSN Forum "Water for Everyone".
Abidjan, Côte d’Ivoire, Nov-Dec 2016. Available at:
6_walker_david-rev1.pdf Retrieved 14 March 2017. RWSN, St Gallen,
Switzerland.
community-based monitoring programme', Journal of Hydrology, 538, pp. 713-
725.
Groundwater (in press).
272
Walraevens, K., Gebreyohannes Tewolde, T., Amare, K., Hussein, A., Berhane, G.,
Baert, R., Ronsse, S., Kebede, S., Van Hulle, L. and Deckers, J. (2015) 'Water
Balance Components for Sustainability Assessment of Groundwater‐Dependent
Agriculture: Example of the Mendae Plain (Tigray, Ethiopia)', Land Degradation
& Development, 26(7), pp. 725-736.
Walraevens, K., Vandecasteele, I., Martens, K., Nyssen, J., Moeyersons, J.,
Gebreyohannes, T., De Smedt, F., Poesen, J., Deckers, J. and Van Camp, M. (2009)
'Groundwater recharge and flow in a small mountain catchment in northern
Ethiopia', Hydrological Sciences Journal, 54(4), pp. 739-753.
Wang, L., O Dochartaigh, B. and Macdonald, D. (2010) A literature review of recharge

estimation and groundwater resource assessment in Africa. British Geological
Survey.
Wani, S.P., Rockström, J. and Oweis, T.Y. (2009) Rainfed agriculture: unlocking the
potential. CABI, Wallingford, UK, 326pp.
Washington, R., Harrison, M., Conway, D. and Black, E. (2004) African climate report:
a report commissioned by the UK Government to review African climate science,
policy and options for action. Department for Environment, Food and Rural Affairs.
Washington, R., James, R., Pearce, H., Pokam, W.M. and Moufouma-Okia, W. (2013)
'Congo Basin rainfall climatology: can we believe the climate models?',
Philosophical Transactions of the Royal Society of London B: Biological Sciences,
368(1625), p. 20120296.
Washington, R., Kay, G., Harrison, M., Conway, D., Black, E., Challinor, A., Grimes, D.,
Jones, R., Morse, A. and Todd, M. (2006) 'African Climate Change: Taking the
Shorter Route', Bulletin of the American Meteorological Society, 87(10), pp. 1355-
1366.
Watt, S.B. and Wood, W.E. (1977) 'Hand dug wells and their construction', in Hand dug
wells and their construction. Billing & Sons Ltd, Worcester, UK, 254pp.
Wenzel, L.K. (1936) 'Several methods of studying fluctuations of ground‐water levels',

Eos, Transactions American Geophysical Union, 17(2), pp. 400-405.
273
Werner, A.D., Zhang, Q., Xue, L., Smerdon, B.D., Li, X., Zhu, X., Yu, L. and Li, L.
(2013) 'An initial inventory and indexation of groundwater mega-depletion cases',
Water resources management, 27(2), pp. 507-533.
White, D.A., Battaglia, M., Mendham, D.S., Crombie, D.S., Kinal, J. and McGrath, J.F.
(2010) 'Observed and modelled leaf area index in Eucalyptus globulus plantations:
tests of optimality and equilibrium hypotheses', Tree Physiology, 30(7), pp. 831-
844.
WHO (2011) Guidelines for drinking-water quality. World Health Organization, Geneva,
Switzerland.
WHO (2015) 25 Years Progress on Sanitation and Drinking Water - 2015 Update and
MDG Assessment. The Joint Monitoring Programme (JMP) of United Nations
Childrens' Fund (UNICEF) and the World Health Organisation (WHO), Geneva,
Switzerland, 90pp.
WHYMAP (2016) 'www.whymap.org World-wide Hydrogeological Mapping and

Assessment Programme', BGR and UNESCO. , Accessed 21 December 2016.
Wiersinga, R.C. and de Jager, A. (2009) Business opportunities in the Ethiopian fruit and
vegetable sector. Wageningen University and Research Centre, Wageningen,
Netherlands.
Wieser, G., Brito, P., Lorenzo, J.R., González-Rodríguez, Á.M., Morales, D. and
Jiménez, M.S. (2016) 'Canary Island pine (Pinus canariensis), an evergreen species
in a semiarid treeline', in Progress in Botany 77. Springer, pp. 415-435.
Willems, P. (2009) 'A time series tool to support the multi-criteria performance evaluation
of rainfall-runoff models', Environmental Modelling & Software, 24(3), pp. 311-
321.
Willmott, C.J., Robeson, S.M. and Feddema, J.J. (1994) 'Estimating continental and
terrestrial precipitation averages from raingauge networks', International Journal
of Climatology, 14, pp. 403-414.
WMO (2003) Twenty-first status report on implementation of the World Weather Watch:
Forty years of World Weather Watch. Report 957, World Meteorologocal
Organization (WMO), Geneva, Switzerland.
274
WMO (2011) Guide to Climatological Practices, WMO-No. 100. World Meteorological
Organisation (WMO), Geneva, Switzerland.
WMO (2012) 'Standardized precipitation index user guide ', (M. Svoboda, M. Hayes and
D. Wood). World Meteorological Organization (WMO-No. 1090), Geneva,
Switzerland.
Woldearegay, K. (2004) 'A study of water resource potential of Aba’ala Wereda',

Research and development experience on dryland husbandry in Ethiopia. OSSREA
and Mekelle University, Addis Ababa, Ethiopia, pp. 93-107.
Woldearegay, K. and Van Steenbergen, F. (2015) 'Shallow Groundwater Irrigation in

Tigray, Northern Ethiopia: Practices and Issues', in Engineering Geology for
Society and Territory-Volume 3. Springer, pp. 505-509.
Wolff, D.B., Marks, D.A., Amitai, E., Silberstein, D.S., Fisher, B.L., Tokay, A., Wang,
J. and Pippitt, J.L. (2005) 'Ground Validation for the Tropical Rainfall Measuring
Mission (TRMM)', Journal of Atmospheric and Oceanic Technology, 22(4), pp.
365-380.
Wondefrash, M. (2003) 'Wetlands, birds and important bird areas in Ethiopia', in Abebe,
Y.D. and Geheb, K. (eds.) Wetlands of Ethiopia. IUCN Wetlands and Water
Resources Programme, Gland, Switzerland, p. 25.
Wood, W.W. (1999) 'Use and misuse of the chloride-mass balance method in estimating
ground water recharge', Ground water, 37(1), pp. 2-5.
Woodroffe, A. (1988) 'Summary of weather pattern development of the storm of 15/16

October 1987', Meteorological Magazine, 117(1389), pp. 99-103.
World Bank (2003) Reaching the Rural Poor : A Renewed Strategy for Rural
Development. Washington DC, USA. [Online]. Available at:
https://openknowledge.worldbank.org/handle/10986/14084.
World Bank (2015) Ethiopia poverty assessment 2014. Washington DC, USA.
World Bank (2017a) Sustainable Land Management Project (SP133133). Available at:
http://projects.worldbank.org/P133133?lang=en (Accessed: 10 December 2017).
275
World Bank (2017b) World Bank Open Data; Free and open access to global
development data Available at: http://data.worldbank.org/ (Accessed: 9 January
2017).
Wright, E.P. and Burgess, W.G. (1992) 'The hydrogeology of crystalline basement
aquifers in Africa', London Special Publication, 66, pp. 1-27.
Wu, C.L. and Chau, K.-W. (2010) 'Data-driven models for monthly streamflow time
series prediction', Engineering Applications of Artificial Intelligence, 23(8), pp.
1350-1367.
Xu, Y. and Beekman, H.E. (2003) Groundwater recharge estimation in Southern Africa.
UNESCO IHP Series. Paris, France.
Xu, Y. and van Tonder, G.J. (2001) 'Estimation of recharge using a revised CRD method',
Water SA, 27(3), pp. 341-343.
Yates, D.N. and Strzepek, K.M. (1998) 'Modeling the Nile Basin under climatic change',
Journal of Hydrologic Engineering, 3(2), pp. 98-108.
Yeshaneh, E., Wagner, W., Exner-Kittridge, M., Legesse, D. and Blöschl, G. (2013)
'Identifying land use/cover dynamics in the Koga Catchment, Ethiopia, from multi-
scale data, and implications for environmental change', ISPRS International
Journal of Geo-Information, 2(2), pp. 302-323.
Yihdego, S. (2003) Hydrogeology of Illala-Aynalem catchments with particular reference

to the chemical variation and aquifer characterization. MSc thesis, Department of
Earth Sciences, Addis Ababa University, Ethiopia.
Zavala, L.M., González, F.A. and Jordán, A. (2009) 'Intensity and persistence of water
repellency in relation to vegetation types and soil parameters in Mediterranean SW
Spain', Geoderma, 152(3), pp. 361-374.
Zegeye, H. (2010) 'Environmental and socioeconomic implications of Eucalyptus in

Ethiopia', Eucalyptus Species Management, History, Status and Trends in Ethiopia.
Addis Ababa: ETH-CANA publishing company, pp. 184-205.
276
Zeleke, G. and Hurni, H. (2001) 'Implications of land use and land cover dynamics for
mountain resource degradation in the Northwestern Ethiopian highlands', Mountain
research and development, 21(2), pp. 184-191.
Zemadim, B., McCartney, M., Langan, S. and Sharma, B. (2014) 'A participatory
approach for hydrometeorological monitoring in the Blue Nile river basin of
Ethiopia', International Water Management Institute (IWMI), IWMI Research
Report 155, Colombo, Sri Lanka, 32pp.
Zeru (2008) Evaluation of groundwater potential using modeling (MODFLOW). MSc

thesis, Mekelle University, Ethiopia.
Zewdie, G. (2010) Baseflow analysis of Rivers in Lake Tana sub basin. MSc thesis,
Zhang, L., Walker, G.R. and Dawes, W. (1999) 'Predicting the effect of vegetation
changes on catchment average water balance', Cooperative Research Centre for
Catchment Hydrology, CSIRO Land and Water, Canberra, Australia.
Zhang, R., Santos, C.A.G., Moreira, M., Freire, P.K.M.M. and Corte-Real, J. (2013)
'Automatic calibration of the SHETRAN hydrological modelling system using
MSCE', Water resources management, 27(11), pp. 4053-4068.
277
Appendix A: Properties of shallow thin regolith aquifers in
sub-Saharan Africa: a case study from northwest Ethiopia
The following section is taken from a peer-reviewed conference paper submitted to and
orally presented at the 7th RWSN Forum "Water for Everyone" in Abidjan, Côte d’Ivoire,
in November/December 2016.
Citation: Walker, D. (2016) 'Properties of shallow thin regolith aquifers in sub-Saharan

Africa: a case study from northwest Ethiopia', 7th RWSN Forum "Water for
Everyone". Abidjan, Côte d’Ivoire, Nov-Dec 2016. Available at:
166_walker_david-rev1.pdf Retrieved 14 March 2017.
Note that the references, as with all the appendices, are included within the main
references list.
A-1 Introduction
It is well discussed that the hydrogeology of sub-Saharan Africa is poorly understood,
particularly regarding shallow groundwater resources (Robins et al., 2006; Calow et al.,
2009b; MacDonald et al., 2009), even though such resources sustain the majority of the
continent’s population (Lapworth et al., 2013). A knowledge of aquifer properties allows
for calculations and models to assess groundwater recharge, abstraction potential,
contamination risk, impacts of future climate variability, and management strategies.
However, few data are available on shallow aquifer properties for this region (Bonsor et
al., 2014).
The most useful shallow aquifer properties are; hydraulic conductivity (K) – the ease by
which water moves through an aquifer, specific yield (Sy) – the volume of water that
drains from the aquifer per unit surface area of aquifer per unit decline of the water table
(the drainable porosity), and well yield – the rate at which water can be abstracted from
a well.
As part of an ongoing project assessing the vulnerability of shallow aquifers in sub-

Saharan Africa, pumping tests were first conducted during a field visit to Ethiopia in
March/April 2015, timed to coincide with the end of the dry season and period of greatest
water scarcity. Further testing was undertaken during a second field visit in
278
October/November 2015 at the end of the wet season. In order to estimate hydraulic
conductivity, specific yield and well yield, tests were conducted on hand-dug wells in two
locations within the Amhara region (Figure A-1). Both drawdown (the drop in water level
within the well caused by pumping) and recovery (the increase in water level back to pre-
test level following cessation of pumping) were monitored then analysed using alternative
methods.
A-2 Testing Locations

Seven wells were tested within Dangila woreda (Figure A-1), benefiting from and
enhancing relationships established by a community-based monitoring programme which
has been ongoing since March 2014 (Walker et al., 2016). A further well was tested in
Robit-Bata kebele, 80 km northeast of Dangila woreda, close to the city of Bahir Dar.
All of the wells tested were located within weathered basalt regolith above variously
massive, vesicular and/or fractured basalt. Tested wells ranged in depth from 3.55 to
10.09 metres below ground level (mbgl) with (often irregular) diameters of around a
metre (+0.2 m). Wells are excavated by hand with picks and shovels until the solid
geology is hit. Therefore, water column height is considered the saturated thickness of
the aquifer and in the tested wells ranges from 0.54 to 3.85 m in the dry season and 1.99
to 6.34 m in the wet season. These ranges of well geometries and water depths are typical
of hand-dug wells in the areas. Wells were selected to cover a range of topographies from
floodplains, to valley slopes, to higher elevations, in an area of moderate relief within the
Ethiopian Highlands.
279
Figure A-1. Geographical distribution of testing locations.
A-3 Methodology
Motor pumps were not available, therefore, water was removed from wells using manual
methods. Water-lifting incorporated a rope and bucket; the bucket being a modified
HDPE water container. At least two people were involved in water-lifting (Figure A-2)
which helped to maintain a largely constant discharge rate.
A pressure transducer measuring every two seconds was placed in the well prior to
starting a test and water levels were also manually measured using a dip-meter. The
volume of the emptying water container was measured in addition to well diameter and
depth. The number of buckets abstracted per minute was monitored to calculate the
pumping rate. The pumping rate varied between wells from 2 litres/minute to 15
litres/minute (with one exception) dependent on the size of the container and the depth to
the water table; the smallest container and deepest water table giving the slowest pumping
rate. The test in Robit-Bata kebele had a pumping rate of 30 litres/minute being the only
tested well with a pulley and double bucket system, which leads to much more efficient
water lifting.
Water was abstracted until the well water column was reduced by at least 10%. The
necessary time to remove the entire well volume or to reach steady-state conditions with
280
the equipment available would have been excessive and extremely labour-intensive. More
importantly, given that the first field visit took place during the period of greatest water
scarcity, it would have been unethical to attempt to reduce the water level in the wells to
near empty. In order not to waste water, all containers that each household possessed were
filled during pumping tests and further water was used for backyard irrigation and for
watering livestock. The recovery of the water level was monitored with no additional
abstraction.
Figure A-2. Photographs of pumping tests.
A-4 Analysis
Selecting analysis methods was not straightforward. The shallow aquifer here may be
unconfined, however, a low-permeability though leaky clay-rich layer is commonly
observed in weathered igneous regolith profiles above more permeable material which
hosts the aquifer (Taylor and Eggleton, 2001; Sharp, 2014). In addition, it is suspected
that fractures in the underlying solid geology are influential. There is further uncertainty
over whether the wells are fully-penetrating; wells are generally excavated until solid
geology is hit though they may be partially penetrating if a boulder was struck (such
boulders are commonly observed in regolith stream bank sections) or where water tables
are shallow (i.e. on floodplains).
The Moench (1985) method was selected because it considers leaky aquifers and large-
diameter wells, i.e. well bore storage is included, and is straightforward to use on
AquiferWin32 software. The method requires: well geometry, aquifer thickness, pumping
rate, and a time-series of drawdown. Given that the period of pumping was quite short
and did not reach steady state, only a small portion of curve was available for matching
to provide values for hydraulic conductivity and specific yield. Therefore, there is a
281
potential error on the results though it is likely to be less than the natural variation of the
aquifer material. However, the specific yield values resulting from this method were often
considered impossibly low (< ~1 x 10-6), as they are computed from early time data which
is considered in pumping test analysis to be the least reliable, and have not been included
when calculating averages (Table A-1).
Recovery data was analysed using nomograms presented by Barker and Herbert (1989)
to facilitate application of the solution of Papadopulos and Cooper (1967) to recovery
tests on large-diameter wells. The method requires: well geometry, pumping rate and
period, drawdown at the end of pumping, and time taken for 25%, 50% or 75% recovery,
and provides values for transmissivity (T) (the rate at which water flows horizontally
through an aquifer; T = K multiplied by aquifer saturated thickness) and specific yield.
Values for hydraulic conductivity were derived by dividing the transmissivity by the
measured saturated thickness.
Potential well yield is considered the maximum continuous abstraction, i.e. pumped to
steady-state, that a well could be subjected to without drying out. In this case, “drying
out” is considered to be a water depth of 0.3 m which is the minimum depth from which
water could be abstracted using bucket and rope methods or without excessive sediment
intake in a motor pump. Well yield (Q) was calculated with the application of the Thiem
(1906) equation:
𝑲(𝑯𝟐 − 𝒉𝟐 ) (A-1)
𝑸=
𝑪 𝐥𝐨𝐠(𝑹⁄𝒓)
Hydraulic conductivity (K) was taken from the pumping test analyses, saturated thickness
(H) and well radius (r) were as measured prior to testing, water depth (h) was fixed at
0.3 m, C is a constant equal to 0.733, and the radius of the cone of depression (R) was
varied from 5-20 m.
A quick yield calculation was also conducted simply by dividing the recovered well
volume by the time taken for recovery and multiplying by the abstractable saturated
thickness.
A-5 Results
Given the uncertainties over well and aquifer geometry, hydraulic conductivity and
specific yield results are sufficiently similar from the drawdown and recovery analyses to
indicate suitability of methods (Table A-1). The largest differences in properties between
testing methods are within the natural variation of the aquifer materials.
282
Table A-1. Aquifer properties determined by pumping tests; K = hydraulic conductivity and Sy = specific
yield. The Sy results in italics are considered unreliable (see text) and were not used to calculate averages.
Note the quick yield estimate is calculated during recovery, as described in the text. NR = no result. The
test labelled “unable to analyse” (Well dw32) resulted in a drawdown time series that could not be
explained or analysed (see plot in Sub-Appendix A-1). Note that an uppercase well ID, e.g. DW73,
denotes a well originally surveyed by Demis Alamirew whereas lower case, e.g. dw6, indicates a well
surveyed only by David Walker.
Well yield (l/s)

Well ID Test length (mins) K (m/d) Sy quick Thiem
End of the dry season

Dangila
Drawdown DW73 (MW2) 16 1.3 0.12 0.02
dw6 21 3.1 0.08 0.02
DW77 (MW4) 32 0.7 0.03 0.12
DW61 32 5.3 0.11 0.01
dw7 48 1.4 NR 0.01
Recovery DW73 (MW2) 76 6.4 0.32 0.10 0.11
dw6 61 3.6 0.13 0.09 0.03
DW77 (MW4) 133 1.5 1x 10-5 0.22 0.26
DW61 56 2.0 NR 0.05 0.01
dw7 114 0.2 0.06 0.05 0.001
Robit-Bata
Drawdown RB 13 1.8 0.03 0.20
Recovery RB 138 0.4 NR 0.26 0.04
End of the wet season

Dangila
Drawdown DW73 (MW2) 27 3.7 3.6 x 10-6 0.29
dw6 18 6.1 3.2 x 10-6 0.27
DW77 (MW4) 14 2.8 7.8 x 10-7 1.5
dw32 29 Unable to analyse
dw33 12 6.8 2.4 x 10-6 0.80
Recovery DW73 (MW2) 41 22.3 0.05 1.0 1.8
dw6 65 19.0 0.1 0.31 0.76
DW77 (MW4) 30 6.2 0.07 2.5 2.8
dw32 46 Unable to analyse
dw33 82 10.3 0.1 0.50 1.2
Significantly, the result from Robit-Bata is consistent with those from Dangila confirming
field observations of the similarity of the regolith of both areas. This outcome suggests
that conclusions reached on the hydrogeology may be transferrable to other shallow
aquifers above basalt bedrock throughout Ethiopia. Continuing research is required to
283
determine if findings on the shallow aquifer in this region are potentially transferrable
across a wider area as studies have shown that regolith has hydrogeologically similar
characteristics across a variety of rock types (Jones, 1985).
The mean dry season hydraulic conductivity values derived from all wells using analysis
of both drawdown and recovery data is 2.3 m/d with a median of 1.6 m/d, a range of 0.2
to 6.4 m/d and a standard deviation of 1.95. The mean wet season hydraulic conductivity
is 9.7 m/d with a median of 6.5 m/d, a range of 2.8 to 22.3 m/d and a standard deviation
of 7.19. This disparity between seasons is not only of transmissivity but hydraulic
conductivity, therefore, is not explained by greater saturated thickness. Layers of higher
hydraulic conductivity must exist within the higher water column during the wet season.
The implication of this finding is significant; not only would wells excavated more deeply
below the water table have higher well bore storage, but they are more likely to intercept
more transmissive (water-bearing) layers providing greater yield, unless such layers only
exist at shallower depths. Estimates of well yield are generally >1 l/s in the wet season
when the water column is high though this may drop an order of magnitude during the
dry season.
The hydraulic conductivity results are comparable to studies of regolith elsewhere in

Africa: Olaniyan et al. (2010) report a range of 0.30 to 9.36 m/d (average: 2.13 m/d) from
a study in Nigeria, Taylor and Howard (1998) report a range of 0.3 to 3.0 m/d in Uganda,
while 0.05 to 1.5 m/d is reported by Chilton and Smith-Carington (1984) in Malawi. A
textbook range for weathered igneous regolith presented by Taylor and Eggleton (2001)
is 0.09 to 1.7 m/d.
It is well reported that there are few data on specific yield of regolith, or indeed any,
aquifers in Africa (MacDonald et al., 2012a). From all seasons and locations the specific
yield range of 0.00001 to 0.32 and mean of 0.09 (median of 0.08 and standard deviation
of 0.079) is similar to the wide range quoted by Jones (1985) of 0.00001 to 0.1 for Central
Africa and higher than the 0.003 reported by Taylor et al. (2010). Bahir Dar University
laboratory assessment of density, porosity and field capacity of five bulk samples enabled
estimation of a specific yield range of 0.052 to 0.219 for weathered basalt regolith from
Robit-Bata kebele (D. L. Yilak, personal communication, March 2015). A textbook range
for specific yield of regolith presented by Fetter (2001) is 0.15 to 0.3.
284
A-6 Conclusions
Pumping tests conducted on hand-dug wells in northwest Ethiopia provide mean
hydraulic conductivity values of 2.3 m/d in the dry season and 9.7 m/d in the wet season
(median = 1.6 and 6.5 m/d), and a mean specific yield value of 0.09 (median = 0.08).
These values contribute to the extremely sparse data available in published literature for
shallow regolith aquifers in sub-Saharan Africa. Calculations of well yield
(average = 0.5 l/s) indicate that penetrating a substantial saturated thickness of aquifer
(>3 m below water table) to maximise water column height is as important for achieving
desirable yield as locating areas of high hydraulic conductivity. A well or borehole fitted
with a handpump must be able to sustain a supply of >0.1 l/s (preferably 0.3 l/s) to supply
a community (MacDonald et al., 2012a). Irrigation demand depends on crop type and
local environmental conditions, though these are less significant when considering
general feasibility. For the range of crops and conditions likely to be encountered at the
study site, daily water use can be calculated as approximately 1 l/s/ha (Brouwer et al.,
1992). Given this calculated irrigation requirement, the well yield estimations give some
optimism that small scale irrigation, in addition to the existing community supply, is
achievable from hand-dug wells in shallow regolith aquifers. Further research is required
to determine the transferability of findings, though similarities in results from wells some
distance apart and with published results suggest the findings may be transferable to other
areas of shallow weathered regolith aquifers across sub-Saharan Africa and certainly to
shallow weathered basalt regolith aquifers within Ethiopia. Knowledge of aquifer
parameters is vital in constructing models for simulation of climate change impacts and
in developing management strategies for sustainable development of shallow
groundwater resources.
285
Sub-Appendix A1 – Plotted pumping test data
Well: DW73 (MW2) - dry season

0
0.05
0.1
0.15
Drawdown (m)
0.2
0.25
0.3
0.35
0.4
Dip meter
0.45
Pressure transducer
0.5
0 1000 2000 3000 4000 5000 6000
Elapsed time (s)
Well: dw6 - dry season

0
0.05
0.1
0.15
Drawdown (m)
0.2
0.25
0.3
0.35
0.4
Dip meter
0.45
Pressure transducer
0.5
0 1000 2000 3000 4000 5000 6000
Elapsed time (s)
286
Well: DW77 (MW4) - dry season
0
0.1
0.2
0.3
Drawdown (m)
0.4
0.5
0.6
0.7 Dip meter

Pressure transducer
0.8
0 2000 4000 6000 8000 10000 12000
Elapsed time (s)
Well: DW61 - dry season

0
0.1
0.2
Drawdown (m)
0.3
0.4
0.5
Dip meter
Pressure transducer
0.6
0 1000 2000 3000 4000 5000 6000
Elapsed time (s)
287
Well: dw7 - dry season
0
0.1
0.2
Drawdown (m)
0.3
0.4
0.5
Dip meter
Pressure transducer
0.6
0 2000 4000 6000 8000 10000 12000
Elapsed time (s)
Well: RB - dry season

0
0.1
0.2
Drawdown (m)
0.3
0.4
0.5
Dip meter
Pressure transducer
0.6
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Elapsed time (s)
288
Well: DW73 (MW2) - wet season
0
0.02
0.04
Drawdown (m)
0.06
0.08
0.1
0.12
Dip meter
0.14 Pressure transducer
0 500 1000 1500 2000 2500 3000 3500 4000 4500

Elapsed time (s)
Well: dw6 - wet season

0
0.05
Drawdown (m)
0.1
0.15
0.2
Dip meter
Pressure transducer
0.25
0 1000 2000 3000 4000 5000 6000
Elapsed time (s)
289
Well: DW77 (MW4) - wet season
0
0.05
Drawdown (m)
0.1
0.15
0.2
Dip meter
Pressure transducer
0.25
0 500 1000 1500 2000 2500 3000
Elapsed time (s)

0
Unable to analyse
0.5
1
Drawdown (m)
1.5
2.5
3 Dip meter
Pressure transducer
3.5
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Elapsed time (s)
290
0
0.02
0.04
0.06
Drawdown (m)
0.08
0.1
0.12
0.14
0.16
Dip meter
0.18
Pressure transducer
0.2
0 1000 2000 3000 4000 5000 6000
Elapsed time (s)
291
Appendix B. Hydrochemistry field investigations at Dangila
woreda
B-1 Introduction
Groundwater sampling and in-situ testing was initially undertaken during the first field
visit in March/April 2015 at the end of the dry season. All samples and tests were taken
from traditional hand-dug wells (HDWs) or springs. Samples were analysed principally
at laboratories in Ethiopia with some samples tested in the UK to assess consistency and
accuracy of the results. Laboratory analysis consisted of major ion and stable isotope
testing. In-situ testing consisted of measurement of field pH, temperature and electrical
conductivity.
Much of the sampling and testing was repeated during the second field visit in
October/November 2015 at the end of the wet season. In addition, the sampling and
testing regime included deep boreholes, surface water, rainwater, and radon-222
measurements.
B-2 Purpose of sampling and in-situ testing

The aims of the sampling and in-situ testing programme were:
 Identification of water types: (i) higher yielding wells may have a particular
chemical signature; (ii) there may be areas where groundwater is chemically
unsuitable for certain uses.
 Assessment of aquifer connectivity: (i) if nearby wells and springs have
contrasting hydrochemistry then it may indicate low connectivity, (ii) similar
hydrochemistry at distant locations may indicate preferential groundwater flow
paths.
 Groundwater aging: (i) very low ionic concentrations would indicate recently
recharged groundwater; (ii) it may be possible to estimate groundwater residence
times to indicate areas of greatest storage and therefore the least vulnerability to
climate variability.
 Recharge mechanism: (i) estimations on the degree of evaporation prior to
recharge could indicate if recharge occurs during intense storm events or is more
diffuse, (ii) the role of floodplains to act as recharge basins could be evaluated,
(iii) it would suggest whether recharge occurs locally or ubiquitously.
292
 Consistency of groundwater: (i) comparisons with testing results from previous
studies will suggest whether the samples are representative, (ii) such comparisons
will also show whether the groundwater is changing over time.
 Identification of losing and gaining reaches of surface water to aid conceptual
model development.
B-3 Sampling locations

Sampling locations were identified prior to the field visit utilising data collected by Demis
Alamirew of GSE during fieldwork in February/March 2014 as part of the AMGRAF
catalyst period. Sampling locations were selected to give a distribution of groundwater
analyses from various catchments, topographic positions, altitudes, kebeles, positions in
the catchment; e.g. close to the watershed or close to rivers, geologies; e.g. from fractured
or vesicular basalt with or without regolith or alluvium, seasonal and perennial sources,
springs and wells, shallow and deeper water tables, zones of various groundwater
potential indicated by SHETRAN modelling of low-flows by Dr Geoff Parkin during the
AMGRAF catalyst period, and zones of various groundwater potential indicated by
Demis Alamirew according to geological formation and field investigations.
It was not possible to keep entirely to the sampling location plan during the field visit due
to occasional springs or wells being dry and a number of pumps being broken.
Furthermore, the selection of sampling points according to geology, particularly
regarding bedrock, was generally unfeasible due to outcrops being restricted to riverbeds
and banks. Superficial geology was occasionally easier to identify from well excavation
arisings or by peering into the well shaft. However, the generally ubiquitous weathered
basalt regolith or clayey sandy floodplain materials above variously massive, vesicular
and/or fractured basalt meant sampling according to precise geology was overly-
ambitious.
The geographical distribution of the sampling locations is presented in Figure B-1. During
the second field visit, ten locations were resampled; more resamples were proposed but
non-functioning wells made this impossible. New sampling sites included transects of
surface and groundwater to test hypotheses from the conceptual model development. The
ease of in-situ testing meant that measurements were taken at many other sites in addition
to the sampling locations; wherever it was thought results could provide useful
information.
293
Figure B-1. Geographical distribution of sampling locations (red triangles) across Dangila woreda (black
outline). The background is a DEM with green being the lowest elevation and blue the highest elevation.
B-4 Sampling methodology

Sampling was conducted in accordance with accepted international standards and
guidelines (Gov. WA, 2009; CCME, 2011; IAEA, -). A polypropylene syringe was rinsed
several times in sample water prior to filling for sample collection. Sample water was
collected directly from the hand pump or rope-and-washer pump outlet, from a collection
bucket on a rope (typically a doctored 10-litre HDPE jerry can) or the point of emergence
of a spring (Figure B-2). New nalgene bottles were used for sampling, which were
brought from the UK. Major ion samples were filtered through 0.2 μm Supor® Membrane
into 125 ml bottles. Two 125 ml samples were collected and filtered at each sampling
location, one for cation analysis and one for anions; the sample for cation analysis having
the addition of three to four drops of nitric acid preservative. Stable isotope samples were
collected in 60 ml bottles with no filtration or preservative. Care was taken to keep the
bottles clean and avoid contamination during sampling. In addition to tightly capping,
samples had their caps sealed further with electrical insulation tape to restrict the
possibility of evaporation from the bottles.
294
Figure B-2. Receiving local assistance sampling and in-situ testing from a developed spring (left) and a
rope and washer pump.
Wells were not purged to commonly proposed standards prior to sampling. Pumping tests
showed that the hydraulic conductivity of the shallow aquifer is quite low and water level
recovery in the wells was slow following abstraction. Given that the first field visit took
place during the period of highest water scarcity it would have been unethical to attempt
to purge several (typically three) well volumes prior to sampling. What’s more, HDWs
in the area have diameters of around a metre and consequently well bore storage is high.
The necessary time to remove several well volumes with the equipment available would
have been excessive and extremely labour-intensive. However, sampling preferentially
took place from wells that experienced heavy use and/or following use in order for the
collection of water by the local community to mimic controlled purging. In addition,
sampling preferentially took place from sealed sources, which were not open to
evaporation, such as hand pumps, rope-and-washer pumps, and springs at the point of
emergence. To restrict the sampling of groundwater that had potentially undergone
evaporation, samples from open wells were collected following abstraction of water
during pumping tests, which in effect was a low volume purge of the well.
On each sample bottle the sample number, site name, sampling date and required analysis
was marked with indelible marker. For each sample location the following information
was recorded: sample number, woreda, kebele, site name, GPS coordinates, GPS altitude,
sampling date, geology, field pH, temperature, electrical conductivity, well depth, depth
to water, pump or cover type, flow estimate (if a spring), adjacent land use, immediate
topography, wider topography, mode of emergence (if a spring), use (e.g. domestic,
irrigation, etc.), seasonality, and any other information that could be pertinent, such as the
history of the feature, either observational or through conversation with local users.
295
Samples were stored in a cool dark place until transportation to the laboratory; the average
storage period being eleven days for the first field visit and three days for the second.
B-5 Major ion hydrochemistry
B-5-1 Analysis
Major ion analysis took place at ADSWE (Amhara Design & Supervision Works
Enterprise) Laboratory in Bahir Dar, Ethiopia. ADSWE is a modern and professional soils
and water testing laboratory currently working towards international accreditation. Anion
analysis was undertaken by a Palintest 2700 photometer utilising Dionex ion
chromatography, with the exception of bicarbonate and carbonate that was analysed by
titration. A Nova 300 Series utilising atomic absorption spectroscopy undertook cation
analysis. Analysis equipment was calibrated in accordance with manufacturer’s
instructions.
Major ion analysis consisted of testing for Ca2+, Mg2+, Na+, K+, HCO3-, Cl-, SO42- and
NO3-. In addition, some samples were analysed for CO32-, F- and Fe2+. Analysis for Fe2+
was deemed pertinent due to common red staining of filters during sampling. Samples
were analysed for F- because Ethiopia is known for having problems with fluorosis caused
by groundwater, though excess F- in groundwater is generally found in the Rift Valley
and from deep wells (Kloos and Haimanot, 1999; Tekle-Haimanot et al., 2006; Kebede,
2013).
B-5-2 Quality assurance

A sample was collected in triplicate with one sample submitted blind to the laboratory
and the third brought to the UK for analysis at Newcastle University. The average
difference between measured concentrations of the samples tested by ADSWE was
16.6%. The average difference between measured concentrations of ADSWE vs
Newcastle University tested samples was 26.2% for anions and 71.1% for cations. The
discrepancies between the triplicate samples appear high; particularly in the case of
cations tested at Newcastle University, which gave significantly higher concentrations
than for all samples analysed in Ethiopia. This sample also showed a high positive ionic
balance error of 15.9%. The cause of the discrepancies was investigated and several
possibilities can be ruled out:
 Contact with the laboratories indicated that all equipment had been recently
calibrated.
296
 A review of manufacturers’ datasheets and discussion with equipment operators
revealed that the discrepancies are in excess of the tolerance of the analysis
equipment.
 Following discussion with operators of such equipment at Newcastle University,
the natural variations in hydrochemistry within a sample could not account for
such discrepancies.
 The sample that appeared to contain excessive cations was retested at Newcastle
University to determine if an erroneous result had been obtained. Almost identical
concentrations were measured.
It is therefore suggested that the sample with excessive cations had become contaminated.
It is worth noting that, given the very low concentrations of all major ions, reduction of
just 5 mg/l of calcium, magnesium and potassium would bring the ionic balance error
within ±5%. However, the cation concentrations would still be significantly higher than
those measured in Ethiopia. A sample collected in duplicate during the second field visit
and analysed at ADSWE and at Newcastle University showed an average difference
between measured concentrations of just 1.3% further suggesting that the triplicate
sample from the first visit had become contaminated.
Regarding the first field visit, ionic balance calculations in only 35% of the samples were
within ±5% and 82% of samples were within ±10% with an average ionic balance error
of 6.6%. From the second field visit, just 12.5% of samples were within ±5% while only
43.8% of samples were within ±10% and the average ionic balance error is 12.9%. Three
possibilities are identified which could (singularly or together) be causing the high ionic
balance errors:
1) Major ion concentrations within the groundwater samples are low, often at trace
level. Therefore, trace elements, which are usually unimportant in calculating
electroneutrality, are having an impact. For the first field visit, 73% of the ionic
balance errors greater than ±5% are in the negative suggesting there are
unanalysed cations affecting the ionic balance. To substantiate this claim, it was
intended that subsequent testing regimes would include trace metal analysis such
as aluminium and silica (considering the mineralogy of the shallow aquifer).
Unfortunately, when the samples from the second visit were delivered to the
ADSWE laboratory they informed us that such analysis could not be undertaken
at that time due to a shortage of necessary equipment consumables. However,
samples from the second visit in all but one case show high positive ionic
297
imbalances and therefore an excess of cations perhaps indicating the presence of
unanalysed anions. This suggestion of unanalysed ions influencing
electroneutrality is supported by the higher TDS of samples with greatest ionic
balance error though analysed concentrations are not significantly different.
2) Because the major ion concentrations are low, small errors in concentration
measurement (due to the equipment, the operator, or minor contamination) would
be amplified when calculating percentage errors.
3) There is of course the third possibility that the major ion analysis is unreliable and
significant conclusions should not be drawn from the groundwater chemistry
analysis.
B-5-3 Results
The results of the major ion analysis are presented in Tbale B-1 and Table B-2.
Considering the first field visit, there was not great variation among the shallow
groundwater analyses. In trying to assign water types there are three samples with very
low EC and dissolved ion concentrations (TDS all <25 mg/l) that could be considered the
youngest, i.e. most recently recharged, groundwater. However, perhaps unexpectedly
given the proposed short groundwater residence time, SI/A/C 1 and 3 are from sources
with very good perennial supply. An older water type could be assigned to the
groundwater with higher concentrations of HCO3- and Ca2+ and highest EC (TDS all <150
mg/l). Such groundwater (SI/A/C 7, 8, 9, 11 and 18) has a longer residence time though
again, similar to the youngest water type, the sampling locations have few similarities.
There are not significant enough differences to assign water types or identify a
hydrochemical signature of the highest yielding sources. Although there is the possibility
that all of the samples represent the water type for good perennial water sources because
the seasonal wells were dry in March/April and couldn’t be sampled.
Considering the second field visit, there is not great variation among and between the
shallow groundwater and surface water analyses though the deep groundwater has a
different signature. The deep groundwater has low calcium and high sodium suggesting
ion exchange. pH is high at 8.8 (possibly due to release of CO2 from the water). The
groundwater is likely to be old – as would be expected from boreholes >100 m deep – as
it has high EC (>300 µS/cm), high bicarbonate and has taken time for Ca-Na exchange.
Nitrate and sulphate are both low due to little human input and reducing waters. It may
have been expected to see bicarbonate to sulphate exchange and this may have occurred
followed by sulphate reduction, though there is no evidence of sulphide (not analysed but
298
no smell). However, sulphide could have precipitated out as FeS2, which is why iron
concentration is lower than may be expected in a (suspected) low to zero dissolved
oxygen water, though this sulphate reduction would have reduced the pH. Generally, the
EC and ionic concentrations are lower than the first field visit, which indicates that at the
end of the wet season the analysed groundwater had been recently recharged. However,
repeat tests show very similar chemistries suggesting a longer (months rather than
days/weeks) residence time. Surface water samples show similar chemistries to shallow
groundwater samples though with EC at the lower end (around 100 µS/cm) and pH at the
upper end (around pH 6.5).
Table B-1. Results of in-situ testing and laboratory analysis from first field visit in March/April 2015. All
the samples are shallow groundwater.
Calculated Ionic Laboratory analysis

Sample Location In-situ field measurement Laboratory analysis (mg/l)
from EC balance (%o VSMOW)
number ID o 2+ 2+ + + 2+ - -
Temp. ( C) pH* EC (μS) TDS (mg/l) Ca Mg Na K Fe Cl F SO42- NO3- HCO3- CO32- error δ18O δ2H
SI/A/C 1 DW43 20.9 5.19 50.62 25 8.83 1.07 0.08 2.65 0.1 1.5 0.31 2.8 2.7 25.8 0 1.70% -2.14 -0.17
SI/A/C 2 dw4 20.8 5.84 99.7 50 10.8 1.44 2.4 7.5 2.93 3.2 0.5 8.2 1.54 35.4 0 8.52% -1.81 0.23
SI/A/C 3 CS42 22.1 5.31 48.67 24 14.1 2.42 0.08 5.13 0.01 0.8 0.8 2.4 2.34 50.2 0 3.11% -1.56 -0.58
SI/A/C 4 DW56 22.9 5.53 171.9 87 12.5 2.28 0.08 0.15 0.01 2.6 0.21 1.5 4.85 48.8 0 -9.63% -1.24 3.64
SI/A/C 5 DW73 20.3 5.83 130.7 66 14.6 2.37 1.73 0.42 0.2 0.5 0.22 1.1 4.37 52.0 0 2.26% -1.55 1.98
SI/A/C 6 dw6 22.3 5.57 144.2 73 15.5 2.23 0.08 0.83 0.03 4.9 0.3 1.2 1.98 40.2 0 6.10% -2.54 -3.58
SI/A/C 7 DW79 25.2 6.88 334.9 172 19.4 3.4 0.08 1.74 0.01 3.7 0.31 1.1 2.9 85.2 0 -10.08% -1.91 no result
SI/A/C 8 DW79 25.2 6.88 334.9 172 17.5 3.07 0.08 3.43 0.01 4.1 0.41 2.3 3.05 80.3 0 -12.03% -1.36 1.67
SI/A/C 9 DW79 25.2 6.88 334.9 172 41.9 16.7 9.0 0.99 0.002 5.597 0.116 1.167 2.822 15.89% no sample
SI/A/C 10 CS12 22 5.98 217.6 111 21.7 4.24 5.03 0.04 0.01 1.2 0.35 1.5 4.95 95.1 0 -2.08% -1.48 0.27
SI/A/C 11 DW18 25.9 6.66 481.9 249 19.8 3.61 17.55 3.21 0.01 2.1 0.43 2.1 1.99 152.2 0 -10.90% no result 0.59
SI/A/C 12 DW2 21.9 5.76 200.4 102 18.4 3.44 0.08 2.13 0.01 1.3 0.32 0.84 3.2 90.4 0 -12.03% -1.83 -2.12
SI/A/C 13 dw10 22.1 5.69 44.29 22 10.1 1.78 2.87 1.03 0.01 1.4 0.33 0.92 2.65 38.5 0 3.37% -1.95 -2.74
SI/A/C 14 DW21 24.7 6.17 174.0 88 17.4 2.81 4.6 1.54 0.01 1.3 0.4 6.2 1.48 60.2 0 5.60% -1.85 0.76
SI/A/C 15 DW22 24 6.31 264.4 135 18.8 2.94 2.54 2.29 1.65 0.7 0.56 1.6 1.22 97.3 0 -9.30% -2.86 -5.04
SI/A/C 16 dw15 23.4 5.69 196.8 100 18.7 3.12 0.1 1.14 4.3 1.7 0.28 1.4 2.37 86.1 0 -5.63% no result no result
SI/A/C 17 cs6 22.2 6.09 189.6 97 23.3 3.87 0.8 1.24 0.01 1.2 0.37 1.1 2.5 91.2 0 -2.00% -1.97 -1.45
SI/A/C 18 dw30 21.9 5.99 309.4 159 27.1 3.18 0.19 9.06 0.7 2.7 0.25 1.2 4.12 125.4 0 -8.67% -2.16 -0.02
* The field pH meter was suspected to have been reading 0.25-0.5 pH too low (see text).
Triplicate samples highlighted grey (SI/A/C7, 8 and 9) with the sample tested at Newcastle University highlighted darkest.
299
Table B-2. Results of in-situ testing and laboratory analysis from second field visit in October/November
2015.
Calculated Ionic Laboratory analysis

Sample Location In-situ field measurement Laboratory analysis (mg/l)
Sample type from EC balance (%o VSMOW)
number ID o 2+ 2+ + + 2+ - -
Temp. ( C) pH* EC (μS) TDS (mg/l) Ca Mg Na K Fe Cl F SO42- NO3- HCO3- error δ18O δ2H
2SI/A/C1 DW56 Groundwater 22.9 5.61 167.7 107 15.88 6.5 8.65 1.91 2.70 1.10 10.50 51.85 22.09% 5.74 -0.47
2SI/A/C2 dw21 Groundwater 30.9 6.71 403.5 258 45.67 5.5 16.21 4.71 2.60 3.20 0.96 140.30 18.34% -0.45 -0.41
2SI/A/C3 cs6 Groundwater 22.9 6 188.70 121 19.85 5.5 13.1 0.85 2.70 5.20 3.76 54.90 27.99% 3.46 -0.01
2SI/A/C4 DTW3 Deep Groundwater 22.9 8.77 315 202 2.55 4.05 34.1 1.84 0.01 0.90 0.47 1.10 0.98 97.60 8.23% -3.98 -1.63
2SI/A/C5 dw31 Groundwater 21.8 5.55 136.4 87 14.73 5.01 10.4 1.24 2.20 8.10 3.70 51.85 17.72% 10.59 0.41
2SI/A/C6 D3 Deep Groundwater 21.8 8.81 335 214 0.97 4.51 34.5 1.86 0.01 2.50 0.64 0.50 1.44 97.60 6.23% -1.02 -1.61
2SI/A/C7 RFL39 Rain 0.380 1.912 0.799 16.96 1.58
2SI/A/C8 DW43 Groundwater 20.5 4.77 46.26 27 4.89 6.02 5.33 1.34 1.20 8.30 4.20 24.40 19.71% 9.57 0.21
2SI/A/C9 CS42 Groundwater 22.2 4.77 58.44 37 6.25 4.03 3.01 1.08 1.20 7.20 4.50 27.45 6.35% 1.46 -0.29
2SI/A/C13 dw6 Groundwater 22.2 5.41 133.8 86 12.48 5.02 1.73 0.57 4.8 7.05 4.2 24.85 19.57% 4.81 0.30
2SI/A/C14 dw18 Groundwater 1.75 -0.04
2SI/A/C15 DW77 Groundwater 22.3 5.24 146.6 94 4.23 0.31
2SI/A/C16 dw2 Groundwater 22.5 6.43 289 185 6.37 0.11
2SI/A/C17 RFL46 Wetland stream 21.8 6.83 99.35 64 10.98 1.59
2SI/A/C18 CS12 Groundwater 22.8 5.98 217.6 139 17.56 5.05 6.27 2.01 1.2 6.1 11.5 60.35 9.50% 6.99 1.03
2SI/A/C19 CS12 Groundwater 22.8 5.98 217.6 139 15.85 6.71 3.92 3.86 2.67 8.22 9.41 60.26 7.55% 6.37 0.96
2SI/A/C20 RFL48 River 23.1 6.39 106.6 68 13.49 0.92
2SI/A/C21 RFL49 Wetland 30.9 6.71 256.1 164 16.42 1.85
2SI/A/C22 RFL50 Rain 0.541 1.225 0.935 51.30 5.90
2SI/A/C23 RFL39 Rain 1.122 1.063 1.102
2SI/A/C24 RFL51 Wetland stream 20.3 6.84 83.26 53 23.14 0.35
2SI/A/C25 RFL52 Wetland stream 26.6 6.04 70.28 45 9.41 7.08 2.64 4.33 1.3 14 2.06 53.25 1.73% 23.44 0.92
2SI/A/C26 RFL53 River 23.1 6.18 91.1 58 21.97 0.64
2SI/A/C27 RFL55 River 22.3 6.27 93.45 60 23.13 1.29
2SI/A/C28 DW79 Groundwater 24.7 6.59 231.3 148 33.0 8.72 4.5 <1 0.001 2.119 1.146 6.485 86.43% 3.31 0.91
2SI/A/C29 cs9 Groundwater 23.2 5.75 124.8 79.8 2.65 0.50
2SI/A/C30 cs10 Groundwater 23.5 6.11 171.7 110 5.39 0.34
2SI/A/C31 RFL61 River 22.9 6.29 104.8 67 9.66 6.12 1.34 1.83 2.8 25.1 2.1 74.55 -26.00% 8.29 1.16
* The field pH meter is no longer thought to have been reading low (see text).
Duplicate samples highlighted grey (2SI/A/C18 and 19) with the samples tested at Newcastle University highlighted darkest.
The piper diagrams in Figure B-3 show that the surface water and shallow groundwater
belong to the “bicarbonate calcium” type typical of recent recharge. The deep
groundwater is of “bicarbonate sodium” type indicative of higher mineralisation due to
longer residence time and greater distance of flow. The shallow groundwater samples
from the wet season plot more closely to the surface water samples indicating a high
degree of interconnectivity between the surface and shallow groundwater. This was
expected in the wet season from the observed very shallow water table.
300
Deep groundwater (wet season)
CATIONS ANIONS
Ca = 1. mg/l HCO3 = 98. mg/l
Mg = 4.5 mg/l CO3 = 0. mg/l
Na = 35. mg/l Cl = 2.5 mg/l
K = 1.9 mg/l SO4 = 0.5 mg/l
F = 0.6 mg/l
100 80 60 40 20 0 0 20 40 60 80 100
Calcium(Ca) Chloride(Cl) + Fluoride(F)
Surface water (wet season)
CATIONS ANIONS
Ca = 9.7 mg/l HCO3 = 75. mg/l
Mg = 6.1 mg/l CO3 = 0. mg/l
Na = 1.3 mg/l Cl = 2.8 mg/l
K = 1.8 mg/l SO4 = 25. mg/l
F = 0. mg/l
100 80 60 40 20 0 0 20 40 60 80 100
301
Shallow groundwater (wet season)
CATIONS ANIONS
Ca = 12. mg/l HCO3 = 25. mg/l
Mg = 5. mg/l CO3 = 0. mg/l
Na = 1.7 mg/l Cl = 4.8 mg/l
K = 0.6 mg/l SO4 = 7.1 mg/l
F = 0. mg/l
100 80 60 40 20 0 0 20 40 60 80 100
Shallow groundwater (dry season)
CATIONS ANIONS
Ca = 11. mg/l HCO3 = 35. mg/l
Mg = 1.4 mg/l CO3 = 0. mg/l
Na = 2.4 mg/l Cl = 3.2 mg/l
K = 7.5 mg/l SO4 = 8.2 mg/l
F = 0.5 mg/l
100 80 60 40 20 0 0 20 40 60 80 100
Figure B-3. Piper diagrams for different sources of water analysed within Dangila woreda. Average
cation and anion concentrations are also given on the plots.
Fluorine levels within all samples, including those from the deep aquifer, were below the
WHO recommended maximum of 1.5 mg/l (WHO, 2011); the maximum measured was
302
0.8 mg/l. Pit latrines, occasionally in close proximity to wells, could elevate nitrate levels
in shallow groundwater though all samples recorded well below WHO guidelines; the
maximum measured was 11.5 mg/l. WHO does not give a guideline value for iron
concentration in drinking water though four samples were above the 0.3 mg/l at which
water discolours and staining can occur. The sodium adsorption ratios (SAR) were
extremely low: ~3.0 for the deep groundwater samples and <1.0 for all other samples.
SAR >3.0 is generally thought potentially problematic for irrigation water (Olson, 2012).
Considering only hydrochemistry and not microbial content, the analyses indicate that all
the groundwater tested is suitable for both irrigation and potable use.
B-6 Stable isotope
B-6-1 Analysis
Dr Seifu Kebede, a renowned isotope hydrogeologist, at the School of Earth Science,
Addis Ababa University, Ethiopia, conducted stable isotope analysis. Analysis was
undertaken by a LGR DLT-100 utilising laser spectroscopy. Analysis equipment was
calibrated in accordance with manufacturer’s instructions. Stable isotope analysis
consisted of testing for δ18O and δ2H.
B-6-2 Quality assurance

A sample was collected in duplicate with one sample submitted blind to the laboratory.
A result was not obtained for δ2H for one of the samples whereas for δ18O the results
differed by 28.8%. This percentage difference seems high though it is a percentage of a
low result, which actually equates to only 0.55%o VSMOW. Both results would plot in a
very similar position on Figure B-4Figure B-4. Plot of all stable isotope results. Units are
%o VSMOW. The dotted line equates to the Addis Ababa local meteoric water line
derived from data presented by Rozanski et al. (1996) and the dashed line is the derived
local evaporation line.. The blind submitted during the second field visit varies by 8.9%
(δ18O) and 6.8% (δ2H) with small absolute differences.
B-6-3 Results
Unfortunately, analysis could not be completed on all of the samples. “No Result” on
Table B-1, according to the laboratory, “means we haven’t gotten good results for those
analysis and we have discarded them”. Each sample is analysed a number of times due to
the sensitivity of the equipment and if the standard deviation of the results is above a
certain threshold the result is rejected. The stable isotope results are presented in Figure
B-4.
303
7.0
6.0
5.0
4.0
3.0
2.0
1.0
d2H
0.0
-1.0 Deep groundwater

-2.0 Shallow groundwater (dry season)
-3.0 Shallow groundwater (wet season)
-4.0 Surface water
-5.0 Rain
-6.0
-10.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0
d18O
Figure B-4. Plot of all stable isotope results. Units are %o VSMOW. The dotted line equates to the Addis
Ababa local meteoric water line derived from data presented by Rozanski et al. (1996) and the dashed
line is the derived local evaporation line.
Figure B-4 shows clear differences between water sources and notably between shallow
groundwater sampled in the dry season and from the wet season. A local evaporation line
can be drawn through the wet season shallow groundwater and surface water results at
odds with the dry season shallow groundwater results, which plot close to the local
meteoric water line. The surface water samples are indicative of evaporation causing
enrichment, which would be expected as rainfall collects in dambo wetlands before
accumulating to form streams and rivers. The wet season shallow groundwater results are
also enriched and similar to the surface water results indicating recently recharged water
that infiltrated from wetlands. The slightly lower enrichment of the shallow groundwater
than the surface water suggests mixing with diffusely recharged and thus less evaporated
and enriched water. The dry season shallow groundwater results plot close to the local
meteoric water line suggesting diffuse recharge. This would be predicted, as the wetlands
are dry outside of the wet season, therefore, there is less opportunity for evaporation at
the time of infiltration in the dry season despite the higher temperatures and lower relative
humidity. The deep groundwater was sampled during the wet season and plots well away
from other wet season groundwater samples indicating little interconnectivity. The deep
samples show the highest depletion meaning recharge could have occurred at a distant
304
high mountain area. The rainwater samples show very high enrichment. The cause of this
enrichment is uncertain. Sampling was directly from falling rain during intense storms
and at night so there was no opportunity for evaporation.
B-7 In-situ testing
B-7-1 Methodology
A handheld Myron L Company Ultrameter II was used for in-situ water testing, calibrated
immediately prior to the field visits. Testing was conducted in accordance with accepted
international standards and guidelines (Gov. WA, 2009; IAEA, -). Testing comprised
measurement of water temperature, pH and electrical conductivity. The sensor was
thoroughly rinsed with sample water and the reading allowed to stabilise prior to a
measurement being recorded.
B-7-2 Results
The results of in-situ testing from the locations where samples were collected are
presented with the laboratory results in Table B-1 and Table B-2. Results from locations
that had been tested a year previously during fieldwork by GSE are presented in Table
B-3; there are eleven locations where testing was repeated. A total of 28 locations were
subjected to testing in March/April 2015 following the 198 in February/March 2014 and
a further 40 in October/November 2015.
Table B-3. Comparison of in-situ testing results from February/March 2014, March/April 2015 and
October/November 2015.
Location Oct/Nov 2015 Mar/Apr 2015 Feb/Mar 2014 Oct/Nov 2015 Mar/Apr 2015 Feb/Mar 2014 Oct/Nov2015 Mar/Apr 2015 Feb/Mar 2014
Sample number
code pH * pH * pH EC (μS) EC (μS) EC (μS) Temp. (o C) Temp. (oC) Temp. (o C)
DW73 SI/A/C 5 & 2SI/A/C 12 5.3 5.83 6.22 110.9 130.7 132.6 20.9 20.3 21.5
DW77 2SI 15 5.24 5.28 6.04 146.6 90.74 104 22.3 22.2 26.4
DW43 SI/A/C 1 & 2SI/A/C 8 4.77 5.19 5.82 46.26 50.62 56.8 20.9 20.9 18.9
DW56 SI/A/C 4 & 2SI/A/C 1 5.61 5.53 6.04 167.7 171.9 182.2 22.9 22.9 25.4
DW61 5.95 6.33 216.6 217 22.1 22.5
DW79 SI/A/C 7 8 9 & 2SI/A/C 28 6.59 6.88 7.35 231.3 334.9 320 24.7 25.2 26.7
DW18 SI/A/C 11 6.66 7.2 481.9 502 25.9 24.2
DW2 SI/A/C 12 5.76 6.32 200.4 203 21.9 23.3
DW21 SI/A/C 14 & 2SI/A/C 11 5.19 6.17 6.54 167.1 174 142 22.1 24.7 21.7
DW22 SI/A/C 15 & 2SI/A/C 10 5.46 6.31 6.69 264.4 264.4 200 24 24 26.4
dw6 2SI/A/C 13 5.41 5.57 133.8 144.2 22.2 22.3
CS42 SI/A/C 3 & 2SI/A/C 9 4.77 5.31 6 58.44 48.67 49.7 22.2 22.1 23.5
cs6 2SI/A/C 3 6 6.09 188.7 189.6 22.9 22.2
CS12 2SI/A/C 18 19 5.98 5.98 225.3 217.6 220 22.8 22.0 22.9
cs1 5.36 5.22 66.35 61.93 20.3 22.8
* The field pH meter was suspected to have been reading 0.25-0.5 pH too low after the first field visit but now it is thought the 2014
readings were too high (see text).
Considering the first field visit, it was suspected that the handheld meter was falling out
of calibration with regard to pH. The field pH results were consistently lower than tested
previously whereas electrical conductivity results were almost identical 12-13 months
after first being recorded and water temperature measurements are similar. However, pH
305
results from the second field visit are similar to that of the first suggesting the anomalous
readings may be those taken by GSE in 2014.
In-situ testing indicates that shallow groundwater in the area generally has pH around 6
and EC 50-300 µS. These values are typical of young groundwater in igneous terrain. The
field tests gave remarkably similar EC measurements on both field visits, and to the
previous period of testing 12-13 months earlier. The consistency of results indicates that
the groundwater has consistent properties and, therefore, the samples are representative
of shallow groundwater from this location.
B-8 In-situ radon-222 testing
B-8-1 Methodology
Radon-222 is one of the radioactive decay products of uranium-238, a radioactive element
that naturally occurs in the minerals of most rocks. As an inert gas, 222Rn readily migrates,
222
through advection and diffusion, into and with groundwater. Groundwater Rn
concentration reaches steady state and declines rapidly upon discharge due to the short
half-life of 3.8 days. Analysis of the spatial distribution of 222Rn concentrations in surface
and groundwater was used to assess infiltration from surface water to aquifers by Hoehn
and Von Gunten (1989) in Switzerland and by Bertin and Bourg (1994) in southern
France, to measure rates of groundwater discharge into a river in the Northern Territory
of Australia by Cook et al. (2003) and into small lakes in Florida, USA, by (Dimova et
al., 2013), and in Central Chile by Oyarzún et al. (2014) to assess groundwater-surface
water connectivity. No previous studies were identified from Africa.
A DURRIDGE RAD7 with the “Big Bottle System” was used for in-situ measurement of
radon-222 concentration in water. The equipment was borrowed from Dr Seifu Kebede
at the University of Addis Ababa. Radon-222 testing was conducted in accordance with
the Durridge Company Inc. manuals, comprising:
1. Equipment set up.

2. Purging of the equipment for at least 10 minutes and until a maximum internal
relative humidity of 10% was achieved.
3. Careful sampling of water directly into the “big bottle” avoiding turbulent flow
and degassing.
4. Running the test for at least 45 minutes with monitoring of radon-222
concentration, internal relative humidity and water temperature.
306
5. Removal of sample and minimum of 8-minute post-test purge before equipment
disassembly.
The initial test was unsuccessful; purging took several hours to reduce the relative
humidity to the correct level and following over an hour of testing the radon-222
concentrations were approaching zero. Considering the sample came from one of Dangila
town’s deep supply boreholes a concentration of several thousand Bq/m3 would be
expected. Later investigation revealed that a one-way valve, in place to prevent water
from entering the RAD7 instrument, had been installed the wrong way round. Once the
valve had been reinstalled correctly, the equipment operated satisfactorily.
Sites for testing were selected along surface and groundwater flow reaches to attempt to
identify areas that were losing to groundwater and areas where groundwater discharged
to surface water. Photographs of the equipment set up and testing in progress are shown
in Figure B-5.
Figure B-5. Photographs showing radon-222 testing. The in-situ EC, pH, and temperature meter is also
visible.
B-8-2 Results
The first phase of testing concentrated on Dangesheta kebele and took place over three
days; the locations and results of the testing can be seen in Figure B-6. The highest radon-
222 concentrations were obtained high on the elevated floodplain close to monitoring
well MW2. The samples were very recently discharged groundwater and the high values
are as expected. The radon-222 concentrations decrease along the surface water flow path
through the wetland towards the Brante River. This decrease is caused by dilution of
222 222
spring-fed surface waters (high Rn) with rain-fed runoff (zero to very low Rn)
through the wetland. The main Brante River always shows lower concentrations than the
307
wetlands due to dilution of recent groundwater input. The pattern shown by the testing
along the Brante River is higher radon-222 concentrations along reaches through
floodplains with alluvium beds and lower concentrations in narrow valleys with basalt
bedrock riverbeds. Radon-222 concentrations increase along the length of floodplains.
The results indicate that groundwater discharge occurs in the floodplains from the regolith
aquifer and not from basalt bedrock along narrow valley reaches. However, it should be
noted that the faster, more turbulent flow through rocky reaches would have a degassing
effect on the river water thus reducing radon-222 concentrations (Cook et al., 2003).
Radon-222 tests
222Rn (Bq/m3)
100 River flow
1,000 direction
10,000
C
! Dangila deep boreholes
1000 m
Figure B-6. Map of radon-222 testing results in Dangesheta within the Brante catchment.
The second phase of testing took place to the northwest of Dangila town; the locations
and results of the testing can be seen in Figure B-7. Quick water balance estimates had
suggested that there might be groundwater loss from the shallow to the deep aquifer. In
addition, an official at the Dangila town water supply office spoke of farmers complaining
that wells close to the town’s deep abstracting boreholes were suffering declining water
levels. If the deep boreholes were drawing water from the shallow aquifer this should
show up in radon-222 concentrations in the vicinity of the boreholes. The three deep
boreholes were drilled in 2009 to depths of 150-192 m in Berayta kebele and provide
Dangila town’s water supply. The boreholes are open hole with electric submersible
pumps at 60 m depth. The deep abstracting boreholes aim to tap groundwater in fractures
and scoriaceous layers. At the time of the second visit, only two boreholes were
abstracting, at 20 l/s and 32 l/s respectively, for 10 h/d. Testing was conducted on samples
308
from the Amen and Kilti rivers close to the abstraction boreholes and also at a distance
beyond their influence. The testing results were inconclusive concerning the impact of
the deep boreholes. The highest radon-222 concentration measured was actually at the
closest point to the northernmost borehole and though the value for a sample from beside
the centre borehole was much lower, a sample from upstream away from the borehole’s
influence was lower still. The results do match the findings from the Dangesheta testing
programme in that radon-222 concentrations increase as a river progresses through a
floodplain, again indicating that groundwater is discharging into the rivers predominantly
at floodplains.
Radon-222 tests
222Rn (Bq/m3)
100
1,000
10,000
C
! Dangila deep boreholes
N
River flow
direction
1000 m
Dangila
Figure B-7. Map of radon-222 testing results near deep boreholes at Kilti and Amen confluence. Note the
southernmost borehole was not abstracting at the time of the second field visit.
B-9 Summary
resource could be vulnerable to drought.
There is no hydrochemistry evidence to suggest mixing between the shallow and deep
groundwaters though there is evidence to suggest mixing between shallow groundwater
and surface water. Furthermore, radon-222 measurements showed the opposite of what
would be expected if surface water and shallow groundwater were being drawndown by
abstraction from the deep boreholes. However, the investigation was not exhaustive and
309
groundwater levels in shallow wells near the deep abstracting boreholes should be
monitored in order to observe any long-term changes in groundwater level.
Large floodplains in river valleys are areas of groundwater discharge, as shown by

radon-222 testing.
Other interesting findings resulted from comparing individual samples collected during
the same visit. Wells reported by the community to have a good year-round supply often
showed greater stable isotope enrichment and higher ionic concentrations than would be
expected from their topographic position close to a flow divide (see samples SI/A/C 4
and 5 in Table B-1). Often, across the flow divide, there was a dambo. The higher
enrichment (through evaporation when inundated) and higher concentrations (due to the
longer residence time) may indicate the sampled water originated in the dambo, which
provides continuous groundwater supply through the dry season with groundwater flow
paths contradicting surface water flow paths.
310
Appendix C. Filling the observational void: scientific value and
quantitative validation of hydrometeorological data from a
community-based monitoring programme – Supplementary
material
C-1 AMGRAF research project

As stated on www.upgro.org: “Unlocking the Potential of Groundwater for the Poor
(UPGro), is a seven-year international research programme (2013-2020) which is jointly
funded by UK’s Department for International Development (DFID), Natural
Environment Research Council (NERC) and the Economic and Social Research Council
(ESRC). Over 130 of the world’s best researchers from 43 organisations across Africa
and Europe are focused on improving the evidence base around groundwater availability
and management in Sub-Saharan Africa. The goal is to ensure that the hidden wealth of
Africa’s aquifers benefit all citizens and the poorest in particular. UPGro projects are
interdisciplinary, linking the social and natural sciences to address this challenge.”
AMGRAF was one of 15 UPGro catalyst projects.
Dangesheta kebele was selected as a focus community within Dangila woreda to assess
the potential of the shallow groundwater resource to support increased irrigation use. A
kebele is the smallest administrative unit in Ethiopia; equivalent to a parish or ward.
Dangesheta is one of 27 rural kebeles within Dangila woreda, a woreda being similar to
a district. The selection of Dangesheta kebele for hydrogeological study followed
collaboration with IWMI, GSE and a field visit in September/October 2013: The rural
kebeles of the woreda were ranked for intervention, according to: (i) access to market, i.e.
proximity to an all-weather road and distance to market: necessary for the adoption of
groundwater irrigation, (ii) experience in small-scale irrigation, and (iii) potential of
shallow groundwater, i.e. evidence of existing shallow groundwater use. Shallow
groundwater is here defined as <25 m; that which is accessible to the poor rural
communities using manual excavation methods.
Further information on the background of the AMGRAF project and the work completed
to date can be found in (Gowing et al., 2016). A key aspect of the research, which is being
conducted as part of a PhD project, is to evaluate the vulnerability of the shallow
groundwater resource. A greater understanding of the shallow groundwater system is
being achieved through fieldwork, water chemistry, stable-isotope and radon-222
311
analysis, recharge assessments and physically-based modelling. The latter two elements
rely on hydrometeorological data though formal rainfall and river flow data is sparse and
groundwater data is non-existent. Recharge must be quantified in order to assess the
sustainability of irrigative agriculture and such quantification can be conducted
(alongside other less direct techniques) with water-table fluctuation and cumulative
rainfall departure (CRD) methods (Xu and van Tonder, 2001) only with time series of
rainfall and groundwater level. Physically-based modelling of the shallow groundwater
system will allow the simulation of potential changes and variations in climate, land use
and abstraction to assess the impacts on the groundwater resource, on surface water and
on downstream users. Time series of river flows and groundwater levels at various
locations within the catchments allow thorough calibration and construction of
representative transient models. The limited available formal data mean it has been
necessary to implement a new hydrometeorological monitoring scheme.
C-2 Dangesheta community-based monitoring network

A workshop was held with the local community in February 2014 and the monitoring
equipment was presented prior to installation. The community were encouraged to handle
and use the equipment, including measuring water level in a nearby well. Following
installation of the equipment (rain and river gauges) the community were shown the
equipment in-situ and took part in surveying the rivers at the gauge locations, measuring
flow velocity with an electrical current-meter, and were tested in their ability to take
measurements.
The community member selected to monitor the wells and host the rain gauge is a known
and respected member of the community due to his occupation in the Dangesheta Service
Cooperative where the local population deliver agricultural produce to be weighed and
transported to market in Dangila. Having such a person involved in the community-
monitoring programme was deemed to be crucial to gaining the acceptance and
continuing support of the local community. The community members who would
monitor the river gauges were chosen because they lived very close to the river gauges
and again were known in the local community. The Irrigation and Water Manager of the
Dangila woreda, who was instrumental in organising community workshops and site
selection, provides support and receives measurement data from the monitors then types
up the data onto an Excel spreadsheet before forwarding it on to IWMI in Addis Ababa.
The rating curve for converting river stage to flow was developed from flow gauging
measurements. Flow gauging was conducted by IWMI and Newcastle University staff
312
during the AMGRAF catalyst project (prior to commencement of the PhD), by David
Walker during each field visit, and by Bahir Dar PhD and master’s students at other times
following training by David Walker. The flow gauging always involved local community
assistance. A current velocity meter was utilised and the velocity-area method was applied
to calculate discharge. Flow gauging was conducted at various seasons to incorporate
variations in river stage. The rating curves for the Brante and Kilti Rivers are presented
in Figure C-1.
10 45
9 y= 8.478x2 - 19.755x + 17.181 40 y = 12.716x2 - 64.679x + 93.718
8 35
7
30
Flow (m3/s)
Flow (m3/s)
6
25
5 y = 4.0193x2 + 0.0435x - 0.0566
20
4
3 15
2 10
1 5
y = 2.0506x2 - 0.041x + 0.0003 y = 2.0691x2 - 0.2932x + 0.0514
0 0
0 0.5 1 1.5 2 0 1 2 3 4 5
Water level above lowest point (m) Water level above lowest point (m)
Figure C-1. Rating curves for the Brante (left) and Kilti Rivers at the locations of the community-
monitored river gauges. Note that the kinks in the rating curves relate to when flow becomes out of bank.
The community-monitoring programme was explicitly designed to be low cost. Previous

such programmes by IWMI, detailed in Zemadim et al. (2014), confirmed that low cost
community monitoring programmes are the most sustainable. The number of wells, rivers
and rain gauges to be monitored was considered the minimum possible in order to obtain
sufficient initial data from this pilot study to allow planning of a longer-term community-
monitoring programme. In addition to utilising minimal low cost equipment, payment of
members of the community involved in the monitoring was designed to be a small
financial incentive rather than a wage. The decision to provide remuneration for
observational duties was made in order to get the project up and running with a longer-
term aim of continuing the monitoring with no payment once the benefit to the community
was seen.
C-3 Formal ground observations – rainfall

The NMA rain gauges in the region of Dangila woreda vary from tipping bucket type
housed within professionally manned meteorological stations, to electronic weather
stations within small fenced compounds, to traditional graduated cylinder rain gauges
313
monitored by local part-time observers. Data quality checks were conducted and, as may
be expected, higher quality data was identified from the professionally manned
meteorological stations of which Dangila is one:
Double mass checks were conducted for the nine NMA rain gauges in the vicinity of
Dangila woreda. The checks indicate that the Dangila rainfall record is the most reliable.
Several of the NMA rainfall records showed breaks of slope when the double mass checks
were conducted. These rain gauges were visited, the person responsible for taking
measurements was interviewed and the regional NMA office in Bahir Dar was visited and
consulted in an attempt to determine the causes of the breaks of slope. The precise reasons
could not always be ascertained though suggestions were proffered, for example; the
breaks of slope seen in the Meshenti rainfall record (Figure C-2) occur approximately
when the data collector changed from father to daughter to sister each of whom perhaps
had a slightly different measurement routine.
a.
Dangila cumulative rainfall (mm)
5000
4000
3000
2000
1000
0
0 1000 2000 3000 4000 5000
Merawi cumulative rainfall (mm)
b. 8000
Meshenti cumulative rainfall (mm)
6000
4000
2000
0
0 2000 4000 6000 8000
Dangila cumulative rainfall (mm)
314
Figure C-2. Double mass checks of rainfall for Dangila and Merawi NMA (a) showing good linear
relationships indicating reliable records. Compare to graph (b) of Dangila and Meshenti NMA rainfall
records showing breaks of slope in early 2008, early 2009 and early 2010.
C-4 Gridded datasets – rainfall

The gridded remote sensing and reanalysis rainfall datasets that have been considered are
TRMM, ERA-Interim, NASA MERRA, JRA-55 and NCEP:
TRMM (Tropical Rainfall Measuring Mission) was a joint research mission between
NASA and JAXA measuring rainfall in the tropics. The satellite was launched in 1997
and re-entered Earth’s atmosphere in 2015 providing a 17-year rainfall dataset at a grid
resolution of 0.25o x 0.25o. The instruments on board the TRMM satellite included a
precipitation radar, a microwave imager, a visible and infrared scanner, a cloud and Earth
radiant energy sensor, and a lightning imaging sensor (Adler et al., 2007).
ERA-Interim is the third generation ECMWF data assimilation reanalysis product

produced using a four-dimensional variational analysis. The rainfall dataset is available
from 1979 to present on a 0.75o x 0.75o grid. Details of the data assimilation method, the
forecast model and the input datasets can be found in Dee et al. (2011).
NASA-MERRA (Modern-Era Retrospective analysis for Research and Applications) was

generated with version 5.2.0 of the Goddard Earth Observing System (GEOS)
atmospheric model and data assimilation system (DAS). The rainfall dataset is available
from 1979 to present at a grid resolution of 0.666667o x 0.5o. Details of the DAS and the
processing strategy can be found in Rienecker et al. (2011).
JRA-55 is a third generation reanalysis product and was the first to apply four-dimensional
variational analysis. The rainfall dataset is available from 1958 to present on a 1.25o x
1.25o grid. Details of the observational datasets and data assimilation can be found in
Kobayashi et al. (2014).
NCEP (the NCEP/DOE AMEP-II Reanalysis) is a first generation reanalysis product.

The rainfall dataset is available from 1979 to present at a grid resolution of 0.5o x 0.5o.
For more details on this product see Kanamitsu et al. (2002).
The spatial resolution of these gridded datasets can be seen in Figure C-3 relative to the
study area. Each grid square evidently comprises quite an altitudinal range and where
multiple NMA rain gauges are present within a grid square the observed variations in
rainfall totals can be very high.
315
TRMM grid size: 0.25o x 0.25o ERA-Interim grid size: 0.75o x 0.75o
NASA-MERRA grid size: 0.666667o x 0.5o JRA-55 grid size: 1.25o x 1.25o
NCEP grid size: 0.5o x 0.5o

Figure C-3. Remote sensing and reanalysis dataset grids overlain on area around Dangila woreda.
C-5 River flow

A double mass check of monthly flow totals from the formal gauges; “Kilti Nr Durbete”
and “Amen @ Dangila” shows two significant breaks of slope; in August 2001 and April
2014 (Figure C-4a). It has been proposed by MoWIE that the more recent break of slope
could be due to the rating-curve becoming unreliable and efforts are ongoing to remedy
this (S. Mamo, personal communication, 10 December 2015).
A double mass check was conducted for the two community monitored rivers (Figure
C-4b) and a significant step in the data can be seen beginning in January 2015. The
explanation for this feature is because the Brante River almost dries up from January to
March 2015 (as it usually does during the dry season) whereas the larger Kilti River has
perennial flow.
316
a. b.
Apr.14
Jan.15
Aug.01
Figure C-4. Double mass checks of river flow
C-6 Formal observations – groundwater

According to the Amhara Regional State Bureau of Water Resources Development, there
are fifteen known boreholes within Dangila woreda of which three are 25-75 m in depth
and twelve are 100-200 m deep. Only four are abstracting for domestic supply, the rest
having been abandoned, typically due to decreased yield. Borehole yield data are
available for some of the boreholes at the time of drilling and sporadically since. Data on
the shallow aquifer is limited to the quantity of hand-dug wells within the woreda;
estimated at 2281 in September 2013 (Gebregziabher and Haile, 2013), though the lack
of regulation on hand-dug wells means this figure is difficult to keep updated.
317
Appendix D. Insights from a multi-method recharge estimation
comparison study – Supporting Information
D-1 Non-selection of certain recharge estimation methods

Recharge estimation methods can be subdivided to the three hydrological zones, namely
surface water, the unsaturated zone and the saturated zone (Scanlon et al. 2002) with a
further category for methods that consider all zones (Table D-1).
Table D-1. Methods of recharge estimation grouped by hydrological zone. Methods applied in this study
are marked with *.
All zones Surface water Unsaturated zone Saturated zone
*Basin water balance Channel water budget *Soil moisture balance (SMB) *Water table
fluctuation (WTF)
*Numerical modelling *Streamflow Lysimeters
hydrograph methods *Rainfall infiltration
*Empirical methods, Zero flux plane
breakthrough
e.g. rainfall-recharge Seepage meters
Infiltration models model (RIB)
relationship
Heat tracers
Darcy’s law
Applied tracers, e.g. bromide or
Remote sensing, e.g.
Isotopic tracers coloured dye
GRACE, InSAR Groundwater dating,
a e.g. tritium or
Numerical modelling Historical tracers, e.g. tritium or
*Large-scale mapping
36Cl CFCs
*Environmental
Environmental tracers, e.g.
tracers, e.g.
chloride
chloride mass
a Numerical modelling balance (CMB)
Empirical methods, e.g. a Numerical

infiltration coefficients modelling
a
SHETRAN modelling used in this study couples surface and subsurface (both unsaturated and saturated zones)
flows.
All the techniques presented in Table D-1 were explored and several were rejected for
this study for the following reasons:
 Insufficient stream gauge locations were available for the channel water budget
method.
 There are no surface water bodies available for seepage meter and thermocouple
(for the heat tracer method) installation with the exception of highly seasonal
318
rivers in which installed equipment would not be secure during wet season
floods.
 Stable isotopes are very useful for providing information on recharge sources
but not for quantitative recharge estimation (Healy, 2010).
 Lysimeters and zero flux plane equipment were excluded due to their high cost
and maintenance requirements.
 Insufficient soil properties data were collected for use in infiltration models or to
estimate infiltration coefficients.
 Applied tracers were not used due to the potential cultural conflicts that could
arise from application of (albeit, harmless) chemicals into groundwater that is
used domestically.
 The high rainfall and shallow thin aquifer leads to low groundwater residence
times, therefore bomb-pulse tritium and 36Cl are unlikely to still be present.
 There is insufficient spatial groundwater level and hydraulic conductivity data to
determine hydraulic gradient for application of Darcy’s law.
 Sampling groundwater for CFCs requires zero atmospheric contact (Gooddy et
al., 2012), which is impossible to achieve from open wells.
 Satellite based remote sensing is increasingly used to evaluate changes in
groundwater storage, in particular with the Gravity Recovery and Climate
Experiment (GRACE) satellite mission (Tapley et al., 2004), and Interferometric
Synthetic Aperture Radar (InSAR) (Galloway et al., 1998). InSAR was not
considered for this study, as its application is most suited to arid and semi-arid
areas (Galloway and Hoffmann, 2007). GRACE measures temporal variations in
the Earth’s gravity field, which are used to estimate changes in terrestrial water
storage. It has a spatial resolution of ~300 km and a monthly temporal resolution
(NASA, 2016), both of which were considered too coarse for this study.
It can be seen in Table D-3 that none of the methods rejected by this study was used in
other Ethiopia recharge studies, with the exception of tritium analysis, though this was
used for source assessments and timing rather than quantification (e.g. Demlie et al.
(2007); Girmay et al. (2015)). However, there may be some bias as many of the studies
share authors, or supervisors in the case of university theses.
319
D-2 Recharge estimates for the study site from large-scale mapping and
modelling
There have been several attempts to produce global scale groundwater recharge maps
beginning with Lʹvovich (1979), whose map was based on the estimation of the baseflow
component of observed river discharge. More recently, Döll and Flörke (2005) introduced
recharge into their WaterGAP Global Hydrological Model (WGHM). The authors state a
shortfall of the model in that it is calibrated against measured river discharge only, due to
the unavailability of direct recharge measurements. The scarcity of independent recharge
estimates meant the model was tuned in arid and semi-arid areas against just twenty-five
recharge assessments (nine of which were in sub-Saharan Africa) using chloride and
isotope profiles. The WGHM was updated by Döll and Fiedler (2007) and incorporated
into the World-wide Hydrogeological Mapping and Assessment Programme (WHYMAP,
2016) by BGR (the German Federal Institute for Geosciences and Natural Resources) and
UNESCO. MacDonald et al. (2012a) used WHYMAP to produce the recharge map in
their “Quantitative maps of groundwater resources in Africa”. The recharge estimates for
the study site from WHYMAP and the other large-scale studies are shown in Table D-2.
The Africa-wide recharge map presented by Altchenko and Villholth (2015) utilised data
from the PCR-GLOBWB global hydrological model (van Beek et al., 2011), which is
calibrated against river discharge and reanalysis evapotranspiration data. At national
scale, Ayenew et al. (2008b) produced a hydrogeological framework of Ethiopia based
on analysis of pertinent information from governmental and non-governmental
organisations and academic institutions, as well as over a decade of field study comprising
conventional hydrogeological mapping, well drilling, geophysics, hydrochemical and
environmental isotope analysis and remote sensing techniques.
Table D-2. Comparison of recharge estimates from large-scale mapping/modelling studies.
Source Scale Resolution Recharge

o
WHYMAP (2016) Global 0.5 20-100 mm/a
Altchenko and Villholth (2015) Africa 0.5o 100-300 mm/a
Ayenew et al. (2008) Ethiopia ~40 km* 250-400 mm/a
* Estimated from the map zonation.
The large-scale mapping and modelling methods share two characteristics that lessen
confidence in those recharge results: 1) a necessarily low spatial resolution, and 2) the
scarcity of recharge studies in the sub-Saharan Africa region that would aid calibration.
Concerning the first point, at the resolution of 0.5o for WHYMAP and the continental
320
recharge map, it is unsurprising that the recharge estimates are dissimilar to those utilising
catchment and local-scale methods. This region of Ethiopia has high variation in
elevation, topography and climate, from 1100 m and hot dry plains immediately to west
of Dangila, to cold wet mountains over 3000 m to the east (Figure 6-1 of Chapter 6).
These environments often appear in the same grid square at these scales. The particular
usefulness of the maps in data scarce regions coincides with a higher uncertainty in those
regions as products frequently cannot be validated. This point is valid not just for recharge
but remote sensing and reanalysis weather data products (Walker et al. 2016). The global-
scale WHYMAP clearly underestimates recharge and was rejected from this study.
D-3 Field data not described in the main text [of Chapter 6] that was used to
develop the conceptual model and to parameterise models
Geological surveys, water point surveys and workshops with the local community
The soils and aquifer properties for input into SHETRAN relied on field investigations as
soil and geological mapping of the area is not available at sufficiently high resolution.
The observed variation in well depths, differences in monitored well responses, variation
in well pumping/recovery test results, geological and soil observations, and discussions
with local communities led to the definition of hydrogeological zones.
Outcrops are visible in riverbeds, occasionally on steeper slopes and in a few man-made
excavations. The basalts are variously massive, fractured and vesicular with variations
occurring in short distances. Above the solid geology lies weathered basalt regolith, itself
overlain by red clayey loam soils. Local communities report that there are rarely problems
with well sidewall collapse. Often the solid geology is reached abruptly and well
excavation is halted. Depth to the top of the solid geology is variable. Wells are typically
excavated until further excavation becomes impossible, therefore, the location of
rockhead can be inferred from well depth. Over the three field visits, 80 wells were
measured for estimation of regolith thickness; more wells were visited but access for
measurement, such as in the case of wells fitted with handpumps, was not always possible.
Rockhead was generally found to be deeper in more steeply sloping areas and shallower
in floodplains. Rivers have often incised to the level of the rockhead where solid basalt
forms the riverbed with banks of only 1 to 3 m in height.
In addition to season, topography appears to govern shallow groundwater availability.

The variations in geology are sufficiently subtle, particularly concerning the regolith,
which forms the shallow groundwater aquifer, to be less of a control on the hydrogeology
321
than geomorphology. Near the end of the dry season in March/April within the
floodplains where the solid geology is at a depth of around 4 m, the water table lies at 2-
4 mbgl. The water table can often be seen as a seepage face at this depth within riverbank
sections in floodplain sediments. However, on the larger and steeper slopes where
rockhead is around 15 m deep the water table is at a depth of 12-15 m. Thus, the shallow
aquifer is thicker on slopes giving deeper water tables and generally greater saturated
aquifer thickness.
It is noted that farmers often talk of a well excavation striking rock at a shallow depth and
being dry, then when the well is relocated a short distance away (~10 m) rock is struck at
greater depth and the well fills with water. Such a situation is commonly ascribed to
heterogeneous rockhead, however, the unsuccessful wells are perhaps more likely to be
due to the presence of large and massive basalt boulders lying higher in the weathered
profile as are often visible in riverbank sections.
Over 200 hand-dug wells were surveyed during field visits; the surveys included GPS
location, depth and water level measurements, description of geology, topography, land
use, pump/lifting device and cover, in-situ measurement of water temperature, pH and
electrical conductivity, and discussions with local community over the well’s use,
seasonality and history. Water point surveys also included assessment of springs, many
of which are used by the local community, whether developed or not, to collect water for
domestic and potable use. Where springs and seepages emerge from gullies they
commonly occur at contacts between regolith and bedrock or gravelly regolith and more
solid regolith. Springs and seepages are also very common around the edges of
floodplains where the water table from the surrounding slopes intercepts the ground
surface.
Hydrochemistry and stable isotope sampling, radon-222 measurements
The conceptual model was developed with the aid of hydrochemistry investigations.
Samples of shallow and deep groundwater, surface water and rainwater were analysed in
the laboratory for major ions, some trace elements, and stable isotopes oxygen-18 and
deuterium (δ18O and δ2H). In-situ testing involved measurement of pH, electrical
conductivity (EC), total dissolved solids (TDS), temperature, and radon-222
concentration.

322
resource could be vulnerable to drought. Surface water and shallow groundwater belong
to the “bicarbonate calcium” type typical of recent recharge. The deep groundwater is of
“bicarbonate sodium” type indicative of higher mineralisation due to longer residence
time and greater distance of flow. The shallow groundwater samples from the wet season
are very similar in chemistry to surface water samples indicating a high degree of and
rapid interconnectivity. This was expected in the wet season from the observed very
shallow water table. There is no hydrochemical evidence to suggest mixing between the
shallow and deep groundwaters; they belong to clearly different water types. What’s
more, Radon-222 measurements showed the opposite of what would be expected if
surface water and shallow groundwater were being drawndown by abstraction from the
222
deep boreholes: Rn concentrations would be lower in the vicinity of the abstracting
boreholes as groundwater discharge would be prevented but the reverse was measured.
Radon-222 measurements did suggest that the large floodplains in river valleys are areas
of groundwater discharge from the shallow regolith aquifer, whereas the narrower valleys
with basalt riverbeds are not discharge areas.
Vegetation properties
Categorisation of vegetation was required for the SHETRAN models. The three
categories were:
 Grassland – floodplains that are almost always and entirely utilised as pasture.
 Arable – where the majority of land is devoted to rainfed agriculture. The crops
planted are 87% cereals (39% maize, 35% teff and 13% millet), with the rest being
pulses, oilseeds, sugarcane, potatoes, vegetables, fruits, onions, garlic, and
tomatoes (Belay and Bewket, 2013), the latter few generally occupying backyard
plots.
 Shrub – areas of higher gradients with characteristic scrub-like vegetation.
How SHETRAN converts the potential evapotranspiration time series into actual
evapotranspiration is specific for each of these categories. Each category within
SHETRAN differs in rooting depth and root density at different depths, ground coverage
at maximum seasonal extent, canopy storage capacity, leaf area index, and AET/PET at
particular soil moisture tensions. These parameter values were estimated from ground
observations in order to generalise vegetation and crop types followed by consultation of
the key instructional texts for calculating water demand; FAO24 (Doorenbos and Pruitt,
1975) and FAO56 (Allen et al., 1998), and other published studies providing detail of
323
particular vegetation types, particularly, Canadell et al. (1996); Dardanelli et al. (1997);
Cain (1998) and; Fan et al. (2016).
D-4 Empirical method

There are numerous published examples of the development of a rainfall-recharge
relationship, often utilising secondary literature sources e.g. Bonsor and MacDonald
(2010), Crosbie et al. (2010), Zhang et al. (1999). A series of Boolean searches were
undertaken in February 2017 on Google Scholar and repeated on Web of Science in order
to identify quantitative recharge studies from Ethiopia. The common Google search
engine was also utilised to identify reports and other documents that may be outside of
the usual scientific literature. The initial search terms were “groundwater AND recharge
AND Ethiopia”, which produced 1000s of documents. Despite including the search term
“Ethiopia”, many of the “hits” were from other countries and could be discarded
immediately. Many other studies that did take place in Ethiopia had mention of recharge
but had no recharge estimate and were also discarded. The first 150 hits were reviewed
for inclusion in the analysis before the searches became more specific by adding a
recharge estimation method in an attempt to exclude the non-quantitative recharge
studies. Specific searches involved adding a fourth term to the search described above.
The added search terms were (independently): “baseflow”, “chloride”, “balance”,
“fluctuation”, “wtf”, “tritium”, “lysimeter”, “zero flux plane”, “channel water budget”,
“seepage meter”, “heat tracer”, “Darcy’s law”, “infiltration model”, “bromide”, “applied
tracer”, “isotropic tracer”, “infiltration coefficient” and “CFC”. The first 100 documents
were reviewed for each search, or all the documents if there were fewer than 100 hits. In
an attempt to fill in the gaps when the study site locations were plotted on a map of
Ethiopia (Figure D-1), further searches were conducted with a geographic location
replacing the third search term (the first two still being “groundwater AND recharge”).
The geographic search terms used were the regions of Ethiopia that were under or not
represented on the map of study sites, namely: “Gambella”, “Benishangul Gumuz”,
“Afar”, “Somali AND Ethiopia” and “SNNPR”. The first 50 documents were reviewed,
seeing as, by this stage of the search, the majority of hits had been reviewed previously.
Table D-3 presents information on all of the studies identified in addition to the reference.
Several studies are included that were found during other literature searches undertaken
for other aspects of the wider AMGRAF (Adaptive Management of shallow
GRoundwater for small-scale irrigation and poverty alleviation in sub-Saharan AFrica)
project. Where a study was cited within another study, an attempt was made to locate the
cited study for review. Occasionally, a cited study could not be found though sufficient
324
detail was provided for its inclusion in this project (listed in Table D-3 as “… cited in…”).
Several times a cited study with or without sufficient detail was not listed in the citing
study’s reference list and as such, after a failed attempt at location, it was not included in
this project. Attempts were made within Ethiopia to access often-cited but seldom seen
reports via university and other organisations’ libraries with limited success. Forty-nine
quantitative studies were located comprising 22 peer-reviewed articles, the rest being grey
literature (predominantly MSc theses). Where a study used multiple methods or multiple
catchments, recharge results were considered independently. Therefore, 102 annual
recharge estimates could be plotted against annual rainfall. Various trendlines were fitted
through the data, excluding points with rainfall below certain thresholds (in an Africa-
wide study, Bonsor and MacDonald (2010) recognised a linear relationship between
recharge and rainfall above 500 mm/a). A quadratic trendline, reflecting an increase in
recharge disproportionate to increasing precipitation, achieved the best R2 and standard
error.
Table D-3. Details of the recharge estimation studies used to develop a new empirical recharge method
for Ethiopia based on the rainfall-recharge relationship. Note that multiple recharge results from the same
study relate to different recharge estimation methods applied and/or to different catchments or areas of the
study site. CMB = chloride mass balance method, SMB = soil moisture balance method, SNNPR =
Southern Nations, Nationalities and Peoples’ Region, WTF = water table fluctuation method.
Location Area Recharge Annual Annual Publication Reference

(km2) estimation rainfall recharge
method (mm) a (mm) a
Raya Valley 2579 Water balance 813 50 Addis Ababa Abdella (2011)
Basin CMB 52 University
Kobo 1351 Water balance 813 63 master’s thesis
Valley CMB 53
Basin,
Tigray
Gilgel Abay 4178 SWAT model 2141 753 Landscape Abiy et al.
Catchment Dynamics, (2016)
Gumera 1418 1424 438 Soils and
Catchment Hydrological
Ribb 2132 1407 453 Processes in
Catchment Varied
Megech 661 1251 407 Climates
Catchment,
Lake Tana
Basin, NW
Ethiopia
Lower 41887 CMB 553 29 Addis Ababa Addisu (2012)

Awash Sub- University
Basin 4092 553 58 master’s thesis
Aysha
Basin,
Afar
325
Geba Basin, 5150 WetSpass model 550 22 KU Leuven Alene (2006)
Tigray (Belgium), cited in
master’s thesis Tesfagiorgis et
al. (2011)
Negelle ? ? 647 31 GSA Ali Jr (2006)

Borena, (Geological
south Society of
Ethiopia America)
Annual
Meeting
presentation
Teji 700 Streamflow 1104 109 Addis Ababa Andualem

Catchment, hydrograph University (2008)
central Water balance 325 master’s thesis
Ethiopia incorporating
SMB and
streamflow
hydrograph
Gilgel 1664 Streamflow 1376 70 ITC Asmerom

Abay, Koga hydrograph (Netherlands), (2008)
and Kilti master’s thesis
Catchments
Megech 514 BASF rainfall- 959- 50-77
Catchment runoff model 1212
Lake Tana 6316 CMB 1094- 45-155
Basin, NW 1730
Ethiopia
Weybo 574 Water balance 1341 88 Addis Ababa Aychluhim

Catchment, incorporating University (2006)
SNNPR SMB master’s thesis
Streamflow 63
hydrograph
(baseflow
separation)
Streamflow 75
hydrograph
(Meyboom
method)
CMB 124
Lake Tana 15339 “CMB, baseflow 1094- 70-120 Addis Ababa Ayalew (2010)
Basin, NW separation, etc.” 1730 University
Ethiopia b
master’s thesis
Meki Basin, 3051 Streamflow 762- 80 SINET: Ayenew

central hydrograph 1138 Ethiopian (2008)
Ethiopia WATBAL water 63 Journal of
balance model Science
Lake 1455 MODFLOW 1030 47 Lakes and Ayenew and

Awassa model Reservoirs: Tilahun (2008)
Basin, Research and
central Management
Ethiopia
326
Raya Valley 2480 WATBAL water 813 86 Water Ayenew et al.
Basin, balance model International (2013)
Tigray
Guder Sub- 7088 SMB 1424 258 Journal of Azagegn et al.

Basin CMB 210 African Earth (2015)
Muger Sub- 8263 SMB 1201 150 Sciences
Basin CMB 163
Jema Sub- 6760 SMB 992 164
Basin CMB 120
Upper 16000 SMB 1112 232
Awash Sub- CMB 133
Basin,
central
Ethiopia
Southern 1664 c Streamflow 1541 c 308 Technical BCEOM

Lake Tana hydrograph report for the (1998) cited in
Basin, NW Ministry of Ayalew (2010)
Ethiopia Water
Resources
(MoWR)
Kulubi area, ? Water balance 626 c 150 Technical BCEOM

Dire Dawa using assumed report for the (2005) cited in
Dengego ? runoff 626 c 50 Ministry of Tilahun and
area, Dire coefficients Water Merkel (2009)
Dawa Resources
Between 260 Spring-area 721 43 (MoWR)
Melko relationship
Jebdu and
Hurso, Dire
Dawa, NE
Ethiopia
Lake 505 WTF 534 42 University of Belay (2009)

Beseka EARTH modelling 47 Bonn
Basin, CMB 1.2 (Germany)
central PhD thesis
Ethiopia
Upper 6735 Water balance 1077 131 Journal of Berehanu et al.

Awash Sub- CMB 135 Geoscience (2017)
Basin Streamflow 91 and
hydrograph Environmental
Muger Sub- 1770 Infiltration model 1077 157 Protection
Basin Water balance 125
CMB 148
Streamflow 158
hydrograph
Jema Sub- 304 Infiltration model 1077 239
Basin, Water balance 130
central CMB 122
Ethiopia Streamflow 86
hydrograph
Infiltration model 239
Akaki 1500 CMB 1254 265 d Hydrological Demlie et al.

Catchment, processes (2007)
central
Ethiopia
327
Akaki 1464 SMB 1254 105 Environmental Demlie (2015)
Catchment, CMB 273 Earth Sciences
central
Ethiopia
Fogera 500 Soil moisture 1360 850- Land Enku et al.

Plain, profiles and 1000 e Degradation & (2017b)
Lake Tana groundwater Development
Basin, NW level/evaporation
Ethiopia relationship
Adama- 1760 MODFLOW 860 123 Environmental Furi et al.

Wonji model Earth Sciences (2011)
Basin,
central
Ethiopia
Werii, 1797 WetSpa 717 30 b Haramaya Gebremeskel

Tekeze WetSpass University (2015)
Basin, (Ethiopia)
Tigray master’s thesis
Geba Basin, 5260 WetSpass model 400-950 41 Journal of Gebreyohannes

Tigray Hydrology et al. (2013)
Dire Jara 85-90 ? 626 c 31 Technical Gibb and

and Hurso, report for the Seureca (1996)
Dire Dawa, Ministry of cited in
NE Ethiopia Water Tilahun and
Resources Merkel (2009)
(MoWR)
Dire Dawa, ? ? 626 c 40 Hebrew Greitzer (1970)

NE Ethiopia University, cited in
Jerusalem, Tilahun and
PhD thesis Merkel (2009)
Upper 4489 SMB 924 21 Addis Ababa Habtamu

Wabe Sub- Water balance 23 University (2009)
Basin, Streamflow 155 master’s thesis
central hydrograph
Ethiopia
Raya Valley 1085 CMB 724 116 ITC Hagos (2010)

Basin, MODFLOW 114 (Netherlands),
Tigray model master’s thesis
Berga 303 Water balance 1119 83 Addis Ababa Hussen (2006)

Catchment, incorporating University
central SMB and master’s thesis
Ethiopia streamflow
hydrograph
Aynalem 104 CMB 670 30-40 ITC Kahsay (2008)

Wellfield, MODFLOW 42 (Netherlands),
Mekelle, model master’s thesis
Tigray
Gedeb 290 SWAT model 1392 467 Proceedings of Koch et al.

Catchment, 2012 (2012)
central international
Ethiopia congress on
328
environmental
modeling and
software
managing
resources of a
limited planet,
sixth biennial
meeting,
Leipzig,
Germany
Gidabo 3302 SWAT model 800 25 Journal of Mechal et al.

Basin, south 1600 410 Hydrology: (2015)
Ethiopia Regional
Studies
Koraro 59 Water balance 549 57 Momona Nedaw (2010)

Area, incorporating Ethiopian
Tigray SMB Journal of
Science
Meki Basin, 1669 Water balance 992 117 Addis Ababa Netsanet
central incorporating University (2007)
Ethiopia SMB master’s thesis
Becho 1552 Water balance 1131 320 Addis Ababa Nuramo (2016)
incorporating University
SMB master’s thesis
Streamflow 81
hydrograph
Koka, 1461 Water balance 879 50
Upper incorporating
Awash, SMB
central Streamflow 104
Ethiopia hydrograph
Koka 1461 Water balance 900 27 Addis Ababa Reys (2016)

Becho, 1552 incorporating 1026 227 University
Upper SMB and master’s thesis
Awash, streamflow
central hydrograph
Ethiopia
Shaya 504 SWAT model 1071 174 HESS Shawul et al.

Watershed, (2013)
SE Ethiopia
Bulbul 508 Water balance 1520 350 Asian Journal Shimelis et al.
Basin, SW incorporating of Applied (2014)
Ethiopia SMB and Science and
streamflow Engineering
hydrograph
Streamflow 395
hydrograph
Upper 2075 Streamflow 1232 129 Addis Ababa Sintayehu

Bilate hydrograph University (2009)
Catchment, Water balance 96 master’s thesis
SNNPR incorporating
SMB and
streamflow
hydrograph
329
May Nugus 15 Water balance 738 19 International Tadesse et al.
Catchment, incorporating Journal of (2010b)
Tigray SMB Earth Sciences
and
Engineering
Illala 340 WetSpass model 550 66 Momona Teklebirhan et

Catchment, Ethiopian al. (2012)
Tigray Journal of
Science
Bilate 5625 Water balance 1146 116 Addis Ababa Tesfaye (2010)
Catchment, incorporating University
SNNPR SMB and master’s thesis
streamflow
hydrograph
Streamflow 201
hydrograph
Dire Dawa, 920 WetSpass model 626 28 Hydrogeology Tilahun and

NE Ethiopia Journal Merkel (2009)
Zenako- 4 MODFLOW 724 167 Hydrogeology Vandecasteele

Argaka model Journal et al. (2011)
Catchment,
Tigray
Zenako- 2 Runoff model “525- 110- Hydrological Walraevens et

Argaka SMB model 900 334 b Sciences al. (2009)
Catchment, MODFLOW (average
Tigray model = 687)”
Mendae 5 CMB 512 18 b Land Walraevens et

Plain, SMB model Degradation al. (2015)
Tigray and
Development
Aba’ala 254 Hydrochemistry 340-480 43 Research and Woldearegay
woreda, development (2004)
Afar experience on
dryland
husbandry in
Ethiopia
Aynalem ? ? 576 53 Addis Ababa Yihdego

and Illala University (2003) cited in
Catchments, master’s thesis Teklebirhan et
Tigray al. (2012)
Aynalem ? MODFLOW 550 c 61 Mekelle Zeru (2008)

Catchment, model University cited in
Tigray master’s thesis Teklebirhan et
al. (2012)
Gilgel Abay 1640 Streamflow 1614 379 Addis Ababa Zewdie (2010)
Catchment hydrograph University
Gumera 1394 1292 158 master’s thesis
Catchment
Megech 492 1081 62
Catchment
Ribb 1592 1263 77
Catchment
330
Koga 302 1410 234
Catchment
Kilti 698 1322 57
Catchment,
Lake Tana
Basin, NW
Ethiopia
a
Where a range is given, the mean rainfall or recharge was used for the plot.
b
It is uncertain which technique gave which result.
c
Unclear from the study; therefore, the value is taken from a study in the same area.
d
For uncertain reasons, this value is stated to be an overestimate.
e
Not included in the analysis as the study-specific conceptual model renders the recharge value
inapplicable.
Recharge
study sites
Dangila
study site
0 300 km
Figure D-1. Location map of the study area with other recharge study sites identified in the literature
review shown on the right (image source: Google.Earth; Imagery ©2017 DigitalGlobe).
D-5 Streamflow hydrograph methods

Recharge estimation using streamflow hydrograph methods typically involves separating
the baseflow component (Figure D-2) and approximating this to groundwater recharge.
In humid regions, according to Döll and Fiedler (2007), who required recharge
assessments for the entire globe for calibration of their WaterGap Global Hydrology
Model, streamflow hydrograph methods are the most commonly applied recharge
estimation method and numerous examples are available in the literature. The methods
are idealised in assuming that groundwater storage remains constant interannually, or is
in balance over longer time periods, and that precipitation entering the aquifer as recharge
must be balanced by groundwater discharge into rivers that forms baseflow.
331
120
Overland flow
100 Interflow
Baseflow
80
River flow (m3/s)
60
40
20
Figure D-2. The components of a streamflow hydrograph. Total flow is the sum of the three components,
or the entire area below the Overland flow curve. The plot is a snapshot of the WETSPRO analysis of
Kilti river flow.
Three streamflow hydrograph methods were used in this study, the baseflow recession
method presented by Meyboom (1961), and two digital recursive filter tools, the Web
GIS based Hydrograph Analysis Tool (WHAT) (Lim et al., 2005) and WETSPRO
(Willems, 2009). For both Meyboom and WHAT, the interflow is part of the quick flow
component whereas for WETSPRO the interflow is separated in a two-step digital filter
method after baseflow filtering. There are many very similar digital filter programs
available for baseflow separation and two were chosen to assess whether they would give
significant differences in recharge result. More in-depth comparison studies of baseflow
separation methods are available, e.g. Chapman (1999), Eckhardt (2008).
The Meyboom method uses analysis of baseflow recession from at least two consecutive
years to estimate recharge. The stream hydrograph is plotted on semi-logarithmic paper
creating a straight-line recession curve. The start and end times of the recession “curve”
are noted manually. According to Meyboom (1961), the total potential groundwater
discharge can be estimated from
𝑸𝟎 𝒕𝟏 (D-1)
𝑽𝒕𝒑 =
𝟐. 𝟑
where Vtp is the total potential groundwater discharge, Q0 is the baseflow at the start of
the recession, and t1 is the time that it takes the baseflow to drop from Q0 to 0.1 Q0. The
332
amount of potential baseflow, Vt, remaining at some time, t, after the initiation of baseflow
may be estimated by:
𝑽𝒕𝒑 (D-2)
𝑽𝒕 =
(𝒕 )
𝟏𝟎 ⁄𝒕𝟏
The difference between the remaining potential groundwater discharge at the end of a
given baseflow recession and the total potential groundwater discharge at the beginning
of the next recession represents the recharge that takes place between these two
recessions, i.e.
𝑽𝒕𝒑 − 𝑽𝒕 = 𝑹 (D-3)
where R is the total quantity of recharge which is divided by the basin area to give a value
in mm/a.
The Meyboom method is well-used, e.g. Mau and Winter (1997), Kumai and Mitamura
(2004), Berhail et al. (2015), though seems inappropriate for this study site. The recharge
quantities calculated using the Meyboom method are the lowest of all methods.
Uncertainty of manually choosing start times of baseflow recessions could lead to
recharge underestimation when early times of interflow recession have been incorrectly
identified as baseflow recession (Figure D-3). Furthermore, when plotted on semi-log
graph paper, baseflow recessions of the three rivers did not always form a straight-line
meaning the Meyboom method is rejected for this study site.
100.0000
Brante Kilti sub-catchment
10.0000
1.0000
Flow (m3/s)
0.1000
Uncertainty over
whether this is the start
of baseflow recession
0.0100 or interflow recession
Baseflow recession does

not form a straight line
on semi-log graph paper
0.0010
0.0001
14
27
40
53
66
79
92
1
105
118
131
144
157
170
183
196
209
222
235
248
261
274
287
300
313
326
339
352
365
378
391
404
417
430
443
456
469
482
495
508
521
534
547
560
573
586
599
612
625
Days
Figure D-3. Snapshot of the Brante and Kilti hydrographs showing uncertainties encountered with the
Meyboom method.
333
The WHAT method requires a filter parameter a (the recession constant) and a BFImax,
which is the maximum value of long-term ratio of baseflow to total streamflow. The
developers of WHAT recommend using representative BFImax values proposed by
Eckhardt (2005): 0.80 for perennial streams with porous aquifers, 0.50 for ephemeral
streams with porous aquifers, and 0.25 for perennial streams with hard rock aquifers. The
local hydrological system meant a BFImax of 0.50 was most appropriate for this study,
which was confirmed by a calculated BFImax 0.57 (average of the three catchments). The
WETSPRO method uses two main parameters to filter out the baseflow and interflow
respectively. For each of the two components a recession constant k, which relates to the
WHAT filter parameter as a = exp(-1/k), and the portion contributing directly to runoff w
are defined. Both parameters are dependent on catchment size and characteristics.
D-6 Soil moisture balance (SMB)

The SMB method has the advantage over other unsaturated zone methods that it is not
point based, rather it is at the scale that the precipitation, potential evapotranspiration
(PET) and soil property inputs remain applicable. Rainfall totals were utilised in addition
to the required meteorological time series to enable PET computation using the Penman-
Monteith FAO-56 method (Allen et al., 1998). The Thornthwaite-Mather (1955, 1957)
(T-M) method additionally requires a value for soil moisture retention (MC), which is
dependent upon vegetation and soil type. MC, or ‘root zone storage’, equals field capacity
multiplied by depth of root zone. MC values were assigned to each LULC class and
recharge was calculated individually for each class (Tables D-4 and D-5). The SMB
method calculates total monthly actual evapotranspiration (AET) and soil moisture
surplus with this surplus contributing to recharge. The direct runoff component is dealt
with in two ways:
(a) It can be subtracted from the precipitation input by applying a runoff factor that could
be taken from literature (e.g. Bakundukize et al. (2011)), derived from streamflow
hydrograph analysis (e.g. Demlie (2015)), or modelled (e.g. Walraevens et al. (2009)).
(b) A portion of the soil moisture surplus is subtracted; Thornthwaite and Mather (1957)
recommend subtracting 50% (e.g. Chishugi and Alemaw (2009), Azagegn et al. (2015).
Both methods were applied in this study; the runoff factor based on a simple flow
separation (IOH, 1980) conducted on the longer time series river flow records (Kilti and
Amen).
334
Calculation of groundwater recharge using a SMB involved the application of the
Thornthwaite and Mather (1955, 1957) method and equating soil moisture surplus to
recharge. The SMB calculations are shown in Table D-4 with a description of the
parameters below.
Table D-4a. Calculation of actual evapotranspiration (AET), soil moisture deficit and soil moisture surplus
(from which 50% forms recharge) using the Thornthwaite-Mather (1955, 1957) method. The year 2000
has been selected and grassland LULC category (MC = 200 mm) as an example. All values are in mm.
J F M A M J J A S O N D Annual
total
P 0 7.1 7.5 77.0 135.3 344.1 313.0 436.1 237.8 265.3 70.0 2.5 1895.7
PET 98.7 105.1 117.4 119.1 114.5 86.3 81.3 84.6 90.1 93.4 92.4 90.0 1172.7
P-PET -98.7 -98.0 -109.9 -42.1 20.8 257.8 231.7 351.5 147.7 172.0 -22.4 -87.5
APWL -208.6 -306.6 -416.5 -458.5 0 0 0 0 0 0 -22.4 -109.9
SM 70.5 43.2 24.9 20.2 41.0 200.0 200.0 200.0 200.0 200.0 178.8 115.5
∆SM -45.0 -27.3 -18.3 -4.7 20.8 159.0 0 0 0 0 -21.2 -63.4
AET 45.0 34.4 25.8 81.7 114.5 86.3 81.3 84.6 90.1 93.4 91.2 65.9 894.0
SMD 53.7 70.7 91.6 37.3 0 0 0 0 0 0 1.2 24.1
SUR 0 0 0 0 0 98.8 231.7 351.5 147.7 172.0 0 0 1001.7
Rech. 0 0 0 0 0 49.4 115.9 175.8 73.9 86.0 0 0 500.9
Table D-4b. Calculation of actual evapotranspiration (AET) with the application of a runoff factor, soil
moisture deficit and soil moisture surplus (which is equated to recharge) using the Thornthwaite-Mather
(1955, 1957) method. The year 2000 has been selected and grassland LULC category (MC = 200 mm) as
an example. All values are in mm.
J F M A M J J A S O N D Annual
total
Peff 0 0.2 10.5 26.9 131.3 230.8 290.4 316.5 208.4 83.4 17.1 1.3 1316.7
PET 98.7 105.1 117.4 119.1 114.5 86.3 81.3 84.6 90.1 93.4 92.4 90.0 1172.7
P-PET -98.7 -104.9 -106.9 -92.2 16.7 144.5 209.1 231.9 118.3 -9.9 -75.3 -88.7
APWL -262.7 -367.6 -474.5 -566.7 0 0 0 0 0 0 -75.3 -164.0
SM 53.8 31.8 18.6 11.8 28.5 173.0 200.0 200.0 200.0 200.0 137.3 88.1
∆SM -34.3 -21.9 -13.2 -6.9 16.7 144.5 27.0 0 0 0 -62.7 -49.2
AET 34.3 22.1 23.7 33.8 114.5 86.3 81.3 84.6 90.1 93.4 79.8 50.5 794.3
SMD 64.4 83.0 93.7 85.3 0 0 0 0 0 0 12.5 39.5
SUR 0 0 0 0 0 0 182.1 231.9 118.3 0 0 0 532.3
Rech. 0 0 0 0 0 0 182.1 231.9 118.3 0 0 0 532.3
335
Where
P = monthly rainfall total as measured by the NMA (National Meteorological Agency of

Ethiopia) at the meteorological station in Dangila town, in the above examples for the
year 2000.
Peff = monthly effective rainfall, which is the monthly rainfall total minus the direct runoff
computed by a simple flow separation (IOH, 1980) applied to river flow data; Peff = P.(1-
RF) where RF is the runoff factor calculated to be 0.145.
PET = potential evapotranspiration calculated using the Penman-Monteith FAO-56

method (Allen et al., 1998) with input parameters measured by the NMA at Dangila, in
the above examples for the year 2000.
APWL = accumulated potential water loss; the summation begins with November, the
first month of the dry season, until end April.
MC = moisture capacity; also known as soil moisture retention and equals field capacity
multiplied by depth of root zone. It is dependent upon vegetation and soil type and, in the
above examples, is specified as 200 mm for the grassland land use/land cover (LULC)
category (see further explanation below).
SM = soil moisture; the soil moisture during dry months is obtained using accumulated
potential water loss by the following formula: SM = MC.exp(APWL/MC). For the wet
months, the soil moisture is calculated by adding the excess rainfall of the current month
to the soil moisture of the previous month; where this exceeds the moisture capacity, the
excess is booked as moisture surplus.
∆SM = change in soil moisture from the previous month.
AET = actual evapotranspiration; for the wet months AET = PET as it is assumed that all
rainfall is available for plants. During dry months, AET = P + moisture loss from the soil
(i.e. negative ∆SM).
SMD = soil moisture deficit; the difference between PET and AET.
SUR = surplus moisture available for infiltration. According to Thornthwaite and Mather
(1955, 1957) and shown in Table D-4a, 50% of SUR = recharge (Rech). Alternatively,
where a runoff factor is applied, shown in Table D-4b, 100% of surplus moisture forms
recharge (SUR = Rech).
336
Rech. = recharge
MC values were assigned to each LULC class and the recharge was calculated for each.
The recharge values were then multiplied by the proportional area of each LULC class
and summed to give the recharge for Dangila district. The LULC information was taken
from ADSWE (2015) and the MC values were based on field identification of soil and
vegetation types then assigned according to published values (see FAO-UNESCO world
soils database and mapping; Batjes (1997) and Nachtergaele et al. (2010)) (Table D-5).
Table D-5. Representative MC values and proportional coverage of LULC classes for Dangila woreda.
LULC class Coverage (%) MC (mm)

Built up 8.72 10 a
Cultivated 71.7 150
Forest 11.0 300
Grassland 7.9 200
Shrub and bush 0.5 250
a
The built up areas have some patches of vegetation; therefore, a nominal MC was applied.
D-7 Basin water balance

The water balance, or water budget, was the most commonly used method identified
during the literature review of Ethiopian recharge studies (see Table D-3 for examples).
AET is not straightforward to estimate and was calculated with three methods for
comparison: (1) The T-M method mentioned in the previous section (Table D-4); (2)
Application of Turc’s formula (Turc 1954), see below for a full description, and; (3) A
value estimated by Allam et al. (2016) for this region of the Tana Basin by combining
remote sensing and river flow records. The average AET values were 789, 831 and 931
mm/a, respectively. Annual average runoff values were obtained using a simple and
standard flow separation method (IOH 1980); the separated baseflow was subtracted from
total flow to give the direct runoff component. The basin water balance can be increased
in complexity by computing at a daily time step and utilising high-resolution soil and
vegetation mapping with which to calculate AET; such data were not available for this
study. Accurate quantification of all the fluxes is always troublesome though is required
in order to leave an accurate residual that is equated to actual recharge (Scanlon et al.
2002). There persists the potential for unaccounted groundwater depletion as described
in the streamflow hydrograph section.
Turc’s formula requires only annual precipitation and temperature and was established
empirically based on 254 watersheds, globally distributed (including 10 in Ethiopia) in
337
different climatic zones (Turc, 1954). Actual evaporation from a catchment, AET, is
defined as
𝑨𝑬𝑻 = 𝑷/((𝟎. 𝟗 + (𝑷𝟐 ⁄𝑳𝟐 ))−𝟎.𝟓 ) (D-4)
Where P is average annual precipitation in mm and
𝑳 = 𝟑𝟎𝟎 + 𝟐𝟓𝑻 + 𝟎. 𝟎𝟓𝑻𝟑 (D-5)
Where T is average annual air temperature in oC.
D-8 Chloride mass balance (CMB)

The CMB method requires: Peff – the average annual effective precipitation (rainfall
minus direct runoff), Clwap – the weight-average chloride concentration in precipitation
including dry deposition, and Clgw – the average chloride concentration in groundwater.
Direct runoff was calculated using a simple flow separation (IOH 1980) for the two longer
time series streamflow records, the Kilti and Amen. Dry deposition at this distance from
the coast is considered negligible (Keywood et al., 1997) and is typically neglected. Clgw
from the 31 shallow groundwater samples was 2.10 mg/l with a standard deviation of 1.33
mg/l. Clwap was 0.68 mg/l (standard deviation = 0.32 mg/l); this is the most uncertain
parameter of Eqn. 6-3 in Chapter 6 given the limited amount of samples as, ideally,
samples from throughout the wet season should be obtained. However, the Clwap value
compares well with other studies (Table D-6) giving confidence that the value used is
representative of rainfall chloride concentrations in this region. The use of few rainfall
samples and corroboration with other studies is not uncommon, e.g. Bazuhair and Wood
(1996), Subyani (2004), and in South Africa Clwap is often unmeasured and simply
approximated to 1 mg/l (Dennis, 2017), e.g. Butler and Verhagen (2001), and many theses
available online.
338
Table D-6. Comparison of rainfall chloride concentrations with other studies.
Clwap Source Region Altitude Distance to Rainfall

(mg/l) (m asl) coast (km) (mm/a)
0.68 This study Dangila, NW ~2000 ~600 1541
Ethiopia
0.50 Kebede et al. (2005) Tana Basin, NW ~2000 ~600 ~1500
Ethiopia
0.86 Asmerom (2008) Tana Basin, NW ~2000 ~600 ~1500
Ethiopia
0.70 Demlie et al. (2007) Addis Ababa, ~2300 ~550 1254
Central Ethiopia
0.71±0.18 Vallet-Coulomb et Lake Ziway, ~1650 ~600 ~900
al. (2001) Central Ethiopia
D-9 Water table fluctuation (WTF) and rainfall infiltration breakthrough

(RIB)
The locations of the hand-dug wells used in the water table fluctuation (WTF) and rainfall
infiltration breakthrough (RIB) analyses are presented in Figure D-4. Details of the five
monitoring wells initially set up as part of the AMGRAF project and the community-
based monitoring programme can be found in Walker et al. (2016). The depth to
groundwater was measured every two days by a community-nominated observer using a
dip meter at 6am (prior to well use). The ILSSI project (Innovation Lab for Small-Scale
Irrigation funded by USAID) community-based monitoring programme operated in a
similar fashion. Two local community members using dip-meters at 6am monitored
groundwater level in twenty-five hand-dug wells. Measurements were made weekly in
the dry season and daily through the wet season.
339
Figure D-4. Location map of the AMGRAF and ILSSI monitoring wells.
The WTF method has the key assumption that, because recharge rates vary substantially
within a catchment due to differences in elevation, geology, slope, vegetation, and other
factors, monitoring wells should be sited so the water levels are representative of the
entire catchment (Healy and Cook 2002). Identifying a “representative” location is
problematic, therefore, the possibility of recording groundwater level change due to
lateral groundwater flow is considerable.
0 100
Daily rainfall
1 90
MW1
2 Projected water table decline 80
Water level rise h
3 70
Well water level (mbgl)
Rainfall (mm)
4 60
5 50
6 40
7 30
8 20
9 10
10 0
Figure D-5. Groundwater hydrograph through the wet season and determination of water table rise for the
WTF method. MW1 refers to the groundwater level in monitoring well 1 from where this snapshot is
taken. (mbgl = metres below ground level).
340
Two RIB method parameters, “lag days” and “length days”, were calibrated within the
model. “Lag days” indicates the time it takes percolating rainwater to reach the water
table. It was assigned as 1 day (pre and post-calibration) because the shallow water table
leads to a short time lag between rainfall and groundwater level peak. “Length days”
refers to the length of related rainfall events and is adjusted to gain the best fit between
simulated and observed groundwater levels (Figure D-6). Lateral groundwater inflows
and outflows can be specified in the model though the difficulty in quantifying these
means they are typically set at zero under the assumption that they are in balance. As with
the WTF method, this generates the possibility of accounting for groundwater level rise
from lateral flows in recharge estimation.
60
rainfall Recharge amount WLF dh(rib)
3
Rainfall, groundwater recharge (mm)

50
Water level fluctuation (m)
2
40
1
30
-1
20
-2 10
-3 0
60
1
121
182
244
305
366
425
486
547
609
670
731
790
851
912
974
Days of simulation
Figure D-6. Graphical output of the RIB model showing observed rainfall, observed groundwater level
fluctuation (WLF), simulated groundwater level fluctuation (dh (rib)) and computed recharge. This plot
shows the simulation of monitoring well MW3.
D-10 SHETRAN modelling

The main advantages of SHETRAN over alternative physically based spatially distributed
river basin modelling systems are its comprehensive nature and capabilities for modelling
subsurface flow and transport. The subsurface is treated as a variably saturated
heterogeneous porous medium, and fully three-dimensional flow and transport can be
simulated for combinations of confined, unconfined, and perched systems. The
‘‘unsaturated zone’’ is modelled as an integral part of the subsurface, and subsurface flow
and transport are coupled directly to surface flow and transport (Ewen et al., 2000).
SHETRAN is well established in the literature, having been applied to a variety of
situations such as predicting climate and land use change impacts on a Mediterranean
341
catchment (Parkin et al., 1996), modelling landslide sediment yield in Scotland (Burton
and Bathurst, 1998), coupling flow and nitrogen transport in Sweden (Birkinshaw and
Ewen, 2000b), and modelling forest impact on floods caused by extreme rainfall and
snowmelt in Latin America (Bathurst et al., 2011a).
Both saturated and unsaturated zones of the subsurface are represented by a single
equation:
       ( k r Kz) (D-6)
 = [K x k r ] + [K y k r ]+ [Kz k r ] + -q
t x x y y z z z
where
 d
 = Ss +
(D-7)
n d
and
kr = relative hydraulic conductivity (-)
(Kx,Ky,Kz) = principal components of saturated hydraulic conductivity (m/s)
n = porosity (-)
q = specific volumetric flow rate out of the porous medium (general source/sink term)
(1/s)
Ss = specific storage (1/m)
t = time (s)
(x,y,z) = ordinates of the position vector (m)
 = volumetric soil water content (-)
 = storage coefficient (1/m)
 = pressure potential (m)
As this equation is continuous across saturated-unsaturated boundaries, fluxes across the

boundary are implicit in the solution, unlike groundwater models such as MODFLOW
where the control volume is the saturated zone defined by the time-varying water table as
its upper boundary, and recharge is defined separately as a boundary input. In SHETRAN,
therefore, recharge is a derived variable, which includes both the flux (vertical flow per
342
unit area) through the moving water table, and the rate of capture (or loss) of water in the
saturated zone as the water table position moves. There is also no explicit use of a variable
equivalent to specific yield, so an approximation of this concept is defined here as the
amount of water available due to drainage of pore water space above the water table,
derived from the shape of the unsaturated zone characteristic functions. Recharge is
therefore calculated as:
𝑸𝒓𝒄𝒉 = −𝑸𝒗 + 𝑺𝒚 𝒅𝑯⁄𝒅𝒕 (D-8)
where
H = phreatic surface (i.e. water table) level (m)
Qrch = recharge rate (m/s)
Qv = vertical velocity (m/s), +ve upwards
dH = change in phreatic surface level (m), approximated as change in pressure

potential (Δover the timestep in the highest saturated cell at the end of the timestep
Sy = specific yield (-), approximated as change in water content (Δ in the cell above
the water table over the timestep, implemented as  Δ
dt = timestep (s)
SHETRAN was manually calibrated using an iterative approach with the adjustment of
geological layer thicknesses, aquifer properties, channel characteristics, Strickler
overland flow roughness coefficient, and evapotranspiration characteristics. The range of
values used for the input parameters was determined from field investigations and
literature review of models set up for similar climates. Calibration aimed to minimise the
error between observed and simulated time series. Calibration and validation periods were
selected to give “typical” ranges of hydrological conditions and ran from the end of a wet
season recession to the same point one, two or three years later. In addition to visual
comparison of plotted observed and simulated data, the following performance indicators
were utilised: Nash-Sutcliffe Efficiency (NSE) coefficient (Nash and Sutcliffe, 1970),
where a value greater than 0.5 is considered acceptable (Moriasi et al., 2007), and root
mean square error (RMSE), with units matching the compared data thus the value should
be as low as possible. NSE is very sensitive to peak flows (Krause et al., 2005), therefore,
given the flashy nature of the rivers with short-lived and relatively extremely high peaks,
NSE (and RMSE) was calculated on low flows following hydrograph separation. For the
343
Brante model, NSE and RMSE were calculated on groundwater levels in the five
monitoring wells selected at the onset of the AMGRAF project. The Brante model was
principally calibrated against monitored groundwater levels (with consideration of river
flow) combined with a semi-quantitative calibration to other areas of the catchment where
simulations were compared against occasional observations and anecdotal evidence of
frequent flooding or wells prone to drying up. The Amen and Kilti models were chiefly
calibrated against river flow, though again with consideration of groundwater level
information from around the catchments. A validation period was run to confirm that the
calibrated parameters still produced a satisfactory simulation for independent input
datasets. Table D-7 shows that the calibration statistics are acceptable for both the
calibration and validation periods for all catchment models.
Table D-7. Details and statistics of the calibration and validation periods for the SHETRAN catchment
models.
Catchment Calibration period No. of days NSE RMSE
Amen 5 Apr 1999 to 4 Apr 2001 (years 2-3) 731 0.79 0.19 m3/s
Kilti 16 Apr 1998 to 15 Apr 2000 (years 2-3) 731 0.78 1.47 m3/s
Brante 12 Mar 2014 to 11 Mar 2015 (year 1) 365 0.69 2.01 m
Catchment Validation period No. of days NSE RMSE
Amen 12 Mar 2010 to 11 Mar 2013 (years 13- 1096 0.75 0.13 m3/s
15)
Kilti 2 Apr 2004 to 2 Apr 2007 (years 8-10) 1096 0.67 2.30 m3/s
Brante 12 Mar 2015 to 11 Mar 2016 (year 2) 365 0.53 2.08 m
D-11 Comparison of recharge results from the three nested catchments

In this case, catchment scale means 37-632 km2 for the three catchments in this study.
Despite being nested, the catchments have slightly different characteristics, in terms of
proportional land cover and topography, therefore, differences in recharge result would
be expected. Additionally, it is useful to evaluate if the different spatial and temporal
scales contribute to the discrepancies in recharge results between catchments. The Amen
and Kilti analyses utilise 17+ years of input data, the Brante only three, with little overlap.
What’s more, those three years include 2014; the wettest year on record with annual
rainfall of 2005 mm, and 2015 with annual rainfall only in the 20th percentile (1390 mm).
Figure D-7 shows the recharge results plotted for each catchment. While no spatial scale
344
dependence can be seen, temporally, the shorter and alternative data periods of the Brante
catchment are contributing to the reduced recharge estimate of the streamflow hydrograph
methods. For each year from 2014-2017, the streamflow hydrograph methods give much
lower recharge estimates than for the longer period analyses of the Amen and Kilti.
However, the Brante catchment is flatter with the greatest proportion of floodplain
wetlands and shallow water tables; therefore, direct groundwater evaporation would be
elevated causing the low minimum recharge computations.
800 50% 800 50%
Median annual recharge %MAP

Median annual recharge mm/a
Amen, 37.0 km2, Kilti, 631.7 km2,

data: 1988-2014 40% data: 1997-2014 40%
600 600
30% 30%
400 400
20% 20%
200 200
10% 10%
0 0% 0 0%
800 50%
Median annual recharge mm/a
Median annual recharge %MAP
Brante, 65.5 km2

Data: 2014-2017 40%
600
30%
Potential recharge
400
20%
Actual recharge
200
10%
0 0%
Figure D-7. Graphical comparison of annual recharge estimates from the catchment-scale techniques
separated into catchments. T-M = Thornthwaite-Mather method of AET estimation.
D-12 Insights gained on the conceptual model in other recharge studies

Other studies exist where fewer methods were applied and useful insights were gained.
These insights are specific to the conceptual model of the study site. King et al. (2017)
applied four recharge estimation methods to an alluvial aquifer in Queensland, Australia.
Low CMB recharge results indicated that inherent assumptions were invalidated and
channel leakage and overland flow were significant at the site. The WTF method gave
the highest recharge results, especially in proximity to a river, which was considered,
since the timing coincided with high stream levels, to be due to temporary influxes of
water (bank storage). While a water balance gave a useful approximation of recharge for
the catchment, only by applying and comparing the additional methods could the
conceptual model be updated and understood. Takounjou et al. (2011) compared a hybrid
345
WTF/water balance method with the CMB method for a humid region of Cameroon. The
discrepancy in results led the authors to suggest the CMB method was overestimating due
to preferential shallow groundwater flow paths; ultimately, they considered the CMB
method to be unsuitable for a humid forested environment). Huang et al. (2017), by
comparing results from a CMB method with groundwater aging and stable isotope
analysis, were able to update the conceptual model for a site in northwest China revealing
that no recharge had occurred for >2,500 years, and, as such, potential abstraction of the
paleaowater would be unsustainable. Misstear et al. (2009) compared SMB, WTF,
numerical groundwater modelling and water balance recharge estimation methods for an
aquifer in Ireland. The variations in WTF recharge estimates at particular locations, in
comparison with more consistent results from other methods, indicated Sy variations
within the aquifer.
D-13 Annual recharge time series

The interannual variation between recharge estimates and between methods is shown in
Table D-8.
346
Table D-8. Annual recharge time series calculated by the methods that do not apply temporally averaged
input data. Meyboom results not shown as they were rejected from the study. All results in mm.
YEAR WHAT WETSPRO SHETRAN WTF RIB

Brante Kilti Amen Brante Kilti Amen Brante Kilti Amen Graph. Simp.
1988 321 288

1989 235 224
1990 202 186
1991 237 219
1992 159 170
1993
1994
1995 440 483
1996 321 350
1997 180 146
1998 171 297 145 259 280 289
1999 225 347 176 325 300 297
2000 243 346 201 318 288 280
2001 148 259 116 298 289 266
2002 169 129 131 111 294 274
2003 162 131 142 119 287 265
2004 184 163 144 144 288 276
2005 153 155 127 133 267 252
2006 303 230 225 203 302 297
2007 277 172 207 145 301 279
2008 306 188 267 201 276 271
2009 190 138 169 121 313 286
2010 191 181 194 162 282 267
2011 213 228 184 219 290 277
2012 257 188 224 165 327 306
2013 182 315 280 286 296 292
2014 135 139 256 314 310 1077 893 288
2015 70 65 293 915 806 620
2016 93 94 270 871 763 636
347
Appendix E. Bahir Dar drought analysis and rainfall comparison
E-1 Bahir Dar drought analysis

A drought analysis was conducted for Bahir Dar with assessment of SPI and SPEI for
both spring and summer, growing season length using both Segele and Lamb (2005) and
Stern et al. (2006) criteria, annual rainfall totals from NMA ground observations, remote
sensing (TRMM) and reanalysis products (NCEP, JRA-55, ERA-Interim and NASA-
MERRA), and comparison with other studies using SPI and SRA. SPI indicates extreme
droughts were spring 2003 and summer 1982 (Figure E-1), likewise SPEI indicates
extreme drought in summer 1982 (Figure E-2), and both indicate very wet conditions in
the early-70s. Assessment of growing season length and SRA from other studies support
the findings from the SPI and SPEI. Table E-1 shows the results from all analysis methods
for all years.
Figure E-1. SPI calculated for Bahir Dar for spring (3-month, March-May) and summer (5-month,
June-October). Note the non-continuous year sequence, which is different to that of Dangila.
Figure E-2. SPEI calculated for Bahir Dar for spring (3-month, March-May) and summer (5-month,
June-October). Note the non-continuous year sequence, which is different to SPI and that of Dangila.
348
Table E-1. Identification of drought/wet years and comparison of drought analysis methods for Bahir Dar.
X = insufficient data for analysis. v = very. For SPI, SPEI and SPI/SRA from other studies: v dry and v
wet are <-2 and >2 respectively, dry and wet are <-1.5 and >1.5 respectively. For growing season length:
v long and v short are >±20% difference from mean length in days, long and short are >±10% difference
from mean length in days. Annual rainfall: dry is rainfall total <10%ile, wet is >90%ile.
Year SPI SPEI Growing season Annual rainfall Viste et al. Bewket Ayalew et
length (2013) NW and al. (2012)
Highlands* Conway Bahir
Segele Stern et (2007) Dar
NASA-MERRA
and al. (2006) Bahir SRA
ERA-Interim
Lamb Dar
summer SPI
ground obs.
spring SPI
(2005) SRA
summer
summer
TRMM
JRA-55
spring
spring
NCEP
1962 X X X X X X X X
1963 v short X X X X X X X X
1964 long wet X X X X X X X wet X
1965 dry X X X X X X X dry X
1966 v long long X X X X X X X X
1967 v short long X X X X X X X X
1969 v short X X X X X X X X
1971 wet wet long wet X X X X X X X wet X
1972 long X X X X X dry X
1973 v wet wet v long wet X X X X X v wet X
1974 v wet v wet wet X X X X X v wet X
1975 dry long long wet X X X X X dry X
1976 short X X X X X X
1977 long X X X X X X
1978 short X X X X X X
1979 long X wet wet wet
1980 short short dry X wet dry dry
1981 short X wet wet
1982 v dry v dry short short dry X wet wet v dry v dry v dry
1983 dry X wet wet wet wet
1984 v short X wet dry dry
1985 X wet wet wet dry
1986 X wet wet v dry
1987 dry dry long X wet v dry dry
1988 dry long X wet
1989 X X X X X X X wet wet
1990 X X X X X X X dry dry
1991 X X X X X X X X dry dry
1992 X X X X X X X v dry
1993 X X X X X X X dry wet
1994 X X X X X X dry X dry dry dry
1995 X X X X X X dry X dry dry dry dry dry
1996 X X X X X X X wet
1997 X X X X X X X v dry dry
1998 X X X X X X X wet wet wet
1999 X X X X X X wet
2000 X X X X X X X
2001 X X X X X X X wet wet dry
2002 X X X X X X X dry dry dry
2003 v dry X X X X X v dry v wet
2004 X X X X X X
2005 X X X X X
2006 X X X X wet wet wet X v wet
2007 X X long X
2008 X X X dry dry wet X
2009 dry X X X dry X dry v dry X X
2010 X X X X dry X X
2011 X X X X X X wet X X X X
2012 X X X X X X X X X X
2014 wet X X short wet X X X X X X
349
E-2 Comparison between Dangila and Bahir Dar rainfall
It was necessary to test whether the Bahir Dar rainfall data could be used to inform the
longer period climate at Dangila, i.e. are the coincident portions of the datasets
sufficiently similar. The correlation testing results shown in Table E-2 indicate a high
correlation between monthly rainfall totals. The poorer daily rainfall correlation is simply
due to timing as weather systems proceed slowly with the generally low winds (Kebede
et al., 2006; Setegn et al., 2010). Annual rainfall totals correlate well though with lower
significance due to few matching complete years.
Table E-2. Comparison between NMA Dangila and Bahir Dar rainfall for all coincident periods.
Period compared Resolution Missing data Pearson P-value Spearman P-value

1 Jan 2007 - 31 Oct Daily 915 out of 3226 0.403 0.000 0.703 0.000
2015 days (28%)
Jan 1922 - Apr 1924, Monthly 101 out of 475 0.890 0.000 0.915 0.000
Feb 1961 - Jun 1969, months (21%)
and Jan 1987 - Oct 2015
1922-1923, 1962-1967, Annual 16 out of 34 0.592 0.010 0.610 0.007
and 1989-2014 years (47%)
The histogram in Figure E-3 shows that the distribution of the daily rainfall data is very
similar for both sites with an almost identical standard deviation and an offset in the Bahir
Dar data of approximately 1 mm/day less than Dangila.
Figure E-3. Histogram of daily rainfall from Dangila and Bahir Dar for 1 Jan 2007 to 31 Oct 2015
(excluding non-matching days).
350
A further correlation was conducted between the wet days per month, here defined as
days with greater than 5 mm of rainfall (i.e. productive rainfall), for the overlapping
period of the daily datasets (Jan 2007 to Oct 2015). Bahir Dar and Dangila correlate very
well with a Pearson correlation coefficient of 0.938 (P-value = 0.000) and Spearman Rho
of 0.929 (P-value = 0.000). This justifies the use of Bahir Dar rainfall years in the Dangila
time series to be used for the climate variability future scenario SHETRAN simulations.
Tables E-3 and E-4 show that the percentile range is greater for the Bahir Dar dataset,
indicating that climatic extremes, both wet and dry, were not captured within the 22-year
Dangila rainfall dataset used in the SHETRAN modelling; confirming what was
discovered during the drought analysis. This can be seen in the histogram in Figure E-4
where Bahir Dar, allowing for its lower annual total offset, has a greater spread. Notably,
the Bahir Dar dataset includes the infamous Ethiopian droughts of the early-80s and the
very wet early-70s. The actual range of annual rainfall totals for Bahir Dar is 890 mm
(1982) to 2035 mm (1973) – two particular years that are not present in the Dangila record
– compared to the Dangila range of 1118 mm (1930) to 2005 mm (2014).
Table E-3. Long-term averages of NMA Dangila rainfall (Jan 1922 – Feb 1934, Aug 1955 – Jun 1969,
and Jan 1987 – Oct 2015). Note that the Total column relates to annual totals rather than the presented
monthly values.
J F M A M J J A S O N D Total
Mean 3.3 3.6 24.6 43.4 126.4 245.6 346.6 348.2 230.6 104.2 40.0 7.4 1533.2
StDev 7.6 5.0 28.6 38.6 74.6 55.8 68.6 71.9 63.5 64.5 38.8 12.3 218.9
Min 0.0 0.0 0.0 0.0 0.0 140.6 220.8 165.0 102.0 12.0 0.0 0.0 1118.0
c10 0.0 0.0 0.0 2.0 45.3 169.2 274.0 262.2 151.6 37.6 3.5 0.0 1245.1
c25 0.0 0.0 2.4 10.9 61.8 197.0 305.5 299.0 184.0 58.0 7.5 0.0 1389.2
Median 0.0 1.0 12.1 33.9 125.4 251.0 339.0 351.7 229.0 80.0 28.8 1.9 1517.8
c75 3.0 6.0 35.3 66.1 169.1 283.0 369.2 384.4 268.1 133.1 58.8 9.4 1652.5
c90 7.0 9.8 69.8 92.8 242.2 320.8 440.2 435.5 320.4 194.4 97.2 23.5 1858.7
Max 40.1 21.0 97.4 153.0 302.9 354.4 570.2 531.0 391.7 270.1 163.0 62.0 2004.7
351
Table E-4. Long-term averages of NMA Bahir Dar rainfall (Aug 1920 – Apr 1924, Jun 1938 – Aug 1939,
and Feb 1961 – Oct 2015). Note that the Total column relates to annual totals rather than the presented
monthly values.
J F M A M J J A S O N D Total
Mean 2.6 1.4 11.8 23.4 84.9 178.0 427.4 372.1 204.0 94.1 19.3 2.8 1429.4
StDev 4.7 4.4 21.8 27.4 70.0 70.7 100.4 105.6 55.7 52.5 25.8 6.1 233.5
Min 0.0 0.0 0.0 0.0 1.6 60.0 208.0 150.8 106.2 0.0 0.0 0.0 894.8
c10 0.0 0.0 0.0 0.0 11.3 86.1 305.6 243.2 138.9 23.2 0.0 0.0 1186.5
c25 0.0 0.0 0.0 1.3 34.3 124.2 350.4 293.4 163.3 55.6 2.1 0.0 1258.5
Median 0.0 0.0 4.0 11.0 76.0 178.2 417.9 368.1 200.9 100.5 8.9 0.0 1422.1
c75 3.1 0.3 13.6 33.5 112.3 216.3 481.9 447.7 240.6 123.2 24.6 2.6 1558.2
c90 9.1 4.1 27.4 66.3 153.7 261.5 557.2 504.1 262.9 168.2 52.4 9.8 1703.6
Max 20.3 26.9 118.5 111.0 363.0 404.9 643.9 648.2 378.5 206.6 107.2 34.8 2035.3
Figure E-4. Histogram of annual rainfall from Dangila and Bahir Dar for all complete years.
352
Appendix F. Water point surveys
The information in Tables F-1, F-2, F-3 and F-4 is from surveys conducted by Demis
Alamirew of GSE as part of the AMGRAF catalyst project in February and March 2014.
Longitude and latitude are in the UTM37N coordinate system, taken from a handheld
GPS along with elevation. SWL = static water level.
353
Table F-1. Hand-dug wells.
ID Woreda Kebele Site name Longitude Latitude Elevation m asl Geology Date of visit Field pH EC μS/cm Temp o C Depth m SWL mbgl Water strike mbgl Pump/cover
DW1 Guangu/Chagni Tigiri Metekel ber 243236 1226624 1887 Weathered basalt + soil 03/02/2014 5.79 59.8 22.8 19 16? hand pump
DW2 Dangila Washa Kulkul (lay Medhanialem) 254911 1245762 2205 soil and weathered basalt 04/02/2014 6.32 203 23.3 6 4 hand pump
DW3 Dangila Washa and Kabilita Washa & Kabilita school well/Robit 252972 1246806 2163 basalt with hematite and laterite stains 04/02/2014 7.34 344 26.5 32 27 hand pump
DW4 Dangila Wondayita Woranti gote) 259028 1244564 2119 weathered basalt and red soil 04/02/2014 6.14 86.5 25.3 13.5 about 4m hand pump
DW5 Dangila Dimsa Emadadi/lay Dimsa 256321 1243793 2225 weathered andesitic basalt/red soil 05/02/2014 0 hand pump
DW6 Dangila Washa Medhanialem Tach Yihun gote 254887 1244238 2199 weathered basalt 05/02/2014 12.5 4m, 5m hand pump
DW7 Dangila Warkit kebele Dekit 265130 1254942 2038 soil and fractured/weathered basalt 06/02/2014 10 7.5 hand pump
DW8 Dangila Warkit kebele Dekit/ Ato Yeshalem Zegey 265323 1255195 2043 weathered basalt and grey soil 06/02/2014 6.78 318 24.4 7 6.7 Metal sheet
DW9 Dangila Dengela Georgis Addis Alem 263753 1255546 2055 weathered basalt with varigated colours 06/02/2014 12.5 8 hand pump
DW10 Dangila Dengela Georgis addis Alem(Masresha Atalele's well 263940 1255494 2053 Soil and andesitic basalt 06/02/2014 6.84 301 24.8 4.5 3.6 3 pot cover
DW11 Dangila Warkit kebele Ato Hayimanot Wonde's well (Deki gote) 264930 1254845 2046 soil and weathered basalt 06/02/2014 6.37 197.6 21.8 11 7.2 9 Metal sheet
DW12 Dangila Warkit kebele Ato Fekadie Berku's well(Dekit gote) 265044 1254813 2045 soil and varigated colour weathered basalt 06/02/2014 7.11 307 24.8 10 8.2 Metal sheet
DW13 Dangila Warkit kebele Mokel 265561 1253511 2057 soil and weathered basalt 06/02/2014 6.27 151.3 27.1 11 6.89 hand pump
DW14 Dangila Warkit kebele Legasta 266550 1254196 2044 soil and weathered basalt 06/02/2014 6.29 124.5 26.4 5 hand pump
DW15 Dangila Warkit kebele Nana Minch/Legasta gote 266344 1253350 2047 soil and weathered basalt 06/02/2014 166.3 27 8 6 hand pump
DW16 Dangila Warkit kebele Nana minch(Ato Wubante Eyasu's well) 266327 1253334 2054 red soil (2m) and weathered basalt 06/02/2014 5.5 4.53 4.5 pot cover
DW17 Dangila Manguda Jabi gote 266925 1253133 2048 weathered basalt 07/02/2014 6.68 339 27.7 8.5 hand pump
DW18 Dangila Badani Delelti 249874 1239907 1922 boulder type basalt and black soil 07/02/2014 7.2 502 24.2 4 2 hand pump
DW19 Dangila Badani Akuacha 250264 1241521 1921 alluvium and weathered basalt 07/02/2014 6.52 680 27.9 7.5 6 hand pump
DW20 Dangila Badani Ambo/Guji 250363 1241161 1917 black loamy clay soil 07/02/2014 10 hand pump
DW21A Dangila Kuandisha Gezewetie 264708 1238299 2190 soil and fractured/weathered basalt 08/02/2014 6.79 162 19.8 9.5 6 hand pump
DW21 Dangila Kuandisha Gezewetie 264965 1237972 2203 loamy-silty soil and weathered basalt 08/02/2014 6.54 142 21.7 11 6 hand pump
DW22 Dangila Abla Mariam Mariam Wuha/Addis Alem gote 264115 1236875 2230 red clay top, silty loam dark soil bottom 08/02/2014 6.69 200 26.4 11 4.5 hand pump
Dw23 Dangila Abla Mariam Abla Mariam school well 264676 1237084 2229 fractured vesicular basalt and soil 08/02/2014 6.28 129.3 24.9 6 hand pump
DW23B Dangila Gayita Tach Gayita 270288 1237497 2159 red soil and regolith 08/02/2014 0 hand pump
DW24 Fagita Lekuma Ashewa Fera Cambo 267140 1231866 2362 silty clay soil and basalt from bottom 10/02/2014 6.89 93.2 23.7 6 hand pump
DW25 Fagita Lekuma Makia Teklehaymanot Makia 268237 1232047 2332 weathered basalt 10/02/2014 0 hand pump
DW26 Fagita Lekuma Abla Mariam Meskelti 265301 1231556 2401 silty loamy clay soil above vesicular basalt 10/02/2014 6.39 126.1 25.1 10 hand pump
DW27 Fagita Lekuma Meskelti Wondie Yenesew's well 265278 1231230 2388 weathered basalt 10/02/2014 6.18 113.3 22.3 6 4.8 5 pot cover
DW28 Fagita Lekuma Shangani Kesisi 263863 1230061 2394 red soil and boulder type basalt 10/02/2014 6.32 123 25.1 14 hand pump
DW29 Fagita Lekuma Gula Azmach Tankuari 269703 1228230 2411 red soil and basalt 11/02/2014 6.52 116.8 18.4 0 hand pump
DW30 Fagita Lekuma Tafoch Damburi Akuta 270339 1231449 2339 weathered basalt with pink tint 11/02/2014 7.73 150.2 19.8 24.5 hand pump
DW31 Fagita Lekuma Tafoch Damburi Dambul Elementry school compound 270147 1232224 2313 Deeply weathered basalt 11/02/2014 20 hand pump
DW32 Fagita Lekuma Tafoch Damburi Wonjela/Adurja 270441 1232461 2296 weathered basalt 11/02/2014 6.68 156.9 23.2 21 9 hand pump
DW33 Fagita Lekuma Giraita and Zembel Mariam Wuha gote 273870 1228651 2358 fractured and weathered basalt 11/02/2014 6.61 119.9 25.5 0 hand pump
DW34 Fagita Lekuma Segila Bambildawna Besena 280373 1224286 2476 weathered basalt 11/02/2014 7.11 131.8 22.4 27 12 hand pump
DW35 Fagita Lekuma Awsa Fenzit Fenzit school 276697 1228065 2347 weathered basalt 11/02/2014 9 hand pump
DW36 Dangila Afessa/Segino Gebeya Segino Gebeya 278706 1237135 2129 aquifer is boulder type basalt 12/02/2014 6.14 107.5 18.5 26 23.5 hand pump
DW37 Dangila Afessa Kes Adugna Fantahun/Arbit Gebreal 277399 1234214 2181 weathered vesicular basalt 12/02/2014 5.89 60.7 23.7 10 8.55 Metal sheet
DW38 Dangila Afessa Arbit Gebreal new 277384 1234070 2182 weathered basalt 12/02/2014 6.56 117 23.3 9.5 7.55 hand pump
DW39 Dangila Ligaba Setto 278169 1242126 2054 weathered basalt above massive basalt 12/02/2014 5.92 46.7 28.2? 16 9 hand pump
DW40 Dangila Ligaba Ligaba 277205 1243337 2034 soil + basalt? 12/02/2014 11 hand pump
354
ID Topography Land use Use Perennial Remark
DW1 flat land near small town domestic/town supply good supply A lot of users/Jericans in line
DW2 sloping area farm land domestic/rural supply good supply saturated water depth in the stream cut below well is 2m
DW3 top land near school compound school water supply good supply andesitic basalt at middle, and sandy material aquifer (oral information)
DW4 foot of gently falling area farm land domestic/rural supply water scarcity may/April Water strike depth is interpolated from spring levele downstream
DW5 gentle slope farm land domestic/rural supply high water scarcity not sampled because of poor storage
DW6 sloping area farm land domestic/rural supply good supply Pump broken on same day morning
DW7 flat land grazing land domestic/rural supply not functional/pump uninstalled the pump is uninstalled for fear of teaft
DW8 flat land house compound domestic,cattle well dries in April-May well couldn't be sunk below due to massive rock, water emerges at contact of weathered and fresh basalt
DW9 flat land near house domestic/rural supply supply was good but well not working due to pump stolen
DW10 gentle slope very near house and cattle shelter good supply andesitic basalt
DW11 flat land very near house domestic, cattle good supply Dug in 1997 E.C. and with 85cm diameter
DW12 flat land house compound domestic, cattle, and irrigation good supply Well EC is relatively higher than near by wells
DW13 flat land near farm land domestic supply good supply but not actively used Full of suspended material
DW14 flat land grazing land domestic good supply The swampy area near this well was wet throught the year till 1989E.C according to local people b/c of climate change
DW15 gentle slope near grziing land domestic supply good supply 6m soil, 2m weathered basalt, surrounding area was wet until 1986 E.C. (equaliptus coverage highly increased and climate changed, local people)
DW16 gentle slope near house domestic/cattle good supply
DW17 flat land 3.5m from house/cattle shelter domestic good supply only about 10m away from swamppy area; dug in may 2013
DW18 about 500m from ridge/gentle slope grazing land domestic supply good supply fresh boulder type jointed basalt is the main aquifer
DW19 flat land near farm land domestic supply well is located few hundred meters away from ridge foot
DW20 foot of hill grazing land domestic supply good supply
DW21A flat land farm land domestic supply The well couldn't be lowered below this depth due to massive basalt; 6m soil
DW21 flat land farm land domestic supply good supply
DW22 flat land grazing land domestic supply good supply main wateris in the black soil and and night storage during digging was 1.5m
Dw23 flat land school compound domestic good supply water storage was about 1.5m in a night and flow from fractured basalt mainly (oral information)
DW23B flat land near farm land Pump not functional. The well is located in higly /actively irrigated area
DW24 flat land grazing land domestic supply good supply The basalt has some regolith on top of it which stores water
DW25 gentle slope farm land not functional/pump uninstalled
DW26 flat land farm land domestic supply good supply
DW27 flat land farm land cattle good supply
DW28 gentle slope farm land domestic supply good supply
DW29 depression grazing land domestic supply good supply A spring located some 800m away at down stream emerges at the contact of weathered basalt and trachyte
DW30 plateau farm land domestic supply good supply weathered basalt with red tint aquifer, a lot of commercial plants for charcol production
DW31 plateau school compound school water supply good supply not sampled
DW32 plateau grazing land domestic and for seedling good supply located in depression
DW33 flat land near farm land domestic supply
DW34 flat land school compound domestic and school supply good supply
DW35 flat land school compound domestic supply good supply while working pump not functional
DW36 gentle slope rural town supply domestic supply scarcity in March soil is dry (local people, but the contact between soil and massive basalt is wet
DW37 flat land individual well cattle good supply
DW38 flat land Rural village supply under construction Under construction, aquifer is weathered basalt under relatively thick red soil
DW39 plateau Rural village supply domestic supply good supply
DW40 flat land Rural village supply domestic good supply
355
DW41 Dangila Ligaba Gilgel Badma 276292 1242522 2030 massive basalt on top and weathered basalt 12/02/2014 6.41 141.7 29.1 15 hand pump
DW42 Dangila Ligaba Asterio 275223 1243555 2011 soil with bottom massive basalt 12/02/2014 6.57 240 25.6 6 5.4 hand pump
DW43 Dangila Zelesa Kilage 268713 1244603 2132 red and black loamy soil intercalation 13/02/2014 5.82 56.8 18.9 11 2 hand pump
DW44 Dangila Enku Densi Fasiledes Kambo gote 273025 1241891 2022 Andesitic basalt 13/02/2014 16 hand pump
DW45 Dangila Misrak Zelesa Sale Egziabhaire 271951 1244264 2055 alluvium and weathered basalt 13/02/2014 0 hand pump
DW46 Dangila Misrak Zelesa Dagnaw Tarekegn's well 271864 1244048 2064 highly weathered red basalt top, yellow bottom 13/02/2014 6.11 197.6 22.5 13 6.6 11.5 Metal sheet
DW47 Dangila Zelesa Wale Tilahun's well 271912 1244201 2057 red soil 13/02/2014 6.54 133.3 21.9 12.5 6.53 pot cover
DW48 Dangila Zeguda Lay Shewaye 270430 1246366 2084 red soil and regolith 14/02/2014 6.22 98.9 23.7 5 hand pump
DW49 Dangila Zeguda Weldehana 271170 1247335 2066 soil and thin regolith 14/02/2014 5.94 88.4 25.2 9 8 hand pump
DW50 Dangila Zeguda Abera Negash's well in Woldehana 271060 1247222 2072 weathered basalt 14/02/2014 697.6 23.1 8.5 7.25 8 Metal sheet
DW51 Dangila Zeguda Kuaja 269811 1248608 2064 regolith 14/02/2014 6.06 114.5 26.2 17 4.5 hand pump
DW52 Dangila Degeshta Kuakurti 269677 1250776 2047 weathered vesicular basalt or regolith? 14/02/2014 6.76 231 25.1 10.5 9 hand pump
DW53 Achefer Weldafecha Mamo Denaenqubar 269062 1251933 2046 red soil, black soil/loamy clay, weathered basalt 14/02/2014 6.86 234 22.1 10 9 hand pump
DW54 Achefer Weldafecha Denaenqubar 269382 1251545 2041 red soil, black soil/loamy clay, weathered basalt 14/02/2014 7.31 322 24.6 6 3 hand pump
DW55 Dangila Zugda Tach Kuaja/Degu Negussie 270226 1249552 2047 Loamy clay soil 14/02/2014 6.13 40.8 21.7 6 1.75 open well
DW56 Dangila Dengesheta Girmaw Malede/Cheba gote 265228 1252431 2057 red soil and weathered basalt 20/02/2014 6.04 182.2 25.4 10 9.5 rope and washer pump
DW57 Dangila Dengesheta Girmaw Malede/Cheba gote red soil and weathered basalt 20/02/2014 6.12 212 23.6 11 10.1 10 open well
DW58 Dangila Dengesheta Ato Arega Wolie/Cheba gote 264608 1252655 2070 regolith, weathered trachy basalt, massive basalt 20/02/2014 6.28 172.5 23.4 5 4.75 under construction
DW59 Dangila Dengesheta Ato Arega Wolie/Cheba gote 264594 1252659 2069 red soil, regolith and weathered trachy basalt 20/02/2014 6.8 37.9 25.2 6.5 5.89 open well
DW60 Dangila Dengesheta Ato Nebretie Abebayehu/Cheba gote 264321 1252387 2067 red soil, regolith and weathered trachy basalt 20/02/2014 6.11 105.6 23.3 6 3.65 4 open well
DW61 Dangila Dengesheta Ato Semahagn Below/Cheba gote 264333 1252397 2075 red and weathered basalt 20/02/2014 6.33 217 22.5 5 3.2 4 open well
DW62 Dangila Dengesheta Ato Semahagn Below/Cheba gote red and weathered basalt 20/02/2014 5.5 2.7 open well
DW63 Dangila Tarra Gebreal/Dengesheta Ato Kassa Wudu (chorka village) 263709 1252406 2075 soil and regolith 20/02/2014 6.44 169.5 26.5 4.5 3 pot cover
DW64 Dangila Tarra Gebreal/Dengesheta W/ro Yayesh Ayinalem 263782 1252437 2074 weathered basalt 20/02/2014 6.36 181 23.1 4.5 3.2 4 pot cover
DW65 Dangila Tarra Gebreal/Dengesheta Ato Abayineh Shawel (Chorka gote) 263541 1252719 2078 weathered basalt 20/02/2014 6.47 107 23.9 5 3.89 open well
DW66 Dangila Tarra Gebreal/Dengesheta Ato Beyene Fekadie(Chorka) 263526 1252707 2080 soil and regolith 20/02/2014 6.46 111.7 22.2 11 4.1 4 Metal sheet
DW67 Dangila Tarra Gebreal/Dengesheta Chorka 263565 1252721 2080 soil and weathered basalt 20/02/2014 6.48 140.4 26 7.5 hand pump
DW68 Dangila Tarra Gebreal/Dengesheta Ato Shibabaw Workneh (Chorka gote) 263577 1252555 2073 weathered trachy basalt 20/02/2014 6.56 124.6 21.6 7.5 3..45 7 open well
DW69 Dangila Tarra Gebreal/Dengesheta Ato Gedefaw Ayalew (Chorka gote) 263358 1252573 2075 regolith and weathered basalt 20/02/2014 6.61 144 23.1 5 3.25 4 pot cover
DW70 Dangila Tarra Gebreal/Dengesheta Ato Kasahun Worku(Abdra gote) 262992 1251801 2092 soil/regolith 20/02/2014 6.78 254 22.7 14 6 pot cover
DW71 Dangila Tarra Gebreal/Dengesheta Ato Necho Anagie (Abadra gote) 262987 1251752 2085 fractured basalt 20/02/2014 6.7 307 23 6.5 4.6 pot cover
DW72 Dangila Tarra Gebreal/Dengesheta Ato Atinkut mulu (Abadra gote) 262338 1252023 2074 weathered basalt 20/02/2014 6.11 177.1 22.7 12.5 8.18 9.5 Metal sheet
DW73 Dangila Dengesheta Ato Melese Worku (Mender 1-Bunteta gote, M-1) 265440 1249776 2091 regolith and weathered trachy basalt 21/02/2014 6.22 132.6 21.5 8 5.75 Metal sheet
DW74 Dangila Dengesheta Mender 1(Bunteta village) 265269 1249847 2087 wethered basalt, with some regolith contribution 21/02/2014 6.08 120.8 23.1 7 5 hand pump
DW75 Dangila Dengesheta Ato Birhanu Shibabaw, Mender 1, M-2) 265056 1250118 2082 fractured basalt with some regolith 21/02/2014 6.84 338 21.7 4.5 3.29 Metal sheet
DW76 Dangila Dengesheta Ato Bazezew Worku(Monitoring well 3) 265588 1250141 2093 contact of weathered basalt and massive basalt 21/02/2014 6.8 280 23.1 9.5 8 Metal sheet
DW76B Dangila Dengesheta Ato Bazezew Worku weathered vesicular basalt 21/02/2014 6.59 192.7 22.1 9.5 8 Metal sheet
DW77 Dangila Dengesheta Ato Getaneh Ayichew (Demekta gote, M-4) 265457 1250733 2075 red soil and weathered basalt 21/02/2014 6.04 104 26.4 11 4.56 Metal sheet
DW78 Dangila Dengesheta Mender 1(Demekta gote) 265360 1250728 2075 Loamy clay soil and basalt 21/02/2014 7.3 381 26.9 0 hand pump
DW79 Dangila Dengesheta Tara Gebreal Primary school 263346 1252081 2088 weathered basalt 21/02/2014 7.35 320 26.7 9 hand pump
DW80 Dangila Dengesheta Ato Tarekegn Tamiru (Abadra gote) 262754 1252201 2084 soil and regolith 21/02/2014 6.34 307 26 9 7.4 7 Metal sheet
DW81 Dangila Dengesheta Ato Degu Ejigu/ Abadra goote 262804 1252184 2093 soil and regolith 21/02/2014 6.21 153.3 23.2 9 7.7 Metal sheet
DW82 Dangila Dengesheta Ato Dessie Sewnet 262802 1252235 2090 red soil, regolith and weathered basalt 21/02/2014 0 7.5 Metal sheet
DW83 Dangila Dengesheta Ato Alehegn Guadie (Abadra gote) 262874 1252330 2091 red soil, weathered andesitic basalt 21/02/2014 6.46 220 24 8.5 7.63 Metal sheet
DW84 Dangila Dengesheta Ato Fenta Guadie 262815 1252071 2090 soil and weathered basalt 21/02/2014 7.5 7.47 Metal sheet
DW85 Dangila Gerargie Ato Alelgn Gebeyehu (Girarge T/Haimanot red soil, regolith and weathered vesicular basalt 21/02/2014 6.77 277 26.5 12.5 6.65 Metal sheet
DW86 Dangila Gerargie Ato Atalay Gebeyehu (Gerargie) 2623306 1250311 2080 weathered vesicular basalt 21/02/2014 6.65 281 23.9 9.5 5 Metal sheet
DW87 Dangila Abadra Tach Mender Abadra Tach Mender 259912 1250977 2075 red soil and regolith 21/02/2014 6.32 110.4 26 7 hand pump
DW88 Dangila Abadra/Hamusit Abadra town 258070 1251989 2096 loamy clay soil, regolith and weathered basalt 21/02/2014 6.52 174.3 25 10 6 hand pump
DW89 Achefer Sebte Guchbigie 272919 1250764 2047 red soil and regolith 03/03/2014 6.05 94.6 23.3 7 5 hand pump
DW90 Achefer Sebte Kes Ajaw 272529 1251633 2041 weathered basalt 03/03/2014 5.92 73.5 22 12.5 6.55 10 pot cover
DW90B Achefer Sebte Kes Ajaw 272550 1251627 2044 weathered basalt 03/03/2014 5.65 52.4 21.3 15 8.5 7 pot cover
DW91 Achefer Gedema Fechito Gedema Fechito 272840 1254285 2035 red soil, regolith and weathered basalt 03/03/2014 6.19 132.8 23.1 16 14 hand pump
DW92 Achefer Gedema Fechito Gedema Fechito 272173 1254395 2016 Loamy clay soil and regolith 03/03/2014 6.17 168.5 22.9 6 2 hand pump
DW93 Achefer Gedema Fechito Gedema Fechito 271664 1254517 2010 red soil and regolith 03/03/2014 5.65 85.4 23.6 7.5 4 hand pump
DW94 Achefer Gedema Dengirsi (Cheba Dure gote) 271406 1254772 2014 weathered basalt 03/03/2014 6.32 222 26.5 14 10 hand pump
DW95 Achefer Gedema Emahoy Zertihun Redie"s well 271539 1254776 2020 regolith and weathered basalt 03/03/2014 6.12 59.4 23.3 16.5 13.75 14 Metal sheet
DW96 Achefer Gedema Mariam Tajorka gote 271046 1255390 2009 regolith and weathered basalt 03/03/2014 0
DW97 Achefer Gedema Tazewarka 271276 1255852 2008 weathered basalt 03/03/2014 6.93 322 25.2 10 5 hand pump
DW98 Achefer Gedema Ato Bazezew 271256 1256196 2021 weathered basalt 03/03/2014 6.39 165.6 23.8 0 11.45 9 pot cover
DW99 Achefer Guta Yetubie Senshaw (Shumbab gote) 269189 1257635 2012 red soil little regolith 03/03/2014 6.23 199.7 24.1 12 6.6 9 pot cover
DW100 Achefer Guta Shumbra gote near DW99 269208 1257630 2011 red soil, regolith and some weathered basalt 03/03/2014 10 8 hand pump
DW101 Achefer Gedema Alemneh Fentie (Tach worka gote) 272387 1256232 2015 soil and regolith 03/03/2014 0
DW102 Achefer Gedema Taje Worka 272847 1256473 2012 red soil and regolith 03/03/2014 6.72 219 27 13 11.5 hand pump
DW103 Achefer Atite Abo Mehal Abo 272921 1266530 1942 black clay loam, regolith mainly weathered basalt 04/03/2014 6.65 232 22.2 5 4 hand pump
DW104 Achefer Atite Abo Ato Azene Guadie/Abalikab gote 272005 1266380 356
1985 soil for this well but the surrounding is basalt 04/03/2014 5.75 148 20.5 12.5 6.6 12 Metal sheet
DW41 flat land Rural village supply domestic supply good supply
DW42 flat land Rural village supply good supply; not used: pollution good supply The well has been opened for measuring water level and polluted and not currently in use
DW43 flat land near grziing land Rural village supply good supply Over 5000 pots are featched every day with no scarcity
DW44 gentle slope grazing land Rural village supply not functional/pump uninstalled not functional currrently due to pump problem
DW45 sloping area grazing land Rural village supply well not functional due to pump problem
DW46 sloping area near house cattle moderate supply Two dug wells close by and varying in two meters depth show some EC variation
DW47 sloping area near house domestic, cattle moderate supply located near house/mule shelter
DW48 flat land domestic moderate supply Scarcity of water during dry period
DW49 flat land near foot of gentle slope domestic supply good supply throughout the year water stored in soil mainly
DW50 flat land near house domestic supply moderate supply
DW51 flat land domestic good supply throught dug in 1991 E.C.
DW52 flat land farm land, 50m from grazing land domestic supply good supply throught Dug in 2000 E.C and sustainable supply all day long throught the year
DW53 gentle slope near swamy area in farm land domestic supply good supply about 3.5m red soil, about 1m black soil/loamy clay and about 5.5m weathered basalt is reported(oral info), dug in 2003 E.C.
DW54 flat land grazing land domestic supply good supply good supply but has bad smell in the rainy period due to the swampy area
DW55 flat land grazing land domestic, cattle good supply Near grazing land and sloping area before well
DW56 gentle slope farm land domestic and chat plantation drops in May-April Depth of weathering around this well is relatively deeper and some of the wells supply good water throught the year
DW57 gentle slope irrigation land/Chat tree irrigation only dries in April 10m soil and one meter weathered basalt
DW58 flat land near house domestic under construction The toilet is 12.5m from the new well and old well is 133m away from new well and toilet is almost at same depth to the new well
DW59 flat land near house domestic and cattle scarcity in May-April The toilet is 12.5m from the new well and old well is 133m away from new well and toilet is almost at same depth to the new well
DW60 flat land near house domestic and cattle Dug before 7 years
DW61 flat land near house irrigation only scarcity in May-April well separation is only 8m and used for irrigating chat plant
DW62 gentle slope irrigation land/Chat tree groundwater flow is towards east
DW63 flat land farm land/grazing land irrigation only good supply top soil is 3m thick and below this is regolith
DW64 domestic Relatively massive basalt at bottom, (dug before 6 years)
DW65 near house domestic Dug in 2003 February and functional
DW66 almost flat land near house cloth washing and cattle water level highly drops in May Aquifer is regolith and water storage recovers though it falls significantly in dry period
DW67 almost flat land grazing land domestic supply good supply Main storage is in the regolith than in basalt or top soil
DW68 almost flat land near house cattle dug in April 2013; top soil is 2m then 1m regolith and 3.5m trachy basalt
DW69 flat land near house domestic and cattle good supply throught Main storage is in weathered basalt
DW70 flat land near house domestic supply high water scarcity As the bottom layer is massive basalt people have difficulty to sunk wells below this depth
DW71 gentle slope near house domestic, cattle supplies better than surrounding wells The aquifer is fractured boulder type basalt and has better yield compared to other wells (1.5m soil, 1.5m regolith and weathered basalt 2.5m
DW72 sloping side near house domestic, cattle good supply good supply though water level significantly drrops (dug in 2004 E.C March)
DW73 flat land near house domestic and cattle good supply dug in May 2004E.C.(red soil 2.65m, then 2m regolith and then 2.35m trachybasalt)
DW74 flat land grazing land domestic supply good but turbid in peak dry period Black soil then regolith and then trachy basalt
DW75 flat land 4m from house domestic, cattle good supply Massive basalt at a depth of 4.5m and wells can't be sunk below(loamy clay soil, regolith, weathered basalt, massive basalt layering)
DW76 flat land near house cooking, cattle better supply Main flow is under varigated colored weathered vesicular basalt at the contact with massive basalt
DW76B flat land near toilet irrigation only better supply
DW77 flat land near house domestic and cattle good potential red soil and weathered basalt at bottom
DW78 flat land grazing land not used because of bad smell good storage, but not fully functional weathered basalt with grey weathering (aquifer), and dug in 2002EC and pump fitted in 2003 E.C., near dry creek
DW79 flat land school compound school water supply better supply The bottom layer is massive basalt and digging was stoped because of massive basalt
DW80 plateau near house domestic supply high water scarcity water well couldn't be sunk below this depth due to massive basalt
DW81 plateau near house domestic, cattle and irrigation high water scarcity well dries in may april, 1m soil 6m regolith and two meter basalt (well log, oral information)
DW82 plateau near house domestic, cattle high water scarcity water scarcity due to limited depth because of massive basalt
DW83 plateau near house domestic, cattle better supply new well (Feb. 2014) and the depth was fairly deep enough, and the aquifer is weathered andesitic basalt
DW84 flat land inside farm land domestic better supply
DW85 flat land near house domestic, cattle better supply the well has better supply and especially wells close to the swampy area
DW86 flat land farm land domestic supply very good supply throught the year the well is over 20 years old but still has great supply
DW87 gentle slope farm land domestic supply better supply
DW88 gentle slope inside small town domestic supply better supply A lot of users/Jericans in line
DW89 flat land near swamy area domestic supply better supply Located near N20W depression
DW90 gentle slope inside irrigation farm irrigation only decreases in May-April water table fluctuation is about 4.55m
DW90B gentle slope inside irrigation farm irrigation only decreases in May-April well separation is 21m
DW91 gentle slope grazing land domestic supply better supply
DW92 flat land near swampy area domestic good supply, but turbid in May The water has bad smell during the wet period ; No massive rock at bottom
DW93 flat land grazing land domestic supply good supply very good supply since digging and good recovery during digining (2002E.C.
DW94 gentle slope grazing land domestic supply very good supply throught the year top 4m is red soil and the rest is weathered basalt since it was dug (May 2003 E.C.)
DW95 gentle slope near house domestic supply very thin weathered basalt seen
DW96 abandoned due to pump problem
DW97 flat land grazing land domestic supply good supply No regolith but loamy clay soil
DW98 gentle slope farm land irrigation only good supply couldn't grow vegetable due to termites which cut the plants at any stage of growth of the plants
DW99 flat land near house domestic and Irrigation good supply Many people in the surrounding area tried to dig but couldn't sunk wells due to massive basalt at shallow depth
DW100 flat land grazing land domestic supply good supply Many people in the surrounding area tried to dig but couldn't sunk wells due to massive basalt at shallow depth
DW101
DW102 flat land grazing land domestic supply good supply Dug in June 2003E.C
DW103 flat land Intermountain depression domestic supply very good supply 357
there is 2msoil, thin layer of gravel and regolith at the creek cut near by
DW104 plateau near house domestic and cattle Shallow water is mainly stored in weathered basalt around this place but in this well the storage is mainly in soil
DW105 Achefer Atite Abo Abalikab 272252 1266509 1970 fractured and weathered basalt 04/03/2014 6.67 285 20.7 7 6 hand pump
DW106 Achefer Atite Abo Engedaw Bezie (Abalikab gote) 272333 1266658 1968 weathered basalt 04/03/2014 6.21 125.3 21.5 7 4.8 4 Metal sheet
DW106B Achefer Atite Abo Engedaw Bezie (Abalikab gote) 272303 1266670 1974 weathered basalt 04/03/2014 7.5 6.2 6 Metal sheet
DW107 Achefer Wulsi Tesfu Minch 274219 1265438 1936 Loamy clay soil and fractured basalt 04/03/2014 6.85 330 24.6 10 8.7 hand pump
DW108 Achefer Wulsi Ato Demeke Atnaw(Tesfu Minch gote) 274228 1265401 1941 loamy clay soil 04/03/2014 5.72 123 23.2 5 2.7 hand pump
DW109 Achefer Nefasa Ashuda Amestya Micheal 272691 1272100 1939 Trachyte 04/03/2014 6.72 259 27 8 5 hand pump
DW110 Achefer Nefasa Ashuda Ato Gerum Atalele (Amestya locality) 272724 1272050 1940 Trachyte 04/03/2014 6 4.05 Metal sheet
DW111 Achefer Nefasa Ashuda Ato Kassa Atalele 272735 1272042 1940 loamy clay soil 04/03/2014 6.2 88.5 23.3 5 3.6 pot cover
DW112 Achefer Nefasa Ashuda Wondi Debelo community well 269749 1270836 2044 weathered basalt 04/03/2014 0 hand pump
DW113 Achefer Nefasa Ashuda Wondi Debelo School supply well 269800 1270485 2063 weathered basalt 04/03/2014 6.43 174.5 22.6 9 almost 9m hand pump
Dw114 Achefer Lalibela Medhanialem Eheri 273098 1272318 1941 loamy silty clay soil and regolith 04/03/2014 6.4 151.1 19.2 9 4.5 hand pump
DW115 Achefer Azena Amede Guma gote 276218 1269822 1900 weathered basalt 04/03/2014 7.49 350 20.3 7.5 6 hand pump
DW116 Achefer Gergista Micheal Gergista 276806 1249357 1986 weathered basalt 04/03/2014 6.45 244 22.4 9 7.5 hand pump
DW117 Achefer Gergista Micheal Kembro 276978 1249136 1991 soil and weathered basalt 04/03/2014 6.4 264 20.5 15 11 hand pump
DW118 Achefer Gergista Micheal Sheferaw Shiti (Kembro Gote) 276621 1249409 1987 red soil 04/03/2014 6.28 102.5 19.7 6 4.12 Metal sheet
DW119 Achefer Sebte Dandie mesk gote 274509 1250559 2003 weathered basalt 05/03/2014 6.7 275 24.2 13 10 hand pump
DW120 Achefer Sebte Ato Semachew Genetie (Tach Sebt) 274604 1250594 2003 weathered basalt 05/03/2014 6.16 132.4 22.2 13 8.85 10 Metal sheet
DW121 Achefer Sebte Ato Mekuanent Amare(Tach Sebt) 274659 1250615 2007 weathered basalt 05/03/2014 6 331 24 12.5 8.3 9.5 Metal sheet
DW122 Achefer Sebte Ato Andualem Worku(Tach Sebt) 274739 1250537 2002 weathered basalt 05/03/2014 6.18 131 23.8 10.5 6.55 Metal sheet
DW123 Achefer Sebte Tach Sebt 274685 1251183 2001 weathered basalt 05/03/2014 6.56 266 26 12.5 10 hand pump
DW124 Achefer Debikan Medhanialem Ato Degu Tayachew (Denka) 281229 1254917 1931 2m black soil +3m regolith 05/03/2014 7.19 227 25 5 4.5 Metal sheet
DW125 Achefer Debikan Medhanialem Bitayita 281359 1255724 1929 loamy clay and regolith 05/03/2014 12.5 11 hand pump
DW126 Achefer Debikan Medhanialem Ato Bekele Degarege(Bitayita gote) 281384 1255775 1932 soil and regolith 05/03/2014 7.24 189 22.3 12.5 5.8 10 Metal sheet
DW127 Dangila Afafe Eyesus Wawi 256420 1256428 2190 red soil and weathered basalt 06/03/2014 0 open well
DW128 Dangila Afafe Eyesus Tach Afafe Eyesus 256313 1255993 2135 fractured basalt 06/03/2014 6.73 273 21.5 7 open well
DW129 Dangila Tach Wawi Ajuri 257967 1256697 2101 alluvium and weathered basalt 06/03/2014 6.46 226 24.2 10 5 hand pump
DW130 Dangila Afafe Eyesus Lay Afafe eyesus 255125 1254862 2249 thin soil and weathered basalt 06/03/2014 6.57 141.8 26 12.5 10.5 hand pump
DW131 Mecha Abro Menore Asana 308402 1244023 2166 weathered basalt 07/03/2014 7.18 553 22.1 14 10 hand pump
DW132 Mecha Abro Menore Asana 308653 1243664 2184 weathered basalt 07/03/2014 7.97 302 25.8 8.5 hand pump
DW133 Mecha Abro Menore Abromenore school 307496 1241242 2373 weathered basalt 07/03/2014 13 12 hand pump
DW134 Mecha Fellegehiwot Debre Mender 306107 1248295 2143 alluvium and weathered basalt 07/03/2014 6.86 209 24 6.5 2.5 hand pump
DW135 Mecha Fellegehiwot Chew Duba 305262 1248163 2155 red soil 07/03/2014 6.82 266 24.5 7 5 hand pump
DW136 Mecha Hulum Selam Ato Yaregal Sheferaw(Gerchech town) 300494 1244287 2060 weathered basalt 07/03/2014 5.95 31.3 23 16 12.6 Metal sheet
DW137 Mecha Hulum Selam Ato Yenework Yayeh(Gerchech) 300492 1244320 2060 weathered basalt 07/03/2014 6.29 61 23.8 16 11.15 13.5 Metal sheet
DW138 Mecha Hulum Selam Ato Yaze Achenefe (Gerchech) 300476 1244381 2060 weathered basalt 07/03/2014 6.15 38 24 19 12.45 Metal sheet
358
DW105 flat land near creek domestic supply reduces in supply during May-April Dug in 1998E.C. The local people complain water table drop due to ecualiptus tree growth
DW106 flat land near house currently for vegetation only
DW106B flat land near house currently for vegetation only Top is slightly-loamy clay then fresh basalt(2.5m) and then deeply weathered basalt
DW107 flat land about 800m away from Kilti river domestic supply good supply flow at soil and rock contact but mainly through fractured basalt
DW108 flat land near hand pump fitted well good supply
DW109 gentle slope domestic supply very low supply There is loam clay soil, little gravel and trachyte in the area. Completely dries in May
DW110 gentle slope domestic supply
DW111 gentle slope domestic supply
DW112 gentle slope/foot of hill grazing land domestic supply very low supply The well was shallow due to massive basalt layer. There is high scarcity of water or many wells failed due to this
DW113 foot of hill/sloping area grazing land school supply very low supply Soil is upto 6m, then thin regolith and then weathered basalt, and fresh rock at a depth of 9m
Dw114 flat land near grazing land domestic supply good supply Black silty loamy clay soil seen in this flat land
DW115 depression near farm land near farm land domestic supply low supply, thin saturated thickness As the well was dug the saturated thickness was 1m and there is high scarcity of water in it
DW116 flat land grazing land domestic supply good supply there is 3.5m thick soil, then regolith and then weathered basalt
DW117 flat land grazing land domestic supply there is massive fresh basalt at the depth of 15m, soil is 3m, 60cm regolith and 2.5m weathered basalt at river sections
DW118 flat land near house domestic supply good supply
DW119 flat land grazing land domestic supply good supply the flow was good at the contact of the weathered and massive basalts, (dug in 2002E.C. January to April)
DW120 flat land near house irrigation only good supply
DW121 flat land near house
DW122 flat land near house irrigation only good supply The basalt is deeply weathered and with brown tint
DW123 gentle slope near stream domestic supply good supply the main flow was along EW direction and along fractures (parallel to EW depression)
DW124 flat land near house domestic supply Almost everyone tried to dig three to four wells but failed because of massive basalt (5 to 6m), and water fluctuation is 4m (oral information of users)
DW125 flat land grazing land domestic supply good supply not functional currrently due to pump problem
DW126 flat land near house domestic and for seedling good supply There are three wells used for seedling of plants for sale
DW127 gentle slope farm land not used yet Under construction, seems to be abandoned
DW128 gentle slope grazing land domestic supply under construction; good storage by night The flow is through fractured of slightly weathered fractured basalt from 6m depth
DW129 foot of hill/sloping area grazing land domestic supply good supply Good storage throught the year, near Ajuri River
DW130 gentle slope near farm land domestic supply low storage in May-April The sloping area has highvariation of water storage depending on slope angle and depth of weathering
DW131 foot of hill/sloping area grazing land domestic supply water in open well stayed for >year before pump installation, used to have bad smell, seepage at soil rock contact; main storage is in weathered basalt
DW132 sloping side near farm land domestic supply good supply There is little contribution of alluvial deposits but main storage is in weathered basalt, dug in 2005E.C.
DW133 small depression in sloping area grazing land domestic supply well is not functional due to pump problem
DW134 foot of hill/sloping area protected area domestic supply Scarcity in May The buffer area for this well is the first kind in the area
DW135 depression, fault related? protected area domestic supply good supply The EC is different from shallow water hosted in soil in the surrounding area
DW136 flat land very near to house(60cm) domestic supply good supply The EC in the area is generally low but very low to this well
DW137 flat land inside the house(kitchen) domestic supply good supply
DW138 flat land very near to house(60cm) domestic supply very good supply Many people in the surrounding area fetch water from this well but the supply is good(good recovery and storage)
359
Table F-2. Boreholes.
ID Woreda Kebele Site name Longitude Latitude Elevation masl Geology Date of visit Field pH EC μS/cm Temp o C
BH1 Guangu/Chagni Tigiri Tigiri Health Center 243381 1226373 1891 Soil and weathered basalt 03/02/2014 7.42 205 25.3
BH2 Dangila Gundri-Ablakena Kuanchinta 265412 1237233 2206 weathered basalt 08/02/2014
BH3 Dangila Gundrie Abo Achirta health center 266847 1237440 2213 basalt and red soil 08/02/2014
BH4 Dangila Gundrie Lay Gundrie 267710 1237463 2213 vesicular basalt 08/02/2014 7.63 322 25.4
BH5 Dangila Ligaba Gilgel badma 277151 1243848 2021 12/02/2014 6.48 156.1 27.5
BH6 Dangila Zeguda Aboyita 268346 1246247 2110 alluvial soil, weathered and fractured basalt 14/02/2014 6.92 238 20.1
BH7 Debub Achefer Sebt Lay Sebt 272432 1251717 2038 weathered basalt 03/03/2014 6.55 254 23.5
BH8 Debub Achefer Debre Tsion Kechinie near Kurbani 278612 1265261 1957 04/03/2014
BH9 Debub Achefer Sebt Tach Sebt 274241 1251221 2017 basalt? 05/03/2014 6.4 165 27
BH10 Debub Achefer Debikan Mariam Chincha 282400 1255478 1926 basalt 05/03/2014 6.91 283
Bh11 Dangila Abadra Medhanialem Godguadit gote 259769 1252085 2069 alluvium and may be basalt 06/03/2014 6.28 332 26
BH12 Mecha Rime town Dima Gote 304790 1250896 2065 red soil and weathered basalt 07/03/2014 6.91 268 25
BH13 Mecha Hulum Selam Gerchech 300559 1244392 2065 basalt and weathered basalt 07/03/2014 6.71 136.8 24.1
ID Depth m Pump Position Pump Topography Land use Use Perennial Remark
BH1 40 hand pump flat land Health center domestic good discharge No protection area
BH2 60 27 hand pump flat land near grazing land domestic new well, not functioning yet
BH3 70 35 hand pump flat land near health center/grazing land new well, not functioning yet
BH4 hand pump flat land near stream bank domestic good discharge Millitery camp and community well
BH5 51? hand pump flat land near stream bank domestic good discharge People are not using it due to bad smell
BH6 hand pump flat land near stream and probably reason not known domestic good discharge Bad smell and bad test reported so people not using
BH7 78 hand pump near Kuchbiye river near stream and probably reason not known domestic
BH8 hand pump flat land near river domestic New well under construction
BH9 100? hand pump gentle slope grazing land domestic good discharge Data to be collected later if possible
BH10 70 hand pump flat land near swampy area close to Gilgel Abay
Bh11 132? moterized flat land near river and swampy area domestic artesian well Not functioning due to running cost problem(fuel)
BH12 hand pump flat land farm land domestic good discharge well data not fully found
BH13 95 16? hand pump flat land center of town town supply good discharge pump tested for 72hrs?(oral info), drilled in 2002EC
360
Table F-3. Springs.
ID Woreda Kebele Site name Longitude Latitude Elevation masl Geology Date of visit Field pH EC μS/cm Temp o C Flow/yield l/s Measuring method Topography
CS1 Dangila Alefa-Kacha Barbash 238051 1230275 1793 boulder type olivine basalt and silty clay soil 03/02/2014 5.56 91.2 22.2 about 0.05 Estimation flat land
CS2 Guangu/Chagni Seragam Micheal Sholal gote 240465 1228133 1821 Highly fractured and jointed basalt + red soil 03/02/2014 5.28 40 21.9 about 0.03 Estimation flat land
CS3 Dangila Kabita Sihai/Deber gote 254945 1246434 2207 weathered amygdaloidal basalt 04/02/2014 6.69 300 18.9 about 0.01 Estimation sloping area, in riverbank
CS4 Dangila Washa Kulkul 254940 1245733 2198 red soil (aquifer), weathered basalt in the surrounding 04/02/2014 8.16 74.5 27 about 0.01 Estimation sloping area, in streambank
CS4 Dangila Washa and Kabilta Agashti 252908 1247159 2103 >4m soil thickness and moderately weathered andesitic basalt 04/02/2014 7.73 224 23.3 0.84 floating method sloping area sliding soil mass
CS5 Dangila Dimsa Emadadi 256241 1243768 2213 andesitic basalt under 4m soil 05/02/2014 6.42 119.8 21.9 about 0.01 Estimation sloping area
CS6 Dangila Washa Medhanialem Tach Yihun gote 254712 1244066 2214 andesitic basalt with minor iron rich basalt 05/02/2014 7.42 242 17.6 about 0.01 Estimation sloping area
CS7 Dangila Dimsa Kanabari/Lamami 257249 1245757 2183 Weathered basalt, amygdaloidal basalt and soil 05/02/2014 7.83 310 18.2 0.5 floating method sloping area
CS8 Dangila Badani Saguma 249425 1241057 1900 alluvial soil 07/02/2014 905 25.6 very low flat land
CS9 Dangila Badani Akuacha 250368 1241676 1924 alluvial material 07/02/2014 524 23.3 about 0.01 Estimation gentle slope
CS10 Dangila Badani Warkit 250903 1236661 1928 red soil (aquifer), basalt in the surrounding 07/02/2014 206 26.6 0.090909091 volumetric flat land
CS11 Dangila Senguri Zerihun Minch 250799 1236196 1923 fractured basalt 07/02/2014 155.6 23.3 about 1 Estimation riverbank
CS12 Dangila Sehara lunk/Astuta gote 248488 1237421 1886 fractured basalt and regolith? 07/02/2014 220 22.9 total flow is over 30 floating method Topographic depression
CS13 Dangila Senguri Dengel 247412 1237304 1868 red soil 07/02/2014 206 23.8 about 3 Estimation Topographic depression
CS14 Dangila Badani Embura/Gizani gote 249831 1236850 1917 Soil and Vesicular basalt 07/02/2014 206 23.4 about 1 Estimation flat land
CS15 Dangila Kuandisha Gezewtie 264708 1238299 2190 Soil and Vesicular basalt
CS16 Dangila Kuandisha Bogalech Mersha/Gezewetie 265238 1237676 2207 Soil and Vesicular basalt 08/02/2014 6.35 106.2 23.2 seepage flat land
CS17 Dangila Gundri-Ablakena Yaba Tegegn Minch (Guachinta) 265342 1237359 2211 weathered basalt and soil 08/02/2014 6.2 152.1 21.6 below 0.1 Estimation flat land
CS18 Dangila Abla Mariam Buna Wuha 263399 1236148 2281 weathered basalt,weathered thickness is over 5m 08/02/2014 6.76 210 21.7 0.071428571 volumetric sloping side
CS19 Dangila Abla Mariam Mariam Wuha 263299 1236779 2283 weathered trachybasalt 08/02/2014 7.61 164.6 18 0.75 floating method sloping side
CS20 Dangila Gayita Georgis Workit Georgis Tsebel 270520 1237079 2163 boulder type vesicular basalt and top soil 08/02/2014 6.54 190.4 20.4 about 1.2 Estimation sloping side
CS21 Dangila Gayita Georgis Workit Domestic spring 270541 1237061 2162 boulder type vesicular basalt and top soil 08/02/2014 6.79 200 19.7 4.3 floating method sloping side
CS22 Dangila Gayita Georgis Minchit/Bambuit 271038 1237805 2149 basaltic regolith and soil 08/02/2014 6.1 113 18.1 about 2 Estimation Topographic depression
CS23 Dangila Lay Gayita Dokmit 269972 1237517 2157 Soil and Vesicular basalt 08/02/2014 6.13 108 23 1 volumetric Topographic depression
CS24 Dangila Gayita Georgis Yashina Micheal holy Water 270082 1236750 2172 vesicular basalt 08/02/2014 6.57 198.3 20.3 about 1.5 Estimation Topographic depression
CS25 Dangila Gayita Georgis Warda Gebit/Ashola Micheal gote 268826 1235330 2253 Basaltic regolith 10/02/2014 6.33 121.8 20.1 about 0.5 Estimation Topographic depression
CS26 Fageta Lekuma Tafoch Dambul Azarama(Macha) 268113 1234334 2281 silty loamy-clay soil with little regolith 10/02/2014 6.59 189.2 20.5 about 3 floating method depression
CS27 Fageta Lekuma Makia Teklehaymanot Godguadit Agegnehush 268221 1232018 2336 thin regolith and weathered vesicular basalt 10/02/2014 6.13 99.7 20.8 about 0.3 Estimation streambank depression
CS28 Fageta Lekuma Shangani Kechisi(Birzana Spring) 263806 1230039 2390 soil and basalt 10/02/2014
CS29 Fageta Lekuma Shangani Dabula spring 263575 1230141 2393 soil and basalt 10/02/2014
CS30 Fageta Lekuma Gula Azmach Kimkima 269450 1228190 2393 contact of weathered basalt and trachyte and along EW fracture 11/02/2014 5.94 60.4 17.2 0.2 Estimation Topographic depression
CS31 Fageta Lekuma Girayita Mariam Wuha 274209 1228955 2336 soil 11/02/2014 6.4 105.4 20.4 below 0.1 Estimation Topographic depression
CS32 Fageta Lekuma Sigla Yohanes Aba Drey. Ashebrite gote 279957 1227054 2375 boulder type basalt under soil 11/02/2014 6.92 85.6 19.7 0.25 volumetric sloping side
CS33 Fageta Lekuma Kuri Jegola Besena 279941 1224254 2452 weathered basalt 11/02/2014 6.65 43.3 25 seepage Topographic depression
CS34 Dangila Wumbre Yesesayitu Minch 278930 1237040 2118 thin regolith and weathered basalt 11/02/2014 6.26 104.6 19.9 about 0.7 Estimation riverbank
CS35 Dangila Wumbre Yaba Semeneh Minch 279090 1236724 2121 fractured and jointed basalt 12/02/2014 6.01 81 22.6 0.21 volumetric riverbank
CS36 Dangila Muksi Selamargi Mariam 277787 1239116 2085 soil and weathered basalt 12/02/2014 5.79 55 28? 2.83 volumetric Topographic depression
CS37 Dangila Zelesa Kes Mender 270117 1243849 2123 regolith? Weathered basalt 13/02/2014 6.77 380 18.1 2 volumetric Topographic depression
CS38 Dangila Zunga Yabagashe Minch 272876 1241213 2023 regolith? Weathered basalt 13/02/2014 6.73 194.1 21.6 0.125 volumetric sloping side
CS39 Dangila Zunga Gedema 272624 1240367 2024 andesitic basalt and aphanitic olivine basalt undifferentiated 13/02/2014 6.51 183 23.7 0.31 volumetric sloping side
CS40 Dangila Misrak Zelesa Sale Egziabhair 272096 1244283 2050 Scoraceous basalt under about 2m soil thickness 13/02/2014 7.51 280 27.9 about 0.01 Estimation sloping side
CS41 Dangila Zelesa Karmarie 272136 1244621 2058 black cotton soil + scoraceous basalt under the soil 13/02/2014 7.44 367 25.5 0.058823529 volumetric sloping side
CS42 Dangila Zeguda Thankisti/Aboyta gote 268522 1246029 2111 emerges from regolith 14/02/2014 6 49.7 23.5 0.333333333 volumetric flat land
CS43 Dangila Zeguda Lay Shewaye 270493 1246381 2088 emerges from soil at riverbank 14/02/2014 5.99 110.6 24.7 about 0.1 Estimation riverbank
CS44 Dangila Dengesheta Ceba gote near new metre.station 265035 1252291 2051 fractured vesicular basalt 20/02/2014 6.16 188.8 24.5 almost seepage Estimation riverbank
CS45 Dangila Chereka minch/Abadra gote 262483 1252024 2063 fractured and jointed basalt 20/02/2014 6.33 128.3 21.6 seepage depression
CS46 Achefer Atite Abo Aba Lika 272149 1266394 1984 loamy clay soil 04/03/2014 6.1 106.3 19 very low Estimation sloping side
CS47 Achefer Sebte Woynwuha (lay Sebte) 274071 1250601 2020 emerges from regolith and weathered basalt 05/03/2014 7.63 133 27 low flow (below 0.01 ) Estimation depression
CS48 Achefer Abchikili Zuria Jirfit 273763 1257658 2029 regolith 05/03/2014 low flow depression
CS49 Dangila Tach Afafe Eyesus Senabo 256914 1255568 2110 foliated basalt 06/03/2014 6.29 101.6 23 low flow/about 0.001 Estimation riverbank
CS50 Dangila Tach Afafe Eyesus Minchitie 257004 1256037 2132 emerges from soil rock contact 06/03/2014 6.95 139.8 22.6 about 0.2 Estimation sloping side
CS51 Dangila Lay Afafe Eyesus Senabo 256112 1256369 2169 emerges from regolith but there is soil and boulder basalt around 06/03/2014 6.15 93 22.7 about 0.3 Estimation sloping side
CS52 Mecha Abromenore Sosna Georgis holy water 308683 1243630 2186 alluvial material 07/03/2014 7.22 400 22.2 no flow sloping side
CS53 Mecha Abromenore Ment(Genbo sub-kebele) 307420 1242491 2307 weathered and fractured trachy basalt 07/03/2014 6.52 174.2 22.4 low low fracture spring
361
ID Fractures Land use Mode of Emergence Use Perennial Remark
CS1 EW grazing land and vegetation Fracture and depression/flat land Domestic and cattle Seasonal since the last 3y The spring starts drying in the dry period due to deforestation and plantation of Ecualiptus tree
CS2 Ew and NS Vegetation and grazing land Fracture and depression/streambed irrigation at downstream perennial vegetation density is good and emerges at junction of EW and NS fractures
CS3 emerges along NS fault vegetation along streambed, farms by riverbank Fracture and depression/streambed cattle perennial dense vegetation along streambank
CS4 farm land, grazing land topographic depression cattle perennial soil thickness is over 3m and spring not developed, not protected
CS4 farm land, grazing land topographic depression cattle and drinking, not protected perennial Emerges at the contact of soil and rock, multiple eye, causes landslide
CS5 N30W and EW near farm land along fractures cattle and some times domestic perennial emerges from fractured rock mainly along N30W and some along EW
CS6 N30W seepage and N20E fractures seen fracture and topogrphic break cattle, minor irrigation and domestic perennial
CS7 EW fracture seen grazing land and vegetation contact Domestic and cattle perennial
CS8 grazing land and vegetation topographic depression Domestic and cattle perennial The yield reduced, or dries in April due to deforestation / population density (local ealderly people)
CS9 farm land, grazing land contact Domestic and cattle perennial The spring emergence depth has been lowered by about 30 to 40 cm away from its orgional level some years back
CS10 grazing land and near swampy area topographic depression Domestic and cattle perennial Yield increases downstream from spring eye
CS11 near farm land contact of soil and fractured basalt domestic supply perennial the water in the river is used for irrigation
CS12 N50W (main flow and N50E minor flow) grazing land ang vegetation fractured and contact town supply, irrigation and cattle perennial It almost forms small stream
CS13 grazing land and vegetation topographic depression Domestic and cattle perennial
CS14 grazing land topographic depression domestic, cattle, irrigation perennial
CS15
CS16 grazing land topographic depression domestic, cattle perennial possibly polluted by No3 from anaimal dung
CS17 grazing land topographic depression Domestic and cattle perennial It was dried in 2013 or 2005 E.C. Dry period
CS18 grazing land and vegetation topographic depression domostic, cattle perennial Depth of weathering is over 5 meter
CS19 N50E church land and vegetation/forest topographic break irrigation, drinking, preying water perennial
CS20 forest topographic break holy water, domestic and irrigation downstream perennial The flow at downstream is high
CS21 forest topographic break domestic and irrigation perennial The flow at downstream is high
CS22 grazing land just at margion of irrigation land topographic break demestic, irrigation, cattle perennial Yield highly increases during rainy period
CS23 grazing land and some vegetation topographic break Domestic and cattle perennial Poor development and pipes are closed and leaks under asunery work
CS24 grazing land and vegetation topographic break holy water, domestic and irrigation downstream perennial Spring eye divided for domestic and holy water
CS25 grazing land and vegetation topographic break domestic and irrigation perennial soil thickness is 1.5m on top of regolith
CS26 grazing land topographic break irrigation, domostic and cattle perennial Yield highly increases during rainy period
CS27 streambank near farm land topographic break domestic, cattle perennial Dries around mid of March, emerges from the contact of regolith and Weathered basalt
CS28 Not measured because of its high turbidity
CS29
CS30 EW grazing land topographic break seedling and for domestic and cattle perennial The spring flow is along bedding joints of weathered basalt and along EW fractures at the contact
CS31 near farm land topographic break Domestic and cattle perennial reddish loamy soil with over 1.5m visible soil thickness
CS32 foot of plateau topographic break domestic and irrigation perennial Local people constructed temporary storage for irrigation
CS33 foot of plateau topographic break cattle perennial Debuki spring which was developed in 1990EC is totally dry due to water table drop/and/or development
CS34 streambank near grazing land contact domestic perennial there is seepage between soil and massive basalt but locally limmited regolith and joints have higher flow
CS35 grazing land fracture domestic perennial flow is along the vertical joint but the horizontal thinner joints are also connected with it
CS36 near river bed topographic break domestic, cattle and irrigation perennial
CS37 topographic break domestic, cattle and irrigation perennial
CS38 grazing land sloping side/topographic break domestic, cattle perennial development of spring just finshed on this date
CS39 grazing land sloping side/topographic break domestic perennial development of spring just finshed on this date
CS40 grazing land contact/ depression domestic perennial yield is low as soil thickness is thin
CS41 grazing land topographic break domestic perennial development was in 2005 E.C. but with very low discharge
CS42 grazing land topographic break domestic, cattle perennial Developed and local people report emergence point shrink by about 30m and emerges from regolith.
CS43 grazing land topographic break
CS44 grazing land topographic break domestic supply perennial emeerges just above the relatively massive vesicular basalt
CS45 N20E, N70W, EW junction (emergence point) and N50W grazing land topographic break domestic supply and cattle perennial emerges at junction point of fractures
CS46 near farm land sloping side/topographic break domestic and irrigation perennial turbid water and used mainly for coffee plantation
CS47 grazing land topographic break doemstic, cattle perennial flow increases downstream of the creek after small seepages along the creek
CS48 forest Along EW depression not used except cattle perennial Not measured because of its high turbidity
CS49 NS foliation,EW and NS fractures, NS is younger riverbank along NS foliation domestic perennial The top red soil is 2.5m thick, Near Senabo stream
CS50 near farm land topographic break domestic perennial Not developed, not fenced. Multiple eye and emerges fromcontact of soil and moderately weathered basalt
CS51 near farm land topographic break domesic supply and little irrigation perennial Not developed, not fenced. Multiple eye and emerges from regolith below soil horizon
CS52 protected land, near creek topographic break holy water perennial it is kind of water hole and no low this time
CS53 EW and NS fractures, NS younger and low is along NS farm land, sloping area sloping side/topographic break not intensively used perennial microfractures controll low and storage but yield little
362
Table F-4. Rivers.
ID Woreda Kebele Site name Longitude Latitude Elevation masl Geology Date of visit Field pH EC μS/cm Temp o C Flow/yield l/s Measuring method
RW1 Dangila Alefa-Kacha Awsi River near Tiski fall 237656 1229863 1779 Olivine basalt slightly weathered from top 03/02/2014 8.16 234 18.3 about 130 Estimation
RW2 Gizani Alefa-Kacha Gizani River (Barbash gote,near Tiski fall) 237734 1229844 1778 Olivine basalt slightly weathered from top 03/02/2014 8.19 117.7 20.1 1.73m3/s floating
RW3 Dangila Washa and Kabilita Fench wuha 252903 1246494 2139 weathered basalt, chonchoidal weathering 04/02/2014 8.21 290 20.3 1.3 Estimation
RW4 Dangila Dimsa Awsa Arka 256547 1245547 2134 amygdaloidal basalt 04/02/2014 7.75 255 23.1 about 1.5 Estimation
RW5 Dangila Dimsa Yaba Yenew shete(stream) 254751 1244691 2205 moderately weathered basalt 05/02/2014 7.15 265 22.3 about 1 Estimation
RW6 Dangila Warkit Kilti at Deki gote 264591 1255464 2029 fresh slightly fractured basalt and soil 06/02/2014 7.98 218 22.4 about 1.5 m3/s floating
RW7 Fageta Lekuma Girayita Muger River 272957 1228851 2332 massive basalt with some joints and minor faults 11/02/2014 7.58 121.2 18.2 about 35 Estimation
RW8 Fageta Lekuma Aswa Fenzit Zuma 276908 1227475 2339 weathered trachybasalt at margin and basalt at River 11/02/2014 7.33 93 23 about 25 Estimation
RW9 Dangila Zelesa-Ligaba Ashare 274212 1244121 2003 basalt 13/02/2014 7.65 169.1 26.9 over 1.8m3/s Estimation
RW10 Dangila Diversion Quasheni 272664 1240223 2024 Andesitic basalt 13/02/2014 8.02 195.4 21.9 about 100 Estimation
RW10 B Dangila Enku Densi Fasiledes Kamo gote 273290 1242819 2007 alluvial soil 13/02/2014 7.44 300 25.1 about 40 Estimation
RW11 Dangila Dengesheta Branti river at New Gauge site 265107 1252291 2050 vesicular basalt 20/02/2014 8.25 256 21.5 about 5 Estimation
RW12 Debub Achefer Dekuli Kilty River 268938 1258318 2006 fresh fine grained massive basalt 03/03/2014 7.61 253 22.3 over 1.5m3/s Estimation
RW13 Debub Achefer Atibara 277039 1249117 1985 massive basalt with some joints 04/03/2014 7.54 154.6 17.6 about 4 Estimation
RW14 Debub Achefer Sebte Ashare river at sebte 275739 1250594 1974 massive basalt with some joints 04/03/2014 7.9 180 21.3 Estimation
RW15 Dangila Tach Wawi Ajuri River 257928 1256840 2099 2.5msoil,0.5m regolith, weathered basalt bottom unexposed 06/03/2014 7.21 131.6 21.9 about 35 Estimation
RW15B Dangila Abadra Ajuri River 259730 1252090 2065 alluvial soil 06/03/2014 7.84 168.4 30 about 60 Estimation
RW16 Mecha Abromenore Koga River 309328 1243278 2171 weathered basalt, stratiied and deeply weathered 07/03/2014 8.02 205 19.3 about 180 Estimation
RW17 Mecha Abromenore Asanat 308004 1240354 2451 deeply weathered basalt at top, bottom is massive basalt 07/03/2014 7.48 178 20.3 about 1.5 Estimation
ID Topography Fractures Land use Bank width Status Use Remark

RW1 flat land near ridge escarp N40E and shallow joints grazing land and vegetation about 30m perennial, turbid Cattle Depth of weathering is between 0.5 to 1.2m
RW2 flat land near ridge escarp N40E grazing land and vegetation about 30m perennial cattle at this site Depth of weathering varies between 2 to 12.5m
RW3 sloppy area N60E and EW fractures farm land 4m perennial cattle and rarely irrigation depth of weathering is between 0.5 to 1m and
RW4 depression N30W and Ew minor fractures vegetation 15m perennial cattle N60E, N30W and NS fractures seen at down stream
RW5 sloppy area EW & NS fractures farm land 3m perennial cattle, irrigation, rarely domestic exposed weathered section is 2.3m at this spot
RW6 flat land grazing land, farm land, vegetation 13m perennial cattle, irrigation, rarely domestic few farmers are irrigating their land using motor pumps
RW7 gentle slope EW, N50E, N30E grazing land about 50m perennial irrigation, cattle
RW8 gentle slope near farm land about 30m perennial irrigation, cattle
RW9 flat land grazing land about 35m perennial irrigation, cattle it is beeing diverted by local people
RW10 flat land grazing land abut 35m perennial irrigation, cattle The river is also diverted by local people from down stream
RW10 B flat land irrigation area 12m perennial irrigation, cattle & domestic Local diversions are made in many places
RW11 flat land near gentle slope grazing land 9.04m perennial irrigation, cattle Slightly jointed vesicular basalt under 1.5m soil cover
RW12 flat land grazing land 50m perennial irrigation, cattle, mill riverbank is 1.4m; surrounding people complain that massive rock hinders well diging
RW13 20m perennial irrigation, cattle, mill
RW14 flat land grazing land 25m perennial irrigation, cattle
RW15 depression grazing land 8m perennial irrigation, cattle exposed section depth is 3.5m
RW15B flat land, deep erosion and landslide grazing land 25m perennial irrigation, cattle The loamy clay soil thickness is obout 4m at river bank
RW16 depression N30W and Ew minor fractures river valley, farm land on sides 31m perennial irrigation, domestic, cattle The river is damed at down stream
RW17 sloppy area N30W and Ew minor fractures 7m perennial irrigation,domestic, sanitation Part of the river segiment is controlled by N30W vertical fracture pattern
363
The information in Tables F-5, F-6 and F-7 is from surveys conducted by David Walker in
February and March 2014. All locations are within Dangila woreda unless stated. Coordinates
are in the WGS84 coordinate system, taken from a handheld GPS along with elevation.
Measurements in italics were received word of mouth. Note that an uppercase ID, e.g. DW73,
denotes a location originally surveyed by Demis Alamirew whereas lower case, e.g. dw6,
indicates a location surveyed only by David Walker.
364
Table F-5. Hand-dug wells and boreholes.
ID (DW = Demis ID) Kebele Site name N E Elevation masl Geology (italics = Demis survey ) Date of visit Pumping test Sample 222Rn test Field pH EC μS/cm Temp o C Depth m SWL mbgl
DW76 Dangesheta Ato Bazezew Worku (MW 5) 11o18.184' 036o51.195' 2090 weathered basalt 19/03/2015 8.44 5.98
dw1 Dangesheta 11o17.985' o
036 51.200' 2091 quite friable fractured vesicular basalt below gravelly soil 19/03/2015 7 5.5
036 51.176' 2091 quite friable fractured vesicular basalt below gravelly soil 19/03/2015 7 5.5
DW73 Dangesheta Ato Melese Worku (MW 2) 11o17.984' o
036 51.120' 2091 regolith and weathered trachyte basalt 19/03/2015 Yes (both visits) SI5 C5 A5 5.83 130.7 20.3 7.03 5.77
DW75 Dangesheta Ato Birhanu Shibabaw (MW 3) 11o18.168' o
036 50.907' 2086 fractured basalt with some regolith 19/03/2015 4.16 3.94
DW77 Dangesheta Ato Getaneh Ayichew (MW 4) 11o18.504' o
036 51.125' 2075 red soil and weathered basalt 19/03/2015 Yes (both visits) 5.28 90.74 22.2 9.19 5.34
dw3 Dangesheta Ato Assaye Molla (MW 1) 11o18.051' o
036 51.268' 2094 19/03/2015 6 5.95
DW43 Zelesa Kilaje 11o15.191' o
036 52.936' 2137 red and black loamy soil intercalation 21/03/2015 SI1 C1 A1 5.19 50.62 20.9 17 2
dw4 Zelesa Kilaje 11o15.246' o
036 52.350' 2136 21/03/2015 SI2 C2 A2 5.84 99.7 20.8 18
dw5 Zeguda Tankishti 11o16.085' o
036 52.734' 2119 gravelly alluvium regolith 22/03/2015 dry
DW57 Dangesheta Girmaw Malede/Cheba gote 11o19.435' o
036 50.999' 2061 red soil and weathered basalt 23/03/2015 11 8.41
DW56 Dangesheta Girmaw Malede/Cheba gote 11o19.424' 036o50.992' 2052 red soil and weathered basalt 23/03/2015 SI4 C4 A4 5.53 171.9 22.9 10 9.5
dw6 Dangesheta Ato Birhanu Shibabaw 11o18.169' 036o50.899' 2082 regolith 24/03/2015 Yes (both visits) SI6 C6 A6 5.57 144.2 22.3 3.55 2.71
DW61 Tara Gabriel Ato Semahagn Abebayehu 11o19.398' 036o50.504' 2070 regolith 25/03/2015 Yes (first visit) 5.95 216.6 22.1 4.12 3.58
DW62 Tara Gabriel Ato Semahagn Abebayehu 11o19.406' 036o50.504' 2070 regolith 25/03/2015 4.17 3.75
dw7 Tara Gabriel Ato Getenet Birehanu 11o19.429' 036o49.755' 2089 regolith 25/03/2015 Yes (first visit) 5.97 149.9 23.6 8.75 7.91
DW79 Tara Gabriel Tara Gabriel School 11o19.227' 036o49.959' 2081 weathered basalt 25/03/2015 SI7-I8 C7-9 A7-9 6.88 334.9 25.2 10
dw8 Sahara Gisa agricultural office 11o12.039' 036o40.067' 1894 black alluvium 26/03/2015 1.3 1
DW18 Badani Delelti 11o12.568' 036o42.610' 1930 massive basalt boulders in black clayey alluvium 26/03/2015 SI11 C11 A11 6.66 481.9 25.9 4 2
dw9 Kwakurta Chara restaurant 11o11.551' 036o45.558' 1975 regolith 26/03/2015 7.7
DW2 Washa Amognita 11o15.763' 036o45.351' 2216 red regolith 26/03/2015 SI12 C12 A12 5.76 200.4 21.9 6 4
dw10 Wondayita Worranty 11o15.103' 036o47.937' 2111 26/03/2015 SI13 C13 A13 5.69 44.29 22.1 8.39
DW21 Kuandisha Gezewetie 11o11.761' 036o50.766' 2200 loamy-silty soil and weathered basalt 27/03/2015 SI14 C14 A14 6.17 174 24.7 11 6
dw11 Kuandisha 11o11.500' 036o50.269' 2218 27/03/2015 6.91
DW22 Abla Mariam Avila 11o10.990' 036o50.442' 2233 red regolith 27/03/2015 SI15 C15 A15 6.31 264.4 24 11 4.5
DW130 Afafe Eyesus Lay Afafe eyesus 11o20.705' 036o45.429' 2251 thin soil and weathered basalt 28/03/2015 12.5 10.5
dw12 Afafe Eyesus Ayesheshem Chakul 11o20.595' 036o45.317' 2267 massive basalt boulders in regolith 28/03/2015 5.64 87.06 19.8 17 12
dw13 Afafe Eyesus Desalign Abyu 11o20.628' 036o45.372' 2255 massive basalt boulders in regolith 28/03/2015 16 14
dw14 Afafe Eyesus Asmara Sonet 11o20.437' 036o45.878' 2195 28/03/2015 5.51 136.7 20.5 14 11.8
DW88 Abadra Abadra town 11o19.154' o
036 47.062' 2091 loamy clay soil, regolith and weathered basalt 28/03/2015 10 6
dw15 Abadra Mangudit 11o19.680' o
036 46.827' 2107 28/03/2015 SI16 C16 A16 5.69 196.8 23.4 8.5
dw16 Dangesheta Girma 11o19.518' o
036 51.343' 2055 regolith 30/03/2015 10.2 9.77
036 51.335' 2055 regolith 30/03/2015 5.59 145.5 21.9 10.39
036 51.342' 2055 regolith 30/03/2015 5.46 133.1 21.3
dw19 Dangesheta Prest Getay 11o19.424' o
036 51.543' 2047 30/03/2015 5.45 166.5 22.2
dw20 Dangesheta Ebenew 11o19.297' o
036 51.710' 2045 red regolith over friable weathered basalt 30/03/2015 5.74 215.5 21.2 3.5 2.4
dw21 Dangesheta Asheshum 11o19.289' o
036 51.721' 2043 red regolith over friable weathered basalt 30/03/2015 5 3
dw22 Dangesheta Mobit Mulu 11o18.638' o
036 51.993' 2063 alluvium and clayey regolith 30/03/2015 4.29 3.35
dw23 Dangesheta Mobit Mulu 11o18.631' o
036 51.991' 2063 alluvium and clayey regolith 30/03/2015
dw24 Dangesheta Wondifro Taye 11o18.638' o
036 51.922' 2077 alluvium and regolith 30/03/2015 4.53 3.33
dw25 Dangesheta Wondifro Taye 11o18.633' 036o51.921' 2077 alluvium and regolith 30/03/2015
dw26 Dangesheta Kindu Asmano 11o18.678' o
036 51.903' 2059 regolith 30/03/2015 4.47 3.73
036 51.913' 2059 regolith 30/03/2015
dw28 Dangila 11o17.074' o
036 49.536' 2082 31/03/2015 6.56 231.4 21.6
dw29 Abadra 11o16.542' o
036 47.886' 2103 31/03/2015 6.49
dw30 Abadra Berayta 11o16.174' o
036 48.068' 2099 31/03/2015 SI18 C18 A18 5.99 309.4 21.9
365
ID (DW = Demis ID) Pump/Cover Topography Land use Use Perennial Remark
DW76 pot cover flat land on edge of floodplain near house between crops, floodplain/pasture and eucalyptus domestic and cattle perennial
dw1 open well flat land on floodplain floodplain pasture will not be used dug in March 2015, 1.5 m diameter, will not be used due to unstable sides, "easy to dig"
dw2 open well flat land on floodplain floodplain pasture will be for potable supply dug in March 2015, 1.5 m diameter, will be used for potable supply, "easy to dig", dug 10 m from previous after that one abandoned
DW73 oil drum cover flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial
DW75 oil drum (no cover) flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial
DW77 oil drum cover fairly flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial
dw3 wood and branches flat land on edge of floodplain near house between pasture and eucalyptus plantations "rarely used because dirty water" seasonal beside house of man who monitors wells - raingauge is also beside his house. Well almost dry but higher level in rainy season.
DW43 hand pump flat land on floodplain floodplain pasture village potable supply perennial dug 19 years ago, always good supply
dw4 hand pump flat land on floodplain floodplain pasture village potable supply inconsistent water very turbid (took about 6 filters to take samples), "sometimes no supply - it was once dry for over a year", only ~400 m from previous well in similar position
dw5 hand pump flat land on floodplain floodplain pasture "contaminated so just used for irrigation" seasonal currently dry so locals use spring ~100 m upstream
DW57 pot cover gentle slope above floodplain chat and coffee irrigation seasonal
DW56 rope and washer pump gentle slope above floodplain chat and coffee potable and irrigation perennial very good supply all year. Water very clear - took no effort to push samples through filter.
dw6 pot cover flat land on edge of floodplain near house between crops and floodplain/pasture irrigation perennial other side of house to MW3. Pump tested this one as too little water in MW3.
DW61 pot cover flat land on edge of floodplain in chat plantation between house and floodplain/pasture irrigation perennial 8 m south from next well in same plot. Adjacent to "hanging" floodplain above quite steep Brante valley. Close to catchment boundary.
DW62 wood and branches flat land on edge of floodplain in chat plantation between house and floodplain/pasture irrigation perennial there is another well 10 m west and many more in nighbouring plots. Some become dry in April/May.
dw7 pot cover slope on hillside near house within crops and eucalyptus plantations domestic and cattle perennial part way down long sloping valley side of Kilti river - not near any floodplain/pasture
DW79 hand pump flat land on edge of floodplain in school grounds between floodplain/pasture and crops school domestic use perennial high, near to catchment boundary. Sampled in triplicate to send 1x to Addis lab, 1x blind to Addis lab, and 1x to UK.
dw8 open well fairly flat land on edge of floodplain floodplain pasture cattle perennial very shallow water table, 5 m diameter open well for farmers to water cattle
DW18 hand pump sloping floodplain at base of large steep hill boulder-strewn floodplain pasture village potable supply perennial locals don't have their own wells as sides collapse
dw9 oil drum cover flat land in town beside restaurant restaurant domestic supply perennial
DW2 hand pump slope on hillside between crops and floodplain/pasture village potable supply perennial high in hills, ~20 m from trickling stream
dw10 rope and washer pump slope on hillside crops irrigation perennial cover has gaps for contamination, visibly only slightly turbid but took many filters to sample
DW21 hand pump flat land on edge of floodplain floodplain pasture domestic perennial compound locked but locals told us it is well used. Filter discoloured red but easy to sample and not turbid.
dw11 pot cover slight slope at base of hill near house within crops domestic perennial
DW22 hand pump flat floodplain at base of steep hill small pasture between nearby river and houses/crops domestic perennial
DW130 broken hand pump slope on hillside crops pump broken
dw12 oil drum cover slope on hillside near house within crops domestic perennial "very good supply" but quite near catchment boundary (crest of big cliff)
dw13 pot cover slope on hillside near house within crops domestic perennial ~100 m downslope from previous well
dw14 pot cover slope on hillside near house within crops domestic perennial
DW88 broken hand pump fairly flat land on edge of floodplain near houses in small town pump broken well W of town near schools broken handpump, handpump in town N of river contaminated, most of town supply now from motorised pump in deep borehole E of town
dw15 hand pump gentle slope on hillside small pasture within crops on edge of Abadra town domestic perennial pump locked with hours restricted to a few in the morning and again at end of day; "to prevent kids breaking pump"
dw16 pot cover and pulley gentle slope coffee, chat, mango, banana, orange, onion irrigation perennial these three wells are all in same plot within 10 m of each other. Solid rock was not struck during excavation.
dw17 pot cover and pulley gentle slope coffee, chat, mango, banana, orange, onion irrigation perennial
dw18 rope and washer pump gentle slope beside house and coffee, chat, mango, banana, orange, onion irrigation and domestic perennial IWMI installed pump in 2014
dw19 rope and washer pump flat land on edge of floodplain in house plot between crops and floodplain/pasture irrigation and domestic perennial IWMI installed pump in 2014
dw20 pot cover and pulley flat land on edge of floodplain beside house between crops and floodplain/pasture irrigation and domestic perennial
dw21 open well flat land on edge of floodplain between crops and floodplain/pasture under construction ~15 m from previous well at other side of house. Spoil still visible - had to chisel through basalt layer - and intend to dig a bit deeper. 1.5 m diameter.
dw22 wood and branches flat land on edge of floodplain between crops and floodplain/pasture irrigation and domestic perennial dug two weeks ago. In same plot as next well.
dw23 rope and washer pump flat land on edge of floodplain between crops and floodplain/pasture irrigation and domestic perennial IWMI installed pump in 2014
dw24 wood and branches, treadle pump flat land on edge of floodplain between crops and floodplain/pasture irrigation perennial on opposite side of floodplain to previous well
dw25 rope and washer pump flat land on edge of floodplain beside house within crops irrigation and domestic perennial IWMI installed pump in 2014. ~7 m from previous well in same plot.
dw26 open well with pulley flat land on edge of floodplain crops irrigation perennial ~20 m from next well in plot of adjacent house
dw27 rope and washer pump flat land on edge of floodplain beside house within crops irrigation and domestic perennial IWMI installed pump in 2014
dw28 hand pump flat land on floodplain pasture domestic perennial
dw29 hand pump half installed gentle slope on edge of large floodplain crops under construction concrete lined, all infrastructure installed except pump headworks
dw30 hand pump gentle slope at base of large hill crops domestic perennial heavy use, bolt broken in pump so not working well, filter turned deep red but easy to sample
366
ID (DW = Demis ID) Kebele Site name N E Elevation masl Geology (italics = Demis survey ) Date of visit Pumping test Sample 222Rn test Field pH EC μS/cm Temp o C Depth m SWL mbgl
dw3 Dangesheta Ato Assaye Molla (MW 1) 11o18.051' 036o51.268' 2094 19/03/2015 6 3.81
DW76 Dangesheta Ato Bazezew Worku (MW 5) 11o18.184' 036o51.195' 2090 weathered basalt 19/03/2015 8.44 4.91
DW56 Dangesheta Girmaw Malede/Cheba gote 11o19.424' 036o50.992' 2052 red soil and weathered basalt 10/10/2015 2SI1 2C1 2A1 5.61 167.7 22.9 9 5.5
dw21 Dangesheta Asheshum 11o19.289' 036o51.721' 2043 red regolith over friable weathered basalt 10/10/2015 2SI2 2C2 2A2 6.71 403.5 30.9 6 1
DTW3 Agaga DTW3 deep BH 11o16.905' 036o48.558' 2076 alluvium 12/10/2015 2SI4 2C4 2A4 8.77 315.0 22.0 150
DTW1 Berayta DTW1 deep BH 11o16.762' 036o48.561' 2074 alluvium 12/10/2015 192
DTW4 Berayta DTW4 deep BH 11o16.491' 036o48.551' 2073 alluvium 12/10/2015 150
dw30 Berayta Berayta 11o16.174' 036o48.068' 2099 12/10/2015
dw31 Agaga Agaga 11o16.961' 036o48.115' 2082 alluvium 12/10/2015 2SI5 2C5 2A5 5.55 136.4 21.8 ? ?
D3 Dangila D3 deep BH 11o15.889' o
036 50.917' 2107 12/10/2015 2SI6 2C6 2A6 Yes 8.81 335.0 21.8 130
DW43 Zelesa Kilaje 11o15.191' o
036 52.936' 2137 red and black loamy soil intercalation 13/10/2015 2SI8 2C8 2A8 4.77 46.26 20.5 18 ?
DW22 Abra Mariam Avila 11o10.990' o
036 50.442' 2233 red regolith 14/10/2015 2SI10 2C10 2A10 5.46 133.3 21.7 18 ?
DW21 Kuandisha Gezewetie 11o11.761' o
036 50.766' 2200 loamy-silty soil and weathered basalt 14/10/2015 2SI11 2C11 2A11 5.19 167.1 22.1 ? ?
036 51.342' 2055 regolith 15/10/2015 2SI14
DW73 Dangesheta Ato Melese Worku (MW 2) 11o17.984' o
036 51.120' 2091 regolith and weathered trachyte basalt 16/10/2015 Yes (both visits) 2SI12 2C12 2A12 5.30 110.9 20.9 6.89 4.27
DW75 Dangesheta Ato Birhanu Shibabaw (MW 3) 11o18.168' o
036 50.907' 2086 fractured basalt with some regolith 16/10/2015 4.18 1.52
dw6 Dangesheta Ato Birhanu Shibabaw 11o18.169' o
036 50.899' 2082 regolith 16/10/2015 Yes (both visits) 2SI13 2C13 2A13 5.41 133.8 22.2 3.41 1.42
DW77 Dangesheta Ato Getaneh Ayichew (MW 4) 11o18.504' o
036 51.125' 2075 red soil and weathered basalt 16/10/2015 Yes (both visits) 2SI15 5.24 146.6 22.3 9.17 2.83
dw2 Dangesheta Nr Asaye New 11o17.986' o
036 51.176' 2091 friable fractured vesicular basalt below gravelly soil 17/10/2015 2SI16 Yes 6.43 289.0 22.5 7 5.5
dw32 Sehara Little Asaye 11o11.117' o
036 41.831' 1842 regolith 18/10/2015 Yes (second visit) 5.42 207.4 23.6 10.09 7.4
dw33 Kwakurta Selam 11o11.430' o
036 44.952' 1990 regolith 18/10/2015 Yes (second visit) 5.82 196.5 23.0 5.91 2.78
DW79 Dangesheta Tara Gabriel School 11o19.227' o
036 49.959' 2081 weathered basalt 21/10/2015 2SI28 2A28 2C28 6.59 231.3 24.7 10
ID (DW = Demis ID) Pump/Cover Topography Land use Use Perennial Remark
dw3 wood and branches flat land on edge of floodplain near house between pasture and eucalyptus plantations "rarely used because dirty water" seasonal
DW76 pot cover flat land on edge of floodplain near house between crops, floodplain/pasture and eucalyptus plantations
domestic and cattle perennial
DW56 rope and washer pump gentle slope above floodplain chat and coffee potable and irrigation perennial very good supply all year. Water very clear - took more effort to filter than March/April.
dw21 hand pump flat land on edge of floodplain between crops and floodplain/pasture potable and domestic perennial was under construction when I was last here, now complete. Very shallow water table (half a pump and water flowed)
DTW3 electric submersible very large flat floodplain pasture (near Kilti/Amen junction) Dangila town supply perennial Drilled in 2009, 150 m deep, open hole, pump at 60 m, Q = 20 l/s, operates 10h/d
DTW1 electric submersible very large flat floodplain pasture (near Kilti) Dangila town supply perennial Drilled in 2009, 192 m deep, open hole, pump at 60 m, Q = 32 l/s, operates 10h/d
DTW4 electric submersible very large flat floodplain pasture (near Kilti) Dangila town supply (not operational) - Drilled in 2009, 150 m deep, open hole, pump at 60 m, Q = 0 l/s, electrical problem so not operational
dw30 hand pump broken gentle slope at base of large hill crops not operational - pump broken; could not re-sample, locals using Kilti river. Apparently school nearby also has HDW but pump is also broken
dw31 hand pump flat land on edge of floodplain between crops and floodplain/pasture potable and domestic perennial Sampled here because it is the closest to broken Berayta HDW for re-sample (though far away), also closest to deep BHs.
D3 electric submersible fairly flat land on edge of town near small river (Fincha) in trees Dangila town supply perennial Drilled 1985, 130 m deep, open hole, pump at 100 m, Q = 3.5 l/s, operates 10h/d. D1 and D2 close but yield decreased till they non-functional
DW43 hand pump flat land on floodplain floodplain pasture village potable supply perennial dug 1992, always good supply
DW22 hand pump flat floodplain at base of steep hill small pasture between nearby river and houses/crops domestic perennial "sometimes bad smell". Was very difficult to filter though not turbid - 3x filters per 125ml bottle.
DW21 hand pump flat land on edge of floodplain floodplain pasture domestic perennial compound no longer locked
dw18 rope and washer pump gentle slope beside house and coffee, chat, mango, banana, orange, onion irrigation and domestic perennial IWMI installed pump in 2014
DW73 oil drum cover flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial Repeat well test now greater saturated thickness
DW75 oil drum (no cover) flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial
dw6 pot cover flat land on edge of floodplain near house between crops and floodplain/pasture irrigation perennial Repeat well test now greater saturated thickness
DW77 oil drum cover fairly flat land on edge of floodplain near house between crops and floodplain/pasture domestic and cattle perennial Repeat well test now greater saturated thickness
dw2 hand pump flat land on floodplain floodplain pasture potable and domestic perennial had just been dug in March 2015, 1.5 m diameter
dw32 plastic drum cover flat land beside house surrounded by crops rarely used (cattle) perennial house is closest to Lunk spring which has very good water so they use that
dw33 oil drum cover flat land beside house surrounded by crops domestic perennial near to town (Chara)
DW79 hand pump flat land on edge of floodplain in school grounds between floodplain/pasture and crops school domestic use perennial not locked as it was during last visit
367
Table F-6. Springs.
ID (CS = Demis ID) Kebele Site name N E Elevation masl Geology Date of visit Sample Field pH EC μS/cm Temp o C Flow/yield Topography
cs1 Minchinat joins Brante 11o18.174' 036o51.599' 2076 regolith 22/03/2015 5.22 61.93 22.8 0.1 l/s slightly sloping floodplain
cs2 11o17.742' 036o51.656' 2085 gravelly alluvium regolith above massive basalt 22/03/2015 steeper sloping narrower valley with narrow floodplain
cs3 11o17.692' 036o51.681' 2084 gravelly alluvium regolith above massive basalt 22/03/2015 steeper sloping narrower valley with deep dry gullies
cs4 11o17.350' 036o52.189' 2110 regolith 22/03/2015 deep gully in floodplain high above main channel
CS42 Zeguda Tankishti/Aboyta gote 11o15.967' 036o52.829' 2118 gravelly alluvium regolith 22/03/2015 SI3 C3 A3 5.31 48.67 22.1 1 l/s flat floodplain
CS45 Chereka minch/Abadra gote 11o19.191' o
036 49.485' 2071 fractured and jointed basalt 23/03/2015 shallow valley sloping towards Kilti gauge
CS12 Sehara Lunk/Astuta gote 11o11.217' o
036 41.861' 1896 vesicular basalt boulders in regolith 26/03/2015 SI10 C10 A10 5.98 217.6 22 gullies forming at topographic break in flat land
CS4 Washa Kulkul 11o15.748' o
036 45.283' 2222 red regolith 26/03/2015 gully in narrow sloping floodplain
CS19 Abla Mariam Mariam Wuha 11o10.879' o
036 49.925' 2275 very deep weathered basalt 27/03/2015 deep gully in hillside
cs5 Dangesheta 11o19.020' o
036 51.978' 2048 alluvium 30/03/2015 flat floodplain
cs6 Dangesheta Kote Labeles 11o18.711' o
036 51.960' 2068 clayey alluvium 30/03/2015 SI17 C17 A17 6.09 189.6 22.2 0.25 l/s flat floodplain
cs7 Dangesheta 11o18.995' 036o50.505' 2075 alluvium 02/04/2015 edge of flat floodplain
cs8 Dangesheta 11o19.432' 036o51.721' 2043 alluvium 10/10/2015 5.22 195 22.1 1 l/s halfway down floodplain
cs6 Dangesheta Kote Labeles 11o18.711' 036o51.960' 2068 clayey alluvium 10/10/2015 2SI3 2C3 2A3 6.0 188.7 22.9 4 l/s flat floodplain
CS42 Zeguda Tankishti/Aboyta gote 11o15.967' 036o52.829' 2118 gravelly alluvium regolith 13/10/2015 2SI9 2C9 2A9 4.77 58.44 22.2 high flat floodplain
CS12 Sehara Lunk/Astuta gote 11o11.217' 036o41.861' 1896 vesicular basalt boulders in regolith 18/10/2015 2SI18-19 2C18-19 2A18-19 5.98 225.3 22.8 very high gullies forming at topographic break in flat land
cs1 Minchinat 11o18.174' 036o51.599' 2076 regolith 20/10/2015 5.36 66.35 20.3 0.5 l/s slightly sloping floodplain
cs9 Workit Workit 11o19.407' 036o51.573' 2036 alluvium above massive basalt 22/10/2015 2SI29 5.75 124.8 23.2 0.5 l/s downstream end of large floodplain
cs10 Dangesheta Brante SB 11o19.331' 036o50.993' 2054 alluvium over fractured basalt 22/10/2015 2SI30 6.11 171.7 23.5 0.5 l/s small floodplain
ID (CS = Demis ID) Land use Mode of Emergence Use Perennial Remark
cs1 pasture with cropland within 30 m of river contact of gravelly alluvium - more solid regolith minor use but potable perennial
cs2 pasture, forest on steeper sloping E bank contact of gravelly alluvium regolith - massive basalt none perennial many springs and seepages in west bank as basalt outcrops in river bed
cs3 pasture, forest on steeper sloping E bank contact of gravelly alluvium regolith - massive basalt none perennial spring ~15 m up floodplain side, between this spring and previous are many springs and seepages in W bank as basalt outcrops in river bed
cs4 cropland topographic break none perennial spring ~30 m up floodplain side where main channel cuts >10 m
CS42 pasture water table in floodplain centre domestic and cattle perennial developed with 2 of 4 pipes flowing and open concrete tank
CS45 pasture water table at bottom of valley and fractures seasonal spring and gully dry and in a visibly drier area
CS12 pasture and forest topographic break Gisa and Chara towns' supply perennial several springs with high flow form quite large stream. Spring developed and piped to tanks.
CS4 pasture water table at bottom of valley cattle perennial "more flow and additional springs in wet season"
CS19 forested church land water table at bottom of gully domestic, cattle and irrigation perennial many seepages emerging from undergrowth forming trickling stream, currently low flow but higher in wet season
cs5 pasture water table in floodplain centre cattle perennial many seepages form stream that flows to Brante
cs6 pasture water table in floodplain centre domestic perennial developed with 2 of 4 pipes flowing from enclosed tank and open concrete tank. More springs and seepages ~30 m away form stream that flows to Brante
cs7 pasture topographic break domestic and cattle perennial many seepages form stream that dries up on floodplain
cs8 pasture water table intercepts floodplain none seasonal many such springs and seepages on most Dangesheta floodplains
cs6 pasture water table in floodplain centre domestic perennial developed with 2 pipes flowing from enclosed tank and open concrete tank. Many more springs and seepages all over small floodplain with high flow
CS42 pasture water table in floodplain centre domestic and cattle perennial pipes and open concrete tank submerged due to spring and stream flow
CS12 pasture and forest topographic break Gisa and Chara towns' supply perennial Very easy to filter. Several springs with very high flow form large stream. Developed; piped to tanks (overflowing into stream) then piped to elevated tanks for town supply
cs1 pasture with cropland within 30 m of river contact of gravelly alluvium - more solid regolith minor use but potable perennial numerous "eyes" all with low flow and very clear water
cs9 pasture contact of alluvium and massive basalt cattle perennial numerous seepages across fairly large area - just minor seepages in dry season
cs10 pasture contact of alluvium and fractured basalt cattle seasonal numerous seepages
368
Table F-7. Rivers and other features.
ID Woreda Kebele Site name N E Elevation masl Geology Date of visit Sample 222Rn pH EC Temp Topography
RFL1 Dangila Dangesheta Bridge over dry stream 11o18.466' 036o50.993' 2075 dry, cracked, gravelly alluvium 19/03/2015 flat floodplain
RFL2 Dangila Dangesheta BDU weather station at Dangesheta Agricultural Development Office 11o18.561' 036o51.068' 2080 19/03/2015 on shallow slope up from floodplain
RFL3 Dangila Dangesheta Ato Assaye Molla - Community monitored rain gauge 11o18.041' 036o51.251' 2094 19/03/2015 flat land on edge of floodplain
RFL4 Dangila Dangesheta Dangeshta Service Cooperative 11o18.237' 036o51.142' 2111 19/03/2015 higher land between two floodplains
RFL5 Merawi Kolga Dam 11o20.758' 037o08.530' 2028 19/03/2015 shallow valley
RFL6 Dangila Dangesheta East of Dangeshta Service Cooperative 11o18.403' 036o51.023' 2076 22/03/2015 bottom of slope at floodplain
RFL7 Dangila Dangesheta Minchinat Stream joins Brante River 11o18.174' 036o51.599' 2076 regolith 22/03/2015 sloping floodplain
RFL8 Dangila Dangesheta Point where Minchinat currently emerges 11o18.174' 036o51.629' 2072 regolith 22/03/2015 sloping floodplain
RFL9 Dangila 11o17.692' 036o51.681' 2084 gravelly alluvium regolith above massive basalt 22/03/2015 steep sloping narrower valley, deep dry gullies
RFL10 Dangila Source of Brante tributary 11o17.279' 036o52.255' 2112 regolith 22/03/2015 deep gully
RFL11 Dangila 11o17.310' 036o51.955' 2118 22/03/2015 sloping valley sides
RFL12 Dangila Brante River road bridge 11o16.572' 036o51.847' 2111 gravelly alluvium and massive basalt boulders 22/03/2015 flat floodplain
RFL13 Dangila 11o16.538' 036o52.059' 2114 fractured vesicular basalt boulder field 22/03/2015 flat floodplain
RFL14 Dangila Zeguda Tankishti 11o16.085' 036o52.734' 2119 gravelly alluvium regolith 22/03/2015 flat floodplain
RFL15 Dangila Dangesheta Brante gauge 11o19.350' 036o50.924' 2050 massive, fractured and vesicular basalt boulders 23/03/2015 gently sloping floodplain
RFL16 Dangila Dangesheta Kilti gauge 11o19.290' 036o49.417' 2045 massive and fractured basalt boulders 23/03/2015 quite steep valley sides
RFL17 Dangila Dangila Amen gauge 11o15.774' 036o50.647' 2106 clayey alluvium 23/03/2015 flat floodplain
RFL18 Dangila Dangila Dangila weather station 11o15.050' 036o50.749' 2105 24/03/2015 flat open field
RFL19 Dangila Sehara Gizani River 11o10.779' 036o41.895' 1903 massive basalt boulders 26/03/2015 flat floodplain
RFL20 Dangila Sehara Gisa agricultural office 11o12.039' 036o40.067' 1894 black alluvium 26/03/2015 flat wetland floodplain
RFL21 Dangila Awsi weir 11o12.072' 036o39.998' 1858 black alluvium, massive basalt boulders in red regolith 26/03/2015 flat wetland floodplain
RFL22 Dangila Awsi River 11o12.309' 036o42.843' 1937 massive basalt boulders in black alluvium 26/03/2015 dry flat floodplain
RFL23 Dangila Amen River? 11o11.668' 036o50.420' 2200 regolith 27/03/2015 flat floodplain
RFL24 Dangila Abadra Gagie River road bridge 11o19.472' 036o46.894' 2080 black alluvium above red regolith 28/03/2015 small floodplain, quite steep valley sides in town
RFL25 Dangila Dangesheta Brante River 11o19.395' 036o51.574' 2047 black clayey alluvium, red regolith, vesicular jointed basalt 30/03/2015 flat floodplain
RFL26 Dangila Dangesheta Tributary enters Brante (currently dry) 11o19.216' 036o51.915' 2044 red regolith and massive basalt boulders 30/03/2015 flat floodplain
RFL27 Durbete Durbete weather station 11o21.545' 036o57.376' 1990 30/03/2015 flat, on slight rise
RFL28 Dangila Amen River road bridge 11o16.245' 036o50.100' 2084 clayey alluvium above red regolith 31/03/2015 flat floodplain
RFL29 Dangila Tributary enters Amen (currently dry) 11o16.915' 036o48.863' 2072 alluvium 31/03/2015 very large flat floodplain
RFL30 Dangila Tributary enters Amen (currently dry) 11o16.881' 036o48.655' 2074 alluvium 31/03/2015 very large flat floodplain
RFL31 Dangila Amen joins Kilti River 11o16.858' 036o48.480' 2070 gravelly sandy alluvium 31/03/2015 very large flat floodplain
RFL32 Dangila Tributary of Kilti 11o16.435' 036o47.969' 2095 clayey alluvium above red regolith, basalt cobbles in bed 31/03/2015 deep gully in slope above floodplain
RFL33 Dangila Tributary of Kilti 11o16.346' 036o47.972' 2096 clayey alluvium 31/03/2015 deep gully in slope above floodplain
RFL34 Dangila Kilti River 11o16.194' 036o48.447' 2083 sandy gravelly pebbly alluvium 31/03/2015 very large flat floodplain
RFL35 Dangila Tributary of Kilti meets large floodplain 11o16.059' 036o49.024' 2091 clayey alluvium and massive basalt boulders 31/03/2015 small floodplain meets large flat floodplain
RFL36 Bahir Dar Bahir Dar National Meteorology Office and weather station 11o35.985' 037o21.602' 1801 06/04/2015 fenced compound within flat field
RFL37 Meshenti Meshenti rain gauge 11o28.263' 037o17.152' 1963 06/04/2015 very small fenced compound within flat field
369
ID Land use Depth m Bank width m Incision m Remark
RFL1 pasture dry 3 1 bridge over dry stream, "wetland in rainy season"
RFL2 next to cropland weather station installed by Abdu and Debebe on 18-3-15
RFL3 between pasture and eucalyptus plantation small tree (3-4 m high) 4 m to NW
RFL4 where crops are weighed and sold; within eucalyptus this where Assaye Molla works
RFL5 large resevoir behind dam and irrigation canals Koga Dam Irrigation Project on Abay River. Finished in 2011/2012, paid for by MoA and Africa Bank. 1750ha resevoir, 1730x21m dam, 9.1m3/s release for irrigation: 7000ha for 14000 people, >1m3/s released: Abay River
RFL6 limit of eucalyptus plantations before pasture all high ground around coop office is eucalyptus
RFL7 pasture with cropland within 30 m of river 0.2 5 2 note there is a spring 10 m upstream of channel intersection (see springs tab)
RFL8 cropland and seedling nurseries immediately adjacent dry 3 3 upstream (dry) the river is highly sinuous and cuts ~4.5 m deep
RFL9 pasture, forest on steeper sloping east bank 0.3 10 2 upstream the channel incision is 5-10 m and much deep gullying (dry) in adjacent sloping floodplain from where springs emerge in wet season
RFL10 cropland and forest 0.1 8 8 source of Brante tributary (I thought this was Brante source because most of downstream Brante flow comes from this tributary)
RFL11 cropland and eucalyptus plantations crossroads of eucalyptus plantations on high ground to SE and cropland between here and river to west
RFL12 pasture and forest dry 6 1.5 small pool under bridge but otherwise Brante is dry. Occasional stagnant pools upstream.
RFL13 pasture 0.05 5 1 extensive boulder field within very large floodplain. Occasional flowing sections within Brante as well as dry sections and stagnant pools.
RFL14 pasture 0.2 2 0.5 Tankishti river fed by spring (see spring tab) but doesn't reach Brante at the moment. In rainy season the whole Brante/Tankishti floodplain floods and locals must use asphalt road to reach Dangila.
RFL15 pasture then houses and cropland ~30 m from river 0.15 5 1 Brante gauge in position as originally placed and still sturdy. Measure twice daily (6am and 6pm). Wet season floods just overtop bank.
RFL16 pasture on north bank, acacia forest on south bank 0.3 15 1.5 Kilti gauge broken after being hit by a floating tree in wet season. Metal supports and upper board snapped. River stage measured twice daily (6am and 6pm) by dipping measuring stick in river.
RFL17 pasture/football pitch/builders yard 0.2 7 2 Lower part of gauge a bit bent. Upstream through town the river is very dirty and full of litter - here it's quite clear.
RFL18 fenced open field weather station actually ~25 m NW of the coordinates. Tall eucalyptus trees ~30 m to north. Visited at 18:30 so nobody present to let me into compound or answer questions.
RFL19 pasture 0.25 20 2 Quite high flow over boulders in bed
RFL20 pasture perennial wetland (one of several) between town of Gisa and Awsi river
RFL21 pasture 1 12 1.5 very turbid upstream of weir - probably raising water table forming wetlands. Downstream is shallower (0.25 m). Weir built in 2014. Two canals (left bank = 50 l/s) irrigate 100 ha.
RFL22 none dry 18 2 however, further upstream east of Chara there is water in the Awsi where it is much smaller
RFL23 pasture and crops 0.05 4 2.5 I would have expected river here to be flowing south but this goes north. Where to?? Small aqueduct (currently dry) irrigates small plot. (Looking at Google Maps later: I think this is the Amen River)
RFL24 pasture, crops and forest 0.25 10 3 quite fast so substantial flow even though it’s shallow. Some abstraction below bridge.
RFL25 pasture 0.2 8 1.5 cattle watered here
RFL26 pasture dry 8 2 several tributaries such as this enter through large floodplain - most currently dry
RFL27 fenced compound within school field weather station within Durbete Primary School, large tree ~15 m north and another tall tree ~ 15 m east
RFL28 pasture 0.1 7 2.5 almost no flow in river
RFL29 pasture 0.1 6 1.5 between here and previous location there are reaches of almost zero flow and reaches of basalt bed. In large floodplain; water table visible in bank at ~1.3 mbgl.
RFL30 pasture 0.25 3.5 1.5 between here and previous location there are dry reaches and reaches of stagnant water
RFL31 pasture 0.1 8 1 fast flow in Kilti, no flow in Amen
RFL32 pasture and crops 0.1 18 8 some flow that dries up as the channel reaches the flat floodplain
RFL33 pasture and eucalyptus 0.05 18 8 some flow (less than previous) that dries up as the channel reaches the flat floodplain
RFL34 pasture 0.2 8 2.5 flowing
RFL35 pasture and crops 0.05 3 1.5 some flow that dries up as the channel reaches the flat floodplain and dries up ~ 100 m upstream
RFL36 weather station and car park nearby pylon and many overhead cables. Small hill ~150 m to the south.
RFL37 unused land around Kebele office and clinic no nearby trees or other obstructions
370
ID Woreda Kebele Site name N E Elevation masl Geology Date of visit Sample 222Rn test field pH EC μS/cm Temp o C
RFL3 Dangila Dangesheta Ato Assaye Molla - Community monitored rain gauge 11o18.041' 036o51.251' 2094 10/10/2015
RFL15 Dangila Dangesheta Brante gauge 11o19.350' o
036 50.924' 2050 massive fractured vesicular basalt boulders 10/10/2015
RFL38 Dangila Dangesheta Abdu/Debebe Brante gauge 11o19.260' o
036 51.876' 2038 alluvium 10/10/2015
RFL39 Dangila Dangila Hahu Hotel Roof 11o15.476' o
036 50.832' 2048 12/10/2015 2SI7 2A7 2A23
RFL40 Dangila Dangila Dangila Water Supply Service Office 11o15.918' 036o51.067' 2109 12/10/2015
RFL41 Wetet Abay Wetet Abay electronic met. station 11o22.026' 036o02.264' 1896 13/10/2015
RFL42 Dangila Dangesheta Brante River road bridge and Debebe river gauge 11o16.572' 036o51.847' 2111 gravelly alluvium, massive basalt boulders 15/10/2015
RFL43 Dangila Dangesheta Flow gauging location (Brante) 11o16.591' 036o51.818' 2108 gravelly alluvium, massive basalt boulders 15/10/2015
RFL44 Dangila Dangesheta Flow gauging location (Brante) 11o19.338' o
036 50.926' 2053 massive fractured vesicular basalt boulders 15/10/2015
RFL45 Dangila Dangesheta Flow gauging location (Brante) 11o19.253' o
036 51.891' 2040 sandy alluvium 15/10/2015
RFL46 Dangila Dangesheta Nr Malese (Dangesheta floodplain stream) 11o18.055' o
036 51.043' 2053 alluvium 17/10/2015 2SI17 Yes 6.83 99.35 21.8
RFL47 Dangila Sehara Gisa and Chara collection tank at Lunk 11o11.239' 036o41.813' 1893 vesicular basalt boulders in regolith 18/10/2015
RFL48 Dangila Sehara Gizani River 11o10.799' 036o41.874' 1901 massive basalt boulders 18/10/2015 2SI20 6.39 106.6 23.1
RFL49 Dangila Sehara Dinkeresh (wetland on Gizani floodplain) 11o10.782' 036o41.910' 1906 alluvium 18/10/2015 2SI21 6.71 256.1 30.9
RFL50 Dangila Sehara Gisa 11o11.164' 036o40.262' 1879 18/10/2015 2SI22 2A22
RFL51 Dangila Dangesheta Nr Brahanu (stream at Dangesheta floodplains' neck) 11o18.348' o
036 50.895' 2063? alluvium 20/10/2015 2SI24 Yes 6.84 83.26 20.3
RFL52 Dangila Dangesheta Nr Getaneh (Dangesheta floodplain stream) 11o18.478' o
036 51.113' 2076? alluvium 20/10/2015 2SI25 2A25 2C25 Yes 6.04 70.28 26.6
RFL53 Dangila Dangesheta Bunteta (Brante river) 11o18.613' 036o51.250' 2079? alluvium 20/10/2015 2SI26 Yes 6.18 91.1 23.1
RFL54 Dangila Dangesheta Brante 11o18.468' 036o51.348' 2075? alluvium 20/10/2015 Yes 6.22 93 23.4
RFL55 Dangila Dangesheta Mandar 2 (Brante river) 11o18.184' 036o51.554' 2076? alluvium, basalt boulders 20/10/2015 2SI27 Yes 6.27 93.45 22.3
RFL16 Dangila Dangesheta Kilti gauge 11o19.290' 036o49.417' 2045 massive and fractured basalt boulders 21/10/2015
RFL56 Dangila Dangesheta Flow gauging location (Kilti) 11o19.292' 036o49.410' 2042? massive and fractured basalt boulders 21/10/2015
RFL57 Dangila Dangesheta Bridge over large wetland stream 11o17.095' o
036 51.210' 2102 alluvium 21/10/2015
RFL58 Dangila Workit Workit (bridge over Brante) 11o19.390' 036o51.611' 2018? alluvium over massive basalt 22/10/2015 Yes 6.26 109.6 21.3
RFL59 Dangila Dangesheta 11o19.191' o
036 51.413' 2045 alluvium 22/10/2015 Yes 6.25 108.8 22.7
RFL60 Dangila Dangesheta Brante SB 11o19.331' 036o50.993' 2054 alluvium over fractured basalt 22/10/2015 Yes 6.25 105.9 23.7
RFL61 Dangila Dangesheta Brante Gorge 11o19.158' 036o50.681' 2061 regolith over fractured basalt 22/10/2015 2SI31 2A31 2C31 Yes 6.29 104.8 22.9
RFL62 Dangila Dangesheta 11o18.991' 036o50.639' 2064 alluvium and basalt boulders 22/10/2015 Yes 6.16 103.4 22.4
RFL63 Dangila Agaga Amen Floodplain 11o17.042' 036o48.984' 2015? alluvium 24/10/2015 Yes 6.13 127.5 18.9
RFL64 Dangila Agaga Kilti Bridge 11o17.130' o
036 48.620' 2067 gravelly alluvium 24/10/2015 Yes 7.03 134 19.8
RFL65 Dangila Agaga Amen by DTW3 11o16.889' 036o48.568' 2068 alluvium 24/10/2015 Yes 7.12 122.5 21.2
RFL66 Dangila Berayta Kilti by DTW1 11o16.741' o
036 48.548' 2074 gravelly alluvium 24/10/2015 Yes 7.37 166.8 23.9
RFL67 Dangila Berayta Kilti Floodplain 11o16.307' 036o48.565' 2081 alluvium 24/10/2015 Yes 7.43 168.7 27.2
371
ID Topography Land use Depth m Bank width m Incision m Remark
RFL3 flat land on edge of floodplain between pasture and eucalyptus plantation No issues to report regarding rain gauge. Protective fence has gone. Small tree still 3-4 m high 4 m to NW.
RFL15 gently sloping floodplain pasture. Houses and cropland ~30 m from river 0.3 7 1.5 No issues regarding Brante gauge. Gauge in deepest part of flow, 1m from bank, 2m from bank top and 5m from other side bank top.
RFL38 quite narrow floodplain unused, crops 15-25 m back from bank top 0.4 8 2 Also a stage board on top of bank for floods (didn't flood in 2015)
RFL39 flat town centre town Rain from Hahu Hotel roof during big storm from ~midnight to ~5am morning of 12-10-15
RFL40 flat edge of town town Very helpful chaps at water supply office
RFL41 beside road bridge at base of quite steep slope trees, road, river Some trees quite close to gauge and also at base of a fairly steep bank sloping to river and road
RFL42 flat floodplain pasture and forest 0.7 6 1.5 Debebe's gauge (monitored by a farmer) is on the parapet on the upstream left bank
RFL43 flat floodplain pasture 0.3 12 1.5 Flow gauging location 30 m downstream of road bridge
RFL44 gently sloping floodplain pasture. Houses and cropland ~30 m from river 0.36 5.5 1 Flow gauging location 30 m downstream of NCL/IWMI Brante gauge
RFL45 quite narrow floodplain unused, crops 15-25 m back from bank top 0.49 6 1.5 Flow gauging location 30 m downstream of Debebe/Abdu Brante gauge
RFL46 centre of flat floodplain pasture 0.1 2 0.4 222Rn sample/test location. Sampled surface water but very close to numerous springs/seepages.
RFL47 gullies forming at topographic break in flat land pasture and forest Collection tank ~80 m d/s of Lunk springs overflowing to stream. Water piped to elevated tanks closer to Gisa and Chara for towns' supply
RFL48 flat floodplain pasture 0.5 20 1.5 High flow. Steep gradient - pool and drop.
RFL49 flat floodplain pasture 0.04 ~50 0 Wetland/floodplain sample location. Forms small stream into Gizani but no obvious springs into wetland.
RFL50 flat town centre town Rain from Gisa cafe roof during heavy shower at 13:30 18-10-15
RFL51 flat neck at floodplain outlet into other floodplain trees surrounded by crops 0.2 3 3 222Rn sample/test location. Sampled surface water.
RFL52 centre flat floodplain before large Brante floodplain pasture 0.3 1.5 1 222Rn sample/test location. Sampled surface water.
RFL53 centre of d/stream end of large Brante floodplain pasture 0.5 4 2 222Rn sample/test location. Sampled surface water.
RFL54 centre of large Brante floodplain pasture 0.5 5 1.25 222Rn sample/test location. Sampled surface water.
RFL55 extreme upstream end of large floodplain pasture and crops 0.5 8 2 222Rn sample/test location. Sampled surface water.
RFL16 quite steep valley sides crops on N bank, acacia forest on S bank 1 15 1.5 gauge broken (again) after hit by floating tree in wet season (again). Metal supports and upper board snapped. Stage measured by dipping measuring
RFL56 quite steep valley sides crops on N bank, acacia forest on S bank 0.7 11 1.5 Flow gauging location 20 m upstream of Kilti gauge
RFL57 very very large flat floodplain/wetland pasture/swamp 0.2 ~500 0 Wetland, no flow observed, locals say no outlet, large marsh in the middle
RFL58 neck between two floodplains crops and pasture 0.6 5 1.5 222Rn sample/test location. Sampled surface water.
RFL59 large floodplain pasture 0.7 7 1.8 222Rn sample/test location. Sampled surface water.
RFL60 small floodplain pasture 0.4 8 1.5 222Rn sample/test location. Sampled surface water.
RFL61 narrow valley, steep fast flow over basalt trees, pasture, crops 0.3 9 1 222Rn sample/test location. Sampled surface water.
RFL62 floodplain u/s of where large side floodplain joins pasture 0.4 8 1.2 222Rn sample/test location. Sampled surface water.
RFL63 huge flat floodplain pasture 0.7 6 1.2 222Rn sample/test location. Sampled surface water.
RFL64 huge flat floodplain with braided river (3x channels) pasture and wetland 0.5 20 0.5 222Rn sample/test location. Sampled surface water.
RFL65 huge flat floodplain pasture 0.8 6 0.4 222Rn sample/test location. Sampled surface water.
RFL66 huge flat floodplain pasture 0.2 4 1.2 222Rn sample/test location. Sampled surface water. There was a bridge here but it has washed away.
RFL67 huge flat floodplain pasture and wetland 0.2 4 2.5 222Rn sample/test location. Sampled surface water.
372

Walker, D 2018

Uploaded by

Copyright:

Available Formats

Walker, D 2018

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Walker, D 2018

Uploaded by

Copyright:

Available Formats

Resilience of shallow groundwater resources and

their potential for use in small-scale irrigation: a

David William Walker

A thesis presented for the degree of

To fill gaps in formal hydrometeorological monitoring, a community-based monitoring

A recharge assessment estimated annual recharge of 280-430 mm, confirming that a

At Newcastle University I would like to thank numerous researchers: especially, Nathan

In Ethiopia, I am very grateful to International Water Management Institute (IWMI) for

Walker, D. (2016) 'Properties of shallow thin regolith aquifers in sub-Saharan Africa: a

Related publications ....................................................................................................... vi

Contents ......................................................................................................................... vii

List of figures ................................................................................................................ xvi

List of tables ................................................................................................................. xxv

List of abbreviations and acronyms ........................................................................ xxvii

Chapter 1. Introduction ............................................................................................. 1

1.0 Chapter summary .................................................................................................. 1

1.1 Context .................................................................................................................. 1

1.2 Aims ...................................................................................................................... 5

1.3 Study site ............................................................................................................... 6

1.4 Methodology ......................................................................................................... 7

1.5 Thesis outline ........................................................................................................ 8

Chapter 2. Literature review................................................................................... 10

2.0 Chapter overview ................................................................................................ 10

2.1 The role of groundwater in Asia’s “Green Revolution” ..................................... 10

2.3 African hydrogeology ......................................................................................... 14

2.4 The African climate ............................................................................................ 16

2.5 Land use change .................................................................................................. 23

2.6 Research gap ....................................................................................................... 24

Chapter 3. Study site background .......................................................................... 29

3.0 Chapter overview ................................................................................................ 29

3.1 Site selection ....................................................................................................... 29

3.3 Site description.................................................................................................... 32

3.4 Agriculture .......................................................................................................... 36

3.5 Climate ................................................................................................................ 38

3.6 Geology ............................................................................................................... 39

3.7 Regolith hydrogeology........................................................................................ 42

Chapter 4. Field investigations and development of conceptual model .............. 45

4.0 Chapter overview ................................................................................................ 45

4.1 Field visits ........................................................................................................... 45

4.2 Aims .................................................................................................................... 45

4.3 Study site observations........................................................................................ 45

4.4 Hydrology ........................................................................................................... 47

4.5 Dambos ............................................................................................................... 48

4.6 Geology ............................................................................................................... 51

4.8 Pumping tests ...................................................................................................... 55

4.10 Conceptual model ............................................................................................... 57

4.11 Conclusions ......................................................................................................... 60

Chapter 5. Filling the observational void: Scientific value and quantitative

5.0 Chapter overview ................................................................................................ 61

5.1 Introduction ......................................................................................................... 62

5.2 Sub-Saharan Africa context ................................................................................ 63

5.3 Community-based monitoring ............................................................................ 65

5.4 Project context..................................................................................................... 67

5.5 Data analysis methods ......................................................................................... 69

5.6 Rainfall ................................................................................................................ 73

5.7 River flow ........................................................................................................... 77

5.8 Groundwater level data ....................................................................................... 80

5.9 Discussion ........................................................................................................... 83

5.10 Conclusions ......................................................................................................... 86

Chapter 6. Insights from a multi-method recharge estimation comparison study