groundwater resource assessment through distributed steady

Groundwater resource assessment through distributed steady-state flow modeling,

Aynalem wellfield (Mekele, Ethiopia)

Gebrerufael Hailu Kahsay

March, 2008

Groundwater resource assessment through distributed steady-state flow modeling, Aynalem wellfield

( Mekele,Ethiopia) by

Gebrerufael Hailu Kahsay

Thesis submitted to the International Institute for Geo-information Science and Earth Observation in

partial fulfilment of the requirements for the degree of Master of Science in Geo-information Science

and Earth Observation, Specialisation: (Groundwater Assessment and Modeling)

Thesis Assessment Board

Chairman Dr.Ir. M.W. Lubczynski WRS,ITC,Enschede

External Examiner Dr.Ir.P. Droogers Future Water,Wageningen

First Supervisor Dr.A.S.M. Gieske WRS,ITC,Enschede

Second Supervisor Dr.Ing. T.H.M. Tom Rientjes WRS,ITC,Enschede

INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION

ENSCHEDE, THE NETHERLANDS

Disclaimer This document describes work undertaken as part of a programme of study at the International Institute for Geo-information Science and Earth Observation. All views and opinions expressed therein remain the sole responsibility of the author, and do not necessarily represent those of the institute.

Dedicated to my father Hailu Kahsay

i

Abstract

The study focused on groundwater recourse assessment through steady-state flow modelling in

Aynalem wellfield northern Ethiopia. Aynalem wellfield is the main source of water for domestic

water supply of Mekele town, capital of Tigray regional state. Despite its importance for the people in

the region, the hydrogeological system of the wellfield is not well understood. The groundwater in the

area is pumped with little consideration to groundwater recharge and effects of climatic forcing on the

recharge. Because the demand for water for domestic and irrigation use is growing fast, the pressure

on the wellfield will be even more serious in the future. Primary and secondary data on geology,

hydrochemistry, geophysics and hydrology are integrated to develop the hydrogeological conceptual

model which is foundation for the development of steady-state groundwater flow model of the area.

The aquifer system was modelled numerically using PMWIN5.3 as pre and post processor of

MODFLOW under a steady-state condition with one layer of 50 meters constant thickness. The model

area which is about one hundred four square kilometres was divided into grid blocks of 250 by 250

meters. The model domain was delineated based on field traverses, topographic maps and DEM

extracted from ASTER image. Aquifer parameters were assigned based on previously reported values

which were then adjusted during the model calibration. The main recharge mechanism considered was

direct recharge from rainfall. Annual recharge of 30-40 mm (4.5-6% of the average annual rainfall) is

estimated by applying the chloride mass balance method. The model was calibrated under non-

pumping and pumping scenarios to static water levels and to averages of three years monitoring water

level respectively. The over all model results were comparable with the measured well data.

The steady-state flow modeling has demonstrated that an average recharge of 42 mm year-1 maintains

the natural equilibrium. On the other hand, the model result with pumping scenario shows that

groundwater abstraction of 7156 m3day-1 resulted in groundwater table decline up to 37 meters in the

wellfield area. The sensitivity of the calibrated model was tested by systematically changing one

parameter or input variable at a time and it was found that the model is highly sensitive to changes of

transmissivity of the aquifer system and recharge rate. The model is associated with a number of

uncertainties resulting from the simplification and assumptions made to the complex field conditions,

poor data quality, and lack of detailed subsurface characterisation of the aquifer system. Hence the

limitations of the model should be taken into consideration prior to applying the model for

groundwater resource management.

Key Words: Aynalem wellfield, Groundwater Resources assessment, Groundwater modeling.

ii

Acknowledgements

I appreciate ITC (International Institute for Geo-information Science and Earth Observation) for

enabling me to pursue my Master program study by providing academic and financial supports. I am

grateful to my organization, Water Resource Development Bureau for granting my leave of absence to

pursue my study.

Above all I am indebted to my first supervisor Dr. Ambro S.M. Gieske for his unceasing support

guidance and encouragement throughout my study period and thesis time. I benefited a lot from

discussions I had with him owing to which I gained a deeper insight into and understanding of the

factors governing groundwater occurrence and movement. I really appreciate his constructive

criticism and valuable advice which helped me to develop research skills and improve my English. I

would like to thank my second supervisor Dr. Ing. T.H.M. Rientjes for he taught me many of the

aspects in groundwater modeling principles and I am grateful for his critical comments during the

thesis preparation time. I acknowledge the support during my laboratory work to Boudewijn de Smeth

and Remco Dost. It is also my pleasure to thank all WREM staff. Without the impartation of their

knowledge; this work would not have been achieved.

I also wish to express my appreciation to all my class mates for their friendship, support, socialization

and help each other in times of pressure and stress. They were my new family during my stay and it

was a pleasure to be a member of them.

I would like to thank the entire Ethiopian community for providing me environment of home feeling

during my stay here at ITC.

I am most thankful to Teklay, Guesh, Gidena, Aregawi, Yemane,Tesfalem,Solomon and Gebremedhin

who helped me a lot in the secondary data collection during my fieldwork. I particularly appreciate

Teklay that he walked around with me with mud up to his knees and helped me during river flow

measurements.

Finally I owe special gratitude to all my family members and friends back home for always being

there for me.

Thank you all

iii

Table of contents

1. Introduction ...................................................................................................................................1 1.1. Background.............................................................................................................................1 1.2. Problem statement...................................................................................................................2 1.3. Research objective ..................................................................................................................3 1.4. Research questions..................................................................................................................3 1.5. Methods and Materials ...........................................................................................................3 1.6. Organization of the thesis .......................................................................................................7

2. Literature review...........................................................................................................................9 2.1. Previous works........................................................................................................................9 2.2. Groundwater modeling .........................................................................................................10 2.3. Recharge ...............................................................................................................................11

3. Description of the study area......................................................................................................13 3.1. Location ................................................................................................................................13 3.2. Geomorphology and drainage...............................................................................................13 3.3. Climate..................................................................................................................................15 3.4. Land use, Vegetation and soil...............................................................................................17 3.5. Geology.................................................................................................................................19

4. Analysis and model input data preparation..............................................................................23 4.1. Hydrometeorology ................................................................................................................23 4.2. Hydrochemistry.....................................................................................................................25

4.2.1. Water sampling and analysis........................................................................................25 4.2.2. Reliability check ..........................................................................................................25 4.2.3. Presentation of results ..................................................................................................27 4.2.4. Water type ....................................................................................................................27 4.2.5. Source rock deduction..................................................................................................30

4.3. Chloride mass balance method (CMB).................................................................................31 4.3.1. Chloride in rainwater ...................................................................................................32 4.3.2. Chloride content in groundwater..................................................................................32

4.4. Well abstraction and groundwater level analysis .................................................................35 4.4.1. Well abstraction ...........................................................................................................35 4.4.2. Groundwater level analysis..........................................................................................37

4.5. Pumping test .........................................................................................................................38 4.6. Aquifer characteristics..........................................................................................................42 4.7. Digital elevation model (DEM) ............................................................................................43

5. Conceptual model ........................................................................................................................45 5.1.1. Well log data and geology............................................................................................45 5.1.2. Geophysics ...................................................................................................................48

5.2. Hydrostratigraphy .................................................................................................................49 5.3. Hydraulic proporties of the stratigraphic units .....................................................................50 5.4. Water budget.........................................................................................................................50 5.5. Groundwater flow system.....................................................................................................53 5.6. Model boundaries .................................................................................................................53 5.7. Simplification of the real world............................................................................................54

6. Numerical model..........................................................................................................................57

iv

6.1. Code selection...................................................................................................................... 57 6.2. Model geometry ................................................................................................................... 57 6.3. Model design........................................................................................................................ 58 6.4. Model calibration................................................................................................................. 61 6.5. Sensitivity analysis............................................................................................................... 66 6.6. Model validation .................................................................................................................. 67

7. Discussion and results ................................................................................................................ 69 7.1. Hydrochemistry.................................................................................................................... 69 7.2. Modeling results .................................................................................................................. 69

7.2.1. Hydraulic properties.................................................................................................... 71 7.3. Groundwater budget............................................................................................................. 72 7.4. Model limitations................................................................................................................. 73

8. Conclusion and recommendations ............................................................................................ 75 8.1. Conclusions.......................................................................................................................... 75 8.2. Recommendations................................................................................................................ 76

9. References.................................................................................................................................... 77 Appendices ........................................................................................................................................... 81

Appendix 1 Hydrometeorological data ............................................................................................. 81 Appendix 1.1. Monthly rainfall (mm) at Mekele airport station.................................................. 81 Appendix 1.2. Long term monthly rainfall (mm) at Mekele airport station................................. 81 Appendix 1.3. Monthly minimum temperatures (0C) ................................................................... 83 Appendix 1.4. Monthly maximum temperature (0C).................................................................... 83 Appendix 1.5. Monthly mean wind speed (m s-1) at 2m height................................................... 84 Appendix 1.6. Mean monthly relative humidity (%) at 1200 local time...................................... 84 Appendix 1.7. Mean monthly sunshine hours .............................................................................. 84 Appendix 1.8. Monthly average piche evaporation (mm)............................................................ 85 Appendix 1.9. Monthly Evapotranspiration (mm) ....................................................................... 85 Appendix 1.10. River discharge Metere gauging station (Aynalem river)................................... 86

Appendix 2 Hydrochemistry ............................................................................................................. 87 Appendix 2.1. Analysis result of rain water ................................................................................. 87 Appendix 2.2. Physical and chemical constituents of water samples .......................................... 87 Appendix 2.3. Comparison of analysis......................................................................................... 87 Appendix 2.4. Major anions and cations ( meq l-1) and water type.............................................. 88

Appendix 3 Well data........................................................................................................................ 89 Appendix 3.1. Well location......................................................................................................... 89 Appendix 3.2. Monthly water production (m3) ............................................................................ 89 Appendix 3.3. Monthly groundwater level monitoring data ........................................................ 91 Appendix 3.4. Static water level record from the wells ............................................................... 91 Appendix 3.5. Lithologic log data of the boreholes ..................................................................... 92

Appendix 4 Geophysical data ........................................................................................................... 97 Appendix 5 Groundcontrol points to correct ASTER DEM......................................................... 100 Appendix 6 Location of all wells in the wellfield........................................................................... 101 Appendix 7 MODFLOW water budget........................................................................................... 102 Appendix 8 Pumping test curve matching....................................................................................... 103 Appendix 9 Plates............................................................................................................................ 105

v

List of figures

Figure 1.1. Map showing estimated distribution of groundwater availability .........................................2 Figure 1.2. Flow chart of methodological approach ................................................................................6 Figure 3.1. Location map of the study area............................................................................................13 Figure 3.2. Geomorphologic features and elevation cross-section ........................................................14 Figure 3.3. Drainage map of Aynalem sub-basin...................................................................................15 Figure 3.4. Mean monthly values of climatic variables .........................................................................17 Figure 3.5. Land use map of the study area (after Teklay, 2006) ..........................................................18 Figure 3.6. Soil map of the study area (WWDSE, 2006).......................................................................18 Figure 3.7. Composite stratigraphy of sedimentary succession in Mekele outlier. ...............................20 Figure 4.1. Long-term annual rainfall of the study area ........................................................................23 Figure 4.2. Hydrograph of Aynalem river..............................................................................................24 Figure 4.3. Location of water sample points..........................................................................................26 Figure 4.4. Piper diagram of water samples from boreholes .................................................................28 Figure 4.5. Stiff diagrams of water samples from upper Aynalem........................................................28 Figure 4.6. Stiff diagrams of water samples from lower Aynalem........................................................29 Figure 4.7. Stiff patterns of water samples from Ilala and Chelekot .....................................................29 Figure 4.8. Groundwater abstraction from selected boreholes ..............................................................36 Figure 4.9. Location map of pumping wells ..........................................................................................36 Figure 4.10. Total production of the wellfield.......................................................................................37 Figure 4.11. Groundwater level at selected boreholes ..........................................................................37 Figure 4.12. Average groundwater level trend.......................................................................................38 Figure 4.13. Time drawdown plot of TW1 (2005).................................................................................39 Figure 4.14. Time drawdown plot of TW2 (2005).................................................................................40 Figure 4.15. Time drawdown plot of TW4 (2005).................................................................................40 Figure 4.16. Time drawdown plot of TW6 (2006).................................................................................40 Figure 4.17. Log hydraulic conductivity values.....................................................................................42 Figure 4.18. Scatter plot of ground elevation Vs elevation from ASTER DEM ...................................44 Figure 5.1. Dolerite dyke dissecting the sedimentary rock....................................................................46 Figure 5.2. Tilted sedimentary layers due to dolerite intrusion .............................................................46 Figure 5.3. Lithological log showing depth to dolerite (After Yehdego, 2003) ....................................47 Figure 5.4. Geological map with east- west cross-section (WWDSE, 2006) ........................................48 Figure 5.5. Groundwater level profile....................................................................................................52 Figure 5.6. ASTER DEM indicating the three basins in the Mekele area .............................................54 Figure 5.7. Pictorial representation of the hydrologic system of Aynalem sub-basin ...........................55 Figure 6.1. Model boundary conditions .................................................................................................60 Figure 6.2. Trial and error calibration procedures (Adapted from Anderson and Woessner, 1992).....62 Figure 6.3. Contour map of simulated heads (non-pumping scenario) ..................................................63 Figure 6.4. Scatter plot of observed and simulated hydraulic heads (m)...............................................63 Figure 6.5. Sensitivity plot of the calibrated model with respect to transmissivity...............................67 Figure 6.6. Sensitivity plot of the calibrated model with respect to recharge .......................................67 Figure 7.1. Distribution of hydraulic heads with non-pumping scenario ..............................................70 Figure 7.2. Distribution of hydraulic heads with pumping scenario......................................................70

vi

Figure 7.3. Comparisons of simulated hydraulic heads for both scenarios .......................................... 70 Figure 7.4. Transmissivity zones applied to the calibrated model ........................................................ 71

vii

List of tables

Table 4.1. Monthly river discharge (Mm3) of Aynalem river................................................................24 Table 4.2. Summary statistics of the major groundwater constituents ..................................................27 Table 4.3. Parameters used for source rock deduction ..........................................................................31 Table 4.4. Chloride concentration in rain ..............................................................................................32 Table 4.5. Statistics of the chloride concentration in groundwater .......................................................32 Table 4.6. Groundwater chloride content and estimated recharge.........................................................34 Table 4.7. Daily maximum abstraction rate from the wellfield .............................................................35 Table 4.8. Details of pumping test on the test wells ..............................................................................39 Table 4.9. Transmissivity and hydraulic conductivity...........................................................................43 Table 4.10. Summary statistics of transmissivity (m2 day-1) ..................................................................43 Table 5.1. Summery of vertical electrical sounding data.......................................................................49 Table 5.2. Spring inventory data ............................................................................................................51 Table 6.1. Observed and calculated heads for non-pumping scenario...................................................64 Table 6.2. Observed and calculated heads for pumping scenario..........................................................64 Table 6.3. Errors of the calibrated model...............................................................................................65 Table 7.1. Model simulated groundwater budget of the area for the non- pumping scenario ...............73 Table 7.2. Model simulated groundwater budget of the area for pumping scenario .............................73

GROUNDWATER RESOURCE ASSESSMENT THROUGH DISTRIBUTED STEADY-STATE FLOW MODELING, AYNALEM WELLFIELD (MEKELE, ETHIOPIA)

1

1. Introduction

1.1. Background

Groundwater is the subsurface water that occurs beneath the water table in the soils and geologic

formations that are fully saturated (Freeze & Cherry, 1979). Groundwater is one of the key natural

resources of the world. Many major cities and small towns in the world depend on groundwater for

water supplies, mainly because of its abundance, stable quality and also because it is inexpensive to

exploit (Morris et al., 2003). Groundwater use has fundamental importance to meet the rapidly

expanding urban, industrial and agricultural water requirement, especially in arid areas where surface

waters are scarce and seasonal. Uneven distribution of surface water resources resulted in an

increased emphasis on development of groundwater resources. An important objective of most

groundwater studies is to make a quantitative assessment of the groundwater resources in terms of the

total volume of water stored in aquifer or long-term average recharge. Groundwater recharge is

determined to a large extent as an imbalance at the land surface between precipitation and evaporative

demand. When precipitation exceeds evaporative demand by an amount sufficient to replenish soil

water storage, any further excess flows deeper into the ground and arrives at the water table as

recharge.

Groundwater systems have been studied by the use of computer based mathematical models

(Brassington, 1998). These essentially comprise a vast array of equations, which describe

groundwater flow and the water balance in the aquifer. Finite difference method is a commonly used

method to solve the equations. The equations are solved for each node and the movement of

groundwater from one node to its neighbor is calculated. As discussed by Scanlon et al. (2003),

numerical groundwater models are one of the best predictive tools available for managing water

resources in aquifers. These models can be used to test or refine different conceptual models, estimate

hydraulic parameters and, most importantly for water-resource management, predict how the aquifer

might respond to changes in pumping and climate. Groundwater abstractions that exceed the average

recharge, results in a continuing depletion of aquifer storage and lowering of the groundwater table.

Hence safe groundwater abstraction and proper groundwater management is crucial for sustainability

of the resource. Safe yield is the amount of naturally occurring groundwater that can be withdrawn

from an aquifer on a sustained basis, economically and legally, without impairing the native ground

water quality or creating undesirable effects, such as environmental damage (Fetter, 2001).

As most semi-arid areas, Ethiopia is also facing water scarcity particularly in the dry season. Because

of the poor permeability of the crystalline rocks and variable water table depths, the country has

limited supply of groundwater (MacDonald, 2001). The groundwater occurrence is mainly governed

by geology, degree of fracture and topography. As indicated in (Fig.1), the low lands, mainly the rift

valley areas, are characterized by a relatively high potential of groundwater availability. Despite its

scarcity, ambitious development plans, urbanization and rapid population growth lead to over

pumping of the groundwater resource.


2

Aynalem wellfield is one of the areas in Ethiopia that face water table lowering due to increased

withdrawals. The groundwater in the area is pumped with little consideration to groundwater recharge

and effects of climatic forcing on the recharge. As described by Jyrkama & Sykes (2007), quantifying

the future evolution of recharge over time requires not only the reliable forecasting of changes in key

climatic variation but also modeling their impact on the spatially varying recharge process.

Figure 1.1. Map showing estimated distribution of groundwater availability (After MacDonald, 2001)

1.2. Problem statement

Aynalem wellfield is the largest wellfield in Tigray region (Northern Ethiopia) and serves as the only

source of water supply for Mekele town, capital of the regional state. The groundwater table of the

wellfield is continuously declining due to abstraction of water mainly for water supply of the town. As

reported from Water Resources Development Bureau of the region, there has been a significant

groundwater level decline since the implementation the wellfield. On the other hand, population

growth and accompanying development over the past few years in the city have led to an increased

demand for groundwater use from the wellfield. Despite its importance for the people in the region,

the hydrogeological system of Aynalem wellfield is not well understood. The groundwater flow

pattern and the effect of climatic variation on groundwater recharge are not well defined. As an

indictor, the water supply project which was planned for twenty years indicate a drastic groundwater

level decline within three years of service. Because the demand for water for domestic and irrigation

use is growing fast, the pressure on the wellfield will be even more serious in the future. According to

Villholth (2006), abstraction of groundwater has an associated impact on the water balance and hence


3

on the availability of water resources on other parts of the water cycle. Thus understanding of the

aquifer system and assessment of the water balance components of the sub-basin is crucial for the

sustainability of the resource. To understand the effects of abstractions and climatic forcing on the

groundwater flow system, it is worthwhile to develop a groundwater flow model simulating not only

the natural groundwater flow but also abstractions from the underlying aquifer.

1.3. Research objective

General objective

The main objective of the study is to assess groundwater resource of the wellfield and to improve the

understanding of groundwater flow pattern in response to recharge and abstractions using a steady-

state groundwater flow model.

Specific objectives

• To develop a conceptual model representing the hydrogeological condition of the wellfield

based on ground observations and data analysis.

• To assess the water balance components of Aynalem sub-basin.

• To estimate recharge using chloride mass balance method and through model calibration.

• To Set up and calibrate a steady-state groundwater flow model of the wellfield.

• To develop scenarios illustrating effect of different stress conditions on the groundwater

resource.

1.4. Research questions

• What is the dominant hydrologic factor that causes the lowering of the water table in the

wellfield?

• How is the natural flow pattern of groundwater in the wellfield?

• Can a steady state groundwater flow model improve our understanding of flow pattern and

predict the effect of future abstractions?

• How accurate can the groundwater recharge of Aynalem sub-basin be estimated using

chloride mass balance method?

• How accurate are the simulation results from the model developed for the wellfield?

1.5. Methods and Materials

The methods followed in the research process are based on the objectives formulated in section 1.3.

The methodology designed for the research work consists of three major phases (Pre-fieldwork,

Fieldwork and Post- fieldwork).

Pre-fieldwork

In this phase of study, review of previous works on the area and literature related to the method of

recharge estimation and principles of groundwater modeling take a major part. Apart from the

literature review, an archive search for ASTER image and SRTM of the area was also part of the pre-


4

fieldwork phase of study. As part of the secondary data collection, Aster satellite image was acquired

for 30 September, 2005 through ITC remote sensing and geo-database. Acquisition of equipment and

preparation of data requirement list and data collection form were the activities conducted before the

field trip.

Fieldwork

A three week field trip starting at the second week of August 2007 was organized to collect relevant

secondary data and ground truth primary data from the field. Meteorological data (Rainfall, Relative

humidity, Pan evaporation, temperature and wind speed at Mekele airport meteorological station was

collected from Ethiopian National Meteorology Service. Three years of groundwater level data with

gaps in the records of some of the boreholes were collected from Water Supply Office of Mekele

town. Monthly River discharge data at Aynalem River gauging station (which is not operational at the

moment and covers only the upper part of Aynalem catchment) was collected for the years 1992 to

2001 from Ethiopian National Meteorology Service. Lithological and geophysical logs, geological

maps and cross-sections, pumping tests data, soil and land use maps were obtained from previous

works mainly conducted by Water Works Design and Supervision Enterprise (WWDSE). As part of

the field work, primary data collection was the major task of the field duration. The primary data

collected include:

• Water samples from boreholes, springs, ponds and rainfall

• Groundwater level measurements at accessible boreholes

• River discharge measurements

• EC, PH, and chloride in situ measurements

• Ground truth observations

Observation points were selected with the help of topographic maps, aerial photographs and ASTER

image of the area. Location readings of the observation points were taken with a hand held Garmin

GPS. The ground truth field observations were focused on the description of the geology,

stratigraphpy, geomorphologic setting, surface water divide, location of discharge, and recharge areas.

Post-fieldwork (data processing and analysis)

At this stage, the primary and secondary data collected during the pre-field stage and fieldwork period

are processed and analyzed. The most important phase here is processing of the data in order to fit the

data input requirement of the intended model (MODFLOW).

For data processing purposes, GIS and geostatistics are used to prepare model input data. For

example, kriging is applied in order to interpolate point measurements of static water levels to prepare

initial hydraulic heads for the entire model. Use of spreadsheet is made to process and prepare input

data for the model. As part of the input data preparation, a DEM is extracted from ASTER image that

was applied to determine surface elevations of boreholes to prepare surface topography cross- sections

and to define top and bottom elevations of layer in the numerical modeling. Pumping test data and

geophysical data from previous studies were analyzed to have an idea about the aquifer parameters

and the geometry of the aquifer system. Well completion data are organized and summarized to

determine the aquifer thickness and vertical extent of layers in combination with geological cross-

section and geophysical log data. The water samples collected during the fieldwork are processed and


5

analyzed for the major cations and anions to understand the geochemical properties and source rock

deduction. Results of the laboratory analysis are processed and presented using AQUACHEM

software. By extensive use of Surfer 8 and Global Mapper 5 maps and cross-sections of results are

made. The determination of the chloride in rainwater and concentration in groundwater is part of this

analysis. As major tasks of the post-fieldwork phase of the study, recharge was estimated using

chloride mass balance method, a conceptual model was built based on the existing data complemented

with the data collected in the field and one layer groundwater flow model was developed and

calibrated for pumping and non-pumping scenarios.

Materials and equipments

To collect the required data, a number of materials and equipments were used during the fieldwork

and office work phase of the research. Topographic maps of south Mekele and Quiha sheets and

ASTER image were used for the delineation of the study domain and to select representative ground

observation and sampling points. Field geological equipments including geological hammer, compass

and Garmin GPS system were applied at all times during site visit. Water level meter and current

meter were used to measure depth to water table and river discharge respectively. EC meter pH meter

and chloride titration reagent were used for in situ measurements of the electrical conductivity, pH

and chloride ion of the water samples respectively. Two sample bottles were used for sampling the

water sample from a given spot. One of the sample bottles was acidified with hydrochloric acid and is

used for the analysis of anions except chloride. The second sample bottle is acidified using nitric acid

and is used for the analysis of chloride ion and cations.

Framework of the research

The research has two major tasks: interpretation and analysis of hydro meteorological data, geology,

and hydrogeology, hydrochemistry of the water samples for recharge estimation and preparation of

model input data. The second task of the research is modeling of the groundwater flow system in the

sub-catchment. The sequence of the study process is indicated in Figure 1.2 below.


6

Figure 1.2. Flow chart of methodological approach


7

1.6. Organization of the thesis

The thesis is divided in eight chapters and the contents are outlined briefly as:

Chapter1: Describes the introduction of the research that includes the problem statement, the

objective of the research and research questions which the research tries to answer on the basis of the

available data and applied methodologies. Methods followed and materials used are also discussed in

this introductory chapter.

Chapter 2: Stresses the review of previous studies in the area and literature review related to

principle of groundwater modeling and recharge estimation methods are discussed.

Chapter 3: General description of the study area, giving the description of the study area in relation

to location, climate, geology, structure, land use, soil cover and hydro geological setting of the area.

Chapter4: Data processing and analysis. This chapter is devoted to the processing and analysis of

data (primary and secondary data) and involved in synthesizing and screening field data and

translating the data to model input. Water chemistry data, water level data, discharge and recharge

condition of the area are parts of the analysis.

Chapter 5: Development of conceptual model. The main task of this chapter is development of

conceptual model of the area by defining the hydrostratigraphy, water budget and flow systems.

Chapter 6: Numerical modeling, this chapter is mainly designed to discuss the code selection, model

design, model calibration and sensitivity analysis.

Chapter 7: Results and discussion. Illustration and discussion of the modeling results and result of

the recharge estimated by different methods.

Chapter 8: Conclusion and recommendations. Conclusions and recommendations will be made on

the basis of the analysis result. In this final chapter, matters which can not be addressed fully or

partially are out lined and limitation of the research and possibilities of further research are indicated.


8


9

2. Literature review

2.1. Previous works

The lowering of groundwater table in Aynalem wellfield creates a great concern to the region. As a

result, a number of studies were conducted under the supervision of the Tigray Water Resources

Bureau. However, most of the studies focused on selection of prospective borehole sites and

deepening of the existing ones. In the studies, the recharge and other components of the water balance

were not well understood and not considered in the design of the wellfield. Hydrogeological study

was conducted by Beyth (1970) to locate borehole sites for Mekele town water supply. He conclude

that there is no information on the regional groundwater regime in Mekele outlier but the groundwater

availability is probably controlled by the regional and local structures and the potential well sites

were located near the fault lines. The Water resource verification report of DEVECON (1992)

describes the geology structure and hydrogeology of Mekele area. The report indicates that the

groundwater is confined because of alternative layers of shale, marl limestone and dolerite. According

to DEVECON (1992), the main aquifer is dolerite unit, which is not in line with recent studies.

Nowadays days there is acceptance that the main aquifer is the limestone unit and that most of the

productive wells are located near the faults and lineaments where the limestone is highly fractured.

The dolerite dyke plays an important role as it is contributes to the fracturing of the limestone unit in

its way up.

Water Works Design and Supervision Enterprise (WWDSE) carried out a good initial work in 2006.

Unlike the previous studies, this study attempted to assess the hydrological components of the sub-

basin. WWDSE (2006) produce a report containing two volumes, the first volume focuses on geology

and hydrogeology of Mekele vicinity, while the second volume deals with the evaluation of the

groundwater potential in Aynalem wellfield. Both reports contain much useful data and information

including inventory of water points, water quality data, geophysical data and groundwater level

monitoring data of Aynalem wellfield boreholes. According to WWDSE (2006), the main lithologic

units that cover the Aynalem wellfield are shale-limestone intercalation, limestone unit, and Mekele

dolerite. Pockets of calcareous sandstone were also identified but they occupy very small areal

coverage as compared to other lithologic units. The dolerite rocks observed in the area occur mainly

as sills that are very much concordant with the sedimentary rocks. However, there are also dykes of

dolerite offshoots emerging from the sills forming discordant relations.

There is an ongoing study by TAHAL Consulting Engineers which focuses on present water supply

source assessment. According to TAHAL (2007), the current abstraction of about 8000 m3day-1

resulted in groundwater table decline of 15 meters and concluded that the natural recharge of

Aynalem wellfield is quite low which is in the order of 3-5 MCM per annum.

A number of researchers have also conducted scientific research with regard to the overall

groundwater condition of the Aynalem wellfield. Hussien (2000) studied the hydrogeology of the

wellfield and concluded that the annual recharge in the Aynalem catchment is 9 % of the annual

rainfall, which equals about 5.7 M m3year-1. The same author has discussed that the variation in the


10

chemical composition of the groundwater in east-west direction along the valley is attributed to the

variation in the geology of the area. The high concentration of sulphate in the down stream part of the

wellfield is due to the presence of thin layer of gypsum. Yehdego (2003) conducted his research on

the hydrogeologic condition of the Ilala-Aynalem catchments, with particular emphasis given to the

variation in the chemical characteristics of the aquifers, with the application of Isotope hydrology.

The Aynalem sub-catchment shows relatively different water ages, but generally, the boreholes within

the catchments receive recharge from recent rainwater (Yehdego, 2003). Gebregziabher (2003)

conducted a research work in the Aynalem wellfield with the aim to outline fractures and faults

through an integrated geophysical investigation. He concluded that most of the detected fractures and

faults are aligned along NW-SE strike direction similar to the regional structures. He pointed out that

the geological structures (faults, fractures and lithologic contacts) play an important role in the

movement and occurrence of groundwater in the study area. Teklay (2006) carried out a study that

focussed on conjugate use of groundwater and surface water in the sub-basin. In his study he

estimated 35 mm year-1 direct recharge from rainfall which is 5.3% of the annual average rainfall in

the catchment.

2.2. Groundwater modeling

A model is any device that represents an approximation of a field situation (Anderson & Woessner,

1992). As described by Fetter (2001), there are two areas of hydrogeology where we need to rely upon

models of real hydrologic systems: to understand why a flow system is behaving in a particular

observed manner and to predict how a flow system will behave in the future. Groundwater models

have been applied in different parts of the world to solve problems related to groundwater flow, and

solute transport. As explained by Anderson & Woessner (1992), groundwater flow models solve the

distribution of head, whereas solute transport models solve for the concentration of solute as affected

by advection (movement of the solute with the average groundwater flow), dispersion (spreading and

mixing of the solute) and chemical reactions.

The main areas of application of groundwater models are the fields of water quantity and water

quality. In quantitative groundwater modeling only the flow and movement of water are modeled. For

such, model algorithms are applied that are based on the equation of motion (Darcy’s law) and the

continuity of mass equation. These equations are combined and result in the groundwater flow

equation. In the past, groundwater flow behavior has been simulated by scale models, analogue

models and analytical models. However, particularly over the past decades, the use of computer code

is common and allows for the design and development of complex mathematical model approaches.

Several techniques have been developed to solve the partial differential equations describing the

behavior of groundwater flow system. The most popular numerical solution methods are Finite

difference, Finite element and Analytical element method. Finite difference method was the first

method to be used for the systematic numerical solution of partial differential equations. As discussed

by Mehl & Hill (2002), many numerical models of groundwater flow use finite-difference methods to

discretize and solve the governing partial differential flow equation. In finite-difference methods, an

aquifer system is replaced by a discretized domain consisting of an array of nodes and associated

finite difference blocks (cells). Visual MODFLOW, which is based on the finite difference method, is


11

currently the most widely used groundwater flow code in the field of geo-hydrology. Within visual

MODFLOW, the groundwater system is modeled by a set of mathematical equations representing the

flow phenomenon. As described by Carrera-Hernandez & Gaskin (2006), the advantage of

MODFLOW is that it provides different modules to undertake 3-D groundwater flow simulations in

confined and unconfined aquifers as well as in aquifers with variable confinement with both constant

and variable transmissivity values. The modules provided by MODFLOW can be used to simulate the

effect on an aquifer system caused by different stresses such as the presence of extraction and

injection wells, of areal distribution of recharge, evapotranspiration or of different hydrological

features. In the practice of groundwater modeling, often the term numerical model is used to

emphasize that a distributed model domain is applied. Simulation of groundwater flow considering

space variability of stresses and aquifer properties is only possible through distributed-parameter

models (Pulido-Velazquez et al., 2007).

MODFLOW can simulate confined, leaky confined and unconfined aquifers and only simulates

saturated flow in a porous medium with uniform temperature and density (Fetter, 2001). MODFLOW

can not simulate flow in the unsaturated zone and flow in fractured media unless it can be considered

to be an equivalent porous media. However, the discrete heterogeneity of fracture distribution and

hydraulic discontinuity are the primary difficulties in the groundwater modeling practices. Forming a

conceptual model of fractured system requires either a gross simplification or detailed description of

the aquifer properties controlling the groundwater flow (Anderson & Woessner, 1992). Fractured

material is represented as an equivalent porous medium by replacing the primary and secondary

porosity and hydraulic conductivity distributions with a continuous porous medium of equivalent or

effective hydraulic properties. According to Yuri Mun et al.(2004), the equivalent porous medium

(EPM) approach has been frequently applied to simulate flow in fractured media due to its ease of

use. This practice results in some severe limitations such as hydraulic head averaging and an inability

to handle preferred fluid pathways. When fractures are few and far between and the fractured block

hydraulic conductivity is low, the EPM approach may not be appropriate. Simulation of flow through

discrete networks is difficult and data intensive (Snow, 1969). For describing groundwater flow in a

fractured environment, porous media models or continuum approach have been used by increasing the

hydraulic conductivity values of cells where fracture flow occurs.

2.3. Recharge

Groundwater recharge is a process of water movement downward through the saturated zone under

the force of gravity or in a direction determined by the hydraulic condition (Simmers, 1988). Natural

recharge of groundwater could be occurring from precipitation, from rivers and canals and from lakes.

As discussed by Simmers et al. (1997), quantifying the current rate of groundwater recharge is a basic

prerequisite for efficient groundwater resource management and is practically vital in arid regions

where such resources are often the key to economic development.

Groundwater recharge quantification is fraught with problems of varying magnitude and hence

substantial uncertainties. It is therefore desirable to always apply and compare a number of

independent approaches. Various techniques are available to quantify recharge; however, choosing

appropriate techniques is often difficult. According to Scanlon et al.(2002), important considerations


12

in choosing a technique include space or time scales, range, and reliability of recharge estimates based

on different techniques. Each of the methods has its own limitations in terms of applicability and

reliability. Techniques of recharge estimation vary based on source and process of recharge

mechanisms. Simmers et al. (1997) indicate that the procedures to quantify recharge from various

sources are direct measurements, water balance methods, tracer techniques and empirical methods. As

it was applied for the estimation of groundwater recharge in semi-arid climate India by Sharda et al.

(2006), a number of methods were used to estimate the recharge. As part of the study, water table

fluctuation and chloride mass balance methods were applied. The water table fluctuation is based on

the principle that the rise in groundwater level in any aquifer is proportional to the water reaching the

water table. The recharge component contributed to groundwater is expressed as:

Rgw = S WWTA∆ (2.1)

Where, S is storativity,WT∆ is change in water table depth, and AW is area of the watershed. The

chloride mass balance method is based on the assumption of conservation of mass between the input

of atmospheric chloride and the chloride flux in the subsurface (Yongxin & Beekman, 2003).

As described by Bear & Verruijt (1987), the basic equation applicable for the estimation of recharge

using chloride mass balance method is:

Rgw=Pyear

gw

p

Cl

Cl (2.2)

Where, Rgw is the annual recharge rate (mm), Pyear is the average annual rainfall (mm) and Clp and Clgw

are the chloride concentrations of the rainfall and groundwater (mg l-1), respectively. The technique

regards chloride as an inert element, and compared with other inorganic ions, it is not added or

removed by water rock interaction. The element is considered as an inert in the hydrological cycle

having its source from the atmosphere. It has the advantage over tracers involving water molecules in

a sense that atmospheric inputs are conserved during recharge processes allowing a mass balance

approach to be used. Commonly water balance method is applied for recharge estimation in many

climatic zones of the world. Estimation of recharge using this method is largely dependent on the

precision with which the water balance components were determined. The application of the water

balance method in arid and sem-iarid regions is more difficult than in humid regions because

precipitation is frequently only slightly different from actual evapotranspiration, small errors in these

two components cause large errors in recharge estimation. Simmers et al. (1997) identified two

different precipitation mechanisms, diffuse and localized recharge. Direct or diffuse recharge results

from wide spread infiltration of rain water at the point of impact whereas localized recharge resulted

from horizontal flows that occurs into local depression that are not connected to any draining water

courses. The same authors discussed the methods available for estimation of groundwater recharge

directly from precipitation include inflow, aquifer response and outflow methods. Lysimeter

measurements, tracers and soil moisture budget models are considered as inflow methods of recharge

estimation. In the aquifer response method of recharge estimation, groundwater level changes are

transformed to the amount of water by using the specific yield concept. In the outflow method of

recharge estimation groundwater recharge and groundwater discharge are considered equal.


13

3. Description of the study area

3.1. Location

Aynalem sub-basin is located in Tigray regional state (northern part of Ethiopia) at about 5kms south

of Mekele town, capital of the regional state. The geographic location of the area is between 39021’ to

39043’East and 13024’to13030’North. Aynalem area is part of the Ethiopian central plateau just to the

west of Afar rift valley, located at about 770 km north of Addis Ababa (The capital city of Ethiopia).

Figure 3.1. Location map of the study area

3.2. Geomorphology and drainage

The study area covers about 104 km2 with a mean altitude of 2200 m above sea level. The altitude of

the catchment varies from 2100 meters above mean sea level at the mouth of the basin to 2540 meters

above mean sea level at the extreme east of the catchment boundary (Fig 3.2). The northern and

southern ends of the catchment are bounded by a chain of dolerite ridges mainly oriented N–W and

the central part is characterized by relatively flat topography of Mesozoic sedimentary terrain. The

Eastern limit of the catchment is physically separated by the dolerite ridge from Afar lowlands.


14

Figure 3.2. Geomorphologic features and elevation cross-section

Aynalem sub-basin is part of the Giba catchment and belongs to the Tekeze drainage system. The

catchment is fed by Aynalem river and a number of ephemeral streams that drains from the adjacent

ridges. The drainage density and pattern is mainly controlled by the geology, topography and

geological structure of the area.


15

Figure 3.3. Drainage map of Aynalem sub-basin

3.3. Climate

The fact that Ethiopia is located in the tropics, latitude 30 N to 180 N, combined with the high range of

altitude, -120 to +4650 meters, and the pressure and air flow pattern determines the tremendous

differences in climate which prevail in different parts of the country (Chernet, 1993). Seasonal

variation in pressure systems and air circulation seem to determine the seasonal distribution of rainfall

in Ethiopia. The distribution of rainfall in Ethiopia is characterized by reference to the position of the

Inter-Tropical Convergence Zone (ITCZ), a low pressure area of convergence between tropical

easterlies and equatorial westerlies along which equatorial wave disturbances take place (Gamachu,

1977). During summer the ITCZ is located in northern Ethiopia. Due to the southeast-northwest axis

of weak high pressure system over Ethiopia, the ITCZ descends southwest in northeast part of

Ethiopia, so that it runs nearly parallel to the Red sea coast. During this time except in south eastern

Ethiopia, the remaining part, including the study area is under the influence of the Atlantic equatorial

westerlies, which produce the main rainy season in Ethiopia (Gamachu, 1977). In spring, the ITCZ is

located in southern Ethiopia. The easterly and south easterly moist currents ascend over the highlands

and produce the main rainy season in south eastern Ethiopia, and light rains of spring to most parts of

the country. The area has a semi-arid climate with little or no variation within the study area with

mean annual rainfall of 670 mm. The long-term mean monthly values of the climatic variables are

presented in the figures below.


16

Mean monthly rainfall at Mekele airport

0

50

100

150

200

250

JAN

FE

B

MA

R

AP

R

MA

Y

JUN

JUL

AU

G

SE

P

OC

T

NO

V

DE

C

Month

Rai

nfal

l (m

m)

Mean monthly relative humidity at 1200 local time

0

10

20

30

40

50

60

70

80

Jan

Feb

Ma

r

Ap

r

Ma

y

Jun

Jul

Au

g

Se

p

Oct

Nov

De

c

Month

Rel

etai

ve h

umid

ity (

%)

Mean monthly wind speed

1

1.52

2.53

3.5

44.5

5

Jan

Fe

b

Ma

r

Ap

r

Ma

y

Jun

Jul

Au

g

Se

p

Oct

Nov

De

c

Month

Win

d sp

eed

(m s

-1)


17

Temperature of the area

0

5

10

15

20

25

30

Jan

Feb

Ma

r

Ap

r

Ma

y

Jun

Jul

Aug

Se

p

Oct

Nov

Dec

Month

Tem

pera

ture

(0 C)

Mean max Mean min Mean

Figure 3.4. Mean monthly values of climatic variables

3.4. Land use, Vegetation and soil

Land use is a major controlling factor in watershed hydrology (Pappas et al., 2008). For example

precipitation that falls on roof tops and pavement results in quick runoff instead of infiltrating into

soil as it would generally do in a natural or farmed landscape. The principal land use in the study area

is rain-fed agriculture and mainly crop farming. Grazing land and settlements occupy a considerable

part of the area (Fig.3.5). Aynalem catchment is sparsely vegetated as a result of excessive

deforestation mainly for agricultural land. The sparse vegetation cover results in excessive soil

erosion and resulting in silt deposition problems in water harvesting structures constructed within the

sub-basin.

WWDSE (2006) conducted classification of the soil type in the catchment based on the grain size

distribution. Accordingly, the dominant soil types identified are classified to four classes: sandy loam,

silty loam, clay loam and clay soils (Fig.3.6). The soil types in the area are strongly related to geology

and geomorphology. The steep cliffs are dominated by sandy loam and silty loam soils, whereas the

clay loam and clay soils cover the flat topography and follow the river banks.


18

Figure 3.5. Land use map of the study area (after Teklay, 2006)

Figure 3.6. Soil map of the study area (WWDSE, 2006)


19

3.5. Geology

Regional geology

According to Mengeasha et al. (1996), the geological units in Ethiopia fall into three major categories.

These are Precambrian basement, late Palaeozoic to early tertiary sediments and Cenozoic volcanic

and associated sediments. During late Palaeozoic to early Mesozoic the northern and eastern parts of

East Africa acted as depositional basins for sediments coming from the higher cratons. This period is

particularly represented by the deposition of the Enticho sandstone and Edagaarbi glacial in Tigray

(Kazmin, 1975). According to this author two major transgression-regression cycles took place during

Mesozoic. It is believed that these cycles are related to major regional tectonic events that have

affected the entire East African region. The Mesozoic sedimentary succession of the Mekele outlier is

the product of transgression-regression cycles and rocks representing a range of sedimentary

environments have been recognized (Bossellini et al., 1997). The first cycle began during early

Jurassic or late Triassic and resulted in the deposition of the Adigrat sandstone consisting mainly of

sandstone and minor lenses of siltstone and Antalo formation consisting mainly fossiliferous

limestone in Tigray Region (Mengesha et al., 1996). The regression of the first phase caused the

deposition of the Agula formation that is constituted of black shale, marl and clay stone with some

beds of black limestone in the Mekele area. Regression of the second phase during late Cretaceous

resulted in the deposition of Amba Aradam formation that is constituted of siltstone, sandstone and

conglomerates. The works of Beyth (1972) and Kazmin (1975), state that these Mesozoic sedimentary

successions unconformably overlies the Precambrian basement, that forms a nearly circular outlier

800 square kilometre, called Mekele outlier. As described by Bosellini et al. (1995), Mekele outlier

consists of a Triassic clastic unit (Adigrat sandstone), Jurassic carbonate-marl-shale succession

(Antalo Suppersequence) and early Cretaceous sandstone (Amba Aradam formation). The stratigraphy

of the Mekele outlier is shown in Figure 3.7. Tertiary flood basalts unconformably overlay the

sedimentary rocks of the area. The basaltic rock around Mekele area is called Mekele dolerite.

According to Levitte (1970) , the dolerite mostly occurs as sills with a thickness of ranging from one

meter to 30 meters and dykes that intruded the sedimentary rocks.

Local geology

Aynalem catchment is located in the central part of Mekele outlier. The main rock formations which

outcropped in the lager part of the study area are shale-limestone intercalation and igneous rocks,

mainly Mekele dolerite, which has intruded into the Mesozoic sedimentary rocks. Stratigraphically,

the shale-lime stone intercalation is the upper most part of the Antalo group. It is composed of shale,

marl and limestone intercalation which rarely contain thin layers of gypsum (Yehdego, 2003). In areas

where there is incoming dolerite intrusion it becomes highly disturbed and the intercalation beds tilted

from horizontal.The dominant sedimentary rock unit outcropping in the study area underlying the

shale and limestone intercalation is the limestone unit. The limestone unit outcrops mainly in the low

land areas of the Aynalem wellfield. Ridge forming dolerite dykes and sill are commonly outcropped

in the study area. The rocks are black, fine to medium grained and in most cases contain phenocrysts

of plagioclase crystals (Hussien, 2000). The existence of dolerite sill is evidenced by the penetration

of the dolerite unit in almost all wells.


20

(1). Marl, (2).Coral-stromatoporoid rich limestone, (3). Cross-bedded sandstone, (4). Limestone,(5). Cross-bedded oolitic

limestone, (6). Transgressive system tract, (7). Highstand system tract and (8). lowstand system tract.

Figure 3.7. Composite stratigraphy of sedimentary succession in Mekele outlier.

(After Bosellini et al., 1995).


21

Hydrogeological setting

Groundwater occurrence is greatly influenced by the geology, topography and climatic factors that

prevailed in a given area. By the same fact the hydrogeologic condition of Aynalem catchment is

mainly controlled by the geology and geological structure. Geological structures (faults, fractures and

lithologic contacts) play a great role in the movement and occurrence of groundwater in the study area

(Gebregziabher, 2003). According to Beyth (1972), the major faults in the area are divided into two

groups: NW trending faults (supposed to be late Jurassic to early Cretaceous age), N-E to NE- SW

trending faults(associated with the rifting phase. These groups of faults are named as Wukro, Mekele,

Chelekot and Fuceamariam fault belts.

Borehole logs and field descriptions of previous works indicate that the main aquifers in the area are

the limestone unit and weathered and fractured dolerite. The works of Hussien (2000) and Yehdego

(2003) confirm that the main aquifer in the area is limestone. The limestone is commonly outcropped

with inter-beds of shale-marl intercalation and rarely with thin gypsum layers particularly in the

western part of the catchment. The highly jointed part of the limestone bed favors groundwater

storage and movement in the area. As discussed by Hussien (2000), the limestone unit has a hydraulic

conductivity of ranging from 29 m day-1 to 74 m day-1. The highest permeability is at the contact

between limestone and dolerite attributed to intense fracturing by the effect of dolerite intrusion.

The dolerite which has a mod of occurrence of dykes and sills can be considered as an aquifer when

it is fractured (Vernier, 1985). However, the dolerite unit generally has low hydraulic conductivity

ranging 0.02 to1m day-1. The dolerite intrusion in the area caused non uniform and complex vertical

and horizontal distribution of the hydrostratigraphic units forming an interbedded system of

permeable and less permeable layers resulted in a confined to semi-confined aquifer system.

The shale-marl unit is fissile and friable in nature. The intrusion of the dolerite into the shale-marl

intercalation resulted in the development of fractures which increases the porosity and permeability.

As reported from wells drilled in the intercalation, the hydraulic conductivity ranges 1 to 2 m day-1.

This unit is considered as an aquitard which acts as a local confining layer.

Water strike records at the time of drilling and water level monitoring data indicate that the

groundwater table varied spatially as well as temporally with water level rising following the rainy

season. The depth of water table ranges from 7 to 51 meters below ground surface.


22


23

4. Analysis and model input data preparation

4.1. Hydrometeorology

Hydrology encompasses the interrelationships of geologic materials and processes with water (Fetter,

2001). According to Dingman (1994), the movement of water on and under the surface is affected by

the physical and chemical interactions with earth materials accompanying that movement. Defining

the hydrologic boundaries, both surface and subsurface, is a crucial step to perform water budget

analysis of a catchment. Generally water flows from the hydrologic boundaries towards a point of

discharge. Fetter (2001) describes the hydrologic inputs to an area and hydrologic outputs from an

area. Hydrologic inputs include precipitation surface water inflow and groundwater inflow, whereas

the hydrologic outputs from an area include evapotranspiration, groundwater and surface water

outflows and artificial abstractions.

Precipitation

As described in section 3.3, the creation and distribution of precipitation in the study area is highly

influenced by the position of the Inter-Tropical Convergence Zone (ITCZ). The rainfall data for the

study area is available since 1960 (Appendix 1.2). The average annual precipitation for Aynalem

catchment is 670 mm with about 80% falling from June to September. The long–term annual rainfall

record shows inter-annual fluctuations with the driest year 1984 (272 mm) and wettest year 1961

(1106 mm).

Long-term annual rainfall at Mekele airport

0

200

400

600

800

1000

1200

19

60

19

62

19

64

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

88

19

92

19

94

19

96

19

98

20

00

20

02

20

04

year

Rai

nfal

l (m

m)

Figure 4.1. Long-term annual rainfall of the study area

Evapotranspiration

The term Evapotranspiration is the total amount of water lost due to the combined effect of

evaporation from the soil and transpiration through the plant leaves. Evapotranspiration could be

explained with the potential and actual evapotranspiration. Potential evapotranspiration is the rate at

which evaporation would occur from a large area completely or uniformly covered with growing

vegetation which has access to unlimited supply of soil water (Dingman, 1994). Actual


24

evapotranspiration is the evapotranspiration at actual field condition. Direct measurement of the

actual evapotranspiration (AET) is difficult, hence it is usually estimated from the potential

evapotranspiration (PET). Teklay (2006) apply the Penman-Monteith method to estimate the

evapotranspiration in the sub-catchment. The estimated evapotranspiration value was 966 mm year-1

which is typical of semi-arid climate.

River discharge

Discharge record of 1992-2001 for Aynalem river is available from the Metere gauging station which

is located at the upper part of the catchment (Appendix 1.10). There is no gauging station at the outlet

of the catchment and thus it was not possible to get river discharge records leaving the catchment. The

Metere gauging station covers only the upper 69 km 2 of the total catchment area of about 104 km 2.

River flow of about 4 m3 s-1 was measured during the fieldwork in August 2007 (at the peak raining

month of the year) right at the outlet of the catchment. But to understand the river flow in response to

different climatic forcing, time series records of the river flow is quite necessary. In the lower part of

the catchment which is down stream of the gauged part there are a number of springs that fed the

river. Thus it is believed that the base flow contribution to the river flow is higher in this part of the

catchment. The average monthly river flow of the Aynalem river at Metere gauging station for the

monitoring period (1992-2001) is shown below.

Table 4.1. Monthly river discharge (Mm3) of Aynalem river

Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Q (Mm3) 0.01 0.00 0.00 0.01 0.02 0.05 0.72 1.63 0.54 0.08 0.05 0.03

Hydrograph of Aynalem River at upper part

0.0

0.5

1.0

1.5

2.0

Jan

Fe

b

Ma

r

Ap

r

Ma

y

Jun

Jul

Au

g

Se

p

Oct

No

v

De

c

Month

Dis

char

ge (

Mm

3 )

Figure 4.2. Hydrograph of Aynalem river

Due to the fact that intense rainfall showers occur while there is sparse vegetation cover during the

raining season, frequent flash floods are generated after rainfall events. As it can be observed from the

hydrograph, the stream flow reacts rapidly to rainfall and the peak of the discharge corresponds to the

raining months of July and August and declines sharply in October.


25

4.2. Hydrochemistry

Hydrochemistry of groundwater aquifer in a region is largely determined by both the natural

processes, such as precipitation, wet and dry depositions of atmospheric salts, evapotranspiration,

soil/rock–water interactions, and the anthropogenic activities, which can alter these systems by

contaminating them or by modifying the hydrological cycle (Singh et al., 2007). Both the natural

processes and the anthropogenic activities vary in time and space. These variations are reflected in

groundwater hydrochemistry variations showing spatial and temporal fluctuations in a region. The

chemical composition of groundwater is the combined result of the composition of water that enters

the groundwater reservoir and the reactions with minerals present in the rock that may modify the

water composition (Appelo & Postma, 1992). Dissolved constituents in the water provide clues on its

geologic history, its influence on the soil rock masses through which it has a pass, the presence of

hidden ore deposits, and its mode of origin within the hydrologic cycle (Freeze & Cherry, 1979). As

mentioned by Anderson & Woessner (1992), water chemistry data can be used to infer flow

directions, identify sources and amount of recharge and to define local and regional flow systems.

4.2.1. Water sampling and analysis

Representative water samples of boreholes, springs and ponds were collected from Aynalem, Ilala

and Chelekot sub-basins for the analysis of major anions and cations. The well location map of the

Aynalem wellfield is given in Appendix 6 as borehole identification will be used regularly in the

subsequent sections. The location of the water samples are indicated in Fig. 4.3 as Upper Ilala, Lower

Ilala, Lower Aynalem, upper Aynalem and Chelekot. Furthermore rain water samples were collected

from the area to determine chloride concentration in rainfall for the application of chloride mass

balance method of recharge estimation. The groundwater samples were analysed in the Central

Laboratory of the Ethiopian Geological Survey. For comparison purpose, control water samples from

boreholes that are representative of the lower catchment (TW2) and the upper catchment (PW8) were

brought to ITC laboratory and the analysis results from both laboratories were compared (Appendix

2.3). The analysis results for most of the groundwater constituents were comparable in both

laboratories. But there are also differences in the analysis results mainly for the chloride and nitrate

content.

4.2.2. Reliability check

To evaluate the data quality, the accuracy of the water analysis was checked with the anion-cation

balance. The principle of the anion–cation balance is that the sum of cations and sum of anions are

equal because the solution must be electrically neutral. In a electrically neutral solution, the sum of

the cations should be equal to the sum of anions in meq l-1 (Hounslow, 1995).

( ) 100*%∑ ∑∑ ∑

+−

=AnionsCations

AnionsCationstralityElectroneu (4.1)


26

Figure 4.3. Location of water sample points

Based on the electroneutrality, analysis of water samples with a percent balance error <5% is

regarded as acceptable (Fetter, 2001). But in very dilute or saline water, up to 10 % error may be

considered as acceptable due to the errors introduced in measuring major ions in dilute groundwater

or in the multiple dilution require for analysis of concentrated groundwater. The analysis result of all

the samples is within the acceptable range of the reliability check of electroneutrality. The cations-

anions balance results are found to be reliable as the balance does not deviate from the 5% criterion

(Appendix 2.4). The analysis results of the water samples indicate that the dominant dissolved cations

in the groundwater of the area are Ca2+, Na+, and Mg2+ with lower levels of K+. And the major

dissolved anions in the groundwater include: SO42- HCO3

- and Cl-. The range and mean of the major

inorganic constituents of groundwater samples from Aynalem and nearby catchments are summarised

below.


27

Table 4.2. Summary statistics of the major groundwater constituents

Constituent Minimum Maximum Mean

EC ( µS cm-1) 688.0 2540.0 1267.2

TDS ( mg l-1) 447.0 2068.0 962.1

Total Hardness 308.7 1486.8 667.2

pH 7.3 7.9 7.5

Ca2+ (mg l-1) 109.2 502.2 244.6

Mg2+ (mg l-1) 1.9 44.9 12.6

Na+ (mg l-1) 18.0 81.0 37.0

K+(mg l-1) 1.0 4.3 2.3

HCO3- (mg l-1) 14.6 351.4 248.0

Cl- ( mg l-1) 12.5 73.3 24.9

SO42- (mg l-1) 47.5 1100.0 427.7

4.2.3. Presentation of results

The water analysis result of the major anions and cations are plotted in Piper diagrams and Stiff

diagrams for quick and tentative conclusion of the water type. According to Hounslow (1995), the

position of an analysis that is plotted on a piper diagram can be used to make tentative conclusion as

to the origin of the water represented by the analysis. However, the bicarbonate to silica ratio must

also be considered when making this deduction. The analysis results are also plotted using Stiff

diagrams in which, cations are plotted in meq l-1 on the left of the zero axis and anions are plotted on

the right. As discussed in Fetter (2001), Stiff diagrams are useful in making a rapid visual comparison

between water from different sources. Selected representative water samples from the three

catchments are presented in the plots of Piper and Stiff diagrams (Fig 4.4 to 4.7).

4.2.4. Water type

The major inorganic constituents of water originate when water in precipitation dissolves atmospheric

gasses such as carbon dioxide and reacts with minerals on the surface of the earth (Hounslow, 1995).

The major water types identified from the hydrochemistry analysis of the groundwater samples are

Ca-SO4, Ca-HCO3-SO4, Ca-SO4-HCO3 and Ca-HCO3. The water type in the basin and its

surroundings is not uniform in composition. Lower Aynalem, Chelekot and lower Ilala are dominated

by a Ca- SO4 type of water whereas the upper Aynalem and upper Ilala are dominated by a Ca-HCO3

and Ca -SO4- HCO3 type of water. The chemical ions of the water samples plotted on a piper diagram

show that the major cations composition has a limited range of variation, mainly calcium rich. On the

other hand, the anions composition has a range of variation from sulphate to bicarbonate. As it can be

clearly observed from the stiff pattern, there is variation in groundwater chemistry in the water

samples from upper and lower Aynalem, Ilala and Chelekot. It is possible to relate the variation in

groundwater chemistry with a change in lithology and groundwater flow direction. From the drilling

well logs at the lower Aynalem, there is evidence for the existence of a gypsum layer which attributes

to the Ca-SO4 dominating water chemistry.


28

Figure 4.4. Piper diagram of water samples from boreholes

Figure 4.5. Stiff diagrams of water samples from upper Aynalem


29

Figure 4.6. Stiff diagrams of water samples from lower Aynalem

Figure 4.7. Stiff patterns of water samples from Ilala and Chelekot


30

4.2.5. Source rock deduction

The purpose of source rock deduction is to gain insight into the possible origin of water analysis. The

initial composition of groundwater originates from rainfall which may be considered to be diluted sea

water (Hounslow, 1995). During its return path to the ocean, the water composition is altered by rock

weathering. During rock weathering the major cations and anions are added to the water. The source

rock deduction for the study area is carried out with the help of AQUACHEM software. Water

samples are selected from different areas within the catchment and out of the catchment for the source

rock deduction analysis. The objective of the selection of samples from different areas is to compare

the source rock within the different sub-basins and to see whether there is a connection of Aynalem

groundwater sub-basin with adjacent river basins. As there is no silica analysis result in any of the

samples, simple ionic comparisons are used for the analysis of source rock deduction. The ionic

comparisons applied in the source rock deduction analysis are:

• Na/ (Na+ Cl)

• Ca/ (Ca+SO4)

• TDS

• Cl/Sum of anions

• HCO3/Sumof anions

This simplistic mass balance approach to deduce the source rock is not perfect, but it can be very

helpful in understanding the origin of the groundwater. Based on the Na / (Na+ Cl) ionic ratio, the

water samples show sodium source other than halite (albite, ion exchange) except for the water

sample from upper Ilala (which shows reverse softening). Applying Ca/(Ca+SO4) ionic ratio indicate

that lower Ilala and Chelekot resulted from gypsum dissolution whereas the water samples from

upper Ilala and Aynalem sub-catchment show calcium source other than gypsum (carbonate or

silicate). The source rock deduced based on the TDS is carbonate weathering or brine except for the

upper Aynalem (from silica weathering). And Cl/ (Sum Anion) ionic ratio shows that all of the

samples are result from rock weathering. On the basis of HCO3/ (Sum of anions), the water samples

result from silicate or carbonate weathering and gypsum dissolution. The ionic ratio analysis show

that the source rock of Aynalem is relatively different from the adjacent sub basins. That is the water

from Ilala and Chelekot is rich in gypsum as compared to the Aynalem sub basin. This is supported by

the fact that the water from boreholes of Ilala has a high hardness while the hardness of the water

from Aynalem especially from the upper part is much lower.


31

Table 4.3. Parameters used for source rock deduction

Sample name

Na/(Na+Cl)

(meq l-1)

Ca/(Ca+SO4)

(meq l-1)

TDS

(mg l-1)

Cl/SumAnion)

(meq l-1)

HCO3/(SumAnion)

(meq l-1)

Lower Ilala

0.747 0.552 2068 0.114 0.821

Upper Ilala

0.439 0.805 708 0.140 0.618

Upper Aynalem

0.690 0.846 447 0.055 0.819

Lower Aynalem

0.709 0.614 812 0.077 0.505

Chelekot 0.616 0.551 1380 0.090 0.256

4.3. Chloride mass balance method (CMB)

Groundwater resource studies require the estimation of the quantity of water moving downwards from

the soil zones as a potential recharge (Rushton et al., 2006). The methodology selected for the

estimation of recharge should be applicable in a wide variety of climatic and hydrologic situations. In

this study, the Chloride Mass Balance method is applied to estimate the groundwater recharge.

According to Simmers et al. (1997), chloride is the most important environmental tracer and has been

used to estimate rates of groundwater recharge under a wide range of climatic, geologic and soil

conditions. Yongxin & Beekman (2003) added that the chloride mass balance method was applied for

recharge estimation worldwide in recent time. The basic equation used to calculate the annual

groundwater recharge with the assumption negligible chloride dry deposition in an area is based on

equation 2.2. Despite the fact that the method is simple and inexpensive, there are a number of

uncertainties associated with the method in estimating recharge.

In most cases, the long-term average chloride in rainfall is not available. Measured atmospheric input

of chloride, often only short term records of chloride is assumed to be representative for a long period.

But an area of concern as rainfall and chloride deposition during the past may be different from today.

As discussed by Yongxin & Beekman (2003), other areas of concern include the uncertainty in the

measured chloride content of rainfall and rainfall amount. The largest uncertainty associated with

recharge estimation that utilises the chloride mass balance approach is the determination of chloride

concentration in the rainfall. Furthermore rainfall amount is generally difficult to measure, and is

highly variable. The absence of long-term rainfall quality data in the present study is one of the main

limiting factors affecting the accuracy of the method. Another uncertainty source for the chloride

mass balance approach is the sampling density and analysis accuracy of the chloride concentration of

the groundwater. As part of the fieldwork, samples from rainwater and groundwater were collected

and analysed for their chloride content that are utilised in the chloride mass balance method of

recharge estimation.


32

4.3.1. Chloride in rainwater

The chloride concentration in the rainwater of the study area has a low detection limit. As a result,

rain water samples that are collected from Mekele and Bahrdar in August 2007 are sent to a special

geochemical laboratory in Utrecht, the Netherlands for the determination of chloride concentration.

Since the rainfall in Bahrdar and Mekele has the same origin (ITCZ), the data set from Bahrdar in

combination to the rainfall water sample from Mekele is used to analyse the standard deviation from

the mean value of the chloride concentration in rain water.

Table 4.4. Chloride concentration in rain

Time of sampling Chloride concentration ( mg l-1) Station

June1-17,2002 0.43 Bahrdar

June17-30,2002 0.30 Bahrdar

July 1-15,2002 0.65 Bahrdar

July 1-15,2002 0.61 Bahrdar

August 9,2007 0.52 Bahrdar

August 10+11,2007 1.42 Bahrdar

August 12,2007 0.62 Mekele

August 12,2007 0.66 Bahrdar

Standard deviation 0.2

Average Cl-1 rain for the samples taken in

August 2007 0.8 ± 0.2 mg l-1

4.3.2. Chloride content in groundwater

Thirty three groundwater samples were collected and analysed for their chloride content. The chloride

content of the collected ground water samples ranges from 10 to 81 mg l-1.

Table 4.5. Statistics of the chloride concentration in groundwater

Chloride concentration in the collected groundwater samples (mg l-1)

Minimum Maximum Arithmetic mean Standard deviation

10 81 24 11

As it is shown by the standard deviation of the chloride concentration, the variability of the chloride

concentration in the groundwater of the area is considerable and this affects the value of the estimated

recharge. Groundwater chloride concentrations may originate from various flow components in the

unsaturated zone, thus the recharge calculation by chloride mass balance gives an average long-term

estimate of recharge. The method has several shortcomings, one of which is that the method can not

be used in environments affected by other sources of chloride other than total atmospheric fallout.

Thus the following assumptions are made in applying the method.

• Precipitation is the only chloride source in groundwater

• Chloride is conservative and will not undergo any chemical reaction with the geologic

material


33

According to Eriksson (1985), the average groundwater chloride content should be calculated as

harmonic mean and is given by equation 4.2.

Clgwave

∑=

N

i gwCl

N

1

1 (4.2)

Where

Clgw is the individual chloride concentration of samples (mg l-1)

N is the total number of observations

Based on the collected groundwater samples, the harmonic mean of the chloride content in the

ground water of Aynalem sub-basin is 18 mg l-1.

Summarizing the above results as:

• Average chloride concentration in rain water (0.8 mg l-1 )

• Harmonic mean of chloride content in groundwater (18 mg l-1)

• Average annual rainfall (670mm)

The estimated recharge is 30 mm year-1 which is 4.5% of the average annual rainfall in the area.

Given that input chloride concentrations can vary significantly from site to site within a region of

investigation, it is not surprising that CMB estimations are site specific (Yongxin & Beekman, 2003).

To see the spatial distribution of the recharge in the sub-basin, the average recharge is estimated at

each sample point as indicated in table 4.6.

The estimated recharge value using the chloride mass balance method is sensitive to concentration

variability of the chloride content in the groundwater and rain water. Taking the maximum and

minimum values of chloride content both in rain and groundwater into consideration, an average

recharge of 37mm year-1 which is 5.5% of the average annual rainfall, is estimated while the average

recharge estimated by applying the harmonic mean of the chloride content in the groundwater samples

and the average chloride content in rain water is 30 mm year-1 (4.5% of the annual rainfall). Thus the

estimated recharge may range between 30 to 40 mm year-1 depending on the range of chloride

concentrations in both rain and groundwater.


34

Table 4.6. Groundwater chloride content and estimated recharge

UTM East UTM North Clgw (mg l-1) Cl rain (mg l-1) Annual

rainfall (mm)

Recharge

(mm year-1)

558941 1489255 22.20 0.81 670 24.45

558284 1489689 42.40 0.81 670 12.80

557809 1488359 12.50 0.81 670 43.42

556722 1487915 18.30 0.81 670 29.66

558268 1488286 12.50 0.81 670 43.42

557115 1487967 17.40 0.81 670 31.19

553941 1488821 20.30 0.81 670 26.73

552945 1488663 17.40 0.81 670 31.19

552219 1488072 22.20 0.81 670 24.45

552590 1488475 17.40 0.81 670 31.19

552313 1488491 19.30 0.81 670 28.12

552490 1489376 18.30 0.81 670 29.66

552506 1489646 19.30 0.81 670 28.12

559953 1484601 64.60 0.81 670 8.40

564070 1486356 9.70 0.81 670 55.95

554920 1480973 34.70 0.81 670 15.64

552650 1485112 81.10 0.81 670 6.69

557115 1487960 12.48 0.81 670 43.49

553706 1488251 13.44 0.81 670 40.38

551965 1487745 56.64 0.81 670 9.58

553320 1488680 15.40 0.81 670 35.24

560968 1487142 20.00 0.81 670 27.14

558901 1489740 19.30 0.81 670 28.12

555901 1486423 36.70 0.81 670 14.79

556519 1487277 11.60 0.81 670 46.78

553549 1488948 17.28 0.81 670 31.41

555526 1487648 15.36 0.81 670 35.33

559405 1487676 22.10 0.81 670 24.56

556050 1487809 17.28 0.81 670 31.41

557487 1488269 15.40 0.81 670 35.24

561072 1484157 15.40 0.81 670 35.24

559545 1480345 21.20 0.81 670 25.60

552160 1485603 36.48 0.81 670 14.88


35

4.4. Well abstraction and groundwater level analysis

4.4.1. Well abstraction

Groundwater has been abstracted from the wellfield since 1999 for the water supply of Mekele town.

At present there are about ten production wells in operation. Additional boreholes are under

construction and the wellfield is expanding to the upstream part of the sub-basin. The pumping

boreholes are concentrated in the lower right bank of the catchment. The abstraction rate of the

boreholes for the water supply of Mekele town indicates that most of the boreholes are not

continuously functional throughout the year. For this study, the average abstraction rate from the

wellfield is determined based on the continuously pumped wells. A significant amount of water is

abstracted from a well implemented in 2005 (TW4-2005). Thus the abstraction rate of this borehole is

considered in the total average of the abstraction rate. Moreover, the abstraction rate of one borehole

(Aviation) which is not owned by the Water Supply Office is considered in the average rate of

abstraction from the wellfield. Based on the abstraction records from the Water Supply office and by

considering the above two boreholes, the average daily abstraction from the wellfield is estimated to

be 7156 m3 day-1. The abstraction rate from each well is summarised in Table 4.7 and the four year

monitoring abstraction data is indicated Appendix 3.2. Table 4.7. Daily maximum abstraction rate from the wellfield

WELL ID UTM E UTM N ALTITUDE (m) Q (m3day-1)

TW-4 (2005) 557116 1487879 2206 1730.4

PW-2 556637 1487727 2224 1045.8

PW-3 553848 1488600 2214 528.2

PW-4B 553608 1488037 2212 472.0

PW-7B 557020 1487774 2233 1107.0

PW-8 557809 1488359 2237 1478.4

PW-11 552379 1489188 2194 161.8

May shibti 554419 1489568 2227 129.9

Abo Tareke 552315 1488492 2175 449.9

Aviation 556596 1488532 2229 52.6

Total daily

production

7156

The abstraction rates of five boreholes that have a continuous record for about three years are

presented in the figure 4.8 below, to illustrate trend of abstraction.


36

Trend of monthly water production

0

10000

20000

30000

40000

50000

60000

70000Ju

l

Aug

Se

p

Oct

No

v

Dec

Jan

Feb

Ma

r

Ap

r

May

Jun

Jul

Aug

Se

p

Oct

No

v

Dec

Jan

Feb

Ma

r

Ap

r

May

Jun

Jul

Aug

Se

p

Oct

No

v

Dec

2004 2005 2006Month

Pro

duct

ion

(m3)

PW3 PW7B PW8 PW11 PW2 Figure 4.8. Groundwater abstraction from selected boreholes

Figure 4.9. Location map of pumping wells

Though, there is no complete production record for all wells that abstract water from the wellfield, a

total average monthly production of about 200000 m3 month-1 is estimated as indicated in Fig 4.10.


37

Total production for the monitored period

0

50000

100000

150000

200000

250000

300000

Jul

Au

Se

pO

ctN

ov

De

cJa

nF

eb

ma

rA

pr

Ma

yJu

nJu

lA

ug

Se

pO

ctN

ov

De

cJa

nF

eb

.m

ar

Ap

rM

ay

Jun

Jul

Au

gS

ep

Oct

No

vD

ec

Jan

Fe

bm

ar

Ap

ril

Ma

yJu

nJu

lA

ug

Se

pO

ctN

ov

De

cJa

nF

eb

ma

rA

pr

Ma

yJu

n

2003 2004 2005 2006

Month

Pro

duct

ion

(m3 )

Figure 4.10. Total production of the wellfield

4.4.2. Groundwater level analysis

Fluctuation of groundwater level in an aquifer is complex and dynamic and is a result of dynamic

responses of the groundwater system to recharge and groundwater abstractions from the system. Three

and half years of groundwater level monitoring data were collected from the Water Supply office of

the region. A trend of groundwater level in selected boreholes which have continuous records for

about three years and the average groundwater level from all wells for the monitoring season are

presented in the figure 4.11 and 4.12 respectively.

Groundwater level at selected boreholes

0

10

20

30

40

50

60

Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec

2003 2004 2005

Wat

er le

vel b

gl (

m)

PW6 (not pumped) PW8 PW11

Figure 4.11. Groundwater level at selected boreholes


38

Average groundwater level for the monitored period

0

10

20

30

40

50

60

Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

Aug Sep Oct

Nov

Dec Jan

Feb

Mar

Apr

May Jun

Jul

2003 2004 2005 2006

Gro

undw

ater

leve

l bgl

(m

)

0

50

100

150

200

250

300

350

Month

Mon

thly

rai

nfal

l (m

m)

Figure 4.12. Average groundwater level trend

As it can be observed from the average groundwater level graph, there is a general groundwater level

decline in response to the ongoing groundwater abstractions. Following the rains of July and August,

there is a groundwater level rise in response to the direct recharge from rainfall. Based on the

groundwater level monitoring data, the average groundwater level is declining at a rate of 4.8 per year

or about 16.6 meters for the monitored period.

4.5. Pumping test

Detailed knowledge of the distribution of hydraulic parameters in the subsurface is a prerequisite for

the solution of many problems in hydrogeology and related fields (Leven & Dietrich, 2006). In

practice, several investigation techniques are commonly employed in order to estimate the distribution

of hydraulic parameters such as hydraulic conductivity, transmissivity, and storage coefficients.

Depending on the hydrogeologic situation and the investigation objectives, short-term or long-term

pumping tests are utilized. The principle of a pumping test is that if we pump water from a well and

measure the discharge of the well and drawdown in the well and in piezometers at a known distances

from the well, we can substitute the measurements into an appropriate well flow equation and can

calculate the hydraulic characteristics of the aquifer (Kruseman & de Ridder, 1991). There are several

equations and associated theoretical models or curve types which have been developed to analyse

pumping test results and choice of the equations depends on the matching of the drawdown pattern of

the well with theoretical models.

In the Aynalem case single well pumping tests at a number of boreholes followed by a few pumping

tests with observation wells were conducted at the implementation time of the wellfield in 1998/99.

Pumping tests were also conducted in the test wells drilled later in 2005 and 2006. For the present

study it was only possible to get raw data on continuous pumping test for the test well boreholes

drilled later in the years of 2005 and 2006. As part of the present work, the collected pumping test

data were analysed to understand the aquifer system behaviour of the area. As there were no reported

step drawdown test data, it was not possible to describe the behaviour response of each well under

varying pumping rates. The test wells which were used for the continuous pumping test analysis

include:


39

• TW1 (2005) and TW2 (2005) that are located upstream of the current wellfield where

there is less effect of the ongoing pumping.

• TW4 (2005) located in the upper clusters of the pumping wells where there is intensive

pumping.

• TW6 (2006) is located near the outlet of the catchment for which there is less effect of

pumping.

The geographic locations of these wells are indicated in the well location map (Appendix 6).

Table 4.8. Details of pumping test on the test wells

Well name SWL bgl

(m)

Constant

discharge

(l s-1)

Duration of

the test

(hours)

Drawdown

at the end of

the test (m)

Date of test Aquifer

thickness

(m)

TW1(2005) 18.15 36.7 72 0.50 27/02/2006 65

TW2 (2005) 6.30 30.0 72 18.34 02/02/2006 72

TW4 (2005) 53.35 25.0 144 11.93 15/01/2006 54

TW6 (2006) 16.75 6.0 18 12.55 25/05/2006 40

To identify the aquifer system which in turn helps to select appropriate model for calculating the

hydraulic characteristics, semi-log plots of drawdown versus time were constructed in each of the

wells as indicated in the figures 4.13 through 4.16. As can be observed from the time drawdown plots,

the stability of the groundwater level was not yet reached when the tests was terminated. The pumping

in most of the wells has less influence at the initial stage but the drawdown increases with pumping

time. The drawdown behaviour as result of pumping is influenced by the type of aquifer and inner and

outer boundary conditions at different times during the test.

Time drawdown plot of TW1 (2005)

0

0.1

0.2

0.3

0.4

0.5

0.6

1 10 100 1000 10000

Log time (minute)

Dra

wdo

wn

(m)

Figure 4.13. Time drawdown plot of TW1 (2005)


40


0

4

8

12

16

20

1 10 100 1000 10000

Log time (minute)

Dra

wdo

wn

(m)


Time dradown plot of TW4 (2005)

0

2

4

6

8

10

12

14

1 10 100 1000 10000

Log time ( minute)

Dra

wdo

wn

(m)



0

2

4

6

8

10

12

14

1 10 100 1000 10000

Log time (minute)

Dra

wdo

wn

(m)



41

The common interpretation of pumping test data is based on the assumption that only one aquifer is

pumped and tested. However, the intensive pumping in nearby wells in the wellfield during the test

causes a significant local pressure drop which may cause deviation from the theoretical curves and

may affect the interpretation results considerably. This condition is clearly shown in TW4 (2005),

more likely, the jump in drawdown prevailed at later time of the test is a result of pumping in nearby

wells at the time of pumping test or jumps in the abstraction rate at the well itself during the test. The

time drawdown plot of TW6 (2006) shows an abrupt change of drawdown behaviour which may be

resulted from barrier boundary, fractured storage or change in abstraction rate. Generally it is apparent

that the various combination and amount of groundwater storage in vicinity of the well can result in

almost any type of time drawdown curves. Deviation from theoretical curves is usually due to specific

boundary condition including partial penetration of the well, recharge or impermeable boundaries

(Kruseman & de Ridder, 1991). When the cone of depression reaches recharge boundary, the

drawdown in the well stabilises and impermeable boundary has the opposite effect on the drawdown.

The degree of fracturing also may play a role in the shape of the time drawdown plot. In evaluating

time drawdown curves, a feature that is very common in wells that abstract water from fracture system

is a high or moderate initial yield that decreases rapidly with time. Usually the cause is insufficient

storage of groundwater in the vicinity of the well (Davis & DeWiest, 1966). As discussed by

Kruseman & de Ridder (1991), two systems are recognised in aquifers with double porosity; the

fractures of high permeability and low storage capacity and the matrix blocks of low permeability and

high storage capacity. A characteristic of the flow in such a system consists of:

• Early pumping time, when all the flow comes from the storage in the fracture.

• Medium pumping time, a transition period during which the matrix blocks feed their

water at an increasing rate to the fractures, resulting in a partly stabilizing drawdown.

• Late pumping time, when the pumped water comes from storage in both the fractures

and the matrix blocks.

The drawdown behaviours revealed in the wells are compared with that of the various theoretical

curves to identify the aquifer type. The diagnostic plots of time drawdown plots of the wells under

consideration resemble a confined fractured aquifer of the double porosity type. Cooper & Jacob time

drawdown method (confined aquifer) was applied to the late time data plot to calculate the

transmissivities in the wells. For the analysis purpose AQUITEST software was utilised and the

transmissivity values obtained were 1720, 118, 65 and 74 m2 day-1 for TW1 (2005), TW2 (2005), TW4

(2005) and TW6 (2006) respectively. The curve matchings used for the mathematical solution to

obtain the transmissivities are given in Appendix 8. The interpretation results of the transmissivity

values of these boreholes from the present work are comparable with the previously reported

transmissivities from the same boreholes (Table 4.9).

The boreholes utilised for the present study are only four and they do not fully represent the aquifer

system. Thus the pumping test data are compiled from previous studies mainly from the works of

Hussien (2000), Yehdego (2003), WWDSE (2006) and THAL (2007) in order to characterise the

aquifer system and to prepare model input data. Yehdego (2003) discussed the well interference from

the pumping tests with observation wells. TW4 was pumped at a rate of 10 ls-1 for 72 hours with

observation wells of TW3 and TW5, which are 400 and 290 meters away respectively. The water

level at TW5 started dropping in less than five minutes after pumping was started and it had dropped


42

2.93 meter by the end of 72 hours. On the other hand TW3 was not affected for over 24 hours after

pumping started and the water level dropped 0.35 meter by the end of 72 hours of pumping. Well

interference test of TW4 and TW5 show that there was high degree of interference which was

observed in less than five minutes after the start of pumping.

4.6. Aquifer characteristics

Quantitative description of aquifers is vital in order to address several hydrological and

hydrogeological problems. Fluid transmissivity, hydraulic conductivity and aquifer depth are

fundamental properties describing subsurface hydrology. The most effective way of estimating

hydraulic properties of an aquifer is the pumping tests that are carried out on certain boreholes sites.

Nevertheless, a probable sparse spatial distribution of the available boreholes gives rise to significant

problems in modeling the hydrogeological systems. The estimation of hydraulic properties is usually

done by fitting with a relevant type curves. The interpretation results of the aquifer parameters are

reported in WWDSE (2006) and are summarised in able 4.9 for the purpose of the present study.

As is evident from the reported analysis result, the area is characterized by a highly variable and wide

range of transmissivity and hydraulic conductivity. The hydraulic conductivity ranges from 0.02 to

81 m day-1. In some of the wells the hydraulic conductivity is extremely high and in others it is very

low which is indicative of the heterogeneous condition of the subsurface geologic system. The

variability in the hydraulic properties mainly results from the intense fracturing and heterogeneity due

to the existence of dolerite dykes. As it is indicated in the plot of point hydraulic conductivities at the

wells (Fig.4.17), the hydraulic conductivity is not uniform over the catchment. The locations of the

wells applied to illustrate the point hydraulic conductivities are shown in the well location map

(Appendix 6).

Log hydraulic conductivity

0.00

0.50

1.00

1.50

2.00

2.50

PW

-12

PW

-3

PW

-11

PW

-1

Adi

se

lest

e

PW

-4

PW

-6

PW

8

PW

-2

PW

9

PW

-7

TW

1(2

005

)

TW

2(2

005

)

TW

4(2

005

)

TW

5(2

005

)

AR

-1

AR

-2

Le

spe

r

TW

4

TW

5

TW

6(2

006

)

TW

1

Well name

Log

K(m

day

-1)

Figure 4.17. Log hydraulic conductivity values


43

Table 4.9. Transmissivity and hydraulic conductivity

Well Id UTM

east

UTM

north

Elevation

(m)

Transmissivity

(m2 day-1)

Aquifer

thickness (m)

Hydraulic

conductivity

(m day-1)

PW-12 553549 1488948 2208 1138 14 81.30

PW-3 553941 1488821 2214 2630 53 49.70

PW-11 552490 1489376 2208 967 30 74.40

PW-1 556050 1487809 2211 138 30 4.60

Adi seleste 555901 1486423 2252 217 12 18.10

PW-4 553706 1488251 2210 92 30 3.10

PW-5 554336 1487216 2189 1 45 0.02

PW-6 555526 1487648 2221 2260 30 74.00

PW-8 557809 1488359 2237 500 34 14.70

PW-2 556722 1487915 2227 24 9 2.82

PW-9 558268 1488286 2243 51 16 3.20

PW-7 557115 1487967 2233 888 20 44.40

TW1(2005) 561057 1487352 2277 1750 66 26.68

TW2(2005) 564439 1485877 2311 100 72 1.39

TW4(2005) 557234 1488028 2228 100 54 1.85

TW5(2005) 552970 1488153 2206 170 60 2.83

AR-1 556406 1488604 2215 409 14 29.20

AR-2 555787 1489875 2256 723 18 40.20

LESPER 551526 1487025 2143 27 24 1.14

TW-4 553140 1488452 2178 24 19 1.28

TW-5 552880 1489018 2183 127 28 4.54

TW6(2006) 549453 1485160 2133 65 40 1.60

TW-1 553845 1487586 2193 23 19 1.25

Table 4.10. Summary statistics of transmissivity (m2 day-1)

Mean Median Mode Standard Deviation Minimum Maximum

540 138 24 753 1 2630

As shown in the summary statistics of the reported well transmissivities (Table 4.10), the

transmissivity ranges from 1 to 2630 m2 day-1.

4.7. Digital elevation model (DEM)

The Advanced Space borne Thermal Emission and Reflection Radiometer (ASTER) on-board

NASA’s Satellite Terra is a high-resolution multi spectral sensor that provides along-track stereo

image data of the Earth in the near-infrared wavelength region. A DEM is extracted from Level 1A

product of ASTER using LPS extension in ERDAS IMAGINE to a 15 meters resolution. Ground

control points are prepared from the topographic maps of the study area to assess the vertical accuracy

of the DEM. The vertical accuracy of the DEM is assessed by comparing the extracted elevation value

at a number of check points with those prepared from the topographic maps as ground control


44

elevation points. The relationship between the ground elevation and the elevation from ASTER DEM

is shown in figure 4.18. The elevations from ASTER DEM and those from the topographic map show

high correlation with R2 of 0.99 and the comparison at the check points for the study area results in a

root mean square error of 13 meters. Before applying the elevation extracted from the DEM for

further analysis, the elevation is corrected by the regression equation obtained by the built comparison

to the ground control point elevations and the extracted elevation. The regression equation obtained

from the comparison is X = (Y- 47.882)/ (0.9746), where Y is the elevation from ASTER and X is the

elevation from Topomap of the area. The correction process of the elevation from the ASTER is

accomplished by applying this formula in the Map calculation of ILWIS. After correction, the

required array of elevations at the well locations is extracted by applying the map value function of

ILWIS to the ASTER DEM.

Topo elevation vs ASTER DEM elevation

y = 0.9746x + 47.882

R2 = 0.99

1700180019002000210022002300240025002600

1700 1800 1900 2000 2100 2200 2300 2400 2500 2600

Topo elevation(m)

AS

TE

R D

EM

ele

vatio

n(m

)

Figure 4.18. Scatter plot of ground elevation Vs elevation from ASTER DEM

The corrected Elevation from the ASTER DEM is applied to define the ground elevation of the

boreholes, to define the aquifer top and bottom, to construct cross-sections and to define elevation of

the bottom of the riverbed for the use of river package. Furthermore the DEM plays a key role in

defining of the model boundary in the conceptual model formulation.


45

5. Conceptual model

The first step in the procedure of modeling is the construction of a conceptual model of the problem

and the relevant aquifer domain. Development of a conceptual model is one of the critical steps in

modeling process. It consists of a set of assumptions that reduce the real problem and the real domain

to simplified versions that are acceptable in view of the objective of the modeling. It is critical that the

conceptual model is a valid representation of the important hydrogeological conditions.

Conceptualization of the system is the basis of the numerical modeling. The nature of the conceptual

model determines the dimension of the numerical model and the design of the grid. Failures of

numerical models to make accurate predictions can often be attributed to errors in the conceptual

model. As discussed by Yihdego (2005), the purpose of developing a conceptual model is to formulate

a better understanding of a site condition, to define the groundwater problem, to develop a numerical

model and to aid in selecting a suitable computer code. The elements of a conceptual model include

defining the extent and characteristic of the aquifer system and developing an understanding of

groundwater flow directions, sources and sinks. The conceptual model of the area is developed by

integrating the available data on hydrostratigraphy, well and geophysical logs, geologic map and

geologic cross–section from previous studies.

5.1.1. Well log data and geology

Well log data has a great importance to develop better understanding of subsurface aquifers and

groundwater flow direction. In other words, the information from well log data contributes for proper

characterization of the hydrogeological condition at a site which is necessary to understand the

relevant flow process. Drilling log data (Appendix 3.5) of the boreholes in the wellfield are collected

from different drilling organizations and applied to demarcate the hydrostratigraphic zones of the

area. The well log data show that the geologic units in the area are not uniform and are highly

heterogeneous in lateral and vertical extent. There are interlayers of permeable and less permeable

geologic units (Limestone shale and dolerite). The geological heterogeneity mainly results from the

existence of dolerite dykes intruded into the sedimentary layers. The occurrence of the dolerite in the

area is mainly in the form of sill. In places, intrusion of dolerite dykes resulted in tilting and fracturing

of the sedimentary layers. The dipping layers in some parts of the Aynalem catchment led to an

increased complexity of the hydrogeological situation. The information from the lithologic log

indicates that the limestone and the fractured and weathered part of the dolerite are the main water

bearing geologic units. In almost all well log data, massive and less permeable dolerite is encountered

at depth which acts as a barrier to groundwater flow.


46

Figure 5.1. Dolerite dyke dissecting the sedimentary rock

Figure 5.2. Tilted sedimentary layers due to dolerite intrusion


47

Figure 5.3. Lithological log showing depth to dolerite (After Yehdego, 2003)

The cross-section constructed from the lithological log of selected boreholes indicates that the lateral

and vertical distribution of dolerite intrusion is highly variable. Considering the same measuring

reference elevation point, the dolerite at pumping well one (PW-1) is encountered at a depth of 71

meter below ground surface whereas at pumping well two (PW-2) which is 600 meters away, the

dolerite is exposed at the surface. Locations of wells applied for the lithological cross-section are

indicated in the well location map (Appendix 6). Apart from the lithological variation, the dolerite has

an effect on the degree of fracturing of the sedimentary layers. The distribution of the geological units

in the area is indicted in the geological map and cross-section constructed in the east-west direction

following the valley.


48

Figure 5.4. Geological map with east- west cross-section (WWDSE, 2006)

5.1.2. Geophysics

Geophysics refers to the study of the earth with special reference to its physical properties, structure

and composition. The geophysical prospecting techniques are based on various fundamental principles

of physics like the law of gravitational attraction, magnetism, optics, refraction and reflection, those

elements of electricity and theory of electromagnetism. The Resistivity method is the method

commonly applied in groundwater studies. Vertical electrical sounding is applied in the study of

horizontal and vertical discontinuity in the electrical properties of the ground, and also in the

detection of three dimensional bodies of anomalous electrical conductivity. The best interpretation

results are generally obtained from a combination of horizontal profiling and electrical sounding data.

Electrical sounding is the process by which depth investigations are made, and horizontal profiling is

the process by which lateral variation in resistivity are detected. However, the results of electrical


49

sounding and of horizontal profiling often are affected by both vertical and horizontal variations in the

electrical properties of the ground. For the present study, geophysical survey results (VES) and

Seismic refraction data were collected from previous studies mainly conducted by WWDSE (2006)

and Gebregziabher (2003). The resulting analysis of these surveys is used as supportive data for the

lithologic log and geological cross-section for identifying the hydrostratigraphic units. The

interpretation results of the vertical electrical soundings are calibrated by comparing with the

lithologic log and the resistivity value of the subsurface material at the same location in order to have

an idea of the resistivity values of geologic formation where there is no well log data. The geophysical

survey data is given in Appendix 4 and the interpretation results for the vertical electrical soundings

are summarised in table 5.1.

Table 5.1. Summery of vertical electrical sounding data

No. Estimated

resistivity range

(Ωm)

Main geological

formation

Description Remarks

1 10-60 Shale 10-25, wet and

25-60 dry

2 60-280 Limestone Weathered and

fractured

Water bearing

3 200-450 Limestone Hard, less

fractured.

4 100-300 Dolerite Fractured

6 300-600 Dolerite Slightly

fractured

7 >600 Dolerite Massive, hard dry

5.2. Hydrostratigraphy

Hydrostratigraphic units comprise geologic units of similar hydrogeological properties. In formulating

the hydrostratigraphy, several geologic formations may combine into a single hydrostratigraphic unit

or a geologic formation may be subdivided into aquifers and confining units (Anderson & Woessner,

1992). Understanding the lateral and vertical extent and relationship between the hydrogeological

units is crucial for constructing an accurate conceptual groundwater flow model. Based on available

resistivity and seismic data, drilling records and geologic knowledge gained from the previous studies

the following hydrostratigraphic units are identified.

• Limestone-shale–marl intercalation

• Limestone

• Dolerite

The main water bearing unit in the area is the limestone unit, but the weathered and fractured part of

the dolerite also acts as a conduit for occurrence and movement of groundwater. As discussed in a

recent study conducted by TAHAL (2007), the groundwater in the area exists under confined to semi-

confined conditions because of the interbedded fractured limestone and shale beds. Moreover, the


50

reported storage coefficients for some of the wells are in the order of 10- 4 which suggests a confined

aquifer system.

5.3. Hydraulic proporties of the stratigraphic units

Hydraulic properties including both horizontal and vertical hydraulic conductivities and

transmissivities are key components of the conceptual groundwater model. The hydraulic property

data for the Aynalem aquifer system is derived from aquifer pumping tests carried out in the previous

studies. The pumping test results have been documented on a number of published and unpublished

reports. Hydraulic properties of the different geologic units have been reported by Hussien (2000).

The hydraulic conductivity of limestone unit ranges from 29 to 74 m day-1. The limestone has a high

hydraulic conductivity at the contact between the limestone and dolerite due to the fracturing effect of

the dolerite intrusion. The Mekele dolerite has low primary porosity and permeability. Due to the

secondary porosity as a result of fracturing, the dolerite can also be considered as an aquifer where it

is fractured Vernier (1985). The hydraulic conductivity of the unfractured dolerite is reported to be

0.02-1 m day-1. As a result of geologic heterogeneity and the effect of fracturing, the Aynalem

catchment does not have uniform hydraulic properties. In the area there are zones with high

contrasting transmissivities and hydraulic conductivities attributed to the geologic heterogeneity and

degree of fracturing. As described in section 4.5, the pumping test analysis result shows wide ranges

of hydraulic properties with an average transmissivity of 540 m2 day-1.

5.4. Water budget

Identifying the source of water to the system as well as outflow from the system is part of the

development of conceptual model. In preparing water budget to the system, the expected outflow

directions and exit points are considered.

Recharge

Major factors contributing to groundwater recharge of an area include rainfall, evapotranspiration and

soil type. Groundwater recharge could take place by direct percolation through the vadose zone in

excess of moisture deficits and evapotranspiration or indirect recharge from streams and reservoirs.

The recharge in the area occurs from rainfall and seasonal floods generated from the topographic

elevated ridges surrounding the area. Isotope analysis of the groundwater in the Aynalem catchment

carried out by Yehdego (2003) indicates that the groundwater recharge mechanism is mainly direct

recharge from rainwater. The recharge process of Aynalem catchment is highly controlled by the

topography, geology and structure which direct the infiltrated water towards the discharge area. As

the mechanism of recharge process differs based on the climatic zones, selection of appropriate

recharge estimation technique requires conceptualisation of the recharge process. For this research,

recharge is estimated by applying the chloride mass balance method (section 4.3). The estimated

recharge by applying this method is 30 to 40 mm year-1 (4.5-6% of the average annual rainfall). As

none of the recharge estimation methods are absolutely accurate, the results of recharge estimated

from different methods will be compared with the recharge obtained from the model calibration to end

up in a reasonable estimate of the recharge.


51

Ephemeral channel transmission loss represents an important groundwater surface water exchange in

arid and semi-arid regions and is potentially a significant source of recharge at the basin scale.

Indirect recharge from ephemeral streams and seepage from reservoirs are other areas of groundwater

recharge mechanisms in the catchment. At the upstream parts of the sub-catchment, there are small

ephemeral streams that loose water downward from the stream bed to the water table as part of

groundwater recharge. Recently a number of small scale reservoirs were constructed in the Aynalem

catchment for the purpose of supplementary irrigation of the rain-fed agriculture. Seepage is a

common phenomenon of the area as the underlying geology is affected by fracturing and bedding

plane of the sedimentary beds. But for the purpose of the modelling it is assumed that the indirect

recharge from reservoirs and ephemeral streams is integrated in the areal recharge estimates, based on

the chloride mass balance method.

Groundwater discharge

The mechanism of groundwater discharge from the aquifer system is mainly discharge to streams,

well abstractions and groundwater outflow through the western outlet. The Aynalem river is well

connected to the aquifer system which feeds water to the river as base flow during the dry season. The

groundwater discharge from the aquifer is expressed as springs and seepages along the river banks

mainly in the lower reaches of the river and marshy areas in the flat laying parts of the valley. As part

of the natural groundwater discharge, a number of low yield springs are identified during the field

work. Most of the springs are gravity springs, discharged right at the contact of permeable and less

permeable units (limestone and shale).

Table 5.2. Spring inventory data

Spring UTM E UTM N Elevation (m) Date of measurement Discharge

(l s-1)

Discharge

(m3 day-1)

1 552160 1485603 2184 24/08/2007 0.3 25.92

2 555605 1491726 2247 24/08/2007 0.2 17.28

3 554457 1485491 2258 24/08/2007 0.2 17.28

4 551805 1487990 2176 24/08/2007 1.0 86.40

5 552077 1485518 2180 24/08/2007 0.2 17.28

6 557061 1486928 2202 25/08/2007 0.5 43.20

7 560160 1487371 2277 25/08/2007 0.2 17.28

8 560226 1487267 2274 25/08/2007 0.2 17.28

The main groundwater abstraction presently takes place through production wells for the water supply

of Mekele town. As described in section 4.4, about 7156 m3 of groundwater is abstracted daily from

the wellfield. Though there are no detailed studies of the unsaturated zone in Aynalem catchment, it is

assumed that the influence of evapotranspiration is limited to a depth of several meters above the

water table. Evapotranspiration from the saturated zone is therefore not considered in the conceptual

model as groundwater discharge.


52

Figure 5.5. Groundwater level profile

A profile showing surface topography, groundwater level elevation before pumping and during

pumping was constructed to have insight into the flow system and to see the effect of pumping in the

wellfield. As can be observed in the profile constructed based on the groundwater level

measurements, the general groundwater gradient follows the topography. Associated errors of

groundwater level measurements mainly, measuring device errors, operator errors and transient effects

during measuring resulted in noisy static water level elevations. However, the overall hydraulic

gradient is from east to west following the topography. Using this profile, the estimated natural

hydraulic gradient in the sub-basin is 0.012 or 1.2%. The groundwater level in the wellfield is

declining due to the ongoing abstractions resulting in a local cone of depression. The groundwater

level monitoring data indicate that there is a groundwater decline of up to 40 meters in response of the

well abstractions.

The major natural groundwater outflow from the sub-basin is through the western outlet under the

saturated aquifer depending on the hydraulic gradient in the wellfield. About 12960 m3 daily

groundwater outflow is estimated by applying Darcy law. Average transmissivity of 540 m2 day-1

(from reported pumping test results table 4.9) by assuming the well transmissivity to be valid also for

the aquifer transmissivity, hydraulic gradient 0.012 and two kilometres groundwater outflow zone are

considered in the calculation. The equation applied to estimate the groundwater outflow is given as:

Q =TW *dh/dx (5.1)

Where,

Q = groundwater outflow (m3 day-1)

T = Transmissivity (m2 day-1)

W = width of the zone of groundwater outflow (m)

dh/dx = hydraulic gradient


53

Similarly TAHAL (2007) applied Darcy’s law to calculate groundwater flux that feeds Aynalem

wellfield from the eastern side. Their basic assumptions were groundwater recharge occurs only from

the eastern part and no interaction between groundwater and surface water. TAHAL (2007) calculated

the groundwater flow that apparently feed Aynalem wellfield by considering 5km front width, average

gradient of 1.5% and transmissivity value of 180 m2 day-1 which they claim as the median value of the

well transmissivities reported by WWDSE (2006). Based on these considerations the calculated

groundwater flow to the wellfield was 13500 m3 day-1.

The present study considerations differ from that of TAHAL (2007) in that:

• Calculation is for the groundwater outflow that leaves the catchment with front width of 2km

groundwater outflow zone in the western end of the catchment.

• The recharge is not only occurring in the eastern part.

• There is groundwater-surface water interaction as the groundwater discharge is expressed by a

number of springs and measurements of stream flow even in dry seasons.

• Transmissivity value was considered as the average value of the reported transmissivities.

5.5. Groundwater flow system

A groundwater flow system is a set of flow paths with common recharge and discharge areas. The

conceptualization of how and where water originates in the groundwater flow system and how and

where it leaves the system is critical to the development of an accurate model. As it is pointed out by

different researchers (Gebregziabher 2003, Hussien, 2000), the groundwater flow system in the area is

mainly controlled by extensive faults and fractures. The local flow system is controlled by the

topography and geology and degree of fracturing. To construct groundwater flow direction, hydraulic

head information is required. Head measurements are also needed to establish initial conditions for

the numerical groundwater modeling and for model calibration. As it can be observed in the static

water level records from the wells (Appendix 3.4), under natural condition, the groundwater level in

the area ranges from 7 to 51 meters below ground surface.

5.6. Model boundaries

Model boundary is the interface between the model calculation domain and surrounding environment.

In modeling, we are interested in a specific part from the continuous real world system. Thus, the

effect of the real world in terms of hydrological influences at the model boundaries must be described.

Correct selection of boundary conditions is a critical step in model design because the boundaries

largely determine the flow pattern (Anderson & Woessner, 1992). In other words, mass exchanges

across the boundaries are simulated and hence wrong boundary conditions generate wrong water

balance of the system under study. In groundwater flow system, model boundaries could be physical

(impermeable geologic formations and surface water bodies) or hydraulic boundaries (groundwater

divides and flow lines). Establishment of the model boundaries is based on the site specific

knowledge acquired from the geology topography and flow system prevailing in the area. Physical

boundaries including impervious geologic formation, tight fault escarpments, topography and surface

water divides are used in defining the boundaries of the model domain. As it can be seen in the map

prepared from the ASTER DEM (Fig.5.6), Aynalem sub-basin is physically separated by the dolerite


54

ridge lines from the adjacent sub-basins of Ilala and Chelekot. Dolerite ridge lines form the Eastern,

Southern and Northern boundaries of the study domain. Those parts of the model boundaries are

defined as no flow boundary assuming that the groundwater fluxes across the water divide are

negligible. As described in section 5.4, the groundwater flow from the sub-basin is considered through

the western outlet depending on the hydraulic gradient in the wellfield. General head boundary is

selected as the appropriate boundary condition for the head dependent flow through the western

outlet.

Figure 5.6. ASTER DEM indicating the three basins in the Mekele area

5.7. Simplification of the real world

Simplification is necessary because complete reconstruction of the field system is not feasible. The

conceptual model should be simplified as much as possible while it is still remains complex enough to

represent the system behaviour (Anderson & Woessner, 1992). Despite heterogeneous hydrogeologic

condition in the sub-basin, a simplified conceptual hydrogeological model of the groundwater system

of Aynalem was developed on the basis of information about geology, hydrogeology and hydrology.

For the modelling purpose, the weathered and fractured dolerite and the limestone unit were combined

to form a single aquifer system. On the other hand, the massive dolerite sill and inter-bedded shale are

considered as barriers to the groundwater occurrence and movement in the wellfield.


55

In order to simplify the complexity of the real hydrogeological system, some basic assumptions about

the study area have to be made. The following plausible assumptions are made about the model area

of Aynalem:

• The system is considered in a steady-state throughout the year.

• The geological formations of concern are considered horizontal.

• Since there are flow measurements, during the dry season the Aynalem river is assumed to be

discharging groundwater (gaining stream) on average.

• There is no groundwater inflow from the adjacent sub-basins.

Groundwater occurrence, distribution and flow regime in the Aynalem sub-basin is highly governed

by dolerite sills which categorise the groundwater into shallow and deep aquifer systems. Due to

absence of data for the deep aquifer system, this study is concentrated on the shallow aquifer system.

Based on the lithological and geophysical logs 50 meters of average aquifer thickness is considered.

The conceptual system of the sub-basin is shown in Fig 5.7 below.

.

Figure 5.7. Pictorial representation of the hydrologic system of Aynalem sub-basin

Q out


56


57

6. Numerical model

6.1. Code selection

Primary governing factor for selection of code depends on the objective and reliability of the code.

Since the main focus of the study is to understand the groundwater flow system in response to

recharge and discharge, the selected model is MODFLOW. MODFLOW is a modular three

dimensional finite-difference groundwater flow model of the U.S. Geological Survey, to describe and

predict the behaviour of groundwater flow system. The code is based on the flow equation of Darcy

and mass continuity equation. The partial differential equation in which MODFLOW is based on is:

t

hSW

z

hk

zy

hk

yx

hk

x szzyyxx ∂∂=−

∂∂

∂∂+

∂∂

∂∂+

∂∂

∂∂

(6.1)

Where,

kxx, kyy, kzz represents hydraulic conductivity along x, y and z coordinate axes (LT-1), which are

assumed to be parallel to the major axes of hydraulic conductivity, h represents the potentiometric

head (L), W is flux per unit volume representing sources and/or sinks of water (T-1), Ss represents the

specific storage of the porous material (L-1) and t is time. For steady state condition there is no change

in storage with time and hence the right-hand side of the equation is set to zero. Furthermore the

consideration of the present study is two-dimensional model, hence the third component of the

equation goes to zero.

MODFLOW numerically uses block-centred finite differences to solve the flow equations in three

dimensions. It consists of a main program and independent subroutines called modules (McDonald &

Harbaugh, 1988). The modules are grouped into packages and each package deals with a specific

hydrogeologic feature to be simulated.

MODFLOW is selected for the following reasons:

• MODFLOW is one of the most widely used groundwater flow code in the field of

hydrogeology and is able to simulate steady state and transient flow conditions in one.

• The model can simulate recharge, flow to wells, flow to drains and flow through riverbeds.

• MODFLOW is extensively tested in various environments under different conditions.

• The theory behind the model is well documented and relatively easy to understand.

6.2. Model geometry

As discussed in the conceptual model development section, the model domain is delineated based on

the surface topography and local physical boundaries.


58

Horizontal extent The horizontal extent of the model domain is 8 by 20 km bounded by 548820 to 569251 m UTM East

1482054 to 1490288 m UTM north. The irregular shape of the study area reduces the model domain

to an area of about 104 square kilometres.

Vertical extent As described in the conceptual model, a simplified one layer model was used to represent the geologic

materials in the study area. In groundwater modeling, the number of model layers, which are

considered in the discretized domain, depends on the hydrogeological stratification of the system. In

many model approaches hydrogeological layers of a real world system can be simulated by a single

model layer. Based on the geologic and geophysical logs and well completion data a layer with a

constant average thickness of 50 meters is considered to model the Aynalem aquifer system. The top

and bottom elevations of the aquifer system are defined based on the lithologic logs and the DEM

extracted from the ASTER image.

6.3. Model design

At this stage of model development, the conceptual model approach is transformed to a form suitable

for mathematical modeling. The most important activities of model design include the design of

spatial domain, selection of initial conditions and setting the boundary conditions.

Discretization

In numerical models, the continuous natural phenomenon is replaced by a discretized domain, the so

called grid. Grid size depends on hydraulic gradient, degree of aquifer heterogeneity, size of the

model area, level of detail required and availability of data. Selecting the size of the nodal spacing is a

critical step in grid design (Anderson & Woessner, 1992). The size of the nodal spacing in horizontal

dimension is a function of the expected curvature in the water table or potentiometric surfaces. Finer

nodal spacing is required to define highly curved surfaces. A grid with a smaller number of nods is

preferred in order to minimise data handling, computer storage and computation time. Yet, it is

desirable to use a large number of nodes to represent the system accurately. By making simplifications

and assumptions of the actual field condition, the model area is discretized to one layer with a regular

grid of (250m by 250m, 32 rows by 82 columns) consisting of 1492 active cells.

Mathematical representation of boundaries

The mathematical representation of the boundaries in the model is important because many hydrologic

boundary conditions can be mathematically represented in more than one way. Boundary conditions

are mathematical statements specifying the dependent variable (head) or the derivative of the

dependent variable (flux) at the boundary of the problem domain. In the field of hydrogeology three

types of mathematical boundaries are applied.

• Specified head (Dirchlet condition) for which head is given.

• A specified flow (Neumann condition) for which the derivative of the flux across the

boundary is given.

• Head dependent flow boundary (Cauchy condition) for which flux across the boundary is

calculated given a boundary head and conductance values.


59

For the present study specified flow (no flow) boundaries and head dependent flow boundaries

(Cauchy boundary conditions) were applied. A general head boundary package was employed to

simulate groundwater outflow through the western outlet of the sub-basin. General head boundary is a

generic form of the head dependent boundary condition. General head boundaries are normally used

along the edge of the model to allow groundwater to flow into or out of the model under a regional

gradient. If the water elevation rises above the specified head, water flows out of the aquifer. The

expression applied for the head dependent flow in the general head flow boundary is:

Qb=Cb*(hb-h) (6.2)

Where

Cb = hydraulic conductance of the boundary (L2 T-1)

hb = hydraulic head at or beyond the boundary( L)

h = hydraulic head in the aquifer (L)

As discussed in the conceptual model formulation, Aynalem river is in hydraulic contact with the

groundwater. In the lower reaches of the river groundwater is discharged to the river as springs and

seepages along the river banks and river beds. The flow of water between an aquifer and overlying

river is commonly simulated using river package. The expression for which river package is based on

is:

QRIV = CRIV * (HRIV - h) h>RBOT (6.3)

QRIV = CRIV * (HRIV - RBOT) h<= RBOT (6.4)

CRIV = (K * L * W) / M (6.5)

Where,

QRIV is the rate of leakage between the river and the aquifer (L3 T-1), CRIV is hydraulic conductance

of the river bed, (L2T-1), HRIV is head in the river (L), h is hydraulic head in cell (L), RBOT is

elevation of the bottom of the riverbed (L), K is the hydraulic conductivity of the riverbed material

(LT -1), L is the length of the river within a cell (L), W is the width of the river (L) and M is the

thickness of the riverbed (L).

The groundwater drained to the river can be simulated by setting RBOT equal to HRV in the river

package, for this case the river package acts the same as the drain package. The drain package works

in a much the same way as the river package, except that leakage from the drain to the aquifer is not

allowed (Anderson & Woessner, 1992). Due to the absence of river level measurement data, drain

package is applied to simulate the groundwater discharge to the gaining reaches of the river. The

recharge to the aquifer from the loosing reaches of the river is assumed to be take place by the vertical

areal recharge which is integrated in the chloride mass balance method.


60

Figure 6.1. Model boundary conditions

Initial conditions

Initial conditions refer to the hydraulic head distribution in the system at the beginning of the

simulation and thus are boundary condition in time (Anderson & Woessner, 1992). The initial

conditions in numerical groundwater models are initial head distributions and have only to be entered

to fulfil the convergence criteria of the numerical scheme. If the hydraulic gradient between heads of

boundary elements and non boundary elements become too large, many computer codes will fail in

their calculations by numeric instabilities. So for steady state models initial heads should be only in

the range with the values of the hydraulic head conditions at the boundary element of the model. For

the present case, the static water level records of the wells are interpolated within the model to obtain

the initial hydraulic heads for the entire model.

Representation of aquifer parameters

The primary hydraulic parameters required by a steady-state groundwater flow model are either

transmissivity or hydraulic conductivity in a distributed fashion across the model grid. Zonation

schemes of the aquifer parameters were applied to better approximate spatial distribution of the

hydraulic parameters and for accommodation of the heterogeneity. Zonation for the input parameters

was carried out based on geological information, point hydraulic conductivity and transmissivity data

of the pumping tests. Initially, the hydraulic parameters estimated from pumping test results of

previous studies were applied, later the parameters were adjusted during the calibration process.

Boreholes and observation wells

The majority of the boreholes are concentrated on the lower right bank of the catchment. The data

base of the boreholes was collected from the Water Supply Office of Mekele town and from previous

reports on the area. There are no proper data records of all the boreholes and it is hardly possible to

obtain a continuous time series monitoring groundwater level data for all of the boreholes. In almost


61

all wells, the pumping well serves as an observation well. That means that the monitoring

groundwater level data is highly affected by the transient effect of the pumping at the time of water

level measurements. In the modelling area the well abstractions indicated in section 4.4, were

implemented using the well package of MODFLOW.

Recharge

Considering direct recharge from rainfall, the recharge estimated by the chloride mass balance

method which is 4.5-6% from the average rainfall of 670mm, was utilised as a model input. This

recharge is uniformly applied to the top most active cell by using the recharge package of

MODFLOW. The recharge value is adjusted to 6% of the average annual rainfall of the area during

the model calibration.

6.4. Model calibration

In most cases, the model will not give satisfactory results because the input data to the model do not

reflect the real world with enough accuracy mainly due to ambiguity of input data. Thus in order to

improve the reliability of the model, adjustments in the model input data are required. In the

procedure of parameter value adjustment, the values are adjusted within a pre-determined range of

error criterion until the model produces results that approximate the set of field measurements

selected as the calibration target.

Calibration target and uncertainty

Prior to the calibration process, setting of calibration target is required. Calibration target is a

calibration value and its associated errors. In the Aynalem case, hydraulic heads obtained from

groundwater level measurement data were used as calibration values. And the calibration target was to

match hydraulic heads calculated by the model with measured head points. Hydraulic heads were

obtained from groundwater level monitoring of three years data and from static water level records

measured during drilling time of 1998/99. Water level measurements from some of the wells (PW2

and PW6) are erratic and are believed as outliers due to large measurement errors and are left out

during calibration. It should be noted that most of the field measured head data are associated with

errors due to the following reasons.

• The static water level measurements were taken just after well completion with out

stabilization.

• There are no independent monitoring wells, the water level observations are carried out on the

pumping wells.

• Measurement errors related to measuring instrument accuracy and operator errors.

Thus the field measurement of head values may not represent the actual water level of the field

condition. The combined effect of these errors is assumed to result in a maximum of 10 meters

difference between the observed and calculated heads.

The model calibration was carried out for the scenarios with and without abstraction. The first step

towards model calibration is to check the model reliability in generating field condition, when it is


62

subjected only to the natural regime. The static water levels records before time of pumping were used

in calibrating the natural regime of the groundwater flow system. Average groundwater levels of three

years monitoring data were used to calibrate the steady-state model with pumping scenario.

Trial and error calibration

Trial and error calibration is the process of manual adjustment of input parameters until the model

produce field measured heads within the range of the error criteria. The model was calibrated for

natural steady-state conditions, assuming constant recharge and steady discharge neglecting seasonal

fluctuations. Calibration was conducted through trial and error by varying aquifer hydraulic

parameters and comparing calculated heads to those measured in wells. During the calibration

transmissivity, recharge and boundary conditions were modified manually and trial runs were carried

out until the model output was within the range of the pre-defined error criterion.

The best fit results were achieved when the study area was divided into regions with different

transmissivity zones. The use of zones of uniform property value as a basis of spatial parameter

definition can be quite unsatisfying for a number of reasons, including the fact that the design of a

tentative zonation scheme for the model must be undertaken ahead of the actual parameter estimation

process. Although geological mapping can help in this regard, there are many instances where

geological boundaries are only approximately known. The transmissivity values applied during the

calibration ranges from 30 to 135 m2 day-1. The transmissivity values attained during the calibration

process are by far smaller than the reported transmissivities from previous studies. The procedure

followed during the trial and error method of calibration is indicated in Figure 6.2.

Figure 6.2. Trial and error calibration procedures (Adapted from Anderson and Woessner, 1992)


63

Calibration results

The calibration results of the model are presented graphically and in a table form showing the

distribution and comparisons of the observed and model calculated hydraulic heads. A listing of

measured and simulated heads together with their differences and average of the differences is a

common way of reporting calibration results.

Figure 6.3. Contour map of simulated heads (non-pumping scenario)

Figure 6.4. Scatter plot of observed and simulated hydraulic heads (m)


64

Table 6.1. Observed and calculated heads for non-pumping scenario

Borehole

name

UTM east UTM north Observed

head (m)

Computed

head (m)

Difference (m)

PW-1 556050 1487809 2209.48 2210.26 -0.78

PW3 553941 1488821 2185.15 2195.59 -10.44

PW4 553706 1488251 2184.95 2192.92 -7.97

PW5 554336 1487216 2186.67 2196.56 -9.89

PW7 557115 1487967 2215.80 2215.39 0.41

PW8 557809 1488359 2222.99 2227.12 -4.13

PW9 558268 1488286 2222.14 2231.66 -9.52

PW11 552490 1489376 2187.60 2184.11 3.49

PW12 552945 1488948 2188.83 2187.14 1.69

TW3 552945 1488955 2177.62 2186.18 -8.56

TW5 553207 1488955 2188.47 2189.43 -0.96

TW1(2005) 561057 1487352 2258.35 2263.56 -5.21

TW2(2005) 564439 1485877 2311.30 2317.73 -6.43

Table 6.2. Observed and calculated heads for pumping scenario

Borehole

name

UTM East UTM North Observed

had (m)

Computed

head (m)

Difference (m)

TW4(2005) 557234 1488028 2182.40 2178.48 3.92

PW3 553941 1488821 2162.70 2169.51 -6.81

PW7b 557188 1488028 2182.90 2176.74 6.16

PW8 557809 1488359 2202.30 2193.81 8.49

PW11 552490 1489376 2162.40 2159.74 2.66

Mu 552511 1489648 2159.60 2160.23 -0.63

TW3 552945 1488955 2160.20 2160.39 -0.19

PW6 555529 1487648 2172.00 2180.17 -8.17

PW12 552945 1488948 2165.80 2161.64 4.16

TW1(2005) 561057 1487352 2258.35 2255.91 2.44

TW2(2005) 564439 1485877 2311.30 2309.12 2.18

Evaluation of calibration

The results of the calibration should be evaluated both qualitatively and quantitatively (Anderson &

Woessner, 1992). The calibrated results were evaluated based on the calibration target and assessment

of the mass balance of the system.

The calibrated model was evaluated qualitatively and quantitatively, in which hydraulic head

distributions have been used as quantitative calibration targets whereas flow directions have been

used as qualitative calibration targets. Qualitatively, the contour maps of the simulated heads were

analyzed to see whether the results are comparable with the actual field condition. Flow direction was

determined based on the simulated head distribution and comparison is made with the flow direction

determined in the conceptual model. A scatter plot of measured against simulated heads is another


65

way of showing the calibrated fit (Fig.6.4). The scatter plots are visually examined whether points in a

plot deviated from the straight line in a random distribution. Deviation of points from the straight line

should be randomly distributed. Furthermore the calibrated model outputs were evaluated by applying

the three common ways of error quantifying methods (Mean error, Mean absolute error and Root

Mean Squared error).

Mean error is the mean of the difference between the measured head (hm) and simulated head (hs)

ME = is

n

im hh

n)(

1

1

−∑=

(6.6)

Mean absolute error is the mean of the absolute value of the difference between the measured head

(hm) and simulated head (hs)

MAE =∑=

−n

iism hh

n 1

)(1

(6.7)

Root Mean Squared error is the square root of the average of the squared difference between the

measured head (hm) and simulated head (hs)

RMSE =

−∑ =ni ism hh

n 12)(

15.0

(6.8)

Where

n = number of calibration values

Table 6.3. Errors of the calibrated model

Error (m)

Model scenario ME MAE RMSE

Non-pumping 4.48 5.35 6.42

Pumping 1.34 4.12 4.92

The evaluation of the calibrated model result shows that:

• Most of the simulated heads were within the pre-established calibration target.

• Water balance discrepancy was zero (Appendix 7).

• The overall results of the groundwater model are comparable with the measured well data and

in agreement with conceptual model.

• The measure of errors evaluated by ME, MAE and RMSE are in the acceptable range

according to the pre-determined error criteria.

Though the overall result of the model was comparable with the measured well data, few observations

which are not uniformly distributed over the model domain are utilised in the calibration process.

Ideally calibration values should be measured at a large number of points uniformly distributed over


66

the model domains. Thus it is not possible to conclude that the calibration is accurate by only

quantifying the errors using ME, MAE and RMSE with out considering the distribution of the

residuals. Comparison between contour maps of measured and simulated heads provides a visual

qualitative measure of the similarity between patterns, thereby giving some idea of the spatial

distribution of errors in the calibration.

6.5. Sensitivity analysis

Sensitivity analysis is an essential step in all modeling applications. As discussed by Anderson &

Woessner (1992), the purpose of a sensitivity analysis is to quantify the sensitivity of the model

simulations in the calibrated model caused by uncertainty in the estimates of aquifer parameters, stress

and boundary conditions. Sensitivity analysis provides information on which model parameters are

most important to the simulated system. Sensitivity analysis is also inherently part of model

calibration. The most sensitive parameters will be the most important parameters for matching the

model result with the observed values.

To assess the sensitivity of the simulations in the calibrated model, a sensitivity analysis was

performed with respect to the pumping scenario. The sensitivity analysis was performed by

systematically changing one calibrated parameter at a time while noting the observed changes in

hydraulic heads. A number of sensitivity analyses were conducted to test the effects on model results

due to changes in input parameters or boundary conditions.

It was identified that the most sensitive factors are recharge depth and transmissivity of the aquifer.

The calibrated values of these input variables were multiplied by factors of 0.5.0.8, 0.9, 1.1, 1.2, 1.3

and 1.5. The resulting hydraulic heads were then compared with the observed hydraulic heads and

mean average error, absolute average error and root mean squared error were calculated for each

parameter. Then the calculated average errors in the hydraulic heads were plotted against the

multiplying factors as shown in figures 6.5 and 6.6. The magnitude of change in heads from the

calibrated solution is the measure of the sensitivity of the solution to that particular parameter. And

the results of the sensitivity analysis are reported as the effects of the parameter change on the average

measure of error selected as a calibration criterion (in this case mean average error, absolute average

error and root mean squared error).

It was found that slight changes in either the aquifer transmissivity or slight changes in recharge rate

affect dramatically the distribution of hydraulic head throughout the area. The sensitivity plots show

that the recharge generates non-linear sensitive response while sensitivity towards transmissivity

generates linear response. The model is equally sensitive to both increase and decrease of

transmissivities on the other hand, the calibrated model is more sensitive to recharge fluxes reduction

than to recharge rate increment.


67

6.6. Model validation

In model validation, the normal procedure is to define a set of measurements or observations of

system variables, where part is used for model calibration and the remaining part is used for model

validation. Unfortunately it is often impossible to validate a model because usually too short set of

observed state data is available, which is already required for calibration. For the same reason model

validation was not accomplished here.

0.00

5.00

10.00

15.00

20.00

25.00

30.00

35.00

0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7

Transmissivity change factor from calibrated value

Ero

rr o

f hy

daru

lic h

ead

(m)

ME MAE RMSE

Figure 6.5. Sensitivity plot of the calibrated model with respect to transmissivity

-40

-20

0

20

40

60

80

100

120

140

0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7

Recharge change factor from calibrated value

Err

or o

f hy

drau

lic h

ead

(m)

ME MAE RMSE

Figure 6.6. Sensitivity plot of the calibrated model with respect to recharge


68


69

7. Discussion and results

This chapter is designed to analyse the findings and discuss the overall results of the study with more

attention given to modelling results.

7.1. Hydrochemistry

Water chemistry differs depending on the source of water, the degree to which it has been evaporated,

the types of rock and mineral it has encountered, and the time it has been in contact with reactive

minerals. The chemical constituents of groundwater give important clues with regard to the geological

history of the enclosing rocks, the velocity and direction of water movement (Freeze & Cherry, 1979).

As discussed in section 4.2, water samples from the existing wells and springs were collected and

their physical and chemical characteristics were analysed. Chemical analysis results of water samples

collected from wells drilled in Agula shale show high concentrations of Ca2+, SO42- and Na+. This

high concentration is caused by dissolved gypsum and limestone minerals which are found

interbedded in the Agula shale. On the other hand, chemical analysis of water samples collected from

the wells drilled in Mekele dolerite generally show low concentrations of Ca2+, SO42- and Na+. The

plots of chemical analysis results on Piper diagram, (Fig 4.4) and Stiff diagrams (Fig 4.5 to Fig 4.7)

show that the groundwater samples are calcium rich. Among the anion facies a majority of the water

samples does not fall in any dominant class and varies from sulphate to bicarbonate. The analysis

shows that groundwater in the sub-basin is Ca-HCO3 type in the upper catchment, changing to Ca-

HCO3-SO4 type along the groundwater flow direction and finally becoming Ca-SO4 type near to the

outlet of the catchment. Groundwater in the upper zone of the catchment has a low concentration of

total dissolved solids (TDS) that ranges from 400 to 700 mg l-1, whereas, the groundwater in the lower

part of the catchment has high total dissolved solids (TDS) that ranges from 800 to 1500 mg l-1. The

groundwater in the lower part or outlet of the catchment is a combination of water that infiltrates

every where in the catchment and has a more chance to interact with the rock materials along its flow

path which contributes for the high concentration of total dissolved solids toward the outlet of the

catchment. In other words, the increase in TDS to ward the western outlet is resulted from water with

longer residence time.

Water type and source rock deduction analysis of groundwater samples from Aynalem and adjacent

catchments was attempted to see whether there is groundwater connectivity between the catchments.

As there is no silica analysis result in any of the samples, simple ionic comparisons were used for the

analysis of source rock deduction. In the water type identification the bicarbonate to silica ratio is not

considered. Thus, with this simple analysis and with the data at hand, it is not possible to give a clear

conclusion whether the similarity in water type is attributed to lithology or groundwater connectivity.

7.2. Modeling results

The maps of spatial distribution of the simulated hydraulic heads for the non-pumping and pumping

scenarios are shown in Figures 7.1 and 7.2 respectively.


70

A

B

Figure 7.1. Distribution of hydraulic heads with non-pumping scenario

A

B

Figure 7.2. Distribution of hydraulic heads with pumping scenario

Hydraulic head for pumping and non-pumping scenarios

2100

2150

2200

2250

2300

2350

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000

distance (m)

hydr

aulic

hea

d (m

)

head with non-pumping head with pumping

A

B

Figure 7.3. Comparisons of simulated hydraulic heads for both scenarios


71

A cross-section is constructed from east to west direction following the valley to see the hydraulic

head distribution in the system with both scenarios. In both cases the general hydraulic gradient in the

sub-basin follows the surface topography and the gradient is east-west which is in agreement with the

flow system defend in the section on the conceptual model of the area. As it is clearly shown in the

hydraulic head cross-section with pumping scenario, there is a groundwater level decline which

results in a cone of depression in the wellfield area. The effect of pumping is largely on the wellfield

area where there is extensive abstraction and with less effect towards the upper and lower end of the

catchment. As is indicated in the cross-section comparing the simulated hydraulic heads for the

pumping and non-pumping scenarios, the groundwater abstraction in the wellfield area results in a

groundwater level decline up to 37 meters. The groundwater level decline as a result of pumping

estimated by the model is in agreement with the observed decline (up to 40 m).

7.2.1. Hydraulic properties

The aquifer parameters obtained from the pumping test result show very high contrasting hydraulic

properties spatially. There are extremely high and low values of transmissivity and hydraulic

conductivity values obtained from nearby wells. The assumption of homogeneity and infinite

horizontal extent of aquifer usually considered in pumping test data analysis are highly violated due

the geological heterogeneity of the area. Thus it is tried to optimize the hydraulic properties during the

calibration process using the calibration values (hydraulic heads). The average transmissivity value

reported from the well pumping test results was 540 m2 day-1. The model calculated transmissivities

were much lower than those reported from pumping test result data of the previous studies. To better

address the heterogeneity, zones of transmissivities were applied (Fig 7.4). The transmissivity values

adjusted in the calibrated model are ranging 30 to 135 m2 day-1.

Figure 7.4. Transmissivity zones applied to the calibrated model


72

7.3. Groundwater budget

The input term to the groundwater considered for the present study is direct recharge from rainfall.

Whereas the output terms considered are well withdrawals, groundwater drains to the river and head

dependent groundwater flow out of the aquifer system through the western boundary.

Recharge

Recharge is very difficult to estimate reliably and in many cases more than one recharge estimation

method is required. There are as many methods available for quantifying groundwater recharge as

there are different sources and processes of recharge. Each of the methods has its own limitations in

terms of applicability and reliability (Yongxin & Beekman, 2003). As described in section 4.3, the

recharge is estimated using chloride mass balance method as part of the present study. The estimated

recharge value may vary depending on errors associated to the method and the standard deviation of

the chloride content measurements both in the groundwater and the rain water. Yongxin & Beekman

(2003) discuss the uncertainties associated with the chloride mass balance method as:

• Uncertainties in the measured chloride content, both in rainfall and groundwater

• Uncertainty in the measured rainfall amount, depending on the type of rain gauge used and

analytical errors introduced.

The optimised recharge rate by model calibration is 42 mm (6% of the mean annual rainfall), while

the chloride mass balance shows that the recharge is in the order of 30-40 mm year-1 (4.5-6% of the

mean annual rainfall in the area). Considering the standard deviation in the chloride concentration of

the groundwater samples, the recharge estimated by chloride mass balance method is in the same

range as the recharge determined by the model calibration.

Previously, the groundwater recharge in the area has been estimated by different studies (Hussien

2000, Yehdego, 2003 and Teklay, 2006). The studies of Hussien (2000) and Yehdego (2003)

estimated the recharge by applying a water balance method as 9% of the average annual rainfall. By

applying the same method, Teklay (2006) estimates the recharge as 5.3% of the annual rainfall. The

reported recharge values vary widely. This report estimates the groundwater recharge in the area in

the range of 4.5 to 6% of the annual rainfall.

Groundwater abstraction

Groundwater pumping is frequently the least measured water balance component (Ruud et al., 2004).

The groundwater abstraction in Aynalem wellfield is poorly documented and the recorded data shows

many gaps. The operating periods of the boreholes are also not well known, thus it is difficult to get

the accurate abstraction rate from the wellfield. By examining and analysing the available abstraction

records, 7156 m3 of average daily groundwater abstraction from the wellfield is estimated and this

value is implemented in the steady-state model through the well package.


73

Model simulated groundwater budget

The groundwater budget can be quantified on the basis of the calibrated model output. The

groundwater flow budgets calculated by the model for the non-pumping and pumping scenarios are

indicated in table 7.1 and table 7.2 respectively.

Table 7.1. Model simulated groundwater budget of the area for the non- pumping scenario

Flow term IN (m3 day-1) OUT (m3 day-1)

Recharge 11982

River 7236

Head dependent flow through the western boundary 4746

Well withdrawals

Total 11982 11982

Table 7.2. Model simulated groundwater budget of the area for pumping scenario

Flow term IN (m3 day-1) OUT (m3 day-1)

Recharge 11982

River 810

Head dependent flow through the western boundary 4015

Well withdrawals 7156

Total 11982 11982

There are several indicators that the overall result of the calibrated steady-state groundwater flow

model developed for the area is realistic.

• The deviation of the simulated heads from the observed heads is within the pre-established

calibration target.

• The simulated groundwater level drawdown due to pumping in the wellfield area is in

agreement with the observed groundwater levels.

• The model calculated inflow and outflow terms are balancing.

• Groundwater flow direction simulated by the model is reasonable and in agreement with the

flow direction defined in the conceptual model.

Nevertheless, there are a number of limitations and uncertainties in the steady-state groundwater flow

model developed here.

7.4. Model limitations

Numerical models of groundwater flow are limited in their representation of the physical system

because they contain simplifications and assumptions that may or may not be valid. Results from

groundwater flow models have a degree of uncertainty primarily because of uncertainties in many

model input parameters (most importantly hydraulic conductivity and transmissivity) and boundary

conditions applied. The various steps in the modelling process may each introduce errors, converting

the real world into conceptual model and converting conceptual model into mathematical model.

The built model of the present research is associated with a number of uncertainties. First of all the

hydro geological heterogeneity caused difficulties in the conceptual simplification of the field


74

condition due to lack of detailed description of the heterogeneity. Despite the complex and

heterogeneous nature of the aquifer system, assumptions and simplifications were made during the

conceptualization of the system. Definitely uncertainties will be introduced as the result of these

assumptions and simplifications of the field conditions. Various forms of heterogeneity in the porous

media properties can be very different from the fluid flow behaviour in the individual zones (Das &

Lewis, 2007). There are generally few locations where observations are available, and the geological

structure of the aquifer is only partially known. Lack of proper site characterization may result in a

model that is calibrated to a set of conditions which are not representative of actual field conditions.

The main constraints in the modeling process were data gaps and poor quality of the available data.

The data which have a key role in defining the model geometry (screen length and aquifer thickness)

are not well documented. The available records of the water level measurements are not continuous

and mostly are only single measurements. Furthermore calibration values (heads) are highly

associated with measurement errors. The available hydraulic head data is applied for model

calibration, thus there was insufficient independent data for model validation.

Another area of uncertainty is resulting from defining the boundary conditions of the model domain.

The boundary conditions were defined based on the surface physical features such as impervious

geology and surface water divide. The locations of the groundwater catchment boundaries are

uncertain since they might not coincide with their surface expressions. This uncertainty is highest in

sedimentary terrain where there is preferential occurrence of secondary porosity along bedding planes

as in the case of the study area. The area is conceptualized as a single layer assuming impermeable

dolerite sill separating the upper aquifer from the deep aquifer system. In reality the separating layer

may be partially impervious layer due to localized fractures. Hence additional uncertainty may be

introduced as the result of this assumption.


75

8. Conclusion and recommendations

8.1. Conclusions

Primary and secondary porosity and permeability of the main water bearing geologic unit (limestone)

plays a most important role in controlling the natural groundwater occurrence and movement in the

Aynalem catchment. Due to the inter-layering of the limestone unit with less permeable shale and

dolerite rocks, the groundwater in the area occurs under confined to semi-confined conditions. In

response to the geological heterogeneity, broad ranges of aquifer parameters (transmissivity and

hydraulic conductivity) are reported from the results of pumping test analysis. As major part of the

study, a steady-state groundwater flow model was developed with and with non-pumping scenarios to

assess the groundwater resource of Aynalem sub-basin.

The principal mechanism of groundwater recharge in the area is direct recharge from rainfall. The use

of chloride mass balance method to estimate recharge shows that the annual recharge in the catchment

is in the order of 30-40 mm year-1 from mean annual rainfall of 670 mm. The model simulates a mean

annual recharge of 42 mm year-1. Although the accuracy of the chloride mass balance method is

dependent on the measurement accuracy of the chloride content both in groundwater and rainfall, it

can be applied to estimate groundwater recharge in the region.

From the hydrochemical data analysis, a conclusion can be made that there exist at least two classes

of water types in Aynalem catchment. Ca-CHO3 dominated water type at the upper catchment and Ca-

SO4 dominated water type at the lower western extreme of the catchment are present with a clear

evolutionary trend between the two types.

The steady-state flow modeling has demonstrated that an average recharge of 42 mm year-1 maintains

the natural equilibrium. On the other hand, the model results with pumping scenario show that a

groundwater abstraction of 7156 m3 day-1 resulted in a groundwater level decline up to 37 meters in

the wellfield area. The high rate of groundwater withdrawals in the lower right bank of the catchment

has created a local cone of depression (Fig 7.3). The cone of depression due to pumping is limited to

the current wellfield area and has less effect on the upstream and downstream areas of the catchment.

According to the overall groundwater resource assessment carried out by integrating geological,

hydrological data and steady-state flow modelling it is concluded that water balance components that

play important role in the groundwater table fluctuation of the wellfield are recharge and groundwater

abstraction from wells. The steady-state numerical model is suitable as a tool to improve our

understanding of the groundwater flow system in response to recharge and abstractions and to

estimate aquifer properties in the area without data. But the limitations of the model discussed in

section 7.4 should be taken into account prior to applying the model for groundwater resource

management.


76

8.2. Recommendations

This study is not the end of groundwater flow modeling in the area rather it is a good starting work for

detailed modeling in the future. With additional data, further refinement of the model is possible,

which is expected to improve the accuracy of the model. Further extensive field-based observations

combined with down hole geophysical well logging and hydraulic testing techniques, detailed

delineation of fractures and other secondary porosity is required to compile the hydrogeologic

framework for each geological sequence. Prior to detail groundwater modeling, detail structural

mapping is required which will have a great importance in aquifer characterization and definition of

boundary conditions.

The transmissivity to which the model is highly sensitive requires better future characterization. The

cell size applied in the discretization of the problem domain (250 by 250m) is not small enough to

adequately represent the drawdown in the wells. Thus the model can be further developed by

integrating additional data and by applying finer grids in the pumping areas. To improve the

uncertainties on the model boundaries, regional scale steady–state groundwater flow modeling is

recommended so that the steady-state solution of the regional model is applied to set the boundary

condition for the model at the local scale.

In the present study, only the upper shallow aquifer system is modeled as there is no data for the

deeper aquifer system, thus future studies should consider the deeper aquifer system and study the

relation to the regional groundwater flow system provided that additional data is obtained. The

fracturing in the limestone resulting from faulting and dolerite intrusion as identified by Hussien

(2000) and Yehdego (2003) shows fracture flow. It was tried to simulate this in the model by

assigning high transmissivity values that may not fully represent the fractured flow system. Thus a

groundwater flow model which accounts for both diffuse and fracture flow should be used to improve

modeling of groundwater flow in the fractured aquifer system.

Once an improved steady-state model is obtained, transient simulation should be carried out for better

predictions of pumping effect and for better recharge modeling (response of groundwater levels to

good/bad raining seasons). To improve calibration values, groundwater level should be monitored at a

set of monitoring wells that are evenly distributed in the sub-catchment. The time series

documentation of the abstraction rates should be improved.

Finally it is advised to consider the following points in relation to groundwater resources development

in the wellfield.

• Future groundwater resource development plans in the wellfield should take into account the

balance between the groundwater recharge and the intended abstraction rates to ensure the

sustainability of the resource in the catchment.

• It is advisable to redistribute the pumping wells from the narrow right lower bank of the

catchment to the upper catchment to reduce the groundwater stress resulting from localised

pumping.

• The drawdown in the wellfield should not become so high that the groundwater levels in the

wellfield is becoming lower than the level at the outlet.


77

9. References

Anderson, M. P., & Woessner, W. W. (1992). Applied groundwater modeling : simulation of flow and advective transport. San Diego Academic Press.

Appelo, C. A. J., & Postma, D. (1992). Geochemistry, Groundwater and Pollution. Lyngby, Amsterdam

Bear, J., & Verruijt, A. (1987). Modelling groundwater flow and pollution. Dordrecht etc.: D.Reidel. Beyth, M. (1970). Hydrogeology of Mekele area, Geological Survey of Ethiopia. Beyth, M. (1972). Paleozoic-Mesozoic Sedimentary basin of Mekele outlier. Memorie dicienze

geolgche,v.49. Beyth, M. (1972). Paleozoic - Mesozoic Sedimentary Basin of Mekele Outlier, Northern Ethiopia,

Amer. Ass. Petrol. Bossellini, A., Russo, A., Tadesse, S., & Assefa, G. ( 1997). The Mesozoic Succession of the Mekele

outlier (Tigre province, Ethiopia). Memo. Sci. Geol., V. 49, pp 95-116. Brassington, R. (1998). Field hydrogeology (Second edition ed.). Chichester etc.: Wiley & Sons. Carrera-Hernandez, J. J., & Gaskin, S. J. (2006). The groundwater modeling tool for GRASS

(GMTG): Open source groundwater flow modeling. Computers & Geosciences, 32(3), 339-351.

Chernet, T. ( 1993). Hydrogeology of Ethiopia and water resource development. EIGS, Addis Ababa. Das, D. B., & Lewis, M. (2007). Dynamics of fluid circulation in coupled free and heterogeneous

porous domains. Chemical Engineering Science, 62(13), 3549-3573. Davis, N. S., & DeWiest, J. M. R. (1966). Hydrogeology. John Wiley Sons, New York. DEVECON. (1992). Five Towns Water Supply and Sanitation Study. Ministry of Water Resources,

Addis Ababa. Dingman, S. L. (1994). Physical hydrology + disk for Lotus version 2 or 3. Upper Saddle River:

Prentice Hall. Eriksson, E. (1985). Principles and applications of hydrochemistry. Chapman & Hall, London-New

York. Fetter, C. W. (2001). Applied hydrogeology + Visual Modflow, Flownet and Aqtesolv student version

software on CD - ROM (Fourth edition ed.). Upper Saddle River: Prentice Hall. Freeze, R. A., & Cherry, J. A. (1979). Groundwater. Englewood Cliffs: Prentice-Hall. Gamachu, D. (1977). Aspects of climate and water budget in Ethiopia. Addis Ababa University press,

Addis Ababa. Gebregziabher, B. (2003). Integrated geophysical methods to investigate the geological structures

and hydrostratigraphic unit of the Aynalem area ,South East Mekele. Unpublished Msc thesis AAU.

Hounslow, A. W. (1995). Water quality data : analysis and interpretation. Boca Raton etc.: CRC Lewis.

Hussien. (2000). Hydrogeology of The Aynalem Wellfield, Tigray, Northern Ethiopia. Unpublished Msc thesis, Addis Ababa University.

Jyrkama, M. I., & Sykes, J. F. (2007). The impact of climate change on spatially varying groundwater recharge in the grand river watershed (Ontario). Journal of Hydrology, 338(3-4), 237-250.

Kazmin, V. (1975). Explanation of geological map of Ethiopia.Geological Survey of Ethiopia. Kruseman, G. P., & de Ridder, N. A. (1991). Analysis and evaluation of pumping test data (Second

edition, completely revised ed.). Wageningen: International Institute for Land Reclamation and Improvement (ILRI).

Leven, C., & Dietrich, P. (2006). What information can we get from pumping tests?-comparing pumping test configurations using sensitivity coefficients. Journal of Hydrology, 319(1-4), 199-215.

Levitte, D. (1970). The Geology of Mekele (Report on the Geology of the central part of sheet ND 37-11). Geological Survey of Ethiopia, Addis Ababa.

MacDonald, A. M. (2001). Ethiopia: water security and drought. BGS Technical Report WC/01/02.


78

McDonald, M. G., & Harbaugh, A. W. (1988). A modular three-dimentional finite difference ground-water flow model,U.S.Geological.Survey.tech.Water resources.Invest.,.

Mehl, S., & Hill, M. C. (2002). Development and evaluation of a local grid refinement method for block-centered finite-difference groundwater models using shared nodes. Advances in Water Resources, 25(5), 497-511.

Mengesha, T., Tadios, C., & Workneh, H. (1996). Explanation of the geological map of Ethiopia,Ethiopian Geological Survey.

Morris, B., Lawrence ARL, Chilton PJC, Adams B, Calow RC, & BA., K. (2003). Groundwater and its susceptibility to degradation: a global assessment of the problem and options for management. Early

Warning and Assessment Report Series, Nairobi, Kenya:United Nations Environment Programme. . RS.03–3.

Pappas, E. A., Smith, D. R., Huang, C., Shuster, W. D., & Bonta, J. V. (2008). Impervious surface impacts to runoff and sediment discharge under laboratory rainfall simulation. CATENA, In Press, Corrected Proof.

Pulido-Velazquez, D., Sahuquillo, A., Andreu, J., & Pulido-Velazquez, M. (2007). A general methodology to simulate groundwater flow of unconfined aquifers with a reduced computational cost. Journal of Hydrology, 338(1-2), 42-56.

Rushton, K. R., Eilers, V. H. M., & Carter, R. C. (2006). Improved soil moisture balance methodology for recharge estimation. Journal of Hydrology, 318(1-4), 379-399.

Ruud, N., Harter, T., & Naugle, A. (2004). Estimation of groundwater pumping as closure to the water balance of a semi-arid, irrigated agricultural basin. Journal of Hydrology, 297(1-4), 51-73.

Scanlon, B., Healy, R., & Cook, P. (2002). Choosing appropriate techniques for quantifying groundwater recharge. Hydrogeology Journal, 10(1), 18-39.

Scanlon, B. R., Mace, R. E., Barrett, M. E., & Smith, B. (2003). Can we simulate regional groundwater flow in a karst system using equivalent porous media models? Case study, Barton Springs Edwards aquifer, USA. Journal of Hydrology, 276(1-4), 137-158.

Sharda, V. N., Kurothe, R. S., Sena, D. R., Pande, V. C., & Tiwari, S. P. (2006). Estimation of groundwater recharge from water storage structures in a semi-arid climate of India. Journal of Hydrology, 329(1-2), 224-243.

Simmers. (1988). Estimation of natural groundwater recharge. Dordrecht . D. Reidel. Simmers, Hendrickx, J. M. H., Kruseman, G. P., & Rushton, K. R. (1997). Recharge of phreatic

aquifers in sem -arid areas. Rotterdam . Balkema. Singh, Kunwar P, Malik, Amrita, Sinha, Sarita, et al. (2007). Exploring groundwater hydrochemistry

of alluvial aquifers using multi-way modeling. Analytica Chimica Acta, 596(1), 171-182. Snow, D. T. (1969). "Anisotropic Permeability of fructured Media." Water Resource Research 5:86-

95. TAHAL. (2007). Mekele Water supply development Project present water supply source assessment. Teklay, Z. (2006). Conjugate use of surface and Groundwater of Aynalem area, Mekle University. Vernier. (1985). Groundwater investigation in Tigray region-Mekele. Water well Drilling

Enterprise,Addis Ababa. Villholth, K. (2006). Groundwater assessment and management: implications and opportunities of

globalization. Hydrogeology Journal, 14(3), 330-339. WWDSE. (2006). Evaluation of Aynalem wellfield and selection of additional prospective boreholes

for Mekele Town Water supply. Yehdego. (2003). Hydrogeology of Ilala-Aynalem Catchments with Particular reference to the

chemical Variation and Aquifer characterization. Unpublished Msc thesis, Addis Ababa University.

Yihdego, Y. (2005). Three dimensional ground water model of the aquifer around lake Naivasha area, Kenya. Unpublished MSc Thesis, ITC, Enschede.

Yongxin, X., & Beekman, H. E. (2003). Groundwater recharge estimation in southern Africa. Paris: United Nations Educational Scientific and Cultural Organization (UNESCO).


79

Yuri Mun, Christopher G, & ., U. (2004). Development and Application of a MODFLOW pre-processor using percolation theory for fructured media. Journal of the American Water Resources Association. 40 (1), 229–239.


80


81

Appendices

Appendix 1 Hydrometeorological data

Appendix 1.1. Monthly rainfall (mm) at Mekele airport station

Appendix 1.2. Long term monthly rainfall (mm) at Mekele airport station


82


83

Appendix 1.3. Monthly minimum temperatures (0C)

Appendix 1.4. Monthly maximum temperature (0C)

Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

1993 21 22 23.6 22.2 24 25.3 22.3 22.9 24.4 23 22 22

1994 22 23 24.2 24.9 26.1 25.7 21.3 21 22.5 23 22 21

1995 22 24 24.6 24.6 25.8 28.2 22.9 22.1 23.9 23 23 23

1996 23 25 24.9 25.6 24.8 24.4 23.2 22.5 24.9 24 22 22

1997 23 24 25.7 25.5 26.6 26.6 22.8 23.1 25.6 23 23 23

1998 24 25 26.2 27.3 27.0 27.9 22.4 21.3 23.9 23 22 22

1999 22 25 25.0 26.3 27.9 27.9 21.7 21.4 23.5 23 22 22

2000 23 24 24.7 25.6 27.5 27.6 23.6 22.4 23.9 24 23 22

2001 23 25 24.5 26.5 28.1 25.5 24.6 21.9 24.6 25 23 23

2002 22 25 25.8 26.6 28.7 27.3 25.5 23.3 24.8 25 24 23

2003 25 26 25.7 26.6 28.2 26.9 23.4 22.3 24.3 24 23 22

2004 25 24 25.0 25.9 28.2 26.5 24.8 22.9 25.1 21 23 23

2005 24 26 26.1 26.3 26.4 27.4 23.2 23.3 24.6 23 23 22

2006 24 25 25.5 25.0 26.0 27.1 23.6 22.3 24.5 24 23 22

mean 23 24 25.1 25.6 26.8 26.7 23.2 22.3 24.3 23 23 22


84

Appendix 1.5. Monthly mean wind speed (m s-1) at 2m height


1996 3.6 4.0 3.79 4.08 2.96 1.92 2.04 1.64 2.00 3.3 3.6 3.7

1997 4.4 4.9 3.86 3.79 3.63 2.25 1.68 1.39 2.32 3.7 3.8 3.8

1998 3.5 3.2 4.36 4.78 3.42 2.39 2.33 2.31 1.58 2.7 3.3 3.6

1999 3.5 5.7 17.5 4.25 2.9 2.83 1.86 2.20 1.49 2.7 3.4 3.6

2000 3.7 4.4 4.72 3.44 2.99 2.17 2.05 1.89 1.71 2.8 3.1 3.3

2001 2.8 3.5 3.22 3.74 2.67 1.89 2.31 1.56 1.79 2.8 3.2 3.5

2002 3.5 3.6 3.19 3.43 2.67 2.01 1.54 1.52 1.96 3.0 3.4 3.1

2003 3.9 3.3 3.81 3.63 3.07 2.01 1.99 1.61 1.49 3.6 3.9 4.1

2004 3.6 4.0 4.16 4.04 3.01 1.87 1.49 1.48 2.01 3.3 4.0 4.1

2005 3.4 4.6 4.51 4.46 2.95 3.78 2.11 1.48 1.45 2.8 3.6 4.2

2006 4.0 4.3 3.79 4.04 2.91 2.00 1.71 1.65 1.47 3.2 3.8 3.9

Appendix 1.6. Mean monthly relative humidity (%) at 1200 local time


1993 48 42 38 68 65 67 72 69 53 50 54 44

1994 43 47 46 34 31 39 69 74 56 34 52 42

1995 38 38 40 45 39 39 63 67 40 36 35 44

1996 47 34 42 35 43 49 60 70 44 33 39 33

1997 37 31 34 32 30 45 68 62 33 42 45 35

1998 45 33 33 29 29 28 69 76 48 36 28 29

1999 36 20 28 24 20 26 72 75 47 45 39 42

2000 30 21 34 28 26 30 63 73 44 42 38 35

2001 38 29 37 29 24 38 70 76 44 38 34 30

2002 45 31 34 27 20 33 56 68 41 36 34 43

2003 31 32 34 34 24 36 64 74 46 36 34 34

mean 40 33 36 35 32 39 66 71 45 39 39 37

Appendix 1.7. Mean monthly sunshine hours


1993 9.7 8.4 9.5 8.4 9.3 7.1 5.4 5.9 7.1 8.5 10.4 10.2

1994 10.3 9.9 9-0 9.5 10.1 6.4 4.4 5.1 7.9 10.5 9.8 10.2

1995 10.3 8.9 9.3 9.1 9.3 9.2 5.4 5.1 8.6 9.8 10.0 9.4

1996 9.0 9.6 8.2 9.2 8.4 5.9 6.1 5.7 7.7 9.8 9.0 9.9

1997 9.5 9.9 8.6 9.1 9.6 8.0 6.0 6.5 8.4 8.1 8.9 10.0

1998 8.4 8.7 9.1 9.4 9.4 7.3 4.9 4.1 7.1 9.2 10.2 10.3

1999 9.3 10.3 9.7 10.4 9.9 6.9 3.9 5.2 8.1 8.9 10.4 9.9

2000 10.1 10.0 10.0 7.8 9.6 7.7 6.7 6.3 6.9 9.0 9.0 9.4

2001 9.6 9.7 6.4 9.6 10.1 9.8 6.5 6.9 8.6 9.3 10.3 10.1

2002 9.3 10.1 8.9 10.4 10.6 11.9 6.2 7.4 8.5 10.3 10.0 9.6


85

Appendix 1.8. Monthly average piche evaporation (mm)

Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual

1993 135 153 212 128 179 149 69 82 135 201 220 225 1887.5

1994 250 194 202 266 308 178 85 111 153 229 195 207 2377.8

1995 220 195 197 219 287 186 93 59 171 249 204 190 2268.5

1996 151 247 177 133 219 174 119 62 203 232 213 262 2192.1

1997 219 295 263 303 353 253 132 118 225 242 188 222 2810.8

1998 166 244 302 374 391 420 128 75 174 240 317 354 3184.4

1999 295 432 328 451 365 287 89 69 140 152 218 178 3003.2

2000 213 264 275 234 257 240 105 63 142 141 153 167 2255.7

2001 119 189 174 258 254 170 74 53 139 173 180 198 1981.2

2002 128 213 208 253 328 240 140 79 162 239 223 142 2354.6

2003 177 196 237 271 323 201 114 69 131 222 245 219 2403.5

2004 182 263 346 210 331 224 145 81 170 240 236 199 2625.5

2005 177 283 233 275 234 235 98 92 129 214 196 278 2442.4

2006 192 185 220 247 254 236 103 65 149 219 192 115 2175.7

mean 187 239 241 259 292 228 107 77 159 214 213 211 2425.9

Appendix 1.9. Monthly Evapotranspiration (mm)


1992 3.7 4.4 5.7 7.5 7.9 8.6 3.6 1.7 3.9 5.8 5.4 4.1

1993 4.3 5.3 6.8 4.3 5.8 5.0 2.2 2.7 4.5 6.5 7.3 7.3

1994 8.1 6.7 6.5 8.9 9.9 5.9 2.7 3.6 5.1 7.4 6.5 6.7

1995 7.1 6.7 6.3 7.3 9.3 x 3.0 1.9 5.7 8.0 6.8 6.1

1996 4.9 8.5 5.7 4.4 7.1 5.8 3.8 2 6.8 7.5 7.1 8.5

1997 7.1 10.2 8.5 10.1 11.4 8.4 4.2 3.8 7.5 7.8 6.3 7.1

1998 5.3 8.4 9.7 12.5 12.6 14 4.1 2.4 5.8 7.7 10.6 11.4

1999 9.5 14.9 10.6 15.0 11.8 9.6 2.9 2.2 4.7 4.9 7.3 5.7

2000 6.9 9.1 8.9 7.8 8.3 8.0 3.4 2.0 4.7 4.6 5.1 5.4

2001 3.8 6.5 5.6 8.6 8.2 5.7 2.4 1.7 4.6 5.6 6.0 6.4

2002 4.1 7.4 6.7 8.4 10.6 8.0 4.5 2.6 5.4 7.7 7.4 4.6

2003 5.7 6.8 7.7 9.0 10.4 6.7 3.7 2.2 4.4 7.2 8.2 7.1

2004 5.9 9.1 11.1 7.0 10.7 7.5 4.7 2.6 5.7 7.8 7.9 6.4

2005 5.7 9.7 7.5 9.2 7.5 7.8 3.1 3.0 4.3 6.9 6.5 9.0


86

Appendix 1.10. River discharge Metere gauging station (Aynalem river)

UTM E-553925

N-1486920 Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

1992 I 0.00 0.00 0.02 0.00 0.07 0.00 0.17 0.79 0.04 0.00 0.02 0.00

II 0.00 0.00 0.13 0.00 0.36 0.00 0.52 2.03 0.07 0.00 0.07 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

1993 I 0.00 0.00 0.00 0.10 0.08 0.01 0.17 1.08 0.44 0.02 0.01 0.00

II 0.00 0.00 0.00 0.41 0.20 0.04 0.20 9.10 4.15 0.01 0.04 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.01 0.00 0.00 0.00

1994 I 0.00 0.00 0.00 0.00 0.00 0.04 0.19 3.84 1.93 0.03 0.00 0.00

II 0.00 0.00 0.00 0.02 0.00 0.36 0.76 15.15 7.59 0.02 0.00 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.00 0.00

1995 I 0.00 0.00 0.00 0.00 0.01 0.10 0.15 0.55 0.99 0.02 0.00 0.00

II 0.00 0.00 0.03 0.00 0.07 0.23 0.20 0.96 3.81 0.01 0.00 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00

1996 I 0.01 0.00 0.00 0.00 0.02 0.08 0.35 0.22 0.04 0.00 0.00 0.00

II 0.00 0.00 0.00 0.00 0.22 0.22 2.47 0.27 0.04 0.00 0.00 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00

1997 I 0.00 0.00 0.00 0.00 0.05 0.04 1.70 0.51 0.26 0.00 0.00 0.00

II 0.00 0.00 0.00 0.00 0.41 0.03 10.19 1.35 0.68 0.00 0.00 0.00

III 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.00

1998 I 0.00 0.00 0.00 0.00 0.00 0.27 2.02 2.35 0.49 0.18 0.20 0.15

II 0.00 0.00 0.00 0.00 0.00 1.10 4.71 6.83 1.35 0.09 0.10 0.07

III 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.08 0.06 0.05 0.03

1999 I 0.05 0.01 0.00 0.00 0.00 0.00 0.98 3.90 0.97 0.47 0.24 0.15

II 0.03 0.01 0.00 0.00 0.00 0.00 1.00 6.68 2.8 1.96 0.10 0.07

III 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.12 0.10 0.07 0.04

2000 I 0.05 0.00 0.00 0.00 0.00 0.00 0.18 0.81 0.04 0.00 0.00 0.00

II 0.04 0.00 0.00 0.00 0.00 0.01 0.72 0.96 0.09 0.00 0.00 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

2001 I 0.00 0.00 0.00 0.00 0.00 0.00 1.32 2.22 0.20 0.06 0.03 0.00

II 0.00 0.00 0.00 0.00 0.00 0.00 6.25 5.45 0.12 0.04 0.02 0.00

III 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.04 0.01 0.00 0.00

I is monthly Runoff in million m3 II is maximum Discharge in m3s-1 III is minimum Discharge in m3s-1


87

Appendix 2 Hydrochemistry

Appendix 2.1. Analysis result of rain water

Appendix 2.2. Physical and chemical constituents of water samples

Appendix 2.3. Comparison of analysis


88

Appendix 2.4. Major anions and cations ( meq l-1) and water type

Well

Na+ K+ Ca 2+ Mg 2+ Cl - SO4 2- NO3

- HCO3 - Water type Ionic

balance

PW11 2.1 0.1 15.90 0.76 0.5 14.29 0.04 2.93 Ca-SO4 -2.9

PW4 0.9 0.0 7.56 1.05 0.4 3.63 0.20 4.37 Ca-HCO3-SO4

-5.3

PW8 0.8 0.0 5.45 0.71 0.4 0.99 0.31 5.23 Ca-HCO3 -1.0

PW2 1.3 0.0 9.64 0.63 0.5 5.22 0.24 4.75 Ca-SO4-HCO3

-3.8

TW2 1.5 0.1 10.50 0.76 0.6 6.59 0.04 4.08 Ca-SO4-HCO3

-6.2

TW3 1.7 0.1 12.60 1.05 0.5 9.89 0.08 3.55 Ca-SO4-HCO3

-5.2

L.Ilala 3.5 0.1 25.10 3.69 2.1 22.92 0.04 4.56 Ca-SO4 -4.6

PW6 2.4 0.1 21.40 0.84 0.4 18.58 0.05 3.31 Ca-SO4 -5.0

PW9 0.9 0.0 7.55 0.46 0.4 3.13 0.23 4.46 Ca-HCO3-SO4

-4.7

TW1 2.6 0.1 15.90 3.23 1.6 18.68 0.06 0.24 Ca-SO4 -3.1

Pw3 1.4 0.0 8.80 0.16 0.6 5.83 0.23 4.71 Ca-SO4-HCO3

4.5

PW7 0.9 0.0 7.13 0.76 0.5 1.32 0.32 5.76 Ca-HCO3 -5.5

U.Ilala 0.9 0.0 8.38 0.84 1.2 2.03 1.05 5.28 Ca-HCO3-SO4

-3.3

Dandera1 0.7 0.1 5.46 1.11 0.3 1.43 0.18 5.62 Ca-HCO3 0.9

Tw1(2005) 0.9 0.1 10.50 0.35 0.6 5.83 0.20 5.24 Ca-SO4-HCO3

0.5

PW-12 1.2 0.2 7.56 1.06 0.5 4.29 0.17 4.87 Ca-HCO3-SO4

-0.8

PW-7B 1.6 0.2 8.40 0.72 0.4 7.69 0.07 2.50 Ca-SO4-HCO3

-1.3

PW-1 1.0 0.1 2.69 1.53 0.5 3.19 0.08 0.82 Ca-Mg-Na-SO4

-7.9

piz-1S 1.4 0.1 7.29 1.07 0.4 5.27 0.23 3.02 Ca-SO4-HCO3

-4.8

Mu 2.4 0.2 18.5 0.81 0.6 16.90 0.05 2.54 Ca-SO4 -4.4

piz-3S 1.1 0.1 6.72 0.55 0.6 3.90 0.55 3.31 Ca-SO4-HCO3

-0.7

Endbotarek 1.6 0.2 10.50 0.55 0.6 6.77 0.05 4.03 Ca-SO4-HCO3

-5.7

piz-2D 0.7 0.1 6.08 0.93 0.4 0.83 0.47 5.28 Ca-HCO3 -5.4


89

Appendix 3 Well data

Appendix 3.1. Well location

Appendix 3.2. Monthly water production (m3)


90


91

Appendix 3.3. Monthly groundwater level monitoring data

Appendix 3.4. Static water level record from the wells


92

Appendix 3.5. Lithologic log data of the boreholes


93


94


95


96


97

Appendix 4 Geophysical data

Vertical electrical sounding data Where

• NMN/2 is potential electrode separation

• AB/2 is electrical electrode separation

• Ra is apparent Resistivity in ohm meter


98

Figures showing the vertical electrical sounding interpretation using Ipi2 software


99

Seismic profile and cross-sections (Gebregziabher, 2003)


100

Appendix 5 Groundcontrol points to correct ASTER DEM X Y Z topo Z STER Corrected

Z

X Y Z topo Z

ASTER

Corrected

Z

540422 1490899 2261 2256.5 2266.2 553616.9 1485035 2285 2251.5 2261

540868.2 1489632 2245 2236.7 2245.9 554988.1 1485712 2275 2256.5 2266.2

540650.5 1491505 2267 2245 2254.4 555695.4 1490205 2304 2272.8 2282.9

541308.9 1491673 2225 2220.2 2228.9 554521.1 1491978 2326 2299 2309.8

542631.1 1491154 2170 2166.2 2173.5 557702.7 1490473 2226 2219.2 2227.9

543726.2 1490352 2147 2145.9 2152.7 559819.3 1491648 2149 2141.9 2148.6

544118 1488900 2162 2140 2146.6 559672.4 1488481 2254 2251.8 2261.4

546674.2 1488451 2110 2109.9 2115.8 559694.2 1486015 2278 2265.3 2275.2

547974.6 1488337 2127 2125.3 2131.6 558399.2 1477005 2231 2227.3 2236.2

548148.7 1489755 2147 2133.4 2139.9 558562.4 1472989 2398 2385.4 2398.4

548932.2 1488976 2167 2157.7 2164.8 556696.1 1470466 2209 2194.5 2202.6

549737.5 1488158 2164 2158.5 2165.6 558078.2 1468716 2148 2145.3 2152.1

547120.3 1486634 2097 2097.9 2103.4 556756 1476148 2292 2296.7 2307.4

548165 1487294 2109 2109.4 2115.2 556951.9 1483344 2295 2288.4 2298.9

548578.6 1486449 2107 2099.2 2104.8 557017.2 1484806 2385 2366.8 2379.4

549628.7 1486439 2110 2110.6 2116.5 563859.5 1488514 2346 2331.3 2342.9

546124.6 1485479 2154 2146.4 2153.2 562368.6 1493072 2290 2278.8 2289.1

548143.3 1484845 2168 2167.4 2174.8 563500.4 1495800 2386 2384.6 2397.6

550920.2 1487249 2154 2147.4 2154.2 564664.8 1499378 2344 2352.1 2364.3

551116.1 1485159 2172 2163 2170.2 564788.2 1496998 2487 2489.9 2505.7

550827.7 1486448 2126 2117.2 2123.2 562932.8 1497925 2418 2419.8 2433.7

564167.9 1501861 2347 2335.1 2346.8

566567.6 1499879 2332 2337.6 2349.4

565016.9 1494060 2384 2385.7 2398.7

565746 1489546 2380 2354.7 2366.9

566866.9 1485316 2385 2368.8 2381.4

564494.5 1484656 2339 2327.2 2338.7

563351.9 1483234 2332 2318.5 2329.8

565517.5 1483261 2430 2412.6 2426.3

565675.3 1482498 2393 2368.2 2380.8

562889.4 1478739 2300 2298.3 2309.1

562867.1 1476554 2267 2263.8 2273.7

560581.8 1475748 2256 2247.7 2257.1

556134 1473132 2335 2330.4 2342

549398.9 1472842 2228 2212.7 2221.2

548544.6 1472815 2337 2303.8 2314.7

549562.1 1471753 2230 2213.6 2222.2

548365.1 1478871 2005 2002.2 2005.3

547707.5 1479872 2048 2038.2 2042.2

548844.7 1480607 2150 2129.8 2136.2

Topo elevation vs ASTER DEM elevation

y = 0.9746x + 47.882

R2 = 0.9933

1700

1900

2100

2300

2500

2700

1700 1900 2100 2300 2500 2700

Topo elevation(m)

AS

TE

R D

EM

ele

vati

on

(m

)


101

Appendix 6 Location of all wells in the wellfield


102

Appendix 7 MODFLOW water budget


103

Appendix 8 Pumping test curve matching


104


105

Appendix 9 Plates


106

groundwater resource assessment through distributed steady

Documents