See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/225572537

Estimation of groundwater recharge using a

GIS-based distributed water balance model in
Dire Dawa, Ethiopia

Article in Hydrogeology Journal · September 2009

DOI: 10.1007/s10040-009-0455-x

CITATIONS READS

14 711

2 authors:

Ketema Tilahun Zeleke Broder Merkel

Charles Sturt University Technische Universität Bergakademie Frei…
28 PUBLICATIONS 239 CITATIONS 133 PUBLICATIONS 560 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Broder Merkel
letting you access and read them immediately. Retrieved on: 19 September 2016
Estimation of groundwater recharge using a GIS-based distributed
water balance model in Dire Dawa, Ethiopia

Ketema Tilahun & Broder J. Merkel

Abstract Sustainable groundwater management requires Groundwater potential is directly dependent on recharge.
knowledge of recharge. Recharge is also an important Groundwater recharge can be defined as the entry into the
parameter in groundwater flow and transport models. saturated zone of water made available at the water-table
Spatial variation in recharge due to distributed land-us.e, surface (Freeze and Cherry 1979). It is the most important
soil texture, topography, groundwater level, and hydro- parameter for groundwater sustainability in arid and
meteorological conditions should be accounted for in semiarid regions because the abstraction from a ground-
recharge estimation. However, conventional point-estimates water reservoir should, in the long term, not be larger than
of recharge are not easily extrapolated or regionalized. In this the long-term average recharge. Quantifying groundwater
study, a spatially distributed water balance model WetSpass recharge is thus a prerequisite for efficient and sustainable
was used to simulate long-term average recharge using land- groundwater resource management. As aquifers are
use, soil texture, topography, and hydrometeorological depleted, recharge estimates have become more essential in
parameters in Dire Dawa, a semiarid region of Ethiopia. determining appropriate levels of groundwater withdrawal
WetSpass is a physically based methodology for estimation (de Vries and Simmers 2002). Unfortunately, there has been
of the long-term average spatial distribution of surface no such study conducted in the Dire Dawa groundwater
runoff, actual evapotranspiration, and groundwater recharge. basin, a semiarid region in eastern Ethiopia.
The long-term temporal and spatial average annual rainfall As there are no surface-water resources, groundwater is
of 626 mm was distributed as: surface runoff of 126 mm the sole source of water supply for domestic, agricultural,
(20%), evapotranspiration of 468 mm (75%), and recharge and industrial use in the Dire Dawa area. Dire Dawa is the
of 28 mm (5%). This recharge corresponds to 817l/s for the second largest industrial and commercial town in the
920.12 km2 study area, which is less than the often-assumed country. The Sabian well field, located in the town,
1,000l/s recharge for the Dire Dawa groundwater catchment. supplies water to the town. In addition, there are a number
of wells owned by individuals for domestic and industrial
Keywords Ethiopia . Geographic information systems . uses. Dire Jara well field is also soon to be operational to
Groundwater recharge/water budget . Water balance . supply water at a rate of 240 l/s to Harar, which is another
WetSpass large and historical town about 65 km south-east of the
Dire Jara area. The groundwater level in the Sabian well
field has been dropping over the years to such an extent
Introduction that at present some of the wells (e.g. PW1, PW2, and
PW3) are dry. The water level measured in four boreholes
Knowledge of groundwater resource potential is important (PW5, PW6, PW7, and PW9) showed an average water
for groundwater management and sustainable use. level decline of 11 m from its initial measured value in
1989 (Dire Dawa Administrative Council integrated
Received: 8 June 2008 / Accepted: 2 March 2009
Published online: 24 March 2009 resource development master plan study project, water
works design and supervision enterprise (WWDSE), Vol. III
© Springer-Verlag 2009 (Water resources), Addis Ababa, Ethiopia, unpublished,
2004; cited hereafter as ‘WWDSE 2004 unpublished’). This
K. Tilahun : B. J. Merkel indicates, at least, the existence of imbalance between
Department of Hydrogeology, abstraction and replenishment. However, the only informa-
Technische Universität Bergakademie Freiberg, tion available on recharge in Dire Dawa groundwater basin is
Gustav-Zeuner Straße 12, 09599, Freiberg, Germany various estimates based on very crude assumptions (Harar
Present Address: water supply project groundwater resources. Ministry of
K. Tilahun ()) Water Resources, Ethiopia, unpublished, 2005 (cited here-
School of Agricultural and Wine Sciences, after as BCEOM-WWDSE-CECE 2005 unpublished);
Charles Sturt University, Wagga Wagga, NSW 2650, Australia Greitzer 1970; WWDSE 2004 unpublished). These studies
e-mail: [email protected]
Tel.: +49-3731-392039 used some simplified approaches to estimate recharge such
Fax: +49-373-1392720 as assumption of runoff coefficient and use of spring

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1444
discharge-area relationship (BCEOM-WWDSE-CECE Smedt 2007) was applied to estimate long-term seasonal/
2005 unpublished) and adoption of coefficients determined annual average recharge as a function of land-cover, soil
in other catchments (Greitzer 1970; WWDSE 2004 unpub- type, topography and hydrometeorological factors. Wet-
lished). Most of these reports have claimed that recharge in Spass is an acronym for Water and Energy Transfer
Dire Dawa groundwater basin is more than 6% of annual between Soil, Plants, and Atmosphere under quasi-Steady
rainfall. State. It is a physically based methodology for estimation
Recharge can be estimated using a number of of the long-term average, spatially varying, water-balance
approaches depending on the availability of data and level components: surface runoff, actual evapotranspiration, and
of accuracy required, as follows: (1) inflow estimation groundwater recharge. The GIS structure of the model,
such as soil moisture budget and tracers (Allison et al. with parameters as attribute tables, has been shown to be
1994); (2) aquifer response (water level fluctuation advantageous in transferring it to other environments
method; Sophocleous 1991); (3) catchment water balance (Batelaan and De Smedt 2007). WetSpass was, for
(Essery and Wilcock 1990) and chloride and bromide mass example, successfully applied in the Dijle-Demer-Nete
balance methods (Allison and Hughes 1978; Bazuhair and catchments of Belgium (Batelaan and De Smedt 2007),
Wood 1996). Recharge can also be estimated with numerical the Upper-Beibrza catchment, Poland (Thijs 2002), and
groundwater models using inverse techniques. However, the Geba Basin in Tigray, Ethiopia (Asfaw 2005).
models do not produce unique solutions, so should not be
relied upon as a sole technique for estimating recharge. In
arid and semiarid areas, recharge is relatively small and Methodology
potentially more variable (Allison et al. 1994). As such, the
use of techniques that can handle spatial variability is Description of the study area
necessary. In this regard, geographic information systems The region known as the Dire Dawa Administrative
(GIS) have emerged as effective tools for handling spatial Council (DDAC), area 1,333 km2, is found at the margin
data and decision making in several areas including of the eastern part of the Ethiopian rift valley. It lies
engineering, geology and environmental fields. between 09°28′ to 09°49′N latitude and 41°38′ to 42°19′E
In this study, a distributed water balance recharge longitude (Fig. 1). It is characterized by very diverse
estimation technique called WetSpass (Batelaan and De spatial variation of topographic features with an altitude

Fig. 1 The study site showing DDAC, Dire Dawa groundwater basin, well fields, towns, and surface drainage

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1445
range of 950–2,600 meters above sea level (masl). The supply well fields (Sabain and Dire Jara, respectively) tap
mountainous areas are located at the southern and south- water from the upper sandstone and limestone aquifers
eastern parts with slopes reaching up to 40%. The flat which are considered to be the major aquifers in the area.
plains lie mainly on the northern and north-western part of The alluvial aquifer is extensive at Dire Dawa town and
the DDAC with a general slope of less than 3%. north of it; west of Dire Dawa town the occurrence of
Two of the weather stations (at Dire Dawa and Hurso) groundwater in this formation is limited along alluvial
are located in the plain area while the other three (at fans and river channel deposits. However, the alluvial
Dengego, Kersa, and Kulubi) are located on the escarp- aquifer has only limited productivity (WWDSE 2004
ment (Fig. 1). The climatic situation is characterized unpublished). The Dire Dawa groundwater basin is a plain
mainly by warm and semiarid climate with relatively low area surrounded by mountains (escarpment). The conceptual
rainfall (Tilahun 2006). The mean annual maximum and model for Dire Dawa groundwater system is that the surface
minimum temperatures are 28 and 17°C respectively and water divide is also the groundwater divide. The outlet of the
mean annual rainfall is 626 mm. The year is divided into surface water flow, north of the plain area, is considered as
three seasons: a main rainy season (Kiremt) from mid- the direction towards which the groundwater flows. The
June to September, a dry season (Bega) from October to hydraulic head in the boreholes has confirmed that the
February, and a small rainy season (Belg) in March and direction of groundwater flow is from the south and south-
April. Potential evapotranspiration, the ability of the western part towards the north and north-eastern part,
atmosphere to remove water from a surface through the following the surface drainage system.
processes of evaporation and transpiration, is about
1,660 mm. Actual evapotranspiration, the quantity of
water that is actually removed from a surface, would be
much lower. The WetSpass methodology
Not all parts of DDAC drain towards the Dire Dawa WetSpass is a methodology for estimating spatially
groundwater basin. In order to determine the catchment distributed, long-term average recharge. It integrates GIS
area for Dire Dawa groundwater basin, the topographic and a water balance method (Batelaan and De Smedt
divide was assumed as a groundwater divide. As such, the 2001; Batelaan and De Smedt 2007). By using long-term
catchment draining towards the Dire Dawa groundwater average standard hydrometrological parameters as input,
basin was differentiated from the rest of the DDAC the model simulates the temporal average and spatial
catchment. The area of Dire Dawa groundwater system is, differences of surface runoff, actual evapotranspiration,
therefore, about 920 km2 (Fig. 1). and groundwater recharge. Since evapotranspiration from
shallow groundwater can be significant, especially in
groundwater dependent wetlands, the position of the water
Surface drainage and hydrogeology table is taken into account, through the use of a coupled
The DDAC has two geographic settings: the escarpment groundwater model, in the estimation of recharge. Wet-
and the plain area. The escarpment occupies the southern, Spass is integrated in GIS ArcView as a raster model.
south-eastern and eastern part of the DDAC. The drainage Parameters such as land-use and related soil type, are
network of DDAC is presented in Fig. 1. Generally, connected to the model using attribute tables of the land-
surface water drains from the southern escarpment area use and soil raster maps.
towards the plain area in the north. The rainfall in Dire The seasonal water balance for a vegetated fraction of a
Dawa area is high intensity, short-duration rainfall result- raster cell (Fig. 2) is
ing in floods that have disastrous effects on property and
human life in Dire Dawa town. After a day or two, there is P ¼ I þ Sv þ Tv þ Rv ð1Þ
no flow in the stream channels draining the high-velocity
water from the escarpment area. As a result, there is no where P is the precipitation [L], I the interception [L], Sv
baseflow in the DDAC catchment area. the surface runoff [L], Tv the actual transpiration [L] and
The escarpment area has highly rugged topography and Rv the groundwater recharge [L] in the vegetated fraction
is intensively faulted by east-west trending faults. On the of the raster cell. The actual evapotranspiration, ETv [L],
escarpment are outcrops of Precambrian rocks, Adigrat is the sum of Tv and the evaporation from the bare soil, Es
(lower) sandstone, Hamanalei (Antalo) limestone, Amba- [L]. The total actual evapotranspiration, ETtot [L], is the
aradam (upper) sandstone, and basalts (Hydrogeology and sum of I, Tv and the evaporation from the bare soil, Es [L].
hydrochemistry of the Dire Dawa area. Ethiopian Institute The surface runoff, Sv, is simulated in two stages. First,
of Geological Survey, Addis Ababa, Ethiopia, Zerai H and the potential surface runoff, Sv–pot, is calculated as:
Sima J unpublished 1986). The foot of the escarpment
occupies the northern and north-western part of the Sv
pot ¼ f1 ðV ; ST ; S; DÞðP
I Þ ð2Þ
DDAC, which is dominantly covered by alluvial deposits.
In general the stratigraphy of the plain area, in sequence where f1 is a runoff factor from a lookup table whose
from top to bottom, can be described as alluvial deposit, value is dependent on vegetation type (V), soil texture
basalt, upper sandstone/limestone, lower sandstone, and (ST), slope (S), and groundwater saturated areas (D), and
basement rock. The Dire Dawa and Harar towns' water is based on characteristic values derived from Smedema

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1446
estimation of spatially distributed recharge as a function of
vegetation, soil texture, slope, groundwater depth, and
precipitation regime. Recharge was determined from the
water balance for bare soil surfaces as:

P ¼ Ss þ Es þ Rs ð4Þ

where the index s refers to bare soil surfaces. The surface

runoff, Ss, was simulated in a similar way to the vegetated
area fraction in two stages—Eqs. (2) and (3). The water
balance for the open water area fraction of a cell is defined
as:

P ¼ So þ Eo þ Ro ð5Þ

where the index o stands for open water surfaces. It is

assumed that the recharge, Ro, derived from the precipi-
tation on the open water fraction, is negligible compared
to the possible recharge from the surface water body itself.
If the Penman open water evaporation, Eo, is smaller than
the precipitation, the remaining water will contribute to
the surface runoff, So (Batelaan and De Smedt 2007). That
Fig. 2 Schematic water balance of a hypothetical raster cell is why there is no Ro term in Eq. (7c).
(adapted from Batelaan and De Smedt 2007) The water balance for impervious surfaces is given as:

P ¼ Si þ Ei þ Ri ð6Þ
and Rycroft (1988), Pilgrim and Cordery (1992), and
Chow et al. (1988). In the second stage, the potential where the index i refers to impervious surfaces. The total
surface runoff is adjusted for recharge areas by taking into water balance, per raster cell and hydrological season is
account the seasonal precipitation intensity distribution stated as:
(Pi) in relation to the soil infiltration capacity (Ic; Rubin,
1966) as: ET c ¼ av ET v þ as Es þ ao Eo þ ai Ei ð7aÞ

Sv ¼ f2 ðPi ; Ic ; DÞSv
pot ð3Þ

where f2 is a factor from a soil texture lookup table that Sc ¼ av Sv þ as Ss þ ao So þ ai Si ð7bÞ

partitions the precipitation for a hydrological season, in an
effective and non-effective part for contribution to the
surface runoff. It can be derived by estimating the fraction
of seasonal precipitation with intensity higher than the Rc ¼ av Rv þ as Rs þ ai Ri ð7cÞ
infiltration capacity of a particular soil type.
Summer and winter hourly rainfall data of Dire Dawa where the index c refers to raster cell, with ETc, Sc, Rc [L]
weather station (1972–2006) was used to determine the respectively, the total evapotranspiration, surface runoff
cumulative rainfall fraction with intensity higher than the and recharge in a raster cell and av, as, ao and ai
infiltration capacity of the respective soil textural classes. respectively the vegetated, bare soil, open water and
This fraction was used to partition the summer and winter impervious area fraction of a raster cell. In a vegetated or
rainfall into effective and non-effective with regard to cropped area, there is evaporation from the soil in between
runoff. The basic infiltration capacity of the soils (mm/h) the plants and transpiration from vegetation, resulting in
was assumed as following: loamy sand (96.7), sandy loam evapotranspiration (ETv) in Eq. (7a). From the portion of
(50.3), loam (15.5), sandy clay loam (11.3), silty clay the catchment on which there is no vegetation (i.e.
(3.7), and sandy clay (1.4; Saxton and Rawls 2006). practically bare land), only evaporation from the bare soil
Reference evapotranspiration was estimated from sea- occurs, Es in Eq. (7a).
sonal climatic data (Allen et al. 1998; Federer 1979).
Actual evaporation and transpiration (evapotranspiration)
is estimated taking into account the soil moisture content; The input data
the detail of which is presented in Batelaan and De Smedt The interaction of climate, geology, morphology, soil
(2007). condition, and vegetation determines the recharge process
Groundwater recharge was determined as the residual (de Vriers and Simmers 2002). These input data were
of the water balance equation (Eq. 1). This resulted in an prepared in the form of maps of selected meteorological,

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1447
hydrological, and geographical elements in the basin. The respective elevation. As shown in Table 1, the two can be
climate data for five stations in the Dire Dawa catchment linearly related. This relation was used to prepare a
was obtained from Ethiopian Meteorological Services potential evapotranspiration map of the catchment
Agency. Two of the stations (Dire Dawa and Hurso) are (Fig. 4). The potential evapotranspiration data obtained
located in the plain area while the other three (Dengego, from Fig. 4 are given in Table 2.
Kersa, and Kulubi) are located on the escarpment (Fig. 1).
Only Dire Dawa provides complete and more than 3. Temperature
50 years of climate data. The other stations have only
daily rainfall and temperature data for periods: Dengego
(1981–2006), Dire Dawa (1953–2006), Hurso (1981– Temperature data are available for all the five weather
2006), Kulubi (1995–2006), and Kersa (1995–2006). stations. A linear regression (Table 1) was used to
The WetSpass model was applied to the study area generate a temperature variation map (Fig. 5) from the
using grid cells of 30×30 m. The model functions on two plain area to the escarpment. The summery temperature
seasonal data sets. For this purpose the year was divided data are given in Table 2.
into two seasons with summer (April to September) and
winter (October to March) with respective input data 4. Wind speed
(land-use, precipitation, potential evapotranspiration, tem-
perature, wind speed, and groundwater depth). In the Dire
Dawa area, the period from mid-June to September is the Wind speed data are available only for Dire Dawa station.
major rainy season. Another, but smaller, rainy period is These data were used uniformly over the catchment: summer
March–April, with April having higher amount of precip- (2.7 m/s), winter (1.5 m/s), and annual (2.1 m/s).
itation. Farmers start cultivation in March. Therefore, it
was assumed that the precipitation in March initiates 5. Land-use
vegetation growth while it is only in April that full
vegetation cover is obtained. May is the dry month The land-use and land-cover of DDAC was mapped in
between the minor and major rainy periods. However, 2003 (WWDSE 2004 unpublished). The same was adapted
the vegetation growth and cultivation started during for the Dire Dawa groundwater basin (Fig. 6). The land-use
March would not be affected by this short dry period. in the study area is: cultivated land (41%), physiognomic
Therefore, from land-cover point of view, it is appropriate vegetation (29%), bare land (26%), and urban and built-up
to take the summer (rainy season in Ethiopian context) as area (3%). Most of the escarpment area is bare land. This
April to September and winter (dry season in Ethiopian area is highly degraded due to the steep topography,
context) as October to March. The input maps preparation deforestation, and intensive rainfall. The highland area
is presented in the following paragraphs. which was covered by sparse forest is now cultivated land.
The plain area is mainly covered by shrubs. In WetSpass, the
1. Precipitation land-cover change from summer to winter was taken into
account in the respective parameter tables.
Rainfall data of five weather stations were used to The cultivated land includes: intensively cultivated
develop a rainfall map of the catchment which is quite land with/without irrigation, moderately cultivated land,
well represented by linear equations as shown in Table 1. and sparsely cultivated land. Physiognomic vegetation
The plain area has low rainfall while the escarpment has includes dense shrub land, open shrub land, and eucalyp-
higher amount of rainfall. The annual rainfall map is tus plantation. Bare land means exposed rock outcrops
shown in Fig. 3. The summarized rainfall distribution and exposed soil surface with scrubs and grass vegetation.
obtained from Fig. 3 is presented in Table 2. The major town in the area is Dire Dawa and the other
small town is Melka Jebdu. The escarpment or southern
2. Potential evapotranspiration part of the study area is mainly cultivated land and bare
land while the physiognomic vegetation dominates the
valley plain or southern part of the catchment. Cultivation
The Penman Monteith method (Allen et al. 1998) was is mainly in the highlands where the rainfall amount is
used to calculate potential evapotranspiration. Evapotrans- higher with average rainfall of 690 mm while the average
piration data of the five stations were regressed against the rainfall in the lowland area is 572 mm.

Table 1 Linear regression between elevation and climatic data of the study area
Precipitation (mm) Temperature (°C) Potential ET (mm) Wind (m/s)
Equation R2 Equation R2 Equation R2
Winter 0.039Z + 107 0.600 −0.0066Z + 31 0.924 −0.176Z + 1107 0.936 1.5
Summer 0.231Z + 132 0.821 −0.0087Z + 37 0.915 −0.291Z + 1428 0.783 2.7
Annual 0.273Z + 238 0.880 −0.0077Z + 34 0.919 −0.467Z + 2535 0.871 2.1
Z = elevation (masl)

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1448

Fig. 3 Average annual precipitation of Dire Dawa catchment

6. Elevation and slope higher recharge rates than do fine-grained soils (Cook et al.
1992). The textural map of the catchment was adapted from
A 90 m digital elevation model (DEM) of the study the map prepared during the Dire Dawa Administrative
area was developed by mosaicking individual 1° blocks of Council master plan study (WWDSE 2004 unpublished).
three arc second (Fig. 7). Elevation varies from 1,028 to The soil textural class with corresponding area coverage is:
2,673 masl with an average of 1,503 masl. The slope was loam (41%), sandy clay loam (28%), sandy loam (19%),
derived from the elevation map using ArcView. The loamy sand (6%), sandy clay (3%), and silty clay (3%)
average slope of the major land-use classes in the study area (Fig. 9).
is: urban and built-up area (4% slope), bare land (10% Almost all of the urban land (94%) is covered with
slope), cultivated land (9% slope), and physiognomic sandy loam soil. About 73% of the bare land is loam soil.
vegetation (7% slope). The urban areas (i.e. Dire Dawa and The cultivated land is covered by 49, 23, and 13% sandy
Melka Jebdu) are located in the plain area. In general, the clay loam, loam, and loamy sand respectively while 41, 35
minimum, average, and maximum slopes of the area are 0, 8, and 16% of physiognomic land is loam, sandy loam, and
and 39% respectively (Fig. 8). sandy clay loam respectively.

7. Soil 8. Groundwater depth

Soil texture and permeability are important in recharge WetSpass can be run without running MODFLOW,
estimation because coarse-grained soils generally result in thus it is not a must that it should be run in conjunction
with MODFLOW as long as one has a good estimate of
Table 2 Hydrometeorological data of the study area the groundwater depth input for WetSpass modeling
Season Minimum Average Maximum (Batelaan and Woldeamlak 2003). In this study, the
groundwater depth is greater than 20 m and as such its
Precipitation (mm)
Winter 148 167 211 effect on WetSpass simulation is insignificant.
Summer 370 480 750
Annual 519 626 968
Potential evapotranspiration (mm) Results and discussion
Winter 637 842 926
Summer 650 990 1129
Annual 1286 1833 2055 Water balance analysis by land-use classes
Temperature (°C) In order to understand the spatial recharge distribution, it
Winter 13 21 24 is useful to first analyze surface runoff and evapotranspi-
Summer 14 24 28
Annual 13 22 26
ration which occur prior to recharge. Runoff simulated by
WetSpass is presented in Fig. 10. A high spatial variation

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1449

Fig. 4 Average annual potential evapotranspiration of Dire Dawa catchment

Fig. 5 Average annual temperature of Dire Dawa catchment

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1450

Fig. 6 Land-use of Dire Dawa catchment (WWDSE 2004 unpublished)

Fig. 7 Topographic map of Dire Dawa catchment (meters above sea level)

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1451

Fig. 8 Slope map of Dire Dawa catchment

Fig. 9 Soil texture of Dire Dawa catchment

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1452

Fig. 10 Average annual surface runoff from Dire Dawa catchment modeled by WetSpass

in runoff rates was observed, which can be attributed to a result potential evapotranspiration is high. The urban and
number of factors. Areas close to the southern catchment built-up areas are mainly impervious surfaces with lower
divide have higher amount of runoff. This is due to the evaporation and transpiration.
high slope and heavy soil texture (sandy clay loam). The Evapotranspiration has two components: evaporation
other area with high runoff is the north eastern periphery, and transpiration. Soil evaporation was highest from bare
an area with bare land, sand clay soil, and high slope. lands and lowest from urban and built-up areas (Fig. 13).
Despite sandy loam soil and plain topography, a higher Transpiration was highest from physiognomic vegetation
runoff was also simulated for the town of Dire Dawa. This and lowest from urban and built-up areas (Fig. 14). As
is due to the impervious surfaces in the urban area which shown in Fig. 15, in the urban and bare land areas,
prohibit infiltration and promote runoff. evaporation dominates while in the cultivated land and
The annual surface water balance of the Dire Dawa physiognomic vegetation areas transpiration dominates.
groundwater basin for major land-use units is presented in The WetSpass simulated recharge to the Dire Dawa
Fig. 11. The average simulated annual runoff in the study groundwater basin is presented in Fig. 16. The average
catchment is 126 mm which is 20% of annual precipitation. annual recharge in the Dire Dawa groundwater basin is
The highest runoff rate occurs from urban and built-up areas 28 mm. This is 5% of the average annual precipitation
because of impervious surfaces in this land-use class. The (626 mm) over the catchment. About 80% (22 mm) of the
next highest runoff was from cultivated lands. This is due to recharge occurs on the escarpment with the remaining
the high slope (9%) and the soil type (49% sandy clay loam) 20% (6 mm) occurring in the plain area. Considering the
both of which are major factors to initiate runoff. Plain areas, 920.12 km2 of the study area, this corresponds to 817 l/s
more sandy soil covers, and areas with vegetation cover have of recharge. For a sustainable groundwater use the
low runoff rates. abstraction from the basin should thus not exceed this
The simulated actual evapotranspiration is presented in rate. From the water balance of the catchment (Fig. 11), it
Fig. 12. The average annual actual evapotranspiration can be seen that high recharge occurs from bare land on
(ET) from the catchment was 468 mm which is 75% of the the escarpment. This land-use unit has lower runoff and
annual precipitation (Fig. 11). The highest actual ET was evapotranspiration compared to the cultivated land.
from physiognomic vegetation followed by cultivated Batelaan and De Smedt (2001) have also found the
land. The least ET was from urban and built-up areas. highest simulated recharges on bare soils in the interfluves
The physiognomic vegetations are mainly in the plain of Gote Nete Basin, Belgium. Another good recharge area
lowland areas where the temperature is high and as a is the high lands near Kulubi. This is due to the high

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1453
800
Precipitation Run off Evapotranspiration Recharge
700

Water balance components (mm)

600

500

400

300

200

100

0
Urban Bare Cultivated Physiognomic
Landuse
Fig. 11 WetSpass simulations for runoff, actual evapotranspiration, and recharge for different land-use classes in the study area

precipitation and low evaporation in these areas. The area. The sandy soil texture and the plain topography of
lowest recharge was found for physiognomic vegetation. the town of Dire Dawa favors recharge. For a similar
Unlike observations in other studies (e.g. De Smedt semiarid region in northern Ethiopia, Asfaw (2005) found
and Batelaan 2003), the urban and built-up areas in this runoff, evapotranspiration, and recharge rates of 16.7,
study have relatively high recharge. This is due to the fact 75.6, and 7.7% of precipitation respectively. Higher
that, unlike the areas in the former studies, the ratio of recharge and lower runoff might be because he used the
bare lands to impervious lands is high in the Dire Dawa same parameters, including rainfall intensity, as used for

Fig. 12 Average annual actual evapotranspiration in Dire Dawa catchment

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1454

Fig. 13 WetSpass simulated average annual soil evaporation from Dire Dawa catchment

Fig. 14 WetSpass simulated average annual transpiration from Dire Dawa catchment

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1455
600
Evaporation Transpiration
(3%) (Fig. 8). The valley plain is mainly sandy loam while
the escarpment area is dominated by loam and sandy clay
500
loam. The average slope of areas occupied by different
soil classes is: sandy clay loam (11%), loam (9%), loamy
Evapotranspiration (mm)

400 sand (9%), sandy clay (4%), sandy loam (3%), silty clay
(1%). The loam area in the escarpment is mainly bare land
300 while the sandy clay loam area close to the catchment
divide is cultivated land.
200 The annual water balance of Dire Dawa catchment
summarized by soil textural classes is presented in Fig. 17.
100
The sandy clay loam area receives the highest rainfall
amount (741 mm). The highest runoff is generated from
sandy clay loam area which is located in the high slope
0
Urban Bare Cultivated Physiognomic and high rainfall area. The runoff rate increases as the soil
Landuse gets heavier: loamy sand (36 mm), sandy loam (50 mm),
Fig. 15 Evaporation and transpiration components of actual loam (68 mm), sandy clay loam (237 mm), sandy clay
evapotranspiration components in different land-use units (280 mm), silty clay (253 mm). The evapotranspiration
seems not to show a clear pattern with soil texture as has
the temperate condition in WetSpass. However, in this also been observed by Zhang et al. (2004). It is rather
study, rainfall intensity, a major factor affecting runoff much affected by elevation, with soils in the plain area
rate, was determined from 30 years hourly rainfall data at (high temperature area) having high evapotranspiration.
Dire Dawa.

Earlier attempts to estimate recharge in the Dire

Dawa groundwater catchment
Water balance analysis by soil type Some reports are available about earlier efforts made to
The soil in the study area is grouped into six textural estimate groundwater recharge in the Dire Dawa area
classes: loamy sand (7%), sandy loam (19%), loam (41%), (BCEOM-WWDSE-CECE 2005 unpublished; Greitzer
sandy clay loam (28%), sandy clay (3%), and silty clay 1970; WWDSE 2004 unpublished). According to the

Fig. 16 WetSpass simulated yearly recharge in Dire Dawa catchment

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

1456
800
Precipitation
700 Runoff
Evapotranspiration

Water balance components (mm)

600 Recharge

500

400

300

200

100

0
Loamy sand Sandy loam Loam Sandy clay Sandy clay Silty clay
loam
Soil texture
Fig. 17 Annual precipitation and WetSpass simulated runoff, actual evapotranspiration, and recharge for soil texture classes in the Dire
Dawa catchment

report of the joint work by BCEOM-WWDSE-CECE approach, the groundwater recharge in DDAC was
(2005 unpublished), a runoff coefficient of 10% was estimated to be 910 l/s. In the steady-state flow approx-
assumed based on a study conducted in the nearby imation or Darcy method, groundwater recharge was
Alemaya catchment. Recharge estimated in this way was estimated by assuming about 5.5 km width of aquifer in
24% of precipitation in Kulubi area and 8% in the the NW–NE groundwater flow direction, its gradient in
Dengego area. It would be obvious that this recharge is this flow direction, and the average transmissivity of the
so high because of the low runoff estimate in the water aquifer. The estimated recharge was 1,018 l/s.
balance (i.e., low runoff coefficient of 10%). Recommended
runoff coefficients for agricultural/croplands are 40%, for
grass lands/herb lands 25%, and for urban area 70%
(Schwab et al. 1993). Using spring-area relationship for the Conclusions
260 km2 area between Melka Jebdu and Hurso, a spring
discharge of 363 l/s (i.e., 44 mm or 6% of the mean annual A water balance of a semiarid region groundwater system
rainfall of 721 mm) was estimated. in Ethiopia was determined using a GIS-based distributed
Sir Alexander Gibb and Seureca consulting engineers hydrologic model WetSpass. The goal was to estimate
(the town of Harar water supply, Hydrogeological study recharge to the economically and socially important Dire
interim working paper. Ministry of Water Resources, Dawa aquifer. Use was made of the available field data
Addis Ababa, Ethiopia, unpublished, 1996), who estimated and GIS technology to determine the seasonal and annual
a yield of 110 l/s for a 90 km2 area for Dire Jara and 100 l/s runoff, actual evapotranspiration, and recharge. The
for the 85 km2 Hurso area, which is adjacent to Dire Jara simulated recharge map reflected the already perceived
area, stated that recharge over 5% of mean annual rainfall is knowledge about the recharge and discharge areas. It was
unlikely. Greitzer (1970) based on the study of recharge in found that the annual rainfall is apportioned as: 75%
Harar area estimated 40 to 50 mm recharge on the evapotranspiration, 20% runoff, and 5% recharge. Con-
sedimentary rocks and 30 mm for the alluvial plains in Dire sidering the 920.12 km2 area of the catchment, this
Dawa. This recharge is about 6.5–5% of the mean rainfall. estimated recharge corresponds to 817 l/s. This is less
WWDSE (2004 unpublished) estimated recharge for than the currently assumed recharge of 1,000 l/s. The fact
Dire Dawa and Dire Jara areas by taking different that the latter is not the true recharge is already evidenced
infiltration coefficients for different rocks based on the by declining groundwater levels in boreholes pumped at
Abbay Basin Master Plan Study Hydrogeology Report by equivalent rate in the Sabian well field.
BCEOM (Abbay River basin integrated development The current knowledge based on the available infor-
master plan project, Vol. II, part 3: hydrogeology. Ministry mation tells us that the recharge to the Dire Dawa aquifer
of Water Resources, Addis Ababa, Ethiopia, unpublished, is less than what was already thought. The future
1997) and Darcy’s approach. The assumed infiltration groundwater development and management in the area
coefficients were: alluvial deposits (3%), basalts (2%), should take this into account. The result is also useful for
upper sandstone (10%), Hamanlei sandstone (10%), groundwater modeling and vulnerability studies in the
Adigrat sandstone (2%), and basement rocks (2%). In this area. However, it should be noted that the current recharge

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x

 1457
can change depending on the change in land-use and Chow VT, Maidment DR, Mays LW (1988) Applied hydrology.
climate. McGraw-Hill, New York
Cook PG, Walker GR, Buselli G, Potts I, Dodd AR (1992) The
application of electromagnetic techniques to groundwater
Acknowledgements The first author is an Alexander von Humboldt recharge investigations. J Hydrol 130:201–229
foundation research fellow at the Technische Universität Bergakademie De Smedt F, Batelaan O (2003) Investigation of the human impact
Freiberg (Germany). The foundation is, therefore, highly acknowl- on regional groundwater systems. In: Ecosystems and sustain-
edged. Dire Dawa Water Supply and Sewerage Authority, Ethiopian able development. Tiezzi E, Brebbia CA, Uso JL (eds)
Water Works Design and Supervision Enterprise, Harar Water Supply Advances in ecological sciences 19, WIT Press, Wessex, UK
Project, National Meteorological Services Agency, and Mr. Million De Vries JJ, Simmers I (2002) Groundwater recharge: an overview
Solomon are acknowledged for the data used in this study. Prof. O. of processes and challenges. Hydrogeol J 10(1):5–17
Batelaan for the provision of WetSpass program. Essery CI, Wilcock DN (1990) Checks on the measurement of
potential evapotranspiration using water balance and independent
measure of groundwater recharge. J Hydrol 120:51–64
Federer CA (1979) A soil-plant-atmosphere model for transpiration
References and availability of soil water. Water Resour Res 15(3):555–562
Freeze RA, Cherry JA (1979) Groundwater. Prentice-Hall, Englewood
Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotrans- Cliffs, NJ
piration: guidelines for computing crop water requirements. Greitzer Y (1970) Stratigraphy, hydrogeology and Jurassic Ammonites
FAO Irrigation and Drainage Paper No. 56, FAO, Rome of Harar and Dire Dawa, Ethiopia. PhD Thesis, Hebrew
Allison GB, Gee GW, Tyler SW (1994) Vadose-zone techniques for University, Jerusalem
estimating groundwater recharge in arid and semiarid regions. Pilgrim DH, Cordery I (1992) Flood runoff. In: Maidment DR
Soil Sci Soc Am J 58:6–14 (ed) Handbook of hydrology. McGraw-Hill, New York, pp
Allison GB, Hughes MW (1978) The use of environmental chloride 9.1–9.41
and tritium to estimate total recharge to an unconfined aquifer. Rubin J (1966) Theory of rainfall uptake by soils initially drier than
Aust J Soil Res 16:181–195 their field capacity and its applications. Water Resour Res 2
Asfaw MG (2005) Groundwater recharge and water balance (4):739–749
assessment in Geba basin, Tigray, Ethiopia. MSc Thesis Saxton KE, Rawls WJ (2006) Soil water characteristic estimates by
IUPWARE, Vrije Universiteit Brussel and Katholieke Universiteit texture and organic matter for hydrologic solutions. Soil Sci Soc
Leuven, Belgium Am J 70:1569–1578
Batelaan O, De Smedt F (2001) WetSpass: a flexible, GIS based, Schwab GO, Fangmeier DD, Elliot WJ, Freveret RK (1993) Soil
distributed recharge methodology for regional groundwater and water conservation engineering. Wiley, New York
modelling. In: Gehrels H, Peters J, Hoehn E, Jensen K, Smedema LK, Rycroft DW (1988) Land drainage. Batsford,
Leibundgut C, Griffioen J, Webb B, Zaadnoordijk W-J (eds) London
Impact of human activity on groundwater dynamics, Publ. no. Sophocleous MA (1991) Combining the soil water balance and
269, IAHS, Wallingford, UK, pp 11–17 water level fluctuation methods to estimate natural groundwater
Batelaan O, De Smedt F (2007) GIS-based recharge estimation by recharge: practical aspects. J Hydrol 124:229–241
coupling surface–subsurface water balances. J Hydrol 337:337–355 Thijs G (2002) Calibration and sensitivity analysis of groundwater
Batelaan O, Woldeamlak ST (2003) ArcView interface for Wet- discharge modeling of the Upper-Biebrza catchment, Poland (in
Spass, user manual, version 1-1-2003. Department of Hydrol- Dutch). MSc Thesis, Vrije Universiteit Brussel, Belgium
ogy and Hydraulic Engineering, Vrije Universiteit Brussel, Tilahun K (2006) The characterization of rainfall in the arid and
Belgium, 50 pp semi-arid regions of Ethiopia. Water SA 32(3):429–436
Bazuhair AS, Wood WW (1996) Chloride mass-balance method for Zhang L, Hickel K, Dawes WR, Chiew FHS, Western AW, Briggs
estimating groundwater recharge in arid areas: examples from PR (2004) A rational function approach for estimating mean
western Saudi Arabia. J Hydrol 186:153–159 annual evapotranspiration. Water Resour Res 40, W02502

Hydrogeology Journal (2009) 17: 1443–1457 DOI 10.1007/s10040-009-0455-x