Application of a spatially distributed water balance model for assessing surface water and groundwater resources in the Geba basin, Tigray, Ethiopia

Journal of Hydrology 499 (2013) 110–123

Contents lists available at SciVerse ScienceDirect

Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol

Application of a spatially distributed water balance model for assessing surface water and groundwater resources in the Geba basin, Tigray, Ethiopia Tesfamichael Gebreyohannes a,b, Florimond De Smedt a,⇑, Kristine Walraevens c, Solomon Gebresilassie b, Abdelwasie Hussien b, Miruts Hagos b, Kasa Amare b, Jozef Deckers d, Kindeya Gebrehiwot b a

Department of Hydrology and Hydraulic Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium Department of Earth Sciences, Mekelle University, PO Box 231, Mekelle, Ethiopia c Laboratory for Applied Geology and Hydrogeology, Ghent University, Krijgslaan 281-S8, 9000 Gent, Belgium d Department of Earth and Environmental Sciences, Division for Land and Water Management, KULeuven, Celestijnenlaan 200E, B-3001 Leuven, Belgium b

a r t i c l e

i n f o

Article history: Received 7 December 2012 Received in revised form 6 June 2013 Accepted 8 June 2013 Available online 4 July 2013 This manuscript was handled by Konstantine P. Georgakakos, Editor-in-Chief, with the assistance of Matthew Rodell, Associate Editor Keywords: Groundwater recharge Water balance WetSpass Geba River Ethiopia

s u m m a r y The Geba basin is one of the most water-stressed areas of Ethiopia, with only a short rainy period from mid-June to mid-September. Because rainfall in this region has been consistently erratic in the last decades, both in time and space, rain-fed agriculture has become problematic. Hence, in order to supplement rain-fed agriculture by irrigation, a detailed understanding of local and regional surface water and groundwater resources is important. The main objective of this study is to assess the available water resources in the Geba basin using a spatially distributed water balance model (WetSpass). Relevant input data for the model is prepared in the form of digital maps using remote sensing images, GIS tools, FAO and NASA databases, field reconnaissance and processing of meteorological and hydrological observations. The model produces digital maps of long-term average, seasonal and annual surface runoff, evapotranspiration and groundwater recharge. Results of the model show that 76% of the precipitation in the basin is lost through evapotranspiration, 18% becomes surface runoff and only 6% recharges the groundwater system. Model predictions are verified against river flow observations and are shown to be reliable. Additional maps are derived of accumulated surface runoff, safe yield for groundwater abstraction and water deficit for crop growth. Comparison of existing reservoirs with the accumulated runoff map shows that many reservoirs have failed because their design capacity is much higher than the actual inflow. Comparison of the safe yield map with the crop water deficit map shows that in most areas groundwater can be safely abstracted to supplement the water deficit for crop growth during the wet summer season. However, in the dry winter season the crop water deficit is too high to be supplemented by groundwater abstraction in a sustainable way. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Ethiopia is commonly referred to as ‘‘the water tower of Africa’’, because of the huge amount of surface runoff from the Ethiopian highlands that make up over 86% of the flow of the Nile River (Yacob and Imeru, 2005). However, the country is constantly affected by shortage of water for rain-fed agriculture, mainly because of lack of proper water resources utilization and management practices. Of the total land area of Ethiopia (about 113 million ha) only 14.8% is under cultivation (EPA, 1998). About 3.8 million ha of the cultivable land area is potentially irrigable but, so far, only about 289,000 ha has been irrigated (Frenken, 2005). According

⇑ Corresponding author. Tel.: +32 34557571; fax: +32 26293022. E-mail address: fl[email protected] (F. De Smedt). 0022-1694/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhydrol.2013.06.026

to the Ethiopian Ministry of Water Resources report (2001), the total renewable freshwater (mean annual flow) of the country is estimated at 122 billion m3, and 54.4 billion m3 of surface water and 2.6 billion m3 of groundwater could be developed for utilization. Currently less than 5% of the surface water potential is used for consumptive purposes while groundwater is virtually untouched. The Geba basin is one of most food insecure areas of the Tigray Regional State, in Northern Ethiopia (Eyasu, 2005). The climate is mainly semi-arid, such that rainfall is limited and erratic and usually insufficient for optimal crop production. The regional government and several non-governmental agencies have been investing in water harvesting activities to supplement subsistence agriculture with small to medium scale irrigation through the construction of micro-dams and hand-dug wells. The principal objectives are to change the agrarian system to widespread small-scale irrigated agriculture and to gradually attain self-sufficiency in food

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

production. However, the success of these initiatives has been very limited due to various reasons, among which the absence of adequate surface and groundwater knowledge is considered to be the main factor. Although existing regional studies (e.g. Yilma and Zanke, 2004; Yilma and Camberlin, 2006) and meteorological records in the basin show that the total rainfall remained more or less constant since the 1950s, the spatial and temporal variation resulted in recurrent draught and continuous impoverishment of the farmers. Hence, it is becoming clear that food self-sufficiency can only be attained by adopting a strategy which encourages conjunctive groundwater and surface water management. In this respect, a better understanding of the availability and distribution of groundwater resources for supplementary irrigation could help to ease the water stress and thereby improving crop production, but groundwater studies in the region have been limited. Chernet and Eshete (1978) undertook a regional scale (1:250,000) hydrogeological mapping of the region around the regional capital Mekelle. DevCon (1992) investigated the water resources potential of the Mekelle area as part of the Five Towns Water Supply and Sanitation project, by the Ministry of Water Resources. NEDECO (1997) conducted drillings up to 300 m deep in the Mekelle area to identify exploitable groundwater resources. Hussein (2000) investigated the hydrogeology of the Aynalem valley south of Mekelle. CoSAERT (2001) presented a hydrogeological study of the Suluh valley in the north of the Geba basin. WWDSE (2006) made a comprehensive investigation of the Aynalem well field, which supplies drinking water for Mekelle. Walraevens et al. (2009) and Vandecasteele et al. (2011) studied a small perched aquifer near Hagere Selam in the west part of the basin during the rainy season of 2006, and estimated the soil water budget for the period 1995–2006, showing that there is a water deficit for on average 10 months per year due to the strong seasonal variation in rainfall. Kibrewossen et al. (2011) made an assessment of the groundwater resources in the Geba basin using a simplified 2D groundwater model, and concluded that possibly 30,000 m3/d of groundwater could be abstracted in the Geba basin in a sustainable way. Sustainable use of groundwater is largely determined by groundwater recharge, i.e. the downward flow of water reaching the water table, which is added to the groundwater storage (Freeze and Cherry, 1979; Lerner et al., 1990; Healy, 2010). Hence, knowledge of rates and locations of recharge is important for determining sustainable yields of groundwater systems (Sophocleous, 2000; Sophocleous and Devlin, 2004; Devlin and Sophocleous, 2005). Many methods have been proposed to determine the groundwater recharge, as described for instance by Lerner et al. (1990), Simmers (1997), Kinzelbach et al. (2002) and Healy (2010), but the largest class of techniques are water-budget methods and models (Healy, 2010). The main objective of this study is to assess the regional surface and groundwater potential in the Geba basin to support crop growth by irrigation. Basically it focuses on understanding the hydrological processes in the basin that determine the water resources, i.e. the partitioning of the precipitation in runoff, evapotranspiration and groundwater recharge in relation to the spatial distribution of hydro-meteorological variables and soil, land-use and topographic conditions. To achieve these objectives various tools and techniques are applied, e.g. remote sensing and GIS for data gathering, and application of the WetSpass model (Batelaan and De Smedt, 2001) for assessment of runoff, evapotranspiration and groundwater recharge. The novelty of the work is to demonstrate how spatial distributed water resources can be assessed in a large, difficult to access, water-stressed region using existing modeling tools and popular global data by well-established methods.

111

2. Methodology 2.1. Description of the study area The Geba River is a major tributary of the Tekeze River, which joins the river Nile at Atbara in Sudan. The Geba basin is situated in the Tigray regional state in north Ethiopia, between latitudes 13°160 and 14°160 north and longitudes 38°380 and 39°490 east. Fig. 1 shows the location map of the study area. The basin covers an area of about 5260 km2 and is surrounded by the Ethiopian Rift escarpment in the east, by the Tekeze River basin in the south, the Mugulat Mountains in the north, and the Werii River basin in the west. The topographic elevation in the basin ranges from about 950 m at the basin outlet to 3300 m in the northern part (Fig. 1), and is characterized by steep volcanic mountains and sharp cliffs with plateaus of sandstones in the north, deep gorges of limestone in the centre and ragged metamorphic terrain in the southwest. The fault-controlled Mekelle, Wukro and Sinkata areas, and the Atsbi horst form the major plains of the Geba basin. The geology of the Geba basin is highly diversified and complicated, as described by Tesfagiorgis et al. (2010); a geological map and a cross-section taken from this publication is presented in Fig. 2. The geology of the basin comprises Precambrian, metamorphosed volcanics/volcanoclastics, intrusives and sediments in the north and southwest, Paleozoic and Mesozoic sediments in the centre, some patches of Tertiary volcanic and shallow intrusives (Dolerites) in the centre and north, and localized quaternary sediments along the valleys of the major rivers. More than half of the basin is covered by Mesozoic sediments, and about a quarter by Precambrian rocks, while the remaining consists of Tertiary volcanics and Dolerites, or alluvial sediments. The climate in the Geba basin is semi-arid with a mean annual precipitation ranging from 400 mm in the eastern part of the basin up to 950 mm in the northern and western parts. The temperature varies from a minimum average of 6.5 °C in the northern highlands and the northeast plateau up to a maximum average of 32 °C in the western lowlands near the basin outlet. Precipitation mainly takes place during the summer, which lasts from mid June till mid September. The long-term annual average reference (potential) evaporation is about 1500 mm, such that the aridity index, i.e. the ratio between precipitation and potential evapotranspiration, is between 0.27 and 0.63, which according to the UNEP classification (Middleton and Thomas, 1992) corresponds to a semi-arid to dry sub-humid climate. The land cover is accordingly, with bare land or shrubs in the semi-arid eastern lowlands, and mostly cultivated land and occasionally forest in the dry sub-humid higher parts in the north and west of the basin. Because of the semi-arid climate, agricultural practice is hampered severely. Generally, there is only one harvest possible by dryland farming in the wet summer season from June to September. Farmers predominantly grow cereals, as wheat, sorghum and teff, which take about three to four months to mature. Cropping starts as soon as the rains arrive in May or June. The growing stage is often insecure due to the erratic nature of the rainfall and the high evaporative demand, which is usually much larger than the rainfall. Hence, harvests are often insufficient and certainly non-optimal. This could be improved by irrigation, as advocated by the authorities and relief agencies, who promoted the construction of small scale reservoirs and hand-dug wells, which however often do not meet expectations. The drainage system of the Geba basin can generally be described as dendritic with some significant influence of major structures like folds and faults. The main tributaries of the Geba River are Suluh, Genfel, Agulae, Illala and Metere (Fig. 1). The east–west flow direction of the Illala in the central part of the basin and north–south orientation of the Genfel in the north are thought to

112

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Fig. 1. Location map of the Geba basin with elevation and locations of the meteorological and river gauging stations.

be dictated by the Mekelle and Chelekot faults and by the Wukro syncline, respectively. Except for the Geba River and its major tributaries, most streams in the Geba basin are ephemeral, flowing only during the rainy season. The hydrogeological conditions of the area are largely unknown, except for the information provided by Tesfamichael (2009) and Kibrewossen et al. (2011). From these studies it can be concluded that there are no major aquifers, but some geological formations can be considered as aquitards and could be used for limited groundwater abstraction. Generally groundwater levels are deep, but there are areas with shallow groundwater, where groundwater can be abstracted for irrigation purposes. Kibrewossen et al. (2011)estimated that possibly 30,000 m3/d of groundwater can be abstracted in the Geba basin in a sustainable way. At present groundwater abstraction is very limited, due to unawareness and limited resources for installing pumping wells.

2.2. WetSpass model WetSpass (an acronym for Water and Energy Transfer in Soil, Plants and Atmosphere under quasi Steady State) is a numerical model to simulate long-term average spatial distributions of hydrological parameters and processes on basin scale (Batelaan and De

Smedt, 2001, 2007). The model makes use of grid GIS technology and digital data to partition the precipitation into surface runoff, evapotranspiration and groundwater recharge. The methodology can be schematised as follows (for details refer to the manual available at http://www.vub.ac.be/WetSpa/introduction_wetspass. htm). The model is based on the long-term average seasonal water balance equation:

P ¼ S þ ET þ R;

ð1Þ

where P is precipitation [L], S is surface runoff [L], ET is evapotranspiration [L], and R is groundwater recharge [L]. Surface runoff is determined as:

S ¼ f1 ðLV; ST; SAÞ  Pn ;

ð2Þ

where f1(.) is a runoff factor depending upon land-use and vegetation characteristics (LV), soil texture (ST) and slope angle (SA), and Pn is the net precipitation [L], i.e. precipitation minus interception, the latter being also a function of LV. Evapotranspiration ET is determined from soil evaporation and transpiration by the vegetation, as:

ET ¼ f2 ðLV; STÞ  EP ;

ð3Þ

where f2(.) is an evapotranspiration factor depending upon land-use and vegetation characteristics (LV) and soil texture (ST), and Ep is

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

113

Fig. 2. Geological map of the Geba basin and geological cross-section along the central part of the basin (roughly E–W direction): tpb is trap basalt volcanic, ups is upper sandstone, dlt is dolerite, shl is shale; sml is shal-marl-limestone, mls is marl-limestone, lsm is limestone-marl, and ast is Adigrat Sandstone.

the potential evaporation of open water [L]. The groundwater recharge is determined as the closing term in Eq. (1). In the equations above, all variables and parameters are digital maps and the calculations and derivations are obtained by means of GIS tools. To operate the model the user has to provide spatial digital data of terrain properties and of seasonal climatic variables. Because, the model was originally developed for conditions in temperate regions in general and Belgium in particular, the user can interfere in the calculations and predictions by modifying default parameters and procedures. 2.3. Data gathering The WetSpass model calculates water balances per seasons, which are by default the 6-months winter and summer season of a humid climate, to enable transfer of accumulated soil moisture storage from the wet winter season to the dry summer season. For application in the Geba basin, this was changed to an 8 months dry winter season from October to May and a 4 months rainy sum-

mer season from June to September. The WetSpass model requires two types of input data, i.e. GIS grid maps and parameter tables (Batelaan and De Smedt, 2001). The grid maps consist of slope angle, land-use, soil texture, groundwater depth, and seasonal meteorological maps of precipitation, potential evapotranspiration, temperature and wind speed. The WetSpass model was applied using GIS grid maps with a cell size of 90  90 m2. All input data were collected between 2004 and 2008 during the Mekelle University Institutional University Cooperation (MU– IUC) Hydrogeology Project (http://www.mu.edu.et/index.php/muiuc-project/mu-iuc-hydrogelogy-project). A 90 m resolution digital elevation model (DEM) of the basin (Fig. 1) was obtained from the Shuttle Radar Topography Mission (SRTM) of the National Aeronautics and Space Administration (NASA). A qualitative accuracy assessment of this dataset was made against a digitised topographic map of the Wukro sheet (Index number 1339 B1) on scale 1:50,000 of the Ethiopian Mapping Agency (EMA, 1996) and no unacceptable discrepancies were noted. A slope angle map (Fig. 3a) of the Geba basin was produced from this DEM using

114

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Fig. 3. GIS maps used as input for the WetSpass model: (a) slope angle, (b) land-use, and (c) soil type in the Geba basin.

ArcView GIS tools, and shows that slope angles range from 0° to 55° with a mean of 8°.

A land-use map of the basin was derived from cloud free Landsat ETM + satellite images of January 27 (path 68, rows 50 and 51) and February 5 (path 69, rows 50 and 51), 2000 using the standard IDRISI GIS supervised image classification procedure. The image classification was based on identifying and delineating at least three training sites for each land-use type, and by interpretation of Google Earth images. Six land-use classes were identified, i.e. cultivated land, bare land, shrubs, forests, settlements and water bodies. The accuracy of the resulting land-use classification was assessed using 668 ground truth data points obtained in the field during the period of 2006–2008. The results are presented in Table 1. A correspondence of more than 80% was obtained for the major land-use classes. The land-use map (Fig. 3b) of the basin shows that nearly 50% of the basin is cultivated land, while bare land and shrubs cover about 20% each, and forest, urban areas and water bodies constitute the remainder. A soil map of the Geba basin was derived from the Soil and Terrain Database for Northeast Africa developed by the Food and Agricultural Organization (FAO, 1998). Missing data were filled in from the Data Exchange Platform for the Horn of Africa (DEPHA) (http:// www.un-spider.org/guide-en/3203/data-exchange-platform-hornafrica-depha). Soil type classes were translated into USGS soil texture classes, using the percentages of coarse, medium and fine particle size fractions in the topsoil. The soil map of the Geba basin (Fig. 3c) shows the soil texture, i.e. clay loam (39%), sandy clay (33%), sandy clay loam (18%), sandy loam (4%), silty clay loam (3%) and loam (3%). Maps of groundwater depth are needed in the WetSpass model for delineation of wetlands, where water balance calculations have to include seepage fluxes. In previous studies with the WetSpass model in humid areas, wetlands have been delineated as areas with a shallow groundwater table for which the user could specify a default value, as for instance less than 2 m deep below the soil surface. However, in case of the Geba basin there are no wetlands due to the semi-arid climate conditions. Hence, there is no need for a groundwater level map in this study. Seasonal meteorological parameters were prepared from the available meteorological data. There are 17 meteorological stations within or close nearby the Geba basin as indicated in Fig. 1. Many of these stations are either recent with recordings starting in 2003, or have a lot of missing data because of instability in the region during the 1970s, 1980s and 1990s. The longest data record in the basin is for the station at Quiha airport near Mekelle, which started in 1959 but misses data from 1989 to 1991. Moreover, only four stations have the facility to record all necessary meteorological parameters required for the WetSpass model. Daily precipitation recordings are available in all stations, but temperature has only been recorded in 11 stations, and wind speed, sunshine hours and relative humidity only in 4 stations. The meteorological stations are not evenly distributed throughout the Geba basin. Most are located along the main road from Mekelle to Adigrat, while there are no stations in the western arid lowlands and in the middle section of the Geba gorge. Mean monthly values of precipitation were calculated from the available data in each station and used to derive mean total precipitation for the wet season (June–September) and for the dry season (October–May). As topography (elevation) is considered to have a significant influence on the climate, an attempt was made to correlate rainfall with altitude, but no such significant correlation could be detected. Hence, digital maps of seasonal precipitation were derived by spatial interpolation of the values observed in the 17 stations, using the universal kriging interpolation module of ArcView GIS. The resulting maps are shown in Fig. 4a and b. The total precipitation ranges between 325 and 699 mm for the summer season and between 37 and 199 mm for the winter season.

115

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Table 1 Results of the ground truth verification of the land-use map obtained by supervised image classification of Landsat ETM + satellite images of January 27 and February 5, 2000. Ground truth data

Land-use class from the map

Land-use

Number

Bare land

Shrubs

Agriculture

Forest

Urban areas

Correspondence (%)

Bare land Shrubs Agriculture Forest Urban areas

53 158 447 2 8

45 0 24 0 0

3 132 26 0 0

5 26 393 2 6

0 0 1 0 0

0 0 4 0 2

85 84 88 0 25

Total

668

Mean correspondence (%)

As temperatures in the Geba basin are very significantly negatively correlated with altitude, seasonal temperature maps were derived in ArcView GIS, using linear regression equations obtained from fitting the temperature recordings in the stations as a function of elevation. The maps are shown in Fig. 4c and d, respectively. The average temperature ranges between 12 °C and 28 °C in the summer season and between 11 °C and 27 °C in the winter season. Because observations needed to derive the potential evapotranspiration (PET) in the Geba basin are very limited, PET was estimated using an approach developed by Yilma (2002), who derived regression equations for monthly PET as a function of altitude for the Tekeze and Mereb basins. The reliability of this approach was first tested by comparison with monthly reference evapotranspiration calculated with the Penman–Monteith based FAO PET Calculator software v. 3.1 (Raes, 2009) for the four meteorological stations (Adigrat, Mekelle, Quiha and Senkata) where wind speed, sunshine hours and relative humidity have been recorded. Table 2 shows that the PET-values derived using the two approaches are very similar. Hence, digital maps of total seasonal PET were prepared based on the regression equations of Yilma (2002). The resulting maps are shown in Fig. 4e and f. The total PET ranges between 862 and 1318 mm for the summer season and between 455 and 594 mm for the winter season. Wind speed has only been recorded at Adigrat, Mekelle, Quiha and Senkata, which is not sufficient to make any meaningful interpolation map for the entire Geba basin, as also no correlation could be detected with elevation. Hence, average values of the four stations were used instead, and the wind speed maps were set uniform to 1.87 m/s for the summer season and 2.00 m/s for the winter season. Land-use, soil type and runoff parameters have to be specified in four look-up tables required for running the WetSpass model. The two land-use attribute tables include parameters related to land-use type and soil type. The former contains parameters such as rooting depth, leaf area index and vegetation height. The latter contains soil parameters for each (USGS) textural soil class such as field capacity, wilting point and permeability. The runoff attribute table contains runoff coefficients for all combinations of land-use, slope angle and soil type. By default these tables contain values that suit conditions in humid regions, especially for vegetation. Hence, in order for the parameter tables to suit the conditions in the Geba basin, some modifications had to be introduced. Modifications were made by expert opinion for the summer and winter land-use parameters related to vegetation, such as leaf area index, crop height and interception capacity. Appropriate modifications were also made for the vegetative, bare-land, impervious and open water area proportions of each land-use class. Major adjustments included: (1) substantial decrease of leaf area index values as vegetation cover is lower in semi-arid regions and plants have less and transpiration reducing foliage; (2) no deciduous vegetation, although vegetation reduces in the dry season; (3) 20% imperviousness of all land covers because of stoniness and hydrophobic soils; (4) increased loss of rainfall by plant interception, depression storage and evaporation due to high evaporative demand, stoniness,

86

hydrophobic soils and rough terrains. For more details reference is made to the WetSpass manual and Tesfamichael (2009).

3. Results and discussion 3.1. Model results The WetSpass model results comprise several annual and seasonal hydrologic outputs. The most important ones are the digital maps of seasonal surface runoff, actual evapotranspiration and groundwater recharge. The seasonal surface runoff maps are shown in Fig. 5a and b. In total, annual surface runoff in the Geba basin ranges from 0 to 268 mm/y with a mean value of 108 mm/y. This amounts to 18% of the annual precipitation in the Geba basin. About 87% of the surface runoff occurs during the wet summer season while the remaining 13% occurs during the dry winter season. Table 3 gives the mean annual surface runoff for different combinations of land-use and soil classes. The largest surface runoff occurs on sandy clay soils with bare land, agriculture or urban land-use, while the lowest values are for sandy loam and loam soils with forest or shrub. The traceability of the soil type boundaries in Fig. 5a and b and the higher standard deviation values of the runoff for different soil types (Table 3) indicate that surface runoff is more influenced by soil type than by land-use. The influence of the precipitation is also noticeable by the fact that areas around EdagaHamus and Hagere-Selam have higher runoff than those in the east around Araguren and in the west near the Geba outlet. Evapotranpiration is calculated in the WetSpass model as the sum of evaporation of the precipitation intercepted by the vegetation, transpiration by the vegetation, and evaporation from bare soil and open water bodies (Batelaan and De Smedt, 2001). The seasonal evapotranspiration maps are shown in Fig. 5c and d. Annual evapotranspiration in the basin ranges from 276 to 1639 mm/y; the latter is obviously potential evapotranspiration from open water. Evapotranspiration is the largest component in the water balance of the Geba basin, constituting 76% of the precipitation or 462 mm/y on average. About 87% of the evapotranspiration takes place during the wet summer season, while the remaining 13% takes place during the dry winter season, which is obviously due to the unequal temporal distribution of the precipitation but also partly to the fact that the vegetation cover is less in the winter season. Larger evapotranspiration values are observed in the north and southwest parts of the basin, as these receive more precipitation. The fact that in Fig. 5c and d the soil texture boundary is weakly visible in the central part of the basin indicates that evapotranspiration in this part of the basin is also strongly influenced by soil type. Table 4 gives the mean annual evapotranspiration values for different combinations of land-use and soil classes. From this table it appears that silty clay loam, sandy loam and loam soils have the highest values of evapotranspiration, while sandy clay soils have the lowest values. Mean annual evapotranspiration for different land uses indicate that forest and shrubs have the highest values, while agriculture, urban areas and bare land have the lowest values. While overall

116

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Fig. 4. GIS maps used as input for the WetSpass model of major meteorological parameters in the Geba basin: (a) total summer precipitation (mm), (b) total winter precipitation (mm), (c) average summer temperature (°C), (d) average winter temperature (°C), (e) total summer potential evapotranspiration (mm), and (f) total winter potential evapotranspiration (mm).

evapotranspiration is highly influenced by precipitation and to some extent by soil texture, the standard deviation values in Table 4 show that evapotranspiration in the Geba basin is more variable within land-use classes than within soil textural classes.

Groundwater recharge in the Geba basin, and elsewhere, is promoted by low evapotranspiration and low surface runoff, e.g. typically for a flat topography and permeable soils. The seasonal groundwater recharge maps are shown in Fig. 5e and f. Annual

117

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Table 2 Comparison of monthly potential evapotranspiration at 4 meteorological stations in the Geba basin, calculated using the FAO-Penman-Montheit PET Calculator v. 3.1 (Raes, 2009) and with the method of Yilma Seleshi (2002) based on regression equations with altitude; the locations of the stations are indicated in Fig. 1. Stations

Adigrat

Month

PET1 (mm)

PET2 (mm)

Mekelle PET1 (mm)

PET2 (mm)

Quiha PET1 (mm)

PET2 (mm)

PET1 (mm)

PET2 (mm)

January February March April May June July August September October November December

4.7 5.0 4.9 4.7 4.4 4.3 3.3 3.3 3.9 4.0 3.7 3.4

3.6 4.0 4.6 5.1 5.1 4.9 3.7 3.5 3.9 4.4 3.6 3.2

4.4 5.0 5.3 5.4 5.5 4.5 3.5 3.2 4.0 4.6 4.3 3.7

3.9 4.4 5.0 5.5 5.4 5.2 3.9 3.6 4.2 4.7 4.0 3.6

5.1 5.5 5.5 5.8 5.8 4.9 3.5 3.1 4.1 4.9 4.5 3.9

3.7 4.2 4.8 5.3 5.3 5.0 4.0 3.5 4.0 4.5 3.8 3.4

4.7 5.6 5.1 4.9 4.5 4.5 3.1 2.9 4.4 4.1 4.0 3.7

3.6 4.0 4.6 5.1 5.1 4.9 3.7 3.5 3.9 4.4 3.6 3.2

Mean error

0.4

0.3

groundwater recharge in the Geba basin ranges from 0 to 191 mm/ y, with a mean value of 37 mm/y. This makes up only 6% of the total annual precipitation in the Geba basin. About 81% of the recharge occurs during the four months rainy season in the summer while the remaining 19% occurs in the eight months dry winter season. The northern part of the basin has generally higher groundwater recharge, likely due to a combination of favorable conditions such as high precipitation, permeable soils, gentle topography and less vegetation. Table 5 gives the mean annual groundwater recharge values for different combinations of landuse and soil classes. The largest groundwater recharge is observed for bare land and agricultural land on sandy loam and loam soils. This is basically because of the high permeability of these soils, but could partly also be due to lower evaporation rates on bare land and less runoff on the relatively gentler slopes of agricultural lands. On the other hand, forest and shrubs on any type of soil are found to yield less groundwater recharge. This is evidently because of the high transpiration rate, especially for forest. The overall summary of the water balance of the Geba basin is given in Table 6. Only a small fraction of the annual precipitation remains to recharge the groundwater reservoirs, the rest leaves the basin mainly through evapotranspiration and to a lesser extend by surface runoff. The small error in the water balance comes from the assumption that water bodies can evaporate unlimitedly at PET rate. Hence, this error should be subtracted from the runoff as open water bodies are supplied by runoff from surrounding areas, and does not leave the basin as river flow but is evaporated instead. 3.2. Model verification Results of the WetSpass model are verified against river flow observations made at six gauging stations in the Geba basin (Fig. 1). The river flow data have been collected by the Ministry of Water Resources, Ethiopia, and are available as monthly values covering different periods from 1968 to 2003, often with several months to years of gaps (Table 7). Unfortunately, most of the gauging stations are situated in the upstream parts of the river system, while there is only one station in centre and none in the downstream western lowlands. Mean monthly river flows were calculated and accumulated to mean annual flows at the six gauging stations, as shown in Table 7. As all river gauging stations are located on streams that are perennial, the river flow consists of direct (surface) runoff and base flow due to groundwater drainage. In order to estimate the annual river discharge at the river gauging stations from the WetSpass results, surface runoff and groundwater recharge are accumulated based on topography. Because surface runoff flows down gradient, it is possible to calculate the accumulated surface runoff in each point in the basin, using standard GIS

0.6

Senkata

0.6

tools. The accumulated surface runoff represents the direct river flow. For the base flow, it follows from the groundwater balance that long-term average drainage of groundwater should equal the groundwater recharge minus possible groundwater abstractions such as pumping wells. The latter can be ignored in the Geba basin, because most wells are only domestic and the pumping rates are very small compared to the total recharge. Relationships between source areas where groundwater recharge is occurring and discharge areas where drainage or seepage takes place depend upon the groundwater flow system(s), which are largely unknown in the Geba basin. However, groundwater flow systems are often strongly conditioned by topography (e.g. Freeze and Cherry, 1979; Haitjema and Mitchell-Bruker, 2005; Tóth, 2009). In particular, topography-controlled groundwater flow will likely occur in low-permeable sediments (Haitjema and Mitchell-Bruker, 2005), hence, not in aquifers but in aquitards as is the case in the Geba basin (Tesfamichael, 2009; Kibrewossen et al., 2011). Hence, as an approximation it is assumed that the groundwater flow also follows topography, and river base flow can be estimated by accumulating the groundwater recharge along topography similar as was done for surface runoff. The accumulated surface runoff and groundwater recharge at the locations of the gauging stations are compared with the observed mean annual flows in Table 7. The model results are close to the observations, but higher simulated values are noted for the Suluh River and the Genfel River and lower values for the Agulae River. The higher simulated values in the upper sections of the Geba basin (Suluh and Genfel) can possibly be due to a combination of surface water abstraction and the assumption that all groundwater recharge re-appears as base flow in the rivers at the gauging stations. As a significantly large portion of the simulated river flow at these locations results from groundwater drainage, it is possible that part of the groundwater flow in the upstream area of the basins moves further down to lower areas and joins the river system further downstream. At the Agulae gauging station, unlike all other stations, the simulated flow is smaller than the observed value. This may be due to the fact that the area upstream of the Agulae gauging station receives the lowest annual precipitation (about 200 mm less than the basin average), which has possibly been estimated as too low. Further downstream at the gauging station on the Geba River in the centre of the basin, the discrepancy between measured and simulated values becomes smaller, probably because the assumption about the contribution from groundwater drainage becomes more plausible with increase in drainage area. It is also important to note that most gauging stations may not be able to catch low flows accurately during the dry season as the gauging instruments are fixed to one side of the riverbanks.

118

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

Fig. 5. GIS maps of water balance components simulated with the WetSpass model: (a) total summer surface runoff (mm), (b) total winter surface runoff (mm), (c) total summer evapotranspiration (mm), (d) total winter evapotranspiration (mm), (e) total summer groundwater recharge (mm), and (f) total winter groundwater recharge (mm).

Another verification of the model results is achieved by river flow measurements performed in January and February 2008 at several places along the river system. Because January and Febru-

ary are the driest months of the year, the observed flows mainly consist of base flow and can be compared with the accumulated groundwater recharge predicted with the model for the dry winter

119

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123 Table 3 Mean annual surface runoff predicted with the WetSpass model for different combinations of land-use and soil texture. Sandy loam

Loam

Sandy clay loam

Silty clay loam

Clay loam

Sandy clay

Mean

Std. dev.

Agriculture Shrubs Bare land Forest Urban areas

13 8 16 4 –

21 12 23 6 –

89 58 104 34 72

125 90 146 53 –

112 75 125 45 97

144 104 168 61 145

84 58 97 34 105

55 40 64 24 37

Mean Std. dev.

10 5

16 8

60 38

104 41

76 47

104 63

62 34

44

Table 4 Mean annual evapotranspiration predicted with the WetSpass model for different combinations of land-use and soil texture. Sandy loam

Loam

Sandy clay loam

Silty clay loam

Clay loam

Sandy clay

Mean

Std. dev.

Agriculture Shrubs Bare land Forest Urban areas

461 612 474 597 –

464 586 464 614 –

443 551 465 548 407

508 640 506 674 –

444 523 431 534 422

411 503 408 512 415

455 569 458 580 415

32 53 34 60 8

Mean Std. dev.

536 80

532 79

483 64

582 88

471 53

450 53

509

45

Table 5 Mean annual groundwater recharge predicted with the WetSpass model for different combinations of land-use and soil texture.

Agriculture Shrubs Bare land Forest Urban areas

Sandy loam

Loam

Sandy clay loam

Silty clay loam

Clay loam

Sandy clay

Mean

Std. dev.

147 5 135 5 –

126 5 94 5 –

65 5 32 5 8

93 5 84 5 9

66 5 30 5 –

19 5 11 5 9

86 5 64 5 9

46 0 48 0 1

73 79

58 62

19 25

39 45

21 28

8 7

36

25

Mean Std. dev.

Table 6 Annual water balance of the Geba basin predicted with the WetSpass model. Water balance component

Annual values (mm/y) Min.

Max.

Mean

Std. dev.

Precipitation (P) Evapotranspiration (ET) Surface runoff (S) Groundwater recharge (R)

406 276 0 0

874 1639 266 191

605 462 108 37

96 85 43 40

Difference

P  ET  S  R = 2

Table 7 Comparison of predicted and observed mean annual river flow at six gauging stations in the Geba basin; the locations of the stations are indicated in Fig. 1. Gauging station

Suluh Genfel Agulae Illala Metere Geba

Recording period

1976–1995 1992–2000 1995–2002 1981–2002 1986–2001 1975–2002

Observed river flow (106 m3/y)

Simulation results (106 m3/y) Surface runoff

Base flow

Tota flow

42 84 59 21 6 359

44 58 39 21 10 254

26 40 9 2 1 115

70 98 48 23 11 369

Difference (106 m3/y)

28 14 11 2 5 10

season. The base flow measurements were made with a current meter at 15 different cross-sections from the most upstream section of the Suluh River near Hawzen to the Geba outlet (Fig. 1). Results are shown in Table 8. The correlation coefficient of the

Table 8 Comparison of predicted and observed river base flow at 15 cross-sections along the Geba River system; the measurements were performed in January–February 2008 at the locations indicated in Fig. 1. River (location)

Agoro (Mekelle-Samre) Suluh (Senkata-Hawzen) Illala (Dollo) Tankua (Geba junction) Agoro (Geba junction) Genfel (Tannery) Genfel (Geba junction) Agulae (Agulae town) Suluh (Abreha-Atsbeha) Agulae (Geba junction) Suluh (Geba junction) Geba (Mekelle) Geba (Avergelle bridge) Geba (Jigike) Geba (Tekeze junction)

GPS coordinates

Base flow (m3/s)

Easting

Northing

Measured

Simulated

545499 552902 560455 486681 532309 563045 545120 562718 585349 545202 544639 540961 502904 482400 460483

1479571 1549136 1492230 1495282 1489347 1521168 1508612 1513261 1530640 1505448 1508987 1503370 1488174 1496986 1503960

0.04 0.08 0.11 0.14 0.18 0.18 0.28 0.45 0.47 0.51 1.36 1.36 2.30 2.86 2.90

0.08 0.35 0.04 0.11 0.26 0.32 0.36 0.11 0.55 0.17 0.64 1.18 1.77 2.05 2.15

measured and simulated base flow is 0.97, indicating that the agreement between observations and model predictions is good. Also, the discrepancy between simulated and measured base flow values diminishes when going further downstream. This is consistent with the overall flow comparison made earlier and the consideration that part of the groundwater flow in upstream areas of the basin may join the river system further downstream. Hence, considering the scale and topographical complexity of the basin and limitations and unreliability of hydrological and

120

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

meteorological data, it may be concluded that the WetSpass model predictions compare reasonably well with river flow observations. 4. Discussion The accumulated runoff volume in the rainy season is an indication of how much runoff water could be harvested every year during the rainy season. This can be estimated as:

V¼

Z

Sr dA  103 ;

ð4Þ

A

where V is the accumulated runoff volume (m3), Sr is the generated surface runoff in the rainy season (mm), and A is the upstream area (m2). The accumulated runoff volume map can be used to estimate the inflow to existing reservoirs or to identify interesting places for water harvesting. Figure 6 shows the summer surface runoff accumulation map of the Geba basin produced using Eq. (4). From the surface runoff accumulation map, it can be observed that small-scale surface water harvesting for irrigation should be possible in the northern part of the basin between Wukro and Edaga-Hamus, especially on the Atsbi plateau. Less surface water is available in the east, central and west part of the basin, except for the main rivers. The existing reservoirs are also shown in the figure. Due to financial and technical constraints and topographic complexity of the area, the focus of the regional government for the moment is to build small-scale water harvesting reservoirs in the order of 0.2 to 2  106 m3. Since 1992, the Commission for Sustainable Agriculture and Environmental Rehabilitation in Tigray (COSAERT) constructed 54 micro-dam reservoirs in Northern Ethiopia, most of which are located in the Geba basin. These micro-dams are built on small streams to retain surface water for irrigation and domestic purposes (Haregeweyn et al., 2006). However, many of these reservoirs are not meeting their expectations. As the availability of sufficient surface runoff is thought to be one of the reasons for failure of some reservoirs in the region, a comparison is made between the design capacity of the reservoirs and the available surface runoff based on the WetSpass results. For instance, the Arato

reservoir, located some 15 km east of Mekelle on the Illala River, is one of the failing reservoirs. Though the design capacity of the reservoir was 2.59  106 m3 (Eyasu, 2005), it practically holds almost no water as the inflow is insufficient to compensate the loss by evaporation and leakage (Gebremedhin et al., 2013). The WetSpass result shows that the accumulated surface runoff at this location is only 0.096  106 m3. Another example is the Hashenge reservoir on the Illala River, about 3 km southeast of Arato, which was designed to hold 2.23  106 m3 of water. This reservoir is mostly dry except for a few weeks during the rainy season (Gebremedhin et al., 2013). The WetSpass result shows that at this location only 0.062  106 m3 of water can be harvested during the rainy season. Rubafeleg is one of the few successful reservoirs on the Genfel River, located on the Atsbi plateau about 15 km north of Atsbi town. It has a design capacity of 2.70  106 m3, while the WetSpass result predicts 2.27  106 m3. In general, the design capacity of all reservoirs east of Mekelle and many others in the northern part of the basin is found to be significantly higher than what the WetSpass model predicts. Of course there is also some surface runoff going to the reservoirs during the winter season (especially in the northern part) and some base flow during the summer season, which were not considered in this evaluation. But nevertheless, overall it seems that many reservoirs have been wrongly designed or wrongly positioned. The groundwater potential for water use to support crop growth by irrigation can be assessed by means of a safe yield map, which indicates how much groundwater can be abstracted in a sustainable way without depleting the groundwater resources. Safe yield is usually expressed as a percentage of the groundwater recharge. Several authors suggested different percentages, from the least conservative 100% to a reasonably conservative 10% (Miles and Chambet, 1995). The former estimate is largely outdated as nowadays it is accepted that, in general, sustainable yield must be considerably less than the groundwater recharge to sustain both the quantity and quality of streams, springs, wetlands and groundwater dependent ecosystems (Sophocleous, 2000). Hence, in this study a reasonably conservative estimate of the sustainable yield of 25% of the groundwater recharge is adopted, i.e.,

Fig. 6. GIS map of accumulated surface runoff (m3) in the Geba basin for the wet summer season derived from WetSpass model results.

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

121

Fig. 7. GIS map of safe yield for groundwater abstraction (m3/d/ha) in the Geba basin derived from WetSpass model results.

SY ¼ 6:85  103 R;

ð5Þ

where SY is the safe yield groundwater abstraction rate expressed in (m3/d/ha) and R is the total annual groundwater recharge expressed in (mm). The resulting safe yield map of the Geba basin is presented in Fig. 7. The values range from 2 to 11 m3/d/ha. The largest values occur in the basin north of Wukro and in the centre around Agbe, Hagere-Selam and Abi-Adi. The eastern part around Agulae and the lowlands in the west at the Geba outlet have the lowest safe yield. The water deficit for crop growth can be estimated as the difference between the crop water requirement and the actual evapotranspiration that is feasible only by rainfall. The crop water requirement, defined as the amount of water needed to meet the water loss through evapotranspiration for optimal crop growth, can be estimated as a crop coefficient times reference evapotranspiration of well-watered grass (Allen et al., 1998). Crop coefficients vary between 0.70 and 1.15 depending on crop type and growing stage. As a first guess, the crop coefficient can be assumed to be one and reference evaporation to equal PET, which allows estimating how much water is needed for supplementary irrigation for optimal crop growth (Kendy, 2003). Such calculation can be made on annual basis, but it is more interesting to do this separately for the wet summer season and the dry winter season. Hence, the water deficit can be calculated as:

WD ¼

10 ðEp  ETÞ; n

ð6Þ

where WD is the water deficit for ideal crop growth expressed in (m3/d/ha), Ep and ET are the potential and actual evapotranspiration in (mm) for the rainy season or for the dry season, and n is the number of days in a season. For the WetSpass model of the Geba basin, June–September are considered as wet months (120 days), while the remaining 8 months (245 days) are considered to be the dry season. Fig. 8a and b shows the crop water deficit maps based on Eq. (6) for the summer and winter season, respectively. In the wet summer season, values range between 49 (negative value indicating excess water) and 23 m3/d/ha, with a mean crop water deficit of 9.5 m3/d/ha. Comparison of the safe yield map with the summer season crop water deficit map shows that groundwater can be safely abstracted to supplement the water deficit by irrigation in most parts of the basin. For the dry winter season, values range between 18 and 52 m3/d/ha with a mean crop water deficit of 42 m3/

Fig. 8. GIS maps of crop water deficit (m3/d/ha) in the Geba basin derived from WetSpass model results: (a) in the wet summer season, and (b) in the dry winter season.

d/ha. Evidently, the crop water deficit in the winter season is highly influenced by the potential evapotranspiration, which in turn is strongly dependent on elevation. Hence, the highest water deficits are found in the low elevation areas such as the lowlands in the west and centre, while the lowest water deficits occur in the northern highlands around Edaga-Hamus and Senkata and to some extent in the Hagere-Selam highlands in the west. It follows that, even for a cropping period of a few months during the dry winter season, the needed water for irrigation in most parts of the Geba basin is much larger than the predicted safe yield, and, hence, too high to be supplemented by groundwater abstraction in a sustainable way.

5. Conclusions The WetSpass model was applied to calculate the water balance of the Geba basin in northern Ethiopia. Specific input data were prepared in the form of digital maps using remote sensing images, various GIS tools, FAO and NASA databases, field reconnaissance

122

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123

and meteorological observations. Parameter attribute tables in the WetSpass model were adjusted to the conditions prevailing in the Geba basin. Results of the model indicate that evapotranspiration constitutes 76% of the total annual precipitation and is mainly influenced by precipitation and potential evapotranspiration and to some extent by soil type. Annual surface runoff in the Geba basin ranges from 0 to 268 mm with a mean of 108 mm, which amounts to 18% of the total precipitation and occurs mainly during the summer season. Bare land, agriculture and urban land-uses on sandy clay soils produce the highest surface runoff in the basin. Annual groundwater recharge in the Geba basin ranges from 0 to 191 mm with a mean of 37 mm, which constitutes only 6% of the annual precipitation. Sandy loam and loam soils on bare land and agriculture produce the highest groundwater recharge in the Geba basin. Comparison of observed and simulated river flows at six gauging station in the basin show that the model predictions are reasonable and realistic. The discrepancies likely result from the errors involved in estimating the base flow due to groundwater drainage, as groundwater flow systems in the basin are largely unknown. Also a good agreement was achieved between predicted and observed river base flows performed in January–February, 2008 at 15 cross-sections along the river system, The accumulated summer surface runoff map shows that the northern part of the basin is most promising for small-scale reservoirs to harvest surface water, while the eastern part of the basin is less promising, although most of the reservoirs have been built there. Comparison of the design capacity and the estimated accumulated summer runoff (inflow) of some failed reservoirs (e.g. Arato and Hashenge) show that either these dams were built at wrong locations or the inflow was overestimated. Based on the assumption that 25% of the annual groundwater recharge can be abstracted safely from the groundwater reservoirs, the northern part of the basin around Wukro and Sinkata and the southwestern part around Hagere-Selam and Abi-Adi are the most promising areas for sustainable groundwater abstraction. Crop water deficit maps indicate that during the summer season there is sufficient water available in many parts of the basin or the deficit is small. But for the winter season, the water deficit is high everywhere, although it is inversely correlated with elevation. According to the WetSpass results, the safe yield of the groundwater reserves in the Geba basin is sufficient to sustain optimal crop growth by supplementary irrigation in the summer season when rainfall is erratic or insufficient, but is too small to sustain crop irrigation in the dry winter season. This study also demonstrated that regional assessment of water resources for management purposes is feasible with only limited local data by using global data and modeling tools that are readily available in contemporary practice. Hence, this approach can have large applicability in other basins, especially in developing regions. Acknowledgements The authors would like to thank two anonymous reviewers for their useful comments and suggestions, which enabled to improve the quality of the paper. This work formed part of the PhD thesis of the principal author Tesfamichael (2009). References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evaporation – guidelines for computing crop water requirements. FAO Irrigation and Drainage Papers 56. FAO, Rome, Electronic version (visited 04.15.09). Batelaan, O., De Smedt, F., 2001. WetSpass: a flexible, GIS based, distributed recharge methodology for regional groundwater modelling. In: Gehrels, H.,

Peters, J., Hoehn, E., Jensen, K., Leibundgut, C., Griffioen, J., Webb, B., Zaadnoordijk, W-J. (Eds.), Impact of Human Activity on Groundwater Dynamics, IAHS Publ. No. 269, pp. 11–17. Batelaan, O., De Smedt, F., 2007. GIS-based recharge estimation by coupling surfacesubsurface water balances. J. Hydrol. 337 (3–4), 337–355. Chernet, T., Eshete, G., 1978. Hydrogeological map of Mekele, scale 1:250,000. Ethiopian Mapping Agency. CoSAERT (Commission for Sustainable Agriculture and Environmental Rehabilitation in Tigray), 2001. Suluh valley integrated rural, agriculture and water resources development study. Identification and Reconnaissance Report, Mekelle, Ethiopia. DevCon (DevConsultants Limited), 1992. Five towns water supply and sanitation study. Ministry of Water Resources, Addis Ababa, Ethiopia. Devlin, J.F., Sophocleous, M., 2005. The persistence of the water budget myth and its relationship to sustainability. Hydrogeol. J. 13, 549–554. EPA, 1998. National action plan program to combat desertification. Volume I: The state of natural resources in arid, semi-arid and dry sub-humid area. Environmental Protection Agency, Accra, Ghana, 158 pp. Eyasu Yazew, 2005. Development and management of irrigated lands in Tigray, Northern Ethiopia. PhD thesis, Wageningen University, The Netherlands, 284 pp. FAO, 1998. Land and water digital media series – the soil and terrain database for northeastern Africa. FAO, Land and Water Digital Media Series no. 2, FAO, Rome. Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice-Hall, Inc., Englewood Cliffs (NJ). Frenken, K. (Ed.), 2005. Irrigation in Africa in figures – AQUASTAT survey 2005. FAO Water Reports 29, FAO, Rome, 74 pp. Gebremedhin Berhane, Martens, K., Al Farrah, N., Walraevens, K., 2013. Water leakage investigation of micro-dam reservoirs in Mesozoic sedimentary sequences in Northern Ethiopia. J. Afr. Earth Sci. 79, 98–110. Haitjema, H.M., Mitchell-Bruker, S., 2005. Are water tables a subdued replica of the topography? Ground Water 43 (6), 781–786. Haregeweyn, N., Poesen, J., Nyssen, J., De Wit, J., Haile, M., Govers, G., Deckers, S., 2006. Reservoirs in Tigray (Northern Ethiopia): characteristics and sediment deposition problems. Land Degrad. Develop. 17, 211–230. Healy, R.W., 2010. Estimating Groundwater Recharge. Cambridge University Press. Hussein, A., 2000. Hydrogeology of the Aynalem well field, Tigray, Northern Ethiopia. MSc thesis, Addis Ababa University. Kendy, E., 2003. The false promise of sustainable pumping rates. Ground Water 41 (1), 2–4. Kibrewossen Tesfagiorgis, Tesfamichael Gebreyohannes, De Smedt, F., Moeyersons, J., Miruts Hagos, Nyssen, J., Deckers, J., 2011. EthiopiaEvaluation of the groundwater resources in the Geba basin, Ethiopia. Bull. Eng. Geol. Environ. 70 (3), 461–466. Kinzelbach, W., Aeschbach, W., Alberich, C., Goni, I.B., Beyerle, U., Brunner, P., Chiang, W.H., Rueedi, J., Zollmann, K., 2002. A survey of methods for groundwater recharge in arid and semi-arid regions. Early Warning and Assessment Report Series, UNEP/DEWA/RS.02-2. United Nations Environment Programme, Nairobi. Lerner, D.N., Isaar, A.S., Simmers, I., 1990. Groundwater recharge: a guide to understanding and estimating natural recharge. In: IAH International Contributions to Hydrogeology, vol. 8. Verlag Heinz Heise, Hanover. Middleton, N., Thomas, D.S.G., 1992. World Atlas of Desertification. United Nations Environment Program, Edward Arnold, London, UK, 69 pp. Miles, J.C., Chambet, P.D., 1995. Safe yield of aquifers. J. Water Resour. Plan. Manage. 121, 1–8. NEDECO (Netherlands Engineering Consultants), 1997. Tekeze river basin integrated development master plan project. Ministry of Water Resources, Addis Ababa, Ethiopia. Raes, D., 2009. ETo Calculator – Evapotranspiration from a reference surface, Reference manual, FAO, Rome, 38 pp. Simmers, I. (Ed.). 1997. Recharge of phreatic aquifers in (semi-)arid areas. Int. Ass. Hydrogeol. 19, A.A. Balkema, Rotterdam. Sophocleous, M.A., 2000. From safe yield to sustainable development of water resources – the Kansas experience. J. Hydrol. 235, 27–43. Sophocleous, M.A., Devlin, J.F., 2004. Discussion of ‘‘The Water Budget Myth Revisited: Why Hydrogeologists Model’’ by John D. Bredehoeft. Ground Water 42, 618. Tesfamichael Gebreyohannes, 2009. Regional groundwater flow modelling of the Geba basin. PhD thesis, Vrije Universiteit Brussel, Belgium, 250 pp. Tesfamichael Gebreyohannes, De Smedt F., Miruts H., Solomon G., Kassa A., Kurkura K., Abdulwassie H., Nyssen J., Bauer H., Moeyersons J., Deckers J., Mitiku H., Nurhussein T., 2010. Large Scale Geological mapping of the Geba basin, Northern Ethiopia. Tigray Livelihood papers no. 9, 40 pp. Tóth, J., 2009. Gravitational Systems of Groundwater Flow. Cambridge Univ. Press. Vandecasteele, I., Nyssen, J., Clymans, W., Moeyersons, J., Martens, K., Van Camp, M., Tesfamichael, Gebreyohannes, De Smedt, F., Deckers, J., Walraevens, K., 2011. Hydrogeology and groundwater flow in a basalt-capped Mesozoic sedimentary series of the Ethiopian highlands. Hydrogeol. J. 19, 641–650. Walraevens, K., Vandecasteele, I., Martens, K., Nyssen, J., Moeyersons, J., Tesfamichael, Gebreyohannes, De Smedt, F., Poesen, J., Deckers, J., Van Camp, M., 2009. Groundwater recharge and flow in a small mountain catchment in northern Ethiopia. Hydrol. Sci. J. 54 (4), 739–753. WWDSE (Water Works Design and Supervision Enterprise), 2006. Evaluation of Aynalem well field and selection of additional prospective boreholes for Mekelle town water supply source. Report vol. II, 140 pp.

T. Gebreyohannes et al. / Journal of Hydrology 499 (2013) 110–123 Yacob Arsano, Imeru Tamrat, 2005. Ethiopia and the Eastern Nile Basin. Aquat. Sci. 67, 15–27. Yilma, Seleshi, 2002. Estimation of monthly reference evapotranspiration (ETo) for areas with adequate meteorological data in Ethiopia. J. Ethiop. Eng. Archit. 19, 21–28.

123

Yilma Seleshi, Camberlin, P., 2006. Recent change in dry spell and extreme rainfall events in Ethiopia. Theor. Appl. Climatol. 83, 181–191. Yilma Seleshi, Zanke, U., 2004. Recent changes in rainfall and rainy days in Ethiopia. Int. J. Clim. 24, 973–983.