Delineation of groundwater potential zones of the transboundary aquifers within the semiarid Bulal catchment, Southern Ethiopia

In the semiarid Bulal transboundary catchment of southern Ethiopia, groundwater is the only reliable drought-resilient water source. The central and southern parts of the catchment are dominantly overlain by the transboundary aquifers of the Bulal basalts, while the basement rocks outcrop in the eastern part. This study uses an integrated geographic information system (GIS), remote sensing (RS), and analytical hierarchal process (AHP) to identify and delineate the groundwater potential zones of the semiarid Bulal catchment within the Ethiopian territory. Based on their relative importance to groundwater occurrence and movement, ten input parameters were chosen. According to Saaty’s AHP approach, the input themes and each of their distinct features were given normalized weights. A composite groundwater potential zone index (GWPZI) map was generated by integrating all the input layers employing the GIS-overlay analysis technique. The map was validated using the yield of wells from the catchment. The GWPZI map depicts four groundwater potential zones: high (representing 27% of the total area), moderate (20%), low (28%), and very low (25%). The geological feature has the greatest influence on the distribution of groundwater potential. Areas with high potential are mainly overlain by the Bulal basaltic flow, while low groundwater potential zones are in the regolith over the basement rocks. Unlike conventional methods, our novel approach is effective in identifying relatively shallow GWPZs throughout the catchment, and it can be applied in similar semiarid regions. The GWPZI map serves as a quick guide for effectively planning, managing, and developing the catchment’s groundwater resources.


Introduction
Most arid and semiarid climate regions in the world underlain by hard rocks face severe water shortages as a result of their physiographic conditions, which are exacerbated by climate change (Ahmed et al., 2007), and groundwater extraction has become an immediate alternative (Fenta et al., 2015). Although the Ethiopian highlands are major sources of recharge to the transboundary aquifers (TBAs) such as the Bulal catchment and have a significant distribution of the spatiotemporal potential of groundwater, access to fresh water is a challenge in the drought-prone peripheral regions (Alemayehu, 2006(Alemayehu, , 2010Moges, 2012). Groundwater is a vital resource that provides more than 80% of the domestic and irrigation uses of Ethiopia (Kebede et al., 2005;Khadim et al., 2020). It is also often considered a reliable, droughtresilient resource (Calow et al., 1997;Taylor et al., 2013) that exists in all geological formations (Alemayehu et al., 2017;Ayenew et al., 2008;Legesse & Ayenew, 2006). It ensures livelihood, especially in the southern parts of the country with prevailing semiarid climatic conditions where a severe drought is common (Mera, 2018;Razak et al., 2020). Groundwater in semiarid regions underlain by hard rocks mainly occurs through indirect recharge from influent riverbeds during high flood seasons, along secondary porosity and intensive fractures and weathering zones (Lloyd, 1999).
The groundwater resource potential of the TBAs in a water-stressed environment must be systematically evaluated and properly monitored to manage groundwater depletion problems and the impacts of climate change (Adiat et al., 2012;Alemayehu, 2010;Hashemi et al., 2015). Therefore, identifying groundwater potential zones is important. Accordingly, in previous studies conducted by Oromia Water Works, Design and Supervision Enterprise (OWWDSE) in 2017, six groundwater potential zones (well fields) have been identified and delineated within the present study area after utilizing intensive and costly conventional investigation techniques through geological, hydrogeological, geophysical, hydrochemical, and isotope investigations, including drilling of test wells and groundwater modeling (Razak et al., 2020). Therefore, although delineation of the groundwater potential zones could be better achieved through conventional approaches, they are costly and time-consuming compared to modern geographical information system (GIS) and remote sensing (RS) techniques (Diwakar & Thakur, 2012). In addition, conventional exploration methods may not also be very reliable as they do not combine major governing factors that can control the occurrence and movement of groundwater (Vidhya & Vinay, 2019).
Currently, modern techniques of statistical probabilistic-based models using an integrated application of GIS and RS are being widely used as efficient and cost-effective tools for delineating potential groundwater areas, especially in a semiarid environment (Sharma, 2016). The analytical hierarchical process (AHP) has proven to be the most effective multi-criteria decisionmaking tool (MCDM) (Jha et al., 2010) among the statistical models/tools such as the weight of evidence, evidential belief (Tahmassebipoor et al., 2016), analytical hierarchical process (Sahoo et al., 2017;Sandoval & Tiburan, 2019), logistic regression , frequency ratio (Hong et al., 2018), weighted overlay (Senthilkumar et al., 2019), and multi-influencing factors (Anbarasu et al., 2019). The AHP involves the integration of multiple geo-environmental and hydrogeological parameters such as lithology, structure, land use land cover, geomorphology, drainage, slope, etc. that can govern groundwater movement and occurrence (Dar et al., 2011;Jha et al., 2010;Singh et al., 2011a;Srivastava & Bhattacharya, 2006). The method has also been proven to be an important tool for preparing thematic maps and conducting multi-criteria decision analysis (Vittala et al., 2005;Madrucci et al., 2008;Chowdhury et al., 2009;Javed & Wani, 2009;Dar et al., 2010;Jha et al., 2010). Integrated RS and GIS techniques for the assessment and delineation of groundwater potential and recharge zones have been successfully applied in some semiarid regions of Ethiopia and Africa (e.g., Abdelkarim et al., 2022;Kisiki et al., 2022;Mengistu et al., 2022;Owolabi et al., 2020;Seifu et al., 2022).
The Bulal catchment underlain by hard rocks of the volcanic and crystalline TBAs shared between Ethiopia and Kenya (Kebede et al., 2010;Razak et al., 2020) is among the 7 major TBAs of Ethiopia (IGRAC & UNESCO-IHP, 2015). This work focuses only on the Bulal TBAs within the Ethiopian territory. The Bulal catchment is characterized by a semiarid climate, which receives low annual rainfall (Razak et al., 2020) and has been severely affected by drought in recent decades (Mera, 2018). As a result, intermittent rivers/streams and lowered groundwater levels are common phenomena in the catchment (Razak et al., 2020). The Bulal catchment is mainly overlain by metamorphic basement complexes in the eastern part, Quaternary rift volcanic rocks in the southern part, and Neogene pre-rift volcanic rocks in the western part covered by thick layers of recent superficial deposits in places (Gebeyehu et al., 2022;Razak et al., 2020). The groundwater is confined in three major aquifer domains: (1) quaternary sediments; (2) Borena basement, and (3) Bulal volcanic rocks (Kebede et al., 2010).
This study aimed at identifying and delineating the groundwater potential zones of the catchment using an integrated application of GIS and RS in combination with Saaty's Analytical Hierarchal Process (AHP) (Saaty, 2008). Ten thematic map layers that have significance for groundwater occurrences have been used and integrated by using GIS-overlay analysis to produce the groundwater potential zone index (GWPZI) map. The method is considered effective in identifying shallow to moderately deep groundwater zones. Subsequently, the results should be verified with ground field checks. In this study, data obtained from detailed geological, hydrogeological, and geophysical field investigations from 2010 to 2017 (OWWDSE, 2017), and again in 2022, as well as from test wells, were utilized to validate the precision of the current GWPZI map output. The results of this study could serve as a quick guide and framework for policy-and decision-makers for developing sustainable groundwater resource studies, development, and integrated management plans on the TBAs of the catchment. The rationale behind this study is to encourage a multilateral agreement for sustainable management and uses of the groundwater resources in a drought-prone TBAs basin.

Location
The Bulal catchment is located in southern Ethiopia within the Borena Zone, Oromia regional state straddling the Ethiopia-Kenya border (Fig. 1). It covers an area of 14,641 km 2 . The catchment can be accessed by a 464 km main asphalt road from Addis Ababa, the capital of Ethiopia, through Bule Hora and Yabelo towns. Several all-weather and dry-weather roads are also available for intersite mobilization.

Climate, drainage, and topography
The Bulal catchment is characterized by an arid to semiarid climate with relatively high temperatures throughout the year with mean monthly maximum and minimum temperatures of 26 °C and 15 °C, respectively, and a mean annual temperature of 20.2 °C. The rainfall pattern is characterized by two dry seasons (from December to February and July to August) and two rainy seasons, from March to May and September to November. The mean annual rainfall in the catchment is ~ 608 mm (NMSA, 2021).
The catchment is mainly drained by intermittent rivers/streams receiving runoff from the western and eastern highlands. The flows of drainages towards the Ethiopia -Kenya border in the south are mainly controlled by the N-S trending Ririba fault zone. The structurally controlled runoff infiltrates from the riverbeds into the groundwater system mainly along the fault zone in the central low-lying areas and at the foot of the Mega fault scarp in the southeastern parts (Razak et al., 2020). The catchment is mainly characterized by rectangular and parallel to sub-parallel drainage patterns (Fig. 1).
The Bulal catchment is part of the Main Ethiopian Rift (MER) forming a topography reflecting volcanotectonic activity. A flat to gently rolling terrain characterizes a substantial portion of the catchment area with a slope gradient of less than 3°. The elevation ranges from 702 m a. s. l. in the south to 2482 m a. s. l. in the centraleastern parts of the catchment (Fig. 1). The catchment exhibits two distinct physiographic regions i.e., the eastern warped plateau and associated ridges, and the central-southern rift floor plain with some isolated hills and valleys ( Fig. 1).

Geological and hydrogeological setting
Quaternary rift volcanic rocks are the main geological formations in the Bulal catchment occupying the central and southern regions (Gebeyehu et al., 2022). They consist mainly of basaltic agglomerates, scoria falls, lappili tuffs, and associated pyroclastic rocks, as well as vesicular and massive basalts (Bulal basalts). The Bulal basalt dominates the volcanic sequences in the catchment and is a horizontally layered basaltic lava flow with an average thickness of 200 m. The pre-rift basaltic flows associated with minor felsic pyroclastic deposits, phonolites, trachytes, and rhyolites occupy a limited area on the western elevated ridges of the catchment (Gebeyehu et al., 2022;Razak et al., 2020). The basement complex terrain occupying the eastern and southeastern parts is underlain by Precambrian basement rocks of quartzo-feldspathic, layered, and granitic gneisses with few granitic intrusions. On the other hand, most low-lying areas of the catchment are covered by relatively thick layers of recent superficial deposits, which comprise alluvial and eluvial sediments, and weathered regolith developed over the crystalline basement rocks (Fig. 2).
The Bulal catchment is highly affected by NW-SE, N-S, and E-W oriented lineaments and faults. The N-S oriented Ririba normal fault system is one of the major fault zones in the catchment which runs from north to south of the whole catchment (Fig. 2). These structures have significance in governing the groundwater occurrence and flows in the catchment (Razak et al., 2020). In addition, crater maars and alluvial fans are common.
The fractured and weathered Bulal basalt is the major TBA unit in the catchment (Aquifer class I) (Gebeyehu et al., 2022;Kebede et al., 2010;Razak et al., 2020). Gebeyehu et al. (2022) indicated that this hydrostratigraphic unit is highly permeable and productive with average hydraulic conductivity (K) and transmissivity (T) values of 13 m/day and 642 m 2 /day, respectively. Its average saturated thickness is ~ 70 m. Localized fractured and weathered basaltic aquifers intercalated with acidic volcanics (Aquifer class II) exhibit moderate productivity with average K and T values of 1 m/day and 115 m 2 / day, respectively. Intergranular alluvial sediments and alluvial fans (Aquifer class III) over a limited area have a high storage potential for the underlying basaltic aquifer and moderate shallow groundwater productivity. A borehole drilled through this aquifer yields ~ 9 l/s. An extensive intergranular hydrostratigraphic unit of unwelded pyroclastic deposits and regolith (Aquifer class IV), and weathered and fractured crystalline basement aquifers (Aquifer class V) act as local aquicludes and leaky The deeper groundwater system of the Bulal catchment occurs under confined and semi-confined conditions mainly within the fractured zones of the Bulal basaltic aquifers (Kebede et al., 2010). The groundwater flow system in this aquifer media is dominantly discrete and fracture-controlled, and the flow converges from the eastern, western, and northern highlands to the southern low-lying areas following the Ririba fault system (Gebeyehu et al., 2022;Kebede et al., 2010;Razak et al., 2020). On the other hand, shallow groundwater confined to the upper thick unconsolidated sediments and regolith has very short but continuous flow paths. Recharge mainly takes place from floods as diffuse and preferential indirect recharge at the riverbeds along the Ririba fault system and the Mega fault belt. Direct recharge from rainfall is very limited. The average annual recharge of the catchment estimated by a water balance model is 54 mm/year (Kebede et al., 2010;Razak et al., 2020), while it varies between 30 mm/year and 45 mm/year as estimated by the chloride-mass balance (CMB) method (Kebede et al., 2010). According to the studies conducted by OWWDSE in 2017 (Razak et al., 2020), areas under the first group were found to be the most prospective potential zones for groundwater development in the catchment (Fig. 4).

Methodology
The delineation of the groundwater potential zones in the Bulal catchment was performed using a GIS-based overlay analysis technique. The overlay analysis was carried out by integrating ten thematic input layers that are important for groundwater occurrence and movement. Input layers include geology, groundwater level, soil texture, lineament density, and remote sensingderived data for land use/land cover (LULC), rainfall, drainage density, slope, topographic wetness index, and  Gebeyehu et al., 2022;Razak et al., 2020) topographic variability. All the maps were prepared using the ArcGIS 10.8 platform and presented in UTM projection, zone 37 with reference datum of WGS84. All input raster layers were resampled to 100-m cell size resolution based on Tobler's rule (Tobler, 1987;Rajinder Nagi, 2010) by taking the output map scale (1:250,000) into consideration. Saaty's AHP as a Multi-Criteria Decision-Making (MCDM) tool was used to assign the normalized weights to all input layers and their respective classes based on their relative influence on groundwater occurrence.

Selection of the input layers and their importance
To delineate the groundwater potential zones of the catchment, the major controlling factors of groundwater movement, storage, and occurrence should be investigated and identified (Kolli et al., 2020;Tolche, 2021). The selection of the input layers for groundwater potential zone mapping was achieved by taking into consideration the relative influence of the input parameters on the groundwater occurrence, data availability, and the semiarid nature of the catchment (Razandi et al., 2014). The occurrence, movement, quality, and quantity of groundwater, particularly in semiarid areas, are governed by factors such as the underlying rock formations and their structural fabric, the thickness of weathered material, the topography, and climatic conditions (Gintamo, 2015;Singh et al., 2011b;Thakur et al., 2011). Accordingly, ten (10) parameters were used. Geology (GG), lineament density (LD), land use/land cover (LULC), slope (SL), topographic variability (TV), depth to groundwater (GD), rainfall (RF), drainage density (DD), soil texture (ST), and topographic wetness index (TWI) were selected and integrated to produce the GWPZI map of the catchment.
The geology of an area is one of the major factors that can affect the occurrence, movement, and quality of groundwater (Janardhana & Reddy, 1998;Rajaveni et al., 2015). It can determine the groundwater storage aquifer media and has a significant impact on groundwater recharge conditions (Shaban et al., 2006). Lineaments such as joints, fractures, and faults are straight to curvilinear geological discontinuities that are widely used in groundwater investigations to locate groundwater prospect sites on fractured bedrock (Edet et al., 1998;Sankar, 2002). They may act as conduits for groundwater movement and storage, leading to increased secondary porosity and, therefore, can serve as groundwater potential zones (Idris et al., 2018;Prasad et al., 2008;Rao, 2006). They can also influence the hydraulic properties (i.e., transmissivity, hydraulic conductivity, storativity) of the geological formations (Idris et al., 2018). Therefore, the connectivity and density of fractures/lineaments are significant determining factors for groundwater occurrence, storage, and flow of catchments such as Bulal, which are underlain by hard rocks (Idris et al., 2018).
Land use/land cover (LULC) is another important factor governing groundwater storage and recharge in a catchment (Singh et al., 2013). It can influence the groundwater recharge and soil moisture of the region (Verma & Patel, 2021).
Soil texture generally has a significant role in controlling the amount and rate of infiltrating water and recharge. The rate of infiltration largely depends on the grain size and permeability of the soils (Jasrotia et al., 2016).
Topographic variability controls groundwater subsurface movement, as elevation variation affects hydrological processes and the occurrence of groundwater potential (Kumar & Krishna, 2018;Muralitharan & Palanivel, 2015;Pourghasemi et al., 2020). The occurrence and flow of groundwater are also strongly governed by the slope gradient (Yeh et al., 2016). It has a strong effect on the infiltration of surface water and surface runoff depending on the slope variations, which in turn controls the groundwater recharge (Rajaveni et al., 2015;Strahler, 1964). Furthermore, the topographic wetness index (TWI) plays an important role in the occurrence and development of groundwater. The TWI was first developed by Beven and Fig. 4 Previously identified groundwater potential zones of the Bulal catchment using conventional investigation methods (modified after Razak et al., 2020) Kirkby (1979). The TWI has been widely used to quantify topographical controls on hydrological processes and groundwater outflow to the surface (Beven et al., 1988;Famiglietti & Wood, 1991).
Precipitation in arid and semiarid regions is one of the major input parameters in determining the availability of groundwater recharge for aquifers (Thomas & Vijayasekaran Duraisamy, 2018;Khan et al., 2022). The amount of recharge varies with the amount and intensity of rainfall (Verma & Patel, 2021).
Drainage density is a function of exposed rock permeability. In areas underlain by a low permeability rock, the drainage density is likely to be high, which in turn leads to low infiltration and greater runoff (Basavaraj Hutti & Nijagunappa, 2020;Verma & Patel, 2021). The depth to groundwater level, on the other hand, is an important indicator of the existence and sustainability of groundwater resources in arid and semiarid regions. Seasonal groundwater level fluctuation in shallow aquifer systems of semiarid regions depends primarily on groundwater discharges, recharge, and the rate of evapotranspiration (Jhariya et al., 2016;Pavelic et al., 2012).

Data acquisition and preparation of the thematic layers
A geological map of the catchment area at the scale of 1:250,000 was obtained from the OWWDSE. It was modified and further reclassified into four major lithological units by taking their hydrogeological significance into consideration (Gebeyehu et al., 2022). Digital lineament features extracted manually from Sentinel-2 images were also obtained. Subsequently, the lineament density map layer was computed using a GIS algorithm with a 3 km buffer radius, which is expressed in terms of the total length of lineament per unit area (km/km 2 ) as expressed in Eq. 1 (Yeh et al., 2016).
where ∑ i = n i = 1 L i represents the total length of lineaments (L) and A represents a unit area (L 2 ).
A vector data set of the land use/land cover (LULC) and soil maps of the catchment area prepared by OWWDSE at the scale of 1:50,000 (OWWDSE, 2017) were used and converted into raster format in a 100 m cell size resolution in ArcGIS 10.8. The 2014 Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM) at 30 m resolution from an open-source product of the United States Geological Survey (USGS, 2021) was used to produce thematic topographic variability and slope maps. The DEM was first gap-filled and resampled to a cell size of 100 m. In addition, the topographic wetness index (TWI) was prepared from the slope (in degrees), and the flow direction and flow accumulation input raster maps that were generated from the gap-filled DEM. Subsequently, the TWI was estimated and prepared in a GIS environment employing a "TOPMODEL" simulation as expressed in Eq. 2 (Beven, 1997).
CHIRPS' monthly precipitation data for 22 years (1999-2021) were downloaded from the USGS website (https:// chc. ucsb. edu/). The remotely sensed rainfall data were calibrated using data collected from the National Meteorological Service Agency (NMSA) at four stations within the catchment (i.e., Mega, Moyale, Teltele, and Yabelo). Then, the point data were interpolated using the inverse distance weighted (IDW) tool in the GIS environment to produce the mean annual rainfall map layer. Drainage of the catchment area was first generated from a gap-filled DEM (30 m), and subsequently, the drainage density map (DD) was produced by using a line density algorithm in ArcGIS 10.8. The DD is expressed in terms of the length of channels per unit area (km/km 2 ). Finally, average values of water level data from 69 wells, including 13 springs measured during both dry and wet seasons from the 2008 to 2017 period, were interpolated with IDW in Arc-GIS 10.8 to produce the groundwater depth map of the Bulal catchment.
Weight assignment for the thematic layers Appropriate weights were assigned to the ten input themes and their individual features after determining their relative importance in causing groundwater occurrence and movement in the Bulal catchment. The normalized weights of the individual themes and their different feature classes were obtained through Saaty's AHP approach (Saaty, 2008). AHP as an MCDM tool of experts' decisions measurement by pairwise (2) TWI = ln α tan β comparison was first introduced by Saaty in 1980. It has been widely used and successfully applied in groundwater potential mapping of arid and semiarid regions in particular (Kaliraj et al., 2014;Machiwal et al., 2011;Mallick et al., 2015;Ouma & Tateishi, 2014;Rahaman et al., 2015). AHP is an efficient group decision-making technique that allows users to assess the relative importance of input parameters based on their experiences (Chowdhury et al., 2009;Saaty, 2008). The Eigen normalized weights were employed for the ten input themes and their associated features as per Saaty's 1-9 scale of assignment (Saaty, 2008), which considers the relative importance of each theme and their classes against one another to determine groundwater availability (Tables 1, 2, 3, and 4). The pair-wise comparison matrix of the assigned weights to the ten thematic layers and their individual classes (Table 3) was constructed using Saaty's AHP technique (Eq. 3). The assigned weights are also normalized by calculating the eigenvector and eigenvalue using Eqs. 4 and 5 (Saaty, 1980).
where A 1 is the normalized pair-wise comparison matrix.
where W & W′ are the eigenvector, and w i and w′ i are Eigenvalues of criterion i.
where λ max is the average eigenvalue of the pair-wise comparison matrix. (3) For the computation of the Eigen normalized weights, an open-source Excel-based software version 15.09 of Goepel (2018) was used. Accordingly, geology (GG) was ranked as the dominant factor with a normalized weight of 0.265, while TWI was the least accounted for parameter with a normalized weight of 0.013 for groundwater occurrence in the catchment (Tables 2, 3, and 4).
To determine the uncertainty of the parameters used, the consistency index (CI) of the assigned weights was also calculated following the procedure suggested by Saaty (1990) using Eq. 6. The consistency ratio (CR), which indicates the probability that the matrix ratings were randomly generated, was also computed using the following relations (Eq. 7): where λ max is the principal eigenvalue, n is the number of criteria or factors, and RI is the random consistency index.
(6) CI = λ max ∕(n − 1) (7) CR = CI∕RI  For this study, the CR for assignments of the normalized weights to all input parameters was estimated to be 0.07 which is below the threshold consistency value of 0.1 (Saaty, 1990). The CR values computed for each sub-class of the theme are shown in Table 4.
Integration of the thematic layers using a weighted GIS-overlay analysis All ten normalized weighted thematic layers were integrated using a sum-weighted overlay analysis tool on the GIS platform to demarcate the groundwater potential zones (GWPZs) of the Bulal catchment. The detailed procedures adopted for the delineation of GWPZ are shown as a flow chart in Fig. 5.
The integration of the normalized weighted input parameters to generate a composite GWPZI map of the catchment area was conducted using the procedure stated in Eq. 8.
where GG is the geology, LD is lineament density, LULC is land use/land cover, SL is the slope, TV is topographic variability, GD is depth to groundwater, RF is rainfall, DD is drainage density, ST is soil texture, and TWI is topographic wetness index, while w is the normalized weight of the theme and wi (i = 1 to n) is the normalized weight of a subclass.
The final GWPZI map was further classified into four classes, i.e., "high," "moderate," "low," and "very low" potential zones based on the obtained index values (8) GWPZI = GG w GG wi + LD w LD wi + LULC w LULC wi + RF w RF wi + SL w SL wi + TV w GM wi + GD w GD wi + DD w DD wi + ST w ST wi + TWI w TWI wi from the combined impacts of the input parameters on the groundwater prevalence, employing a natural breaks method in ArcGIS 10.8 platform.
Validation and sensitivity analysis of the GWPZI map A total of 40 wells with yield information were collated from previous studies by OWWDSE (2017). These yield data were used to check the accuracy of the classified GWPZI map of the catchment. In addition, sensitivity analysis (SA) was performed to determine the influence of each input parameter on the GWPZ output model using a map-removal technique (Lodwick et al., 1990). Accordingly, one input layer was removed at a time while the remaining parameters were calculated by using Eq. 9. The obtained sensitivity index (SI) values were then used to assess the impact of the removed layer in delineating the catchment area's GWPZ map.
where SI = sensitivity index of the removed parameter/ layer, GWPI = the groundwater potential index calculated using all input layers, GWPI′ = the groundwater potential index obtained by excluding each input parameter at a time, and N and n are the numbers of input parameters used to delineate GWPZ and GWPZ′, respectively.
A removed input layer with the highest SI indicates the most sensitive parameter while a parameter with the lowest SI is considered the least sensitive layer in delineating the GWPZI map of the catchment.

Grouped geology/lithology
In general, the Bulal catchment is overlain by the transboundary Borena basement complex in the eastern region, while volcanic terrain occupies the western, central, and southern parts of the catchment (see Fig. 2). The Quaternary Bulal basalt dominates the rift volcanics covering an extensive area of the catchment. It is also a major transboundary lithological unit that extends beyond the border to the Kenyan side (Kebede et al., 2010;Razak et al., 2020). The lithological units in the catchment are further classified into four major groups based on their hydraulic properties, which affect their relative hydrogeological importance and productivity ( Fig. 6a and Table 4): (i) Bulal basaltic flow; (ii) alluvial deposits; (iii) felsic volcanics; and (iv) crystalline basement rocks and associated regolith.
The Bulal basalt is mainly characterized by scoriaceous, vesiculated, and highly fractured lava flows. The fractures are highly penetrative and interconnected throughout the basaltic unit which favors groundwater circulation and storage in most places (Gebeyehu et al., 2022;Razak et al., 2020). However, the vesicles and fractures of the basalt at places are filled with secondary minerals (such as calcite and zeolites), and impervious weathered clay or laterites which may hinder groundwater movement and limit its productivity (Gebeyehu et al., 2022). Based on its high permeability and productivity, a higher normalized weight (0.54) was assigned to this unit ( Fig. 6a and see Table 4). The recent superficial alluvial sediments and fans occupy limited areas in the northern and eastern regions, at the foot of hills and rift scarps. The sediment is dominantly composed of highly porous gravel and sand. This lithological unit is more than 80 m thick. Although the alluvial deposits form an unsaturated aquifer, they have high groundwater storage and transmitting potential to the underlying basaltic rocks (Gebeyehu et al., 2022;Razak et al., 2020). It is also highly productive and favorable for shallow groundwater development. Accordingly, a weight of 0.28 was given to this unit ( Fig. 6a and Table 4).
The pre-rift volcanics occupying the western elevated ridges and hills, and the Quaternary volcanics in most of the southern low-lying areas are mainly composed of acidic rocks of trachytes and rhyolites associated with pyroclastic deposits and minor basalts  (Gebeyehu et al., 2022;Razak et al., 2020). The groundwater potential in the fractured and weathered sections of the acidic and basaltic rocks is relatively high. However, the presence of interlayered impervious pyroclastic deposits may decrease their permeability, and as they occupy elevated grounds, they may act as local barriers to groundwater flow, which can limit their productivity in most places (Gebeyehu et al., 2022;Razak et al., 2020). Therefore, a moderate normalized weight of 0.12 was assigned to this unit. The metamorphic basement rocks of various gneissic and granitic rocks mostly outcrop along the eastern peripheral regions of the catchment. They have relatively little hydrogeological importance and influence on groundwater availability as they act as a regional aquiclude except where they develop secondary porosity through fractures/joints and on thick weathered regolith deposits (Gebeyehu et al., 2022;Razak et al., 2020). Because of its low productivity and groundwater prospect, a lower weight of 0.06 was assigned to this unit ( Fig. 6a and Table 4).

Lineament density
The Bulal catchment is highly affected by lineaments and/or fractures because of rift-forming volcanotectonic activities. NW-SE-and N-S-trending lineaments are prominent in the catchment (Gebeyehu et al., 2022;Razak et al., 2020). The lineament density (LD) in the catchment varies between 0 and 1.5 km/km 2 (Fig. 6b and Table 4). For most of the catchment area, the LD is high, varying from 0.5 to 1.5 km/km 2 . Although the lineaments are widespread across the catchment, the volcanic terrain has relatively higher LD than areas underlain by crystalline basement rocks. The localities with higher LD are regarded as having good prospects for groundwater storage and movement compared to those with the lowest LD. Accordingly, a higher weightage factor of 0.46 was assigned to localities with high LD and low normalized weight (0.09) to areas with low LD (Fig. 6b and Table 4).

Land use/land cover
According to the LULC classification of the Bulal catchment, dense bushes and shrubs are the dominant types that cover most of the western, northern, and southern parts followed by scattered bushes and shrubs ( Fig. 6c and Table 4). Settlements, dense forests, rocky, and marshy areas, and cultivated, grass and bare lands occupy very limited localities of the catchment. Together with the moderate annual precipitation (mean = 608 mm/year), the distribution of the LULC is expected to enhance the groundwater recharge depending on the underlying soil and geologic conditions (Olutoyin et al., 2013). Accordingly, areas covered by dense vegetation of trees and shrubs were given the highest weights of 0.27 and 0.25, respectively ( Fig. 6c and Table 4).

Rainfall
The mean annual rainfall of the Bulal catchment ranges from 359 to 858 mm/year. The western and eastern highlands receive a higher mean annual rainfall of 735 mm due to localized orographic effects (Anders et al., 2006;Colberg & Anders, 2014;Garreaud et al., 2016), while the southern lowlands receive a mean annual rainfall of 468 mm/year ( Fig. 6d and Table 4). Areas with high and moderate annual rainfall have weightage factors of 0.46 and 0.29, respectively, signifying very good and moderate groundwater potential. On the other hand, areas with the lowest annual rainfall are given a 0.09 normalized weight, suggesting low groundwater potential.

Slope
The slope thematic layer (Fig. 6e) shows that the land gradient of the catchment varies between less than 2% and more than 31%. Most of the low-lying rift floor along the central and southern regions is characterized by flat to gently sloping land with slopes < 7%. On the other hand, the elevated areas with slope gradients > 16% (steep to very steep) constitute most of the western volcanic ridges and the northern and southeastern basement terrain. Accordingly, based on the land slope's influence on infiltration and groundwater recharge, areas with slopes of < 2% (i.e., nearly flat surfaces) were rated higher in terms of groundwater availability (Rajaveni et al., 2015;Strahler, 1964), and assigned a normalized weight of 0.47 compared to areas with slopes > 31% with weightage factor of 0.04 ( Fig. 6e and Table 4).

Topographic variability
In the catchment area, four major topographic variability classes (landforms) were identified and demarcated based on their elevation variation and hydrogeological importance ( Fig. 6f and Table 4). Most of the catchment is characterized by lowlands and plains with elevations ranging between 702 and 1250 m a. s. l. These landforms generally represent the low-lying rift floor covered by volcanic rocks and unconsolidated sediments. On the other hand, uplands and inselbergs representing the chain of high elevated ridge crests along the rift escarpment and hills constitute most of the northern and southeastern basement terrain. The plains and lowlands of the catchment are the most favorable landforms where most structurally controlled runoff accumulates and preferential indirect recharge occurs from the stream beds into the subsurface (Razak et al., 2020). Therefore, these landforms were assigned higher normalized weights (0.41 and 0.48, respectively). The uplands and inselbergs were given 0.11 and 0.0 weights, respectively, as they are considered to be unfavorable land features for the availability of groundwater potential and unsuitable for groundwater development.

Depth to groundwater level
The depth to groundwater level of the Bulal catchment varies between 0 and 159 m below ground level (b. g. l.) with a mean of 70 m b. g. l. (Fig. 7a and Table 4). The spatial map (Fig. 7a) indicates that the shallower groundwater system is mainly confined to the regolith developed over the basement rocks and unconsolidated sediments in the eastern parts, while the relatively deeper groundwater is developed on the fractured basaltic rocks in the central parts of the catchment (Gebeyehu et al., 2022;Razak et al., 2020). Due to the arid condition of the catchment and high evapotranspiration rate, the shallow groundwater system is an unreliable resource and tends to be vulnerable to seasonal water level fluctuations due to the impacts of climate change on groundwater storage (Gribovszki et al., 2010;Lautz, 2008;Martinet et al., 2009). As a result, most shallow wells in the catchment are found to be dry and the unconsolidated sediment aquifer becomes unsaturated during the dry seasons (Razak et al., 2020). Therefore, areas with shallow groundwater levels are assigned the lowest weight (0.09) compared to the areas with deeper groundwater levels (0.46) ( Fig. 7a and Table 4).

Drainage density
An area with a very high drainage density represents more closeness of drainage channels and high runoff, while a lower DD indicates lower run-off and a higher probability of recharge and groundwater potential (Olutoyin et al., 2013). Most of the drainage originates from the volcanic ridges and inselbergs in the west, and the basement ridges in the northern and eastern parts of the catchment. Figure 7b shows that the DD of the catchment ranges from < 0.0001 to 0.0011 km/km 2 . Areas with the highest DD are weighted relatively lower (0.09) compared to very low drainage density areas, which are given a high weightage factor (0.46). However, the generally moderate to high drainage density in the catchment implies low or moderate infiltration and recharge potentials.

Soil texture
Fine-grained soils limit infiltration due to their lower permeability, unlike coarse-grained soil materials where water can infiltrate easily because of high permeability (Olutoyin et al., 2013). In this study, six major soil texture units were identified: clay, clayey loam, loam, sandy-clayey loam, loamy sand, and sandy loam ( Fig. 7c and Table 4). As shown in Fig. 7 c, sandy loam is the dominant soil texture covering an extensive area of the Bulal catchment. Due to its higher sand content (coarsegrained materials) and permeability, a higher weightage factor was given to the sandy loam soil units (0.36) compared to areas covered by massive, unfractured rocks and clayey soil with a normalized weight of 0.02 and 0.04, respectively ( Fig. 7c and Table 4).

Topographic wetness index
In the catchment area, the TWI value varies between 4 and 20 with a mean of 9.5 ( Fig. 7d and Table 4). The slightly elevated areas with gentle sloping drainage systems where runoff waters from the highlands accumulate have a relatively higher TWI value. On the other hand, the well-elevated areas with steep sloping drainages have relatively lower TWI or surface accumulation. Therefore, an area with a higher value of TWI which indicates the presence of a relatively high soil moisture accumulation tends to enhance groundwater recharge (Naghibi & Dashtpagerdi, 2017;Owolabi et al., 2020) and is assigned a high weight (0.46), whereas areas with the lowest TWI value are of low groundwater prospect and given a low weight of 0.095 ( Fig. 7d and Table 4). Sensitivity analysis of the GWPZI map A sensitivity analysis was performed to determine the influence of each input parameter in delineating the GWPZ map of the Bulal catchment. Accordingly, the sensitivity index (SI) shows that geology (GG) with a mean SI value of 1.7% and topographic wetness index (TWI) with a mean of -1% are the most and least sensitive parameters in delineating the GWPZI map, respectively (Table 5). In other words, the geological layer has a significant impact on the spatial distributions of groundwater potentiality in the catchment. As a result, the removal of geology reduces the area coverage of the high and very low zones by ~ 7% each, while increasing the moderate zone by ~ 14%. The topographic wetness index is the least influential factor on the catchment's groundwater potential, and removing it from the delineation of the GWPZI map does not affect the spatial extents of the potential zones (Table 5).

Classification of groundwater potential zones
The GWPZI map of the Bulal catchment shows four distinct potential zones (i.e., very low, low, moderate, and high) ( Fig. 8 and Table 6). The potential map provides a quick perspective on the storage and availability of groundwater resources in the catchment. The GWPZI map indicates that the western and eastern peripheral regions have low to very low groundwater potential, while most of the northern, central, and southern low-lying rift floor areas generally exhibit moderate to high potential. The map shows a highly productive deeper aquifer system of basaltic rocks, dominantly the Bulal basalts. Among the major lithologic units in the catchment, the Bulal basaltic flow overlain by thick sediments in places forms an extensive and highly to moderately productive major TBA system, as revealed by the GWPZI map (Fig. 8). The alluvial sediments in fans along the feet of rift scarps and ridges and the weathered regolith deposits over the crystalline basement also have the potential to store groundwater at shallow depths (Gebeyehu et al., 2022;Razak et al., 2020). Weathered and fractured acidic volcanic rocks can also form deeper potential aquifers along lineaments/fractures and fault planes.  On the other hand, the low and very low GWP areas show limited aquifer capabilities of the metamorphic basement rocks associated with regolith and minor pyroclasts (Gebeyehu et al., 2022;Razak et al., 2020). The GWPZI map indicates that the distribution of groundwater potential is a function of the lithological input, while lineament density, slope, topographic variability, and groundwater depth also play a significant role. Areas underlain by the Bulal basaltic flows associated with acidic volcanics and alluvial sediments along the northern, western, central, and southern parts forming plains with flat to gently rolling slopes, with high lineament density, and covered by bushes and shrubs have high to moderate groundwater potential. On the other hand, areas underlain by the metamorphic basement rocks covered by regolith deposits in the eastern part of the catchment forming rugged and highly elevated topography with relatively steep slopes, high drainage densities, low topographic wetness index, and lower lineament densities, exhibit very low to low shallow groundwater  potential. Moreover, the predominantly ridge-forming, steeply sloped felsic rocks (trachytes and rhyolites) with minor basalts in the western part of the catchment show low to very low groundwater potential. In short, high to moderate groundwater potential in the Bulal catchment is associated with a combination of the following factors: (i) fractured and weathered basaltic rocks, (ii) high lineament density, (iii) dense vegetation cover, and (iv) low gradient plains.

Validation of the GWPZI map
The accuracy of the GWPZI map was evaluated by superimposing yield data from 40 boreholes within the catchment. Well yields varying between 8 l/s and 71 l/s with a mean of 25 l/s and 3.1 l/s and 8 l/s with a mean of 5 l/s are attributed to the high and moderate potential zones, respectively, which is also in agreement with the high potential zones identified by OWWDSE in 2017 through conventional investigation (see Fig. 4) (Razak et al., 2020). On the other hand, well yields varying from 1 to 3 l/s (average 2.6 l/s) and yields below 1 l/s (average 0.8 l/s) fall in the low and very low potential zones, respectively ( Fig. 8 and Table 6).
The well yields also show clustering with the lithologies (Fig. 9). Most wells with yields ≥ 8 l/s (high yield) are mainly on the Bulal basaltic flows and associated alluvial sediments. On the other hand, most wells within the regolith developed over the metamorphic basement, and pyroclastic deposits have low (1-3 l/s) and very low (≤ 1 l/s) yields. The moderate yield (3.1-8 l/s) wells mostly plot on the fractured basaltic and felsic rocks covered by some alluvial sediment. The well yield distribution very well validates the GWPZI map of the catchment.
The validation clearly justifies the efficiency of the integrated RS and GIS-based overlay analysis technique employed in the delineation of the groundwater potential zones. This technique is a useful modern approach for proper groundwater resource investigation and development. However, it should be noted that the developed GWPZI map can only provide a quick perspective guide for the purpose of regional groundwater exploration and development. Detailed ground-truthing and further verifications should be considered for sitespecific groundwater investigation and development.

Conclusions
Using Saaty's AHP approach as an MCDM tool, a composite GWPZI map is produced and delineates the groundwater potential zones of the Bulal catchment. The work shows the following: 1. There are four groundwater potential zones in the catchment: high, moderate, low, and very low. The high and moderate groundwater potential zones represent 27% and 20% of the total catchment area, respectively, while low and very low potential zones together account for approximately 53% of the total area. 2. The geological (lithology, structure density) and geomorphological (topography, slope) features are the most dominant influencing factors in the distribution of groundwater potential in the catchment, as also shown by the sensitivity analysis, where the geology is the most influencing factor while soil texture and topographic wetness index are the least sensitive. 3. Areas underlain by the transboundary Bulal basaltic rocks in most of the central and southern parts of the catchment are characterized by high to moderate groundwater potential zones, while areas with low and very low groundwater potential are typically associated with the regolith deposits over the metamorphic basement rocks in the eastern periphery of the catchment. 4. High to moderate groundwater potential zones are associated with a combination of fractured and weathered basaltic rocks, high lineament density, dense vegetation cover, and low slope gradients. 5. The accuracy of the GWPZI map is well validated by the groundwater well yields from the catchment where areas with high and moderate groundwater potential are characterized by average well yields of 25 l/s and 5 l/s, respectively, while those with average yields of 2.6 l/s and 0.8 l/s represent low and very low potential zones, respectively. 6. The integrated RS and GIS-based overlay analysis technique used to delineate the groundwater potential zones is efficient and could be a useful technique for proper regional groundwater exploration and development. Our method could be reliably used to delineate GWPZs in similar semiarid regions. The method will be useful in locating relatively shallow GWPZs of transboundary aquifers across the catchment. The results will also help to locate promising borehole drilling sites and serve as a quick guide for efficiently planning, managing, and developing the catchment's groundwater resources. 7. However, this method has limitations in accurately identifying deeper groundwater system. The GWPZs map produced should therefore be used for regional studies, in this respect. For site-specific deep groundwater system delineation studies, the produced GWPZI map must be verified and accompanied by detailed ground-truthing investigations. Furthermore, it is recommended that water quality be considered as an input parameter for future GWPZ delineation processes.