Deep low-salinity groundwater in sedimentary basins: petrophysical methods from a case study in Somalia

Oil and gas exploration data for the Northern Somalia and Horn of Africa region have been utilised for deep freshwater resources exploration. This unique geophysical and petrophysical dataset is integrated into a regional hydrogeological study and it allows for targeting of unmapped and unknown deep aquifers that would otherwise be beyond the conventional reach of the groundwater sector. The database includes open-hole well logging surveys (resistivity, gamma ray, and spontaneous potential), mud logging and drill stem test (DST) data. The results from the data analysis are combined with the geological interpretation to assess the freshwater potential of each mapped hydrostratigraphic unit in the region. The results highlight the presence of several previously unknown deep low-to-medium salinity (<9 g/L total dissolved solids (TDS)) aquifers within the Jurassic to Eocene units, including the Adigrat, Hamanlei, Gabredarre, Gumburo, Jessoma, and Auradu formations. With tested intervals of water containing less than 3.8 g/L total dissolved solids, the Auradu, Jessoma, and Gumburo formations are the most promising deep freshwater aquifers. Additional results on the analysis of groundwater parameters, such as hydraulic head, provide the basis for further groundwater modelling. The results show that the method has good potential, in particular for use in drought-stricken areas of arid regions. Ultimately, the results from this region have global significance, as the method can provide a new means of boosting fresh groundwater resources in water-poor regions, and supporting sustainable development and utilization of the resources in the medium and long term.


Introduction
Deep stacked aquifers represent a valuable resource of water, especially in arid regions, where shallow aquifers are usually scarce, intermittent, overexploited and/or contaminated (Margat and Van der Gun 2013). However, the characterization of deep aquifers requires time-consuming and expensive data acquisition and drilling methods; therefore, their evaluation is limited to a few, very productive aquifers worldwide. Examples include the North Africa Nubian Sandstone and North Western Sahara Aquifer Systems (El-Rawy and De Smedt 2020; Gonçalvès et al. 2013;Mazzoni et al. 2018;Sultan et al. 2007), reaching depths of 3,500 m; the Arabian Platform aquifer system, more than 450 m deep (Abotalib et al. 2019), the Great Artesian Basin in Australia, up to 3,000 m deep (de Rijke et al. 2016;Fensham et al. 2021;Habermehl 2020), and the Dogger and Albian Aquifer of the Paris Basin (Innocent et al. 2021;Marty et al. 2020;Matray et al. 1994) reaching depths of 1,900 m bsl. In many of these aquifers, exploitation of groundwater, geothermal, hydrocarbon resources and CO 2 storage often coexists.
One possible solution to the problem of data availability resides in the vast geological and geophysical database created by the oil & gas (O&G) industry, which can be repurposed for hydrogeological studies. Due to the depths of the targeted O&G reservoirs, which are greater than conventional aquifers exploited for groundwater resources, these data provide a deeper subsurface reach compared to conventional hydrogeological datasets. Additionally, due to decades of O&G exploration, seismic and well data have already been acquired in most sedimentary basins worldwide, both onshore and offshore (Bertoni et al. 2020). Although most of these data are proprietary and confidential, they are becoming increasingly available for academic studies, and appear in many publications.
To address data limitations in exploring for deep fresh and low-salinity water aquifers, this study uses a data analysis approach, which brings together methods developed and established for both hydrogeology and O&G studies. In this study, results are presented for the analysis of a suite of geological and petrophysical parameters from wells drilled up to 3 km depth, which helps determine the physical properties of the aquifers and the related aquitards. The most recent well data include rock samples (from cuttings and cores), pressure measurements (DSTs and MDTs), wireline logs (spontaneous potential, resistivity, gamma ray, sonic, density and neutron porosity) and fluid geochemistry (salinity, Fig. 1 a Geological map of the Horn of Africa, compiled and modified from several sources (Abbate et al. 1993;Azzaroli and Merla 1957;Kazmin 1972;Merla et al. 1979) and the Geological Map of Kenya obtained from the National Atlas of Kenya (Githaiga et al. 2021). The Daban Formation (Oligocene) is only present in the north of Somalia and therefore not included in the general stratigraphic column. Transboundary formation transitions can be abrupt, due to no entire alignment between different geological maps; b Location and detailed study area map (SAM). The Nogal and Darror basins are highlighted with white semitransparent shading, and red lines represent main basin faults (Abbate et al. 1993); c Regional stratigraphy and hydrogeological data. The current exploited reservoirs and aquifers reported in the literature are highlighted with light blue dots. The newly reported deep aquifers (results from this study) are highlighted with light blue boxes. The potential aquitards are highlighted with brown boxes; d Interpreted seismic section showing the main stratigraphic units, their thickness in Two-Way-Travel-Time (TWT), the smaller-scale faults and the main bounding fault in the northern part of the Nogal basin (modified from Ali and Lee 2019a, b;Ali and Watts 2016). Note that, in this section, the Bihendula Formation corresponds to the upper Hamanlei/lower Uarandab Formation (Fig. 1c) organic geochemistry). A suite of petrophysical methods was used that includes (1) the characterization of potential aquifer zones applying the Spontaneous Potential method and Archie's law, and the identification of aquitards using Gamma Ray (GR) and lithology logs; (2) the estimation of hydraulic properties (permeability, hydraulic conductivity, porosity); and (3) the identification of potential crossflow between stacked aquifers, through the analysis of the hydraulic head from pressure tests. The multidisciplinary approach used in this study is included in a larger project which combines GIS technologies and remote sensing, threedimensional (3D) visualisation and processing tools, finite difference method modelling, development of high-resolution recharge algorithms, analysis and processing of geophysical data, in cooperation between the private sector, academia and UN-institutions (FAO-SWALIM; Quiroga et al. 2022).
This approach was applied to the Northern Somalia region, in East Africa (Fig. 1), which was selected for a combination of reasons: (1) the recognised need for new water resources; (2) the vast upstream recharge areas; and (3) the high geological potential for deep groundwater resources. The scarcity of water resources in large parts of East Africa is caused by low precipitation, high evapo-transpiration, and contamination of shallow resources (Conway et al. 2015;FAO-SWALIM 2012;Godfrey et al. 2019;MacDonald et al. 2012;O'Dochartaigh et al. 2016;Taylor et al. 2013). These factors affect the region in different ways and cause imbalance in adequate resources across the region (Edmunds 2012;Hughes 2019) with saline intrusion, demographic pressure and environmental change particularly affecting coastal areas (Comte et al. 2016;Steyl and Dennis 2010;Walraevens et al. 2015).
Extractable groundwater resources in Somalia have traditionally been limited to aquifers less than 200 m depth (FAO-SWALIM 2012). Exploration boreholes in the area may reach depths of up to ca. 300 m, mainly focusing on exploiting easily identifiable shallow aquifers. Conversely, the extent and areal distribution of deeper aquifers is at present unknown. However, wells drilled by the O&G industry provide an indication that deep freshwater aquifers exist in the region (Ali and Lee 2019a, b). In Tanzania, a deep aquifer exploration project presented a search model based on the integration of hydrogeological data with geophysical and geological data acquired by the O&G industry (Moe et al. 2017;Ruden 2007Ruden , 2009). The resulting modelling of large regional aquifers has shown potential for additional freshwater resources at previously unexplored depths (Moe et al. 2017;Ruden 2007Ruden , 2009. Other studies highlighted how O&G data, combined with geological, geophysical and engineering surveys, can be used to interpret geology, structure and stratigraphy of sedimentary basins, and therefore identify possible deep onshore and offshore aquifers and their capacity (Bertoni et al. 2020;Micallef et al. 2021;Nahm 2002;Varma and Michael 2012). This well dataset is focused in particular on the Darror and Nogal sedimentary basins (Fig. 1), a water-stressed region where the currently exploited shallow aquifers are largely contaminated by salts related to leaching of gypsiferous formations embedded within a dominantly carbonate sedimentary series. The main recharge of these aquifers in northern Somalia is presumed to come from direct precipitation that occurs in the local mountainous terrain (such as the Northern Somalia Highlands, Karkaar and Cal Miskaat). The main contributor to shallow recharge in the valleys is runoff water from the short-lasting thunderstorms (FAO-SWALIM 2012). Previous hydrogeological studies of Somalia (Faillace 1983;FAO-SWALIM 2012) focus on the presence of aquifers in the formations of the Nogal group, the Karkar and Auradu formations (Fig. 1), which are penetrated with 100-300-m-deep wells. These water resources in the north of Somalia can be heavily contaminated by human and agricultural activities, while weathered limestones and karstification processes contribute to water infiltration into deeper formations (FAO-SWALIM 2012). Shallow coastal and alluvial aquifers are consequently not included in the survey but may play a role in the recharge of the upper reaches of the catchment areas.
The Horn of Africa receives a large part of its groundwater through recharge from the Ethiopian highlands, in the west of the study region. A gradient between western summit in Ethiopia and the Indian Ocean is typically 2.5 km/1,000 km. A number of high-yielding freshwater springs have been reported along this route and are currently used for potable water and agriculture. Springs are generally connected to faults, and possibly augmented by extensive karst developments along these, as indicated by numerous reports in oil well drilling reports about lost circulation problems (Faillace 1983). This analytic approach is applicable to other comparable settings worldwide. While these are tested techniques in the fields of hydrogeology and petroleum geology, the opportunity to combine them in the detection of multiple, stacked aquifers provides new insights for well planning in areas with limited data availability, such as the Horn of Africa. The results of this study aim to support sustainable development and utilization of water resources in the Horn of Africa, and can be used to guide the near, medium, and long-term development of the deep aquifer potential in the region.

Regional geology
The geological and hydrogeological setting of the northern part of Somalia is closely linked to the opening of the nearby Afar Rift, and of the Aden Gulf, which was initiated in the Tertiary (Bosellini 1992;Carmignani et al. 2002;Fantozzi and Sgavetti 1998). Structurally, this region belongs to a faulted Oligo-Miocene extensional system, which reactivated pre-existing Mesozoic faults (Ali and Lee 2019a, b;Ali and Watts 2021, 2016Bellahsen et al. 2006Bellahsen et al. , 2013Leroy et al. 2012). The Nogal and Darror (alternative spellings: Nugal, Dharoor, Darroor) basins are part of this system (Fig. 1). Their geological history is punctuated by rifting stages followed by quiescence (Ali and Watts 2016;Bosellini 1992;Fantozzi and Ali Kassim 2002;St John 2016). These basins are characterised by WNW-ESE-striking half-grabens, which formed mainly during Late Cretaceous-Paleogene and Oligocene-Miocene intracontinental rifting (Ali and Lee 2019a, b;Fantozzi and Ali Kassim 2002).
The total sedimentary cover of these basins is over 7 km thick (Nogal basin). Significant thickness variations are observed along the larger fault systems, such as the northern bounding fault of the Nogal basin ( Fig. 1; Ali and Lee 2019a, b). In the Darror basin, extensional NW-SE-striking faults have often thrown greater than 1,000 m (Fig. 1). These faults have divided the whole region into horsts, vast rapidly eroding plateaux and narrow grabens that received syn-tectonic sediments of the Oligo-Miocene depositional cycle. Plio-Quaternary basalt magma flows occur in western and central northern areas (Fantozzi and Ali Kassim 2002).
Successions of two main depositional cycles connected with the two rifting stages crop out in the area ( This geological structure influences the distribution of the aquifers in the subsurface, with the Mesozoic to early Paleocene formations virtually unexplored for fresh groundwater. The pre-Palaeozoic and Palaeozoic basement is composed of metamorphic rocks and granitoids (Fantozzi and Ali Kassim 2002;Piccoli et al. 1986). The Adigrat Formation is a fine to coarse-grained feldspathic sandstone with common limestone inter-beds from Early Jurassic to Late Triassic (Fig. 1). The Hamanlei Formation was deposited in the Middle to Late Jurassic and consists primarily of limestone with some intercalations of shale and sandstone. The Gabredarre Formation, also called Dawan, is generally found in northern Somalia as a massive anhydrite with inter-bedded limestone that deposited in the Late Jurassic. The Gumburo Formation was formed in the Late Cretaceous, and in the east part of northern Somalia consists mainly of highly fractured limestone. The Jessoma Formation is the shallower sandstone formation in both basins (Fig. 1).
The most recent sedimentary units (Nogal group, Karkar Formation and upper Auradu Formation) are currently exploited for groundwater resources (Fig. 1). The Nogal group encompasses Neogene and Quaternary deposits. It is composed mostly of sandstones, red soil, continental carbonate crust, limestones, and conglomerates. Boreholes connected to Quaternary deposits have usually a low yield (less than 61 m 3 /day; FAO-SWALIM 2012). The Karkar Formation is composed of bedded and marly limestones, and white marls. The formation thickness varies between 150-450 m and water quality from springs and wells is usually good with an EC value of 1,490-1,800 μS/cm (FAO-SWALIM 2012), that is 0.954-1.152 g/L total dissolved solids (TDS), using 1 EC (μS/ cm) = 0.00064 TDS (g/L; at 25 °C) conversion. The upper part of the Auradu Formation consists of massive limestones alternated with chalky and gypsiferous beds and calcareous shales. Unless the overlying Taleh aquifer (mainly evaporites) is present, groundwater from Auradu limestones shows low mineralization with an EC value generally below 1,000 uS / cm (FAO-SWALIM 2012), that is 0.640 g/L TDS, using 1 EC (μS/cm) = 0.00064 TDS (g/L) (at 25 °C) conversion.
O&G studies have evidenced that beyond the most recent aquifers, good reservoir properties such as porosity and permeability characterise various deeper Upper Jurassic-Miocene formations, in particular the continental Jessoma Sandstones (Ali and Lee 2019a;Purcell 1979;St John 2016). This formation is generally overlain by shallow-water marine deposits of the Auradu Limestones, Taleh Evaporites (mostly composed of gypsum and anhydrites) and the Karkar Formation (Ali and Lee 2019a, b;St John 2016). The evaporites and the tight deep-water carbonates are potentially a good seal for underlying fluids (Ali and Lee 2019a, b).

Well data
Collectively, seven wells drilled for O&G exploration in Somalia form the core data for the petrophysical analyses of this study ( Fig. 1; Table 1). Logs were available for four of the wells-Shabeel-1 and Shabeel North-1 in the Daroor basin, and Kalis-1 and Nogal-1 in the Nogal basin. For wells Buran-1 (Daroor basin), Yaguri-1 and Burhisso-1 (Nogal basin), only well reports were available (Table 1). Depths reported in this section are given in measured depth (MD) from the kelly bushing (KB), unless otherwise specified.

Methodology
The characterization of rock and fluid properties for oil exploration is only possible through a deep understanding of the behaviour of water in rock. In the water field, rock characterization and the concept of good reservoir properties remain the same as for the oil industry; gamma ray, lithology logs, photoelectric effect, and velocity logs are interpreted to determine the aquifer system. However, as the fluid of interest in the context of hydrogeology is different, the interpretation of measurements that are directly affected by the formation water salinity content such as spontaneous potential (SP) and resistivity require a different approach. In this study, an assessment of the rock properties is first performed to identify good potential reservoirs, followed by an evaluation of the salinity of the formation water within the selected potential reservoirs. Finally, formation pressure is converted to equivalent hydraulic head to identify potential cross flow between the aquifers cut across by the same well.

Identification of potential aquifers and aquitards based on rock quality
The identification of potential aquifers is initially done based on rock properties. Sandstone and carbonate provide the best settings for aquifers systems in Northern Somalia and Somaliland. Formations with high content of shale are considered potential aquitards. Gypsum and anhydrite rocks are classified as potential aquifers or aquitards as they can be both depending on the porosity and permeability. However, in more shallow aquifers, water from gypsum and anhydrite formations have been found to be highly mineralized (FAO-SWALIM 2012).
Lithology logs were available for all the wells and were contrasted with the gamma ray log in each case. A low to relatively low gamma ray response is expected in carbonate and sandstone formations respectively, and higher values in more shale-rich formations. Figures 2 and 3 summarize the results of this step for the Daroor and Nogal basins respectively.
The capacity of the rock to store fluids (porosity), and its capacity to allow fluid to flow through it (permeability) are also considered in the evaluation of the potential aquifers. For this study, porosity data were available from the logs and reports for most of the potential aquifers (Figs. 2 and 3). Permeability data were available only for some formations ( Table 2).

Identification of potential aquifers based on formation water salinity
In Somalia and Somaliland, shallow water boreholes (100-400 m) are tapping water with salinity content up to 46 g/L TDS (FAO-SWALIM 2012). At the same time, O&G exploration wells have recovered pore water with salinities varying from Table 1 Available data for the study. Spontaneous potential (SP), gamma ray, resistivity, density, and porosity logs were available for four boreholes Pressure tests are from drill stem tests (DSTs) or repeat formation tests (RFTs). Formation water analysis is the salinity reported in the DST or core analysis Available Available Available <0.3 g/L up to 60 g/L TDS at depths >400 m, thus deeper than the traditional water boreholes. The World Health Organization has not established official health-based guideline values for total dissolved solids (TDS) in drinking water. However, they state that water containing TDS concentrations below 1 g/L is usually acceptable for human consumption.
Two methods are generally used by the O&G sector to determine the salinity of formation water: (1) chloride titration with silver nitrate, to determine the chloride content of the solution; and (2) directly measuring the resistivity of the solution at a given temperature to obtain the equivalent NaCl concentration using the Schlumberger Chart Gen-9. The latter expresses formation water salinity in the form of sodium chloride (NaCl) concentration, i.e., the total ionic conductivity of the solution is put in terms of sodium chloride concentration. The salinity in NaCl from oil analyses is referred to in this study as "NaCl TDS-equivalent" and is comparable to the TDS. A more conservative approach to estimating the total dissolved solids (TDS in mg/L) is by converting formation water resistivity (R w ) at 25 °C, into electrical conductivity (EC in μS/cm) as follows.
(1) EC = 10,000∕R w Depending on the method used to estimate salinity, this study refers to three types of salinities: -Cl: chloride content expressed in g/L, unless specified.
Determined with the chloride titration method. -NaCl TDS-equivalent: the total electrical conductivity of the formation water, assuming that the solution contains NaCl only. Determined using the Schlumberger Gen-9 chart and expressed in g/L, unless specified. -TDS: total dissolved solids estimated from the resistivity of the formation water, using a factor of 0.64 times the electrical conductivity. Expressed in g/L, unless specified.
For the purposes of this study, formation water with less than 1 g/L TDS is considered of low salinity, and water with salinity values below 9 g/L TDS is considered of interest for low-to-medium salinity water aquifers investigations, due to the potential for livestock and industrial uses, and for more affordable desalination costs. The threshold value of 9 g/L TDS was selected to best cover the most representative of the lower spread of the observed salinity values in the oil wells and is below the marginal limit for camels proposed by UNICEF (FAO-SWALIM 2012) of total dissolved solids of 10-15 g/L. However, further analysis of water quality to determine other or potentially harmful dissolved solids is required during the exploration of these deep aquifers.
As expressed in the preceding, the resistivity of the formation water (R w ) is a direct indicator of its ionic content. R w can be directly measured from formation tests or indirectly determined from resistivity measurements by applying Archie equation, and/or from the spontaneous potential log by applying the SP method.
R w derived from resistivity readings With few exceptions, electricity can pass through a formation only because of the ionic content of the water it contains. Subsurface formations have finite, measurable resistivity because of the water in their pores or absorbed in their interstitial clay (Schlumberger 1991). As opposed to formations saturated with saline water, hydrocarbon-filled formation resistivity readings are  Daroor basin, K (mD) Nogal basin, K (mD) Upper 280  generally higher. In a noninvaded, water-bearing zone with no clay or shale content, the Archie equation (Archie 1942) can be applied to determine the formation water resistivity. Whereby: S w is the water saturation n is the water saturation exponent a is an empirically derived constant referred to tortuosity of the formation R w is the resistivity of the formation water, Ω.m ∅ is the porosity m is the cementation factor R t is the true resistivity of the formation Ω.m In a 100% water-saturated zone, the equation is reduced to: The typical values applied are: Carbonates: a = 1, m = 2 Consolidated sands: a = 0.81, m = 2, and unconsolidated sands a = 0.62, m = 2.15 Porosity ( ∅ ) can be obtained from neutron porosity logs or derived from density or sonic logs. R t can be read directly from the resistivity logs. R w derived from spontaneous potential log Spontaneous potential (SP) measures the difference between the electrical potential of a movable electrode in the borehole and the electrical potential of a fixed surface electrode (Schlumberger 1963, p 3-1). The movement of ions, which causes the SP phenomenon, is only possible in formations having a certain minimum permeability (Schlumberger 1991). Therefore, this log is normally used to differentiate potentially porous and permeable reservoir rocks, correlate beds, and permit the determination of formation water resistivity.
For this electric potential to occur, three conditions are required: (1) a conductive fluid is present in the borehole, (2) a sandwich of a porous and permeable bed between two formations of low-porosity and very-low-permeability, and (3) a difference in salinity or pressure between the borehole fluid and the formation fluid. The first condition is already met as the seven wells were drilled utilizing conductive mud. For the second condition, the volume of shale (Vshale) is calculated, zones with a high volume of shale are considered of very low permeability, and zones with a low volume of shale are contrasted with the lithology data to validate that it is a potentially permeable formation. Finally, for the third condition, in the absence of a high-pressure difference, whenever a shift of the spontaneous potential log with respect to its zero or control point is observed, it is assumed to be due to a difference in salinity between the formation and borehole fluids . An example of this methodology applied to identify low-salinity pore water in hydrogeology has been presented by previous authors (Varma and Michael 2012) for an area in southern Australia.
To identify pore freshwater, the following methodology was applied to each of the wells: 1. Vshale gamma-ray-based was calculated for each well using the linear method, and lithology reports were utilized to identify potential reservoirs in between very low permeability layers.

A shale baseline was defined based on the Vshale value.
Then, the spontaneous potential log was shifted by setting the log to zero at the low permeability layers, as in a low permeability layer no ion interaction between the borehole and the formation is expected. 3. The response of the spontaneous potential log to the baseline shifting is key for determining the zones with formation water with less salinity than the mud filtrate. When the mud filtrate is conductive, a positive deflection of the SP line indicates less saline formation water, and a negative deflection indicates more saline formation water (Schlumberger 1963).
In this study, the SP method was used for a quick characterization of zones where the salinity of the formation water is lower than that of the drilling mud filtrate. A positive deflection (to the right side of the shale baseline) indicates formation water resistivity (Rw) with salinity less than of 15-25 g/L NaCl TDS-equivalent, which is the salinity of the mud filtrate in the studied wells. The higher the deflection, the less saline the formation water is (see Fig. 4).

Formation water salinity from DST and cores
The identification of potential low-salinity aquifers derived from the SP method and Archie equation was contrasted with the formation water analysis in the drill stem tests (DST), cores and repeat formation tester (RFT) when available. Figure 4 shows the results of the applied methodology in Shabeel North-1. Table 3 summarizes the results for all wells.

Hydraulic head from pressure tests
The movement of groundwater is highly controlled by the hydraulic gradient. The hydraulic head is the height of the water column that the formation can withstand at a given point due to its pore pressure, and it is measured above a datum plane such as sea level. In hydrogeology, the standard method for measuring pore water pressure employs a piezometer, which measures the height to which a column of the liquid rises against gravity (Lohman 1979). In the O&G industry, the most common testing technique is to isolate the region of interest with temporary packers and opened valves to produce the reservoir fluids through the drill pipe and allow the well to flow for a time (Aadnoy and Looyeh 2019). The use of pore pressure recorded in O&G wells to derive equivalent hydraulic head has been previously documented (Varma and Michael 2012). While in hydrogeology the hydraulic head is measured directly with a piezometer, the pore pressure measured in the O&G wells can be converted to an equivalent hydraulic head measured from the sea level, using the following equation (Lusczynski 1961): where P is the pore pressure, ρ is the density of the water, g is the acceleration due to gravity, and z is the elevation at the point of measurement of the pressure measured from the sea level. Assuming a water density of 1 g/cm 3 , the equivalent hydraulic head and subsequent depth to water can be calculated from the formation pressure. Equivalent hydraulic heads are presented in Table 4 for the wells with available pressure tests. The depth to water shows that although the aquifers are confined, no artesian flow is expected.
Water density variations associated to borehole temperature and total dissolved solids can be expected. The (6) h = z + P∕ g Fig. 4 Example of aquifer identification based on formation water salinity analyses. (*) Freshwater indications from the spontaneous potential log. Three methods confirm the presence of low salinity formation water: (1) freshwater presence indicated by the positive deflection from the SP log; (2) formation water salinity estimated using Archie, expressed in NaCl TDS-equivalent and TDS, and; (3) the drill stem test in the Jessoma Formation, expressed in Cl content Table 3 Summary of freshwater indications in Northern Somalia according to the available oil exploration boreholes data and the method used to estimate salinity Reference elevation is the working depth reference for measuring depths (MD) relative to this point. The salinity threshold for this study, to be considered low-medium saline water, was set to <9 g/L TDS (9,000 mg/L) standard deviation of the molarity on the reported salinities is 0.0359 mol/kg NaCl. A sensitivity analysis maintaining the temperature constant, shows a variation in density of 0.00423 g/cm 3 , which is a negligible difference for this case. However, more important variations in density are seen with temperature variations. Maintaining a constant molarity of 0.2 NaCl M 11.76 g/L, when temperature varies, assuming Tmin 25 °C and Tmax 85 °C, the maximum difference in density is approximately 0.02862 g/cm 3 . Thus, as the calculations were made with a constant density of 1 g/cm 3 , the margin variation in the equivalent hydrostatic head calculation due to potential density variation is expected to be between 0 to 40 m for every 1,000 psi of formation pressure.

Daroor basin
Petrophysical analysis indicates three main aquifers in the Daroor basin: (1) Auradu limestone formation with 13-20% porosity, approximately 80-mD horizontal permeability, and salinities between <0.64 and 8.5 g/L TDS; (2) Jessoma sandstone formation with 14-40% porosity, very high vertical and horizontal permeability up to 1 Darcy, and salinities between <0.64 and 8 g/L TDS; and (3) Gumburo limestone formation with 2-10% porosity, 8 mD permeability, and salinities of 0.5 g/L Cl and 3.76 g/L TDS. The Adigrat Formation also showed the presence of unusually low values of salinity in Buran-1, with chlorides below 1.6 g/L. This could be related to the placement of the borehole on the flank of the basin targeting a shallower Adigrat Formation with possible meteoric water influx.

Well Shabeel North-1
The well Shabeel North-1 was drilled in 2012 by African Oil Corporation and reached total depth at 3,919 m, drilling 29 m into the basement. Positive spontaneous potential deflection indicates low salinity water in the Auradu, Jessoma and Gumburo formations, which is verified by the Pickett plot method for water resistivity estimations, as shown in Fig. 4. Formation water salinities estimated from Pickett-plots varied with depth, from a low salinity value of 3 Ω.m at 70 °C, corresponding to 0.7 g/L NaCl TDS-equivalent in the Auradu Formation to 0.04 Ω.m at 114 °C, corresponding to 60 g/L NaCl TDS-equivalent in the Qishn sandstone member at the top of the Gabredarre Formation.
Freshwater occurrences are also reported by the DST performed in the Jessoma Formation, with two open hole tests performed by using a jet pump across the interval 1,909.5-1,967 m MD. These tests indicated fairly good flow rates (1,780-1,990 bbl/day) with low salinity water and no oil indications. The salinity of the recovered fluid was estimated to be <2.1 g/L Cl. These low-to-medium salinity values agree with the freshwater salinities (i.e., 1.35-2.5 g/L TDS) that are reported in regional reports for the Jessoma aquifer (Horn Petroleum, unpublished 'Shabeel North-1 geological final well report', 2012). Table 4 Formation pressure converted to equivalent hydraulic heads in some wells from Northern Somalia, assuming water density of 1 g/cm 3 . Gauge pressure is the absolute pressure measured by the device, and Pressure is the pressure corrected to the atmospheric pressure a Measured from the sea level, positive values mean water level below the sea level, negative values mean water level above the sea level b Measured from the topography, positive values mean water level below the topography, negative values mean water level above the topography

Well Shabeel-1
The well Shabeel −1 was drilled by Horn Petroleum, a subsidiary of Africa Oil Corporation, in 2012 and reached total depth at 3,470 m, after drilling 40 m into the basement. The SP method suggests three different intervals with low salinity water: the Gumburo, Jessoma and Auradu formations (Fig. 5). These results coincided with several findings from the well operators, as detailed in the following. The Jessoma Formation reported 874.73 kg/m 3 EMW (equivalent mud weight) of pressure gradient to surface, which is significantly lower than the hydrostatic pressure gradient. The low pressure was interpreted to be due to the relatively fresh formation water composition of the formation water (<6 g/L NaCl TDS-Equivalent or 6.4 g/L TDS) and a low water table near to 300 m below ground level.
Due to logistics problems, no open hole tests, conventional cores or sidewall cores were taken on this well. Using the Archie relation (Schlumberger 1991) with assumed constants of a = 1.0, m = 2 and n = 2, formation salinities were found to vary throughout the well from fresh to brackish to sea-water values. The presence of freshwater at deep depths was unexpected and the oil operator noted in the final report that additional tools would have been helpful  Table. 3) to better distinguish oil from fresh formation water <5 g/L NaCl TDS-equivalent.
The Auradu and Gumburo formations are predominantly composed of carbonates and show SP activity in zones that were reported as highly fractured and vuggy from cuttings. Assuming no polarization effect in the resistivity measurements (because of a relatively vertical well), the separation of the shallow and deep resistivity measurements in the Auradu and Gumburo formations suggests some permeability as a conductive mud filtrate invasion profile can be seen.

Well Buran-1
The well Buran-1 was drilled in 1958 by Amerada Petroleum Corporation and reached TD at 2,437 m after drilling 11 m in the basement. Borehole logs from this well were not available for this study; however, indications of freshwater are noted in the well reports.
A detailed coring analysis reported water of 0.7 g/L Cl in the Auradu Formation, <0.64-1.15 g/L TDS in the Jessoma Formation, 0.5-0.6 g/L Cl in the Gumburo Formation, 0.6-0.9 g/L Cl in the Gabredarre Formation, and 0.9-1.6 g/L Cl in the Adigrat Formation. Additionally, a DST performed in the Auradu Formation (319-331 m) recovered 48 m of mud-cut freshwater with a formation pressure of 100 psi, suggesting a local groundwater level of 246 m below surface. The lower Jessoma Formation (937-947 m) was also tested through a DST and recovered 259 m total fluid as follows: 30 m water-cut drilling mud and 228 m slightly muddy freshwater, with 1.23 g/L TDS.

Nogal Basin
The wells Kalis-1 and Nogal-1 are located in the center of the Nogal basin, while Yaguri-1 and Burhisso-1 are located at its flanks (Fig. 1). Auradu, Jessoma, Gumburo, Gabredarre, Hamanlei and Adigrat are the formations with indications of freshwater from the resulting interpretations. Figure 6 summarizes the occurrences of low to mediumsalinity pore water encountered while drilling the four wells and through subsequent data analysis.

Well Kalis-1
The well Kalis-1 (Fig. 6) was drilled in 1990 by Conoco Somalia Ltd. and reached TD at 2,947 m after drilling 224 m into the basement. A relevant geological finding from this well is that all Jurassic formations, except the Adigrat Formation, are missing. The absence of the Hamanlei Formation is explained by previous authors (Ali and Lee 2019a, b) as due to pre-Cenomanian deformation and erosion of a large part of the Jurassic deposits.
Based on the SP method, the Jessoma sandstone interval 944-992-m TVDSS, exhibits the lowest formation water salinity of the whole section ( Fig. 6; Table 3). Also, the Auradu, Jessoma and Gumburo formations, which were drilled with a mud filtrate resistivity of 0.159 Ω.m at 85 °F, show positive spontaneous potential deflection in some intervals, meaning that the water of the formation has a lower salinity than that of the mud filtrate. This is quantitatively confirmed by a DST test in Jessoma Formation (1503-1515 m MD) that recovered formation water in the 7 g/L TDS range. An important observation is that the SP log deflection in the upper aquifers is noticeably more positive than the one tested with the DST, i.e., the salinity of those intervals is expected to be lower than 7 g/L. No raw resistivity measurements or further operator reports were available for the upper formations.

Well Burhisso-1
The well Burhisso-1 was drilled by Amerada Petroleum Corporation in 1958 and reached TD at 1551 m after drilling 11 m within the basement. Core analysis reported water of low-to-medium salinity in two formations: 1 g/L chlorides in Gumburo (1,364-1,371 m MD) and 1 g/L chlorides in Adigrat (1,510.28-1,518.8 m MD).

Well Nogal-1
The well Nogal-1 was drilled between 1989 and 1990 by Conoco Somalia. The well explored the center of the basin where formations are deeper. However, due to operational issues, the well could not reach the basement and called total depth (TD) within the Gumburo Formation at 3,272 m.
The SP log shows positive activity in the Jessoma and Gumburo formations, particularly within the sandstone where the Kalis-1 well reported low saline water. Previous studies interpreted the Jessoma Formation to be flushed by freshwater, probably as a result of migration from the tilted fault block, across which the Auradu Formation was juxtaposed against the Jessoma Formation (Ali and Lee 2019a, b). No DST (Drill Stem Test) tests were carried out due to severe hole problems, limiting the interpretation for finding freshwater prospects.

Well Yaguri-1
The well Yaguri-1 (Fig. 1) was drilled in 1958 by Amerada Petroleum Corporation and reached TD at 1,441 m, after drilling 15 m in the basement.
A DST in the Hamanlei Formation (1,390.80-1,408.48 m) indicated formation water salinity of 1.8 g/L chlorides (Fig. 5). Yaguri-1 is the only well that proved relatively lowto-medium saline water in the Hamanlei Formation in the Nogal basin; however, it is also the only well where a DST was performed in this formation.
A DST in the Adigrat Formation (Table 3) tested lowto-medium salinity water of 3.5 g/L NaCl TDS-Equivalent using the GEN-9 Schlumberger chart (Schlumberger 1997), and 1.6-g/L chlorides with silver nitrate test (chlorinity titration). There is a good agreement between the wells Yaguri-1 and Burhisso-1, both reporting relatively low-to-medium water salinity in the Adigrat Formation, also possibly because the formation is found shallower as the boreholes were drilled in the flanks of the basin as the Buran well in the Daroor basin (Fig. 1).

Observations on the potential stacked aquifers
This study identified a series of hydrostratigraphic units which are deeper than the well-known Paleocene-to-recent freshwater-hosting stratigraphic units. The deep freshwater bearing formations are, from older to younger, the Adigrat, Hamanlei, Gabredarre, Gumburo, Jessoma and Auradu formations (Fig. 1).
A difference in the salinity of the Adigrat Formation water is observed between the wells drilled in the center and those drilled on the flanks of the basins. Water salinity in the wells at the flanks (Buran-1, Yaguri-1 and Burhisso-1) exhibits <4 g/L TDS and < 1.6 g/L Cl, while Shabeel-1, drilled in the center, reports salinities above 9 g/L TDS, which was not included in this study as it is higher than the defined threshold of 9 g/L TDS. The other wells in the center of the basins show little or no freshwater potential after applying the SP method.
In the Nogal basin, the upper Hamanlei Formation in the Yaguri-1 well tested low-to-medium salinity water of 1.8 g/L Cl. In the Darror basin, the SP method did not show freshwater potential and the tested interval in the upper part of the Hamanlei Formation in the well Shabeel-1 showed salinities >9 g/L TDS on average.
The Gabredarre Formation was observed in the Daroor basin in the well Buran-1 well and tested low-to-medium salinity water of <0.9 g/L Cl. Shabeel-1 and Shabeel North-1 did not exhibit freshwater potential in this formation and were not tested when drilled. In the Nogal basin this formation was not present. The Jessoma Formation was not tested in the Yaguri−1 and Burhisso-1 wells. No SP or resistivity logs were available for those two wells; therefore, the interpretation remains open to further inves-tigations on whether the aquifer is present in the flanks of the basin. Formation water salinity based (*), when at least one of the applied methods indicated low-to-medium salinity (see Table. 3) The Gumburo Formation is penetrated by the Shabeel-1, Buran-1, and Burhisso-1 wells, with a water salinity of 0.6 g/L Cl and 3.76 g/L TDS. Furthermore, the SP methodderived salinity suggests freshwater potential in different intervals within this formation in Kalis-1, Nogal-1, Shabeel North-1, and Shabeel-1.
The Jessoma Formation has indications of low-to-medium salinity water in almost all of the wells, making it a potential aquifer for exploration. Freshwater that is observed in the Jessoma Formation in Kalis-1 did possibly migrate from the Auradu Formation by juxtaposition across a nonsealing fault. (Ali and Lee 2019a, b). This could be the case for most of the wells in the Nogal basin.
Hydrogeological reports have consensually described the upper part of the Auradu Formation as an aquifer with good water quality, with EC values varying from 1,034 to 5,840 μS/cm (0.662-3.74 g/L TDS using 1 EC (μS/ cm) = 0.00064 (g/L) TDS at 25 °C conversion; SWALIM 2012). New results from this study suggest that freshwater could exist in the deeper Auradu Formation with salinities below 1 g/L TDS.
Overall, these results show potential for the presence of a series of deep low-to-medium salinity aquifers, hosted in the Mesozoic to Paleogene formations of the northern part of Somalia. However, further studies are required to understand the deep aquifers in the Darror and Nogal basins, in particular concerning their recharge volume and provenance, and the role of fractures and karst systems. Moreover, some of the aquifers could potentially coexist in equilibrium with evaporitic formations. The modification of the local hydraulic gradients induced by the exploitation of this water should be addressed as it could lead to contamination of the aquifers.

Well cross flow
When a series of vertically juxtaposed (stacked) aquifer systems are intercepted by a single well, the dynamics of the aquifer system become complex and in extreme cases drilling operation can be detrimental for the aquifer. The classic scenario is when a poorly cemented and/or corroded casing in an oil well contaminates freshwater aquifers with saline water or with hydrocarbons (Khalfi and Ahmadi 2022). These hazards come in addition to the more common problems related to shallow gas blow outs and wash-outs due to excessive artesian pressures. In the study area, in the Horn of Africa, some 70 years of exploratory O&G well drilling has left behind a legacy of nearly 70 abandoned deep wells, and the potential for regional water depletion, and/or groundwater contamination by cross flow due to corrosion, and/or poor well abandonment practices can be expected.
Cross flow occurs when the wellbore encounters permeable zones whose pore pressures are not hydrostatically balanced. This often occurs when permeable formations are separated from underlying very low permeability layers (Bauer et al. 2005). The well Buran-1 in the Daroor basin, and Kalis-1 and Yaguri-1 in the Nogal basin were subjected to pressure tests in different formations providing insight on the potential cross flow to be considered in future investigations.
The Buran-1 well had three successful drill stem tests in three different formations-Auradu, Jessoma and Adigrat formations. All formations tested dry for hydrocarbons, and in contrast, they have shown low saline formation water. Results show that if open to production at the same time, a natural cross flow from Auradu and Jessoma into the Adigrat Formation will be created whenever the water level is above ca. 220 m above the sea level. The salinity levels of the formation water in the three formations are relatively similar (Table 3); however, further analysis of the quality of the water is required to discard potential cross flow contamination (Fig. 7).
Several pressure tests were performed also in the Kalis-1 well. The Jessoma and Gumburo aquifers show approximately the same formation pore pressure (Fig. 8). The slight differences can be related to differences in density; however, Auradu aquifer exhibits higher formation pressure with an equivalent hydraulic head of ca. 100 m above Jessoma and Gabredarre. If the three aquifers are open at the same time, a natural cross flow will go from Auradu to Jessoma and Gabredarre.
Two DSTs were performed in the Yaguri-1 well, both in the Adigrat Formation. The differences in the aquifer energy are relatively low and could be related to differences seen in the density/salinity (Fig. 9).

Sustainability and socioeconomical factors
Available data indicate the potential presence of significant regional aquifer systems. More important, these resources may be located in areas presently not exploited or inhabited, due to the lack of water. The demographic effects of a decentralised water supply system will be invaluable. At present, during the all-too-frequent droughts, people living off the land migrate towards the alluvial aquifers along the main rivers Juba and Shebelle. These areas normally accommodate large numbers of drought migrants from the countryside, causing over-exploitation of both land and the limited shallow aquifers. With more decentralised water sources spread over the entire country, much pressure will be taken off the alluvial aquifers and their environs, while enabling a larger part of the country to be utilised for settlements, agriculture and grazing.
A cost-benefit analysis prior to the exploration of these aquifers is required to evaluate the feasibility of implementing deep groundwater in the region. A more general analysis of the feasible solutions to address water scarcity in Somalia was previously undertaken (Faillace 1984). The study recommended focusing more on water harvesting; however, as surface water depends on localized rain, with the climate change of the past decades this resource has become scarcer. An example of this is the levels of the rivers Juba and Shabelle, which are currently 30% below average short-term average, with limited water available to support irrigation of crops and other uses (FAO-SWALIM 2022). Boreholes tapping from shallow aquifers also depend on seasonal rain. As the country experiences severe drought, shallow boreholes are being over exploited showing an abnormal drop in groundwater levels (FAO-SWALIM 2022).
Recently, a new analysis of deep groundwater as an alternative source of water in the Ogadan Jessoma Sandstone Aquifers of Somalia and Ethiopia was undertaken (Godfrey et al. 2019). Economically, an initial capital expenditure (CAPEX) investment in deeper aquifers is considered to provide a more resilient water source which in the long run is less costly than producing from shallow aquifers in the region. The study compares the low yield from shallow boreholes (259.2 m 3 /day) which produce brackish water with the observed yield (1,296.00 m 3 /day) from the deeper Jessoma Formation with freshwater at depths of 600 m below ground level (Godfrey et al. 2019).
Reliability and sustainability are also important factors to consider. The travel time in deep regional aquifers is generally high and involves large quantities of water. These systems are therefore per definition insensitive to short-term climate variations, which is helpful during shortterm droughts where shallow aquifers are significantly more vulnerable. In terms of sustainability, it will be possible to plan future water abstractions according to long-term aquifer replenishment, in addition to what is an accepted abstraction from aquifer storage.
The distance from recharge areas within Ethiopia and the final reservoir under the Indian Ocean is in the order of 1,000 km. Preliminary estimates, considering the vast distances involved, indicate that water is steadily moving at a slow rate towards the Indian Ocean (Ruden 2017), rather than being of a fossil nature. Due to large cross-sectional area, the Darcy velocity of water particles can be in the order of 10 cm/year; at this rate the journey will take 1 million years; with increased porosity and permeability created by fracture/fault-flow and karst, which is likely considering the observed karstic phenomena on the surface (REF), the transit time would probably be much reduced.
The salinity of some of the waters encountered in the deep aquifers covered by this study are at the upper end or exceed the values considered suitable for human drinking, which is normally less than 1 g/L TDS. However, higher values can be suitable for irrigation <2 g/L TDS, for livestock Fig. 7 Buran-1 well lithology column with pressure tests performed in potential freshwater aquifers in blue circles. To the right, equivalent hydraulic heads derived from formation pressure are calculated at each testing point assuming 1 g/cm 3 water density Fig. 8 Kalis-1 well lithology column with pressure tests performed in potential freshwater aquifers in blue circles. The pressure test number 5 (white circle) has no evidence of freshwater in the available data.
To the right, equivalent hydraulic heads derived from formation pressure are calculated at each testing point assuming 1 g/cm 3 water density Fig. 9 Yaguri-1 well lithology column with pressure tests performed in potential freshwater aquifers in blue circles. To the right, equivalent hydraulic heads derived from formation pressure are calculated at each testing point assuming 1 g/cm 3 water density up to 10 or 15 g/L TDS, and for some industrial uses, if exceeding these values. Additionally, it must be considered that the high costs of obtaining freshwater by desalinizing seawater can be substantially reduced if used as an input water with salinity inferior to sea water.
While this study shows promising results with respect to the presence of deep regional freshwater aquifers in the Horn of Africa, more studies are needed in order to estimate the renewability of these potential freshwater sources for their possible future use. This is a key aspect, since groundwater extraction must be lower than aquifer replenishment to be sustainable, regardless of aquifer permeability and/or porosity.

The use of O&G data and methods in hydrogeology
Resistivity, gamma ray, density, neutron-porosity, and photoelectric effect are key logs in determining hydrocarbon presence; however, these logs respond similarly to formations filled with freshwater. This can benefit hydrogeological studies, because the apparent response to hydrocarbons causes the oil industry to run more tests such as pressure tests and fluid samples for a more detailed fluid characterization, allowing, as in this case, the discovery of freshwater aquifers, with abundant information on freshwater indications in reports. Moreover, the resistivity of the formation water can be obtained from the resistivity logs by applying Archie's law and compared with the qualitative response of the spontaneous potential log to confirm results as both principles are of different nature. The petrophysical methods can therefore be repurposed for the exploration of deep aquifers.
The application of O&G petrophysical principles to hydrogeology has been previously discussed by a number of authors (Patten and Bennett 1963;Varma and Michael 2012;Vonhof 1966). Discrepancy among authors arises when it comes to the properties of the borehole fluid in relation to the formation water. The equations for the SP method assume that both the formation water and the borehole fluid are pure sodium chloride solutions, and therefore applying this method to freshwater-based borehole fluid can result in incorrect interpretations in traditional water boreholes, drilled with freshwater. However, in hydrocarbon exploration, wells are generally drilled with brine when using waterbased mud. This makes the available information previously acquired by the O&G industry suitable for utilizing this log to estimate the salinity of the formation water.

Conclusions
In this study, regional geological and hydrogeological data have been combined with previously acquired O&G industry well data, to understand the distribution of potential deep, unexplored freshwater aquifers in the northern part of the Horn of Africa. The discovery of new deep aquifers could help ease the water shortage in this arid region, where the currently exploited aquifers are relatively shallow (10s-100s of meters) and largely contaminated by salts related to leaching of evaporitic formations, and by human and/or farming agricultural activities. Conversely, the sporadic presence of high-yield freshwater springs suggests the presence of deep sources of uncontaminated groundwater.
To evaluate the presence of these groundwater resources, deep penetration (up to >3 km) well data were used, which were acquired and processed by the O&G industry. Gamma ray, resistivity, and mud logging data were analysed, to highlight freshwater saturated units in the subsurface. In addition, a methodology, commonly applied to O&G well analysis, was tested by combining the analysis of spontaneous potential (SP) and Archie's law derived resistivity. This revealed the presence of a series of deep hydrostratigraphic units in the Nogal and Darror basins, in Jurassic to Paleocene/ Eocene formations, which were virtually unexplored for fresh groundwater, and beyond the well-known Paleoceneto-recent freshwater-hosting stratigraphic units.
Freshwater bearing intervals are present in the Adigrat, Hamanlei, Gabredarre, Gumburo, Jessoma and Auradu formations. With some tested intervals revealing water of less than 3.8 g/L TDS, the Auradu, Jessoma and Gumburo are the formations that have consistently shown indications of freshwater in both basins and are the most promising as a potential additional resource for the region. These results allowed us to obtain fundamental parameters such as hydraulic head and salinity, and provide the base for further groundwater modelling. Ultimately, these new results have global significance, as they have the potential of both boosting groundwater fresh groundwater resources in water-poor regions, and supporting sustainable development and utilization of the resources in the medium and long term. Additionally, this study contributes to an improved understanding of mechanisms for freshwater emplacement and preservation in the deep subsurface, which is incomplete in current global water budgets.