Groundwater flow system of the Abijata-Langano-Ziway lakes basin, Ethiopia Authors: Debele Abera, Behailu Berehanu, Tenalem Ayenew Published: 2024 ABSTRACT AbstractThe Abijata-Langano-Ziway Lakes Basin (ALZLB) is situated in the Central part of the Main Ethiopian Rift. The availability and dynamics of groundwater in the Abijata-Langano-Ziway Lakes Basin (ALZLB) are primarily controlled by its geological and hydrogeological characteristics, shaped by volcanic-tectonic and sedimentary processes. The basin faces significant challenges, including drastic change in land use pattern, rapid population growth sustained by subsistence farming, over-extraction of water resources, and vulnerability to climate change and fragile ecosystems. These issues emphasize the urgent need for effective water resource management.To understand this complex system, a numerical groundwater flow model was employed to characterize the groundwater flow system within the ALZLB and examine its interaction with surface water bodies. The MODFLOW model translates the conceptual understanding of the basin's hydrogeology into a mathematical representation, allowing for numerical analysis. The model incorporates input parameters such as hydraulic conductivity and boundary conditions representing groundwater inflow and outflow. Steady-state numerical calculation was used to characterize the qualitative hydrogeological conceptual model into numerical representation and thereby describe the groundwater system.The calibrated model exhibited excellent agreement between simulated and observed groundwater levels. Statistical measures indicated a strong correlation (R2 = 0.98) and high efficiency (NSE = 0.97) in replicating the observed data. Additionally, the Mean Error (ME) of −8.3 m suggests minimal bias in the simulations. Further analysis of the histogram residuals revealed that a significant portion of the simulated values (65 % and 82 %) fell within ±20 m and ±30 m of the observed groundwater levels, respectively. This revealed the model's accuracy in capturing the groundwater system's behavior.The model identified groundwater recharge and constant head boundaries as the primary sources of groundwater inflow, contributing 445 million cubic meters per year (MCM/year) and 90 MCM/year, respectively. Conversely, constant head boundaries represented the most significant outflow pathway, with a simulated discharge of 519 MCM/year. The calibrated balance between inflow and outflow (discrepancy of −0.75 %) confirms that the model effectively simulates steady-state groundwater flow conditions. The calibrated model demonstrates the model's capability to accurately represent the basin's groundwater system.Keywords: Water budget, Volcano-tectonic, Finite difference, groundwater hydrology, Water resources Introduction Numerical systems modeling can reveal the complex interactions between groundwater and surface water bodies, including rivers, lakes, wetlands, and estuaries. These models demonstrate how groundwater discharges contribute to baseflow, sustaining surface water during arid periods. Conversely, surface water bodies can act as recharge zones for underlying aquifers via infiltration processes [1]. A thorough understanding of groundwater-surface water interaction is paramount for effective water resource management [2]. This knowledge facilitates the development of sustainable water allocation strategies, the evaluation of groundwater pumping impacts on surface water availability, and the implementation of measures to safeguard and restore surface water bodies reliant on groundwater inputs. Groundwater flow systems underpin the health and function of aquatic ecosystems. They act as a perennial source of water for surface water bodies, maintaining stable flow regimes and suitable habitat conditions for aquatic organisms [3]. Groundwater discharge also influences temperature regulation, nutrient availability, and the overall ecological integrity of surface water ecosystems, ultimately shaping species distribution and fostering diverse aquatic communities [4]. However, climate change presents a significant challenge to water resources globally [5]. Alterations in precipitation patterns, temperature, and hydrological cycles can disrupt the availability and timing of groundwater recharge and discharge, consequently impacting surface water availability [6,7]. The effect of climate change on the basin under investigation has been addressed by numerous studies and showed variable results. For instance, a decline in runoff and streamflow was reported [[8], [9], [10]]. On the contrary, up scaling in future precipitation projection results were published by numerous authors [[11], [12], [13], [14]]. Even some authors revealed mixed results, both rising and falling in trends of precipitation in the basin [15,16]. Modeling groundwater flow systems is imperative for anticipating the implications of climate change on water resources and subsequently, its impacts on aquatic ecosystem dynamics, in the basin where the optimal balance between human and ecological water uses is adequately addressed. The groundwater flow system has been widely used to examine the interaction of surface water and subsurface water [17]. Besides, the response of aquifer under fluctuating rainfall condition groundwater flow system in fracture controlled hydraulic system [18] and topographically controlled groundwater flow system are among the others [19]. Similarly, there are numerous groundwater modeling studies that have been published in the region with multiple objectives [[20], [21], [22], [23]]. In the study basin, groundwater is a crucial source of water supply for agricultural, domestic and industrial uses. The water supply for villages and towns predominately relies on groundwater sources and also more than 25 % of water demand for irrigation has been directly abstracted from aquifers in the basin [24,25]. Groundwater may be manifested as springs, streams, and wetlands. The perennial streams would not have permanent flow without groundwater contribution as baseflow [1]. Besides this, it has a great role in sustaining water resources in the basin. For instance, groundwater is believed to have a connection with the chain of lakes located in the rift floor of the basin and surrounding ecosystems. The highlands and plateau areas are mainly considered as potential recharge areas. On the contrary, the rift floor is the region where groundwater is found at shallow depth and is regarded as a discharge zone in the basin. The baseflow is considered as groundwater discharged to the streams and a simple baseflow separation method shows that its contribution to Ketar and Mike rivers flows [26,27]. There is a continuously increasing demand for groundwater development that may lead to over-exploitation of groundwater resources and growing impacts of human activities on aquifers, such as a decline in hydraulic heads and degrade groundwater quality. The basin supports a growing human population. These inhabitants depend critically on the basin's ecosystem services, primarily for agricultural production, livestock grazing, and fisheries. However, these anthropogenic activities present a potential threat to the basin's sustainability, potentially leading to water quality deterioration, biodiversity decline, and land degradation. The study basin has also been facing the challenges of satisfying demands for irrigation, domestic and minimum water flows to sustain the ecological health of the basin [[28], [29], [30], [31]]. Hence, the continuity of water resources is severely challenged in the phases of stressors such as population growth, increased demand for agricultural and urban uses, the need to maintain adequate environmental flows to sustain healthy ecology, and potential impacts of mankind-induced modifications such as land use change and climate change. Agricultural constituted the key pillar in supporting the economy in the basin and moreover, there has been sharp growing of irrigation based investments in the basin [32]. Furthermore, the escalation of unregulated irrigation practices causing excessive flooding is resulting in the salinization of soil [33]. As a consequence, there has been a growing trend of converting forested areas into agricultural land as farmers abandon the salt-affected lands and search for new, unaffected lands [34]. The decrease in the size of Lake Abijata is attributed to the extraction of water from rivers and lakes upstream, which is further intensified by climate change and land development [8,35]. This reduction in water availability has led to a decline in the area coverage of Lake Abijata from 179.2 km2 in 1972 to 131.8 km2 in 2015 [36]. As a result, there has been a loss of biodiversity in the freshwater ecosystems due to human impacts on water and the extraction of water is often carried out without a proper understanding of the complex hydrological and hydrogeological systems and the delicate nature of the rift ecology [3]. Furthermore, the study conducted in the basin provides a concise overview of the key water-related issues that require effective water resource management. These include the unsustainable extraction of water, which poses a significant threat to the local ecosystem and leads to intense competition for water resources without proper differentiation in their usage; the decline in water quality, affecting both irrigation and domestic purposes; the heavy dependence of the population on water resources to sustain their livelihoods [37]. So far, a few studies on a numerical groundwater model have been carried out with the objective mainly focusing on its potential [38] and groundwater interaction with the lakes [39]. Therefore, the purpose of this study was to bridge notable research gaps regarding how the groundwater flow systems are configured spatially across the basin and its interaction with surface water using model-based approach. Moreover, model-based informed decision-making tool is very critical for water resource management given the present dynamic situation in the basin. Characterization of the study basin Description of the study basin The Abijata-Langano-Ziway Lakes Basin (ALZLB) is located in the central section of the Central Main Ethiopian Rift valley, which is part of Great East African Rift Valley System. It encompasses part of Oromia and Southern Nations and Nationalities and Peoples’ Administrative Regions. It is a part of the Ethiopian Rift Valley Lakes Basin, and it is bounded by geographic coordinates of 414500–545180 mE longitude and 797930-93592 mN latitude (Fig. 1) using projection of Universal Transverse Mercator (UTM), Adindan N37 zone. The Great East African Rift Valley splits the Ethiopian highland into northern and southern parts, creating a rift floor inhabited by rows of lakes with different sizes, hydrological and hydrogeological settings [40]. Study area, (a) Ethiopia, (b) Rift valley basin, (c) ALZLB. The basin endowed by Ziway, Langano and Abijata lakes lodging in the axis of the rift and on both sides of the basin, there are numerous persistent streams (Meki, Ketar, Gedemso, Lepsi, and Huluka) draining into the lakes. The basin supports a population of more than 2 million whose livelihoods mainly rely on subsistent farming [41] with the annual growth rate of 3 % [42]. According to Ref. [32], more than 83,000 ha of lands was irrigated using surface and groundwater sources and about 1.8 million of livestock [43]. The basin covers an area of more than 10, 000 km2. The basin is characterized by humid to sub-humid in the plateaus and semi-arid in the rift floor. The mean yearly precipitation over the entire basin is around 900 mm, showing high spatial variations as no more than 700 mm/year and as high as 1200 mm/year in the rift floor and highlands respectively [44]. The Rift floor region of the central rift valley basin is occupied by a row of lakes: Ziway, Langano, and the terminal lake Abijata (shown in Fig. 1), while their areal extension, depth and volume in the 1ate 1990s’ (Table 1). However, the current size of the Lake Abijata has been modified to the large extent. As of 2020, the estimated area of the lake from the image of Google Earth was 166 km2. This variation may be associated with a temporal change of the lake area coverage. These lakes and perennial rivers are thought to have existential connections with the groundwater system of the basin because of its geological and hydrogeological frameworks (see Table 2). Table 1The basic lakes data.Table 1LakeAltitude (masl)Lake area (km2)Maximum Depth(m)Mean Depth (m)Volume (MCM)Abijata158018014.27.6954Langano158523047.9173800Ziway16354408.92.51466Source: (Ayenew, 1998) The basic lakes data. Source: (Ayenew, 1998) Table 2Definition.Table 2SymbolDefinitionUnitQRIVRateofleakagebetweentheriverandtheaquifer(L3T−1)CRIVHydraulicconductanceoftheriverbed(L2T−1)HRIVHeadintheriver(L)hHydraulicheadincell(L)RBOTElevationofthebottomtheriverbed(L)KHydraulicconductivityoftheriverbedmaterial(LT−1)LLengthoftheriverwithinacell(L)WWidthoftheriver(L)MThicknessoftheriverbed(L) Definition. Geological and hydrogeological settings The geology of the basin is constituted by three major rock categories documented in Refs. [[45], [46], [47]] in the basin: (i) Basement and Mesozoic Sedimentary rocks; rarely outcropped in few localities in the western part of the basin; (ii) Volcanite and pyroclastic rocks and older lacustrine deposits and (iii) Quaternary Sedimentary Deposits. The volcanite and pyroclastic groups of rocks comprise several units and cover about 71 % of the basin and in age ranging from Oligocene to Middle Pleistocene. These groups of rocks have great significance in determining the hydrological and hydrogeological flow regimes of the basin. The volcanite unit is made of basalts and rhyolitic ignimbrites and tuffs. Especially in the western part of the basin where the Meki River originated, the flat-lying plateau and the strongly faulted major plateau – rift escarpment in the basin are dominantly constituted by Pliocene age ignimbrites and tuffs. These rocks are less permeable, and are used as a groundwater divider on the western border of the basin. They are formed from thick beds of flood basalts, with rare beds of basic tuffs, agglomerates, and volcanic breccia, associated with some interlayered silicic volcanic rocks. The upper unit is dominantly rhyolite and trachyte flows with minor basalts, while flood basalts with minor silicic flows constitute the lower parts. Arsi Shield Volcanoes (Eastern side) is the Late Miocene to Pliocene age forms shield volcanoes that largely contains basalts, mugearites and minor trachyte, and phonolites, but they also contain inter-bedded basaltic breccia (agglomerates). They serve as groundwater divide in the eastern boundary and have a significant influence on meteorological and hydrological parameters, because of their occurrence at the highly elevated Eastern part of the basin and are characterized by low permeability. Rift Floor Ignimbrites unit is composed of peralkaline-rhyolitic- ignimbrites and loosely fused ignimbrites and pumice with subordinate lava flow of basaltic composition. This group of rocks is found abundantly in low lying rift floor region. They also occupy the Wonji Fault Belt (WFB) and the Siltie-Debre Zeyit Fault Zone (SDZFZ) faulting zones. Hence, this unit could be a zone of high groundwater potential that has significant control over the groundwater flow regime. In addition, it significantly influences the hydrochemistry of the basin because of the rhyolitic composition of rocks. Quaternary Sedimentary Units, Late Pleistocene to Late Holocene a complex sequence of lacustrine and fluvial deposits, and recent colluvial and fluvial deposits exposed at the foot of ridges (alluvial fans) occupy the rift floor. These groups of rocks have limited exposure but where they occur, they can influence groundwater flow regimes and strongly determine the hydrochemistry. Besides, there are other rock units that occupy a relatively small portion of the basin that control groundwater movement locally. The hydrogeology of the basin is made up of the dominant and vast aquifer-forming groups of rocks, which are volcanic and volcanoclastic and Quaternary sedimentary units. These volcanics and volcanoclastic rocks comprise rocks of alkali trachyte, rhyolites, ignimbrite, tuffs and basalts that are found on the eastern and western escarpment and plateau of the basin. These were highly affected by rift faulting and intensively fractured and jointed and these structural features facilitate fast infiltration into the groundwater system [39]. Quaternary sedimentary units occur in the central (lakes region) and low-lying part of the basin and it is the second largest unit in surface area coverage. These units, in places where, there are secondary porosity and have an enormous capacity to store and transmit the groundwater. In contrast to these, they may have an impermeable layer that affects the region's hydrologic and hydrogeologic characteristics in some places [28]. Conceptual model of the study basin Developing a conceptual model for the groundwater-flow system in the ALZLB Basin requires the definition of the hydrogeological framework: the identification of hydrogeological units within the lacustrine and alluvial deposits and volcanic units that are characterized as aquifer systems. The top layer is considered as an unconfined aquifer whereas the bottom layer is considered a confined aquifer system. The second part of the conceptual model consists of the hydrologic behavior of the flow systems in terms of shallow local flow and deep local flow, and deep regional groundwater flows. The presence of faults, especially in the escarpment of the basin, affects the groundwater flow system largely. The flow systems are characterized by sources of water (recharge, movement of water from surface water to groundwater, release of water from storage) and sinks (discharge to surface waters, addition of water to storage). Springs emerged at lithological contacts and faulted zones. There are varying, amounts of sources and sinks in the case of lakes that may have a large influence on water levels and water movement in the basin. The third part of the conceptual model element is the connection of groundwater and surface water. The extracting of groundwater can appreciably alter the natural pattern of local and regional discharge into the lakes in the region. This, in turn, affects the levels of lake water. Finally, the calibration was carried out against the observed borehole data using the manual method through the trial-and-error method. The following assumptions were made based on the above schematic conceptual model (Fig. 2). The source water (groundwater, spring and rivers) in the basin is from direct rainfall and rainfall on both sides of the escarpments and plateau. The occurrence of groundwater in the western and eastern escarpment is because of the hydrological and hydrogeological properties within various geomorphological units and lakes: The groundwater recharge zone is considered the western and eastern mountains and flows towards the lakes’ region; Butajira Crescent is assumed to be recharged from runoff coming from the escarpment, and direct rainfall on the plain; Lake Ziway receives water form groundwater, surface water, and direct rainfall on the Lake Surface and outflows groundwater at its south-eastern sides and through Bulbula River; and Lake Langano gain water from groundwater, surface water, and direct rainfall on the Lake surface and outputs through Horakelo River and groundwater. In addition to these, the groundwater flow direction is directed towards the lakes. However, as it approaches the lakes, the flow is directed to the south to the Lake Shalla on the eastern side and the main persistent rivers are believed to have links to groundwater. Schematic Conceptual model for groundwater flow condition. Methodology Governing groundwater flow equations Governing flow equations are a physically based mathematical models originated from Darcy's law and the mass conservation [48]. The equation assumes that a flow system that permits both horizontal and vertical flows via the aquifers system in three-dimensional outlines [49]. The governing partial differential equation executed by the MODFLOW to designate the groundwater flow for a steady-state condition, there is no change in the head with time, therefore, change in storage become zero, whereas for the transient condition when a change in storage with time exists. The MODFLOW model is designed for the condition of a saturated flow aquifer system, where Darcy's law is applicable and the primary flow direction of horizontal hydraulic conductivities don't alter within the system. Therefore, steady-state and transient flow equations are given in Ref. [50] (1.1) and (1.2) respectively. and for the transient model the equation became: Where, Kxx, Kyy, and Kzz are values of hydraulic conductivity in the x, y and z directions. Hydraulic conductivity (LT−1), h is the hydraulic head (L), W is a volumetric flux per unit volume and represents Sinks and/or sources (T−1), Ss is the specific storage of the porous material (L−1), and t is time (T). Boundary and initial conditions The model boundary conditions are delineated by defined hydraulic heads, flux and head dependent flux. It determines the flux of water enters and leaves the model domain. Similarly, in the study basin, the groundwater flow system is mainly controlled by geologic and hydrologic boundaries. Hence, a no-flow boundary is used to determine the geographic extent of the groundwater flow system and inflows and outflows along the model boundary and the general head boundary is used to represent the groundwater outflow and inflow depending on the head at the boundary in the southern part of the model domain. In general, there are three types groundwater model boundary domain [51]. ➢Specified (constant) head boundaries (which are examples of Dirichlet or first-type boundary conditions). A specified head boundary is implemented by fixing head values at the nodes along the boundary; hence, specified boundary heads do not change in response to hydrologic stresses. Specified (constant) head boundaries (which are examples of Dirichlet or first-type boundary conditions). A specified head boundary is implemented by fixing head values at the nodes along the boundary; hence, specified boundary heads do not change in response to hydrologic stresses. Example: Direct surface water-groundwater connection, lakes. ➢Specified flux boundaries (which are examples of Neumann or second-type boundary conditions) Specified flux boundaries (which are examples of Neumann or second-type boundary conditions) Example: impermeable bodies (no-flow); wells; recharge. ➢Mixed or head-dependent flux boundaries (which are examples of Cauchy or type-three boundary conditions) Mixed or head-dependent flux boundaries (which are examples of Cauchy or type-three boundary conditions) Indirect connection (e.g., river/lake, leaky aquifers, drains, distant boundary), C is hydraulic conductance of the boundaries, h are heads, q denotes discharge or recharge flux and ho and hi is hydraulic head outside and inside the model boundary respectively. River leakage Discharge to streams was simulated using the Modflow River packages [50]. Perennial streams were represented with the River package. The River package simulates the connection between surface water and groundwater. The hydraulic conductance between the streambed and aquifer (streambed conductance) is defined as the hydraulic conductivity of the riverbed material times the cross-sectional area of the stream reach is divided by the streambed thickness [52]. River cells can remove water from or add water to the aquifer depending on the direction of the gradient between the river stage and the aquifer hydraulic head. Initially, a cell representing a perennial stream depending on the order of rivers streambed hydraulic conductance given ranging from 8.0 m2/s to 0.43 m2/s was assigned. The mathematical expression for the river package is given as follows: Where, The outside of the model sub-basin was considered as a special type of Specified flux boundaries (which are examples of Neumann or second-type boundary conditions). It is specified as an inactive boundary cell (no-flow boundary) except in a few extents in the southern part where the groundwater outflow is evident [39] and treated as the general head boundary (Fig. 3a). For the initial model input purpose, river properties (depth, width, and hydraulic conductivity) were estimated as shown in Table 3. However, these values are subjected to be modified during model calibration process. (a) Boundary conditions of the model domain and (b) Initial hydraulic head [masl]. Table 3Estimated rivers parameters.Table 3Elevation (m)Depth (m)Width (m)Hydraulic conductivity>250013Similar to the underlain aquifers2500–200025“<2000310“ Estimated rivers parameters. General head boundary The head-dependent boundary was simulated using Modflow packages of the General Head Boundary (GHB) Package [50]. It simulates a source of water outside the model area that removes water from or adds water to adjacent cells depending on the hydraulic-head difference between the source and the model cell [52]. In the study area, groundwater outflow in the southern part from the model boundary shown in (Fig. 3 a) and general head boundary condition was implemented to estimate the amount of outflow. The rate of flow water between the model cell and the outside source or sink is given by where, Q is the rate of flow into or out of the model cell [L3/T], HB is the head assigned to the external source or sink [L], h is the hydraulic head within the model cell [L] values of C. K is the hydraulic conductivity between the model cell & the boundary head [L/T], and L is the length of the general head boundary within a cell. For initial the value of C is 9.5 m2/day for the top layer and 6 m2/day for the bottom layer. In addition, the initial hydraulic head was generated using interpolation of static groundwater elevation from observations to characterize the initially prescribed head and used as the initial condition in the system at the beginning of the simulation (Fig. 3 b). Model design (model domain and Specification) In order to put the conceptual model in a suitable form of numerical model, the model designing step is mandatory. This step comprises the layout of the grid cells, simulation time steps, boundary and initial conditions, and preliminary selection of values of aquifer parameters and hydrologic stresses. The Finite-Difference grid model consisted of two layers made up of 90 rows-oriented East to West and 95 columns-oriented North to South with equal dimension of 1500 m grid-cell. The thicknesses of the aquifers vary from 150 m to 200 m, and 250 m–300 m in the rift floor to the highland for the top layer and bottom layer respectively. Model parameter Hydraulic properties During groundwater modeling, it is a very challenging activity to obtain realistic hydraulic conductivity values at appropriate spatial distribution in the field, preferably by pumping tests. So that, hydraulic conductivity is often the most important and variable modeling parameter [53]. The initial horizontal hydraulic conductivity of the basin is compiled from different literature [38,54]. Based on the above studies, the horizontal hydraulic conductivity map of the basin was prepared. The highest hydraulic value is assigned to the ignimbrite unit at the eastern escarpment and followed by and the rift floor alluvial-lacustrine deposits where primary porosity plays a significant role in groundwater storage and transmission. The least value was given to balchi with insignificant areal coverage. The vast Oligocene-Early Pliocene Volcanites and Late Miocene-Pliocene Volcanites groups of rocks located on the highlands on both sides (at Eastern and western) have been assigned values of 5–10 m/day and 0.1–1 m/day at initial stage and later will be changed through calibration processes. However, there are locally higher or lower hydraulic conductivity values than the hydraulic conductivity zones shown in the (Fig. 4a). (a) Hydraulic conductivity map and (b) recharge estimation map. Groundwater recharge zones Groundwater recharge rate is one of the most variable and yet the most important parameter in groundwater modeling, despite its difficulty to accurately measure and estimate it because it occurs in a wide spatio-temporal dimension. In the case of the study basin, the major source of recharge to the aquifer comes from precipitation and river channel losses. As the main recharge mechanism, direct or diffuse recharge is supposed to take place in almost areas of the basin with exception to where low permeable lacustrine soil (diatomite) exists in the rift floor, around Lake Ziway [32]. The mean annual recharge estimated using catchment soil-water balance mode with 50 mm soil moisture-holding capacity is about 10 % and 13 % of precipitation for Ketar and Meki subbasins respectively [40]. In another similar study revealed that the average annual recharge assessed using the chloride mass balance method at Butajira (Meki subbasin) and Ziway area were about 10 % and 0.68 % of the precipitation respectively) [55] and the recharge estimates from different methods were summarized in Table 4. Table 4Summary recharge estimates using different methods.Table 4SubbasinWEAP modelSimple water balance methodBucket modelChloride methodAverageKetar (Eastern)4.6 %–1 %–3 %Meki (Western)10 %13 %13 %10 %12 %Rift floor (Central)–––0.68 % Summary recharge estimates using different methods. Existing studies on groundwater recharge estimated about 5 % and 1 % of rainfall at the highlands and lowlands areas respectively [56]. Hence, for this study, groundwater recharge was estimated using topographic regions that were higher than 2000 masl, 1750–2000 masl and lower than 1750 masl. The rates of recharge were estimated as 5 %, 3 %, and 1 %, respectively. In this study the initial recharge has been allocated in the model ranges from 3.86 × 10−5 to 3.1 × 10−4 m/day based on the topographic zones (Fig. 4b). The least recharge is in areas where less permeable clay and lacustrine soils exist. Much of the watersheds’ highland plateaus and the permeable volcanic areas get the maximum recharge. Model calibration The study basin groundwater flow model was calibrated using the trial-and-error technique by altering initial estimates of aquifer properties, hydraulic conductivities, recharge, and discharge to become the best fit among simulated hydraulic heads and observed water levels. It was calibrated under steady-state conditions. The overall mean variance between simulated and observed heads was expressed, as given in Ref. [49] using statistical comparison. These are commonly used statistical comparison criteria: Mean Error (ME), Absolute Mean Error (AME), Root Mean Squared Error (RMSE), Nash-Sutcliff coefficient (NSE), and Coefficient of Determination (R2). Results and discussions Model calibration The objective of model calibration is to obtain parameter values that minimize the difference between calculated and field measured variables by changing model input values in an effort to match simulated and actual conditions within an acceptable range. Model calibration is usually an iterative process that is attempted. The major parameters to be adjusted or modified during model calibration were the hydraulic conductivity and groundwater recharge flux inputs. Finally, the simulated hydraulic conductivity and recharge flux were presented in (Fig. 5) and (Fig. 6), respectively. Simulated hydraulic conductivity of layer-1(a) and layer-2 (b). Simulated recharge flux. Steady-state model simulation The model performance was assessed by (1) model performance statistics (2) comparison of groundwater hydraulic heads of simulated and observed data as scatter plot or spatial distribution (3) histogram of residuals estimated. The calibrated result indicated that the performance statistics between the simulated groundwater head and observed head of 163 the observations were presented in Table 5 and (Fig. 7). These, model performance statistics showed that a good-fit between simulated and measured heads. The simulated hydraulic head distribution for layer-1 and layer-2 are presented in (Fig. 9). The groundwater flow contours gradient showed groundwater flow direction in the aquifers and subdued replica of the basin topographic configuration. The comparison of simulated and observed groundwater heads were presented in (Fig. 10). The result of histogram residual error was shown in (Fig. 8). The histogram residual of simulated and observed hydraulic heads revealed that 65 % and 82 % of wells were found to be bounded within ±20 m and ±30 m error, respectively. Table 5Model performance statistics.Table 5Model performance StatisticUnitValueR2–0.98NSE–0.97MEmeter−8.3MAEmeter17.5RMSEmeter21.9NRMSE%1.28 % Model performance statistics. Linear regression of observed and simulated hydraulic heads. Simulated hydraulic head distribution of steady state for (a) layer 1 and (b) layer 2. (a) Simulated groundwater heads and (b) observed groundwater heads. Histogram of residuals of estimated. Model sensitivity analysis Following successful model calibration, a sensitivity analysis was undertaken to quantify the uncertainties associated with input variables, stresses, and boundary conditions [49]. This analysis served to reveal the model's response to variations within the calibrated parameter ranges, thereby identifying those parameters that exert the most significant influence on the simulated outcomes. These insights can be employed to guide future data collection efforts, ultimately minimizing model errors. The model's sensitivity to a particular parameter is then determined by evaluating the relative change in the output variable in response to a unit change in the input parameter holding all the others constant. A greater change in the output signifies a more responsive model, indicating a higher degree of influence from that specific parameter on the overall results. In this study, a sensitivity analysis of the groundwater model was conducted to assess the overall model efficiency, with a specific focus on aquifer parameters. The analysis involved systematically increasing or decreasing the horizontal hydraulic conductivity and groundwater recharge rates by a predetermined percentage. The resulting changes in hydraulic heads were then compared against observed data and established model performance metrics to evaluate the sensitivity of the model to each parameter. The sensitivity analysis revealed that ±20 % perturbations in either the hydraulic conductivity values or the recharge rates exerted a significant influence on the distribution of groundwater heads within the basin. The model demonstrated sensitivity to both increases and decreases in these parameters; however, a greater responsiveness was observed for increased recharge flux compared to equivalent increases in hydraulic conductivity (Fig. 11). Sensitivity analyses of hydraulic conductivity and recharge. Groundwater balance of model domain This study employs water budget analysis to assess the efficiency of the groundwater flow model. The model calculates the total volume of water entering and exiting the aquifer systems across the entire modeling domain. This approach serves as an indirect validation technique, where each component of the water budget (inflow and outflow terms) is evaluated relative to the simulated total inflows and outflows. Under natural conditions, the inflow components encompass constant head boundaries, general head boundaries, and river leakage. Conversely, outflows consist of constant head boundaries, general head boundaries, river leakage, and well abstractions. The model results indicate a high degree of consistency between inflow and outflow terms, signifying a steady-state modeling condition with a minimal discrepancy of −0.75 % (Table 6). This suggests that the model effectively replicates the basin's natural water balance. The results of the water budget of layer −1 and Layer-2 are shown in Table 7, Table 8. Table 6Water budget (daily) of the entire study basin.Table 6Water budget of the whole model domain:Flow TermInOutIn-OutStorage0.0000000E+000.0000000E+000.0000000E+00Constant head2.4593317E+051.4222762E+06−1.1763431E+06Wells0.0000000E+001.6293270E+04−1.6293271E+04Recharge1.2205944E+060.0000000E+001.2205944E+06ET0.0000000E+000.0000000E+000.0000000E+00River leakage8.8733083E+021.9116510E+04−1.8229180E+04Head Dependent Bounds0.0000000E+002.0778653E+04−2.0778652E+04Sum1.4674149E+061.4784646E+06−1.1049750E+04Discrepancy [%] −0.75 Water budget (daily) of the entire study basin. Table 7Water budget (daily) of layer-1 of the basin.Table 7Water budget of layer-1Flow TermInOutIn-OutStorage0.0000000E+000.0000000E+000.0000000E+00Constant head1.0203565E+051.2808739E+06−1.1788382E+06Exchange (Lower)6.2322428E+056.3750075E+05−1.4276472E+04Wells0.0000000E+001.6293270E+04−1.6293271E+04Recharge1.2256769E+060.0000000E+001.2256769E+06ET0.0000000E+000.0000000E+000.0000000E+00River leakage8.8441306E+022.1701840E+04−2.0817428E+04Head Dependent Bounds0.0000000E+006.5097116E+03−6.5097114 E+03Sum of the layer1.9518212 E+061.9628795E+06−1.1058250E+04Discrepancy [%] −0.56 Water budget (daily) of layer-1 of the basin. Table 8Water budget (daily) of layer-2 of the basin.Table 8Water budget of layer-2Flow TermInOutIn-OutStorage0.0000000E+000.0000000E+000.0000000E+00Constant head0.0000000E+000.0000000E+000.0000000E+00Horiz.Exchange0.0000000E+000.0000000E+000.0000000E+00Exchange (Upper)6.3750075E+056.2322428E+051.4276472E+04Recharge0.0000000E+000.0000000E+000.0000000E+00ET0.0000000E+000.0000000E+000.0000000E+00Head Dependent Bounds0.0000000E+001.4276435E+04−1.4276436E+04Sum of the layer6.3750075E+056.3750069E+05−6.2500000E-02Discrepancy [%] 0.00 Water budget (daily) of layer-2 of the basin. The calibrated findings from the steady state model water balance indicated that the modeled groundwater recharge stood at approximately 445 million cubic meters (MCM). In contrast, the initial estimate for groundwater recharge was 555 MCM. The primary inputs into the groundwater system were groundwater recharge and the fixed head boundary, contributing 445 MCM/year and 90 MCM/year, respectively. Conversely, the principal outflow (discharge) from the groundwater system was through the fixed head boundary, with a simulated rate of 519 MCM/year. Research limitations The study basin is made up of volcano-sedimentary rock groups highly modified by volcano-tectonic events since its formation to now. This creates variable hydrogeological and hydrological parameters. Besides limitation of data availability, the numerical groundwater model is a device intended to approximate the field scale conditions of groundwater with a reasonable limit. It may not reproduce the complexity of nature as a whole. Hence, the following limitations may apparently exist: hydrogeological and hydrological properties, variability, boundary conditions, simplification used in conceptual models, spatial discretization de-emphasizes local variability and simultaneous calibration of fluxes and conductivities may bring uncertainties. Conclusions The Abijata-Langano-Ziway-Lake Basin (ALZLB) exhibits a geologically complex setting within the Ethiopian Rift Valley, leading to highly variable aquifer properties and intricate hydrogeologic characteristics. To elucidate these complexities, a numerical groundwater flow model was constructed utilizing MODFLOW software. Model parameters and boundary conditions were accurately selected to align with the established conceptual model of the basin. Calibration of the model outputs confirmed its efficacy in accurately representing aquifer properties, resulting in a strong fit. Notably, the water balance components derived from the calibrated model exhibited closely consistent with the conceptual model and associated assumptions, with a minimal discrepancy of −0.75 %. This result suggests that the model effectively replicates the basin's natural water balance. Furthermore, the model simulations revealed a significant contribution of the basin's groundwater system to sustaining lakes, wetlands, and river ecosystems. It is imperative to acknowledge the ALZLB's status as one of Ethiopia's most environmentally sensitive areas, characterized by fragile ecosystems. Given its designation as a closed hydrological basin, even relatively minor interventions in water resource management strategies have the potential to induce far-reaching ecological consequences. A key outcome of this study is the establishment of a robust groundwater flow model. This model incorporates critical hydrogeological parameters, including hydraulic conductivity, aquifer characteristics, and recharge rates, to provide a comprehensive understanding of the basin's groundwater system. The model's outputs provide valuable insights for water resource management and planning. These include the direction and magnitude of groundwater flow, identification of recharge and discharge zones, and potential areas susceptible to contamination or depletion. Furthermore, the model's predictive capabilities allow for the simulation of various scenarios, such as changes in pumping or land use practices. This enables informed decision-making towards sustainable groundwater management, mitigating potential adverse impacts. In essence, the model serves as a cornerstone for effective groundwater resource assessment and management, providing a scientific foundation for sustainable water practices. CRediT authorship contribution statement Debele Abera: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Behailu Berehanu: Writing – review & editing, Validation, Methodology, Formal analysis, Conceptualization. Tenalem Ayenew: Writing – review & editing, Visualization, Validation, Supervision, Project administration, Conceptualization. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Ethics approval and consent to participate Not applicable. Consent for publication All authors agreed and approved the manuscript for publication in journal of Heliyon. Funding No funding has been received for this study. Declaration of competing interest The authors declared that there are no known competing personal or financial relationships that could have appeared to influence the publication of “Groundwater Flow System in the Abijata-Langano-Ziway Lakes Basin, Ethiopia, on Heliyon. Acknowledgments We want to express our greatest appreciation to the National Meteorological Agency (NMA), the Ministry of Water and Energy (MOWE), Oromia water Bureau, The Ministry of Mines and Petroleum (MoMP) of Ethiopia, Ethiopian Institute of Geological Survey for providing the necessary data. The opinions expressed herein are the authors' own and do not necessarily express the view of above mentioned institutions. Abbreviations the Abijata-Langano-Ziway Lakes Basin The Water Evaluation and Planning Tool meter Meter above sea level Million Cubic Meter References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.