Advertisement

Hydrogeology Journal

, Volume 27, Issue 6, pp 2121–2136 | Cite as

An integrated approach for aquifer characterization and groundwater productivity evaluation in the Lake Haramaya watershed, Ethiopia

  • Haile A. ShishayeEmail author
  • Douglas R. Tait
  • Kevin M. Befus
  • Damien T. Maher
Report

Abstract

Accurate characterization of aquifers remains challenging for large-scale systems because of the spatial heterogeneity of hydraulic properties and temporal variability of hydrologic inputs. This study highlights the importance of integrating geological, hydrogeological and geophysical approaches to characterize an aquifer and evaluate groundwater productivity. Data from geological maps, drill logs, a pumping test, vertical electrical soundings (VES) and different field hydrogeological studies were combined and applied to a heavily extracted aquifer system—Lake Haramaya watershed, Ethiopia. From the geological characterization, the aquifer was found to be a single heterogeneous and anisotropic unconfined unit. Significant differences were found between the three-dimensional geological models of the aquifer developed from the drill logs and VES data; the VES data were likely affected by moisture content. The pumping-test and VES data were combined to estimate transmissivity (T; 126.5 ± 25.8 m2/day) and hydraulic conductivity (K; 4.1 ± 1.0 m/day). This combined use allowed for a reduction in uncertainty (40.1% for T and 33.3% for K) compared with values estimated from the VES data alone. The combined approach also allowed for much greater spatial coverage and a higher resolution characterization of the aquifer. The available volume of groundwater resource in the system was estimated at ~0.62 ± 0.09 km3. The groundwater extraction rate was ~30,120 m3/day, approximately double the estimated sustainable yield of the aquifer (15,720 m3/day). This showed that the current exploitation rate could exhaust groundwater resources in 27–32 years and should be reduced by 50% to ensure sustainability of the groundwater resource.

Keywords

Ethiopia Groundwater potential Hydrogeology Hydraulic properties Lithological modelling 

Une approche intégrée pour la caractérisation et l’évaluation de la productivité d’un aquifère dans le bassin versant du lac Haramaya, Ethiopie

Résumé

La caractérisation précise des systèmes aquifères reste difficile pour les grands systèmes, en raison de l’hétérogénéité spatiale des propriétés hydrauliques et de la variabilité temporelle des apports hydrologiques. Cette étude souligne l’importance d’approches intégrant la géologie, l’hydrogéologie et la géophysique pour caractériser un aquifère et évaluer sa productivité. Les données issues des cartes géologiques, des logs de forages, d’un pompage d’essai, de sondages électriques verticaux (SEV), et de différentes études hydrogéologiques de terrain ont été associées et appliquées à un système aquifère intensément exploité: le bassin versant du lac Haramaya, en Ethiopie. A partir de la caractérisation géologique, l’aquifère est apparu comme étant une unité hétérogène et anisotrope à nappe libre. Des différences significatives ont été mises en évidence entre les modèles géologiques 3D de l’aquifère, réalisés à partir des logs de forages et des données SEV; les données SEV étaient susceptibles d’être influencées par la teneur en eau. Le pompage d’essai et les données SEV ont été associées pour estimer la transmissivité (T; 126.5 ± 25.8 m2/jour) et la conductivité hydraulique (K; 4.1 ± 1.0 m/jour). Cette utilisation conjointe a permis de réduire les incertitudes (40.1 % pour T et 33.3 % pour K) en comparaison des valeurs estimées à partir des seules données SEV. Cette approche conjointe a également permis une bien meilleure couverture spatiale et une caractérisation de l’aquifère avec une plus grande résolution. Le volume disponible de ressources en eau souterraine dans le système a été estimé à environ 0.62 ± 0.09 km3. Le taux de prélèvement d’eau souterraine a été d’environ 30,120 m3/jour, soit à peu près le double de l’estimation du taux de renouvellement de l’aquifère (15,720 m3/jour). Cela a montré que le taux d’exploitation actuel pourrait épuiser les ressources en eau souterraine, d’ici 27 à 32 ans, et devrait être réduit de 50% pour assurer la durabilité de la ressource en eau souterraine.

Un enfoque integrado para la caracterización de acuíferos y la evaluación de la productividad del agua subterránea en la cuenca del Lago Haramaya, Etiopía

Resumen

La caracterización precisa de los acuíferos sigue siendo un reto para los sistemas a gran escala debido a la heterogeneidad espacial de las propiedades hidráulicas y a la variabilidad temporal de los aportes hidrológicos. Este estudio destaca la importancia de integrar métodos geológicos, hidrogeológicos y geofísicos para caracterizar un acuífero y evaluar la productividad del agua subterránea. Los datos de mapas geológicos, registros de perforación, ensayos de bombeo, sondeos eléctricos verticales (VES) y diferentes estudios hidrogeológicos de campo se combinaron y aplicaron a un sistema acuífero de gran extracción: la cuenca del Lago Haramaya, Etiopía. A partir de la caracterización geológica, se encontró que el acuífero era una sola unidad no confinada heterogénea y anisotrópica. Se encontraron diferencias significativas entre los modelos geológicos tridimensionales del acuífero desarrollados a partir de los registros de perforación y los datos VES; los datos VES probablemente se vieron afectados por el contenido de humedad. Los datos del ensayo de bombeo y de los VES se combinaron para estimar la transmisividad (T; 126.5 ± 25.8 m2/día) y la conductividad hidráulica (K; 4.1 ± 1.0 m/día). Este uso combinado permitió una reducción de la incertidumbre (40.1% para T y 33.3% para K) en comparación con los valores estimados a partir de los datos VES solamente. El enfoque combinado también permitió una cobertura espacial mucho mayor y una caracterización del acuífero de mayor resolución. El volumen disponible de recursos de agua subterránea en el sistema se estimó en ~0.62 ± 0.09 km3. La tasa de extracción de agua subterránea fue de ~30,120 m3/día, aproximadamente el doble del rendimiento sostenible estimado del acuífero (15,720 m3/día). Esto mostró que la tasa de explotación actual podría agotar los recursos de agua subterránea en 27–32 años y debería reducirse en un 50% para asegurar la sostenibilidad de los recursos de agua subterránea.

埃塞俄比亚哈拉玛雅湖流域含水层特征和地下水生产力的综合评估方法

摘要

由于水力特性的空间异质性和水文输入信息的时变性, 大区域尺度含水层系统的准确表征是难点。该研究强调了整合地质, 水文地质和地球物理方法以表征含水层和评估地下水生产力的重要性。将来自地质图, 钻探测井, 抽水试验, 垂直电测深(VES)和不同野外水文地质研究的数据综合考虑, 并将方法应用于埃塞俄比亚哈拉玛亚湖流域的高度开采含水层系统。从地质特征来看, 含水层被认为是一个单一的非均质各向异性无限含水单元。钻井记录和VES数据建立的含水层三维地质模型之间存在显著差异; VES数据可能受水分含量的影响。将抽水试验和VES数据一起考虑估计出导水系数(T; 126.5 ± 25.8 m 2 /天)和渗透系数(K; 4.1 ± 1.0m /天)。与仅从VES数据估计结果相比, 综合分析方法降低了不确定性(T为40.1%, K为33.3%)。这种综合分析方法还可考虑更大空间范围和更高分辨率的含水层特征。系统中可利用的地下水资源量估计为0.62 ± 0.09 km3。地下水开采量约为30,120立方米/天, 约为含水层可持续开采量的两倍(15,720立方米/天)。这表明目前的开采量可能会在27–32年耗尽地下水资源, 应减少50%, 以确保地下水资源利用的可持续性。

Uma abordagem integrada para caracterização de aquífero e avaliação da produtividade de água subterrânea na bacia do Lago Haramaya, Etiópia

Resumo

A caracterização precisa de um aquífero continua a ser um desafio para sistemas de grande escala devido à heterogeneidade espacial das propriedades hidráulicas e a variabilidade temporal das recargas. Este estudo destaca a importância de uma integração da abordagem geológica, hidrogeológica e geofísica para caracterizar um aquífero e avaliar a produtividade da água subterrânea. Os dados extraídos dos mapas geológicos, perfil de poços, teste de bombeamento, sondagens elétricas verticais (SEV) e diferentes estudos de campo hidrogeológicos foram associados e utilizados para uma intensa extração do sistema aquífero – bacia do Lago Haramaya, Etiópia. A partir da caracterização geológica, o aquífero foi descrito como uma única unidade heterogênea e anisotrópica não confinada. Diferenças significativas foram encontradas entre os modelos geológicos de três dimensões do aquífero desenvolvidos a partir dos perfis de poço e dos dados de SEV; os dados de SEV provavelmente foram afetados pelo teor de umidade. O teste de bombeamento e dados de SEV foram combinados para estimar a transmissividade (T; 126.5 ± 25.8 m2/dia) e condutividade hidráulica (K; 4.1 ± 1.0 m/dia). Esta combinação de dados permitiu a redução da incerteza (40.1% para T e 33.3% para K) quando comparado apenas com os valores estimados a partir dos dados de SEV. A abordagem integrada também possibilitou uma maior cobertura espacial e uma caracterização de alta resolução do aquífero. O volume disponível de água subterrânea no sistema foi estimado em aproximadamente 0.62 ± 0.09 km3. A taxa de extração de água subterrânea foi ~30,120 m3/dia, aproximadamente o dobro do valor estimado de produtividade sustentável para o aquífero (15,720 m3/dia). Isso mostrou que a atual taxa de explotação poderia exaurir o recurso hídrico subterrâneo dentro de 27–32 anos e deveria ser reduzida em 50% para assegurar a sustentabilidade deste recurso.

Introduction

The quantification of groundwater resources depends on accurate hydrogeological information that can be collected through a wide range of approaches. Reliable knowledge of subsurface geologic characteristics is essential in designing effective and sustainable groundwater management strategies because groundwater flow is controlled by the geological framework of the aquifer (Shishaye 2015; Ezeh 2012). Proper evaluation of the storage and transmission properties of the different geological materials is necessary to measure the groundwater potential of an aquifer system (Kebede 2013). This includes characterizing the source and flows of water in aquifers (Nwosu et al. 2013; Kebede 2013; Wang et al. 2010), estimating groundwater productivity (Demlie and Titus 2015; Masvopo 2008) and predicting groundwater sustainability (Cao et al. 2013).

A number of analytical and conventional methods are available to characterize an aquifer and the groundwater availability in an aquifer system (Paola et al. 2005). One of the techniques can be labeled broadly as “geological characterization”, which includes three-dimensional (3D) lithological and stratigraphic mapping. Three-dimensional geological mapping can be applied to estimate aquifer hydraulic properties, delineate aquifer volumes and boundaries, and for adequate understanding of the structural and lithological complexity of hydrogeological units (Marchant et al. 2011; Wang et al. 2010). Tartarello et al. (2017) applied this method to characterize the underground reservoirs in Sardinia (Italy), where they identified an alluvial plain overlaid on low-grade metamorphic sandstones and carbonate rocks, reporting that the potential reservoir was a heterogeneous, dual porosity/permeability system, which is represented by matrix and fracture network. Israde-Alcantara et al. (2005) also used the geological method to characterize underground reservoirs in a municipal landfill and identify the associated environmental implications in central Mexico, where they found that the site lacks impermeable subsurface layers; however, in order to quantify the hydraulic properties of an aquifer system, hydrogeological techniques must be applied.

Hydrogeological techniques are used to define the hydraulic parameters of an aquifer system such as hydraulic conductivity (K), transmissivity (T), specific yield (SY), specific storage (SS) and storage coefficient (S) (Hunkeler 2010). This can be achieved through techniques that use well drill logs, pumping test analysis and/or groundwater flow modelling (Okiongbo and Akpofure 2012). Turk et al. (2015) applied these techniques along with 3D geological modelling approaches to characterize the groundwater storage in a trans-boundary aquifer that stretches between Slovenia and Italy. They found a new explanation for the hydrogeological setup of a local spring with a much larger water catchment identified. Alexander et al. (2011) also used hydrogeological techniques including pumping test analysis to estimate the extent of the hydraulic properties of a heterogeneous aquifer system in Waterloo, Canada, where they reported that the method was helpful to delineate the distribution of aquifer hydraulic properties in highly heterogeneous fluvial deposits. However, the hydrogeological method relies on the presence of labour intensive and potentially costly wells that are not always available.

Recently, surface geophysical methods have been developed as complementary approaches to field hydrogeological techniques (Herckenrath et al. 2012) and have been applied to search for potable water, evaluate formation strata and solve hydrogeological problems (Adeoti et al. 2012). One method, vertical electrical sounding (VES), provides a relatively simple and effective way of estimating aquifer properties (Abdullahi et al. 2014). It can provide valuable information regarding the vertical successions of subsurface geo-materials in terms of their individual thicknesses, corresponding resistivity values and the protective capacity of overlying confining units using the similarity between electrical current and hydraulic transmissivities (Anomohanran 2013; Henriet 1976). Ifabiy et al. (2016) used this method to understand the influence of hydrogeological characteristics on groundwater yield of shallow wells in a regolith aquifer in Nigeria, and reported that aquifer T and the rate of drawdown should be taken into account when constructing a well to analyze aquifer productivity and sustainability of the resource. Korowe et al. (2011) also used VES to estimate the T and K values of an aquifer in Jangaon sub-watershed, India, by determining a theoretical relationship between geo-electrical data and hydraulic parameters including the linear relationship between T and a formation factor. However, in order to reduce uncertainties in identifying the subsurface resistive properties, the VES method requires additional information from geological maps and pumping test data (Nwosu et al. 2013).

The groundwater in the Lake Haramaya watershed, Ethiopia, has been under intensive use in the past two decades. With little variation in climatological drivers in the watershed over that period (Alemayehu et al. 2006), the declining water-table elevations in the region suggest that current groundwater use may be in disequilibrium with recharge (Shishaye and Nagari 2016). However, despite the importance of the groundwater resource in the area, the geologic and hydrogeologic settings, and the groundwater productivity and safe yield of the aquifer remain uninvestigated. This study uses an integrated aquifer characterization approach, where a suite of geological, hydrogeological and geophysical techniques were employed to characterize the system, reduce uncertainties associated with each method, increase sampling efficiency and provide for a more comprehensive characterization of the aquifer system. Aquifer characterization results from geological mapping were integrated with well drill logs, pumping test, and VES data to construct the 3D hydrogeological model of the Lake Haramaya watershed and to analyze the available water resources within a sustainability framework.

Materials and methods

Description of the study area

The Lake Haramaya watershed (4646208–4721625E, 1,035,000–1,071,089 N, WGS 1984 Web Mercator (auxiliary sphere)) is located in Haramaya Woreda, Oromia Regional State, Ethiopia (Fig. 1). It is 21 km northwest of Harar town and 505 km east of Addis Ababa, Ethiopia. The total area of the watershed is 51.7 km2 and the elevation ranges from 1,976 to 2,382 m above sea level. It is characterized by a sub-tropical agro-climatic zone that receives mean annual rainfall of 776 mm and temperatures that range from 10 to 25 °C, with rainy season extending from April to September.
Fig. 1

Location map of the study area, Lake Haramaya watershed, Ethiopia, showing the locations of 16 wells, 2 coring sites and 41 VES stations. The world projected coordinate system, WGS 1984 Web Mercator (auxiliary sphere), is used

The Lake Haramaya watershed is part of the Harar Plateau, in the upper Wabi-Shebele Basin. There are six different land use types in the watershed: cultivated land (78.3%), grazing land (7.6%), forest (0.6%), settlement (4.5%), shrub (4.6%) and wetlands (4.5%) (Tadesse and Abdulaziz 2009). Khat and vegetables are the major crop types grown in the area and extensive irrigation is used.

Surface geology

To create an accurate surface geological map, topographic maps (Ethiopian Mapping Agency), basin geological maps (Ethiopian Geological Survey) and ground-truth data from more than 800 GPS coordinates within the watershed were digitized using Global Mapper 18 software (Blue Marble Geographics 2018). The surface geology map was used as a validation tool for the surface boundary of the 3D geological models and to characterize the geology of the catchment boundaries because the terrain near the catchment boundaries was too steep for well construction or to conduct geophysical surveys.

Aquifer characterization and 3D geological modelling

Drill logs and coring data

Drill logs were sourced from individual drilling companies, from 17 wells ranging in depth from 43 to 63 m, with records at 2-m intervals. Two 27-m-deep sediment cores were also collected from the lacustrine deposit area (Fig. 1) and sectioned at 1.5-m internals to have a more accurate profile of the clay layer. The unsampled areas between the wells were interpolated from the collected data using a lateral blending algorithm and superface and subface dimensions to produce the 3D lithological model, and inverse distance weighting with a base-to-top modelling sequence to construct the 3D stratigraphic model of the aquifer system using Rockworks-17 (RockWorks 2017). The lateral blending method provides a means for creating models of data that are horizontally contiguous but numerically discrete. This method first assigns the voxels immediately surrounding each borehole the closest lithology or a real number value. Then, the analysis moves out by a voxel and assigns the next kernel of voxels to the closest lithology value spatially. After it reaches a point about a third of the way to the neighboring data points, it applies a randomizing algorithm in the center areas to minimize the abrupt changes between material types. In this case, the superface and subface dimensions are the lithology (top and bottom, respectively) menus used to clip the block model into a surface based on the uppermost elevations and the lowermost elevations of the rock types, respectively. Therefore, the superface dimension used the digital elevation model (DEM) of the area, while the subface dimension used the maximum depths of the wells and VESs to clip the model. Finally, the surface geological map was used to validate the model results (lithological layers, especially of the upper most layers).

Pumping test

The pumping test data (constant discharge and recovery tests) from 17 wells in the alluvial deposit area were used to calculate T, K, SY, aquifer anisotropy ratio (KV/KH) and the specific yield to storativity ratio (SY/S) using AquiferTest Pro software (Schlumberger Water Services 2015) with the Neuman method for an unconfined aquifer (Neuman et al. 2007):
$$ s=\frac{Q}{4\uppi T}\cdotp W\left({U}_{\mathrm{A}},{U}_{\mathrm{B}},\mathrm{\varTheta}\right) $$
(1)
where s is drawdown, Q is well discharge (m3/s), T is transmissivity (m2/s), W(UA, UB, Ɵ) is the unconfined well function, \( {U}_{\mathrm{A}}\ \mathrm{is}\ \frac{r^2S}{4 Tt} \) (type A curve for early time), \( {U}_{\mathrm{B}}\ \mathrm{is}\frac{r^2{S}_{\mathrm{y}}}{4 Tt} \) (type B curve for later time), Ɵ is \( \frac{r^2{K}_{\mathrm{V}}}{b^2{K}_{\mathrm{H}}} \) (type A curve for early time), r is the distance of observation (m), S is storativity, t is flow time (s), KV is the vertical hydraulic conductivity (m/s) and KH is the horizontal hydraulic conductivity (m/s). The input data for the model were the drawdown versus time records, constant discharge rates and the dimensions of each well. The model then estimated the average values of the aquifer hydraulic properties throughout the productive aquifer thickness.

Vertical electrical sounding (VES)

The electrical resistivity survey involved VES, which is based on transmitting current between one electrode pair while measuring the potentials between another electrode pair. The stratification of the ground was investigated by varying the electrode separations as the depth of penetration is proportional to the separation of the electrodes. The measurements were taken with a portable resistivity meter (SAS 1000). The equipment was first calibrated and validated with the help of the drill logs to maintain instrument accuracy. The Schlumberger electrode configuration with maximum half-current electrode separation of 220 m was used in 41 VES stations in the study watershed (Fig. 1). The resistivity data were inverted using IPI2win software (Kurniawan 2009) to estimate the thickness of each geo-electric layer, map the basement topography and define the vertical transformations of the geological materials in the area. The apparent resistivity (ρa) values were calculated using (Ibuot et al. 2013):
$$ {\rho}_{\mathrm{a}}=\left[\frac{{\left(\frac{\mathrm{AB}}{2}\right)}^2-{\left(\frac{\mathrm{MN}}{2}\right)}^2}{\mathrm{MN}}\right]\cdotp {R}_{\mathrm{a}} $$
(2)
where AB is the distance between the two current electrodes, MN is the distance between the two potential electrodes, and Ra is the apparent electrical resistance (Ω) measured from the equipment. The IPI2win software calculates the standard errors for layer resistivity and thicknesses as random errors during inversion and displays it as a percent (%) error (Khalid et al. 2018). The errors ranged from 0.23 to 2.82%, and averaged 1.2 ± 0.1%, which shows defendable inversion results (Khalid et al. 2018). The inverted VES data, in turn, were interpreted in terms of geology and hydrogeology of the area; however, the interpretation of geo-electrical resistivity data depends on different environmental factors and various geological parameters such as the mineral content, inter-granular compaction, porosity, degree of water saturation in the rocks, etc. (Gonzalez-Alvarez et al. 2016; Archie 1942). Therefore, interpreting the inverted resistivity data into geological materials was supported by the drill logs of nearby wells and resistivity interpreting guidelines (Gonzalez-Alvarez et al. 2016). The interpreted VES data were then used to develop the 3D lithological and stratigraphic models of the aquifer system using RockWorks 17, following similar interpolation procedures with the drill logs. Furthermore, the interpreted VES and the drill log data were also combined to produce more comprehensive 3D lithological and stratigraphic models, which cover larger area because of the more expansive spatial coverage of the VES data (Fig. 1) and a relatively validated model (compared to the use of only VES data) because of the addition of actual drill log data as an input to the software (RockWorks 17).
The VES data were also used to estimate the hydraulic properties (T and K) of the aquifer system through the Dar-Zarrouk parameters (transverse resistance and longitudinal conductance) using the equations given by Sabet (1975) in a given geo-electric layer using:
$$ \mathrm{TR}=h\cdotp {\rho}_{\mathrm{a}}=0.19{(T)}^{1.28} $$
(3)
$$ S=\frac{h}{\rho_{\mathrm{a}}} $$
(4)
$$ K=\frac{T}{b} $$
(5)
and equations developed for n number of layers using:
$$ \sum S={\sum}_{i=1}^n\left(\frac{h_i}{\rho_i}\right)=\frac{h_1}{\rho_1}+\frac{h_2}{\rho_2}+\frac{h_3}{\rho_3}+\dots \frac{h_n}{\rho_n} $$
(6)
$$ \sum \mathrm{TR}={\sum}_{i=1}^n\left({h}_i\cdotp {\rho}_i\right)=\left({h}_1\cdotp {\rho}_1\right)+\left({h}_2\cdotp {\rho}_2\right)+\left({h}_3\cdotp {\rho}_3\right)+\dots \left({h}_n\cdotp {\rho}_n\right) $$
(7)
where, TR is transverse resistance (Ωm2), S is electrical longitudinal conductance (Ω−1), a is apparent resistivity (Ωm), h is thickness of a geo-electric layer (m), T is transmissivity (m2/day) and b is thickness of the aquifer (m). The sum of all the electrical longitudinal conductance (∑S) and of the transverse resistance (∑TR) for a layered ground are called the Dar-Zarrouk “function” and “variable”, respectively. The uncertainty levels of the calculated aquifer hydraulic properties caused by the resistivity inversion errors (average of 1.2 ± 0.1%) were ignored in these calculations, as they were assumed to be negligible. Therefore, the estimated values of the hydraulic properties were compared with those estimated from pumping test data and were averaged to get the mean values for groundwater modelling purposes. The protective capacity of the aquifer system was also estimated from the electrical longitudinal conductance values. In this case, aquifer protective capacity is the ability of the upper geo-electrical layer of an aquifer to protect the aquifer system from surface contaminants (Obiora et al. 2015).

Once the hydraulic properties of the aquifer system were estimated from the pumping test and VES data, the grain sizes of the productive layers were identified from the 3D geological maps produced from the integrated drill logs and VES data. Then, in order to constrain overestimations and identify data outliers, comparisons were made to make sure that the estimated aquifer hydraulic property values are within the expected ranges of similar lithologies from the literature (Domenico and Schwartz 1990; Hwang et al. 2017).

Groundwater productivity and sustainability evaluations

Groundwater productivity

The potential productivity of the aquifer in terms of water supply was evaluated based on the grain sizes of each formation, hydraulic properties (K and SY) and geomorphological settings. The productive aquifer formations (fine-coarse sand, gravely sand and weathered granite) were evaluated from the 3D geological models, water level measurements and hydrogeological characterizations. The available amount of groundwater in the aquifer system, at the time of measurement, was then calculated using:
$$ \mathrm{GP}=V\bullet {S}_{\mathrm{Y}} $$
(8)
where, GP is the potentially available water volume in the aquifer system, V is volume of potential aquifer formations and SY is the specific yield of the aquifer system.

Groundwater sustainability

The response of the aquifer to pumping and the sustainability of the groundwater resource over a 50 years simulation period (stress period) was modelled using MODFLOW (Harbaugh 2005). The average hydraulic properties of the aquifer system (K and SY) from the hydrogeological and geophysical investigations and calculated groundwater recharge rates from previous studies (Tadesse and Abdulaziz 2009) were used as inputs to the software. A no-flow boundary condition was used along all model sides, for the watershed is bounded by mountains where the basement rock outcrops in all directions and were assumed to represent hydrologic divides. The recharge boundary condition was also used to simulate surficially distributed recharge to the aquifer system, while the model grid used 100 m × 100 m cells and the thickness of the aquifer system, and the hydrostratigraphic model was used to set the thickness of the unconfined aquifer. Ten years of water level records from three recorded monitoring wells (W7, W8, W9; Fig. 1) in the alluvial deposit area—Table S1 of the electronic supplementary material (ESM)—were used for model calibration and validation (Fig. S1 of the ESM). This was because the records in the remaining wells were not reliable; however, to show the seasonal changes in water levels, average monthly water-level data were plotted and evaluated accordingly, where a decreasing trend was depicted, even though there were seasonal fluctuations because of the direct recharge effects (Fig. S1 of the ESM). Therefore, the trend suggested that using annual averages for model calibration and validation purposes would minimally affect the final long-term predictions of the model.

As the water table in the area is in disequilibrium due to high pumping rates, water-table levels could not be used as a boundary condition to model groundwater flow direction. In areas where extensive pumping is occurring, the direction of groundwater flow can be controlled by the extents of pumping (Barackman and Brusseau 2003). As such, surface elevations were used to create the final grid for groundwater flow direction (Gleeson et al. 2011; Haitjema and Mitchell-Bruker 2005; Chebet 2012; Jenson and Domingue 1988; Greenlee 1987) and mapped using ESRI ArcMap 10.5. This was validated by ensuring that the water table was topographically controlled by estimating the dimensionless criterion introduced by Haitjema and Mitchell-Bruker (2005), called the water-table ratio (WTR):
$$ \log \left(\mathrm{WTR}\right)=\log \left(\frac{R{L}^2}{mKHd}\right)\left\{\begin{array}{c}>0,\mathrm{for}\ \mathrm{topofraphy}\ \mathrm{controlled}\ \\ {}<0,\mathrm{for}\ \mathrm{recharge}\ \mathrm{controlled}\kern1.50em \end{array}\right. $$
(9)
where, R (m/day) is the areal recharge rate, L (m) is the distance between surface water bodies (500 m), K (m/day) is the hydraulic conductivity, H (m) is the average vertical extent of the groundwater flow system, d (m) is the maximum terrain rise (100 m), and m (dimensionless) is either 8 or 16, depending on the flow problem being one-dimensional (1D) or radially symmetric, respectively. WTR was found to be 4.1 × 10−4, which is greater than zero and implies that the water table in the area is topography controlled and it is a replica of the topography.
Therefore, the groundwater flow model was achieved by calculating the direction of flow from every cell in the DEM of the watershed based on the maximum drop principles (Greenlee 1987):
$$ \mathrm{Maximum}\ \mathrm{drop}=\frac{\mathrm{Change}\ \mathrm{in}\ \mathrm{elevation}}{\mathrm{Distance}}\cdotp 100 $$
(10)

If two cells flow to each other, and the method considers them as sinks and assigns undefined flow direction (Jenson and Domingue 1988). However, to obtain an accurate representation of the flow direction across the watershed, the sinks were filled using the spatial analyst tool in ArcGIS.

Results

Geological characterization

The majority of the watershed (28.4 km2 or 55% of the total watershed) contained the granite-gneiss bedrock formation at the surface (Fig. 2). The bedrock was variably weathered with sand grains ranging from fine to coarse, and formed regolith ranging from fertile sandy soils to loamy soils in areas where it outcrops. The other surficial formations were the alluvial deposits (21.2%), sandstone (6.9%), limestone (10%) and the lacustrine deposits (6.9%; Fig. 2). The low elevation, central part of the watershed was filled by alluvial and lacustrine deposits, while limestone was prominent in the eastern uplands and sandstone formations dipped westward on the western side of the watershed. The dry lakebed was covered by a clay layer with an average thickness of 25 m (Figs. 2 and 3a–c). The majority of the watershed, except the sandstone and limestone areas, had a clay layer that varies in thickness, which is overlaid by black cotton soils which also varies in thickness (Fig. 3d).
Fig. 2

Surface geology of Lake Haramaya watershed. The world projected coordinate system, WGS 1984 Web Mercator (auxiliary sphere), is shown

Fig. 3

Lithological and stratigraphic models of the Lake Haramaya watershed using: a drill logs and coring, b VES, and c combined drill logs and VES. The stratigraphic model of the aquifer (d) groups similar lithological units to show the orientation of the layers. Distances and elevations on the scales of the figures are in meters (m). The world projected coordinate system, WGS 1984 Web Mercator (auxiliary sphere), is shown

The alluvial deposits were arranged in horizontal layers with sands and gravels inter-bedded with layers of silts and clays, likely caused by meandering stream channels and associated alluvial deposition (Fig. 3a–d). The 3D lithological modelling showed eight lithological facies (Fig. 3a–c). The wells drilled within the watershed are located at the center of the watershed, and mostly in the alluvial deposit area (Fig. 1); hence, the 3D lithological model produced from the drill log data was limited in characterizing the whole aquifer system due to its limited spatial coverage (Fig. 3a). In this case, the horizontal scale in Fig. 3a is different from the other plots in Fig. 3 because it is restricted to the area within a convex hull defined by the drilled wells. Even though the areal coverage of the VES data was larger than that of drill log data, the 3D lithological model produced from the VES data was found to have some variations/incongruities in material types compared to areas that have wells (Fig. 3b). Therefore, an integrated approach that uses both the available drill logs and VES data was used to produce a more constrained heterogeneous 3D lithological model of the aquifer system that shows the spatial variations of the geological variables (Fig. 3c), confirming the complementary nature of the methods (Fig. S2 of the ESM). Thus, the final 3D lithological model produced from the composite methods (Fig. 3c) related more clearly to the surface geological map (Fig. 2) and the heterogeneous nature of the alluvial deposit. The topographic highs shown in Fig. 3a are not shown in Fig. 3c, produced by combining the drill log and VES data. This is because Fig. 3c was the entire study area, while Fig. 3a was only a small subsample that had only the drill log data; additionally, the vertical exaggeration used in Fig. 3a was 30, while it was 21 in Fig. 3b–d.

The 3D stratigraphic model showed that sediment layers are arranged based on the size of the depositional materials. The coarse sand was the thickest followed by gravely sand, medium sand, clay, sandy clay, fine sand and the black cotton soil, respectively. The stratigraphic model does not show the heterogeneous nature of the aquifer system as it homogenizes the different layers based on the dominant formation types within the layer. It is based on stacked 2D models of each geologic layer and was only used to identify the thickness of the potential aquifer and the alignment of the basement topography. The model also showed that the aquifer system in the area was a single, shallow and unconfined unit (Fig. 3d). The only impermeable layer detected was the basement (granite). The top layers (clay and black cotton soil) are not compacted, inferring there is no confining layer within the aquifer system.

The basement topography in the area, deduced from the stratigraphic model, was found to be undulating, with the deepest part at the center of the watershed and the shallowest in the boundaries, and a rippling bedding throughout the aquifer system (Fig. 4a). The undulating nature of the basement has affected the bedding orientations of the layers overlaying it, including the top layer (Fig. 4b). This is likely due to the bedding orientation of sediment layers above the basement in alluvial deposit aquifers closely following the undulating nature of the basement topography (Chatterjee et al. 2009).
Fig. 4

a Basement and b surface and basement topographies of the aquifer in the Lake Haramaya watershed. Distances and elevations on the scales of the figures are in meters (m)

Hydrogeological characterization

Pumping test analysis

The hydrogeological properties of the aquifer system were estimated from pumping test records of the wells in the alluvial and lacustrine deposit areas (Table 1). The average T, K and SY values of the parts of the aquifer system that have wells were estimated to be 36.2 ± 7 m2/day, 1.9 ± 0.4 m/day and 25.8 ± 1.5%, respectively (Table 1). The average values of the KV/KH and Sy/S were estimated to be 0.5 ± 0.07 and 11.3 ± 0.71, respectively. The spatially variable extents of the hydraulic properties of the aquifer (T, K, SY) showed that the aquifer is heterogeneous, while the <1 values of KV/KH showed that the aquifer is anisotropic (Alakayleh et al. 2018; Fetter 2001). In most of the wells, the SY/S was found to be one-tenth of the specific yield, which is within the ranges for typical unconfined aquifers of 0.01–0.3 (Fetter 2001).
Table 1

Hydraulic properties of the aquifer in the Lake Haramaya watershed estimated from pumping test data. The well screen was located 7 m above the well bottom in all wells. SE standard error

Well ID

Location

Depth (m)

T (m2/day)

K (m/day)

Sy (%)

K V /K H

S Y /S

East

North

W1

173,622

1,042,715

51.0

26.1

0.4

30.0

1.0

16.0

W2

173,614

1,042,776

52.0

5.18

0.2

21.6

0.6

10.0

W3

174,516

1,041,850

47.0

84.0

2.1

29.3

0.2

10.0

W4

174,860

1,041,865

43.0

13.1

0.3

29.3

0.9

10.0

W5

175,046

1,041,815

59.0

112

2.7

20.0

0.2

18.0

W6

174,372

1,041,901

51.0

16.4

0.4

14.6

0.1

10.0

W7

173,918

1,041,901

50.0

23.5

0.5

29.2

1.0

10.0

W8

173,819

1,040,831

49.0

23.1

0.4

35.1

1.0

10.0

W9

173,951

1,040,859

47.0

8.3

0.2

30.0

0.4

10.0

W10

173,181

1,040,820

44.0

66.1

4.4

18.2

0.3

18.0

W11

174,451

1,041,271

60.0

28.2

2.2

34.0

0.6

10.0

W12

174,699

1,041,095

60.0

29.1

4.2

21.1

0.5

10.0

W13

174,456

1,040,939

61.0

65.3

0.9

23.0

0.4

10.0

W14

173,815

1,040,815

63.0

26.0

5.1

27.1

0.5

10.0

W15

173,948

1,040,875

61.0

32.1

4.3

25.0

0.5

10.0

W16

173,915

1,041,900

62.0

38.3

1.8

32.0

0.5

10.0

W17

175,006

1,041,574

55.0

18.4

2.4

19.6

0.5

10.0

Ave.

53.8

36.2

1.9

25.8

0.5

11.3

SE

1.63

7.0

0.4

1.5

0.07

0.7

Hydrogeophysical characterization

The hydraulic properties quantified using VES data were variable depending on the underlying geology with T values ranging from 12.1 to 588.4 m2/day and averaging 215.6 ± 44.9 m2/day, while K ranged from 1.4 to 22.4 m/day and averaged 6.2 ± 1.5 m/day (Table 2). The average thickness of the potential zone of the aquifer system was estimated to be 39.4 ± 4.7 m (Table 2). The average T and K values of the aquifer system from the combined pumping test and VES data were 126.5 ± 25.8 and 4.05 ± 1.0 m/day, respectively.
Table 2

Estimated Dar-Zarrouk parameters and hydraulic properties of the aquifer in the Lake Haramaya watershed from VES readings. SE standard error

VES station

a (Ωm)

b (m)

S−1)

∑S−1)

TR (Ωm2)

TR (Ωm2)

T (m2/day)

K (m/day)

V1

6.5

33.2

5.1

5.1

215.8

241.1

243.8

7.3

V2

7.2

43.5

6.1

6.1

313.2

503.6

326.1

7.5

V3

8.1

52.3

6.4

6.5

423.6

507.2

412.9

7.9

V4

6.3

39.3

6.3

6.4

247.6

351.3

271.4

6.9

V5

3.6

15.5

4.3

5.7

55.8

218.1

84.7

5.5

V6

17.9

35.1

2.0

5.9

628.3

721.0

561.7

16.0

V7

20.5

12.2

0.6

2.4

250.1

308.6

273.5

22.4

V8

11.8

56.5

4.8

4.9

666.7

689.1

588.4

10.4

V9

1.6

50.6

33.4

32.5

81.0

182.1

113.3

2.2

V10

0.8

5.5

6.5

6.8

4.4

376.9

11.6

2.1

V11

1.5

66.8

43.3

43.4

100.2

280.9

133.9

2.0

V12

1.5

50.1

33.4

33.5

75.2

181.1

106.9

2.1

V13

1.4

57.3

41.5

41.5

80.2

220.2

112.5

2.0

V14

1.5

52.4

35.9

35.9

78.6

198.3

110.7

2.1

V15

0.6

12.4

19.3

19.4

7.4

120.1

17.6

1.4

V16

1.1

48.1

33.5

32.5

52.9

178.1

81.3

1.7

Ave

5.7

39.4

17.7

18.0

205.1

329.9

215.6

6.2

SE

1.6

4.7

4.0

3.9

52.3

46.1

44.9

1.5

The protective capacity of the aquifer top was estimated through the electrical longitudinal conductance (S) values derived from VES and was found to range from 0.6 to 43.3 Ω−1, which is in the weak to excellent range of aquifer protective capacity (Abiola et al. 2009). The area that had a thick clay layer (lacustrine deposit) was found to have high protective capacities (>10 Ω−1), while the granite-gneiss area showed a relatively low protective capacity (<1 Ω−1).

Groundwater potential and sustainability

The groundwater development potential of the aquifer system was calculated by upscaling the volume of the potential aquifer formations (volume of fine-to-coarse sand, gravely sand and weathered granite; 2.4 km3), estimated using the lithology volumetrics model in the RockWorks software, by the average specific yield value (25.8 ± 1.5%) which equated to 0.62 ± 0.09 km3. This estimation excludes the two least productive formations (sandstone and limestone) and the 25-m-thick clay layer at the top most part of the lacustrine deposit area (Fig. 2). As the deepest screen position of the extraction wells is 7 m above the basement, the adjusted available (useable) groundwater is 0.53 ± 0.07 km3.

To predict the sustainability of the groundwater resource, the MODFLOW model was calibrated and validated using the water level records from three continuously recorded wells (2008–2017)—Fig. 5; Table S1 of the ESM. The calibration process resulted root mean square error (RMSE) values of 2.6, 2.0 and 3.4 in wells 7, 8 and 9, respectively (Figs. 1 and 5). This implies that the RMSE values were found to be 22.4, 20.6 and 26.6% of the maximum measured values in wells 7, 8 and 9, respectively, which shows a good fit between the measured and modelled values. The wells used for the calibration had almost equal depths (50, 49 and 47 m, respectively). Taking the variations in aquifer hydraulic properties and a 5% uncertainty in groundwater discharge into consideration, the MODFLOW model predicted that the current water use scenario (~30,120 m3/day: Haramaya Woreda Water Supply Bureau) will cause the water table to reach below minimum pumping position in approximately 27–32 years (Fig. 6). However, factors like population growth, industrial expansion and mismanagement of the resource will perhaps exacerbate the situation and shorten the predicted time period until the minimum pumping position is reached. Based on the predicted water levels, the safe yield of the aquifer is 15,720 m3/day, meaning a decrease in the current exploitation rate by ~50% is required to maintain a sustainable resource.
Fig. 5

MODFLOW model calibration and validation using the measured depth to water table from three wells over ten consecutive years in the Lake Haramaya watershed

Fig. 6

Predicted water levels of the groundwater in the Lake Haramaya watershed. Predictions were made using the current extraction rate (~30,120 m3/day ±5%/annum) and average K value ± standard error (SE)

Discussion

Comparisons among the methods

To compare the efficiencies and limitations of the drill log and VES data in producing the 3D lithological models, 11 wells that had both VES and drill log data were used (Fig. 7). The two methods produced similar 3D shapes of the aquifer system; however, the VES data considered some parts of the gravely sand layer modeled from the drill logs (Fig. 7a) as a coarse sand, the coarse sands as medium sand, and the medium sand as fine sand layers (Fig. 7b). This could be caused by differences in moisture content with the resistivity of a saturated material being less than that of unsaturated similar material (Koster 2005).
Fig. 7

A comparison of 3D lithological models sourced from: a drill log data, and b VES data. Distances and elevations on the scales of the figures are in meters (m)

Alternatively, the overlapping resistivity values of the different formations could be responsible for variations between the two models. The resistivity values of sand and gravel materials, for example, lay within the same range (Gonzalez-Alvarez et al. 2016). VES separates these materials based on the dominant formation type within the layer causing an overlap between the coarse sand and gravely sand layers (Fig. 7b). This caused the thickness of the layer covered by gravely sand from the drill logs (Fig. 7a) to be larger than that of from VES (Fig. 7b), which implies that the coarse sand materials are the dominant materials in the gravely sand layer. Similarly, the resistivity values of the clay and sandy clay materials lay within the same range in the resistivity ranges of the earth materials (Gonzalez-Alvarez et al. 2016), suggesting a thicker sandy clay layer than actually existed (Fig. 7b). The presence of clay and black cotton soil at the surface also reduced the resistivity of the medium sand at the top surface of the model from drill logs (Fig. 7a) to fine sand in the model produced from VES (Fig. 7b). A 1D profile of the VES data in comparison with the drill log data (Fig. S2 of the ESM) indicates how the VES data responded to the changes in lithology.

Similarly, the pumping test records from the eleven wells and their corresponding VES readings (Fig. 1; Table S2 of the ESM) were used to compare the extents of the hydraulic properties estimated from the two approaches. The average T and K values estimated using the data from the eleven VES stations were 95.0 ± 20.3 and 2.7 ± 0.6 m/day, while those estimated from the pumping test data of the eleven wells were 48.3 ± 9.0 and 2.6 ± 0.5 m/day, respectively (Fig. 8). The combined use of the VES and pumping test data resulted in average T and K values of 71.6 ± 12.0 and 2.6 ± 0.4 m/day, respectively. Therefore, by using the two methods simultaneously, uncertainties associated with T and K estimates were reduced from using only the VES survey by 40.1 and 33.3%, respectively. Similarly, it reduced the uncertainties of the K estimates from the pumping test by 20% and increased the levels of uncertainties of the T estimation by 25%. The combined method increased the uncertainties of T values in comparison to those estimated from pumping test data because T varies with aquifer thickness and the combined method uses a relatively larger aquifer thickness than the pumping test method, as VES likely overestimates the aquifer thickness (Marechal et al. 2010).
Fig. 8

Comparisons of: aT and bK values estimated from the geological, VES and pumping test methods

Transmissivity is the integration of hydraulic conductivity and the productive aquifer thickness; thus, geo-electrical resistivity can detect the thickness of the unproductive (low hydraulic conductivity) partially weathered part of the bedrock, while the depth of the fully penetrating wells is limited to above the weathered bedrock layer. Furthermore, the VES survey may incorporate the unsaturated zone of the aquifer as a potential aquifer thickness (Khalid et al. 2018), while potential aquifer thickness from the wells is bounded by the water table as it could be measured from the wells. This is because soil-moisture contents in an unsaturated zone could influence aquifer resistivity to show water strikes before it reaches the actual location of the water table (Khalid et al. 2018). Therefore, the differences in aquifer thicknesses (Tables 1 and 2) detected from the fully penetrating wells and the VES data could be the reason for the uncertainties associated with the T estimates from the VES data. T values estimated from the VES and pumping test data were significantly different (two-way t-test; p = 0.01), while K values were not significantly different (p = 0.5), suggesting that the VES data are effective in estimating K values, while it requires pumping test data to effectively estimate T values. Water-table measurements should also be incorporated when estimating T using the Darzarouk parameters to reduce the uncertainties by excluding the thickness of the unsaturated zone of the aquifer system.

Both the T and K values from the VES data were found to be higher than those estimated from pumping test data (Fig. 8). This may be due to the VES method relying on moisture and the presence of conductive materials (Koster 2005). The pumping test method is the most direct method of parameter estimation, while slight over- or under-estimations are expected from the VES survey (Korowe et al. 2011). That is also why combined use of data and comparison of results from the two approaches is important when characterizing an aquifer system; however, it is also important to note that the median and the 75% percentile of the T values estimated from the VES data were within the ranges estimated from the pumping test method (Fig. 8). The median K values derived from VES and pumping test data were also almost equal (Fig. 8).

The ranges of K values for silt and coarse sand formations (the ranges of the materials in the productive layers) from the literature are very wide, 0.08–518.4 m/day (Domenico and Schwartz 1990). Warren et al. (1996) made comparisons of hydraulic conductivities estimated from three methods (permeameter tests, grain-size analysis and slug tests) for a sand and gravel aquifer in southeastern Massachusetts, USA, and reported corresponding hydraulic conductivity ranges of 0.9–86, 0.5–206 and 0.6–79 m/day, with average values of 22.8, 40.7 and 32.9 m/day, respectively. Estimating the K values based on literature values of the materials detected in the study area resulted a range in K values from 0.1 to 45.8 m/day, with a median value of 5.0 m/day, which was considerably higher than both the VES and pumping test derived K values (Fig. 8), and thus can only be used to give general assumptions of the aquifer properties.

The importance of using composite methods

Composite use of the drill log and VES data was found to be very important for a comprehensive characterization of the aquifer system. The quality of drill log data depends on the type of drilling technologies used—for instance, air rotary can produce a relatively better sample quality than direct rotary (Lomberg 2014), whereas on the other hand, drilling a high number of wells may not be economically feasible. There may also be areas that are not suitable to drill a well in, but VES techniques can be applied. However, VES data can also be affected by different factors such as moisture content, conductive materials and overlapping electrical resistivity values of lithological layers, especially when saturated—for instance, Koster (2005) investigated the effects of water content on resistivity survey in an unconfined fluvial aquifer in Columbus, USA, and reported that water saturation decreases resistivity by 0–600 Ω-m. The ranges of T and K values estimated from the VES data covered a higher range than those estimated from pumping test data (Fig. 8), which may be due to different factors such as geophysical equipment calibration and other environmental factors that affect the VES data. Hence, there is a need of geophysical equipment calibration with the help of pumping test data.

The VES survey provided important parameter estimates like aquifer protective capacity and the extent of weathering of the different formations, which cannot be directly extracted from pumping test data. Pumping test data on the other hand are helpful to estimate parameters like KV/KH and Sy, which cannot be estimated easily and accurately from VES data due to the high variabilities of the estimates caused by the uncertainties in porosities (Frohlich and Kelly 1988). However, the comparisons between the values estimated from VES and pumping test datasets could only give relative conclusions about the aquifer’s hydraulic properties that can be driven from the two methods, as the factors which govern the water and current flow and conduction into the soil are extremely variable (Soupios et al. 2007). These can include variations in lithology types, mineral contents, packing and orientation of grains, shape and geometry of pores and pore channels (Salem and Chilingarian 1999). Therefore, integrating hydrogeological and geophysical methods with geological methods was found helpful to minimize these problems and to confirm that the aquifer property values estimated from pumping test and VES data are within the expected ranges for each lithology.

Groundwater productivity evaluation

The most productive formation in the area, based on K, T, and SY estimates, was the shallow alluvial formation composed of loose coarse-grained sand, pebbles and rock fragments, followed in decreasing productivity by the lacustrine deposits, granite-gneiss, limestone and sandstone. The granite-gneiss area potentially receives recharge for the whole watershed as the thickness of the clay layer in this area is relatively thinner and most of the grains in the top layers are coarse grained. The internal hydrogeological condition of the limestone formation is still unknown, as the area was not suitable for VES surveying because of the steep slope hills and no wells were drilled in the limestone formation; however, karst limestone topography can be characterized by underground drainage systems with sinkholes and caves (Doerfliger et al. 1998). These large conduits allow precipitation and surface runoff to penetrate to the soil horizons and fill subterranean caverns, while groundwater resources from other karst aquifers play a major role in meeting water demands throughout the world (Doerfliger et al. 1998). Due to the difficulty in defining protection zones in karst environments, protecting groundwater resources remains a challenge (Thomsen et al. 2004). Future research in the area should focus on understanding the hydrogeological conditions of the limestone area with the help of conversant technologies such as high-resolution satellite imageries and remote sensing (Meijerink 1996).

The gravely sand, coarse sand and medium sand layers in the potential aquifer zones, were found to be the most productive layers, while the remaining layers with fine grains were the least (Fig. 3d). Based on the pumping test data, the productive aquifer zone had a yield ranging from 400 to 1,383 m3/day. This is similar to the study results by Demlie and Titus (2015), in a hydrogeological study in the Natal Group Sandstone formations within South Africa, which reported borehole yields of 260–1,728 m3/day to be in the good range.

Groundwater management implications

The combined use of aquifer characterization methods provided the basis for site-specific groundwater protection and recharge optimization zones. An unconfined aquifer can receive recharge in many ways such as via direct infiltration from rainfall and lateral flow from nearby aquifers with higher water levels (Scanlon et al. 2002). However, local confining layers and geomorphology (e.g., slope and drainage lines) can affect groundwater recharge rates and locations (Kebede 2013; Bouwer 2002), whereby the 25-m-thick clay layer in the lacustrine deposit area is a good example of local confining layers that decrease the likelihood for vertical groundwater recharge. The alluvial deposit area also has an overlying clay layer that ranged from 4 to 10 m in thickness that may reduce recharge and make the granite-gneiss formation (Fig. 2), which has thin clay layer (<1 m) at the top-most surface, the potential recharge site. The groundwater flow in the area (Fig. 9) shows the uplands of the watershed in the eastern and southwestern areas are the potential recharge sites and correspond to where the clay layer is thinnest. At elevations above 2,100 m, granite gneiss formation starts; therefore, locating recharge optimization structures such as collector wells, pits and infiltration galleries in these areas may be effective measures to maximize groundwater recharge rates (Martin-Rosales et al. 2007). While the protective capacity of the majority of the aquifer system is good, the construction of the recharge optimization structures should still consider aquifer contamination threats throughout the watershed. There are several studies that suggest proper planning and design of groundwater recharge optimization structures is very important to safeguard an aquifer system from contamination from different anthropogenic activities. Omeiza and Dary (2018) suggested that anthropogenic activities should be properly planned to abolish groundwater pollution in areas that are prone to contamination. Oni et al. (2017) also suggested that groundwater management and development activities should take the possible sources of groundwater contamination into consideration.
Fig. 9

Groundwater flow direction map of the Lake Haramaya watershed

Conclusion

Geological, hydrogeological and geophysical approaches were integrated for a complete characterization of the aquifer system and evaluation of the groundwater productivity and sustainability in the Lake Haramaya watershed. The aquifer system was characterized as a single shallow unconfined aquifer with a heterogeneous and anisotropic nature and moderately high to low productivity ranges. Groundwater sustainability predictions revealed that, with the current water-use scenario, the groundwater level will reach below minimum pumping position within 27–32 years. It then suggested that the current exploitation rate in the area should be reduced by 50% for a sustainable use of the resource; therefore, policy makers and development agents need to develop mechanisms to reduce extraction rates and at the same time construct recharge optimization structures in the potential recharge sites.

The study also showed the limitations and advantages of each aquifer characterization method. Developing an integrated approach using geological mapping, drill logs, pumping test and VES data for aquifer characterization and groundwater productivity evaluation reduces the uncertainties associated with the single method uses and resulted in a complete characterization of the aquifer system.

Notes

Acknowledgements

The authors thank two anonymous reviewers, the editor Jean-Michel Lemieux and the technical editorial advisor Sue Duncan for their insightful suggestions which greatly improved the manuscript.

Funding information

This project was funded through a Haramaya University Research grant (HURG-2015/16-01-03) with the modelling component funded by the Australian Research Council (DE180100535) and the Herman Slade Foundation.

Supplementary material

10040_2019_1956_MOESM1_ESM.pdf (482 kb)
ESM 1 (PDF 482 kb)

References

  1. Abdullahi MG, Toriman ME, Gasim MB (2014) The application of vertical electrical sounding (VES) for groundwater exploration in Tudun Wada Kano state, Nigeria. J Geol Geophys 4:1–3Google Scholar
  2. Abiola O, Enikanselu PA, Oladapo MI (2009) Groundwater potential and aquifer protective capacity of overburden units in Ado-Ekiti, southwestern Nigeria. Int J Phys Sci 4(3):120–132Google Scholar
  3. Adeoti L, Alile OM, Uchegbulam O, Adegbola RB (2012) Geo-electrical investigation of the groundwater potential in Mowe, Ogun state, Nigeria. Br J Appl Sci 2:58–71Google Scholar
  4. Alakayleh Z, Clement TP, Fang X (2018) Understanding the changes in hydraulic conductivity values of coarse and fine-grained porous media mixtures. Water 10:1–13Google Scholar
  5. Alemayehu T, Wagari F, Dagnachew L (2006) Impact of water overexploitation on highland lakes of eastern Ethiopia. Environ Geol 52:147–154Google Scholar
  6. Alexander M, Berg SJ, Illman WA (2011) Field study of hydrogeologic characterization methods in a heterogeneous aquifer. Ground Water 49(3):365–382Google Scholar
  7. Anomohanran O (2013) Geophysical investigation of groundwater potential in Ukelegbe, Nigeria. J Appl Sci 13(1):119–125Google Scholar
  8. Archie GE (1942) The electrical resistivity log as an aid in determining some reservoir characteristics. Pet Trans AIME 146(01):54–62Google Scholar
  9. Barackman M, Brusseau ML (2003) Groundwater sampling. Poll Eng 35(5):16–18Google Scholar
  10. Blue Marble Geographics (2018) Getting started guide: GlobalMapper. Blue Marble Geographics, Hallowell, ME, 26 ppGoogle Scholar
  11. Bouwer H (2002) Artificial recharge of groundwater: hydrogeology and engineering. Hydrogeol J 10:121–142Google Scholar
  12. Cao G, Zheng C, Scanlon BR, Liu J, Li W (2013) Use of flow modeling to assess sustainability of groundwater resources in the North China plain. Water Resour Res 49:159–175Google Scholar
  13. Chatterjee R, Gupta BK, Mohiddin SK, Singh PN, Shekhar S, Purohit R (2009) Dynamic groundwater resources of National Capital Territory, Delhi: assessment, development and management options. Environ Earth Sci 59:669–686Google Scholar
  14. Chebet C (2012) Estimating water table elevations by spatial interpolation: a case study of Kamariny division, Keiyo North District, Kenya. J Emerg Trends Educ Res Policy Stud 3(5):741–774Google Scholar
  15. Demlie M, Titus R (2015) Hydrogeological and hydrogeochemical characteristics of the Natal Group sandstone, South Africa. S Afr J Geol 118(1):33–44Google Scholar
  16. Doerfliger N, Jeannin PY, Zwahlen F (1998) Water vulnerability assessment in karst environments: a new method of defining protection areas using a multi-attribute approach and GIS tool. Environ Geol 39(2):2–10Google Scholar
  17. Domenico PA, Schwartz FW (1990) Physical and chemical hydrogeology. John Wiley and Sons, New York, pp 824Google Scholar
  18. Ezeh CC (2012) Hydro-geophysical studies for the delineation of potential groundwater zones in Enugu state, Nigeria. Res Inst J Geol Min 2(5):103–112Google Scholar
  19. Fetter CW (2001) Applied hydrogeology, 4th edn. Prentice Hall, Upper Saddle River, NJGoogle Scholar
  20. Frohlich RK, Kelly WE (1988) Estimates of specific yield with the geoelectric resistivity method in glacial aquifers. J Hydrol 97(1–2):33–44Google Scholar
  21. Gleeson TL, Smith ML, Manning AH (2011) Classifying the water table at regional to continental scales. Geophys Res Lett 38:1–6Google Scholar
  22. Gonzalez-Alvarez I, Ley-Cooper AY, Salama W (2016) A geological assessment of airborne electromagnetics for mineral exploration through deeply weathered profiles in the southeast Yilgarn Cratonic margin, western Australia. Ore Geol Rev 73:522–539Google Scholar
  23. Greenlee DD (1987) Raster and vector processing for scanned linework. Photogramm Eng Remote Sens 53(10):1383–1387Google Scholar
  24. Haitjema HM, Mitchell-Bruker S (2005) Are water tables a subdued replica of the topography? Ground Water 43(6):781–786Google Scholar
  25. Harbaugh AW (2005) MODFLOW-2005, The U.S. Geological Survey Modular Ground-Water Model: the ground-water flow process. In: Modeling techniques, book 6, section A, Ground water. US Geological Survey, Reston, VA, 253 ppGoogle Scholar
  26. Henriet JP (1976) Direct application of Dar Zarrouk parameters in groundwater surveys. Geophys Prospect 24(2):344–353Google Scholar
  27. Herckenrath D, Legaz-Gazoty A, Fiandaca G, Auken E, Christensen M, Balicki M, Bauer GP (2012) Sequential and coupled hydro-geophysical inversion of a groundwater model using geo-electric and transient electromagnetic data. J Hydrol 123–137Google Scholar
  28. Hunkeler D (2010) Geological and hydrogeological characterization of subsurface. In: Timmis KN (ed) Handbook of hydrocarbon and lipid microbiology. Springer, BerlinGoogle Scholar
  29. Hwang HT, Jeen SW, Suleiman AA, Lee KK (2017) Comparison of saturated hydraulic conductivity estimated by three different methods. Water 9(942):1–15Google Scholar
  30. Ifabiy IP, Ashaolu ED, Omotosho O (2016) Hydrogeological characteristics of groundwater yield in shallow wells of the regolith aquifer: a study from Ilorin, Nigeria. Momona Ethiopian J Sci 8(1):23–36Google Scholar
  31. Ibuot JC, Akpabio GT, George NJ (2013) A survey of the repository of groundwater potential and distribution using geoelectrical resistivity method in Itu local government area, Akwa Ibom state, southern Nigeria. Cent Eur J Geosci 5(4):538–547Google Scholar
  32. Israde-Alcantara I, Buenrostro DO, Carrillo CA (2005) Geological characterization and environmental implications of the placement of the Morelia dump, Michoacán, central Mexico. J Air Waste Manag Assoc 55(6):755–764Google Scholar
  33. Jenson SK, Domingue JO (1988) Extracting topographic structure from digital elevation data for geographic information system analysis. Photogramm Eng Remote Sens 54(11):1593–1600Google Scholar
  34. Kebede S (2013) Groundwater in Ethiopia: features, numbers and opportunities. Springer, Heidelberg, GermanyGoogle Scholar
  35. Khalid P, Ullah S, Faird A (2018) Application of electrical resistivity inversion to delineate salt and freshwater interfaces in Quaternary sediments of northwest Himalaya, Pakistan. Arab J Geosci 11(112):1–10Google Scholar
  36. Korowe MO, Nyadawa MO, Singh VS, Ratnakar D (2011) Hydrogeophysical parameter estimation for aquifer characterization in hard rock environments: a case study from Jangaon sub-watershed, India. J Oceanogr Mar Sci 2(3):50–62Google Scholar
  37. Koster JW (2005) Effects of water saturation on a resistivity survey of an unconfined fluvial aquifer in Columbus, Colorado State University, Department of Geosciences. Hydrology Days (2005):111–120Google Scholar
  38. Kurniawan A (2009) Basic IP2WIN tutorial: basic principles in using IP2 Win Software. Hydrogeology World, 32 pp. http://www.academia.edu/7756576/Basic_IP2_Win_Tutorial. Accessed February 2019
  39. Lomberg K (2014) Best practice sampling methods, assay techniques, and quality control with reference to the platinum group elements (PGEs). J South Afr Inst Min Metall 114:53–62Google Scholar
  40. Marchant AP, Banks VJ, Royse K, Quigley SP, Wealthall GP (2011) An initial screening tool for water resource contamination due to development in the Olympic Park 2012 site, London. Environ Earth Sci 64(2):483–495Google Scholar
  41. Marechal J, Vouillamoz J, Kumar MS, Dewandel B (2010) Estimating aquifer thickness using multiple pumping tests. Hydrogeol J 18:1787–1796Google Scholar
  42. Martin-Rosales W, Gisbert J, Pulido-Bosch A, Vallejos A, Fernandez-Cortes A (2007) Estimating groundwater recharge induced by engineering systems in a semiarid area (southeastern Spain). Environ Geol 52:985–995Google Scholar
  43. Masvopo TH (2008) Evaluation of the groundwater potential of the Malala alluvial aquifer, lower Mzingwane River, Zimbabwe. MSc Thesis, University of Zimbabwe, Harare, Zimbabwe, pp 29–65Google Scholar
  44. Meijerink AM (1996) Remote sensing applications to hydrology: groundwater. Hydrol Sci J 41(4):549–561Google Scholar
  45. Neuman SP, Blattstein A, Riva M, Tartakovsky DM, Guadagnini A, Ptak T (2007) Type curve interpretation of late-time pumping test data in randomly heterogeneous aquifers. Water Resour Res 43:1–15Google Scholar
  46. Nwosu LI, Ekine AS, Nwankwo CN (2013) Evaluation of groundwater potential from pumping test analysis and vertical electrical sounding results: case study of Okigwe District of Imo state Nigeria. Pac J Sci Technol 14(1):536–548Google Scholar
  47. Obiora DN, Ajala AE, Ibuot JC (2015) Evaluation of aquifer protective capacity of overburden unit and soil corrosivity in Makurdi, Benue state, Nigeria, using electrical resistivity method. J Earth Syst Sci 124(1):125–135Google Scholar
  48. Okiongbo KS, Akpofure E (2012) Determination of aquifer properties and groundwater vulnerability mapping using geo-electric method in Yenagoa City and its environs in Bayelsa state, South Nigeria. J Water Resour Protect 4:354–362Google Scholar
  49. Omeiza AJ, Dary DM (2018) Aquifer vulnerability to surface contamination: a case of the new millennium city, Kaduna, Kaduna state Nigeria. World J Appl Phys 3(1):1–12Google Scholar
  50. Oni TE, Omosuyi GO, Akinlalu AA (2017) Groundwater vulnerability assessment using hydrogeologic and geoelectric layer susceptibility indexing at Igbara Oke, southwestern Nigeria. NRIAG J Astron Geophys 6:452–458Google Scholar
  51. Paola MA, Margiotta S, Mazzone F, Negri S (2005) An integrated geological, hydrogeological and geophysical approach to the characterization of the aquifer in a contaminated site. Hydrol Earth Syst Sci Discuss 2:229–263Google Scholar
  52. Rockworks (2017) RockWorks 17 training manual. Rockware, Colorado Springs, CO, pp 521–2500Google Scholar
  53. Sabet M (1975) Vertical electrical resistivity soundings to locate ground water resources: a feasibility study. Virginia Water Resources Research Center Bull 73. Virginia Polytechnic Institute and State University, Blacksburg, VA, pp 3–51Google Scholar
  54. Salem HS, Chilingarian GV (1999) The cementation factor of Archie’s equation for shaly sandstone reservoirs. J Pet Sci Eng 23:83–93Google Scholar
  55. Scanlon BR, Healy RW, Cook PG (2002) Choosing appropriate techniques for quantifying groundwater recharge. Hydrogeol J 10:18–39Google Scholar
  56. Schlumberger Water Services (2015) AquiferTest pro user manual. Waterloo Hydrogeologic, Waterloo, 300 ppGoogle Scholar
  57. Shishaye HA, Nagari A (2016) Hydrogeochemical analysis and evaluation of the groundwater in the Haramaya well field, eastern Hararghe zone, Ethiopia. J Hydrogeol Hydrol Eng 5(4):1–15Google Scholar
  58. Shishaye HA (2015) Technical note: groundwater flow modeling in coastal aquifers: the influence of submarine groundwater discharge on the position of the saltwater–freshwater interface. Hydrol Earth Syst Sci Discuss 12:3753–3785Google Scholar
  59. Soupios PM, Kouli M, Vallianatos F, Vafidis A, Stavroulakis G (2007) Estimation of aquifer hydraulic parameters from surficial geophysical methods: a case study of Keritis Basin in Chania. J Hydrol 338:122–131Google Scholar
  60. Tadesse N, Abdulaziz M (2009) Water balance of Haramaya watershed, Oromia regional, eastern Ethiopia. Int J Earth Sci Eng 2:484–498Google Scholar
  61. Tartarello MC, Plaisant A, Bigi S, Beaubien SE, Graziani S, Lombardi S, Ruggiero L, Angelis D, Sacco P, Maggio E (2017) Preliminary results of geological characterization and geochemical monitoring of Sulcis Basin (Sardinia), as a potential CCS site. Energy Procedia 125:549–555Google Scholar
  62. Thomsen R, Sondergaard VH, Sorensen KI (2004) Hydrological mapping as a basis for establishing site-specific groundwater protection zones in Denmark. Hydrogeol J 12:550–562Google Scholar
  63. Turk J, Malard A, Jeannin P, Petric M, Gabrovsek F, Ravbar N, Vouillamoz J, Slabe T, Sordet V (2015) Hydrogeological characterization of groundwater storage and drainage in an alpine karst aquifer (the Kaninmassif, Julian Alps). Hydrol Process 29:1986–1998Google Scholar
  64. Wang L, Tye A, Hughes A (2010) Riverine floodplain groundwater flow modelling: the case of Shelford (UK). British Geological Survey, Nottingham, UK, 28 ppGoogle Scholar
  65. Warren LP, Church PE, Turtora M (1996) Comparison of hydraulic conductivities for a sand and gravel aquifer in southeastern Massachusetts, estimated by three methods. US Geol Surv Water Resour Invest Rep 95-4160, 20 ppGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  • Haile A. Shishaye
    • 1
    Email author
  • Douglas R. Tait
    • 1
  • Kevin M. Befus
    • 2
  • Damien T. Maher
    • 1
  1. 1.Southern Cross GeoscienceSouthern Cross UniversityLismoreAustralia
  2. 2.Civil and Architectural EngineeringUniversity of WyomingLaramieUSA

Personalised recommendations