Introduction

The typical characteristic of an urban area is its rapid growth in population and high industrialization (Todd and Mays 2005; Magesh et al. 2012; Adiat et al. 2013; Oikonomidis et al. 2015) and these have often resulted to increase in urban water demands (Mukherjee et al. 2009). Apart from the insufficiency of surface water resource being the most available and renowned major sources of water supply in both urban and rural areas, the impact of the industrial technological advancement resulting to climatic change effect has also reduced the surface water freshness reliability and sustainability (Graham-Tomasi and Yacov 2004; Abdulla and Al-Shareef 2009). Consequently, these have aroused challenges and a threat to global water security (UNESCO and UNESCO-WSSM 2019). In order to address this impending global water crisis, the attentions toward the development of groundwater resources as alternative to surface water resource have been on the increase (Vörösmarty et al. 2010; Marsalek et al. 2006; George et al. 2009). According to several researchers, groundwater is one of the most important natural resources for providing freshwater sources worldwide (Pham et al. 2019). This could probably due to groundwater resources’ unique attributes such as: good quality, constant temperature, low vulnerability to pollution, and low susceptibility to catastrophic events (Chowdhury et al. 2009). However, groundwater is a hidden natural resource, whose occurrences and accumulation in an area varies largely from one place to another (Adiat et al. 2012; Akinlalu et al. 2017; Mogaji and Lim 2017). Hence, to optimize groundwater resources exploration, the need arises for carrying out comprehensive hydrogeologic, geophysical evaluation and geospatial modeling assessment for these important resources irrespective of any geological setting of interest.

The renowned conventional approach for developing groundwater resources is via borehole drilling and some other hydrogeologic means. Nevertheless, these aforementioned techniques are uneconomical (Todd 1980; Roscoe Moss Co. 1990; Fetter 1994) due to high cost of drilling, time-consuming, labor intensive, destructive and lack of regionalized predictive assessment. In alternative, the methodological approaches involving geophysical and geospatial and data mining techniques according to Madan et al. (2010). Mogaji (2016) can largely addressed these shortcomings.

Thus, this study explored the state–of-the-art approach involving geospatial and data mining modeling techniques with the view of proposing a conceptualized data mining indexing algorithm to maximize groundwater resource development in this problematic geologic terrain effectively is timely. Among the renowned data mining technique in the field of groundwater hydrology, the AHP-MCDA is the most prominent in the field of groundwater hydrology. This is probably because the AHP method has a robust capability for the conjunctive and integrated analysis of multidisciplinary data sets (Chowdhury et al. 2010). Arising from previous works, the efficacy of application of GIS-based AHP method to endogenous induced groundwater potentiality condition factors (GPCFs) in the field of groundwater hydrology is relatively limited. In accordance with Jha et al. (2010), the endogenous factors are in situ subsurface lithology physical properties that have direct bearing with groundwater accumulation and movement.

This study is aimed at delineation of potential groundwater zones using hydrogeologic and geophysical data that are obtainable from the analyzed primary and secondary geoelectric parameters (Dar Zarrouk parameters) namely: regolith variation, groundwater head, hydraulic head, and coefficient of anisotropy, aquifer resistivity and aquifer thickness. These multi variables GPCFs will be multi-criterially synthesized based on the established groundwater potential prediction index (GPPI) mathematical modeling equation. Exploring the unique in situ physical properties of the endogenous data from which aquifer hydraulic parameters can be quantitatively determined will largely established a wider application of AHP technique of MCDA method in the field of groundwater hydrology. Thus, this research output will give a broader perspective of developing a more efficient and reliable decision support system model whose output will provide technical support to government agencies as well as private sectors for groundwater development decision-making and assessment in the investigated area.

Description of the study area

Geography, geology, and hydrogeology

The area studied falls within the South-western part of Nigeria. It is situated between 7° 18′ 6″ to 7° 18′ 43″ Latitudes and 5° 13′ 16″ to 5° 14′ 18″ Longitudes with an area extent coverage of about 2 km2 (Fig. 1). The topographic elevation of the area ranges from 301 to 412 m a.m.s.l. The average annual rainfall amount characterizing the area is about 1333.2 mm (NIMETS 2011). The geologic setting of the area falls within the Precambrian Basement Complex of southwestern Nigeria (Rahaman 1976). Though, based on the geological field survey carried out, the pronounced lithological unit among others is the charnockite rock. Thus, this area can be inferred to be a monolithic geologic setting (Fig. 1). A charnockitic rocky unit is known for undergoing spheroidal weathering, whose end product often resulted in clay lithology with little or no evidence of fracture development (Keller and Frischnecht 1966). However, according to Satpathy and kanungo (1976) and Lewis (1990), the basement terrain aquifer is not only inhomogeneous but also occurs in a discrete and discontinuous manner. In crystalline rock terrains, the weathered layer and the fractured basement rocks constitute the main hydrogelogic units. However, significant amount of water can be stored in the weathered layer/fractured basement rock aquifer units when the thickness is sufficiently high Bayode (2013).

Fig. 1
figure 1

Location, geological and data acquisition map of the study area

Full size image

Methodology

Geophysical survey

Geoelectrical survey was deployed along Six (6) traverses of 260–660 m established in an approximately E–W direction (Fig. 1). The geoelectrical measurement was carried out adopting the dipole–dipole profiling technique in the company of vertical electrical sounding (VES) technique as a followed up. The combined 2-D subsurface imaging and 1-D VES measurement approaches often allow delineation of subsurface geologic structures as well as mapping of the continuous vertical, horizontal variation of resistivity within the subsurface and for subsurface lithologic correlation with the geoelectric sections. On the established 2-dipole–dipole traverses, measurements were made at electrode spacing of a = 20 m and expansion factor n, varying from 1 to 5. The acquired dipole–dipole data were inverted into a 2-D resistivity structure using the DIPRO for windows (2000) software. Upon the 2-D subsurface imaging resistivity structure interpretation, 5–9 VES measurements were carried out along the traverses. Figure 2 presents a typical 2-D resistivity structure with the VES location. In this study, the remaining VES stations were approximately evenly distributed along roads and other available spaces across the study area. A total of ninety-six (96) VES measurements were carried out engaging the Ohmega resistivity meter across the investigated area. Among the observed data, were some parametric sounding measurements as typify at VES 48 location, purposely for lithologic correlation and data interpretation constraint. The acquired 1-D VES data were interpreted following the standard partial curve matching technique coupled with the use of the Win RESIST 1.0 (Vender Velper 2004) software for the 1-D forward modeling. Presented in (Fig. 3) are the representatives VES data interpreted resistivity curves. Based on this interpretation was the determined primary layer geoelectric parameters such as resistivities (r) and thicknesses (h) which were further processed and used to generate both 1-D subsurface geoelectric sections as well as the secondary geoelectric parameter modeling (The Dar-Zarrouk parameters). Typical of the produced 1-D subsurface geoelectric section is shown in Fig. 4. Besides, the generated secondary geoelectric parameters include Transverse unit resistance (T) and longitudinal unit conductance (S) from which the coefficient of Anisotropy (λ) was evaluated.

Fig. 2
figure 2

The inverted 2-D resistivity structure images generated for the established traverses 1–6

Full size image
Fig. 3
figure 3

Typical 2-D geoelectric section generated in the area

Full size image
Fig. 4
figure 4

The correlative panel results for the 2-D profiling and 1-D geoelectrical imaging across traverses 1–6

Full size image

Hydrogeologic investigation

A hydrogeologic study involving depth to water level (DWL) assessment from the accessible hand-dug wells in the area was carried out involving the use of measuring tape, etc. Showing in (Fig. 1) are the located sixty-five (65) wells occupied in the area. Two seasons including the wet season (at the peak of the raining season in August, 2011) and dry season (at the peak of the dry season in February, 2012) were considered for the hydrogeologic measurement. The record of the hydrogeological measurement is presented in Table 2.

Modeling of groundwater potential conditioning factors (GPCFs)

In line with several researchers, the concepts of deriving either subsurface or surface-induced aquifer potential controlled factors have greatly enhanced groundwater resources development in diverse geological terrains (Adiat et al. 2013; Akinlalu et al. 2017; Mogaji and Lim 2018). Thus, this study derived possible groundwater potential influencing factors using the processed geophysical and hydrogeological data acquired. The brief of the procedural approach for the considered vital parameters modeling is as highlighted:

Re-analysis of geoelectrical properties for Dar Zarrouk parameters modeling

The Dar Zarrouk parameters (DSPs) have its fundamental basis from the determined primary geoelectrical parameters such as resistivity and thickness obtainable from the interpreted geophysical data (Table 1). In accordance with Tizro et al. (2010), Oborie and Udom (2014) and Mogaji (2017), the Dar Zarrouk parameter is the secondary parameter indices derived from the determined primary geoelectrical parameters. According to Oladapo et al. (2009), the varying subsurface layer/lithological layer are often inferred from the aforementioned primary geoelectrical parameters. Nominally, the DSPs are the longitudinal unit conductance (S), the transverse unit resistance (T), the average longitudinal resistivity (ρL) and the average transverse resistivity (ρT). The hybrid re-analysis of S, T, λ ρL and ρT is considered in modeling of coefficient of anisotropy (λ) parameter (A). The λ can be defined for a geoelectric section consisting of several (n) layers of a unit cube of rock with the total thickness (H). The degree of subsurface fracturing or inhomogeneity can be established through the determination of electrical coefficient of anisotropy (λ) particularly, in a typical basement complex terrain, where permeability is enhanced by faults, fractures, joints and shear zones. In the studies of Olorunfemi et al. (1991), Bayode and Akpoarebe (2011) and Ojo et al. (2015), the above-mentioned geological significance of faults, fractures, joints and shear zones largely enhanced groundwater yield of an area. Thus, the concept of modeling the coefficient of anisotropy (λ) in the area will largely contribute to its groundwater potentiality assessment. Mathematically, the S and T parameters can be defined by Eqs. 1 and 2 (Keller and Frischnecht 1966):

$$\rho_{{\text{T}}} = \frac{T}{H}$$
(1)
$$\rho_{{\text{L}}} = \frac{H}{S}$$
(2)
$$\lambda = \sqrt {\frac{{\rho_{{\text{T}}} }}{{\rho_{{\text{L}}} }}}$$
(3)
$$\lambda = \frac{{\sqrt {TS} }}{H}$$
(4)
Table 1 The result summary of the geophysical data interpretation
Full size table

The inputs from Eqs. 1 and 2 are engaged in Eqs. (3) and (4) to define the coefficient of anisotropy (λ) parameter for the area subsurface lithology.

The hydrogeologic parameter modeling

The subsurface data were hydrogeologically observed from the available hand dug wells. On each located wells, the depth to water level (DWL) below ground surface is measured using measuring tape. In order to determine the height (H) of the groundwater table above mean sea level in each hand dug well, a similar approach adopted in the study of Acharya et al. (2017) was used. This entails, subtraction of DWL value from the in situ observed height on the ground surface above mean sea level recorded by the GPS device at each occupied location. As explained in the study of Ge and Gorelick (2015), “H” is one of the key variables in describing a groundwater system. In the same study, the Hydraulic head (H) has been defined as a fundamental parameter in describing groundwater flow in rocks. It is a function of the ratio of mechanical energy to the unit weight of fluid in the aquifer system because water flows between two points in response to unequal distributions of the water’s mechanical energy. In the present study, hydraulic head (H) for the study area was computed considering the DWL readings observed during wet season (Column 5 of Table 2) and surface elevation measurements undertaken for the 65 wells occupied. Using Eq. (5), the degree of hydraulic head (H) variation for the area was computed.

$$H = {\text{EASL}} - {\text{DWL}}$$
(5)
Table 2 The observed well measurement
Full size table

Regolith (Re) parameter modeling

The term Regolith (Re) is synonymous to overburden that described all the materials which overlie the bedrock. Assessing this parameter often gives insight into knowing the sizable column thickness of weathered materials characterizing the layers including vadose zone, aquifer unit, etc. In accordance with Olorunfemi, et al. (1999), Bala and Ike (2001) and Bayode (2018), practical evaluation of “Re” parameter can be used in the hydrogeological investigation. For the ‘Re’ parameter modeling, the acquired depth sounding or VES data were processed and interpreted for subsurface lithological units delineation (Table 3). The encompassing lithological materials occasioned at varying depth to the bedrock beneath each occupied VES locations were summed up to determine the ‘Re’ variation in the investigated area.

Table 3 The geoelectrical-based GPCFs modeled results
Full size table

Bedrock relief (B) parameter modeling

The insight into knowing the nature and subsurface topographic configuration characterizing an area has often been defined by the bedrock relief (B) variation (Omosuyi et al. 2007; Bayode et al. 2017; Adagunodo et al. 2018). Studies have established that the estimation of “B” parameter often varied from one place to another (Mogaji 2016; Omosuyi et al. 2007). This variation largely depends on the geological setting and the degree of weathering depth experienced in an areas. The variations in the degree of B in a typical hard rock terrain is determined by the intense of tectonic activity resulting to faults, fractures, joints, etc. occurring in such geological terrain. To model the “B” parameter, the geoelectrical layer thickness for the regolith materials and the corresponding terrain topographic elevation variation measurement were considered. In the area, the terrain topography elevation was observed using the integrated topographic map of the area and the standard GPS instrument measurement, while the regolith thickness was determined from the processed and interpreted VES geophysical data.

Aquifer unit resistivity and thickness (R) and (T) parameters modeling

These parameters were modeled from the geoelectrical delineated subsurface lithologic units (Fig. 2). In a typical Basement Crystalline Terrain, the immediate layer above the possible bedrock is often referred to as the aquifer unit lithology. Considering the geoelectrical determined physical properties, such as resistivity and thickness of the delineated aquifer unit beneath each occupied VES locations (Table 1), both “R” and “T” parameters values were modeled/estimated, respectively. The records of the determined R and T are detailed in Table 3

Geospatial modeling of GPCFs’ themes and their hydrological significance

According to Table 3 records, numerical estimate of varying values was determined per VES location for the considered GPCFs in this study. Exploring the efficacy of GIS technique, the georeferenced coordinates in columns 2 and 3 against their determined GPCFs continuous numerical values were processed to produce spatial model hydrologic themes. Figure 5 presents the generated hydrologic themes for the area. In the studies of Chowdhury et al. (2009), Adiat et al. (2013) and Nampak et al. (2014), some groundwater potentiality influencing factors similar to the above discussed GPCFs have been reported for their appeal hydrological relevance. The generated “Re” spatial model theme modeled from the Re determined values beneath the VES location (Table 3 and Fig. 5a) often depict zones of high and low Re variations. According to Bala and Ike (2001), Omosuyi et al. (2013) and Bayode (2018), such depicted high and low Re variation zones have found to be associated with areas of high and low groundwater potential in their previous studies. The implication of B parameter hydrogeologically is such that it has great influence on dischargeability, rechargeability, and storage of an aquifer unit and thus the fluid subsurface movement in and out of varying lithology is greatly controlled. The B parameter theme (Fig. 5b) was generated via spatial modeling of the computed B values in Table 3. The “H” parameter spatially model in (Fig. 5c, Table 2) enables identifying zones of varying degrees of run-off viz-a-viz high or low variation. Its significance can be liken to water that naturally flows from a zone of high hydraulic head to a region of lower hydraulic head along a hydraulic gradient in the same manner in which heat would also flow toward lower temperature regions. Higher ranks were assigned to the regions with lower hydraulic head as compared to the zones with higher hydraulic heads was scored low. The nature and storativity potential of the area underlain and delineated aquifer units was spatially modeled (Fig. 5e and f) using the determined geoelectrical property values in columns 7 and 8 of Table 3. With the knowledge of the aquifer resistivity “R”, inference of the existence of a weathered, fractured, fault, jointed and sheared condition gives insight to the existence of conduit medium for surface water recharge while the thickness values “T” for this unit tells much about the available aquifer unit space for fluid accumulation and containment.

Fig. 5
figure 5

The hand dug well lithologic and parametric VES data interpretation correlation

Full size image

The applied multi-criteria decision methods (MCDMs) theory/principle

The basis of a multi-criteria modeling method revolved around the knowledge-driven expert and criteria analysis synthesis to aid planners and policy/decision-makers in environmental studies. In accordance with Malczewski and Rinner (2005) and Malczewski (2006), the challenges of spatial mapping involving a large set of feasible alternatives with multiple conflicting and incommensurate evaluation criteria have been effectively addressed using the MCDMs' algorithm. Typical of such MCDMs analyzed result has aided in natural/mineral resources precision and prediction modeling (De Araoujo and Macedo 2002; Feizizadeh and Blaschke 2013; Al-Abadi and Shahid 2016). The two keys processes involved in this study’ MCDMs applied technique include (1) Normalized weighting assignment/Rating scores apportioning and (2) weight linear combination mechanism.

The normalized weighting assignment/rating scores apportioning

The considered GPCFs discussed in “Geospatial modeling of GPCFs’ themes and their hydrological significance” section played an essential role in this index modeling. Each of these GPCFs' themes was assigned weight, depending on their influences on the groundwater potential prospect. For the effective weighting assignment, the analytical hierarchy processes (AHP) technique whose mechanism largely driven by the principle of Saaty’s scale was adopted in this study. Using this Saaty’s scale standard the Criteria Pair-wise Comparison Matrix Analysis of these GPCFs was generated. The rating scores on other hand for each of the GPCFs’ class boundaries regarding their contributing influence toward the aquifer expected productivity potential is based on the similar approach adopted in the study of Al-Saud (2010). According to Al-Saud (2010), the rating is basically numerical values in the range of 1–5, where (1) Very low, (2) Low, (3) Medium, (4) Medium–high and (5) High.

The AHP multi-criteria modeling theory

A good account of the analysis and the modeling theory of AHP technique has been reported in the studies of Chowdary et al. (2009), Madan et al. (2010), Machiwal et al. (2015), Adiat et al. (2012) and Mogaji and Lim (2017). However, below are some of the simple mathematical relationships that substantiate the uniqueness of the AHP MCDA method in terms of its capability to (or) measure the degree of consistency among the pairwise comparisons of contributing factors in decision making. Such sound mathematical basis equations can be defined as Eqs. 6 and 7 as reported in the studies of Adiat et al. (2012) and Thirumalaivasan et al. (2003):

$${\text{CI}} = \frac{{\lambda_{\max } - n}}{n - 1}$$
(6)
$${\text{CR}} = \frac{{{\text{CI}}}}{{{\text{RI}}}}$$
(7)

where CI: consistency index. This index determines the inconsistencies in the pair-wise judgments and is a measure of departure from consistency based on the comparison matrix developed. The λmax is the average value of the consistency vector and n is the number of criterion in the matrix (Vahidnia et al. 2009; Garfì et al. 2009). The CR is the ratio of the matrix, which shows the degree of consistency achieved when comparing the criteria or the probability that the matrix rating was randomly generated, whereas the RI is the random index. The value of the RI largely depends on the number of criteria being compared (Garfì et al. 2009).

The multi-criteria weight linear average technique application

The mechanism of weight linear average (WLA) according to Eastman (1996) is driven in terms of normalized weight of criteria and normalized scores/rating for all options relative to each of the GPCFs themes considered (Fig. 5). The WLA simple mathematical modeling expression is ably expressed in Eq. 8:

$$Y = \sum\limits_{t = 1}^{n} {{\text{NW}}_{i} } R_{i}$$
(8)

where Y is the expected output, NW: the normalized weighted value for each theme (Fig. 5) and Ri is the assigned score for the class boundary in each theme (criterion). The synthesis of the GPCFs' themes (Fig. 5), for developing groundwater potential prediction index modeling equation can be established through applying Eq. (8) algorithm.

The developed groundwater potential prediction index (GPPI) modeling equation

In order to establish a potentiality index modeling algorithm in this study, a multi-criteria synthesizing model is conceptualized for the purpose of combining the groundwater potential influencing factors obtainable from both hydrogeological and geophysical measurements. The proposed GPPI model mathematical equation explored the efficacy of the WLA and AHP weightage assessment techniques for the integration of both hydrogeo-physical electrical derived-based GPCFs. The regionalized capability of this developed groundwater potential prediction index (GPPI) model was effectively executed in the GIS platform. Referencing the established Eq. (8), the GPPI modeling algorithm can be developed as modified thus in Eq. 9.

$${\text{GPPI}} = {\text{Re}}_{{{\text{NW}}_{i} }} {\text{Re}}_{{X_{i} }} + B_{{{\text{NW}}_{I} }} B_{{X_{i} }} + H_{{{\text{NW}}_{i} }} H_{{X_{i} }} + A_{{{\text{NW}}_{I} }} A_{{X_{i} }} + R_{{{\text{NW}}_{i} }} R_{{X_{I} }} + T_{{{\text{NW}}_{i} }} T_{{X_{i} }}$$
(9)

where Re: regolith; B: Bedrock relief; H: Hydraulic head; A: coefficient of anisotropy; R: aquifer Resistivity and T: aquifer thickness; NW: AHP-based normalized weight; R: rating GPPI: groundwater potential prediction index.

Results and discussion

2-D Resistivity structures and geoelectric sections results

The 2-D resistivity inversion assisted in imaging the subsurface geologic sequence, the attributes and the structural disposition of the geo-electric layers in the subsurface. Figure 2 presents the 2-D resistivity structures generated for the established traverses. Three subsurface geoelectric layers were delineated. The layers are generally classified as the topsoil/weathered layer, fractured basement and the fresh basement. Depth to topsoil and the weathered layer may have been merged because of overlapping resistivity values and the thickness of the topsoil. This geoelectric unit is characterized by light-deep bluish/greenish/yellowish-reddish color. The values of the resistivity range between 12 and 179 Ω m, while the thickness varies from 0.5 to 2.5 m. The top soil/weathered unit comprises clay, clayey sand, sand and laterite. The weathered layer is characterized by clay, sandy clay, clayey sand and lateritic clay/lateritic concretion. Low resistivity values ranging from 12 to 100 Ω m are characterized as clay, while the resistivity values of 100–250 Ω m are interpreted as sandy clay and clayey sand units. The second layer, is denoted as the fractured basement layer. Its resistivity and thickness values generally range between 121 and 650 Ω m and about 25–> 40 m, respectively. Moreover, the depth ranges for the delineated fractured basement is about 40–> 60 m. Besides, the fractured basement is characterized by vertical discontinuity and break within the relatively resistive basement bed rock (between distances > 370 m; 480–530 m; 180–220 m; 90–120 m along traverses 1, 2, 5 and 6, respectively. The delineated fractured basement zones which are made up of partially weathered materials exhibits relatively low resistivity zones denoted by green color band overlapping the highly resistive basement host rock with depth extent greater than 60 m. Fresh basement bedrock is delineated as the last layer. It is characterized by yellowish–reddish color bands. The basement bedrock’s resistivity generally ranges from 198 to ∞ Ω m along the six (6) traverses. The typical 2-D geoelectric section image produced based on the interpreted VES data along traverse T2 is presented in Fig. 4. The sections assisted in categorizing the area into five subsurface geologic layers/sequences. The correlation panel of the 2-D Resistivity Structures and 2-D Geoelectric Sections subsurface images along Traverses 1–6 is presented in Fig. 6. This figure shows that unlike the results obtained from the 2-D resistivity imaging of the subsurface, the geoelectric sections delineated five subsurface layers comprising the topsoil, laterite, weathered layer, fractured basement and the fresh basement bedrock. Table 4 contains the summary of the geoelectric parameters. The fractured basement is localized beneath VES 40 and 43.

Fig. 6
figure 6

Typical depth sounding curves obtained from the study area

Full size image
Table 4 Geoelectric parameters beneath the study area
Full size table

Figure 7 presents the parametric sounding’s result of interpretation obtained beside a located hand dug well in the study area. The figure equally shows the comparison between the hand-dug well section dug to a depth of 9.5 m and the columnar section for VES 48 which delineated a total depth to bedrock of 15.7 m within the investigated area. The lithologic correlation between the well log and the VES 48 interpretation result shows that the thickness of the lithologic units obtained for the well log are topsoil 1.5 m, laterite 1.5 m, weathered layer > 6.5 m while the VES delineated corresponding thicknesses of 1.1 m, 1.8 m, 12.8 m, respectively. Therefore, this correlation further confirmed the interpretation accuracy of the VES technique adopted and hence, its reliability in this work. The weathered layer and the fractured basement with relatively low resistivity values which range from 12–400 Ω m and 121–650 Ω m; and 23–222 Ω m and 338–714 Ω m, respectively, delineated by the 2-D resistivity structures and the geoelectric sections constitute the aquifer units with hydrogeological significance in the investigated area.

Fig. 7
figure 7

The groundwater controlling factors themes: a regolith; b bedrock relief; c hydraulic head; d coefficient of anisotropy; e aquifer resistivity; f aquifer thickness

Full size image

The 1-D depth sounding and hydrogeological investigation results

The results of the interpreted 1-D depth sounding data were analyzed in terms of the distribution of the apparent resistivity and thicknesses values from which the earth subsurface models' strata/layers were delineated. The summary of the interpreted 1-D depth sounding and the hydrogeological (well data) results are presented in Tables 1 and 2. Table 1 results were geoelectrically analyzed for lithologic layers delineation such as Regolith (Re), Bedrock relief (B), the aquifer unit's physical properties i.e. aquifer layer Resistivity (R) and aquifer layer Thickness (T). Furthermore, applying Eqs. 1 and 2 to Table 1 content, the coefficient of Anisotropy (λ) parameter “A” was determined. The Re, B, A, R, and T estimated values beneath each VES location are presented in Table 3. However, the “H” estimated values beneath each well are presented in column 7 of Table 2. According to Tables 2 and 3, the estimated values for Re, B, H, A, R, and T are in the range of 5.1–42.8 m, 198–30,040 m, 320.2–392.5 m, 0.31–1.83, 13–1641 Ω m and 2.5–41.6 m, respectively. Tables 2 and 3 results were processed in a GIS environment to produce the thematic maps for the Hydro-geophysical geoelectrical-derived Groundwater Potential Conditioning Factors (GPCFs) (Fig. 5). Table 4 presents a summary of the geoelectric parameters beneath the study area.

The GPCFs weighting assignment result

The standard scale of Saaty’s developed in 1980 (Table 5) was adopted, coupled with the expert opinion gathered based on information answered in a keenly assessed questionnaire forms administered. Coupled with the intuitive knowledge of the researchers and analyzed assigned GPCFs weight from experts, the GPCFs multivariable criteria were compared for the study area groundwater potential assessment. Evolving from this criteria comparison was a pair-wise comparison matrix presented in Table 5. Adopting similar approaches reported in the studies of Jha et al. (2010), Akinlalu et al. (2017) and Mogaji and Lim (2017), etc., Table 5 was solved to determine the criterion normalized weight (NW). Applying Eqs. (6 and 7), the consistency of the criteria normalized weights were established. The result of the applied Eq. (7,) gives the consistency ratio (CR) of 0.05 (column 8 of Table 5). In accordance with Saaty (1980), if the CR ≤ 0.1 (10%), then the CR is acceptable, implying that the matrix is consistent. Thus, the determined criteria normalized weights can be assigned to the considered GPCFs themes as shown in Table 6.

Table 5 A matrix of pair-wise comparisons of groundwater potential conditioning factors (GPCFs) for the AHP process
Full size table
Table 6 The probabilistic weightage scoring application to the GPCFs themes
Full size table

The groundwater potential prediction index (GPPI) model application results

The basic components of the developed GPPI algorithm (Eq. 9) are the normalized weight (NW) and assign rating (R) for the classes in each thematic maps (Fig. 5). These assigned NW and R have crucial hydrological implications considering that the NW component which gives insight to the relative importance of the GPCFs in contributing toward groundwater potential assessment in the area, while R defined the ranges of the aquifer expected productivity potential for groundwater storage interpretation based on the order of each factors map's class boundary influences. The quantitative analysis of the maps (Fig. 5) is presented in column 2 of Table 6. In a typical theme of the regolith (Re) factor (Fig. 5a), the class boundary with high Re values ranges of 25.63–42.75 m (column 2) is assigned the highest rating score of 4 (column 4 of Table 6). According to Oladapo et al. (2009), Mogaji (2016) and Bayode (2018), an area underlain with thicker Re unit is often observed to be characterized by high/medium groundwater yield. Thus, the aquifer expected productivity potential interpretation (Column 3 of Table 6) is in agreement. This trend is similar across the other GPCFs themes. Considering the GPCFs' weight (NW) assessment result in column 5 of Table 6 and the spatial attribute scoring values (R) based on the GIS designed template model (Fig. 8) on each GPCFs' themes, the developed GPPI multi-criteria modeling algorithm in Eq. (9) was explored for the groundwater potentiality index computation in the area. The GPPI modeling application result is presented in Table 7.

Fig. 8
figure 8

GIS-based spatial attribute modeling template

Full size image
Table 7 The GIS-based developed GPPI model application results
Full size table

Modeling of groundwater potentiality prediction zones (GPPZs) map

The estimated GPPI values resulting from the applied GPPI multi-criteria modeling equation (Eq. 9) are as presented in Table 7 (Column 9). The computed GPPI values for each grid center (Fig. 8) are noted to be a series of continuous values in the range of 1.59–3.65 for the study area. These GPPI record values were processed in a GIS environment to spatially model the groundwater potential prediction indexing (GPPI) map for the area (Fig. 9). For the potential zone demarcation, the appropriateness of the quantile classification method as applied in the studies of Razandi et al. (2015) and Rahmati and Melesse (2016) was employed. Table 8 presents the results of the potentiality zones' classification. Furthermore, the areal and percentage distribution of these predicted potential zones were evaluated in GIS environment and established that the coexisting percentage areas under Low (L), Medium (M), Medium–High (MH), and High (H) categories are 24%, 47%, 22%, and 7%, respectively (Column 3 of Table 8).

Fig. 9
figure 9

Groundwater potentiality zonation map (GPZM) of the study area

Full size image
Table 8 The potentiality classifications based on the applied GPPI model equation
Full size table

Validation of the groundwater potential prediction index model map

The functionality of a regionalized model map for practical and field usage in any environmental impact assessment is of the essence in the field of groundwater hydrology (Mogaji 2016, 2017). According to Jha et al. (2010), the most appropriate approach for reliability assessment of a predictive model map is the step-draw down test at various locations within each predicted zones. With this approach, the location-specific safe aquifer yields would have been determined. But then, the non-availability of such data made this approach implementation difficult. Thus, well data record (Table 2) obtained via hydrogeological survey measurement were explored for the map model output's reliability evaluation. This approach is in agreement with the validation technique adopted in the studies of Manap et al. (2011) and Akinlalu et al. (2017), etc. This study engaged the Relative Operating Characteristic (ROC) technique, which can be highly efficient in examining the quality of deterministic and probabilistic detection and forecast systems for performance assessment of our produced GPPI model map. The appropriateness of the ROC technique proficiency in environmental decision-making analysis has been established in the studies of Al-abadi and Shahid (2015), Al-Abadi (2015) and Mogaji (2017). The basic mechanism of ROC for performance evaluation involving the use of ROC module of the IDRISI Selva software is explored in this study. It is important to note that the Depth to water level (DWL) records observed during wet seasons (Column 5 of Table 2) was considered as a training data used in preparing the GPPI model map while the Depth to water level (DWL) records observed at the peak of dry seasons (Column 6 of Table 2) not used in GPPI model map generation (Fig. 9) were used as testing data in GIS environment. The ROC curve application result is presented in (Fig. 9). Figure 9 shows that the success and prediction rates of the produced GPPI map are 0.855 (86%) and 0.812 (81%), respectively. According to Pradhan (2013), the uniqueness of the determined success rate is that it establishes the performance of resulting GPPI map (Fig. 9) in classifying the areas’ existing wells, whereas the prediction rate describes the performance of the model and the predictor variable in anticipating the area groundwater occurrence. With these success and prediction accuracy results, the produced AHP-based GPPI map is reasonably suitable for decision-making to maximize groundwater resource development in the study area.

Furthermore, through overlaying the GPCFs' indices parameters' themes (Fig. 2) on the GPPI map (Fig. 9) in a GIS environment, further evaluation of the precision prediction based on the proposed GIS-Based GPPI multi-criteria model was carried out. For the Re, H, A, T themes, a qualitative correlation assessment across the predicted potential zones i.e., from low to high, the medium–high and high potential rate classified zones are correlated with the highest Re, H, A, T predicted values reducing down to the other classified zones (medium high, medium and low). Moreover, the medium–high and high groundwater potential classes mainly cover the southern, western and extend to the northern parts, with two smaller isolated areas observed in the eastern part of the study areas and these areas are typically underlain with saturated aquifer formation (R) (Fig. 5e, f) and relatively low bedrock relief zones (B) (Fig. 5b). In line with Omosuyi et al. (2007), Oladapo et al. (2009), Bayode (2018) and Adagunodo et al. (2018), zones characterized with saturated aquifer formation and low bedrock relief i.e., a depression zone often served as groundwater collecting center and are zones of high groundwater potential zone in a typical crystalline rock area. The correlation of (Fig. 9) spatial attribute prediction output with these aforementioned GPCFs further substantiates the appropriateness of the proposed GIS-Base GPPI multi-criteria model in environmental decision-making studies.

Conclusions

This study has successfully explored the efficacy of geospatial tool in modeling the geophysically/hydrogeological derived groundwater potentiality conditioning factors (GPCFs) such as regolith thickness, bedrock relief, hydraulic head, coefficient of anisotropy, aquifer resistivity, and aquifer thickness toward groundwater potentiality assessment in the study area. The data sources for the derived GPCFs include the 2-D, 1-D geophysical imaging data and hydrogeological well data acquired in a typical hard rock terrain. Through robust analysis of the geophysical data using appropriate software, geoelectrical parameters were determined. The interpreted geoelectric parameters enable subsurface lithological imaging of geologic features including major vertical discontinuities characterized by relatively low resistivity zones within the highly resistive basement bed rock (between distances > 370 m; 480–530 m; 180–220 m; 90–120 m along the established 2-D resistivity imaging traverses (1, 2, 5 and 6, respectively) typical of the delineated fracture zones that are of hydrogeological significance in the investigated area. The reprocessing of these geoelectrical parameters (resistivity and thickness) lead to the establishment of Dar–Zarrouk Parameters. The hydrogeological well data and the primary and secondary geoelectric parameters were processed in a GIS environment to obtain the area groundwater potentiality conditioning factors ‘themes. Applying the standard Saaty’s scale in the context of analytical hierarchy process (AHP) data mining approach, the rated and weighted thematic layers were integrated in GIS environment to compute groundwater potential prediction index (GPPI) for the area. Based on the results of the estimated GPPI in the range of between 1.59 and 3.65, the study area was classified into four groundwater potential zones. The map revealed that about 71% of the areal extent accounts for the low, and medium predicted groundwater potential classes and 29% of the area covers the moderate-high to high groundwater potential classes. The reliability evaluation of the produced GPPI map was investigated using the Reacting Operating Characteristics (ROC) success and prediction rates deterministic approach. The success and the prediction rates for the produced GPPI map are 0.855 and 0.812, respectively. Besides, the precise correlation of this GIS-Base GPPI multi-criteria model map with the GPCFs themes’ spatial attributes output further substantiates its appropriateness in environmental decision-making studies. Thus, information on the produced GPZM could be used by local authorities and water policy makers as a preliminary reference in selecting suitable sites for drilling new boreholes. The findings affirmed the identification of areas where aquifers can be developed. The research output will greatly contribute to the precise location of borehole site that could address the groundwater resources development challenges in the investigated area and another area with similar geology.